ESMO 2026 Lung Cancer Slides

16:30 - 18:15 Presidential Symposium I CHAIRS: FABRICE ANDRÉ, JAMES LARKIN Tarlatamab plus durvalumab significantly improved overall survival versus durvalumab alone as first-line maintenance therapy DeLLphi 305 | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 281) Median OS, months (95% CI) | 21.1 (20.8–NE) | 12.7 (11.2–14.8) Median FU, months | 15.9 | 16.1 HR (95% CI); P-value (2-sided) | 0.42 (0.32–0.54); P < 0.001 Curve labels: 92% | 75% | 79% | 53% Percentage of Patients | Months 0 3 6 9 12 15 18 21 24 27 Number of patients at risk: Tarlatamab + Durvalumab | 282 | 276 | 258 | 246 | 222 | 132 | 55 | 9 | 0 Durvalumab | 281 | 252 | 208 | 171 | 146 | 81 | 33 | 8 | 1 | 0 [footnotes small; partially legible] Data cut-off date: [illegible] ... [illegible] events had occurred in the tarlatamab-durvalumab arm, and [illegible] events had occurred in the durvalumab arm. P value was calculated using a stratified log-rank test. CI, confidence interval; FU, follow-up; HR, hazard ratio; NE, not estimable; OS, overall survival Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study Jacob Sands — Tarlatamab plus durvalumab (durva) vs durva as 1L maintenance (1LM) therapy following durva, platinum, and etoposide (1L durva + EP) in extensive-stage small cell lung cancer (ES-SCLC): Primary analysis of the phase 3 DeLLphi-305 trial ALICANTE AUDITORIUM - HALL 6 Survival table Endpoint: Overall survival Arm | N | Median | Median CI Tarlatamab + Durvalumab | 282 | 21.1 months | 20.8–NE Durvalumab | 281 | 12.7 months | 11.2–14.8 HR: 0.42 HR CI: 0.32–0.54 P: P < 0.001 (2-sided) Landmark rates: 6 months: Tarlatamab + Durvalumab 92% | Durvalumab 75% 12 months: Tarlatamab + Durvalumab 79% | Durvalumab 53%

[Slide 1] DeLLphi-305: Study design A Phase 3, Open-label, Multicenter, Randomized Study of Tarlatamab in Combination With Durvalumab vs Durvalumab Alone in Patients With ES-SCLC Following Platinum, Etoposide and Durvalumab (NCT06211036) 1L Chemo-IO: Platinum-etoposide + durvalumab(a) induction therapy (4 cycles) Key Inclusion Criteria: - Age ≥ 18 years - No disease progression - ECOG PS 0–1 - Asymptomatic untreated or treated brain metastasis R 1:1, N = 563 Treatment until progression or unacceptable toxicity Tarlatamab (10 mg Q2W)(b) + Durvalumab (1500 mg Q4W)(c) n = 282 Durvalumab (1500 mg Q4W)(c) n = 281 Primary Endpoint: OS Key Secondary Endpoint: PFS Secondary Endpoints: ORR, DOR, safety Stratification Factors: - Brain and/or liver metastasis (Yes/No) - Response to 1L chemo-IO (PR/CR vs SD) - Prior platinum type (cisplatin vs carboplatin) Interim analysis for OS: 243 (61% IF) OS events, 0.008 two-sided P-value boundary Endpoints were tested sequentially according to a prespecified hierarchical testing strategy: OS → PFS → ORR Data cut-off date: 7 JUL 2026. (a)Eligible patients must have received ≥ 3 cycles of concomitant durvalumab as 1L treatment of ES-SCLC. (b)Patients were monitored on site for a minimum of 1 to 2 hours, or 6 to 8 hours in certain regions after the first two doses of tarlatamab. (c)Each treatment cycle consists of 28 days. ClinicalTrials.gov. https://clinicaltrials.gov/study/NCT06211036. Accessed 31 August 2026. Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study [Slide 2] Tarlatamab plus durvalumab significantly improved overall survival versus durvalumab alone as first-line maintenance therapy | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 281) Median OS, months (95% CI) | 21.1 (20.8–NE) | 12.7 (11.2–14.8) Median FU, months | 15.9 | 16.1 HR (95% CI); P-value (2-sided)(a) | 0.42 (0.32–0.54); P < 0.001 Curve labels: Tarlatamab + Durvalumab 92% (6 mo), 79% (12 mo); Durvalumab 75% (6 mo), 53% (12 mo) Y axis: Percentage of Patients; X axis: Months 0-27 Number of patients at risk: Months | 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 | 24 | 27 Tarlatamab + Durvalumab | 282 | 276 | 258 | 246 | 222 | 132 | 55 | 9 | 0 Durvalumab | 281 | 252 | 208 | 171 | 146 | 81 | 33 | 8 | 1 | 0 Data cut-off date: 7 JUL 2026. At the time of data cut-off, 87 events had occurred in the tarlatamab-durvalumab arm, and 156 events had occurred in the durvalumab arm. (a)P value was calculated using a stratified log-rank test. CI, confidence interval; FU, follow-up; HR, hazard ratio; NE, not estimable; OS, overall survival. Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study Survival table Endpoint: OS Arm | N | Median | Median CI Tarlatamab + Durvalumab | 282 | 21.1 mo | 20.8–NE Durvalumab | 281 | 12.7 mo | 11.2–14.8 HR: 0.42 HR CI: 0.32–0.54 P: P < 0.001 (2-sided, stratified log-rank) Landmark rates: 6 mo: Tarlatamab + Durvalumab 92% | Durvalumab 75% 12 mo: Tarlatamab + Durvalumab 79% | Durvalumab 53% [Slide 3] Tarlatamab plus durvalumab significantly improved PFS versus durvalumab | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 281) Median PFS, months (95% CI) | 7.4 (5.6–9.1) | 1.9 (1.9–2.0) Median FU, months | 13.9 | 13.9 HR (95% CI); P-value (2-sided)(a) | 0.34 (0.28–0.42); P < 0.001 Curve labels: Tarlatamab + Durvalumab 53% (6 mo), 38% (12 mo); Durvalumab 17% (6 mo), 10% (12 mo) Y axis: Percentage of Patients; X axis: Months 0-24 Number of patients at risk: Months | 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 | 24 Tarlatamab + Durvalumab | 282 | 225 | 147 | 122 | 86 | 32 | 14 | 1 | 0 Durvalumab | 281 | 106 | 46 | 36 | 18 | 8 | 2 | 0 Data cut-off date: 7 JUL 2026. At the time of data cut-off, 182 events had occurred in the tarlatamab-durvalumab arm, and 250 events had occurred in the durvalumab arm. (a)P-value was calculated using a stratified log-rank test. CI, confidence interval; FU, follow-up; HR, hazard ratio; PFS, progression-free survival. Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study Survival table Endpoint: PFS Arm | N | Median | Median CI Tarlatamab + Durvalumab | 282 | 7.4 mo | 5.6–9.1 Durvalumab | 281 | 1.9 mo | 1.9–2.0 HR: 0.34 HR CI: 0.28–0.42 P: P < 0.001 (2-sided, stratified log-rank) Landmark rates: 6 mo: Tarlatamab + Durvalumab 53% | Durvalumab 17% 12 mo: Tarlatamab + Durvalumab 38% | Durvalumab 10% [Slide 4] Tarlatamab plus durvalumab demonstrated significantly improved objective response rate versus durvalumab | Tarlatamab + Durvalumab (n = 235)(a) | Durvalumab (n = 244)(a) BOR, n (%)(a,b) Confirmed CR(c) | 2 (0.9) | 1 (0.4) Confirmed PR(c) | 60 (25.5) | 9 (3.7) Stable disease | 127 (54.0) | 74 (30.3) Progressive disease | 43 (18.3) | 150 (61.5) ORR, % (95% CI) | 26.4 (20.9–32.5) | 4.1 (2.0–7.4) P-value (2-sided) | < 0.001 Median DOR, months (95% CI) | 12.7 (7.4–NE) | 5.6 (3.6–NE) Duration of response [KM curve]: Responders Without Progression or Death (%) vs Months 0-18 Curve labels: Tarlatamab + Durvalumab 70% (6 mo), 52% (12 mo); Durvalumab 38% (6 mo) Number of patients at risk: Months | 0 | 3 | 6 | 9 | 12 | 15 | 18 Tarlatamab + Durvalumab | 62 | 58 | 36 | 27 | 11 | 5 | 0 Durvalumab | 10 | 10 | 3 | 0 As of data cut-off, 43.6% and 11.4% of patients remained on treatment in tarlatamab plus durvalumab and durvalumab arm, respectively Data cut-off date: 7 JUL 2026. Median follow-up: 10.2 months (tarlatamab + durvalumab arm); 8.3 months (durvalumab arm). (a)Based on RECIST 1.1 per investigator assessment in patients with measurable disease at baseline at the time of randomization. (b)Postbaseline scan data was not available for 3 (1.3%) patients in tarlatamab + durvalumab arm and 10 (4.1%) in durvalumab arm. (c)Confirmation of a CR or a PR was required and had to take place no fewer than 4 weeks after the initial documentation of a CR or a PR. 1L, first line; BOR, best overall response; CI, confidence interval; CR, complete response; DOR, duration of response; NE, not estimable; ORR, objective response rate; PR, partial response; RECIST, Response Evaluation Criteria in Solid Tumors. Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study

[Slide 1] Proffered Paper session LBA3 - Tarlatamab plus durvalumab (durva) vs durva as 1L maintenance (1LM) therapy following durva, platinum, and etoposide (1L durva + EP) in extensive-stage small cell lung cancer (ES-SCLC): Primary analysis of the phase 3 DeLLphi-305 trial Speakers Jacob Sands (Boston, United States of America) Lecture Time 17:18 - 17:30 Abstract Background Adding tarlatamab to 1L ES-SCLC therapy showed encouraging efficacy in the phase 1b DeLLphi-303 study (NCT05361395). We report the primary analysis of the randomized phase 3 DeLLphi-305 trial evaluating tarlatamab + durva vs durva as 1LM therapy for ES-SCLC. [Slide 2] Conclusions Tarlatamab + durva significantly improved OS, PFS, and ORR in 1LM ES-SCLC. The overall incidence of adverse events was comparable between groups. CRS and ICANS events were primarily low grade and no fatal events occurred. These results establish tarlatamab + durva as a new standard of care 1LM therapy for ES-SCLC.

[Slide 1] TARLATAMAB–DURVALUMAB IN EXTENSIVE-STAGE SCLC Tarlatamab–durvalumab | Durvalumab No. of Patients | Median Overall Survival (95% CI), mo Tarlatamab–Durvalumab | 282 | 21.1 (20.8–NE) Durvalumab | 281 | 12.7 (11.2–14.8) Hazard ratio for death, 0.42 (95% CI, 0.32–0.54) P<0.001 Curve labels: 91.5 and 74.6 at 6 months; 79.4 and 53.3 at 12 months No. at Risk (months 0, 3, 6, 9, 12, 15, 18, 21, 24, 27) Tarlatamab–durvalumab | 282 | 276 | 258 | 246 | 222 | 132 | 55 | 9 | 0 Durvalumab | 281 | 252 | 208 | 171 | 146 | 81 | 33 | 8 | 1 | 0 Figure 1. Overall Survival (Intention-to-Treat Population). Shown are Kaplan–Meier curves of overall survival in the intention-to-treat population, which included all the patients who had undergone randomization. Overall survival was defined as the time from randomization to death from any cause. Tick marks indicate censored data, and the horizontal dashed line the median. A hazard ratio below 1.0 indicates better overall survival with tarlatamab–durvalumab than with durvalumab. The P value was calculated with the use of a stratified log-rank test controlled for the presence of brain or liver metastases and response to first-line chemoimmunotherapy. NE denotes could not be estimated. Survival table Endpoint: Overall survival (ITT) Arm | N | Median | Median CI Tarlatamab–durvalumab | 282 | 21.1 mo | 20.8–NE Durvalumab | 281 | 12.7 mo | 11.2–14.8 HR: 0.42 HR CI: 0.32–0.54 P: <0.001 Landmark rates: 6 mo: Tarlatamab–durvalumab 91.5% | Durvalumab 74.6% 12 mo: Tarlatamab–durvalumab 79.4% | Durvalumab 53.3% [Slide 2] Figure S3. Overall Survival Analysis in Prespecified Subgroups Subgroup | Tarlatamab + Durvalumab | Durvalumab (no. of events / no. of patients) | Hazard Ratio for Death (95% CI) Age < 65 years | 34 / 139 | 69 / 133 | 0.36 (0.24–0.54) ≥ 65 years | 53 / 143 | 87 / 148 | 0.48 (0.34–0.67) Sex Male | 64 / 197 | 120 / 201 | 0.38 (0.28–0.51) Female | 23 / 85 | 36 / 80 | 0.56 (0.33–0.94) Race White | 54 / 163 | 108 / 181 | 0.40 (0.29–0.56) Black | 0 / 4 | 0 / 4 | - Asian | 31 / 109 | 45 / 92 | 0.47 (0.30–0.74) Other | 2 / 6 | 3 / 4 | - Region Americas | 13 / 36 | 16 / 39 | 0.82 (0.39–1.71) Asia-Pacific | 32 / 114 | 51 / 101 | 0.43 (0.28–0.67) Europe | 42 / 132 | 89 / 141 | 0.36 (0.25–0.52) Presence (previous or current) of brain metastases Yes | 15 / 55 | 26 / 56 | 0.49 (0.26–0.93) No | 72 / 227 | 130 / 225 | 0.41 (0.30–0.54) Presence (previous or current) of liver metastases Yes | 43 / 88 | 69 / 95 | 0.41 (0.28–0.61) No | 44 / 194 | 87 / 186 | 0.40 (0.28–0.57) Presence (previous or current) of brain metastases and/or liver metastases Yes | 52 / 131 | 84 / 134 | 0.45 (0.32–0.64) No | 35 / 151 | 72 / 147 | 0.38 (0.25–0.57) Response to initial therapy CR/PR | 77 / 251 | 135 / 248 | 0.42 (0.32–0.56) SD | 10 / 31 | 21 / 33 | 0.40 (0.19–0.85) Type of platinum first received for ES-SCLC Cisplatin | 11 / 44 | 23 / 44 | 0.33 (0.16–0.67) Carboplatin | 76 / 238 | 132 / 236 | 0.44 (0.34–0.59) Number of durvalumab cycles prior to enrollment 3 | 13 / 56 | 27 / 51 | 0.36 (0.19–0.70) 4 | 74 / 226 | 129 / 230 | 0.44 (0.33–0.58) Received prophylactic cranial irradiation (PCI) Yes | 3 / 14 | 1 / 4 | - No | 84 / 268 | 155 / 277 | 0.42 (0.32–0.55) Disease stage at diagnosis Stage IV | 87 / 269 | 151 / 269 | 0.44 (0.33–0.57) Non-Stage IV | 0 / 13 | 5 / 12 | - Tarlatamab + Durvalumab Better | Durvalumab Better [Slide 3] Table S3. Subsequent Anti-cancer Treatment | Tarlatamab + Durvalumab (N = 282) | Durvalumab* (N = 278) Patients discontinuing study treatment, n | 159 | 246 Patients reporting subsequent anti-cancer treatment†, n | 104 | 179 All chemotherapy | 91 | 135 Anti DLL3 Therapy | 6 | 40 T cell engager | 3 | 34 ADC | 2 | 1 Unspecified | 1 | 5 Radiation | 17 | 20 Other | 4 | 5 ADC other than DLL3 | 7 | 21 PD-(L)1 Checkpoint Inhibitor | 12 | 20 Tyrosine Kinase Inhibitor | 7 | 21 Concomitant medications were coded using the WHO Drug Dictionary version GLOBALB3Mar26. *Three patients were randomized to the durvalumab arm but did not receive treatment. One patient withdrew full consent, and two patients partially withdrew consent following randomization and remained on study for survival follow up. †The subsequent anti-cancer therapy was based on pre-defined list of categories and preferred terms. Some of the treatments were given in combination but listed separately. Concomitant medications were coded using the WHO Drug Dictionary version GLOBALB3Mar26. ADC, antibody-drug conjugate; DLL3, Delta-like ligand 3; N, number of patients in the analysis set; n, number of patients with observed data; PD-(L)1, programmed cell death-(ligand) 1; WHO, World Health Organization. [Slide 4] Table S4. Subsequent Anti-Cancer Treatment by Region | Tarlatamab + Durvalumab: USA (N = 16) | Japan (N = 28) | EU (N = 132) | Others (N = 106) | Durvalumab: USA (N = 22) | Japan (N = 21) | EU (N = 140) | Others (N = 95) Patients reporting subsequent anti-cancer treatment, n (%) | 8 (50.0) | 17 (60.7) | 35 (26.5) | 44 (41.5) | 16 (72.7) | 17 (81.0) | 85 (60.7) | 61 (64.2) All chemotherapy | 5 (31.3) | 15 (53.6) | 31 (23.5) | 40 (37.7) | 5 (22.7) | 14 (66.7) | 69 (49.3) | 47 (49.5) Anti DLL3 Therapy | 0 (0.0) | 3 (10.7) | 1 (0.8) | 2 (1.9) | 14 (63.6) | 8 (38.1) | 10 (7.1) | 8 (8.4) T cell engager | 0 (0.0) | 3 (10.7) | 0 (0.0) | 0 (0.0) | 14 (63.6) | 8 (38.1) | 7 (5.0) | 5 (5.3) ADC | 0 (0.0) | 0 (0.0) | 1 (0.8) | 1 (0.9) | 0 (0.0) | 0 (0.0) | 1 (0.7) | 0 (0.0) Unspecified | 0 (0.0) | 0 (0.0) | 0 (0.0) | 1 (0.9) | 0 (0.0) | 0 (0.0) | 2 (1.4) | 3 (3.2) Radiation | 2 (12.5) | 3 (10.7) | 4 (3.0) | 8 (7.5) | 1 (4.5) | 2 (9.5) | 7 (5.0) | 10 (10.5) Other | 1 (6.3) | 2 (7.1) | 0 (0.0) | 1 (0.9) | 0 (0.0) | 2 (9.5) | 0 (0.0) | 3 (3.2) ADC other than DLL3 | 0 (0.0) | 0 (0.0) | 2 (1.5) | 5 (4.7) | 0 (0.0) | 2 (9.5) | 3 (2.1) | 16 (16.8) PD-(L)1 Checkpoint Inhibitor | 1 (6.3) | 1 (3.6) | 4 (3.0) | 6 (5.7) | 2 (9.1) | 0 (0.0) | 5 (3.6) | 13 (13.7) Tyrosine Kinase Inhibitor | 0 (0.0) | 0 (0.0) | 0 (0.0) | 7 (6.6) | 0 (0.0) | 0 (0.0) | 1 (0.7) | 20 (21.1) The subsequent anti-cancer therapy was based on pre-defined list of categories and preferred terms. Some of the treatments were given in combination but listed separately. Concomitant medications were coded using the WHO Drug Dictionary version GLOBALB3Mar26. ADC, antibody-drug conjugate; DLL3, Delta-like ligand 3; EU, Europe; N, number of patients in the analysis set; n, number of patients with observed data; PD-(L)1, programmed cell death-(ligand) 1; USA, United States of America; WHO, World Health Organization.

[Slide 2] On-screen broadcast chrome: '16:30 - 18:15 Presidential Symposium I'; 'CHAIRS: FABRICE ANDRÉ, JAMES LARKIN'; speaker card 'Jacob Sands'; 'ALICANTE AUDITORIUM - HALL 6'. Tarlatamab plus durvalumab significantly improved overall survival versus durvalumab alone as first-line maintenance therapy DeLLphi 305 | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 281) Median OS, months (95% CI) | 21.1 (20.8–NE) | 12.7 (11.2–14.8) Median FU, months | 15.9 | 16.1 HR (95% CI); P-value (2-sided)a | 0.42 (0.32–0.54); P < 0.001 Curve labels: Tarlatamab + Durvalumab 92% (6 mo), 79% (12 mo); Durvalumab 75% (6 mo), 53% (12 mo) Percentage of Patients / Months (0–27) Number of patients at risk: Months | 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 | 24 | 27 Tarlatamab + Durvalumab | 282 | 276 | 258 | 246 | 222 | 132 | 55 | 9 | 0 | Durvalumab | 281 | 252 | 208 | 171 | 146 | 81 | 33 | 8 | 1 | 0 Data cut-off date: 7 JUL 2026. At the time of data cut-off, 87 events had occurred in the tarlatamab-durvalumab arm, and 156 events had occurred in the durvalumab arm. aP value was calculated using a stratified log-rank test. CI, confidence interval; FU, follow-up; HR, hazard ratio; NE, not estimable; OS, overall survival. Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study Survival table Endpoint: Overall survival (ITT) Arm | N | Median | Median CI Tarlatamab + Durvalumab | 282 | 21.1 months | 20.8–NE Durvalumab | 281 | 12.7 months | 11.2–14.8 HR: 0.42 HR CI: 0.32–0.54 P: P < 0.001 (2-sided, stratified log-rank) Landmark rates: 6 months: Tarlatamab + Durvalumab 92% | Durvalumab 75% 12 months: Tarlatamab + Durvalumab 79% | Durvalumab 53% [Slide 3] On-screen broadcast chrome: '16:30 - 18:15 Presidential Symposium I'; 'CHAIRS: FABRICE ANDRÉ, JAMES LARKIN'; speaker card 'Jacob Sands'; 'ALICANTE AUDITORIUM - HALL 6'. Tarlatamab plus durvalumab significantly improved PFS versus durvalumab DeLLphi 305 | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 281) Median PFS, months (95% CI) | 7.4 (5.6–9.1) | 1.9 (1.9–2.0) Median FU, months | 13.9 | 13.9 HR (95% CI); P-value (2-sided)a | 0.34 (0.28–0.42); P < 0.001 Curve labels: Tarlatamab + Durvalumab 53% (6 mo), 38% (12 mo); Durvalumab 17% (6 mo), 10% (12 mo) Percentage of Patients / Months (0–24) Number of patients at risk: Months | 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 | 24 Tarlatamab + Durvalumab | 282 | 225 | 147 | 122 | 86 | 32 | 14 | 1 | 0 Durvalumab | 281 | 106 | 46 | 36 | 18 | 8 | 2 | 0 | Data cut-off date: 7 JUL 2026. At the time of data cut-off, 182 events had occurred in the tarlatamab-durvalumab arm, and 250 events had occurred in the durvalumab arm. aP-value was calculated using a stratified log-rank test. CI, confidence interval; FU, follow-up; HR, hazard ratio; PFS, progression-free survival. Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study Survival table Endpoint: Progression-free survival Arm | N | Median | Median CI Tarlatamab + Durvalumab | 282 | 7.4 months | 5.6–9.1 Durvalumab | 281 | 1.9 months | 1.9–2.0 HR: 0.34 HR CI: 0.28–0.42 P: P < 0.001 (2-sided, stratified log-rank) Landmark rates: 6 months: Tarlatamab + Durvalumab 53% | Durvalumab 17% 12 months: Tarlatamab + Durvalumab 38% | Durvalumab 10% [Slide 4] On-screen broadcast chrome: '16:30 - 18:15 Presidential Symposium I'; 'CHAIRS: FABRICE ANDRÉ, JAMES LARKIN'; speaker card 'Jacob Sands'; 'ALICANTE AUDITORIUM - HALL 6'. CRS and ICANS events profile is consistent with the known safety profile of tarlatamab CRSa,b, n (%) | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 278) Any grade | 140 (49.6) | 4 (1.4) Grade 2 | 21 (7.4) | 2 (0.7) Grade 3 | 1 (0.4) | 0 (0.0) Grade 4 | 0 (0.0) | 0 (0.0) Serious adverse events | 36 (12.8) | 2 (0.7) Leading to discontinuation of tarlatamab | 1 (0.4) | NA Fatal adverse events | 0 (0.0) | 0 (0.0) Utilization of any CRS interventionsc | 65 (23.0) | 2 (0.7) ICANSd, n (%) | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 278) Any grade | 3 (1.1) | 0 (0.0) Grade 2 | 1 (0.4) | 0 (0.0) Grade 3 | 0 (0.0) | 0 (0.0) Grade 4 | 0 (0.0) | 0 (0.0) Serious adverse events | 0 (0.0) | 0 (0.0) Leading to discontinuation of tarlatamab | 0 (0.0) | NA Fatal adverse events | 0 (0.0) | 0 (0.0) • Most CRS events resolved with supportive care (and antipyretics) – Interventions included corticosteroids (21.6%), tocilizumab (3.9%), and low-flow oxygen (1.4%) Data cut-off date: 7 JUL 2026. aCRS events included one patient with cytokine storm. bCRS and ICANS events were graded using ASTCT 2019 criteria. cOther supportive care interventions included intravenous fluids, vasopressor, and supplemental (low-flow [≤ 6 L/min]) oxygen. dPotentially associated neurologic events (including ICANS) occurred in 27 (9.6%) and 19 (6.8%) patients, respectively. Events were predominantly grade 1 or 2 and no fatal events occurred. [footnote very small; figures 27 (9.6%) / 19 (6.8%) read with moderate confidence] ASTCT, American Society for Transplantation and Cellular Therapy; CRS, cytokine release syndrome; ICANS, immune effector cell-associated neurotoxicity syndrome; NA, not applicable. Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study

CRS and ICANS events profile is consistent with the known safety profile of tarlatamab CRSᵃ,ᵇ, n (%) | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 278) Any grade | 140 (49.6) | 4 (1.4) Grade 2 | 21 (7.4) | 2 (0.7) Grade 3 | 1 (0.4) | 0 (0.0) Grade 4 | 0 (0.0) | 0 (0.0) Serious adverse events | 36 (12.8) | 2 (0.7) Leading to discontinuation of tarlatamab | 1 (0.4) | NA Fatal adverse events | 0 (0.0) | 0 (0.0) Utilization of any CRS interventionsᶜ | 65 (23.0) | 2 (0.7) ICANSᵈ, n (%) | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 278) Any grade | 3 (1.1) | 0 (0.0) Grade 2 | 1 (0.4) | 0 (0.0) Grade 3 | 0 (0.0) | 0 (0.0) Grade 4 | 0 (0.0) | 0 (0.0) Serious adverse events | 0 (0.0) | 0 (0.0) Leading to discontinuation of tarlatamab | 0 (0.0) | NA Fatal adverse events | 0 (0.0) | 0 (0.0) • Most CRS events resolved with supportive care (and antipyretics) – Interventions included corticosteroids (21.6%), tocilizumab (3.9%), and low-flow oxygen (1.4%) Data cut-off date: 7 JUL 2026. [Footnotes in small print, partially legible: ᵃCRS events included one patient with cytokine storm. ᵇCRS and ICANS events were graded using ASTCT 2019 criteria. ᶜOther supportive care interventions included intravenous fluids, vasopressor, and supplemental (low flow [≤ 6 L/min]) oxygen. ᵈPotentially associated neurologic events (including ICANS) occurred in 27 (9.6%) and 19 (6.8%) patients, respectively. Events were predominantly grade 1 or 2 and no fatal events occurred. ASTCT, American Society for Transplantation and Cellular Therapy; CRS, cytokine release syndrome; ICANS, immune effector cell-associated neurotoxicity syndrome; NA, not applicable.] Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study

[Slide 3] low grade (Table). Table: LBA3 | Tarlatamab + durva (n = 282) | Durva (n = 281) OS mOS, mo (95% CI) | 21.1 (20.8, NE) | 12.7 (11.2, 14.8) HR (95% CI); P value | 0.42 (0.32, 0.54); P < 0.001 12 mo OS rate, % (95% CI) | 79.4 (74.2, 83.7) | 53.3 (47.3, 59.0) Median follow-up, mo | 15.9 | 16.1 PFS mPFS | 7.4 (5.6, 9.1) | 1.9 (1.9, 2.0) HR (95% CI); P value | 0.34 (0.28, 0.42); P < 0.001 ORR, % (95% CI) | 26.4 (20.9, 32.5) | 4.1 (2.0, 7.4) P value | P < 0.001 TEAEs, n (%) | (n = 282) | (n = 278) Any grade | 281 (99.6) | 258 (92.8) Gr ≥ 3 | 132 (46.8) | 113 (40.6) CRS^a, n (%) Any grade | 140 (49.6) | 4 (1.4) Gr ≥ 3 | 1 (0.4) | 0 ICANS^a, n (%) Any grade | 3 (1.1) | 0 Gr ≥ 3 | 0 | 0 ^aNo fatal events occurred. NE, not estimable; Gr, grade; TEAE, treatment-emergent adverse event. [Slide 4] Methods 563 patients (pts) with ES-SCLC without progression after 1L durva + EP were randomized (1:1) to receive either tarlatamab + durva (n = 282) or durva (n = 281) until progression or unacceptable toxicity. Tarlatamab was given 1 mg IV cycle 1 day 1 (C1D1); 10 mg IV C1D8, C1D15, then 10 mg IV Q2W; durva was given 1500 mg IV Q4W. The primary endpoint was overall survival (OS) with key secondary endpoint progression-free survival (PFS) per investigator assessment, each measured from time of randomization. Additional secondary endpoints were objective response rate (ORR) and safety. Results With median (m) follow-up time of 15.9 and 16.1 months (mo), respectively, tarlatamab + durva significantly improved OS (HR: 0.42; 95% CI 0.32, 0.54; P <0.001; mOS: 21.1 [95% CI, 20.8 to NE] vs 12.7 mo [95% CI, 11.2 to 14.8]). Tarlatamab + durva significantly improved PFS (HR: 0.34; 95% CI 0.28, 0.42; P < 0.001; mPFS: 7.4 vs 1.9 mo). ORR was 26.4% (95% CI 20.9%, 32.5%) for tarlatamab + durva and 4.1% (95% CI 2.0%, 7.4%) for durva (P < 0.001). Grade ≥ 3 TEAEs occurred in 46.8% (tarlatamab + durva) and 40.6% (durva) of pts. Cytokine release syndrome (CRS) and immune-effector cell-associated neurotoxicity syndrome (ICANS) occurred in 49.6% and 1.1% of tarlatamab + durva pts and were mostly low grade (Table). Table: LBA3

[Slide 1] MADRID 2026 ESMO congress Tarlatamab plus durvalumab vs durvalumab as first-line maintenance therapy following first-line durvalumab, platinum, and etoposide in extensive-stage small cell lung cancer: primary analysis of the phase 3 DeLLphi-305 trial Jacob Sands, Tatsuya Yoshida, Myung-Ju Ahn, Carl M. Gay, Martin Schuler, Haifeng Liu, Ahmet Sezer, Ahmet Bilici, Ippokratis Korantzis, Sun Min Lim, Isamu Okamoto, Oscar Juan-Vidal, Yuanbin Chen, Gustavo Schvartsman, Anne-Marie C. Dingemans, Jeffrey M. Clarke, Miguel Marigil, Yuyang Zhang, Bazle Abidi, Maurice Pérol Presenter: Jacob Sands, MD Dana-Farber Cancer Institute Boston, MA, United States of America 24 October 2026 [Slide 3] Tarlatamab plus durvalumab significantly improved overall survival versus durvalumab alone as first-line maintenance therapy | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 281) Median OS, months (95% CI) | 21.1 (20.8-NE) | 12.7 (11.2-14.8) Median FU, months | 15.9 | 16.1 HR (95% CI); P-value (2-sided)(a) | 0.42 (0.32-0.54); P < 0.001 Curve labels: 6 mo: 92% (T+D), 75% (D); 12 mo: 79% (T+D), 53% (D) Number of patients at risk: Months | 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 | 24 | 27 Tarlatamab + Durvalumab | 282 | 276 | 258 | 246 | 222 | 132 | 55 | 9 | 0 | Durvalumab | 281 | 252 | 208 | 171 | 146 | 81 | 33 | 8 | 1 | 0 Data cut-off date: 7 JUL 2026. At the time of data cut-off, 87 events had occurred in the tarlatamab-durvalumab arm, and 156 events had occurred in the durvalumab arm. (a)P value was calculated using a stratified log-rank test. Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study Survival table Endpoint: OS Arm | N | Median | Median CI Tarlatamab + Durvalumab | 282 | 21.1 | 20.8-NE Durvalumab | 281 | 12.7 | 11.2-14.8 HR: 0.42 HR CI: 0.32-0.54 P: < 0.001 (2-sided, stratified log-rank) Landmark rates: 6 months: Tarlatamab + Durvalumab 92% | Durvalumab 75% 12 months: Tarlatamab + Durvalumab 79% | Durvalumab 53% [Slide 4] Tarlatamab plus durvalumab significantly improved PFS versus durvalumab | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 281) Median PFS, months (95% CI) | 7.4 (5.6-9.1) | 1.9 (1.9-2.0) Median FU, months | 13.9 | 13.9 HR (95% CI); P-value (2-sided)(a) | 0.34 (0.28-0.42); P < 0.001 Curve labels: 6 mo: 53% (T+D), 17% (D); 12 mo: 38% (T+D), 10% (D) Number of patients at risk: Months | 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 | 24 Tarlatamab + Durvalumab | 282 | 225 | 147 | 122 | 86 | 32 | 14 | 1 | 0 Durvalumab | 281 | 106 | 46 | 36 | 18 | 8 | 2 | 0 | Data cut-off date: 7 JUL 2026. At the time of data cut-off, [illegible] events had occurred in the tarlatamab-durvalumab arm, and [illegible] events had occurred in the durvalumab arm. Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study Survival table Endpoint: PFS Arm | N | Median | Median CI Tarlatamab + Durvalumab | 282 | 7.4 | 5.6-9.1 Durvalumab | 281 | 1.9 | 1.9-2.0 HR: 0.34 HR CI: 0.28-0.42 P: < 0.001 (2-sided) Landmark rates: 6 months: Tarlatamab + Durvalumab 53% | Durvalumab 17% 12 months: Tarlatamab + Durvalumab 38% | Durvalumab 10%

LBA3 - Tarlatamab plus durvalumab (durva) vs durva as 1L maintenance (1LM) therapy following durva, platinum, and etoposide (1L durva + EP) in extensive-stage small cell lung cancer (ES-SCLC): Primary analysis of the phase 3 DeLLphi-305 trial Speakers Jacob Sands (Boston, United States of America) Lecture Time 17:18 - 17:30 Abstract Background Adding tarlatamab to 1L ES-SCLC therapy showed encouraging efficacy in the phase 1b DeLLphi-303 study (NCT05361395). We report the primary analysis of the randomized phase 3 DeLLphi-305 trial evaluating tarlatamab + durva vs durva as 1LM therapy for ES-SCLC. Methods 563 patients (pts) with ES-SCLC without progression after 1L durva + EP were randomized (1:1) to receive either tarlatamab + durva (n = 282) or durva (n = 281) until progression or unacceptable toxicity. Tarlatamab was given 1 mg IV cycle 1 day 1 (C1D1); 10 mg IV C1D8, C1D15, then 10 mg IV Q2W; durva was given 1500 mg IV Q4W. The primary endpoint was overall survival (OS) with key secondary endpoint progression-free survival (PFS) per investigator assessment, each measured from time of randomization. Additional secondary endpoints were objective response rate (ORR) and safety. Results With median (m) follow-up time of 15.9 and 16.1 months (mo), respectively, tarlatamab + durva significantly improved OS (HR: 0.42; 95% CI 0.32, 0.54; P <0.001; mOS: 21.1 [95% CI, 20.8 to NE] vs 12.7 mo [95% CI, 11.2 to 14.8]). Tarlatamab + durva significantly improved PFS (HR: 0.34; 95% CI 0.28, 0.42; P < 0.001; mPFS: 7.4 vs 1.9 mo). ORR was 26.4% (95% CI 20.9%, 32.5%) for tarlatamab + durva and 4.1% (95% CI 2.0%, 7.4%) for durva (P < 0.001). Grade ≥ 3 TEAEs occurred in 46.8% (tarlatamab + durva) and 40.6% (durva) of pts. Cytokine release syndrome (CRS) and immune-effector cell-associated neurotoxicity syndrome (ICANS) occurred in 49.6% and 1.1% of tarlatamab + durva pts and were mostly low grade (Table). Table: LBA3 | | Tarlatamab + durva (n = 282) | Durva (n = 281) | OS mOS, mo (95% CI) | 21.1 (20.8, NE) | 12.7 (11.2, 14.8) HR (95% CI); P value | 0.42 (0.32, 0.54); P < 0.001 | 12 mo OS rate, % (95% CI) | 79.4 (74.2, 83.7) | 53.3 (47.3, 59.0) Median follow-up, mo | 15.9 | 16.1 PFS mPFS | 7.4 (5.6, 9.1) | 1.9 (1.9, 2.0) HR (95% CI); P value | 0.34 (0.28, 0.42); P < 0.001 | ORR, % (95% CI) | 26.4 (20.9, 32.5) | 4.1 (2.0, 7.4) P value | P < 0.001 | TEAEs, n (%) | (n = 282) | (n = 278) Any grade | 281 (99.6) | 258 (92.8) Gr ≥ 3 | 132 (46.8) | 113 (40.6) CRS^a, n (%) Any grade | 140 (49.6) | 4 (1.4) Gr ≥ 3 | 1 (0.4) | 0 ICANS^a, n (%) Any grade | 3 (1.1) | 0 Gr ≥ 3 | 0 | 0 ^aNo fatal events occurred. NE, not estimable; Gr, grade; TEAE, treatment-emergent adverse event. Conclusions Tarlatamab + durva significantly improved OS, PFS, and ORR in 1LM ES-SCLC. The overall incidence of adverse events was comparable between groups. CRS and ICANS events were primarily low grade and no fatal events occurred. These results establish tarlatamab + durva as a new standard of care 1LM therapy for ES-SCLC.

[Slide 1] Abstract number: LBA33 FIH study of GFH276 (NCT07198321) • As of 15 June 2026, a total of 98 patients were enrolled in the Phase Ia study and treat with GFH276 5~320 mg QD, including 73 PDAC, 18 NSCLC, 5 CRC and 2 others. Key eligibility criteria: • Pathologically confirmed, advanced solid tumors harboring RAS mutations using local testing results • Previously treated with standard therapies • ECOG PS 0~1 Phase Ia: Dose escalation Orally QD, 21 days/cycle 5 mg | 10 mg | 20 mg | 40 mg | 80 mg | 120mg | 160 mg | 200 mg | 320 mg Accelerated titration BOIN design with backfill RDE Phase Ib: Dose expansion NSCLC | PC | CRC | Other solid tumors Treatment until progressive disease, intolerable toxicity or other reasons to withdrawal Study Endpoints: - Safety and tolerability - Investigator-assessed ORR, DoR, DCR, PFS per RECIST v1.1 - Pharmacokinetics - Exploratory biomarkers Abbreviations: ECOG, Eastern Cooperative Oncology Group. PS, performance status. QD, once daily. BOIN, Bayesian optimal interval. RDE, recommended dose for expansion. NSCLC, non-small cell lung cancer. PC, pancreatic cancer. CRC, colorectal cancer. DCR, disease control rate. DoR, duration of response. NSCLC, non-small cell lung cancer. ORR, objective response rate. PFS, progression-free survival. PK, pharmacokinetics. RDE, recommended dose for expansion. RECIST 1.1, Response Evaluation Criteria in Solid Tumors 1.1. Yuxiang Ma, MD [Slide 2] Abstract number: LBA33 Demographics and Disease Characteristics PDAC (N=73) Age, median (range), years | 61.0 (20.0, 74.0) ≥60, n (%) | 43 (58.9) Male, n (%) | 36 (49.3) ECOG PS 0/1, n (%) | 24 (32.9) / 49 (67.1) Stage IV at study entry, n (%) | 71 (97.3) Baseline metastasis, n (%) Liver | 57 (78.1) Peritoneum | 23 (31.5) Lung | 12 (16.4) Bone | 6 (8.2) Number of prior lines of antitumor therapy, median (range) | 2 (0, 6+) 0/1/≥2, n (%) | 1 (1.4) / 19 (26.0) / 53 (72.6) Prior antitumor therapy, n (%) Chemotherapy | 72 (98.6) Gemcitabine-containing | 66 (90.4) Taxane-containing | 63 (78.8) Fluorouracil-containing | 57 (78.1) Irinotecan-containing | 52 (71.2) Platinum-containing | 41 (56.2) Immune checkpoint inhibitor | 24 (32.9) RAS inhibitor | 5 (6.8) PDAC (N=73) pie: KG12D 40% | KG12V 35% | KG12R 11% | RAS Others 14% NSCLC (N=18) pie: KG12D 34% | KG12V 22% | KG12C 11% | RAS others 33% NSCLC (N=18) Age, median (range), years | 61.5 (53.0, 72.0) ≥60, n (%) | 12 (66.7) Male, n (%) | 12 (66.7) ECOG PS 0/1, n (%) | 5 (27.8) / 13 (72.2) Stage IV at study entry, n (%) | 18 (100) Pathological type Adenocarcinoma/squamous carcinoma/other | 16 (88.9) / 1 (5.6) / 1 (5.6) PD-L1 (TPS) High/intermediate/ negative/missing | 4 (22.2) / 2 (11.1) / 1 (5.6) / 11 (61.1) Baseline metastasis, n (%) Lung | 14 (77.8) Bone | 13 (72.2) Brain | 7 (38.9) Liver | 4 (22.2) Number of prior lines of antitumor therapy, median (range) | 2 (0, 6+) 0/1/≥2, n (%) | 1 (5.6) / 6 (33.3) / 11 (61.1) Prior antitumor therapy, n (%) Chemotherapy | 17 (94.4) Immune checkpoint inhibitor | 15 (83.3) RAS inhibitor | 2 (11.1) Yuxiang Ma, MD [Slide 3] Abstract number: LBA33 Objective Response • Partial responses were observed starting from the 40 mg dose cohort. • 61 PDAC and 14 NSCLC received 40~160 mg: - ORR was 41.0% (25/61, 90% CI: 30.3-52.3) in PDAC, and 45.5% (5/11, 90%CI: 20.0-72.9) in NSCLC. - In RAS inhibitor naïve patients, ORR was 43.1% (25/58) in PDAC, and 55.6% (5/9) in NSCLC. PDAC n (%) | 40 mg (N=18) | 80 mg (N=16) | 120 mg (N=13) | 160 mg (N=14) | 40-160 mg (N=61) ORR | 8 (44.4) | 5 (31.3) | 7 (53.8) | 5 (35.7) | 25 (41.0) DCR | 18 (100) | 14 (87.5) | 12 (92.3) | 12 (85.7) | 56 (91.8) NSCLC n (%) | 80 mg (N=2) | 120 mg (N=5) | 160 mg (N=4) | 80-160 mg (N=11) ORR | 2 (100) | 1 (20.0) | 2 (50.0) | 5 (45.5) DCR | 2 (100) | 3 (60.0) | 3 (75.0) | 8 (72.7) TRT legend: 10 mg | 20 mg | 40 mg | 80 mg | 120 mg | 160 mg | 200 mg | 320 mg Prior RASi: KRAS G12D / KRAS G12C cPR = confirmed PR Data cutoff date: 15 September 2026 ORR and DCR were calculated in the evaluable population, defined as patients with target lesions at baseline and at least one post-dose evaluable tumor assessment. Among the 14 NSCLC, 3 were not evaluable: 1 at 40 mg had no baseline target lesions; 1 at 120 mg had no evaluable post-dose tumor assessment but was still on treatment; 1 at 160 mg dropped out without post-dose tumor assessment due to G3 stomatitis. Yuxiang Ma, MD [Slide 4] Abstract number: LBA33 Safety Profile • No prophylactic medications were used in the Phase Ia part. • GFH276 is well tolerated in dose range 40-160mg QD without DLTs observed. • The most common TRAEs were cutaneous, hematological, mucosal, and gastrointestinal, most being grade 1 or 2. • G3 TRAEs occurred in 27.5%, most commonly rash (7.5%), followed by anemia, diarrhea, platelet count decreased, and WBC count decreased (3.8% each). • No G4 or G5 TRAEs were observed. n (%) | 40-160 mg (N=80)* Any | G3 TRAE | 79 (98.8) | 22 (27.5) TRSAE | 9 (11.3) | 8 (10.0) TRAE leading to death | 0 | 0 TRAE leading to dose modifications Leading to interruption | 18 (22.5) | 11 (13.8) Leading to dose reduction | 14 (17.5) | 7 (13.3) Leading to discontinuation | 1 (1.3) | 1 (1.3) Common AEs in 40-160 mg (TRAE | TEAE, any grade %): Rash* | 85% | 84% Anemia | 66% | 64% Stomatitis ⊕ | 55% | 55% Hypoalbuminemia | 55% | 45% Diarrhea | 49% | 49% Platelet count decreased | 38% | 38% Aspartate aminotransferase increased | 32% | 29% Alanine aminotransferase increased | 26% | 24% White blood cell count decreased | 24% | 22% Pyrexia | 25% | 16% Hyponatremia | 20% | 18% Paronychia | 19% | 18% Neutrophil count decreased | 19% | 16% Data cutoff date: 15 July 2026. * Including 61 PDAC, 14 NSCLC, 3 CRC, and 2 others. Median follow-up time (range) was 8.7 (4.4, 31.1) weeks. Median duration of exposure (range) was 8.4 (1.4, 31.0) weeks. Abbreviations: TRAE, treatment-related adverse event. TRSAE, treatment-related serious adverse event. WBC, white blood cell. Common AEs were presented by TRAE ≥15%. No G4 or G5 TRAEs were observed. * Pooled PTs: Rash, dermatitis acneiform, rash maculopapular, erythema nodosum, dermatitis exfoliative generalized, rash pustular, drug eruption, genital rash. Multiple types of rash may have occurred in the same patient. ⊕ Pooled PT: Stomatitis, and mouth ulceration. Yuxiang Ma, MD

[Slide 1] Small Cell Lung Cancer • Multiple randomized phase III trials have reinforced these results Trial | Agent | mOS | OS HR | FDA approval | EMA approval IMpower 133 | Atezolizumab | 12.3m | 0.70 | March 18, 2019 | September 6, 2019 CASPIAN | Durvalumab | 13.0m | 0.73 | March 27, 2020 | September 1, 2020 ASTRUM-005 | Serplulimab | 15.4m | 0.63 | N/A | February 3, 2025 • Long-term survival possible for a small subset of patients [Inset KM curve, small print:] | Atezo + CP/ET (n=201) | Placebo + CP/ET (n=202) OS events, n (%) | 159 (79.1) | 169 (83.7) Median OS (95% CI), months | 12.3 (10.8, 15.8) | 10.3 (9.3, 11.3) Atezo + CP/ET median follow-up: 59.4 months (range, 0.0-72.6) Placebo + CP/ET median follow-up: 26.4 months (range, 0.0-34.4) 5-year OS, 12% [Number-at-risk rows too small to read reliably — [illegible]] OS rate (95% CI), % | IMpower133 and IMbrella Aᵃ Atezolizumab + CE (n=201)ᵇ | IMpower133 only Placebo + CE (n=202) 1 yearᵇ,⁶ | 51.9 (44.6, 59.1) | 39.0 (32.1, 45.9) 2 yearsᵇ,¹² | 21.6 (15.5, 27.6) | 16.1 (10.8, 21.3) 3 yearsᶜ,¹³ | 15.9 (10.5, 21.3) | NE 4 yearsᶜ,¹³ | 12.8 (7.7, 17.9) | NE 5 yearsᶜ,¹³ | 11.9 (6.8, 16.9) | NE 6 yearsᶜ,¹⁴ | 10.8 (5.8, 15.8) | NE 7 yearsᶜ | 10.8 (5.8, 15.8) | NE Stephen Liu, MD Horn, NEJM 2018; Paz-Ares, Lancet 2019; Cheng, JAMA 2022; Liu, WCLC 2023; Reck, ELCC 2026 [Slide 2] Conclusions • Adding tarlatamab to maintenance durvalumab significantly improves survival • Better safety profile and monitoring requirements in the maintenance setting • Global access to agents like tarlatamab will be critical Stephen Liu, MD

[Slide 2] Tarlatamab: DeLLphi-305: Study Design IMforte: -27% (A —— B annotation line above induction box) 1L Chemo-IO: Platinum-etoposide + durvalumab(a) induction therapy (4 cycles) Key Inclusion Criteria: • Age ≥ 18 years • No disease progression • ECOG PS 0–1 • Asymptomatic untreated or treated brain metastasis Stratification Factors: • Brain and/or liver metastasis (Yes/No) • Response to 1L chemo-IO (PR/CR vs SD) • Prior platinum type (cisplatin vs carboplatin) R 1:1, N = 563 Treatment until progression or unacceptable toxicity Tarlatamab (10 mg Q2W)b + Durvalumab (1500 mg Q4W)c, n = 282 Durvalumab (1500 mg Q4W)c, n = 281 Primary Endpoint: OS Key Secondary Endpoint: PFS Secondary Endpoints: ORR, DOR, safety Interim analysis for OS: 243 (61% IF) OS events, 0.008 two-sided P-value boundary Endpoints were tested sequentially according to a prespecified hierarchical testing strategy: OS → PFS → ORR Sands, ESMO 2026 Stephen Liu, MD [Slide 3] EnLiST ESMO ADAPTATION OF LINES OF SYSTEMIC THERAPY Annals of Oncology — SPECIAL ARTICLE — K.S. Saimi, M. Koopman et al., 2026 ESMO adaptation of Lines of Systemic Therapy (EnLiST): a consensus framework for standardising the designation of lines of therapy in solid tumours EnLiST (ESMO adaptation of Lines of Systemic Therapy) is a consensus framework developed to standardize how lines of systemic anticancer therapy (LoT) are assigned in solid tumors. Importantly, a change in treatment does not necessarily mean a new line of therapy. EnLiST distinguishes between three situations: • Same LoT: Treatment is continued or simplified (e.g., chemotherapy is stopped and maintenance immunotherapy continues), without disease progression. • Modified LoT: A new anticancer agent is added to an ongoing regimen without disease progression or inadequate response. • New LoT: Treatment is changed because of clinical disease progression or an inadequate response to the previous regimen. EnLiST: Assigning Lines of Therapy (LoT) Change in systemic anticancer therapy (SACT) can result in New LoT or Modified LoT or Same LoT Reason for Change of SACT New LoT: Clinical progression of disease (cPD); OR Lack of adequate tumour response Modified LoT: Intolerability; OR Adding anticancer agent(s) to ongoing SACT (A→A + B) e.g. after performance status improves; OR Miscellaneous (Financial, logistic, drug (non)availability, delayed availability of biomarker results, etc.) Same LoT: Prospectively planned; OR Change in dose, schedule, route; OR Dropping anticancer agent(s) from ongoing SACT (A+B→A); OR Re-introduced after 'drug holiday' LoT is recorded separately for three settings: eLoT: Early-stage disease setting; aLoT: Advanced-stage disease setting; iLoT: Assigned to SACT comprised of only investigational agents, irrespective of disease setting eLoT, aLoT, and iLoT are each recorded separately as X.Y (i.e., two numerals separated by a decimal point) New LoT: Assigned the next sequential value to the left of the decimal point; E.g. following eLoT 1.0, next New eLoT will be eLoT 2.0; X refers to total number of New LoTs received by the patient to date Modified LoT: Assigned the next sequential value to the right of the decimal point; E.g. following aLoT 1.0, next Modified aLoT will be aLoT 1.1; Y refers to total number of Modified LoTs received by the patient since the most recent New LoT EnLiST Framework is applicable to routine patient care and clinical trials Reference: ESMO adaptation of Lines of Systemic Therapy (EnLiST), Annals of Oncology, 2026. [Slide 4] Other Questions • Should we be starting even earlier? • Integration into induction could increase impact but does it impact efficacy/safety? • Limited stage consolidation? • Would other induction strategies be more effective? • Antibody drug conjugates can induce very deep responses • Better partners for T-cell engagers? • Role for bispecific antibodies to replace checkpoint inhibitors? • Optimal dosing schedule and duration of therapy? • Patient selection and biomarkers Stephen Liu, MD

[Slide 1] Tarlatamab: DeLLphi-305: Efficacy Stats box: | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 281) Median OS, months (95% CI) | 21.1 (20.8–NE) | 12.7 (11.2–14.8) Median FU, months | 15.9 | 16.1 HR (95% CI); P-value (2-sided)a | 0.42 (0.32–0.54); P < 0.001 Curve labels: 92% | 79% Tarlatamab + Durvalumab | 75% | 53% Durvalumab Number of patients at risk: Tarlatamab + Durvalumab 282 276 258 246 222 132 55 9 0 Durvalumab 281 252 208 171 146 81 33 8 1 0 | DeLLphi-305 Tarla/Durva vs Durva | IMforte Lurbi/Atezo vs Atezo Median OS | 21.1 vs 12.7m | 13.2 vs 10.6m 1y OS rate | 79% vs 53% | 56% vs 44% OS HR | 0.42 | 0.73 Median PFS | 7.4 vs 1.9m | 5.4 vs 2.1m 1y PFS rate | 38% vs 10% | 21% vs 12% PFS HR | 0.34 | 0.54 • Real signs of durable benefit • With 16m median follow up • 79% of patients alive at 1 year • 44% of patients remain on tarlatamab • Is tarlatamab changing SCLC biology? • Treatment beyond progression permitted • Local therapy could be employed to address oligoprogression • CNS efficacy? • QoL, PROs? • Long-term, landmark survival? Sands, ESMO 2026 Stephen Liu, MD Survival table Endpoint: Overall survival Arm | N | Median | Median CI Tarlatamab + Durvalumab | 282 | 21.1 | 20.8–NE Durvalumab | 281 | 12.7 | 11.2–14.8 HR: 0.42 HR CI: 0.32–0.54 P: P < 0.001 (2-sided) Landmark rates: 6 months: Tarlatamab + Durvalumab 92% | Durvalumab 75% 12 months: Tarlatamab + Durvalumab 79% | Durvalumab 53% [Slide 3] Tarlatamab: DeLLphi-305: Delivery • Initial tarlatamab package insert required 22-24h observation C1D1 and C1D8 • Significant barrier to access; many sites lack experience and infrastructure • Updated tarlatamab package insert reduces observation to 6-8 hours • DeLLphi-305 • Tarlatamab given C1D1, C1D8, C1D15 and then Q2W • Premedication with dexamethasone 8mg IV C1D1 and C1D8 only • Hydration with 1L NS after tarlatamab C1D1 and C1D8 • All treatment delivered as an outpatient • Post-infusion monitoring for 1-2 hours (6-8 hours in EU) Sands, ESMO 2026 and author communication Stephen Liu, MD

[Slide 1] 16:30 - 18:15 Presidential Symposium I CHAIRS: FABRICE ANDRÉ, JAMES LARKIN Attrition and the Rationale for Maintenance Strategies - Despite high rates of relapse, most patients do not receive any second-line therapy - Prospective LALUCA registry - 203 patients with ES-SCLC - 37% received second-line therapy - Maintenance strategies appealing in SCLC - Expect deepening of RR & prolongation of PFS at the cost of added toxicity - Relevant endpoint is survival Trial | Maintenance Agent | Overall Survival E7593 Schiller, JCO 2001 | Topotecan | OS 8.9 vs 9.3m (ns) NCIC BR.20 Arnold, JCO 2007 | Vandetanib | OS HR 1.43 (ns) CALGB 30504 Ready, JCO 2015 | Sunitinib | OS HR 1.28 (ns) CheckMate 451 Owonikoko, JCO 2021 | Nivolumab / Nivolumab + Ipilimumab | OS HR 0.84 (ns) / OS HR 0.92 (ns) STOMP Woll, Lung Cancer 2022 | Olaparib bid / Olaparib tid | OS HR 0.85 (ns) / OS HR 1.03 (ns) Moser, Lung Cancer 2026 [Sankey chart of treatment lines 1-3 by regimen group: labels illegible] Stephen Liu, MD Stephen Liu - Invited Discussant LBA3 ALICANTE AUDITORIUM - HALL 6 [Slide 2] 16:30 - 18:15 Presidential Symposium I CHAIRS: FABRICE ANDRÉ, JAMES LARKIN Tarlatamab: Predictive Biomarkers? - Positive or negative predictive biomarkers could help guide strategies - DLL3 expression has potential in this setting - DeLLphi-304: 2L tarlatamab vs chemotherapy DLL3 IHC | Tarlatamab OS | Chemo OS | OS HR < 25% | 7.8 months | 7.9 months | 0.954 ≥ 25% | 17.1 months | 9.6 months | 0.505 - DLL3 expression on CTCs being explored - DLL3 high: RR 75% - DLL3 low: RR 11% [Inset DeLLphi-304 OS KM curve and waterfall plot: small text illegible] Rudin, ASCO 2025; EMA EPAR; Mishra, Cancer Disc 2026 Stephen Liu, MD Stephen Liu - Invited Discussant LBA3 ALICANTE AUDITORIUM - HALL 6

[Slide 2] Conclusions - Tarlatamab plus durvalumab significantly improved OS, PFS, and ORR versus durvalumab in patients with ES-SCLC without progression after initial chemo plus durvalumab - 58% reduction in the risk of death; P < 0.001 - 66% reduction in the risk of disease progression or death; P < 0.001 - ORR: 26.4% vs 4.1%; P < 0.001 - Survival benefit extended to all pre-defined subgroups including patients with brain metastases, liver metastases, or minimal response to initial chemo plus durvalumab - Tarlatamab plus durvalumab demonstrated a manageable safety profile with low rates of treatment discontinuation due to adverse events DeLLphi-305 outcomes support tarlatamab plus durvalumab as a new standard of care first-line maintenance therapy in ES-SCLC Chemo, chemotherapy; CRS, cytokine release syndrome; DOR, duration of response; ES-SCLC, extensive-stage small cell lung cancer; ICANS, immune effector cell-associated neurotoxicity syndrome; ORR, objective response rate; OS, overall survival; PFS, progression-free survival. Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study

Tarlatamab: DeLLphi-305: Efficacy | DeLLphi-305 Tarla/Durva vs Durva | IMforte Lurbi/Atezo vs Atezo Median OS | 21.1 vs 12.7m | 13.2 vs 10.6m 1y OS rate | 79% vs 53% | 56% vs 44% OS HR | 0.42 | 0.73 Median PFS | 7.4 vs 1.9m | 5.4 vs 2.1m 1y PFS rate | 38% vs 10% | 21% vs 12% PFS HR | 0.34 | 0.54 • Real signs of durable benefit • With 16m median follow up • 79% of patients alive at 1 year • 44% of patients remain on tarlatamab • Is tarlatamab changing SCLC biology? • Treatment beyond progression permitted • Local therapy could be employed to address oligoprogression Sands, ESMO 2026 Stephen Liu, MD [Embedded KM curve thumbnail: legend and number-at-risk text illegible]

The Changing Face of Small Cell Lung Cancer • Small cell lung cancer being seen more in patients with no smoking history • Biologically distinct subtypes • Typical TP53/Rb1 loss • Atypical TP53/Rb1 intact • Chromothripsis of ch11/12 • Extrachromosomal amplification • CCND1 • CCND2/CDK4/MDM2 Cohort Selection: SCLC with tNGS (n=600) never/ light (<10 py) smoker RB1+/TP53− smoker SCLC | 6% RB1−/TP53+ smoker SCLC | <1% RB1+/TP53+ "atypical SCLC" (aSCLC) | 3% RB1−/TP53− never-smoker SCLC (nsSCLC) | 3% RB1−/TP53− smoker SCLC | ~88% Rekhtman, Cancer Disc 2024 Stephen Liu, MD Stephen Liu, Invited Discussant LBA3

Tarlatamab plus durvalumab significantly improved PFS versus durvalumab DeLLphi 305 | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 281) Median PFS, months (95% CI) | 7.4 (5.6–9.1) | 1.9 (1.9–2.0) Median FU, months | 13.9 | 13.9 HR (95% CI); P-value (2-sided)a | 0.34 (0.28–0.42); P < 0.001 Curve labels: 53% (Tarlatamab + Durvalumab, 6 mo) | 17% (Durvalumab, 6 mo) | 38% (Tarlatamab + Durvalumab, 12 mo) | 10% (Durvalumab, 12 mo) Percentage of Patients; Months 0 3 6 9 12 15 18 21 24 Number of patients at risk: Tarlatamab + Durvalumab | 282 | 225 | 147 | 122 | 86 | 32 | 14 | 1 | 0 Durvalumab | 281 | 106 | 46 | 36 | 18 | 8 | 2 | 0 Data cut-off date: 7 JUL 2026. At the time of data cut-off, 182 events had occurred in the tarlatamab-durvalumab arm, and 250 events had occurred in the durvalumab arm. aP-value was calculated using a stratified log-rank test. CI, confidence interval; FU, follow-up; HR, hazard ratio; PFS, progression-free survival. Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study [Side panel] Jacob Sands | Tarlatamab plus durvalumab (durva) vs durva as 1L maintenance (1LM) therapy following durva, platinum, and etoposide (1L durva + EP) in extensive-stage small cell lung cancer (ES-SCLC): Primary analysis of the phase 3 DeLLphi-305 trial Survival table Endpoint: Progression-free survival (PFS) Arm | N | Median | Median CI Tarlatamab + Durvalumab | 282 | 7.4 months | 5.6–9.1 Durvalumab | 281 | 1.9 months | 1.9–2.0 HR: 0.34 HR CI: 0.28–0.42 P: P < 0.001 (2-sided, stratified log-rank) Landmark rates: 6 months: Tarlatamab + Durvalumab 53% | Durvalumab 17% 12 months: Tarlatamab + Durvalumab 38% | Durvalumab 10%

[Slide 3] Tarlatamab demonstrated durable responses compared with chemotherapy | | Tarlatamab (n = 90)a | Chemotherapy (n = 52)a | Median DOR (95% CI), months | 7.1 (4.5–12.4) | 5.5 (4.2–5.7) Labels on curve: 42% / 10% at 12 months Proportion of responders without disease progression or death / Months Number of patients at risk: Tarlatamab 90 72 50 41 37 33 22 14 4 0 Chemotherapy 52 40 15 6 5 4 2 1 0 Outcomes (%) | Tarlatamab (n = 254) | Chemotherapy (n = 255) ORR (95% CI) | 35 (30–42) | 20 (16–26) KM estimate of DOR (95% CI) at 6 months | 56 (45–66) | 30 (18–43) KM estimate of DOR (95% CI) at 12 months | 42 (31–52) | 10 (4–20) Sustained DC (≥ 52 weeks) | 24 | 4 mFU for tarlatamab: 20.7 months (95% CI, 18.2–23.5) mFU for chemotherapy: 19.2 months (95% CI, 19.2–NE) Data cutoff date: 18 Feb 2026. • Tarlatamab demonstrated longer responses, with a greater proportion of ongoing responses at 12 months • Approximately six-fold more patients achieved sustained DC (≥ 52 weeks) with tarlatamab vs chemotherapy aNumber of confirmed responders as assessed by investigator (ITT analysis). CI, confidence interval; DC, disease control; DOR, duration of response; ITT, intention-to-treat; KM, Kaplan-Meier; mFU, median follow-up; NE, not estimable; ORR, objective response rate. Survival table Endpoint: Duration of response (confirmed responders, investigator-assessed) Arm | N | Median | Median CI Tarlatamab | 90 | 7.1 months | 4.5–12.4 Chemotherapy | 52 | 5.5 months | 4.2–5.7 KM DOR at 6 months: 56% (45–66) vs 30% (18–43) KM DOR at 12 months: 42% (31–52) vs 10% (4–20) [Slide 4] Tarlatamab demonstrated no new or increase in toxicity with longer-term treatment TRAEs reported by time period for patients on treatment Tarlatamab (< 3 months / 3–12 months / ≥ 12 months), Patients (%) CRS 56.3 / 0 / 1.6 Decreased appetite 21.8 / 10.7 / 3.2 Dysgeusia 21.4 / 6.0 / 4.8 Pyrexia 20.6 / 0 / 0 Fatigue 18.3 / 6.0 / 3.2 Nausea 16.3 / 2.7 / 4.8 On treatment < 3 months: Tarlatamab (n = 252) On treatment 3–12 months: Tarlatamab (n = 149) On treatment ≥ 12 months: Tarlatamab (n = 62) Chemotherapy (< 3 months / 3–12 months / ≥ 12 months), Patients (%) Anemia 56.1 / 45.9 / 40.0 Neutropenia 40.6 / 30.3 / 60.0 Thrombocytopenia 34.4 / 28.4 / 20.0 Leukopenia 29.1 / 22.9 / 60.0 Nausea 25.4 / 8.3 / 10.0 Fatigue 23.0 / 15.6 / 10.0 Decreased appetite 17.2 / 5.5 / 10.0 Lymphopenia 7.8 / 7.3 / 30.0 On treatment < 3 months: Chemotherapy (n = 244) On treatment 3–12 months: Chemotherapy (n = 109) On treatment ≥ 12 months: Chemotherapy (n = 10) Data cutoff date: 18 Feb 2026. • The differing temporal patterns of TRAEs support the favorable long-term tolerability of tarlatamab relative to chemotherapy • No new safety signals were identified with extended follow-up TRAEs with incidence of ≥ 15% by treatment period are presented. CRS, cytokine release syndrome; TRAE, treatment-related adverse event.

[Slide 1] 16:30 - 18:15 Presidential Symposium I CHAIRS: FABRICE ANDRÉ, JAMES LARKIN Tarlatamab: DeLLphi-305: Efficacy [Thumbnail of DeLLphi-305 OS KM curve: 21.1 (20.8–NE) vs 12.7 (11.2–14.8); HR 0.42 (0.32–0.54); P < 0.001; 92%/75% at 6 mo, 79%/53% at 12 mo] | DeLLphi-305 (Tarla/Durva vs Durva) | IMforte (Lurbi/Atezo vs Atezo) Median OS | 21.1 vs 12.7m | 13.2 vs 10.6m 1y OS rate | 79% vs 53% | 56% vs 44% OS HR | 0.42 | 0.73 Median PFS | 7.4 vs 1.9m | 5.4 vs 2.1m 1y PFS rate | 38% vs 10% | 21% vs 12% PFS HR | 0.34 | 0.54 • Real signs of durable benefit • With 16m median follow up • 79% of patients alive at 1 year • 44% of patients remain on tarlatamab • Is tarlatamab changing SCLC biology? • Treatment beyond progression permitted • Local therapy could be employed to address oligoprogression • CNS efficacy? • QoL, PROs? • Long-term, landmark survival? Sands, ESMO 2026 Stephen Liu, MD Stephen Liu — Invited Discussant LBA3 ALICANTE AUDITORIUM - HALL 6 [Slide 2] 16:30 - 18:15 Presidential Symposium I CHAIRS: FABRICE ANDRÉ, JAMES LARKIN Tarlatamab: DeLLphi-304 • Cytokine release syndrome (CRS) • Immune effector cell neurotoxicity syndrome (ICANS) • Initial recommendation for 22-24h observation on cycle 1, days 1 and 8 [Bar chart] Percentage of patients, % | CRS | ICANS Overall* | 56 | 6 Grade 1 | 42 | 3 Grade 2 | 13 | 3 Grade 3 | 1 | 0 Serious | 17 | 4 Discontinuations | 0.4 | 0.4 Fatal | 0 | 0.4 Table 1. Recommended Dosage and Schedule of IMDELLTRA Dosing Schedule | Day | Dose of IMDELLTRA | Administration Instructions | Recommended Monitoring Step-up Dosing Schedule Cycle 1 | Day 1a | Step-up dosea 1 mg | Administer IMDELLTRA as a 1-hour intravenous infusion in an appropriate healthcare setting. | Monitor patients from the start of the IMDELLTRA infusion for 22 to 24 hours on Cycle 1 Day 1 and Cycle 1 Day 8 in an appropriate healthcare setting. Recommend that patients remain within 1-hour of an appropriate healthcare setting for a total of 48 hours from start of the infusion with IMDELLTRA, accompanied by a caregiver. | Day 8a | 10 mga | | | Day 15 | 10 mg | | Observe patients for 6-8 hours post IMDELLTRA infusiona. Rudin, ASCO 2025; FDA.gov Stephen Liu, MD [Speaker panel] Stephen Liu | Invited Discussant LBA3 ALICANTE AUDITORIUM - HALL 6 [Slide 4] 16:30 - 18:15 Presidential Symposium I CHAIRS: FABRICE ANDRÉ, JAMES LARKIN Tarlatamab: DeLLphi-305: Safety CRSa,b, n (%) | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 278) Any grade | 140 (49.6) | 4 (1.4) Grade 2 | 21 (7.4) | 2 (0.7) Grade 3 | 1 (0.4) | 0 (0.0) Grade 4 | 0 (0.0) | 0 (0.0) Serious adverse events | 36 (12.8) | 2 (0.7) Leading to discontinuation of tarlatamab | 1 (0.4) | NA Fatal adverse events | 0 (0.0) | 0 (0.0) Utilization of any CRS interventionsc | 65 (23.0) | 2 (0.7) • CRS seen in 50% of patients • 84% of CRS cases were grade 1 • Likely under-reporting with less observation but did not translate into high grade CRS • CRS leading to discontinuation of tarlatamab in one patient (0.4%) • Due to less disease burden? Sands, ESMO 2026 Stephen Liu, MD [Speaker panel] Stephen Liu | Invited Discussant LBA3 ALICANTE AUDITORIUM - HALL 6

[Slide 3] T-CELL ENGAGERS TARLATAMAB: MOST COMMON AEs (DeLLphi-301) CRS 10 mg tarlatamab 51% of participants 100 mg tarlatamab 61% of participants Fever 10 mg tarlatamab 35% of participants 100 mg tarlatamab 33% of participants Anemia 10 mg tarlatamab 26% of participants 100 mg tarlatamab 25% of participants Decreased appetite 10 mg tarlatamab 29% of participants 100 mg tarlatamab 44% of participants Constipation 10 mg tarlatamab 27% of participants 100 mg tarlatamab 25% of participants 13% vs 29% Dose interruption or reduction for treatment-related AEs Tarlatamab 10 mg vs 100 mg 8% vs 28% ICANS and associated neurological events Tarlatamab 10 mg vs 100 mg ICANS and associated neurological events: broad definition; none severe at 10 mg, 5% severe at 100 mg; no deaths from ICANS; median onset 5 days, median duration 6.5 days. ICANS itself 5% (10 mg) and 10% (100 mg). 10 mg is the approved dose. Adapted from Ahn MJ, et al. N Engl J Med 2023;389:2063–75 LUNG SUMMIT · ESMO 2026 [Slide 4] CYTOKINE RELEASE SYNDROME RISK CRITERIA FOR CRS Clinical risk factors • Advanced age (>75 years) • Multiple comorbidities • Primary tumor size ≥7 cm • Diffuse or extensive liver metastases, or multiple coalescing hepatic lesions Laboratory risk factors • LDH ≥500 U/L • CRP ≥4 mg/dL • Ferritin ≥400 ng/mL • Uric acid ≥8.4 mg/dL Joshi U, et al. Clin Lung Cancer 2026;27:48–58 LUNG SUMMIT · ESMO 2026

[Slide 2] Baseline characteristics Characteristics | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 281) Age, median (range), years | 65.0 (35–87) | 65.0 (29–86) Male, % | 69.9 | 71.5 Race, White/Non-white, % | 57.8 / 42.2 | 64.4 / 35.6 Region, Americas/Asia-Pacific/Europe, % | 12.8 / 40.4 / 46.8 | 13.9 / 35.9 / 50.2 ECOG PS, 0/1, % | 45.7 / 54.3 | 43.1 / 56.9 Smoking history, Yes/Never, % | 90.4 / 9.2 | 90.8 / 8.5 Limited-Stage SCLC at initial diagnosis, % | 5.3 | 2.5 Type of platinum first received, cisplatin/carboplatin(a), % | 15.6 / 84.4 | 15.7 / 84.0 Response to initial therapy, CR/PR/SD, % | 3.2 / 85.8 / 11.0 | 3.6 / 84.7 / 11.7 Presence of current/prior brain metastasis, % | 19.5 | 19.9 Presence of current/prior liver metastasis, % | 31.2 | 33.8 - At baseline, key characteristics were well balanced between the treatment arms - Prior or current brain and/or liver metastases were observed in a meaningful proportion of patients (a)In durvalumab arm, one patient received carboplatin, but the dosage is not available. CR, complete response; ECOG PS, Eastern Cooperative Oncology Group performance status; SCLC, small cell lung cancer; PR, partial response; SCLC, extensive-stage small cell lung cancer; SD, stable disease. Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study [Slide 4] OS benefit was consistent across all key prespecified subgroups Subgroup | Tarlatamab + Durvalumab (no. of events / no. of patients) | Durvalumab (no. of events / no. of patients) | Hazard Ratio for Death (95% CI) Age < 65 years | 34 / 139 | 69 / 133 | 0.36 (0.24–0.54) ≥ 65 years | 53 / 143 | 87 / 148 | 0.48 (0.34–0.67) Sex Male | 64 / 197 | 120 / 201 | 0.38 (0.28–0.51) Female | 23 / 85 | 36 / 80 | 0.56 (0.33–0.94) Race White | 54 / 163 | 108 / 181 | 0.40 (0.29–0.56) Black | 0 / 4 | 0 / 4 | - Asian | 31 / 109 | 45 / 92 | 0.47 (0.30–0.74) Other | 2 / 6 | 3 / 4 | - Region Americas | 13 / 36 | 16 / 39 | 0.82 (0.39–1.71) Asia-Pacific | 32 / 114 | 51 / 101 | 0.43 (0.28–0.67) Europe | 42 / 132 | 89 / 141 | 0.36 (0.25–0.52) Presence (previous or current) of brain metastases Yes | 15 / 55 | 26 / 56 | 0.49 (0.26–0.93) No | 72 / 227 | 130 / 225 | 0.41 (0.30–0.54) Presence (previous or current) of liver metastases Yes | 43 / 88 | 69 / 95 | 0.41 (0.28–0.61) No | 44 / 194 | 87 / 186 | 0.40 (0.28–0.57) Response to initial therapy CR/PR | 77 / 251 | 135 / 248 | 0.42 (0.32–0.56) SD | 10 / 31 | 21 / 33 | 0.40 (0.19–0.85) Type of platinum first received for ES-SCLC Cisplatin | 11 / 44 | 23 / 44 | 0.33 (0.16–0.67) Carboplatin | 76 / 238 | 132 / 236 | 0.44 (0.34–0.59) X axis: 0.1 0.2 0.5 1 1.5 2 — Tarlatamab + Durvalumab Better ← → Durvalumab Better Data cut-off date: 7 JUL 2026. Median follow-up: 15.9 months (tarlatamab + durvalumab arm); 16.1 months (durvalumab arm). CI, confidence interval; ES-SCLC, extensive-stage small cell lung cancer; CR, complete response; OS, overall survival; PR, partial response; SD, stable disease. Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study

Tarlatamab plus durvalumab showed a manageable safety profile consistent with the individual agents DeLLphi 305 | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 278) Median duration of treatment, weeks (IQR) | 46.9 (16.4–65.6) | 8.1 (4.1–20.1) Any grade TEAEs, n (%) | 281 (99.6) | 258 (92.8) Grade ≥ 3, n (%) | 132 (46.8) | 113 (40.6) Serious, n (%) | 118 (41.8) | 88 (31.7) Fatal, n (%) | 13 (4.6) | 13 (4.7) Leading to discontinuation of tarlatamab, n (%) | 16 (5.7) | NA Leading to discontinuation of durvalumab, n (%) | 16 (5.7) | 16 (5.8) Any grade, TRAEsa, n (%) | 268 (95.0) | 145 (52.2) Grade ≥ 3, n (%) | 59 (20.9) | 30 (10.8) Serious, n (%) | 63 (22.3) | 26 (9.4) Fatal, n (%) | 0 (0.0) | 2 (0.7) Leading to discontinuation of tarlatamab, n (%) | 9 (3.2) | NA Leading to discontinuation of durvalumab, n (%) | 9 (3.2) | 10 (3.6) • The grade ≥ 3 TRAEs occurring in ≥ 1% of the patients included neutrophil count decreased, neutropenia, and decreased appetite in the tarlatamab plus durvalumab arm, and pneumonitis, AST increased and lipase increased in the durvalumab arm • Treatment with tarlatamab plus durvalumab resulted in a low frequency of treatment discontinuations and no treatment-related fatalities Data cut-off date: 7 JUL 2026 aRelated to any treatment. AST, aspartate aminotransferase; IQR, interquartile range; NA, not applicable; TEAE, treatment-emergent adverse events; TRAE, treatment-related adverse events. Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study 11

Tarlatamab: Predictive Biomarkers? • Positive or negative predictive biomarkers could help guide strategies • DLL3 expression has potential in this setting • DeLLphi-304: 2L tarlatamab vs chemotherapy [Small embedded DeLLphi-304 OS KM curve, tarlatamab vs chemotherapy — values too small to read] DLL3 IHC | Tarlatamab OS | Chemo OS | OS HR < 25% | 7.8 months | 7.9 months | 0.954 ≥ 25% | 17.1 months | 9.6 months | 0.505 • DLL3 expression on CTCs being explored • DLL3 high: RR 75% • DLL3 low: RR 11% [Small embedded waterfall plot with DLL3% row — values too small to read] Rudin, ASCO 2025; EMA EPAR; Mishra, Cancer Disc 2026 Stephen Liu, MD

Tarlatamab–durvalumab Durvalumab Percentage of Patients Months ORIGINAL ARTICLE MEETING OF THE EUROPEAN SOCIETY FOR MEDICAL ONCOLOGY Tarlatamab–Durvalumab in Small-Cell Lung Cancer M. Pérol and Others In a prespecified interim analysis of a phase 3 trial involving patients with extensive-stage SCLC, the median overall survival was 21.1 months with tarlatamab–durvalumab maintenance therapy and 12.7 months with durvalumab alone. Survival table Endpoint: Overall survival (per summary text; axis unlabelled) Tarlatamab–durvalumab | 21.1 Durvalumab | 12.7 6 mo: 91.5 (tarlatamab–durvalumab) vs 74.6 (durvalumab) 12 mo: 79.4 vs 53.3

[Slide 1] Tarlatamab: DeLLphi-305: Safety ICANSa, n (%) | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 27[?]) Any grade | 3 (1.1) | 0 (0.0) Grade 2 | 1 (0.4) | 0 (0.0) Grade 3 | 0 (0.0) | 0 (0.0) Grade 4 | 0 (0.0) | 0 (0.0) Serious adverse events | 0 (0.0) | 0 (0.0) Leading to discontinuation of tarlatamab | 0 (0.0) | NA Fatal adverse events | 0 (0.0) | 0 (0.0) • ICANS seen in 1% of patients • Only 3 cases (2 grade 1 and 1 grade 2 with no grade 3+) • Due to less disease burden? • Due to omitting hospitalization? Stephen Liu, MD Sands, ESMO 2026 [Slide 2] Tarlatamab: DeLLphi-305: Safety CRSa,b, n (%) | Tarlatamab + Durvalumab | Durvalumab Any grade | [illegible] | [illegible] Grade 2 | [illegible] | [illegible] Grade 3 | 1 (0.4) | [illegible] Grade 4 | 0 (0.0) | 0 (0.0) Serious adverse events | 36 (12.8) | 2 (0.7) Leading to discontinuation of tarlatamab | 1 (0.4) | NA Fatal adverse events | 0 (0.0) | 0 (0.0) Utilization of any CRS interventionsc | 65 (23.0) | 2 (0.7) • CRS seen in 50% of patients • 84% of CRS cases were grade 1 • Likely under-reporting with less observation but did not translate into high grade CRS • CRS leading to discontinuation of tarlatamab in one patient (0.4%) • Due to less disease burden? Sands, ESMO 2026

Bispecific T-Cell Engagers • Strong rationale that BiTEs perform better in low disease burden states • Greater effector molecule to target ratio • Greater tumor accessibility for T-cells • Less antigen mass may decrease CRS and ICANS • Less T-cell exhaustion from chronic antigen stimulation • Decreased heterogeneity and potentially fewer antigen-negative clone [Diagram: T cell – CD3 – DLL3 – SCLC cell] Goebeler, Nat Rev Clin Oncol 2020; Owonikoko, ASCO 2021 Stephen Liu, MD

[Slide 3] Table 3. Adverse Events (Safety Population).* Event | Tarlatamab–Durvalumab (N=282) | Durvalumab (N=278) number of patients (percent) Any adverse event | 281 (99.6) | 258 (92.8) Grade ≥3 adverse event | 132 (46.8) | 113 (40.6) Serious adverse event | 118 (41.8) | 88 (31.7) Fatal event | 13 (4.6) | 13 (4.7) Adverse event leading to interruption of tarlatamab | 98 (34.8) | NA Adverse event leading to discontinuation of tarlatamab | 16 (5.7) | NA Adverse event leading to interruption of durvalumab | 65 (23.0) | 35 (12.6) Adverse event leading to discontinuation of durvalumab | 16 (5.7) | 16 (5.8) Adverse event of any grade reported in ≥8% of the patients in either group† Cytokine release syndrome‡ | 140 (49.6) | 4 (1.4) Dysgeusia | 138 (48.9) | 7 (2.5) Decreased appetite | 101 (35.8) | 48 (17.3) Anemia | 86 (30.5) | 69 (24.8) Constipation | 83 (29.4) | 38 (13.7) Fatigue | 69 (24.5) | 34 (12.2) Nausea | 68 (24.1) | 39 (14.0) Weight decreased | 60 (21.3) | 8 (2.9) Headache | 58 (20.6) | 30 (10.8) Pruritus | 50 (17.7) | 20 (7.2) Neutrophil count decreased | 46 (16.3) | 24 (8.6) Asthenia | 46 (16.3) | 22 (7.9) White-cell count decreased | 42 (14.9) | 19 (6.8) Pyrexia | 42 (14.9) | 12 (4.3) Cough | 38 (13.5) | 31 (11.2) Rash | 38 (13.5) | 23 (8.3) Alanine aminotransferase level increased | 37 (13.1) | 29 (10.4) Hyponatremia | 36 (12.8) | 28 (10.1) Aspartate aminotransferase level increased | 35 (12.4) | 22 (7.9) Diarrhea | 34 (12.1) | 29 (10.4) Creatine kinase level increased | 32 (11.3) | 13 (4.7) Upper respiratory tract infection | 31 (11.0) | 12 (4.3) Vomiting | 29 (10.3) | 13 (4.7) Back pain | 28 (9.9) | 24 (8.6) Arthralgia | 28 (9.9) | 22 (7.9) Dizziness | 26 (9.2) | 24 (8.6) Insomnia | 25 (8.9) | 16 (5.8) Pneumonia | 23 (8.2) | 19 (6.8) Amylase level increased | 23 (8.2) | 14 (5.0) Neutropenia | 23 (8.2) | 8 (2.9) Hypothyroidism | 22 (7.8) | 26 (9.4) Dyspnea | 20 (7.1) | 26 (9.4)

[Slide 1] DeLLphi-305: Study design DeLLphi 305 A Phase 3, Open-label, Multicenter, Randomized Study of Tarlatamab in Combination With Durvalumab vs Durvalumab Alone in Patients With ES-SCLC Following Platinum, Etoposide and Durvalumab (NCT06211036) 1L Chemo-IO: Platinum-etoposide + durvalumab induction therapy (4 cycles) Key Inclusion Criteria • Age ≥ 18 years • No disease progression • ECOG PS 0–1 • Asymptomatic untreated or treated brain metastasis R 1:1 N = 563 Treatment until progression or unacceptable toxicity Tarlatamab (10 mg Q2W) + Durvalumab (1500 mg Q4W) n = 282 Durvalumab (1500 mg Q4W) n = 281 Primary Endpoint: OS Key Secondary Endpoint: PFS Secondary Endpoints: ORR, DOR, safety Stratification Factors • Brain and/or liver metastasis (Yes/No) • Response to 1L chemo-IO (PR/CR vs SD) • Prior platinum type (cisplatin vs carboplatin) Interim analysis for OS: 2[illegible]3 (61% IF) OS events, 0.00[illegible] two-sided P-value boundary Endpoints were tested sequentially according to a prespecified hierarchical testing strategy: OS → PFS → ORR [footnotes illegible] [footer, blurry:] Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study 4 [Slide 2] Tarlatamab plus durvalumab significantly improved overall survival versus durvalumab alone as first-line maintenance therapy DeLLphi 305 Tarlatamab + Durvalumab (n = 2[illegible]) | Durvalumab (n = 281) Median OS, months (95% CI): 21.1 (20.8–NE) | 12.7 (11.2–14.8) Median FU, months: [illegible] | [illegible] HR (95% CI); P-value (2-sided)a: 0.42 (0.32–0.5[illegible]); P < 0.001 Curve labels: Tarlatamab + Durvalumab 92% (6 mo), 79% (12 mo); Durvalumab 75% (6 mo), 53% (12 mo) Percentage of Patients / Months Number of patients at risk: Tarlatamab + Durvalumab: 282, 276, 258, 246, 222, 132, 55, 9, 0 Durvalumab: 281, 252, 208, 171, 146, 81, 33, 8, 1, 0 [footnotes illegible: data cut-off date, event counts] [footer, blurry:] Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study 6 Survival table Endpoint: OS Arm | N | Median | Median CI Tarlatamab + Durvalumab | 282 | 21.1 months | 20.8–NE Durvalumab | 281 | 12.7 months | 11.2–14.8 HR: 0.42 P: P < 0.001 (2-sided) Landmark rates: 6 mo: Tarlatamab + Durvalumab 92% | Durvalumab 75% 12 mo: Tarlatamab + Durvalumab 79% | Durvalumab 53% [Slide 3] Tarlatamab plus durvalumab significantly improved PFS versus durvalumab DeLLphi 305 Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 281) Median PFS, months (95% CI): 7.4 (5.6–9.1) | 1.9 (1.9–2.0) Median FU, months: 13.9 | 13.9 HR (95% CI); P-value (2-sided)a: 0.34 (0.28–0.42); P < 0.001 Curve labels: Tarlatamab + Durvalumab 53% (6 mo), 38% (12 mo); Durvalumab 17% (6 mo), 10% (12 mo) Number of patients at risk: Tarlatamab + Durvalumab: 282, 225, 147, 122, 86, 32, 14, 1, 0 Durvalumab: 281, 106, 46, 36, 18, 8, 2, 0 [footnotes illegible] [footer, blurry:] Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study 8 Survival table Endpoint: PFS Arm | N | Median | Median CI Tarlatamab + Durvalumab | 282 | 7.4 months | 5.6–9.1 Durvalumab | 281 | 1.9 months | 1.9–2.0 HR: 0.34 HR CI: 0.28–0.42 P: P < 0.001 (2-sided) Landmark rates: 6 mo: Tarlatamab + Durvalumab 53% | Durvalumab 17% 12 mo: Tarlatamab + Durvalumab 38% | Durvalumab 10%

Tarlatamab plus durvalumab significantly improved PFS versus durvalumab DeLLphi 305 Median PFS, months (95% CI) | Tarlatamab + Durvalumab (n = 282): 7.4 (5.6–9.1) | Durvalumab (n = 281): 1.9 (1.9–2.0) Median FU, months | 13.9 | 13.9 HR (95% CI); P-value (2-sided)a | 0.34 (0.28–0.42); P < 0.001 Landmark labels: 53% / 17% at 6 months; 38% / 10% at 12 months Number of patients at risk: Tarlatamab + Durvalumab | 282 | 225 | 147 | 122 | 86 | 32 | 14 | 1 | 0 Durvalumab | 281 | 106 | 46 | 36 | 18 | 8 | 2 | 0 Data cut-off date: 7 JUL 2026. At the time of data cut-off, [illegible small print] events had occurred in the tarlatamab-durvalumab arm, and [illegible] events had occurred in the durvalumab arm. Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study Survival table Endpoint: Progression-free survival Arm | N | Median | Median CI Tarlatamab + Durvalumab | 282 | 7.4 | 5.6–9.1 Durvalumab | 281 | 1.9 | 1.9–2.0 HR: 0.34 HR CI: 0.28–0.42 P: P < 0.001 (2-sided) Landmark rates: 6 months: Tarlatamab + Durvalumab 53% | Durvalumab 17% 12 months: Tarlatamab + Durvalumab 38% | Durvalumab 10%

Poster 3821P Sustained Survival With Extended Follow-Up of Tarlatamab Plus Anti–PD-L1 as First-Line Maintenance (1LM) After Chemoimmunotherapy for Extensive-Stage Small Cell Lung Cancer (ES-SCLC): DeLLphi-303 Phase 1b Study Martin Wermke, Myung-Ju Ahn, Mor Moskovitz, Michael Pogorzelski, Simon Häfliger, Seoyoon Cho, Edith Sonnenberg, Andy Itsara, Miranda Brown, Kelly G. Paulson BACKGROUND • First-line maintenance (1LM) regimens with programmed death-ligand 1 (PD-L1) inhibitor ± lurbinectedin have demonstrated limited survival benefit in ES-SCLC, with median overall survival (OS) of approximately 11–13 months from maintenance initiation • Tarlatamab, a DLL3-targeting BiTE® immunotherapy, has demonstrated superior OS versus chemotherapy in patients with previously treated ES-SCLC • In an interim analysis of phase 1b DeLLphi-303 study presented at WCLC 2025, at a median follow-up of 18.4 months, tarlatamab combined with a PD-L1 inhibitor as 1LM therapy demonstrated encouraging antitumor activity and manageable safety, with a median OS of 25.3 months • More recently, tarlatamab plus durvalumab demonstrated a clinically meaningful improvement in OS and progression-free survival (PFS) versus durvalumab alone in the 1LM setting in the phase 3 DeLLphi-305 study • Here, we report extended follow-up results from the DeLLphi-303 study, with a median follow-up of 30.8 months (updated data cutoff: 3 August 2026) METHODS Key Inclusion Criteria ✓ No disease progression following 4–6 cycles of 1L chemoimmunotherapy (chemo-IO)* ✓ ECOG PS 0 or 1 ✓ Treated, asymptomatic brain metastases *Patients who did not have access to PD-L1 inhibitor were eligible. Non-randomized; Switching to a different anti-PD-L1 than received in 1L chemo-IO was permitted 1L Maintenance: Tarlatamab (10 mg IV Q2W) + Atezolizumab (1680 mg IV Q4W) n = 48; Tarlatamab (10 mg IV Q2W) + Durvalumab (1500 mg IV Q4W) n = 40 Primary Objectives: • Safety, tolerability, and RP2D of tarlatamab in combination with anti-PD-L1 Secondary Objectives: • PFS, OS, ORR, DOR, and DCR • Tarlatamab pharmacokinetics RESULTS Baseline Demographics and Disease Characteristics (Tarlatamab + Atezolizumab (n = 48) | Tarlatamab + Durvalumab (n = 40) | Overall Tarlatamab + Anti-PD-L1 (N = 88)) Age, years, median (range) 64 (27–79) | 64 (37–85) | 64 (27–85) Male, % 67 | 58 | 63 Race (Asian / White / Other), % 19/71/10 | 20/70/10 | 19/71/10 ECOG PS (0 / 1), % 38/63 | 45/55 | 41/59 Smoking history (never / current / former), % 2/23/75 | 8/25/68 | 5/24/72 Median sum of lesion diameters, mm 29.2 | 37.0 | 33.0 Stage at initial diagnosis (limited / extensive / unknown), % 15/81/4 | 10/90/0 | 13/85/2 Prior 1L platinum-etoposide cycles (4 / 5 / 6), % 83/10/6 | 83/3/15 | 83/7/10 Prior anti-PD-L1, % 88 | 88 | 88 Prior radiotherapy, % 25 | 35 | 30 Brain / liver metastases, % 19/31 | 33/43 | 25/36 Median time from start of 1L chemo-IO to 1LM, months (range) 3.6 (2.9–5.8) | 3.5 (2.9–5.4) | 3.6 (2.9–5.8) Median OS of 37.9 Months With Tarlatamab + Anti–PD-L1 in 1LM Median follow-up was 30.8 months; survival follow-up remains ongoing mOS (95% CI), months: Tarlatamab + Atezolizumab 30.9 (20.5–NE) | Tarlatamab + Durvalumab 37.9 (20.1–NE) | Overall 37.9 (22.5–NE) mFU, months 30.0 | 34.3 | 30.8 OS curve annotations at 24 months: 61.7% (atezolizumab), 58.1% (overall), 54.3% (durvalumab); at 36 months: 54.3% (durvalumab), 50.1% (overall), 45.5% (atezolizumab) Number of Patients at Risk (0–45 months, every 3): Atezolizumab 48 46 43 37 32 28 26 19 17 15 13 6 3 2 1 0; Durvalumab 40 40 38 36 31 25 23 17 15 14 13 12 10 2 0; Overall 88 86 81 73 63 53 49 36 32 29 26 18 13 4 1 0 Median PFS of 5.6 Months With Tarlatamab + Anti–PD-L1 in 1LM mPFS (95% CI), months: 5.6 (2.0–14.9) | 5.3 (3.5–11.2) | 5.6 (3.5–9.3) mFU, months 22.4 | 33.9 | 29.5 PFS curve annotations at 24 months: 25.9% (atezolizumab), 22.6% (overall), 19.0% (durvalumab); at 36 months: NE (atezolizumab), 19.7% (overall), 15.2% (durvalumab) Number of Patients at Risk (0–39 months, every 3): Atezolizumab 48 26 19 15 14 12 8 7 5 5 4 3 0; Durvalumab 40 23 15 14 10 9 7 6 6 6 5 4 2 0; Overall 88 49 34 29 24 21 15 13 11 11 9 7 2 0 Durable Disease Control Observed With Tarlatamab + Anti–PD-L1 as 1LM Baseline reflects target lesion sum at the start of 1LM, following completion of 4–6 cycles of 1L chemo-IO. Changes from baseline do not include new lesions or non-target lesions. • Overall DCR: 59.1% (3 CR, 18 PR, 31 SD); median DODC: 14.6 months (95% CI: 8.5–16.8) • ORR from post-chemo-IO baseline was 24% (21/88; 3 CR, 18 PR); median DOR: 16.6 months (95% CI: 7.2–NE) Incidence of Tarlatamab-Related AEs Occurring in ≥ 10% of Patients by Time to Onset Among Patients on Tarlatamab + Anti-PD-L1 During Each Time Period (0 to <3 Months (n = 88) | 3 to <12 months (n = 63) | 12 to <24 months (n = 36) | 24 to <36 months (n = 16)), % CRS 56 | 5 | 3 | 0; Dysgeusia 35 | 22 | 0 | 6; Fatigue 26 | 13 | 0 | 13; Pyrexia 24 | 2 | 0 | 0; Decreased appetite 16 | 8 | 6 | 0; Nausea 15 | 2 | 6 | 6; Myalgia 14 | 3 | 0 | 0; Headache 13 | 0 | 0 | 0; Chills 11 | 3 | 0 | 0; Asthenia 10 | 6 | 0 | 0 *n refers to the number of patients on tarlatamab + anti-PD-L1 at start of time interval. Tarlatamab-Related Adverse Events With Onset ≥ 12 Months Among Patients Treated With Tarlatamab + Anti-PD-L1 for ≥12 Months Were Primarily Grade 1-2 (Overall Tarlatamab + Anti-PD-L1 (n = 36)) All-grade tarlatamab-related AEs with onset ≥ 12 months after treatment initiation 15 (41.7); Grade ≥ 3 3 (8.3); Serious 2 (5.6); Fatal 0 (0.0); Leading to dose interruption 2 (5.6); Leading to tarlatamab discontinuation 1 (2.8) • Two patients experienced tarlatamab-related AEs leading to dose interruption (G3 decreased appetite and G2 paresthesia) • One patient had G3 immune-mediated encephalopathy that led to tarlatamab discontinuation. Other G3 events included decreased appetite, gait disturbance, and syncope • No participant with G4 or fatal events CONCLUSIONS In this extended follow-up of DeLLphi-303, tarlatamab + anti–PD-L1 as 1LM therapy demonstrated encouraging long-term survival, durable disease control, and a manageable long-term safety profile in patients with ES-SCLC: • Sustained survival with tarlatamab + PD-L1 inhibitor as 1LM – At a median follow-up of 30.8 months, median OS was 37.9 months (95% CI, 22.5–NE) from the start of 1LM therapy. Further follow-up is ongoing • Durable disease control was observed – DCR was 59.1%, with median duration of disease control of 14.6 months (95% CI, 8.5–16.8) • Manageable long-term safety with no new safety concerns – Among the 36 patients treated with tarlatamab + anti–PD-L1 for ≥ 12 months, three (8.3%) experienced grade 3 tarlatamab-related AEs with onset after 12 months These extended follow-up data support the potential for durable survival benefit with tarlatamab + PD-L1 inhibitor as 1LM and complement the phase 3 DeLLphi-305 results (NCT06211036) Study Information: DeLLphi-303 was not designed as a comparative study between anti-PD-L1 agents and no randomization between treatment arms occurred. The study is funded by Amgen, Inc. (Study Sponsor); durvalumab drug is provided by AstraZeneca. CONTACT | Martin.Wermke@ukdd.de ESMO 23-27 October 2026, Madrid, Spain. © 2026 Amgen Inc. All rights reserved

Tarlatamab: DeLLphi-305: Survival | Tarlatamab + Durvalumab (n = 282) | Durvalumab (n = 281) Median OS, months (95% CI) | 21.1 (20.8–NE) | 12.7 (11.2–14.8) Median FU, months | 15.9 | 16.1 HR (95% CI); P-value (2-sided)a | 0.42 (0.32–0.54); P < 0.001 Curve labels: 92% / 79% (Tarlatamab + Durvalumab); 75% / 53% (Durvalumab) at dashed reference lines (~6 and ~12 months) Percentage of Patients; Months (0, 3, 6, 9, 12, 15, 18, 21, 24, 27) Number of patients at risk: Tarlatamab + Durvalumab | 282 | 276 | 258 | 246 | 222 | 132 | 55 | 9 | 0 Durvalumab | 281 | 252 | 208 | 171 | 146 | 81 | 33 | 8 | 1 | 0 Stephen Liu, MD Sands, ESMO 2026 Survival table Endpoint: Overall survival Arm | N | Median | Median CI Tarlatamab + Durvalumab | 282 | 21.1 months | 20.8–NE Durvalumab | 281 | 12.7 months | 11.2–14.8 HR: 0.42 HR CI: 0.32–0.54 P: P < 0.001 (2-sided) Landmark rates: 6 months (dashed line; inferred): Tarlatamab + Durvalumab 92% | Durvalumab 75% 12 months (dashed line; inferred): Tarlatamab + Durvalumab 79% | Durvalumab 53%

Table S4. Subsequent Anti-Cancer Treatment by Region | Tarlatamab + Durvalumab: USA (N = 16) | Japan (N = 28) | EU (N = 132) | Others (N = 106) | Durvalumab: USA (N = 22) | Japan (N = 21) | EU (N = 140) | Others (N = 95) Patients reporting subsequent anti-cancer treatment, n (%) | 8 (50.0) | 17 (60.7) | 35 (26.5) | 44 (41.5) | 16 (72.7) | 17 (81.0) | 85 (60.7) | 61 (64.2) All chemotherapy | 5 (31.3) | 15 (53.6) | 31 (23.5) | 40 (37.7) | 5 (22.7) | 14 (66.7) | 69 (49.3) | 47 (49.5) Anti DLL3 Therapy | 0 (0.0) | 3 (10.7) | 1 (0.8) | 2 (1.9) | 14 (63.6) | 8 (38.1) | 10 (7.1) | 8 (8.4) T cell engager | 0 (0.0) | 3 (10.7) | 0 (0.0) | 0 (0.0) | 14 (63.6) | 8 (38.1) | 7 (5.0) | 5 (5.3) ADC | 0 (0.0) | 0 (0.0) | 1 (0.8) | 1 (0.9) | 0 (0.0) | 0 (0.0) | 1 (0.7) | 0 (0.0) Unspecified | 0 (0.0) | 0 (0.0) | 0 (0.0) | 1 (0.9) | 0 (0.0) | 0 (0.0) | 2 (1.4) | 3 (3.2) Radiation | 2 (12.5) | 3 (10.7) | 4 (3.0) | 8 (7.5) | 1 (4.5) | 2 (9.5) | 7 (5.0) | 10 (10.5) Other | 1 (6.3) | 2 (7.1) | 0 (0.0) | 1 (0.9) | 0 (0.0) | 2 (9.5) | 0 (0.0) | 3 (3.2) ADC other than DLL3 | 0 (0.0) | 0 (0.0) | 2 (1.5) | 5 (4.7) | 0 (0.0) | 2 (9.5) | 3 (2.1) | 16 (16.8) PD-(L)1 Checkpoint Inhibitor | 1 (6.3) | 1 (3.6) | 4 (3.0) | 6 (5.7) | 2 (9.1) | 0 (0.0) | 5 (3.6) | 13 (13.7) Tyrosine Kinase Inhibitor | 0 (0.0) | 0 (0.0) | 0 (0.0) | 7 (6.6) | 0 (0.0) | 0 (0.0) | 1 (0.7) | 20 (21.1) The subsequent anti-cancer therapy was based on pre-defined list of categories and preferred terms. Some of the treatments were given in combination but listed separately. Concomitant medications were coded using the WHO Drug Dictionary version GLOBALB3Mar26. ADC, antibody-drug conjugate; DLL3, Delta-like ligand 3; EU, Europe; N, number of patients in the analysis set; n, number of patients with observed data; PD-(L)1, programmed cell death-(ligand) 1; USA, United States of America; WHO, World Health Organization. Figure S3. Overall Survival Analysis in Prespecified Subgroups Subgroup | Tarlatamab + Durvalumab no. of events / no. of patients | Durvalumab | Hazard Ratio for Death (95% CI) Age < 65 years | 34 / 139 | 69 / 133 | 0.36 (0.24-0.54) Age ≥ 65 years | 53 / 143 | 87 / 148 | 0.48 (0.34-0.67) Sex Male | 64 / 197 | 120 / 201 | 0.38 (0.28-0.51) Sex Female | 23 / 85 | 36 / 80 | 0.56 (0.33-0.94) Race White | 54 / 163 | 108 / 181 | 0.40 (0.29-0.56) Race Black | 0 / 4 | 0 / 4 | - Race Asian | 31 / 109 | 45 / 92 | 0.47 (0.30-0.74) Race Other | 2 / 6 | 3 / 4 | - Region Americas | 13 / 36 | 16 / 39 | 0.82 (0.39-1.71) Region Asia-Pacific | 32 / 114 | 51 / 101 | 0.43 (0.28-0.67) Region Europe | 42 / 132 | 89 / 141 | 0.36 (0.25-0.52) Presence (previous or current) of brain metastases Yes | 15 / 55 | 26 / 56 | 0.49 (0.26-0.93) Presence (previous or current) of brain metastases No | 72 / 227 | 130 / 225 | 0.41 (0.30-0.54) Presence (previous or current) of liver metastases Yes | 43 / 88 | 69 / 95 | 0.41 (0.28-0.61) Presence (previous or current) of liver metastases No | 44 / 194 | 87 / 186 | 0.40 (0.28-0.57) Presence (previous or current) of brain metastases and/or liver metastases Yes | 52 / 131 | 84 / 134 | 0.45 (0.32-0.64) Presence (previous or current) of brain metastases and/or liver metastases No | 35 / 151 | 72 / 147 | 0.38 (0.25-0.57) Response to initial therapy CR/PR | 77 / 251 | 135 / 248 | 0.42 (0.32-0.56) Response to initial therapy SD | 10 / 31 | 21 / 33 | 0.40 (0.19-0.85) Type of platinum first received for ES-SCLC Cisplatin | 11 / 44 | 23 / 44 | 0.33 (0.16-0.67) Type of platinum first received for ES-SCLC Carboplatin | 76 / 238 | 132 / 236 | 0.44 (0.34-0.59) Number of durvalumab cycles prior to enrollment 3 | 13 / 56 | 27 / 51 | 0.36 (0.19-0.70) Number of durvalumab cycles prior to enrollment 4 | 74 / 226 | 129 / 230 | 0.44 (0.33-0.58) Received prophylactic cranial irradiation (PCI) Yes | 3 / 14 | 1 / 4 | - Received prophylactic cranial irradiation (PCI) No | 84 / 268 | 155 / 277 | 0.42 (0.32-0.55) Disease stage at diagnosis Stage IV | 87 / 269 | 151 / 269 | 0.44 (0.33-0.57) Disease stage at diagnosis Non-Stage IV | 0 / 13 | 5 / 12 | - Tarlatamab + Durvalumab Better | Durvalumab Better

[Slide 2] JOURNAL ARTICLE Tarlatamab Added to Durvalumab Improves OS and PFS in Extensive-Stage Small-Cell Lung Cancer Renée Ruthel For patients with extensive-stage small-cell lung cancer, adding tarlatamab to durvalumab during maintenance therapy significantly improves overall survival and progression-free survival, according to the industry-funded DeLLphi-305 trial. Results of the prespecified interim analysis were reported at the conference and published in the New England Journal of Medicine. [Slide 3] Our clinician-guide Dr. Narjust Florez offered her take: In extensive-stage small-cell lung cancer, progress has come in small steps, and many patients never receive second-line therapy, so moving tarlatamab into first-line maintenance makes clinical sense and means patients get the most active agent while still eligible to benefit. In this study, 80% of patients were alive one year after starting maintenance. This benefit was seen in patients with brain and liver metastases, which traditionally have a worse prognosis. Toxicity was manageable, with no serious neurologic events, and CRS was seen mainly after the first two doses.

Conclusions DeLLphi 305 • Tarlatamab plus durvalumab significantly improved OS, PFS, and ORR versus durvalumab in patients with ES-SCLC without progression after initial chemo plus durvalumab • 58% reduction in the risk of death; P < 0.001 • 66% reduction in the risk of disease progression or death; P < 0.001 • ORR: 26.4% vs 4.1%; P < 0.001 • Survival benefit extended to all pre-defined subgroups including patients with brain metastases, liver metastases, or minimal response to initial chemo plus durvalumab • Tarlatamab plus durvalumab demonstrated a manageable safety profile with low rates of treatment discontinuation due to adverse events DeLLphi-305 outcomes support tarlatamab plus durvalumab as a new standard of care first-line maintenance therapy in ES-SCLC Jacob Sands, MD | Tarlatamab plus durvalumab as 1L M therapy in ES-SCLC: primary analysis from DeLLphi-305 phase 3 study

[Slide 2] Attrition and the Rationale for Maintenance Strategies • Despite high rates of relapse, most patients do not receive any second-line therapy • Prospective LALUCA registry • 203 patients with ES-SCLC • 37% received second-line therapy • Maintenance strategies appealing in SCLC • Expect deepening of RR & prolongation of PFS at the cost of added toxicity • Relevant endpoint is survival [Sankey chart: N patients by treatment line 1–3; line 1: Chemoimmunotherapy (88.7%), Chemotherapy (11.3%); line 2: Anthracycline combination (57.3%), Chemo+/-immunotherapy (16.0%), Topotecan (26.7%); line 3: Anthracycline combination (32.1%), Chemo+/-immunotherapy (32.1%), Topotecan (28.6%), Other (7.2%)] Trial | Maintenance Agent | Overall Survival E7593 (Schiller, JCO 2001) | Topotecan | OS 8.9 vs 9.3m (ns) NCIC BR.20 (Arnold, JCO 2007) | Vandetanib | OS HR 1.43 (ns) CALGB 30504 (Ready, JCO 2015) | Sunitinib | OS HR 1.28 (ns) CheckMate 451 (Owonikoko, JCO 2021) | Nivolumab / Nivolumab + Ipilimumab | OS HR 0.84 (ns) / OS HR 0.92 (ns) STOMP (Woll, Lung Cancer 2022) | Olaparib bid / Olaparib tid | OS HR 0.85 (ns) / OS HR 1.03 (ns) Moser, Lung Cancer 2026 Stephen Liu, MD [Slide 3] Tarlatamab: Predictive Biomarkers? • Positive or negative predictive biomarkers could help guide strategies • DLL3 expression has potential in this setting • DeLLphi-304: 2L tarlatamab vs chemotherapy [DeLLphi-304 OS KM inset: Tarlatamab (n = 254) vs Chemotherapy (n = 255); Median OS, months 13.6 vs 8.3; HR (Tarlatamab/Chemotherapy) (95% CI) 0.60 (0.47, 0.77); p-value (2-sided) p < 0.001; labels 76% / 62% at 6 months, 53% / 37% at 12 months; Number at risk: Tarlatamab 254, 220, 192, 131, 60, 17, 0; Chemotherapy 255, 210, 156, 97, 42, 9, 2, 0] DLL3 IHC | Tarlatamab OS | Chemo OS | OS HR < 25% | 7.8 months | 7.9 months | 0.954 ≥ 25% | 17.1 months | 9.6 months | 0.505 • DLL3 expression on CTCs being explored • DLL3 high: RR 75% • DLL3 low: RR 11% [Waterfall plot: RECIST change from baseline by patient with DLL3 % row] Rudin, ASCO 2025; EMA EPAR; Mishra, Cancer Disc 2026 Stephen Liu, MD [Slide 4] Select T-Cell Engagers in Development for SCLC Trial | T-Cell Engager | Study Phase | Description DeLLphi-312 | Tarlatamab | Phase III | First-line carboplatin + etoposide + durvalumab +/- tarlatamab DeLLphi-306 | Tarlatamab | Phase III | Consolidation tarlatamab vs placebo post chemoradiation DeLLphi-315 | Tarlatamab | Phase III | Second-line subcutaneous vs IV DAREON-Lung-1 | Obrixtamig | Phase III | First-line carboplatin + etoposide + atezolizumab +/- obrixtamig DAREON-Rosetta35 | Obrixtamig | Phase Ib/II | Obrixtamig + pumitamig +/- carboplatin + etoposide MK-6070-003 | Gocatamig | Phase Ib/II | Gocatamig + I-DXd +/- atezolizumab vs chemo-IO MK-6070-002 | Gocatamig | Phase Ib/II | Gocatamig, gocatamig + I-DXd, gocatamig + durvalumab Stephen Liu, MD

[Slide 1] Percentage of Patients vs Months Tarlatamab–durvalumab | Durvalumab Curve labels: 91.5 and 74.6 at 6 months; 79.4 and 53.3 at 12 months No. of Patients | Median Overall Survival (95% CI) mo Tarlatamab–Durvalumab | 282 | 21.1 (20.8–NE) Durvalumab | 281 | 12.7 (11.2–14.8) Hazard ratio for death, 0.42 (95% CI, 0.32–0.54) P<0.001 No. at Risk (0,3,6,9,12,15,18,21,24,27 mo) Tarlatamab–durvalumab | 282 | 276 | 258 | 246 | 222 | 132 | 55 | 9 | 0 Durvalumab | 281 | 252 | 208 | 171 | 146 | 81 | 33 | 8 | 1 | 0 Survival table Endpoint: Overall survival Arm | N | Median | Median CI Tarlatamab–Durvalumab | 282 | 21.1 mo | 20.8–NE Durvalumab | 281 | 12.7 mo | 11.2–14.8 HR: 0.42 HR CI: 0.32–0.54 P: P<0.001 Landmark rates: 6 months: Tarlatamab–durvalumab 91.5 | Durvalumab 74.6 12 months: Tarlatamab–durvalumab 79.4 | Durvalumab 53.3 [Slide 2] Percentage of Patients vs Months Tarlatamab–durvalumab | Durvalumab Curve labels: 53.4 and 17.2 at 6 months; 38.4 and 9.5 at 12 months No. of Patients | Median Progression-free Survival (95% CI) mo Tarlatamab–Durvalumab | 282 | 7.4 (5.6–9.1) Durvalumab | 281 | 1.9 (1.9–2.0) Hazard ratio for disease progression or death, 0.34 (95% CI, 0.28–0.42) P<0.001 No. at Risk (0,3,6,9,12,15,18,21,24 mo) Tarlatamab–durvalumab | 282 | 225 | 147 | 122 | 86 | 32 | 14 | 1 | 0 Durvalumab | 281 | 106 | 46 | 36 | 18 | 8 | 2 | 0 Survival table Endpoint: Progression-free survival Arm | N | Median | Median CI Tarlatamab–Durvalumab | 282 | 7.4 mo | 5.6–9.1 Durvalumab | 281 | 1.9 mo | 1.9–2.0 HR: 0.34 HR CI: 0.28–0.42 P: P<0.001 Landmark rates: 6 months: Tarlatamab–durvalumab 53.4 | Durvalumab 17.2 12 months: Tarlatamab–durvalumab 38.4 | Durvalumab 9.5

[Slide 1] United States | Wake Forest University Health Sciences Clinical Research Unit | William Petty Yale Cancer Center | Anne Chiang Rutgers Cancer Institute of New Jersey | Missak Haigentz Swedish Cancer Institute | Kelly Paulson University Hospitals Cleveland Medical Center | Afshin Dowlati Dana Farber Cancer Institute | Jacob Sands Montefiore Medical Center - Bronx | Balazs Halmos Medical University of South Carolina | Mariam Alexander Mayo Clinic | Konstantinos Leventakos Baptist Cancer Center | Philip Lammers University of Texas MD Anderson Cancer Center | Carl Gay Franciscan Health Indianapolis | Nibal Saad Cancer and Hematology Centers of Western Michigan | Eric Santos; Yuanbin Chen (previous investigator) Sarah Cannon Research Institute Oncology Partners | Melissa Johnson Allegheny Health Network | Ariel Lopez-Chavez Astera Cancer Care | Bruno Fang Oncology Hematology Care Incorporated | Jayadev Mettu Virginia Oncology Associates | John Paschold University of Illinois Chicago | Frank Weinberg University Cancer and Blood Center LLC | Charles Bodine Oncology Associates of Oregon, PC | Bo Wang [Slide 2] Figure S3. Overall Survival Analysis in Prespecified Subgroups Subgroup | Tarlatamab + Durvalumab (no. of events / no. of patients) | Durvalumab (no. of events / no. of patients) | Hazard Ratio for Death (95% CI) Age < 65 years | 34 / 139 | 69 / 133 | 0.36 (0.24–0.54) ≥ 65 years | 53 / 143 | 87 / 148 | 0.48 (0.34–0.67) Sex Male | 64 / 197 | 120 / 201 | 0.38 (0.28–0.51) Female | 23 / 85 | 36 / 80 | 0.56 (0.33–0.94) Race White | 54 / 163 | 108 / 181 | 0.40 (0.29–0.56) Black | 0 / 4 | 0 / 4 | - Asian | 31 / 109 | 45 / 92 | 0.47 (0.30–0.74) Other | 2 / 6 | 3 / 4 | - Region Americas | 13 / 36 | 16 / 39 | 0.82 (0.39–1.71) Asia-Pacific | 32 / 114 | 51 / 101 | 0.43 (0.28–0.67) Europe | 42 / 132 | 89 / 141 | 0.36 (0.25–0.52) Presence (previous or current) of brain metastases Yes | 15 / 55 | 26 / 56 | 0.49 (0.26–0.93) No | 72 / 227 | 130 / 225 | 0.41 (0.30–0.54)

16:30 - 18:15 Presidential Symposium II CHAIRS: GIUSEPPE CURIGLIANO, YELENA JANJIGIAN Drug access [World map] Sotorasib only: Japan • Germany • Spain • Canada • Netherlands • Norway • Austria • Israel • Sweden • Czech Republic • Slovenia • South Korea • Finland Sotorasib + adagrasib: United States • Italy • Poland Chinese KRAS G12C inhibitors: China — Fulzerasib • Garsorasib • Glecirasib • Sosimerasib • Excluding China, 87% of the global population has no access to KRASi • Will these countries accept sotorasib / adagrasib as a control arm ? • What will be the impact of new regulations ? Image generated with copilot Benjamin BESSE Speaker panel: Benjamin Besse — Invited Discussant LBA5 ALICANTE AUDITORIUM - HALL 6

[Slide 1] Krascendo 1: a phase III, global, randomised, open-label trial in previously treated KRAS G12C-mutated advanced or metastatic NSCLC Key eligibility criteria • KRAS G12C inhibitor naive • ECOG PS 0–1 • Advanced KRAS G12C-mutated NSCLC • ≥1 but ≤3 lines of prior systemic therapy, including platinum-based chemo and a PD-(L)1 inhibitor* • Measurable disease per RECIST v1.1 • Patients with active, untreated CNS metastases were excluded R 1:1 N=338 Divarasib (400 mg QD) Treatment crossover not permitted Sotorasib (960 mg QD) OR adagrasib (600 mg BID)† Treatment until disease progression, unacceptable toxicity, death, withdrawal of consent, or study termination Treatment beyond progression was permitted Stratification factors • History of CNS metastases • Intended choice of control arm KRAS G12C inhibitor† • KEAP1 mutation status‡ Primary endpoint • PFS per BICR Key secondary endpoints • OS • Confirmed ORR per BICR • TTCD Other secondary endpoints • DoR per BICR • Safety • Other PROs NCT06497556 PD-L1 all-comers population. *In the advanced or metastatic setting. †The control arm drug depended on local approvals and availability; if both drugs were locally approved and available, investigators designated their intended control drug for each patient prior to randomisation and patients received the same drug for the duration of the study. ‡Detected versus not detected/unknown. BICR, blinded independent central review; BID, twice a day; chemo, chemotherapy; CNS, central nervous system; DoR, duration of response; ECOG PS, Eastern Cooperative Oncology Group performance status; PD-1, programmed cell death protein 1; PD-L1, programmed cell death ligand 1; PROs, patient-reported outcomes; TTCD, time to confirmed deterioration; QD, once daily; RECIST, Response Evaluation Criteria in Solid Tumors Dr Ferdinandos Skoulidis [Slide 2] Overall survival | Divarasib (n=167) | Control (n=171) Patients with event, n (%) | 63 (37.7) | 93 (54.4) Median OS, months (95% CI) | 17.7 (16.3–NE) | 12.0 (10.0–15.2) Stratified HR (95% CI) | 0.56 (0.41–0.78) p-value* | 0.0005 12-month OS, % | 67.1 | 49.2 17.7 months 12.0 months 67.1% 49.2% Patients at risk Time (months) 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 Divarasib 167 | 160 | 142 | 120 | 95 | 41 | 6 | 0 Control 171 | 159 | 128 | 94 | 65 | 31 | 6 | 0 Data cut-off: 19 May 2026. Median duration of follow up: 14.3 months. Minimal duration of follow up: 5.6 months. Control arm included treatment with sotorasib or adagrasib *Stratified log-rank test. NE, not estimable Dr Ferdinandos Skoulidis Survival table Endpoint: Overall survival Arm | N | Median | Median CI Divarasib | 167 | 17.7 | 16.3–NE Control (sotorasib or adagrasib) | 171 | 12.0 | 10.0–15.2 HR: 0.56 HR CI: 0.41–0.78 P: 0.0005 Landmark rates: 12 months: Divarasib 67.1% | Control 49.2% [Slide 3] Progression-free survival by BICR | Divarasib (n=167) | Control (n=171) Patients with event, n (%) | 101 (60.5) | 128 (74.9) Median PFS, months (95% CI) | 9.8 (8.1–12.9) | 5.7 (4.2–6.8) Stratified HR (95% CI) | 0.58 (0.44–0.76) p-value* | <0.0001 12-month PFS, % | 43.9 | 24.0 9.8 months 5.7 months 43.9% 24.0% Patients at risk Time (months) 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 Divarasib 167 | 142 | 108 | 80 | 53 | 27 | 3 | 0 Control 171 | 116 | 73 | 54 | 26 | 14 | 0 | 0 Data cut-off: 19 May 2026. Median duration of follow up: 14.3 months. Minimal duration of follow up: 5.6 months. Control arm included treatment with sotorasib or adagrasib *Stratified log-rank test. CI, confidence interval Dr Ferdinandos Skoulidis Survival table Endpoint: Progression-free survival by BICR (primary endpoint) Arm | N | Median | Median CI Divarasib | 167 | 9.8 | 8.1–12.9 Control (sotorasib or adagrasib) | 171 | 5.7 | 4.2–6.8 HR: 0.58 HR CI: 0.44–0.76 P: <0.0001 Landmark rates: 12 months: Divarasib 43.9% | Control 24.0%

| Diva n=167 | Control* n=171 Efficacy Median PFS,† mos (95% CI) | 9.8 (8.1–12.9) | 5.7 (4.2–6.8) HR‡ (95% CI) | 0.58 (0.44–0.76) p value‡ | <0.0001 Median OS, mos (95% CI) | 17.7 (16.3–NE) | 12.0 (10.0–15.2) HR‡ (95% CI) | 0.56 (0.41–0.78) p value‡ | 0.0005 Confirmed ORR,† % | 58.4 | 32.3 n/n§ | 94/161 | 54/167 95% CI | 50.4–66.1 | 25.3–40.0 Safety Median tx duration, mos (range) | 11.3 (0–19.8) | 5.0 (0–19.8) AE/TRAE, n (%) All grade | 166 (99.4) / 155 (92.8) | 167 (97.7) / 137 (80.1) Grade 3–4 | 72 (43.1) / 41 (24.6) | 88 (51.5) / 59 (34.5) Serious | 51 (30.5) / 13 (7.8) | 70 (40.9) / 26 (15.2) Leading to discontinuation | 12 (7.2) / 8 (4.8) | 25 (14.6) / 20 (11.7) *Soto: n=156; ada: n=15. †Per BICR. ‡Stratified. §Response-evaluable set per BICR. ¶No Grade 5 TRAEs.

KRASCENDO-1 - population Key eligibility criteria • KRAS G12C inhibitor naive • ECOG PS 0–1 • Advanced KRAS G12C-mutated NSCLC • ≥1 but ≤3 lines of prior systemic therapy, including platinum-based chemo and a PD-(L)1 inhibitor* • Measurable disease per RECIST v1.1 • Patients with active, untreated CNS metastases were excluded R 1:1, N=338 Divarasib (400 mg QD) Treatment crossover not permitted Sotorasib (960 mg QD) OR adagrasib (600 mg BID)† Treatment until disease progression, unacceptable toxicity, death, withdrawal of consent, or study termination Stratification factors: • History of CNS metastases • Intended choice of control arm KRAS G12C inhibitor† • KEAP1 mutation status‡ Primary endpoint: • PFS per BICR Key secondary endpoints: • OS • Confirmed ORR per BICR • TTCD Other secondary endpoints: • DoR per BICR • Safety • Other PROs Prior therapy: Chemo-IO 69%; Chemo-IO → Docetaxel / IO → Chemo 30%; 9%; IO → Chemo → Docetaxel 1% KRAS G12C INHIBITOR: 91% SOTORASIB | ADAGRASIB 9% KEAP1 MUTATION STATUS: 22% PRESENT | 78% NOT DETECTED / UNKNOWN HISTORY OF BRAIN METASTASES: 30% YES (History present) rwPFS | sotorasib vs. adagrasib | rwOS [embedded real-world KM and forest plots; small text illegible] Skoulidis ESMO 2026, Barsouk ASCO 2026 All baseline-characteristic graphics are confined to the lower half of the slide. ORR: objective response rate, PFS: progression free survival, OS: overall survival, IO: immunotherapy, rw: real world

LBA5 - Krascendo 1: Efficacy and safety of divarasib (diva) vs sotorasib (soto) or adagrasib (ada) in previously treated locally advanced or metastatic KRAS G12C-mutated NSCLC Speakers Ferdinandos Skoulidis (Houston, United States of America, TX) Lecture Time 16:30 - 16:42 Abstract Background Diva is an oral, next-generation KRAS G12C inhibitor that is more potent and selective for KRAS G12C in vitro than soto or ada. We report the primary analysis of Krascendo 1 (NCT06497556), a phase 3, global, randomised, open-label trial of diva vs soto or ada in advanced/metastatic KRAS G12C-mutated NSCLC. Methods Eligible patients (pts) aged ≥18 years had an ECOG PS of 0/1 and ≥1 prior systemic treatment (tx; including anti-PD-1/PD-L1 and platinum-based chemotherapy) for advanced/metastatic KRAS G12C-mutated NSCLC. Pts were randomised 1:1 to diva (400 mg PO QD) or investigator's choice of soto (960 mg PO QD) or ada (600 mg PO BID). Stratification factors: KEAP1 status, history of CNS metastases and intended KRAS G12C inhibitor (soto/ada). Tx continued until unacceptable toxicity, clinical benefit loss, consent withdrawal or death. Primary endpoint: PFS per blinded independent committee review (BICR). Secondary endpoints included OS, confirmed ORR per BICR (RECIST 1.1), patient-reported outcomes, DoR and safety. Results In total 338 pts were randomised to diva (n=167) or control (n=171); baseline characteristics were generally balanced. At data cutoff (19 May 2026), median follow up was 14.3 months. A statistically significant benefit in PFS (HR: 0.58; 95% CI: 0.44–0.76) and OS (HR: 0.56; 95% CI: 0.41–0.78) was observed for the diva vs control arm (Table). A statistically significant improvement in ORR was seen for diva (58.4%) vs control (32.3%; table). Diva had a manageable safety profile consistent with previous reports for diva and other KRAS G12C inhibitors (Table). Tx duration was twice as long in the diva vs control arm, and discontinuation rates due to TRAEs were lower with diva vs control (Table). Table: LBA5 | Diva n=167 | Control* n=171 Efficacy Median PFS,† mos (95% CI) | 9.8 (8.1–12.9) | 5.7 (4.2–6.8) HR‡ (95% CI) | 0.58 (0.44–0.76) p value‡ | <0.0001 Median OS, mos (95% CI) | 17.7 (16.3–NE) | 12.0 (10.0–15.2) HR‡ (95% CI) | 0.56 (0.41–0.78) p value‡ | 0.0005 Confirmed ORR,† % | 58.4 | 32.3 n/n§ | 94/161 | 54/167 95% CI | 50.4–66.1 | 25.3–40.0 Safety Median tx duration, mos (range) | 11.3 (0–19.8) | 5.0 (0–19.8) AE/TRAE, n (%) All grade | 166 (99.4)/155 (92.8) | 167 (97.7)/137 (80.1) Grade 3–4 | 72 (43.1)/41 (24.6)¶ | 88 (51.5)/59 (34.5)¶ Serious | 51 (30.5)/13 (7.8) | 70 (40.9)/26 (15.2) Leading to discontinuation | 12 (7.2)/8 (4.8) | 25 (14.6)/20 (11.7) *Soto: n=156; ada: n=15. †Per BICR. ‡Stratified. §Response-evaluable set per BICR. ¶No Grade 5 TRAEs. Conclusions Diva significantly improved PFS and OS vs control with a manageable safety profile, establishing diva as a new standard of care for pts with previously treated advanced/metastatic KRAS G12C-mutated NSCLC.

Confirmed ORR and DoR by BICR [Waterfall plots: Divarasib; Control - Best change from baseline (%); Best overall response: PD, SD, PR, CR] n (%), unless otherwise specified | Divarasib (n=161) | Control (n=167) Confirmed ORR* | 94 (58) | 54 (32) 95% CI† | 50.4–66.1 | 25.3–40.0 Difference,‡ % (95% CI) | 25.8 (15.6–36.0) CR | 7 (4) | 6 (4) PR | 87 (54) | 48 (29) SD | 56 (35) | 80 (48) PD | 7 (4) | 19 (11) Missing | 4 (2) | 14 (8) DoR | n=94 | n=54 Median, months (95% CI) | 11.9 (9.7–NE) | 6.9 (5.3–8.3) Median time to confirmed response, days | n=94 43.0 | n=54 44.0 Median best change from baseline,§ % | n=157 -47.9 | n=152 -27.1 Divarasib induced a statistically significant 25.8 percentage point increase in ORR versus control (58% vs 32%, respectively) Data cut-off: 19 May 2026. Median duration of follow up: 14.3 months. Minimal duration of follow up: 5.6 months. Control arm included treatment with sotorasib or adagrasib. ORR was considered statistically significant if the 95% CI difference excluded zero. *BICR-assessed in the response-evaluable per BICR population (n=328; ten patients [divarasib arm, n=6; control arm, n=4] did not have measurable disease per BICR at baseline and were excluded from the response-evaluable set). †Calculated using the Clopper-Pearson method. ‡Covariate adjusted difference, adjusted for the stratification factors included in randomisation. §All patients with baseline and a post-baseline tumour assessment prior to study drug discontinuation and prior to subsequent anti-cancer therapy. CR, complete response; PD, progressive disease; PR, partial response; SD, stable disease Dr Ferdinandos Skoulidis

[Slide 1] [KM curve] Percentage of Patients without an Event Divarasib | Control 6 months: Divarasib 65.7; Control 47.8 12 months: Divarasib 43.9; Control 24.0 Months ORIGINAL ARTICLE MEETING OF THE EUROPEAN SOCIETY FOR MEDICAL ONCOLOGY Divarasib for KRAS G12C–Mutated NSCLC F. Skoulidis and Others Survival table Endpoint: Event-free (endpoint not named on image; likely PFS) Divarasib Control Landmark rates: 6 months: Divarasib 65.7% | Control 47.8% 12 months: Divarasib 43.9% | Control 24.0% [Slide 2] The NEW ENGLAND JOURNAL of MEDICINE Progression-free Survival in the Full Analysis Population No. of Events/No. of Patients | Median Progression-free Survival (95% CI), mo Osimertinib–Savolitinib | 117/171 | 7.6 (5.6–9.7) Platinum–Pemetrexed | 113/167 | 5.4 (4.2–5.8) Hazard ratio for disease progression or death, 0.55 (95% CI, 0.42–0.72) P<0.001 12 months: Osimertinib–savolitinib 36; Platinum–pemetrexed 10 Months since Randomization ORIGINAL ARTICLE MEETING OF THE EUROPEAN SOCIETY FOR MEDICAL ONCOLOGY Osimertinib plus Savolitinib in Advanced NSCLC S. Lu and Others Survival table Endpoint: PFS (full analysis population) Arm | N | Median | Median CI Osimertinib–savolitinib | 171 | 7.6 | 5.6–9.7 Platinum–pemetrexed | 167 | 5.4 | 4.2–5.8 HR: 0.55 HR CI: 0.42–0.72 P: P<0.001 Landmark rates: 12 months: Osimertinib–savolitinib 36% | Platinum–pemetrexed 10%

Progression-free survival by BICR subgroup analysis Baseline risk factors | Subgroup | Total n | Divarasib n | Divarasib median (months) | Control n | Control median (months) | HR (95% CI) All Patients | | 338 | 167 | 9.8 | 171 | 5.7 | 0.57 (0.44–0.74) Sex | Female | 147 | 72 | 11.1 | 75 | 5.7 | 0.50 (0.33–0.77) Sex | Male | 191 | 95 | 8.3 | 96 | 5.4 | 0.61 (0.43–0.85) Race | Asian | 43 | 20 | 10.7 | 23 | 6.7 | 0.58 (0.28–1.21) Race | Non-Asian | 295 | 147 | 9.8 | 148 | 5.4 | 0.57 (0.43–0.75) Age | <65 | 142 | 70 | 8.4 | 72 | 5.5 | 0.50 (0.33–0.75) Age | ≥65 | 196 | 97 | 10.3 | 99 | 5.7 | 0.63 (0.45–0.89) Geographic region | Europe | 209 | 106 | 8.3 | 103 | 4.3 | 0.54 (0.39–0.75) Geographic region | North America | 55 | 28 | 10.9 | 27 | 5.7 | 0.79 (0.38–1.64) Geographic region | Rest of World | 74 | 33 | 13.2 | 41 | 6.9 | 0.52 (0.29–0.92) KEAP1 mutation status* | Detected | 72 | 35 | 4.6 | 37 | 2.8 | 0.56 (0.33–0.92) KEAP1 mutation status* | Not detected/unknown† | 266 | 132 | 12.9 | 134 | 6.7 | 0.55 (0.40–0.75) History of CNS metastases* | Yes | 99 | 48 | 6.8 | 51 | 5.7 | 0.74 (0.47–1.17) History of CNS metastases* | No | 239 | 119 | 11.2 | 120 | 5.4 | 0.52 (0.37–0.71) ECOG PS at baseline | 0 | 133 | 65 | 13.4 | 68 | 8.5 | 0.58 (0.37–0.92) ECOG PS at baseline | 1 | 205 | 102 | 8.2 | 103 | 4.4 | 0.54 (0.39–0.74) Line of therapy | Second line | 239 | 121 | 10.9 | 118 | 5.4 | 0.56 (0.40–0.76) Line of therapy | Third/fourth line | 99 | 46 | 8.3 | 53 | 6.5 | 0.62 (0.39–1.01) PD-L1 status (local testing)‡§ | TC/TPS <1% | 127 | 57 | 9.8 | 70 | 6.9 | 0.71 (0.46–1.09) PD-L1 status (local testing)‡§ | TC/TPS 1–<50% | 127 | 62 | 13.2 | 65 | 4.4 | 0.41 (0.26–0.65) PD-L1 status (local testing)‡§ | TC/TPS ≥50% | 75 | 43 | 8.1 | 32 | 4.4 | 0.61 (0.35–1.06) Liver lesion at baseline¶ | Yes | 36 | 17 | 5.5 | 19 | 3.0 | 0.39 (0.18–0.83) Liver lesion at baseline¶ | No | 302 | 150 | 10.9 | 152 | 6.5 | 0.57 (0.43–0.76) Divarasib better <- -> Control better PFS benefit with divarasib was generally consistent across subgroups Data cut-off: 19 May 2026. Median duration of follow up: 14.3 months. Minimal duration of follow up: 5.6 months. Control arm included treatment with sotorasib or adagrasib All analyses shown in the forest plot are unstratified. *Per IxRS. †Three patients had unknown KEAP1 status (all control arm). ‡PD-L1 testing was conducted locally. PD-L1 TC/TPS negative: <1%, PD-L1 TC/TPS low: 1–<50%, PD-L1 TC/TPS high: ≥50% §Data missing for 9 patients. ¶Per BICR Dr Ferdinandos Skoulidis

[Slide 2] 16:30 - 18:15 Presidential Symposium II CHAIRS: GIUSEPPE CURIGLIANO, YELENA JANJIGIAN Ferdinandos Skoulidis — Krascendo 1: Efficacy and safety of divarasib (diva) vs sotorasib (soto) or adagrasib (ada) in previously treated locally advanced or metastatic KRAS G12C-mutated NSCLC ALICANTE AUDITORIUM - HALL 6 Progression-free survival by BICR | Divarasib (n=167) | Control (n=171) Patients with event, n (%) | 101 (60.5) | 128 (74.9) Median PFS, months (95% CI) | 9.8 (8.1–12.9) | 5.7 (4.2–6.8) Stratified HR (95% CI) | 0.58 (0.44–0.76) p-value* | <0.0001 12-month PFS, % | 43.9 | 24.0 9.8 months 5.7 months 43.9% 24.0% Patients at risk Time (months) 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 Divarasib 167 | 142 | 108 | 80 | 53 | 27 | 3 | 0 Control 171 | 116 | 73 | 54 | 26 | 14 | 0 | 0 Data cut-off: 19 May 2026. Median duration of follow up: 14.3 months. Minimal duration of follow up: 5.6 months. Control arm included treatment with sotorasib or adagrasib *Stratified log-rank test. CI, confidence interval Dr Ferdinandos Skoulidis Survival table Endpoint: Progression-free survival by BICR (primary endpoint) Arm | N | Median | Median CI Divarasib | 167 | 9.8 | 8.1–12.9 Control (sotorasib or adagrasib) | 171 | 5.7 | 4.2–6.8 HR: 0.58 HR CI: 0.44–0.76 P: <0.0001 Landmark rates: 12 months: Divarasib 43.9% | Control 24.0% [Slide 4] 16:30 - 18:15 Presidential Symposium II CHAIRS: GIUSEPPE CURIGLIANO, YELENA JANJIGIAN Ferdinandos Skoulidis — Krascendo 1: Efficacy and safety of divarasib (diva) vs sotorasib (soto) or adagrasib (ada) in previously treated locally advanced or metastatic KRAS G12C-mutated NSCLC ALICANTE AUDITORIUM - HALL 6 Safety summary Safety parameter, n (%) unless otherwise specified | Divarasib (n=167) | Control (n=171) Median treatment duration, months (range) | 11.3 (0.0–19.8) | 5.0 (0.0–19.8) Patients with ≥1 AE | 166 (99.4) | 167 (97.7) Grade 3–4 | 72 (43.1) | 88 (51.5) Serious | 51 (30.5) | 70 (40.9) Grade 5 | 10 (6.0)* | 7 (4.1)† Patients with ≥1 TRAE | 155 (92.8) | 137 (80.1) Grade 3–4 | 41 (24.6) | 59 (34.5) Serious | 13 (7.8) | 26 (15.2) Grade 5 | 0 | 0 AE leading to: Treatment discontinuation | 12 (7.2) | 25 (14.6) Dose reduction | 44 (26.3) | 52 (30.4) Dose interruption | 57 (34.1) | 62 (36.3) TRAEs leading to: Treatment discontinuation | 8 (4.8) | 20 (11.7) Dose reduction | 44 (26.3) | 51 (29.8) Dose interruption | 33 (19.8) | 40 (23.4) Divarasib had a favourable safety profile, with fewer grade 3–4 TRAEs and SAEs versus control While median treatment duration was twice as long with divarasib versus control, approximately half as many patients had TRAEs leading to treatment discontinuation Data cut-off: 19 May 2026. Median duration of follow up: 14.3 months. Minimal duration of follow up: 5.6 months. Control arm included treatment with sotorasib or adagrasib [footnotes * and † listing grade 5 event causes — too small to transcribe reliably] Dr Ferdinandos Skoulidis

[Slide 1] Krascendo 1: a phase III, global, randomised, open-label trial in previously treated KRAS G12C-mutated advanced or metastatic NSCLC Key eligibility criteria • KRAS G12C inhibitor naive • ECOG PS 0–1 • Advanced KRAS G12C-mutated NSCLC • ≥1 but ≤3 lines of prior systemic therapy, including platinum-based chemo and a PD-(L)1 inhibitor* • Measurable disease per RECIST v1.1 • Patients with active, untreated CNS metastases were excluded R 1:1 N=338 Divarasib (400 mg QD) Treatment crossover not permitted Sotorasib (960 mg QD) OR adagrasib (600 mg BID)† Treatment until disease progression, unacceptable toxicity, death, withdrawal of consent, or study termination Treatment beyond progression was permitted Stratification factors: History of CNS metastases; Intended choice of control arm KRAS G12C inhibitor†; KEAP1 mutation status‡ Primary endpoint: PFS per BICR Key secondary endpoints: OS; Confirmed ORR per BICR; TTCD Other secondary endpoints: DoR per BICR; Safety; Other PROs NCT06497556 [footnotes illegible at this distance] Dr Ferdinandos Skoulidis

[Slide 1] Proffered Paper session Introduction Lecture Time 16:30 - 16:30 Proffered Paper session LBA5 - Krascendo 1: Efficacy and safety of divarasib (diva) vs sotorasib (soto) or adagrasib (ada) in previously treated locally advanced or metastatic KRAS G12C-mutated NSCLC Speakers Ferdinandos Skoulidis (Houston, United States of America, TX) Lecture Time 16:30 - 16:42 Abstract Background Diva is an oral, next-generation KRAS G12C inhibitor that is more potent and selective for KRAS G12C in vitro than soto or ada. We report the primary analysis of Krascendo 1 (NCT06497556), a phase 3, global, randomised, open-label trial of diva vs soto or ada in advanced/metastatic KRAS G12C-mutated NSCLC. [Slide 2] Methods Eligible patients (pts) aged ≥18 years had an ECOG PS of 0/1 and ≥1 prior systemic treatment (tx; including anti-PD-1/PD-L1 and platinum-based chemotherapy) for advanced/metastatic KRAS G12C-mutated NSCLC. Pts were randomised 1:1 to diva (400 mg PO QD) or investigator's choice of soto (960 mg PO QD) or ada (600 mg PO BID). Stratification factors: KEAP1 status, history of CNS metastases and intended KRAS G12C inhibitor (soto/ada). Tx continued until unacceptable toxicity, clinical benefit loss, consent withdrawal or death. Primary endpoint: PFS per blinded independent committee review (BICR). Secondary endpoints included OS, confirmed ORR per BICR (RECIST 1.1), patient-reported outcomes, DoR and safety. [Slide 4] Conclusions Diva significantly improved PFS and OS vs control with a manageable safety profile, establishing diva as a new standard of care for pts with previously treated advanced/metastatic KRAS G12C-mutated NSCLC. Clinical trial identification NCT06497556.

Krascendo 1: efficacy and safety of divarasib versus sotorasib or adagrasib in previously treated locally advanced or metastatic KRAS G12C-mutated NSCLC Ferdinandos Skoulidis,1 Myung-Ju Ahn,2 Shirish Gadgeel,3 Sanjay Popat,4 Niels Reinmuth,5 Bianca van Veggel,6 Farastuk Bozorgmehr,7 Victoria J Bray,8 Michael Duruisseaux,9 Wade T Iams,10 Julien Mazieres,11 Laura Bonanno,12 Lore Decoster,13 Bo Wang,14 Youyou Hu,15 Sarah M Shagan,16 Verena Lam,16 Uyi Osaghae,16 Tien Hoang,16 Benjamin J Solomon17 1UT MD Anderson, Houston, TX, USA; 2Samsung Medical Center, Seoul, Republic of Korea; 3Emory University, Atlanta, GA, USA; 4Royal Marsden Hospital & Institute of Cancer Research, London, UK; 5German Centre for Lung Research, Munich-Gauting, Germany; 6Jeroen Bosch Hospital, Hertogenbosch, Netherlands; 7Heidelberg University and National Center for Tumor Diseases (NCT), NCT Heidelberg, a partnership between DKFZ and Heidelberg University Hospital, Heidelberg, Germany and Translational Lung Research Centre Heidelberg, LRC-H, German Center for Lung Research (DZL), Heidelberg, Germany; 8Liverpool Hospital, Liverpool, NSW, Australia; 9Louis Pradel Hospital, Hospices Civils de Lyon Cancer Institute, Bron, France; TheraLung Lab, Cancer Research Center of Lyon, UMR INSERM 1052, CNRS 5286, Lyon, France and Université Claude Bernard Lyon 1, Université de Lyon, Lyon, France; 10Greco-Hainsworth Centres for Research, Tennessee Oncology, Nashville, TN, USA; 11Centre Hospitalier Universitaire de Toulouse, Université de Toulouse, Toulouse, France; 12Oncology and Gastroenterology, University of Padova, Padova, Italy; Medical Oncology 2, Istituto Oncologico Veneto IOV IRCCS, Padova, Italy; 13Universitair Ziekenhuis Brussel, Vrije Universiteit Brussel, Jette, Belgium; 14Willamette Valley Cancer Institute and Research Center, Eugene, OR, USA; 15F. Hoffmann-La Roche Ltd, Basel, Switzerland; 16Genentech Inc., South San Francisco, CA, USA; 17Peter MacCallum Cancer Centre, Melbourne, VIC, Australia

[Slide 1] 16:30 - 18:15 Presidential Symposium II CHAIRS: GIUSEPPE CURIGLIANO, YELENA JANJIGIAN Krascendo 1: a phase III, global, randomised, open-label trial in previously treated KRAS G12C-mutated advanced or metastatic NSCLC Key eligibility criteria • KRAS G12C inhibitor naive • ECOG PS 0–1 • Advanced KRAS G12C-mutated NSCLC • ≥1 but ≤3 lines of prior systemic therapy, including platinum-based chemo and a PD-(L)1 inhibitor* • Measurable disease per RECIST v1.1 • Patients with active, untreated CNS metastases were excluded R 1:1, N=338 Divarasib (400 mg QD) Treatment crossover not permitted Sotorasib (960 mg QD) OR adagrasib (600 mg BID)† Treatment until disease progression, unacceptable toxicity, death, withdrawal of consent, or study termination Treatment beyond progression was permitted Stratification factors: • History of CNS metastases • Intended choice of control arm KRAS G12C inhibitor† • KEAP1 mutation status‡ Primary endpoint: • PFS per BICR Key secondary endpoints: • OS • Confirmed ORR per BICR • TTCD Other secondary endpoints: • DoR per BICR • Safety • Other PROs NCT06497556 [footnotes small, partially illegible] PD-L1 all-comers population. *In the advanced or metastatic setting. †The control arm drug depended on local approvals and availability; if both drugs were locally approved and available, investigators designated their intended control drug for each patient prior to randomisation and patients received the same drug for the duration of the study. ‡Detected versus not detected/unknown. Dr Ferdinandos Skoulidis Speaker panel: Ferdinandos Skoulidis — Krascendo 1: Efficacy and safety of divarasib (diva) vs sotorasib (soto) or adagrasib (ada) in previously treated locally advanced or metastatic KRAS G12C-mutated NSCLC ALICANTE AUDITORIUM - HALL 6 [Slide 4] 16:30 - 18:15 Presidential Symposium II CHAIRS: GIUSEPPE CURIGLIANO, YELENA JANJIGIAN Summary • Krascendo 1 is the first and only global trial evaluating a next-generation KRAS G12C inhibitor versus first-generation inhibitors • The Krascendo 1 trial met its primary endpoint BICR-assessed PFS and key secondary endpoints OS and ORR, demonstrating a statistically significant and clinically meaningful improvement with divarasib versus control KRAS G12C inhibitor (sotorasib or adagrasib) – Median PFS: 9.8 vs 5.7 months with divarasib versus control (HR: 0.58; 95% CI: 0.44–0.76) o Investigator-assessed PFS was consistent with BICR-assessed PFS – Median OS: 17.7 vs 12.0 months for divarasib versus control (HR: 0.56; 95% CI: 0.41–0.78) – Confirmed ORR: 58% vs 32% for divarasib versus control (difference: 25.8%; 95% CI: 15.6–36.0) • Divarasib exhibited a manageable safety profile consistent with previous reports, with no emergent new safety signals Divarasib represents a new standard of care for patients with previously treated advanced/metastatic KRAS G12C-mutated NSCLC Dr Ferdinandos Skoulidis ALICANTE AUDITORIUM - HALL 6

[Slide 2] 16:30 - 18:15 Presidential Symposium II CHAIRS: GIUSEPPE CURIGLIANO, YELENA JANJIGIAN Benjamin Besse - Invited Discussant LBA5 Conclusion • KRASCENDO-1 demonstrates robust superiority of divarasib over sotorasib/adagrasib for the primary endpoint, PFS • A meaningful overall survival benefit is reported • The next frontier will be first-line development in combination with immunotherapy ± chemotherapy • Clinical trial paradigms must evolve to keep pace with therapeutic progress • Systematic cross-over should be reconsidered when PFS is the primary endpoint • Toxicity assessment should capture chronic treatment burden, not only acute adverse events • Eligibility criteria should be broadened to better reflect real-world populations, rather than a highly selected minority ALICANTE AUDITORIUM - HALL 6

KRAS (OFF) G12Ci in Lung cancer Phase III study in 2nd line, after chemo-immunotherapy, vs docetaxel CODEBREAK Sotorasib vs docetaxel [embedded KM plots, small text illegible] 5.6 mo; 10.6 mo KRYSTAL Adagrasib vs docetaxel [embedded KM plot, small text illegible] 5.5 mo Phase I/II data | ORR (%) | PFS (mo) | OS (mo) DIVARASIB. N=44 (400 mg) | 59 | 15.3 | ADAGRASIB. N=116 | 43 | 6.9 | 14.1 SOTORASIB. N=126 | 41 | 6.3 | 12.5 De Langen Lancet 2023, Barlesi Lancet 2025, Sacher NEJM 2023, Remon Cancer Treat Rev 2026 ORR: objective response rate, PFS: progression free survival, OS: overall survival Benjamin BESSE

Krascendo 1: efficacy and safety of divarasib versus sotorasib or adagrasib in previously treated locally advanced or metastatic KRAS G12C-mutated NSCLC Ferdinandos Skoulidis,1 Myung-Ju Ahn,2 Shirish Gadgeel,3 Sanjay Popat,4 Niels Reinmuth,5 Bianca van Veggel,6 Farastuk Bozorgmehr,7 Victoria J Bray,8 Michael Duruisseaux,9 Wade T Iams,10 Julien Mazieres,11 Laura Bonanno,12 Lore Decoster,13 Bo Wang,14 Youyou Hu,15 Sarah M Shagan,16 Verena Lam,16 Uyi Osaghae,16 Tien Hoang,16 Benjamin J Solomon17

Safety summary Safety parameter, n (%) unless otherwise specified | Divarasib (n=167) | Control (n=171) Median treatment duration, months (range) | 11.3 (0.0–19.8) | 5.0 (0.0–19.8) Patients with ≥1 AE | 166 (99.4) | 167 (97.7) Grade 3–4 | 72 (43.1) | 88 (51.5) Serious | 51 (30.5) | 70 (40.9) Grade 5 | 10 (6.0)* | 7 (4.1)† Patients with ≥1 TRAE | 155 (92.8) | 137 (80.1) Grade 3–4 | 41 (24.6) | 59 (34.5) Serious | 13 (7.8) | 26 (15.2) Grade 5 | 0 | 0 AE leading to: Treatment discontinuation | 12 (7.2) | 25 (14.6) Dose reduction | 44 (26.3) | 52 (30.4) Dose interruption | 57 (34.1) | 62 (36.3) TRAEs leading to: Treatment discontinuation | 8 (4.8) | 20 (11.7) Dose reduction | 44 (26.3) | 51 (29.8) Dose interruption | 33 (19.8) | 40 (23.4) Divarasib had a favourable safety profile, with fewer grade 3–4 TRAEs and SAEs versus control While median treatment duration was twice as long with divarasib versus control, approximately half as many patients had TRAEs leading to treatment discontinuation Data cut-off: 19 May 2026. Median duration of follow up: 14.3 months. Minimal duration of follow up: 5.6 months. Control arm included treatment with sotorasib or adagrasib *Pneumonia (n=3, 1.8%), urinary tract infection, gastrointestinal haemorrhage, upper gastrointestinal haemorrhage, transitional cell carcinoma, depressed level of consciousness, confusional state, chronic obstructive pulmonary disease (all n=1 [0.6%]). †Death (n=2, 1.2%), device-related sepsis, acute myeloid leukaemia, ischaemic stroke, myocardial infarction, road traffic incident (all n=1 [0.6%]). AE, adverse event; SAE, serious adverse event; TRAE, treatment-related adverse event Dr Ferdinandos Skoulidis

[Slide 2] Percentage of Patients without an Event Months Divarasib Control No. of Patients | No. of Patients with Event (%) | Median Progression-free Survival (mo) Stratified hazard ratio for disease progression or death, 0.58 (95% CI, 0.44–0.76) Stratified P<0.001 by log-rank test No. at Risk Figure 1. Progression-free Survival in the Full Analysis Population. Shown is the progression-free survival, as assessed by blinded independent central review, among the patients who received divarasib and those who received control treatment (sotorasib or adagrasib). The full analysis population included all the patients who underwent randomization. Survival table Endpoint: PFS by BICR, full analysis population Divarasib | 167 | 9.8 mo Control (sotorasib or adagrasib) | 171 | 5.7 mo HR: 0.58 HR CI: 0.44–0.76 P: <0.001 (stratified log-rank) 6 mo: divarasib 65.7%, control 47.8% (as labelled on figure) 12 mo: divarasib 43.9%, control 24.0% (as labelled on figure)

CHAIRS: GIUSEPPE CURIGLIANO, YELENA JANJIGIAN Most common TRAEs TRAEs occurring in ≥10% of patients by grade* Divarasib (n=167) | Control (n=171) Diarrhoea | 66% | 38% Nausea | 53% | 20% Vomiting | 39% | 8% ALT increased | 15% | 23% AST increased | 14% | 21% Fatigue | 10% | 13% Decreased appetite | 13% | 10% Lipase increased | 10% | 12% Blood ALP increased | 5% | 16% Constipation | 13% | 3% Grade 1 2 3 Divarasib; 1 2 3 4 Control Patients (%) • The safety profile of divarasib monotherapy was consistent with the known class effects of KRAS G12C inhibitors • The most common TRAEs observed with divarasib were diarrhoea, nausea and vomiting, and were mainly grade 1 (35%, 32% and 28%, respectively) – None of the TRAEs occurring in ≥10% of patients receiving divarasib were grade 4 – Hepatotoxicity rates were lower with divarasib than control treatment • No treatment-related QT prolongation or pneumonitis occurred with divarasib, only in the control arm, in 2 and 3 patients, respectively TRAEs for divarasib were mostly low grade, manageable and reversible Data cut-off: 19 May 2026. Median duration of follow up: 14.3 months. Minimal duration of follow up: 5.6 months. Control arm included treatment with sotorasib or adagrasib *Median duration of treatment: 11.3 months for divarasib, 5.0 months for control ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase Dr Ferdinandos Skoulidis

Next generation RASi Acquired resistance to divarasib [Oncoprint heatmap: NSCLC (Patients 10, 1, 37), Colorectal Cancer (Patients 4, 81, 76, 32, 86, 9, 133, 77, 96), Other Solid Tumors (Patients 104, 111, 105, 106); columns Best Response, Dose, PFS; KRAS Alterations (KRAS, KRAS G12A, G12D, G12R, G12V, G13D, R68S, H95R, Y96D, Y96H, Y96N, Q99L, A146T); MAPK or PI3K Alterations (BRAF G469A, L597Q, L597R, V600E, K601E, MAP2K1 Q56P, K57N, E203K, NRAS G13R, PIK3CA E545K, PTEN); RTK Alterations (EML4-ALK, ERBB2, MET); Other Alterations (RB1)] PFS values (NSCLC): 13.6, 9.1, 4.1; (CRC): 8.2, 8.2, 7.0, 6.9, 5.6, 5.4, 3.9, 3.9, 3.7; (Other): 5.6, 5.4, 5.4, 4.0 Legend — Best Response: Partial response; Stable disease; Non-complete response or non-progressive disease. Dose: 50 mg; 100 mg; 400 mg. Progression-free Survival (PFS): 0 to 4 mo; >4 to 8 mo; >8 to 12 mo; >12 to 16 mo. Type of Alteration: Copy number gain or amplification; Copy number loss; Rearrangement; Short variant KRAS G12C-selective small-molecule inhibitors: DIVARASIB locks KRAS in the OFF state, preventing SOS1 binding [structure image: DIVARASIB, GDP, KRAS-G12C (OFF)] Pan KRASi — Tri-complex inhibitors (TCIs): DARAXONRASIB binds cyclophilin A, forms a ternary complex with ON-state KRAS, and blocks effector access. [structure image: CypA, DARAXONRASIB, GTP, KRAS wt (ON)] Pan KRASi → pan RASi: RM-055 Markedly Stimulates Hydrolysis of RAS-GTP to RAS-GDP [G12V mutant RAS(OFF) / RAS(ON), hydrolysis, CYPA:TCI; plot KRAS G12V-GTP (% of control) vs Time (min): DMSO, Daraxonrasib, RM-055] Post-Progression Treatment (NSCLC, KRAS G12C)²: Daraxonrasib → RM-055 [Mean tumor volume (mm³) vs Days on study] KRAS G12Ci + pan KRASi + IO: Daraxonrasib 25 mg/kg po qd + Anti–PD-1 10 mg/kg ip biw — 3/10 CR; Elironrasib 30 mg/kf po qd + Daraxonrasib 25 mg/kg po qd + Anti–PD-1 10 mg/kg ip biw — 10/10 CR [tumor volume vs Days on study] Sacher NEJM 2023, Saemon AACR 2026, Way Cancer Disc 2026, images : the Computer-Aided Molecular Engineering Group Department of Fundamental Oncology Lausanne University. Benjamin BESSE

Patient demographics and baseline characteristics Characteristic, n (%), unless otherwise specified | Divarasib (n=167) | Control (n=171) Median age, years (range) | 66.0 (42–87) | 66.0 (39–89) Sex Female | 72 (43) | 75 (44) Sex Male | 95 (57) | 96 (56) ECOG PS 0 | 65 (39) | 68 (40) ECOG PS 1 | 102 (61) | 103 (60) Smoking status Never smoked | 4 (2) | 6 (4) Smoking status Current/former smoker | 163 (98) | 165 (96) Race White | 127 (76) | 135 (79) Race Asian | 17 (10) | 23 (13) Race Black or African American | 5 (3) | 1 (1) Race Other | 18 (11) | 12 (7) Geographic region Europe | 106 (63) | 103 (60) Geographic region North America | 28 (17) | 27 (16) Geographic region Rest of World | 33 (20) | 41 (24) Prior lines of therapy 1 | 121 (72) | 118 (69) Prior lines of therapy 2 | 41 (25) | 51 (30) Prior lines of therapy 3 | 5 (3) | 2 (1) Histology Adenocarcinoma | 158 (95) | 164 (96) Histology Squamous | 2 (1) | 4 (2) Histology Other | 7 (4) | 3 (2) KEAP1 mutation status* Detected | 35 (21) | 37 (22) KEAP1 mutation status* Not detected or unknown† | 132 (79) | 134 (78) PD-L1 status‡ <1% | 57 (34) | 70 (41) PD-L1 status‡ 1–<50% | 62 (37) | 65 (38) PD-L1 status‡ ≥50% | 43 (26) | 32 (19) History of CNS metastases*§ Yes | 48 (29) | 51 (30) History of CNS metastases*§ No | 119 (71) | 120 (70) Intended KRAS G12C inhibitor in control arm* Sotorasib | 153 (92) | 156 (91) Intended KRAS G12C inhibitor in control arm* Adagrasib | 14 (8) | 15 (9) Prior CPI status Prior CPI¶ | 165 (99) | 171 (100) Prior CPI status No prior CPI | 2 (1) | 0 Data cut-off: 19 May 2026. Median duration of follow up: 14.3 months. Minimal duration of follow up: 5.6 months. Control arm included treatment with sotorasib or adagrasib *Per IxRS. †Three patients had unknown KEAP1 status (all control arm). ‡PD-L1 testing was conducted locally. PD-L1 TC/TPS negative: <1%, PD-L1 TC/TPS low: 1–<50%, PD-L1 TC/TPS high: ≥50% §History of CNS metastases per eCRF: yes: n=48 for divarasib and n=54 for control; no: n=119 for divarasib and n=117 for control. ¶The timing of prior CPI for 1 patient is unknown CPI, checkpoint inhibitor; eCRF, electronic case report form; IxRS, interactive voice/web response system; TC, tumour cell; TPS, tumour proportion score Dr Ferdinandos Skoulidis

Progression-free survival by BICR PFS by BICR (%) Time (months) Patients at risk Divarasib Control 9.8 months 5.7 months 43.9% 24.0% Divarasib (n=167) | Control (n=171) Patients with event, n (%) 101 (60.5) | 128 ([illegible]) Median PFS, months (95% CI) 9.8 (8.1–[illegible]) | 5.7 (4.2–[illegible]) Stratified HR (95% CI) 0.58 (0.44–[illegible]) p-value* <0.0001 12-month PFS, % 43.9 | 24.0 Data cut-off: [illegible] May 2026. Median duration of follow up: [illegible] months. Minimal duration of follow up: [illegible] months. Control arm included treatment with sotorasib or adagrasib *Stratified log-rank test. CI, confidence interval Dr Ferdinandos Skoulidis Survival table Endpoint: PFS by BICR Arm | N | Median | Median CI Divarasib | 167 | 9.8 mo | 8.1-[illegible] Control (sotorasib or adagrasib) | 171 | 5.7 mo | 4.2-[illegible] HR: 0.58 (stratified) HR CI: 0.44-[illegible] P: <0.0001 12-month PFS 43.9% vs 24.0%

Krascendo 1: efficacy and safety of divarasib versus sotorasib or adagrasib in previously treated locally advanced or metastatic KRAS G12C-mutated NSCLC Ferdinandos Skoulidis,¹ Myung-Ju Ahn,² Shirish Gadgeel,³ Sanjay Popat,⁴ Niels Reinmuth,⁵ Bianca van Veggel,⁶ Farastuk Bozorgmehr,⁷ Victoria J Bray,⁸ Michael Duruisseaux,⁹ Wade T Iams,¹⁰ Julien Mazieres,¹¹ Laura Bonanno,¹² Lore Decoster,¹³ Bo Wang,¹⁴ Youyou Hu,¹⁵ Sarah M Shagan,¹⁶ Verena Lam,¹⁶ Uyi Osaghae,¹⁶ Tien Hoang,¹⁶ Benjamin J Solomon¹⁷ [affiliations small, partially illegible] ¹UT MD Anderson, Houston, TX, USA; ²Samsung Medical Center, Seoul, Republic of Korea; ³Emory University, Atlanta, GA, USA; ⁴Royal Marsden Hospital & Institute of Cancer Research, London, UK; ... ¹⁵F. Hoffmann-La Roche Ltd, Basel, Switzerland; ¹⁶Genentech Inc., South San Francisco, CA, USA; ¹⁷Peter MacCallum Cancer Centre, Melbourne, VIC, Australia

Progression-free survival by investigator* | Divarasib (n=167) | Control (n=171) Patients with event, n (%) | 98 (58.7) | 126 (73.7) Median PFS, months (95% CI) | 11.1 (8.8–14.1) | 5.6 (4.4–7.0) Stratified HR (95% CI) | 0.57 (0.43–0.75) 12-month PFS, % | 47.9 | 27.4 11.1 months 5.6 months 47.9% 27.4% Patients at risk Time (months) 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 Divarasib 167 | 144 | 105 | 90 | 58 | 29 | 3 | 0 Control 171 | 118 | 74 | 55 | 34 | 17 | 0 | 0 Investigator-assessed PFS was consistent with BICR-assessed PFS Data cut-off: 19 May 2026. Median duration of follow up: 14.3 months. Minimal duration of follow up: 5.6 months. Control arm included treatment with sotorasib or adagrasib *A sensitivity analysis was performed for PFS as determined by the investigator Dr Ferdinandos Skoulidis Survival table Endpoint: PFS by investigator (sensitivity analysis) Arm | N | Median | Median CI Divarasib | 167 | 11.1 | 8.8–14.1 Control (sotorasib or adagrasib) | 171 | 5.6 | 4.4–7.0 HR: 0.57 HR CI: 0.43–0.75 Landmark rates: 12 months: Divarasib 47.9% | Control 27.4%

[Slide 1] Divarasib / Control Percentage of Patients without an Event / Months Labels on curve: 65.7 / 47.8 at 6 months; 43.9 / 24.0 at 12 months | No. of Patients | No. of Patients with Event (%) | Median Progression-free Survival (mo) Divarasib | 167 | 101 (60.5) | 9.8 Control | 171 | 128 (74.9) | 5.7 Stratified hazard ratio for disease progression or death, 0.58 (95% CI, 0.44–0.76) Stratified P<0.001 by log-rank test No. at Risk Divarasib 167 142 108 80 53 27 3 0 Control 171 116 73 54 26 14 0 0 Survival table Endpoint: PFS Divarasib | 167 | 9.8 months Control | 171 | 5.7 months HR: 0.58 (stratified) HR CI: 0.44–0.76 P: <0.001 (stratified log-rank) 6 months: divarasib 65.7%, control 47.8% (printed on curve) 12 months: divarasib 43.9%, control 24.0% (printed on curve) [Slide 2] Percentage of Patients Alive Months Divarasib Control No. of Patients | No. of Patients Who Died (%) | Median Overall Survival (mo) Divarasib | 167 | 63 (37.7) | 17.7 Control | 171 | 93 (54.4) | 12.0 Stratified hazard ratio for death, 0.56 (95% CI, 0.41–0.78) Stratified P<0.001 by log-rank test No. at Risk Divarasib 167 160 142 120 95 41 6 0 Control 171 159 128 94 65 31 6 0 Survival table Endpoint: Overall survival Divarasib | 167 | 17.7 mo Control | 171 | 12.0 mo HR: 0.56 HR CI: 0.41–0.78 P: <0.001 (stratified log-rank)

Efficacy endpoints | Osi + savo (n = 171) | Plat–pem (n = 167) Median PFS by BICR, mo (95% CI) | 7.6 (5.6, 9.7) | 5.4 (4.2, 5.8) HR (95% CI); p-value | 0.55 (0.42, 0.72) p < 0.001 | — Median PFS by investigator assessment, mo (95% CI) | 7.4 (5.6, 9.7) | 4.2 (3.4, 4.5) HR (95% CI) | 0.47 (0.36, 0.61) | — Median OS, mo (95% CI) | 17.8 (15.6, 23.7) | 12.9 (10.6, 16.8) HR (95% CI); p-value | 0.68 (0.50, 0.91) p = 0.010 | — Median DoR by BICR, mo (95% CI) | 9.7 (7.6, 12.5) | 5.7 (4.4, 8.1)

[Slide 1] SAFFRON study design Key inclusion criteria • Aged ≥18 years (≥19 years in the Republic of Korea) • EGFRm (Ex19del, L858R, and/or T790M) locally advanced or metastatic NSCLC • MET overexpression (IHC3+≥90%) and/or amplification (FISH10+)* centrally confirmed with post-progression tissue biopsy • Progression on 1L or 2L osimertinib as the most recent anticancer treatment† • ECOG PS 0 / 1 • Stable CNS metastases were allowed N = 324, R 1:1 Stratified by: Race (Asian / non-Asian); Baseline brain metastases (Yes / no); Line of treatment (2L / 3L) Osimertinib 80 mg QD + savolitinib 300 mg BID‡ Treatment beyond PD allowed per investigator discretion¶ Pemetrexed 500 mg/m² + carboplatin AUC5 / cisplatin 75 mg/m² (Q3W for 4 cycles) → Maintenance pemetrexed 500 mg/m² (Q3W)‡ Investigator-assessed PD (per RECIST v1.1) or discontinuation Endpoints — Primary: PFS by BICR per RECIST v1.1 in the FAS Key secondary / exploratory: • PFS by investigator, OS, ORR, DoR, DCR and tumour shrinkage in the FAS • PFS in patients in the MET-overexpressed (IHC3+≥90%) or MET-amplified (FISH10+) analysis sets • Safety in the SAF *MET IHC and FISH cut-offs were identified in preliminary analyses of the Phase II SAVANNAH study.¹ IHC test: Investigational assay for MET (SP44; Roche Diagnostics, Tucson, AZ, USA); FISH test: investigational assay for MET (Vysis MET FISH Kit; Abbott Molecular, Inc., Des Plaines, IL, USA); †Patients who received osimertinib in the adjuvant setting were included if progression occurred <6 months after the last dose; ‡Patients received treatment until objective progressive disease by RECIST v1.1 as assessed by investigator, unacceptable toxicity, withdrawn consent or another discontinuation criterion was met; ¶Patients continued to receive osimertinib plus savolitinib or osimertinib monotherapy (if savolitinib was stopped earlier) or pemetrexed maintenance beyond RECIST v1.1-defined progressive disease (if eligible) if they continued to receive clinical benefit in the opinion of the investigator. Crossover between treatment arms was not allowed. 1. Ahn, et al. Clin Cancer Res 2026; Online ahead of print. [abbreviations: 1L, first-line; 2L, second-line; 3L, third-line; AUC; BICR; BID; CNS; DCR; DoR; ECOG PS; EGFRm; Ex19del; FAS, full analysis set; FISH10+, ≥10 MET gene copies; IHC3+≥90%, IHC 3+ staining intensity in ≥90% of tumour cells; ... SAF, safety analysis set] [Slide 2] Progression-free survival by BICR Median PFS, months (95% CI) Osimertinib + savolitinib | 7.6 (5.6, 9.7) Platinum–pemetrexed | 5.4 (4.2, 5.8) HR (95% CI); p-value | 0.55 (0.42, 0.72); p<0.001 Curve labels at 12 months: 36%; 10% No. at risk (Months since randomisation 0–38) Osimertinib + savolitinib | 171 166 144 124 120 98 82 75 70 64 54 49 39 34 29 25 24 22 21 14 14 8 8 6 5 2 1 1 1 1 1 1 1 1 1 1 1 1 0 | Number of patients / number of events 171 / 117 Platinum–pemetrexed | 167 139 109 87 77 61 43 34 30 22 17 12 8 7 7 4 4 3 3 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 | 167 / 113 Osimertinib plus savolitinib provided a statistically significant 45% reduction in the risk of disease progression or death compared with platinum–pemetrexed Data cut-off: 06 July 2026. Median follow-up for PFS by BICR: osimertinib + savolitinib 5.6 months, platinum–pemetrexed 3.5 months. Overall maturity: 68%. BICR, blinded independent central review; CI, confidence interval; HR, hazard ratio; PFS, progression-free survival. Survival table Endpoint: PFS by BICR (FAS) Arm | N | Median | Median CI Osimertinib + savolitinib | 171 | 7.6 | 5.6, 9.7 Platinum–pemetrexed | 167 | 5.4 | 4.2, 5.8 HR: 0.55 HR CI: 0.42, 0.72 P: <0.001 Landmark rates: 12 months: Osimertinib + savolitinib 36% | Platinum–pemetrexed 10% [Slide 3] Overall survival Median OS, months (95% CI) Osimertinib + savolitinib | 17.8 (15.6, 23.7) Platinum–pemetrexed | 12.9 (10.6, 16.8) HR (95% CI); p-value | 0.68 (0.50, 0.91); p = 0.010* Overall maturity: 53% Curve labels at 12 months: 70%; 53% No. at risk (Months since randomisation 0–38) Osimertinib + savolitinib | 171 170 166 162 157 152 144 138 133 130 119 114 98 93 89 74 63 59 51 46 43 39 31 27 22 17 14 14 13 9 7 6 5 5 3 2 1 1 0 | Number of patients / number of events 171 / 87 Platinum–pemetrexed | 167 154 150 142 137 134 128 119 113 96 86 70 62 51 48 40 36 32 27 23 21 20 20 19 18 18 16 12 11 9 7 4 4 3 2 2 1 1 0 | 167 / 93 Osimertinib plus savolitinib provided a statistically significant 32% reduction in the risk of death, with a 4.9 month improvement in median OS, compared with platinum–pemetrexed Data cut-off: 06 July 2026. Median follow-up for OS: osimertinib + savolitinib 14.2 months, platinum–pemetrexed 10.2 months. *Boundary to declare statistical significance: p<0.03047. CI, confidence interval; HR, hazard ratio; OS, overall survival. Survival table Endpoint: Overall survival Arm | N | Median | Median CI Osimertinib + savolitinib | 171 | 17.8 | 15.6, 23.7 Platinum–pemetrexed | 167 | 12.9 | 10.6, 16.8 HR: 0.68 HR CI: 0.50, 0.91 P: 0.010 (boundary p<0.03047) Landmark rates: 12 months: Osimertinib + savolitinib 70% | Platinum–pemetrexed 53% [Slide 4] Tumour response by BICR | Osimertinib + savolitinib (n = 171) | Platinum–pemetrexed (n = 167) | Odds ratio (95% CI) Patients with a confirmed objective response, n (%) | 88 (51) | 45 (27) | 2.90 (1.85, 4.61) Median DoR, months (95% CI) | 9.7 (7.6, 12.5) | 5.7 (4.4, 8.1) | Median best percentage change in target lesion size, % (min, max) | −44 (−100, 50) | −25 (−100, 35) | Patients with disease control,* n (%) | 145 (85) | 108 (65) | 3.18 (1.88, 5.50) [Waterfall plots: Best % change in target lesion size from baseline — Osimertinib + savolitinib; Platinum–pemetrexed] Best overall response: Complete response; Partial response; Stable disease; Progressive disease; Not evaluable Tumour response endpoints all favoured the osimertinib plus savolitinib arm Data cut-off: 06 July 2026. *Based on unconfirmed response; †Values imputed to 20%. BICR, blinded independent central review; CI, confidence interval; DoR, duration of response.

LBA6 - Osimertinib (osi) + savolitinib (savo) vs platinum–pemetrexed (plat–pem) in EGFRm MET-overexpressed (OE) and/or -amplified (AMP) advanced NSCLC post-osi: SAFFRON phase (Ph) 3 primary results Speakers Shun Lu (Shanghai, China) Lecture Time 16:54 - 17:06 Abstract Background Targeted treatments (txs) for EGFRm MET-OE and/or -AMP advanced NSCLC post-osi are limited. In SAVANNAH (Ph 2, NCT03778229), osi (3rd-gen, irreversible, CNS-active EGFR-TKI) + savo (oral, potent, highly selective, CNS-penetrant MET-TKI) demonstrated high and clinically meaningful response rates in this setting. We report primary results from the global Ph 3 SAFFRON study (NCT05261399), assessing efficacy and safety of osi + savo vs plat–pem in EGFRm MET-OE and/or -AMP advanced NSCLC post-osi. Methods Eligible pts: ≥18 years, EGFRm (Ex19del/L858R and/or T790M), MET-OE and/or -AMP (MET IHC3+ intensity in ≥90% of tumour cells and/or ≥10 MET gene copies by FISH) advanced NSCLC, post-1/-2L osi. Pts (randomised 1:1) received osi 80 mg QD + savo 300 mg BID or plat–pem (pem + carboplatin/cisplatin Q3W, 4 cycles, then maintenance pem Q3W) until RECIST 1.1 progression/discontinuation. Primary endpoint: PFS by BICR (RECIST 1.1); secondary/exploratory endpoints: OS, investigator-assessed PFS and safety. DCO: 06 Jul 2026. Results Overall, 338 pts were randomised: osi + savo n=171, plat–pem n=167. Osi + savo significantly improved PFS and OS vs plat–pem (PFS HR: 0.55; 95% CI 0.42, 0.72; p<0.001; OS HR: 0.68; 95% CI 0.50, 0.91; p=0.010; table). PFS benefit was consistent in the MET-OE and MET-AMP subgroups. All causality tx-emergent AEs (TEAEs): osi + savo 99% (Gr ≥3 64%); plat–pem 97% (Gr ≥3 63%). TEAEs leading to tx discontinuation: osi + savo 14%; plat–pem 8%. Serious TEAEs: osi + savo 41%; plat–pem 28%. Tx-related AEs: osi + savo 93% (Gr ≥3 40%); plat–pem 92% (Gr ≥3 52%). Conclusions Osi + savo showed statistically significant and clinically meaningful improvements in PFS and OS vs plat–pem, with a manageable safety profile, in pts with EGFRm MET-OE and/or -AMP advanced NSCLC post-osi. These data support osi + savo as an effective, oral, targeted tx for these pts and establish a new global standard of care. Table: LBA6 Efficacy endpoints | Osi + savo (n=171) | Plat–pem (n=167) Median PFS by BICR, mo (95% CI) | 7.6 (5.6, 9.7) | 5.4 (4.2, 5.8) HR (95% CI); p-value | 0.55 (0.42, 0.72); p<0.001 Median PFS by investigator assessment, mo (95% CI) | 7.4 (5.6, 9.7) | 4.2 (3.4, 4.5) HR (95% CI) | 0.47 (0.36, 0.61) Median OS, mo (95% CI) | 17.8 (15.6, 23.7) | 12.9 (10.6, 16.8) HR (95% CI); p-value | 0.68 (0.50, 0.91); p=0.010 Median DoR by BICR, mo (95% CI) | 9.7 (7.6, 12.5) | 5.7 (4.4, 8.1) BICR, blinded independent central review; CI, confidence interval; DoR, duration of response; HR, hazard ratio; OS, overall survival; PFS, progression-free survival.

[Slide 2] Are the results from SAFFRON clinically meaningful? YES Progression-free survival (BICR) Median PFS, months (95% CI) Osimertinib + savolitinib: 7.6 (5.6, 9.7) Platinum-pemetrexed: 5.4 (4.2, 5.8) HR 0.55 (95% CI 0.42-0.72); p<0.01 12-month: 36% vs 10% No. at Risk (months 0-38, every 2 months) Osimertinib + savolitinib: 171 144 120 82 70 54 39 29 24 21 14 8 5 1 1 1 1 1 1 0 Platinum pemetrexed: 167 109 77 43 30 17 8 7 4 3 1 0 0 0 0 0 0 0 0 0 Median follow-up PFS: osimertinib+savolitinib 5.6 months, platinum-pemetrexed 3.5 months. Overall maturity: 68%. Overall survival Median OS, months (95% CI) Osimertinib + savolitinib: 17.8 (15.6, 23.7) Platinum-pemetrexed: 12.9 (10.8, 16.8) HR 0.68; 95% CI, 0.50-0.91; p=0.010* 12-month: 70% vs 53% Subsequent therapy: Osi+Savo – 60%; 41% ChT; Chemo – 66%; 13% MET+EGFR TKI No. at Risk (months 0-38, every 2 months) Osimertinib + savolitinib: 171 166 157 144 133 119 98 89 63 51 43 31 22 14 13 7 5 3 1 0 Platinum pemetrexed: 167 150 137 128 113 86 62 48 36 27 21 20 18 16 11 7 4 2 1 0 Median follow-up (OS): osimertinib+savolitinib 14.2 months, platinum-pemetrexed 10.2 months. Overall maturity 53%. *Boundary to declare statistical significance p=0.03047. Objective response rate higher with osimertinib + savolitinib (51% v. 27%), odd ratio 2.9; Median duration of response 9.7 months v. 5.7 months Lu S et al. ESMO Congress. 2026 LBA6; Passaro A et al. Ann Oncol 2023; Popat S et al. Ann Oncol 2024; Fang W et al NEJM 2026 Natasha B. Leighl Survival table Endpoint: PFS (BICR) and OS Arm | N | Median | Median CI Osimertinib + savolitinib (PFS) | 171 | 7.6 | 5.6, 9.7 Platinum-pemetrexed (PFS) | 167 | 5.4 | 4.2, 5.8 Osimertinib + savolitinib (OS) | 171 | 17.8 | 15.6, 23.7 Platinum-pemetrexed (OS) | 167 | 12.9 | 10.8, 16.8 HR: PFS 0.55; OS 0.68 HR CI: PFS 0.42-0.72; OS 0.50-0.91 P: PFS p<0.01; OS p=0.010 (boundary p=0.03047) Landmark rates: 12 months: Osimertinib + savolitinib (PFS) 36% | Platinum-pemetrexed (PFS) 10% | Osimertinib + savolitinib (OS) 70% | Platinum-pemetrexed (OS) 53% [Slide 3] Have we MET a new standard of care? YES! - SAFFRON is the first study in osimertinib resistant EGFR mutated lung cancer to demonstrated improved PFS and OS (HR 0.55, 0.68) with an oral combination in a molecularly defined population (IHC3+>=90% or FISH10+) - Tomorrow we must start to ensure that each patient with resistance has molecular testing (tissue preferred) and is matched to the best possible therapy or clinical trials - Then we must ensure drug access, addressing issues of regulatory authorities and cost - MET IHC, MET FISH -> from tier ESCAT 2B to tier ESCAT 1p (proven) - SAFFRON MCBS score 3 -> to be incorporated into next guideline update [ESMO-MCBS scorecard thumbnail, Final Score 3; other text illegible] Natasha Leighl

[Slide 2] Resistance Patterns After Osimertinib Treatment On-target: EGFR Mut/amp — C797X, L792X, L718Q, G796X, G724S, Ex20ins, Off-target (EGFR-independent): Amplification (MET, HER2, FGFR) | Mutation (MET, HER2, FGFR) | Upregulation | Fusions MDM2 → TP53 → RB1 → Cyclin/CDK Downstream signaling pathways: JAK → STAT3; RAS (Mut/gain) → RAF (Mut/fusion) → MEK → ERK (Amp); PI3K (Mut) → AKT ⊣ PTEN (Del) → mTOR Cell proliferation + survival, apoptosis, angiogenesis, migration/invasion Cell survival, cell proliferation, cell invasion, drug resistance, EMT, histological transformation (i.e. SCLC, NEC, SCC) Pie chart: Small cell transformation 15%; 6–10%; 8–17%; 1–8%; 13–14%; 10%; 40–50%; C797X; L718Q; EGFRamp; G724S; S768I; METamp (7–15%) [boxed]; HER2amp (1–2%); SPTBN1-ALK; RET fusions; BRAF fusions; BRAFV600E (3%); PI3KCA (7%); KRAS (3–4%); HER2 (1%); CCND1amp; CCND2amp; CCNE1amp; CDK4amp; CDK6amp ELIOS: Paired plasma + tissue NGS in 51 patients with EGFR TKI resistance • 56% successful tissue NGS at PD • More treatable alterations in tissue (31% v 13% plasma) • Higher if both tissue, plasma used | Tissue | Plasma EGFR C797S | 13% | 28% MET amplification | 31% | 13% RB1 | 16% | 3% TP53 | 81% | 84% MTAP/CDKN2A loss | 22-25% | -- Mut, mutation; Amp, Amplification; Upreg, Upregulation; Overexp, Overexpression; Activ, Activation; Made in Biorender.com Schmid S, Li JJN, Leighl N. Lung Cancer. 2020;147:123-129; Adapted from Ferro A et al. CROH. 2024 Leonetti et al Br J Cancer 2019; Piotrowska et al. Cancer Disc 2026 Natasha B. Leighl [Slide 3] Is SAFFRON well-designed with valid results? ✅ YES and ✅ YES SAFFRON study design Key inclusion criteria • Aged ≥18 years (≥19 years in the Republic of Korea) • EGFRm (Ex19del, L858R, and/or T790M) locally advanced or metastatic NSCLC • MET overexpression (MET IHC3+≥90%) and/or amplification (FISH10+)* centrally confirmed with post-progression tissue biopsy • Progression on 1L or 2L osimertinib as the most recent anti-cancer treatment† • ECOG PS 0/1 • Stable CNS metastases were allowed N=324 R 1:1; Stratified by: Race (Asian/non-Asian); Baseline brain metastases (Yes/no); Line of treatment (2L/3L) Osimertinib 80 mg QD + savolitinib 300 mg BID‡ Treatment beyond PD allowed per investigator discretion¶ Pemetrexed 500 mg/m2 + carboplatin AUC5/ cisplatin 75 mg/m2 (Q3W for 4 cycles) → Maintenance pemetrexed 500 mg/m2 (Q3W)‡ Investigator-assessed PD (per RECIST v1.1) or discontinuation → 28-day safety follow-up visit** → Survival follow-up Endpoints Primary: PFS by BICR per RECIST v1.1 in the FAS Key secondary/exploratory: • PFS by investigator, OS, ORR, DoR, DCR and tumour shrinkage in the FAS • PFS in patients in the MET-overexpressed (MET IHC3+≥90%) or MET-amplified (FISH10+) analysis sets • Safety in the SAF Patients screened, N = 1746*; Patients randomised, N = 338 Osimertinib + savolitinib, n = 171; Received treatment, n = 170 (99%); Ongoing any treatment at DCO, n = 45 (26%); Discontinued any treatment, n = 127 (75%)† Platinum–pemetrexed, n = 167; Received treatment, n = 145 (87%); Ongoing any treatment at DCO, n = 3 (2%); Discontinued any treatment, n = 142 (98%)† 0 v 8% withdrew post randomization 99% v 87% started protocol therapy Patient withdrawal 6% v 13% Lost to follow up 0 v 1% ✓ Molecular selection, dual targeting strategy informed by prior studies1-4 ✓ Is the control arm the current standard of care? (it was…) ✓ Randomized, power 90%, intent-to-treat-analysis, no crossover ✓ Unblinded but primary endpoint PFS by BICR ✓ PFS primary endpoint acceptable, (OS preferred) ✓ Patient characteristics balanced (1/3 male, 2/3 Asian, 31% CNS metastases, 16% post 2 TKIs) ✓ Disproportionate withdrawals in control arm ✓ Fixed sequence hierarchical testing appropriate for multiple testing (Type 1 error controlled) – IHC+ subgroup prioritized→ relevant for implementation Lu S et al. ESMO Congress. 2026 LBA6

[Slide 1] Proffered Paper session LBA6 - Osimertinib (osi) + savolitinib (savo) vs platinum–pemetrexed (plat–pem) in EGFRm MET-overexpressed (OE) and/or -amplified (AMP) advanced NSCLC post-osi: SAFFRON phase (Ph) 3 primary results Speakers Shun Lu (Shanghai, China) Lecture Time 16:54 - 17:06 Abstract Background Targeted treatments (txs) for EGFRm MET-OE and/or -AMP advanced NSCLC post-osi are limited. In SAVANNAH (Ph 2, NCT03778229), osi (3rd-gen, irreversible, CNS-active EGFR-TKI) + savo (oral, potent, highly selective, CNS-penetrant MET-TKI) demonstrated high and clinically meaningful response rates in this setting. We report primary results from the global Ph 3 SAFFRON study (NCT05261399), assessing efficacy and safety of osi + savo vs plat–pem in EGFRm MET-OE and/or -AMP advanced NSCLC post-osi. [Slide 2] Osi + savo showed statistically significant and clinically meaningful improvements in PFS and OS vs plat–pem, with a manageable safety profile, in pts with EGFRm MET-OE and/or -AMP advanced NSCLC post-osi. These data support osi + savo as an effective, oral, targeted tx for these pts and establish a new global standard of care. Table: LBA6 Efficacy endpoints | Osi + savo (n=171) | Plat–pem (n=167) Median PFS by BICR, mo (95% CI) | 7.6 (5.6, 9.7) | 5.4 (4.2, 5.8) HR (95% CI); p-value | 0.55 (0.42, 0.72); p<0.001 | Median PFS by investigator assessment, mo (95% CI) | 7.4 (5.6, 9.7) | 4.2 (3.4, 4.5) HR (95% CI) | 0.47 (0.36, 0.61) | Median OS, mo (95% CI) | 17.8 (15.6, 23.7) | 12.9 (10.6, 16.8) HR (95% CI); p-value | 0.68 (0.50, 0.91); p=0.010 | Median DoR by BICR, mo (95% CI) | 9.7 (7.6, 12.5) | 5.7 (4.4, 8.1) BICR, blinded independent central review; CI, confidence interval; DoR, duration of response; HR, hazard ratio; OS, overall survival; PFS, progression-free survival. [Slide 3] Conclusions Osi + savo showed statistically significant and clinically meaningful improvements in PFS and OS vs plat–pem, with a manageable safety profile, in pts with EGFRm MET-OE and/or -AMP advanced NSCLC post-osi. These data support osi + savo as an effective, oral, targeted tx for these pts and establish a new global standard of care. Table: LBA6 Efficacy endpoints | Osi + savo (n=171) | Plat–pem (n=167) Median PFS by BICR, mo (95% CI) | 7.6 (5.6, 9.7) | 5.4 (4.2, 5.8) HR (95% CI); p-value | 0.55 (0.42, 0.72); p<0.001 | Median PFS by investigator assessment, mo (95% CI) | 7.4 (5.6, 9.7) | 4.2 (3.4, 4.5) HR (95% CI) | 0.47 (0.36, 0.61) | Median OS, mo (95% CI) | 17.8 (15.6, 23.7) | 12.9 (10.6, 16.8) HR (95% CI); p-value | 0.68 (0.50, 0.91); p=0.010 | Median DoR by BICR, mo (95% CI) | 9.7 (7.6, 12.5) | 5.7 (4.4, 8.1) BICR, blinded independent central review; CI, confidence interval; DoR, duration of response; HR, hazard ratio; OS, overall survival; PFS, progression-free survival. [Slide 4] Methods Eligible pts: ≥18 years, EGFRm (Ex19del/L858R and/or T790M), MET-OE and/or -AMP (MET IHC3+ intensity in ≥90% of tumour cells and/or ≥10 MET gene copies by FISH) advanced NSCLC, post-1/-2L osi. Pts (randomised 1:1) received osi 80 mg QD + savo 300 mg BID or plat–pem (pem + carboplatin/cisplatin Q3W, 4 cycles, then maintenance pem Q3W) until RECIST 1.1 progression/discontinuation. Primary endpoint: PFS by BICR (RECIST 1.1); secondary/exploratory endpoints: OS, investigator-assessed PFS and safety. DCO: 06 Jul 2026. Results Overall, 338 pts were randomised: osi + savo n=171, plat–pem n=167. Osi + savo significantly improved PFS and OS vs plat–pem (PFS HR: 0.55; 95% CI 0.42, 0.72; p<0.001; OS HR: 0.68; 95% CI 0.50, 0.91; p=0.010; table). PFS benefit was consistent in the MET-OE and MET-AMP subgroups. All causality tx-emergent AEs (TEAEs): osi + savo 99% (Gr ≥3 64%); plat–pem 97% (Gr ≥3 63%). TEAEs leading to tx discontinuation: osi + savo 14%; plat–pem 8%. Serious TEAEs: osi + savo 41%; plat–pem 28%. Tx-related AEs: osi + savo 93% (Gr ≥3 40%); plat–pem 92% (Gr ≥3 52%).

[Slide 2] 16:30 - 18:15 Presidential Symposium II CHAIRS: GIUSEPPE CURIGLIANO, YELENA JANJIGIAN Shun Lu - Osimertinib (osi) + savolitinib (savo) vs platinum–pemetrexed (plat–pem) in EGFRm MET-overexpressed (OE) and/or -amplified (AMP) advanced NSCLC post-osi: SAFFRON phase (Ph) 3 primary results ALICANTE AUDITORIUM - HALL 6 Progression-free survival by BICR Median PFS, months (95% CI) Osimertinib + savolitinib | 7.6 (5.6, 9.7) Platinum–pemetrexed | 5.4 (4.2, 5.8) HR (95% CI); p-value | 0.55 (0.42, 0.72); p<0.001 12-month labels on curve: 36% | 10% Months since randomisation No. at risk Osimertinib + savolitinib: 171 166 144 124 120 98 82 75 70 64 54 49 39 34 29 25 24 22 21 14 14 8 8 6 5 2 1 1 1 1 1 1 1 1 1 1 1 1 0 | 171 / 117 Platinum–pemetrexed: 167 139 109 87 77 61 43 34 30 22 17 12 8 7 7 4 4 3 3 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 | 167 / 113 (Number of patients / number of events) Osimertinib plus savolitinib provided a statistically significant 45% reduction in the risk of disease progression or death compared with platinum–pemetrexed Data cut-off: 06 July 2026. Median follow-up for PFS by BICR: osimertinib + savolitinib 5.6 months, platinum–pemetrexed 3.5 months. Overall maturity: 68%. BICR, blinded independent central review; CI, confidence interval; HR, hazard ratio; PFS, progression-free survival. Survival table Endpoint: PFS by BICR Arm | N | Median | Median CI Osimertinib + savolitinib | 171 | 7.6 | 5.6, 9.7 Platinum–pemetrexed | 167 | 5.4 | 4.2, 5.8 HR: 0.55 HR CI: 0.42, 0.72 P: p<0.001 Landmark rates: 12 months: Osimertinib + savolitinib 36% | Platinum–pemetrexed 10% [Slide 3] 16:30 - 18:15 Presidential Symposium II CHAIRS: GIUSEPPE CURIGLIANO, YELENA JANJIGIAN Shun Lu - Osimertinib (osi) + savolitinib (savo) vs platinum–pemetrexed (plat–pem) in EGFRm MET-overexpressed (OE) and/or -amplified (AMP) advanced NSCLC post-osi: SAFFRON phase (Ph) 3 primary results ALICANTE AUDITORIUM - HALL 6 Overall survival Median OS, months (95% CI) Osimertinib + savolitinib | 17.8 (15.6, 23.7) Platinum–pemetrexed | 12.9 (10.6, 16.8) HR (95% CI); p-value | 0.68 (0.50, 0.91); p = 0.010* Overall maturity: 53% 12-month labels on curve: 70% | 53% Months since randomisation No. at risk Osimertinib + savolitinib: 171 170 166 162 157 152 144 138 133 130 119 114 98 93 89 74 63 59 51 46 43 39 31 27 22 17 14 14 13 9 7 6 5 5 3 2 1 1 0 | 171 / 87 Platinum–pemetrexed: 167 154 150 142 137 134 128 119 113 96 86 70 62 51 48 40 36 32 27 23 21 20 20 19 18 18 16 12 11 9 7 4 4 3 2 2 1 1 0 | 167 / 93 (Number of patients / number of events) Osimertinib plus savolitinib provided a statistically significant 32% reduction in the risk of death, with a 4.9 month improvement in median OS, compared with platinum–pemetrexed Data cut-off: 06 July 2026. Median follow-up for OS: osimertinib + savolitinib 14.2 months, platinum–pemetrexed 10.2 months. *Boundary to declare statistical significance: p<0.03047. CI, confidence interval; HR, hazard ratio; OS, overall survival. Survival table Endpoint: Overall survival Arm | N | Median | Median CI Osimertinib + savolitinib | 171 | 17.8 | 15.6, 23.7 Platinum–pemetrexed | 167 | 12.9 | 10.6, 16.8 HR: 0.68 HR CI: 0.50, 0.91 P: p = 0.010 (boundary p<0.03047) Landmark rates: 12 months: Osimertinib + savolitinib 70% | Platinum–pemetrexed 53% [Slide 4] 16:30 - 18:15 Presidential Symposium II CHAIRS: GIUSEPPE CURIGLIANO, YELENA JANJIGIAN Shun Lu - Osimertinib (osi) + savolitinib (savo) vs platinum–pemetrexed (plat–pem) in EGFRm MET-overexpressed (OE) and/or -amplified (AMP) advanced NSCLC post-osi: SAFFRON phase (Ph) 3 primary results ALICANTE AUDITORIUM - HALL 6 Safety summary | Osimertinib + savolitinib (n = 170) | Platinum–pemetrexed (n = 145) Median total duration of exposure, months (range) | 8.5 (0.7–37.9) | 3.5 (0–20.3) TEAEs (all causality), n (%) | 169 (99) | 140 (97) Grade ≥3 | 108 (64) | 91 (63) Serious | 69 (41) | 40 (28) Fatal | 12 (7)* | 0 Leading to discontinuation of any treatment | 23 (14) | 12 (8) Leading to dose modification | 104 (61) | 64 (44) TRAEs,† n (%) | 158 (93) | 134 (92) Grade ≥3 | 68 (40) | 75 (52) Serious | 24 (14) | 23 (16) Fatal | 1 (1) | 0 Leading to discontinuation of any treatment | 21 (12) | 9 (6) Leading to dose modification | 82 (48) | 55 (38) [footnotes illegible]

[Slide 2] 16:30 - 18:15 Presidential Symposium II CHAIRS: GIUSEPPE CURIGLIANO, YELENA JANJIGIAN Is SAFFRON well-designed with valid results? ✅ YES and ✅ YES SAFFRON study design [embedded schematic small; see clean copy 2114392179212951939_s3] 0 v 8% withdrew post randomization 99% v 87% started protocol therapy Patient withdrawal 6% v 13% Lost to follow up 0 v 1% ✓ Molecular selection, dual targeting strategy informed by prior studies1-4 ✓ Is the control arm the current standard of care? (it was…) ✓ Randomized, power 90%, intent-to-treat-analysis, no crossover ✓ Unblinded but primary endpoint PFS by BICR ✓ PFS primary endpoint acceptable, (OS preferred) ✓ Patient characteristics balanced (1/3 male, 2/3 Asian, 31% CNS metastases, 16% post 2 TKIs) ✓ Disproportionate withdrawals in control arm ✓ Fixed sequence hierarchical testing appropriate for multiple testing (Type 1 error controlled) – IHC+ subgroup prioritized→ relevant for implementation Lu S et al. ESMO Congress. 2026 LBA6 Natasha Leighl Invited Discussant LBA6 ALICANTE AUDITORIUM - HALL 6 [Slide 4] 16:30 - 18:15 Presidential Symposium II CHAIRS: GIUSEPPE CURIGLIANO, YELENA JANJIGIAN Resistance post Combination Therapy → Different Biology? (Not for cross trial comparison) FLAURA2 - no new resistance mechanisms, reduction in emergent mutations, fusions (ctDNA) Osimertinib + Chemo vs Osimertinib: Acq EGFR mt 4% vs 14%; MET amplification 12% vs 11%; Her2 amplification 4% vs 1%; PI3K 7% vs 6; RAS/RAF 4% vs 13%; Other RTK 4% vs 7%; Fusions 8% vs 15%; Cell Cycle 13% vs 10%; TP53/RB1 loss 3% vs 4% MARIPOSA - no new resistance mechanisms, reduction in MET, EGFR driven resistance (ctDNA) Osimertinib (n=198) vs Amivantamab+Lazertinib (n=148): MET amplification 13.1% vs 3.4% (P=0.002); Secondary EGFR resistance mutations (C797S, L718X, G724X) 7.6% vs 1.4% (P=0.01); HER2 amplification 4.5% vs 8.8%; PI3K 9.1% vs 8.1%; RAS/RAFa 11.6% vs 10.1%; Other RTK 4.5% vs 2.0%; Cell cycleb 9.6% vs 12.8%; TP53/RB1 loss 2.5% vs 1.4% aIncludes BRAF, KRAS, NRAS, PTPN11, and RAF1. bIncludes CCNE1, CDK4, CDKN2A, CCND2, CDK6, and CCND1. Note: Differences between other pathways were not statistically significant. Figure created from Yang JC et al WCLC 2024; Hayashi et al. J Thorac. Oncol. 2026:103894. Natasha B. Leighl Natasha Leighl Invited Discussant LBA6 ALICANTE AUDITORIUM - HALL 6

[Slide 1] 16:30 - 18:15 Presidential Symposium II CHAIRS: GIUSEPPE CURIGLIANO, YELENA JANJIGIAN Conclusions • SAFFRON is the first global phase 3 study showing statistically significant and clinically meaningful improvements in PFS and OS with osimertinib plus savolitinib versus platinum–pemetrexed, after progression on osimertinib, in EGFRm MET-overexpressed and/or -amplified advanced NSCLC PFS HR 0.55; 95% CI 0.42, 0.72; p<0.001 | OS HR 0.68; 95% CI 0.50, 0.91; p = 0.010 • The PFS benefit with osimertinib plus savolitinib versus platinum–pemetrexed was consistent across prespecified subgroups • The combination demonstrated a tolerable safety profile, which was consistent with the established profiles for each treatment1–5 • SAFFRON establishes the clinical value of an all-oral, biomarker-directed treatment approach targeting both EGFR and MET following progression on osimertinib, reinforcing osimertinib as a backbone treatment across lines of therapy Osimertinib plus savolitinib establishes a new global standard of care for patients with EGFRm advanced NSCLC with MET overexpression and/or amplification after progression on osimertinib [footnote references illegible] Shun Lu Osimertinib (osi) + savolitinib (savo) vs platinum–pemetrexed (plat–pem) in EGFRm MET-overexpressed (OE) and/or -amplified (AMP) advanced NSCLC post-osi: SAFFRON phase (Ph) 3 primary results ALICANTE AUDITORIUM - HALL 6 [Slide 3] 16:30 - 18:15 Presidential Symposium II CHAIRS: GIUSEPPE CURIGLIANO, YELENA JANJIGIAN Patient disposition 35% of patients evaluable for MET testing had MET overexpression (IHC3+≥90%) and/or MET amplification (FISH10+) Patients screened, N = 1746* Patients randomised, N = 338 Osimertinib + savolitinib, n = 171 | Platinum–pemetrexed, n = 167 Received treatment, n = 170 (99%)† | Received treatment, n = 145 (87%)† Ongoing any treatment at DCO, n = 45 (26%)‡ | Ongoing any treatment at DCO, n = 3 (2%)‡ Osimertinib + savolitinib, n = 43 (25%); Savolitinib, n = 43 (25%); Osimertinib, n = 45 (26%) | Pemetrexed, n = 3 (2%) Discontinued any treatment, n = 127 (75%)‡¶ | Discontinued any treatment, n = 142 (98%)‡ Progressive disease, n = 89 (52%) | Progressive disease, n = 110 (76%) Adverse event, n = 25 (15%) | Patient decision, n = 15 (10%) Death, n = 9 (5%) | Adverse event, n = 13 (9%) Patient decision, n = 7 (4%) | Consent withdrawal, n = 3 (2%) Other, n = 4 (2%) | Death, n = 1 (1%) Consent withdrawal, n = 2 (1%) Data cut-off 06 July 2026. [remaining footnotes illegible] Shun Lu Osimertinib (osi) + savolitinib (savo) vs platinum–pemetrexed (plat–pem) in EGFRm MET-overexpressed (OE) and/or -amplified (AMP) advanced NSCLC post-osi: SAFFRON phase (Ph) 3 primary results ALICANTE AUDITORIUM - HALL 6 [Slide 4] 16:30 - 18:15 Presidential Symposium II CHAIRS: GIUSEPPE CURIGLIANO, YELENA JANJIGIAN First subsequent treatment Platinum-based chemotherapy was the most common FST in the osimertinib plus savolitinib arm Non-platinum-based chemotherapy was the most common FST in the platinum–pemetrexed arm Osimertinib + savolitinib (n = 126)*: No FST (n = 51); FST (n = 75); 60% patients received FST† Received FST‡ bar: 33% | 5% | 15% | 5% | 41% Platinum–pemetrexed (n = 164)*: No FST (n = 56); FST (n = 108); 66% patients received FST† Received FST‡ bar: 23% | 19% | 14% | 33% | 10% Legend: Platinum-based chemotherapy | Non-platinum-based chemotherapy | EGFR-TKI | MET-TKI | Other¶ Data cut-off: 06 July 2026. [remaining footnotes illegible] Shun Lu Osimertinib (osi) + savolitinib (savo) vs platinum–pemetrexed (plat–pem) in EGFRm MET-overexpressed (OE) and/or -amplified (AMP) advanced NSCLC post-osi: SAFFRON phase (Ph) 3 primary results ALICANTE AUDITORIUM - HALL 6

A Progression-free Survival in the Full Analysis Population No. of Events/No. of Patients | Median Progression-free Survival (95% CI) mo Osimertinib–Savolitinib 117/171 7.6 (5.6–9.7) Platinum–Pemetrexed 113/167 5.4 (4.2–5.8) Hazard ratio for disease progression or death, 0.55 (95% CI, 0.42–0.72) P<0.001 B Progression-free Survival in the MET-Overexpressed Analysis Population Osimertinib–Savolitinib 77/124 8.5 (5.7–11.1) Platinum–Pemetrexed 80/118 5.6 (4.4–6.7) Hazard ratio for disease progression or death, 0.51 (95% CI, 0.37–0.71) P<0.001 C Progression-free Survival in the MET-Amplified Analysis Population Osimertinib–Savolitinib 89/129 7.6 (5.6–9.7) Platinum–Pemetrexed 79/116 4.2 (3.1–5.5) Hazard ratio for disease progression or death, 0.48 (95% CI, 0.35–0.66) Percentage of Patients Months since Randomization No. at Risk Osimertinib–savolitinib Platinum–pemetrexed Survival table Endpoint: PFS (full analysis population; panel A of 3) Arm | N | Median | Median CI Osimertinib–savolitinib | 171 | 7.6 | 5.6–9.7 Platinum–pemetrexed | 167 | 5.4 | 4.2–5.8 HR: 0.55 HR CI: 0.42–0.72 P: P<0.001 12-month PFS 36 vs 10 (label on curve, %)

[Slide 2] How does MET amplification relate to protein expression? polysomy | amplification | gene copy number 5 | MET/CEP7 ratio 1 | MET/CEP7 ratio 5 FISH: MET/CEP7 ≥2 → Focal → MET amplification positive; MET/CEP7 <2 → MET CN/Tumor cell ≥5 → Polysomy → MET amplification positive; MET CN/Tumor cell <5 → Negative chr7: CEP7 probe; EGFR; CDK6; MET probe; MET; BRAF NGS: KL(MET,BRAF) ≥7 AND KL(MET,EGFR/CDK6) ≥7 → Focal → MET amplification positive; KL(MET,BRAF) <7 OR KL(MET,EGFR/CDK6) <7 → GCN ≥5 → Polysomy; GCN <5 → Negative c-MET: H-score < 200 (negative); H-score ≥ 200 (positive) Venn: NGS | FISH | IHC — 1, 4, 14, 73, 22, 10, 21 E | FISH + | FISH - | Concordance NGS + | 87 | 5 | PPA: 66.9% NGS - | 43 | 96 | NPA: 95.1%; OPA: 79.2% IHC (c-MET) + | 94 | 14 | PPA: 72.9% IHC (c-MET) - | 35 | 87 | NPA: 86.1%; OPA: 78.7% MET alteration type | Description | Detection methods | Challenges MET Amplification | Results from polysomy and local gene amplification | NGS, FISH, qRT-PCR | Definitive amplification thresholds challenging to establish METex14 Skipping | Induced by mutations disrupting the regulatory juxtamembrane domain | Sanger Sequencing, RNA-based NGS, RT-PCR | / MET Fusions | Arise from chromosomal rearrangements, often involving exon 15 and kinase domains | Hybridization-based technologies, AMP in RNA-NGS, FISH, qRT-PCR | DNA-NGS is less sensitive for detecting fusions MET Overexpression | Associated with ligand-independent phosphorylation and aberrant signaling activation | IHC | Inconsistent truncation point settings for MET overexpression AMP, Anchored Multiplex PCR; FISH, Fluorescence in Situ Hybridization; IHC, immunohistochemical; NGS, Next-Generation Sequencing; qRT-PCR, Quantitative Reverse Transcription Polymerase Chain Reaction. Xiang et al Mod Path 2024; Cappuzzo et al J Clin Oncol 2009; Tsao et al 2025; Guo et al NRCO 2020

[Slide 1] 1L Options For EGFRm NSCLC (Not for cross-comparison purposes) FLAURA1,2 (Osimertinib): mPFS 18.9; mOS 38.6 FLAURA3-5 (Osimertinib+ChT): mPFS 25.5; mOS 47.5 MARIPOSA6,7 (Amivantamab +Lazertinib): mPFS 23.7; mOS Not Reached mPFS (months) | mOS (months) ESMO Guidelines8 Stage IV mNSCLC with EGFR-activating mutation [ESCAT I-A; ESCAT I-B for uncommon EGFR mutations]9 PS 0-2 [I, A]; PS 3-4 for all following options [III, A] Osimertinib–pemetrexed–platinum-based doublet ChT [I, A; MCBS 3] Lazertinib–amivantamab-vmjw [I, A; MCBS 4] EGFR TKI monotherapy [I, A-B; MCBS 3-4] Erlotinib–anti-VEGF(R) Ab [I, B; MCBS 1-2] Gefitinib–carboplatin–pemetrexed [I, B] Disease progression Oligoprogression → Local treatment (surgery or RT) and continue targeted systemic treatment [IV, C] Systemic progression 1. Soria JC et al New Engl J Med 2018;378:113-125; 2. Ramalingam S et al New Engl J Med 2020;382:41-50; 3. Janne PA et al. ESMO Congress. 2025; 4. Janne PA et al. WCLC. 2023; 5. Planchard D et al. WCLC. 2025; 6. Yang JCH et al. ELCC 2025; 7. Cho BC, et al., ESMO. 2023; 8. Hendriks L et al. Ann Oncol 2023 (ESMO Oncogene Addicted Metastatic NSCLC Living Guideline v1.3, Feb 2026); 9. Elamin YY et al. ESMO 2025 NCT03410043 Natasha B. Leighl [Slide 2] MET amplification - a common mechanism of resistance in oncogene addicted lung cancer EGFR — EGF — EGFR-TKI → MET ← ALK — ALK-TKI RAS → RAF → MEK → ERK; PI3K → AKT → mTOR Targets involved in: · Survival · Proliferation · Cell-cycle progression Donut chart: ~15%; METex14 3%; METamp; METamp-high 1%; EGFR 19%; ALK 3%; RET 1%; ROS1 1%; NTRK 0.2%; KRAS 34%; BRAF 3%; Others 35%; METamp; METamp; METamp Coleman N, et al. ESMO Open 2021; Qin K, et al. Cancers 2023 [Slide 4] Is MET Overexpression Enough or Do We Need MET Addiction? PFS for MET-overexpressed analysis set Median PFS months (95% CI): Osimertinib + savolitinib 8.5 (5.7 [cut off]); Platinum–pemetrexed 5.6 (4.4 [cut off]) HR 0.51 (95% CI 0.37-0.71); p<0.[cut off] 40% / 11% (at 12 months) Randomised patients N=338 (100%) FAS IHC90+ n=242 (71.6%) | FAS FISH10+ n=245 (72.5%) IHC90+ only n=91 (27%) | IHC90+ and FISH10+ n=151 (45%) | FISH10+ only n=94 (28%) IHC90- and FISH10- n=2 (0.5%) PFS for MET-amplified analysis set* Median PFS, months (95% CI): Osimertinib + savolitinib 7.6 (5.6, 9.7); Platinum–pemetrexed 4.2 (3.1, 5.5) HR 0.48 (95% CI 0.35-0.66) *No formal statistical testing was carried out for the MET-amplified analysis set. 33% / 8% (at 12 months) No. at Risk (MET-overexpressed): Osimertinib + savolitinib 124 106 92 64 55 40 29 22 17 16 10 6 4 1 1 1 1 1 1 0; Platinum–pemetrexed 118 77 63 33 23 13 7 6 3 2 1 0 0 0 0 0 0 0 0 0 No. at Risk (MET-amplified): [illegible] Lu S et al. ESMO Congress. 2026; Courtesy J. Pinchevsky, AstraZeneca (internal data) Natasha B. Leighl Survival table Endpoint: PFS (BICR) in MET-overexpressed and MET-amplified analysis sets MET-OE set: Osimertinib + savolitinib | 124 | 8.5 MET-OE set: Platinum–pemetrexed | 118 | 5.6 MET-amp set: Osimertinib + savolitinib | 7.6 | 5.6, 9.7 MET-amp set: Platinum–pemetrexed | 4.2 | 3.1, 5.5 HR: MET-OE 0.51; MET-amp 0.48 HR CI: MET-OE 0.37-0.71; MET-amp 0.35-0.66 P: MET-OE p<0.[cut off by Venn overlay]; MET-amp no formal testing Landmark rates: 12 months: MET-OE: Osimertinib + savolitinib 40% | MET-OE: Platinum–pemetrexed 11% | MET-amp: Osimertinib + savolitinib 33% | MET-amp: Platinum–pemetrexed 8%

Progression-free survival by BICR: MET subgroups MET-overexpressed analysis set Median PFS, months (95% CI) Osimertinib + savolitinib | 8.5 (5.7, 11.1) Platinum–pemetrexed | 5.6 (4.4, 6.7) HR (95% CI); p-value | 0.51 (0.37, 0.71); p<0.001 12-month labels on curve: 40% | 11% No. at risk / number of events: [illegible] ... 127 / 77; [illegible] ... 118 / 80 MET-amplified analysis set* Median PFS, months (95% CI) Osimertinib + savolitinib | 7.6 (5.6, 9.7) Platinum–pemetrexed | 4.2 (3.1, 5.5) HR (95% CI) | 0.48 (0.35, 0.66) 12-month labels on curve: 33% | [illegible] No. at risk / number of events: [illegible] ... 129 / [illegible]; [illegible] ... 116 / 79 PFS by BICR in the MET-overexpressed and MET-amplified analysis sets was generally consistent with the full analysis set Data cut-off: 06 July 2026. *No formal statistical testing was carried out for the MET-amplified analysis set. Survival table Endpoint: PFS by BICR (MET-overexpressed analysis set; MET-amplified analysis set) Osimertinib + savolitinib (MET-overexpressed) | 8.5 | 5.7, 11.1 Platinum–pemetrexed (MET-overexpressed) | 5.6 | 4.4, 6.7 Osimertinib + savolitinib (MET-amplified) | 7.6 | 5.6, 9.7 Platinum–pemetrexed (MET-amplified) | 4.2 | 3.1, 5.5 HR: MET-overexpressed 0.51; MET-amplified 0.48 HR CI: 0.37, 0.71; 0.35, 0.66 P: MET-overexpressed p<0.001; MET-amplified no formal testing Landmark rates: 12 months: Osimertinib + savolitinib (MET-overexpressed) 40% | Platinum–pemetrexed (MET-overexpressed) 11% | Osimertinib + savolitinib (MET-amplified) 33%

Progression-free survival by BICR across subgroups Subgroup | n, events / patients: Osimertinib + savolitinib | Platinum–pemetrexed | PFS HR (95% CI) All patients | 117 / 171 | 113 / 167 | 0.55 (0.42, 0.72) Age <65 years | 63 / 89 | 56 / 91 | 0.65 (0.45, 0.94) Age ≥65 years | 54 / 82 | 57 / 76 | 0.45 (0.31, 0.66) Sex Male | 45 / 61 | 43 / 58 | 0.53 (0.35, 0.81) Sex Female | 72 / 110 | 70 / 109 | 0.56 (0.40, 0.78) ECOG PS 0 | 38 / 60 | 35 / 49 | 0.40 (0.25, 0.64) ECOG PS 1 | 79 / 111 | 78 / 118 | 0.65 (0.47, 0.89) Race Asian | 72 / 107 | 75 / 110 | 0.57 (0.41, 0.79) Race Non-Asian | 45 / 64 | 38 / 57 | 0.50 (0.32, 0.78) Line of therapy Second-line | 96 / 143 | 91 / 141 | 0.57 (0.42, 0.76) Line of therapy Third-line | 21 / 28 | 22 / 25 | 0.45 (0.25, 0.84) Baseline brain metastases Yes | 38 / 53 | 31 / 51 | 0.50 (0.31, 0.82) Baseline brain metastases No | 79 / 118 | 82 / 116 | 0.56 (0.41, 0.77) EGFR mutation status at study entry* Ex19del | 77 / 111 | 76 / 116 | 0.57 (0.41, 0.79) EGFR mutation status at study entry* L858R | 39 / 59 | 37 / 50 | 0.47 (0.30, 0.74) EGFR mutation status based on ctDNA at C1D1 Positive | 65 / 94 | 66 / 82 | 0.49 (0.35, 0.70) EGFR mutation status based on ctDNA at C1D1 Negative | 12 / 23 | 18 / 25 | 0.55 (0.26, 1.14) EGFR mutation status based on ctDNA at C1D1 Missing | 40 / 54 | 29 / 60 | 0.62 (0.38, 1.02) MET FISH10+ Yes | 89 / 129 | 79 / 116 | 0.46 (0.33, 0.63) MET FISH10+ No / Unknown | 28 / 42 | 34 / 51 | 0.73 (0.44, 1.21) MET IHC3+≥90% Yes | 77 / 124 | 80 / 118 | 0.51 (0.37, 0.70) MET IHC3+≥90% No / Unknown | 40 / 47 | 33 / 49 | 0.64 (0.40, 1.02) MET IHC and FISH status by enrolment assay FISH10+ and IHC3+≥90%+ | 50 / 83 | 46 / 68 | 0.39 (0.26, 0.59) MET IHC and FISH status by enrolment assay FISH10– and IHC3+≥90%+ | 24 / 38 | 32 / 46 | 0.69 (0.40, 1.17) MET IHC and FISH status by enrolment assay FISH10+ and IHC3+≥90%– | 37 / 44 | 21 / 33 | 0.67 (0.39, 1.18) Favours osimertinib + savolitinib <- -> Favours platinum–pemetrexed PFS benefit was consistent across prespecified subgroups Data cut-off: 06 July 2026. The shaded area denotes the 95% confidence interval for the overall hazard ratio from the stratified log-rank analysis in the full analysis set. For subgroups with fewer than 20 events across the two groups, hazard ratios were not produced. *Based on historical data at study sites.

[Slide 2] Progression-free survival by BICR Median PFS, months (95% CI) Osimertinib + savolitinib 7.6 (5.6, 9.7) Platinum–pemetrexed 5.4 (4.2, 5.8) HR (95% CI); p-value 0.55 (0.42, 0.72); p<0.001 36% / 10% (at 12 months) No. at risk Osimertinib + savolitinib | 171 166 144 124 120 98 82 75 70 64 54 49 39 34 29 25 24 22 21 14 14 8 8 6 5 2 1 1 1 1 1 1 1 1 1 1 1 0 | 171 / 117 Platinum–pemetrexed | 167 139 109 87 77 61 43 34 30 22 17 12 8 7 7 4 4 3 3 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 | 167 / 113 Osimertinib plus savolitinib provided a statistically significant 45% reduction in the risk of disease progression or death compared with platinum–pemetrexed Data cut-off: 06 July 2026. Median follow-up for PFS by BICR: osimertinib + savolitinib 5.6 months, platinum–pemetrexed 3.5 months. Overall maturity: 68% Shun Lu Osimertinib (osi) + savolitinib (savo) vs platinum–pemetrexed (plat–pem) in EGFRm MET-overexpressed (OE) and/or -amplified (AMP) advanced NSCLC post-osi: SAFFRON phase (Ph) 3 primary results Survival table Endpoint: PFS by BICR Arm | N | Median | Median CI Osimertinib + savolitinib | 171 | 7.6 | 5.6, 9.7 Platinum–pemetrexed | 167 | 5.4 | 4.2, 5.8 HR: 0.55 HR CI: 0.42, 0.72 P: <0.001 Landmark rates: 12 months: Osimertinib + savolitinib 36% | Platinum–pemetrexed 10% [Slide 4] Progression-free survival by BICR: MET subgroups MET-overexpressed analysis set Median PFS, months (95% CI): Osimertinib + savolitinib 8.5 (5.7, 11.1); Platinum–pemetrexed 5.6 (4.4, 6.7) HR (95% CI); p-value 0.51 (0.37, 0.71); p<0.001 MET-amplified analysis set* Median PFS, months (95% CI): Osimertinib + savolitinib 7.6 (5.6, 9.7); Platinum–pemetrexed 4.2 (3.1, 5.5) HR (95% CI) 0.48 (0.35, 0.66) [number-at-risk rows illegible] PFS by BICR in the MET-overexpressed and MET-amplified analysis sets was generally consistent with the full analysis set *No formal statistical testing was carried out for the MET-amplified analysis set. Shun Lu Osimertinib (osi) + savolitinib (savo) vs platinum–pemetrexed (plat–pem) in EGFRm MET-overexpressed (OE) and/or -amplified (AMP) advanced NSCLC post-osi: SAFFRON phase (Ph) 3 primary results Survival table Endpoint: PFS by BICR, MET subgroups MET-OE set: Osimertinib + savolitinib | 8.5 | 5.7, 11.1 MET-OE set: Platinum–pemetrexed | 5.6 | 4.4, 6.7 MET-amp set: Osimertinib + savolitinib | 7.6 | 5.6, 9.7 MET-amp set: Platinum–pemetrexed | 4.2 | 3.1, 5.5 HR: MET-OE 0.51; MET-amp 0.48 HR CI: MET-OE 0.37, 0.71; MET-amp 0.35, 0.66 P: MET-OE <0.001; MET-amp not formally tested

Conclusions • SAFFRON is the first global phase 3 study showing statistically significant and clinically meaningful improvements in PFS and OS with osimertinib plus savolitinib versus platinum–pemetrexed, after progression on osimertinib, in EGFRm MET-overexpressed and/or -amplified advanced NSCLC PFS HR 0.55; 95% CI 0.42, 0.72; p<0.001 | OS HR 0.68; 95% CI 0.50, 0.91; p = 0.010 • The PFS benefit with osimertinib plus savolitinib versus platinum–pemetrexed was consistent across prespecified subgroups • The combination demonstrated a tolerable safety profile, which was consistent with the established profiles for each treatment1–5 • SAFFRON establishes the clinical value of an all-oral, biomarker-directed treatment approach targeting both EGFR and MET following progression on osimertinib, reinforcing osimertinib as a backbone treatment across lines of therapy Osimertinib plus savolitinib establishes a new global standard of care for patients with EGFRm advanced NSCLC with MET overexpression and/or amplification after progression on osimertinib 1. Hartmaier et al. Cancer Discov 2023;13:98–113; 2. Le et al. J Thorac Oncol 2026;21:318–327; 3. Yang et al. Clin Lung Cancer 2026;27:38–46; 4. de Marinis et al. Ann Oncol 2025;36:920–933; 5. Lu et al. Lancet. 2026 24;407:375–38[illegible]

Osimertinib plus savolitinib versus platinum–pemetrexed in patients with EGFRm advanced NSCLC with MET overexpression and/or amplification after progression on osimertinib: SAFFRON Phase III primary results Shun Lu,1 Nicolas Girard, Yusuke Okuma, Frances A. Shepherd, Xiaorong Dong, Surein Arulananda, Giulio Metro, Hidetoshi Hayashi, Laura Bonanno, Yu Jung Kim, Zhe Liu, Myung-Ju Ahn, Yuping Sun, Le Thuong Vu, Silvia Novello, Rafał Lolo, Balakumar Swaminathan, Laura Roncoroni, Francis Maria Esposito, James Chih-Hsin Yang 1Department of Medical Oncology, Shanghai Chest Hospital, Shanghai Jiao Tong University, School of Medicine, Shanghai, China 25 October 2026

[Slide 1] Progress in Classical EGFR-mutated NSCLC (ex19del, L858R) Regulatory approval: USA | China | EU Selected milestones: Platinum doublet | [star] | 1st gen EGFR TKI | 2nd gen EGFR TKI | 3rd gen EGFR TKI | Anti-VEGF | Adjuvant EGFR TKI | 1L combinations | Consolidation post CRT | TROP2 ADC VEGF-PD1 bsAb 2004 - EGFR mutation discovered; 2005 – EGFR T790M resistance mutation described; 2007 – MET amplification as resistance mechanism; 2015 – EGFR C797S as resistance mechanism Median PFS and OS in EGFRm NSCLC (selected trials), mPFS / mOS (months): Platform doublet ~6 / ~23; G1/G2 EGFR TKI 9.5 / ~30; G3 EGFR TKI 18.9 / ~37; G3 EGFR TKI Combos 25.5 / 48+ [Drug/approval labels as in the clean capture 2114392179212951939_s1] Natasha B. Leighl [Slide 2] Current Guidelines For Acquired Resistance In EGFRm NSCLC ESMO guidelines Stage IV mNSCLC with EGFR-activating mutation progressing after a third-generation TKI Genomic analysis by NGS (tissue or cfDNA followed by tissue, if no target is found with cfDNA) [III, C] Repeat biopsy (includes plasma as screening) Resistance mechanism identified → Clinical trial [III, B] or refer to no resistance mechanism → Systemic progression No resistance mechanism identified → Amivantamab-vmjw–platinum-based doublet ChT [I, A; MCBS 2]; Platinum-based doublet ChT [III, A]; Atezolizumab–bevacizumab–paclitaxel–carboplatin [III, B; MCBS 3] → Datopotamab deruxtecan (a) [III, B; MCBS 2]; T-DXd [III, B] for patients with HER2+ (IHC 3+) (b) NCCN guidelines Limited/CNS progression → Local therapy and continue first-line treatment Consider repeat biopsy to r/o small cell transformation, consider NGS (tissue or plasma) Multifocal Systemic progression → Preferred: Amivantamb + ChT post osimertinib; Datopotamab Deruxtecan; Also: amivantamab+Lazertinib, docetaxel+ramucirumab, TDxD (HER2IHC3+), other single agent chemotherapy ASCO guidelines Repeat biopsy to r/o small cell, NGS (plasma if tissue not feasible) Second-Line Therapy For patients who develop EGFR T790M resistance alterations in tumor after first- or second-generation EGFR TKIs: Osimertinib [S] For patients who have progressive disease on osimertinib or other third-generation EGFR TKIs without emergent T790M or other targetable alterations: Platinum-based chemotherapy + amivantamab [S] For patients who are not candidates for platinum-based chemotherapy ± amivantamab: Platinum-based chemotherapy ± osimertinib [C] For patients who have progressive disease on osimertinib and platinum chemotherapy: Datopotamab deruxtecan [C] For patients who have progressive disease on EGFR TKI with acquired MET amplification: Osimertinib + tepotinib or savolitinib [C] Hendriks LEL et al. ESMO Living Guideline. 2026 (EGFR-activating Mutation After Systemic Progression with Third-generation TKI v1.3); Riley GJ et al. JNCCN. 2026;24(4):e260017 (NSCLC, Version 4.2026); Puri S et al...Leighl J Clin Oncol. 2025;44:e15-e55 (ASCO Living Guidelines, 2026.3.0). Natasha B. Leighl [Slide 3] Multiple studies of EGFR TKI + MET targeting agents (mostly Type Ib TKIs) 3 lessons to teach… Regimen | Prior EGFR TKI | Criteria of MET Dysregulation | n | ORR | mPFS | mDoR Gefitinib + capmatinib11: NCT01610336 | Gefitinib | IHC 3+ > 50% tumour cells or MET GCN ≥ 4 | 36; 78 | 47%: GCN ≥ 6; 32%: IHC 3+ | 5.49 m: GCN ≥ 6; 5.45 m: IHC 3+ | N/A Gefitinib + tepotinib10 (INSIGHT): NCT01982955 | 1/2 G TKI | IHC 2+ or 3+ and/or FISH + (GCN≥5 or MET/CEP7≥2) | 19 MET amplified | 67% MET amplified | 16.6 m MET amplified | 19.9 m MET amplified Osimertinib + tepotinib (INSIGHT 2)14: NCT03940703 | 3 G TKI (first-line) | FISH (GCN ≥ 5 or MET/CEP7 ≥ 2) | 90 | N/A | N/A | N/A Gefitinib + savolitinib59 NCT02374645 | 1/2/3 G TKI | FISH (GCN ≥ 5 or MET/CEP7 ≥ 2) | 51 | 31% | 4.0 months (EGFR T790M-negative: 4.2 m) | 5.6 months (EGFR T790M-negative: 7.2 m) Osimertinib + savolitinib (TATTON)60: NCT02143466 | 1/2 G TKI, EGFR T790M-negative | IHC 3+; FISH (GCN ≥ 5 or MET/CEP7 ≥ 2); NGS(GCN ≥5) | 42 | 62% | 9.0 m | 8.0 m Osimertinib + savolitinib (TATTON)60: NCT02143466 | 3 G TKI (any line) | IHC 3+; FISH (GCN ≥ 5 or MET/CEP7 ≥ 2); NGS(GCN ≥5) | 69 | 33% | 5.5 m | 7.9 m Osimertinib + savolitinib12,62 (ORCHARD): NCT03944772 | 3 G TKI (first-line) | NGS (GCN 7 - 68) | 17 MET amplified | 41% | N/A | N/A Lazertinib + amivantamab13 (CHRYSALIS): NCT02609776 | 3 G TKI (any line) | Tissue/liquid NGS EGFR/MET- based resistance | 16 | 36%; 47% (IHC for EGFR/MET) | 4.9 m; 6.7 m (IHC for EGFR/MET) | Not reached 1. Continued EGFR inhibition is needed SAVANNAH: Savolitinib + osimertinib vs Savolitinib + placebo; PFS HR (95% CI): 0.27 PFS events / patients: Savolitinib + osimertinib 26 / 48 (54%); Savolitinib + placebo 14 / 25 (56%) INSIGHT-2 – Tepotinib +/- Osimertinib | Tissue GCN≥5 Plasma GCN≥2.3 N=128) | GCN≥10 N=53 | GCN <10 N=45 | Tepotinib alone (N=12) ORR | 50% | 57% | 42% | 8% Median DOR | 8.5 m | NR | NR | NA Ahn MJ et al Clin Lung Cancer 2022; Liam CK, et al. Clin Cancer Res 2023; Hartmaier RJ et al. Cancer Discov. 2023; Ahn M-J et al. ELCC 2025; Lu S et al. Lancet. 2026

Patient demographics and baseline characteristics Characteristic | Osimertinib + savolitinib (n = 171) | Platinum–pemetrexed (n = 167) Sex: male / female, % | 36 / 64 | 35 / 65 Median age (range), years | 64 (34–86) | 64 (37–86) Race: Asian / non-Asian,* % | 63 / 36 | 66 / 28 ECOG PS: 0 / 1, % | 35 / 65 | 29 / 71 Extent of disease: locally advanced / metastatic, % | 5 / 95 | 5 / 95 Brain metastases at baseline, yes / no, % | 31 / 69 | 31 / 69 Prior line of therapy: 1 / ≥2, % | 84 / 16 | 84 / 16 MET status: IHC3+≥90% / FISH10+ / both, % | 73 / 75 / 49 | 71 / 69 / 41 EGFR mutation at randomisation:† Ex19del / L858R / other,‡ % | 65 / 35 / 1 | 69 / 30 / 1 Smoking status: never / current / former, % | 68 / 2 / 30 | 68 / 4 / 28 Data cut-off: 06 July 2026. Percentages may not total 100 due to rounding. *Data missing or not reported for n = 3 (osimertinib + savolitinib) and n = 11 (platinum–pemetrexed); †Patients with both Ex19del and L858R mutations were included in the Ex19del group (osimertinib + savolitinib arm: n = 4; platinum–pemetrexed arm: n = 6); ‡Other EGFR mutations included T790M (osimertinib + savolitinib arm: n = 1; platinum–pemetrexed arm: n = 1).

[Slide 1] Is SAFFRON well-designed with valid results? [check] YES and [check] YES SAFFRON study design [Embedded study-design schematic and CONSORT diagram - small print largely illegible in this photo. Legible fragments: Osimertinib 80 mg QD + savolitinib 300 mg BID; Pemetrexed 500 mg/m2 + carboplatin AUC5/cisplatin 75 mg/m2 (Q3W for 4 cycles); Maintenance pemetrexed 500 mg/m2 (Q3W); Endpoints - Primary: PFS by BICR per RECIST v1.1 in the FAS; Key secondary/exploratory: PFS by investigator, OS, ORR, DoR, DCR and tumour shrinkage in the FAS; PFS in patients in the MET-overexpressed [illegible] or MET-amplified [illegible] analysis sets; Safety in the SAF; Osimertinib + savolitinib, n = 171; Platinum-pemetrexed, n = 167; Patients screened, N = [illegible]; Patients randomised, N = [illegible]] 0 v 8% withdrew post randomization 99% v 87% started protocol therapy Patient withdrawal 6% v 13% Lost to follow up 0 v 1% ✓ Molecular selection, dual targeting strategy informed by prior studies1-4 ✓ Is the control arm the current standard of care? (it was...) ✓ Randomized, power 90%, intent-to-treat-analysis, no crossover ✓ Unblinded but primary endpoint PFS by BICR ✓ PFS primary endpoint acceptable, (OS preferred) ✓ Patient characteristics balanced (1/3 male, 2/3 Asian, 31% CNS metastases, 16% post 2 TKIs) ✓ Disproportionate withdrawals in control arm ✓ Fixed sequence hierarchical testing appropriate for multiple testing (Type 1 error controlled) – IHC+ subgroup prioritized→ relevant for implementation Lu S et al. ESMO Congress. 2026 LBA6 [Slide 3] Is MET Overexpression Enough or Do We Need MET Addiction? PFS for MET-overexpressed analysis set Median PFS, months (95% CI) Osimertinib + savolitinib | 8.5 (5.7, 11.1) Platinum–pemetrexed | 5.6 (4.4, 6.7) HR 0.51 (95% CI 0.37-0.71); p<0.001 Landmark labels at 12 months: 40% | 11% No. at Risk Osimertinib + savolitinib 124 106 92 64 55 40 29 22 17 15 10 6 4 1 1 1 1 1 1 0 Platinum-pemetrexed 118 77 63 33 23 13 7 6 3 2 1 0 0 0 0 0 0 0 0 0 PFS for MET-amplified analysis set* Median PFS, months (95% CI) Osimertinib + savolitinib | 7.6 (5.6, 9.7) Platinum–pemetrexed | 4.2 (3.1, 5.5) HR 0.48 (95% CI 0.35-0.66) *No formal statistical testing was carried out for the MET-amplified analysis set Landmark labels at 12 months: 33% | 8% No. at Risk Osimertinib + savolitinib 129 108 91 63 54 39 28 20 19 16 11 6 4 1 1 1 1 1 1 0 Platinum-pemetrexed 116 72 42 23 16 9 4 3 1 0 0 0 0 0 0 0 0 0 0 0 Lu S et al. ESMO Congress. 2026; Courtesy J. Pinchevsky, AstraZeneca (internal data) Natasha B. Leighl Survival table Endpoint: PFS by MET analysis set Arm | N | Median | Median CI MET-overexpressed: osimertinib + savolitinib | 124 | 8.5 mo | 5.7, 11.1 MET-overexpressed: platinum-pemetrexed | 118 | 5.6 mo | 4.4, 6.7 MET-amplified: osimertinib + savolitinib | 129 | 7.6 mo | 5.6, 9.7 MET-amplified: platinum-pemetrexed | 116 | 4.2 mo | 3.1, 5.5 HR: MET-overexpressed 0.51; MET-amplified 0.48 HR CI: 0.37-0.71; 0.35-0.66 P: MET-overexpressed p<0.001; MET-amplified not formally tested Landmark rates: 12 months (dotted line): MET-overexpressed: osimertinib + savolitinib 40% | MET-overexpressed: platinum-pemetrexed 11% | MET-amplified: osimertinib + savolitinib 33% | MET-amplified: platinum-pemetrexed 8%

Progression-free Survival in the Full Analysis Population | No. of Events/No. of Patients | Median Progression-free Survival (95% CI), mo Osimertinib–Savolitinib | 117/171 | 7.6 (5.6–9.7) Platinum–Pemetrexed | 113/167 | 5.4 (4.2–5.8) Hazard ratio for disease progression or death, 0.55 (95% CI, 0.42–0.72) P<0.001 12-month landmark: Osimertinib–savolitinib 36; Platinum–pemetrexed 10 Months since Randomization No. at Risk Osimertinib–savolitinib 171 144 120 82 70 54 39 29 24 21 14 8 5 1 1 1 1 1 1 0 Platinum–pemetrexed 167 109 77 43 30 17 8 7 4 3 1 0 0 0 0 0 0 0 0 0 Survival table Endpoint: PFS, full analysis population Arm | N | Median | Median CI Osimertinib–savolitinib | 171 | 7.6 mo | 5.6–9.7 Platinum–pemetrexed | 167 | 5.4 mo | 4.2–5.8 HR: 0.55 HR CI: 0.42–0.72 P: <0.001 Landmark rates: 12 mo: Osimertinib–savolitinib 36% | Platinum–pemetrexed 10%

No. of Events/No. of Patients; Median Overall Survival (95% CI), mo Osimertinib–Savolitinib 87/171 17.8 (15.6–23.7) Platinum–Pemetrexed 93/167 12.9 (10.6–16.8) Hazard ratio for death, 0.68 (95% CI, 0.50–0.91) P=0.01 12-month landmark labels: 70 vs 53 Months since Randomization Percentage of Patients No. at Risk: Osimertinib–savolitinib, Platinum–pemetrexed Survival table Endpoint: OS Arm | N | Median | Median CI Osimertinib–savolitinib | 171 | 17.8 mo | 15.6–23.7 Platinum–pemetrexed | 167 | 12.9 mo | 10.6–16.8 HR: 0.68 HR CI: 0.50–0.91 P: P=0.01 12-mo OS label: 70 vs 53

The NEW ENGLAND JOURNAL of MEDICINE October 25, 2026 Osimertinib–Savolitinib in Advanced NSCLC after Progression during EGFR-TKI Therapy Authors: Shun Lu, M.D., Nicolas Girard, M.D., Ph.D., James Chih-Hsin Yang, M.D., Ph.D., Frances A. Shepherd, M.D., Xiaorong Dong, M.D., Ph.D., Surein Arulananda, Ph.D., Giulio Metro, M.D., Progression-free Survival in the Full Analysis Population No. of Events/No. of Patients | Median Progression-free Survival (95% CI), mo Osimertinib–Savolitinib | 117/171 | 7.6 (5.6–9.7) Platinum–Pemetrexed | 113/167 | 5.4 (4.2–5.8) Hazard ratio for disease progression or death, 0.55 (95% CI, 0.42–0.72) P<0.001 12 months: Osimertinib–savolitinib 36; Platinum–pemetrexed 10 No. at Risk Osimertinib–savolitinib 171 144 120 82 70 54 39 29 24 21 14 8 5 1 1 1 1 1 1 0 Platinum–pemetrexed 167 109 77 43 30 17 8 7 4 3 1 0 0 0 0 0 0 0 0 0 Months since Randomization Survival table Endpoint: PFS (full analysis population) Arm | N | Median | Median CI Osimertinib–savolitinib | 171 | 7.6 | 5.6–9.7 Platinum–pemetrexed | 167 | 5.4 | 4.2–5.8 HR: 0.55 HR CI: 0.42–0.72 P: P<0.001 Landmark rates: 12 months: Osimertinib–savolitinib 36% | Platinum–pemetrexed 10%

We Need A Deeper Understanding Of The Drivers Of Treatment Sensitivity, Resistance And How To Improve The Patient's Journey EGFR TKI Progression Re-biopsy • Target EGFR-dependent mutations • Block bypass pathways • Anti-angiogenesis • ADCs • Immunotherapy Progression Re-biopsy Unmet clinical needs Non-responsive disease (primary/intrinsic resistance) Limited DOR Toxicity management Novel resistance mechanisms Predict Novel biomarkers Cellular level Transcriptional level Protein/metabolite level Artificial intelligence Multi-omics Epigenetics Genomics Transcriptomics Proteomics Metabolomics Radiomics Overcome Structure-based drug discovery Superior strategies Cell therapy ctDNA-guided adaptive therapy ctDNA+ Treatment intensification More-potent drugs Superior combination therapy ctDNA- Treatment de-intensification Drug holiday EGFR TKI resistance Mechanism of resistance EGFR dependent EGFR independent Histologic transformation Unknown Molecular/cellular mechanism EGFR mutation (e.g. C797S) EGFR amplification Oncogenic fusions MET or HER2 amplification Downstream pathway activation Cell cycle gene alterations SCC SCLC Unknown Potential treatment approach Targeted therapy (e.g. next-generation EGFR TKIs) Targeted therapy specific to the identified gene or pathway involved Platinum-based CT CT ± ICIs ± anti-VEGF Zhao J, et al Nat Rev Clin Oncol 2026; Le et al 2026 [illegible]dated from Zhao J, et al Nat Rev Clin Oncol 2026 [illegible]tasha B. Leighl

The addition of chemotherapy to pembrolizumab significantly prolonged PFS compared with pembrolizumab monotherapy in treatment-naïve patients with advanced NSCLC and high PD-L1 expression (≥50% of tumor cells) PERSEE is an academic, prospective, multicenter, randomized, phase III trial conducted across 28 sites in France comparing pembrolizumab plus chemotherapy and pembrolizumab monotherapy as first-line in treatment-naïve patients with NSCLC. Key eligibility criteria • Age ≥18 years • Stage IV NSCLC OR Unresectable Stage III NSCLC not eligible for radiotherapy • PD-L1 expression on ≥50% of tumor cells • No known EGFR mutations, no ALK or ROS1 rearrangements • ECOG PS 0-1 | Pembro + CT | Pembro Randomized ITT population | N = 160 | N = 161 Received ≥1 dose Safety population | n = 158 (Carbo + pacli (n = 32); Carbo + pemet (n = 111); Cispl + pemet (n = 15)) | n = 159 Planned treatment completed | n = 45 | n = 27 Conclusion: The addition of platinum-based chemotherapy to pembrolizumab significantly prolonged PFS compared with pembrolizumab monotherapy. Median follow-up: 38.7 months Primary endpoint: Progression-Free Survival* Arm | Events n (%) | Median PFS (95% CI), months | HR (95% CI) Two-sided P value Pembro + CT | 93 (58.1) | 16.1 (11–34.1) | 0.56 (0.43–0.74) Pembro | 117 (72.7) | 6.9 (5.5–9.3) | P < 0.001 Landmarks: 54.6%, 45.2%, 39.2% (Pembro + CT); 33.9%, 24.9%, 20.8% (Pembro) *PFS according to blinded central read Overall Survival Arm | Events n (%) | Median OS (95% CI), months | HR (95% CI) Two-sided P value Pembro + CT | 74 (46.3) | 50.2 (32.6–NC) | 0.77 (0.56–1.05) Pembro | 89 (55.3) | 29.9 (23.4–47.9) | NC Landmarks: 78.8%, 64.4%, 55.4%, 50.9% (Pembro + CT); 73.3%, 56.2%, 44.9%, 38.9% (Pembro) Treatment response (ITT population) Objective response rate (ORR): OR 3.48 (95% CI, 2.15–5.72); 124 (77.5%) vs 81 (50.3%) Early disease progression: OR 0.17 (95% CI, 0.07–0.35); 9 (5.8%) vs 41 (26.5%) Adverse Events (Safety population) | Pembro + CT (N = 158) | Pembro (N = 159) Treatment-related AE: Any grade | 149 (94) | 119 (75) Grade ≥3 | 86 (54) | 31 (19) Grade 5 | 3 (1.9) | 3 (1.9) Immune related AE: Any grade | 39 (25) | 32 (20) Grade ≥3 | 21 (13) | 20 (13) Grade 5 | 0 (0) | 2 (1.3) Treatment action (All AEs) leading to Pembro discontinuation | 44 (28) | 38 (24) Pembro pause | 78 (49) | 39 (25) CT discontinuation | 76 (48) | NA CT pause | 72 (46) | NA CT dose reduction | 72 (46) | NA

[Slide 1] Study Design A phase III, multicenter, open-label, randomized trial Key eligibility criteria • Stage IV and unresectable or not amenable to definitive radiotherapy stage III NSCLC • TPS PD-L1 ≥50% (local assessment) • ECOG Performance Status ≤1 • No EGFR, ALK or ROS1 alteration R 1:1 PEMBROLIZUMAB 200 mg, Q3W -> Up to 35 cycles • Non-squamous NSCLC : PEMBROLIZUMAB 200mg Q3W + PEMETREXED 500 mg/m2 + CARBOPLATIN AUC5 Q3W or CISPLATIN 75 mg/m2 Q3W • Squamous NSCLC : PEMBROLIZUMAB 200 mg Q3W + PACLITAXEL 200 mg/m2 Q3W + CARBOPLATIN AUC 6 Q3W Up to 4 cycles Maintenance treatment • Non-squamous: PEMBROLIZUMAB up to 31 cycles and PEMETREXED • Squamous: PEMBROLIZUMAB up to 31 cycles Stratification factors : • Squamous vs non-squamous • Brain metastases yes vs no Primary endpoint: PFS by BCR per RECIST v1.1. Key Secondary endpoints: PFS by investigators, OS, ORR, DoR, DCR, efficacy according to PD-L1 sub-groups expression, safety. Key Exploratory endpoints: incidence of early progression, PFS2, QoL, translational analyses PFS : Progression Free Survival; OS : Overall Survival; ORR : Objective Response rate; BCR : Blinded Central Review,, DoR : Duration of Response, DCR: Disease Control Rate; QoL : quality-of- life questionnaires Early progression rate, defined as the proportion of patients with clinical and/or radiological tumor progression according to RECIST v1.1 before or at the first radiological evaluation at 6 weeks Chantal Decroisette [Slide 2] Primary endpoint: PFS by BCR (RECIST 1.1) Median follow-up was 38.7 (35.3-41.6) months Arm | Events n (%) | Median PFS (95% CI), months | HR (95% CI) Two-sided P value Pembro + CT | 93 (58.1) | 16.1 (11–34.1) | 0.56 (0.43–0.74) Pembro | 117 (72.7) | 6.9 (5.5–9.3) | P < 0.001 Landmarks (Pembro + CT): 54.6% (47.3–63.1); 45.2% (37.9–53.9); 39.2% (31.8–48.4) Landmarks (Pembro): 33.9% (27.1–42.5); 24.9% (18.7–33.1); 20.8% (15.0–28.9) Months since Randomization No. at Risk Pembro + CT 160 141 113 90 82 77 71 66 60 52 48 41 34 28 20 15 10 9 6 1 0 Pembro 161 99 84 63 47 44 40 36 32 29 26 24 22 19 12 10 5 2 2 0 0 PFS assessed by BCR was significantly longer in the pembro-CT group than in the pembro group, with a clinically meaningful benefit sustained over time. The PFS benefit was consistent with investigators assessment with a HR 0.59 (0.46-0.77). Chantal Decroisette Survival table Endpoint: PFS by BCR (RECIST 1.1) Arm | N | Median | Median CI Pembro + CT | 160 | 16.1 months | 11–34.1 Pembro | 161 | 6.9 months | 5.5–9.3 HR: 0.56 HR CI: 0.43–0.74 P: <0.001 (two-sided) Landmark rates: 12 months: Pembro + CT 54.6% (CI 47.3–63.1) | Pembro 33.9% (CI 27.1–42.5) 24 months: Pembro + CT 45.2% (CI 37.9–53.9) | Pembro 24.9% (CI 18.7–33.1) 36 months: Pembro + CT 39.2% (CI 31.8–48.4) | Pembro 20.8% (CI 15.0–28.9) [Slide 3] The PFS benefit was observed regardless of PD-L1 expression level PD-L1 50% - 74% Arm | Events n (%) | mPFS months (IC 95 %) | HR (IC 95 %) Pembro + CT | 32 (59.3) | 17.8 (9.3–NC) | 0.46 (0.29–0.74) Pembro | 44 (75.9) | 7.0 (3.6–10) | Landmarks: Pembro + CT 57.2%, 45.4%, 39.7%; Pembro 32.8%, 21.9%, 15.3% No. at Risk Pembro + CT 54 49 42 32 29 27 25 22 19 18 18 15 14 10 7 4 3 2 2 1 Pembro 58 34 30 21 15 14 13 12 10 8 8 8 7 7 3 1 1 0 0 0 PD-L1 75% - 100% Arm | Events n (%) | mPFS months (IC 95 %) | HR (IC 95 %) Pembro + CT | 51 (56.0) | 16.9 (11.1–NC) | 0.62 (0.43–0.91) Pembro | 65 (68.4) | 7.1 (5.1–11.2) | Landmarks: Pembro + CT 56.2%, 46.9%, 41.9%; Pembro 36.5%, 29.0%, 26.4% No. at Risk Pembro + CT 91 80 64 54 49 46 42 40 37 31 27 23 19 17 12 10 6 6 4 0 Pembro 95 61 50 39 31 30 27 24 22 21 18 16 15 12 9 9 4 2 2 0 PD-L1 90% - 100% Arm | Events n (%) | mPFS months (IC 95 %) | HR (IC 95 %) Pembro + CT | 26 (50.0) | 25.4 (11–NC) | 0.6 (0.36–1) Pembro | 39 (68.4) | 7.1 (4.9–18.8) | Landmarks: Pembro + CT 58.3%, 52.0%, 49.4%; Pembro 40.2%, 30.3%, 28.2% No. at Risk Pembro + CT 52 44 36 32 29 29 26 26 23 18 17 16 14 12 8 8 4 4 2 0 0 Pembro 57 39 29 24 22 21 18 15 14 14 11 10 10 7 6 6 3 2 2 0 0 Landmark CIs: [illegible - very small print] Chantal Decroisette [Slide 4] Secondary endpoint : Overall survival (ITT) Arm | Events n (%) | Median OS (95% CI), months | HR (95% CI) Pembro + CT | 74 (46.3) | 50.2 (32.6–NC) | 0.77 (0.56–1.05) Pembro | 89 (55.3) | 29.9 (23.4–47.9) | Landmarks (Pembro + CT): 78.8% (72.7–85.4); 64.4% (57.3–72.2); 55.4% (47.9–63.9); 50.9% (43.1–60.1) Landmarks (Pembro): 73.3% (66.8–80.5); 56.2% (49.0–64.5); 44.9% (37.5–53.7); 38.9% (30.8–49.0) Months since Randomization No. at Risk Pembro + CT 160 151 144 136 126 121 110 108 102 90 84 64 55 46 39 28 26 15 11 5 1 Pembro 161 145 134 126 118 108 103 97 87 74 68 57 45 39 31 22 19 10 4 2 0 A numerically longer median OS of 50.2 months in the pembro-CT group was observed compared with 29.9 months in the pembro group. The trial was not powered to demonstrate an overall survival difference. Chantal Decroisette Survival table Endpoint: Overall survival (ITT) Arm | N | Median | Median CI Pembro + CT | 160 | 50.2 months | 32.6–NC Pembro | 161 | 29.9 months | 23.4–47.9 HR: 0.77 HR CI: 0.56–1.05 Landmark rates: 12 months: Pembro + CT 78.8% (CI 72.7–85.4) | Pembro 73.3% (CI 66.8–80.5) 24 months: Pembro + CT 64.4% (CI 57.3–72.2) | Pembro 56.2% (CI 49.0–64.5) 36 months: Pembro + CT 55.4% (CI 47.9–63.9) | Pembro 44.9% (CI 37.5–53.7) 48 months: Pembro + CT 50.9% (CI 43.1–60.1) | Pembro 38.9% (CI 30.8–49.0)

[Slide 2] A PD-L1 between 50% and 74% Arm | Events n (%) | Median PFS (95% CI), months | HR (95% CI) Two-sided P value Pembro + CT | 32 (59.3) | 17.8 (9.3–NC) | 0.46 (0.29–0.74) Pembro | 44 (75.9) | 7 (3.6–10) | NC Landmark labels (12 / 24 / 36 months): Pembro + CT 57.2% / 45.4% / 39.7%; Pembro 32.8% / 21.9% / 15.3% No. at Risk Pembro + CT 54 49 42 32 29 27 25 22 19 18 18 15 14 10 7 4 3 2 2 1 0 Pembro 58 34 30 21 15 14 13 12 10 8 8 8 7 7 3 1 1 0 0 0 0 B PD-L1 between 75% and 100% Pembro + CT | 51 (56.0) | 16.9 (11.1–NC) | 0.62 (0.43–0.91) Pembro | 65 (68.4) | 7.1 (5.1–11.2) | NC Landmark labels (12 / 24 / 36 months): Pembro + CT 56.2% / 46.9% / 41.9%; Pembro 36.5% / 29.0% / 26.4% No. at Risk Pembro + CT 91 80 64 54 49 46 42 40 37 31 27 23 19 17 12 10 6 6 4 0 0 Pembro 95 61 50 39 31 30 27 24 22 21 18 16 15 12 9 9 4 2 2 0 0 C PD-L1 between 90% and 100% Pembro + CT | 26 (50.0) | 25.4 (11–NC) | 0.6 (0.36–1) Pembro | 39 (68.4) | 7.1 (4.9–18.8) | NC Landmark labels (12 / 24 / 36 months): Pembro + CT 58.3% / 52.0% / 49.4%; Pembro 40.2% / 30.3% / 28.2% No. at Risk Pembro + CT 52 44 36 32 29 29 26 26 23 18 17 16 14 12 8 8 4 4 2 0 0 Pembro 57 39 29 24 22 21 18 15 14 14 11 10 10 7 6 6 3 2 2 0 0 Months since Randomization Survival table Endpoint: PFS by PD-L1 subgroup Arm | N | Median | Median CI PD-L1 50–74%: Pembro + CT | 54 | 17.8 | 9.3–NC PD-L1 50–74%: Pembro | 58 | 7 | 3.6–10 PD-L1 75–100%: Pembro + CT | 91 | 16.9 | 11.1–NC PD-L1 75–100%: Pembro | 95 | 7.1 | 5.1–11.2 PD-L1 90–100%: Pembro + CT | 52 | 25.4 | 11–NC PD-L1 90–100%: Pembro | 57 | 7.1 | 4.9–18.8 HR: 50–74%: 0.46; 75–100%: 0.62; 90–100%: 0.6 HR CI: 0.29–0.74; 0.43–0.91; 0.36–1 P: NC Landmark rates: 12 mo: 50–74% Pembro + CT 57.2% | 50–74% Pembro 32.8% | 75–100% Pembro + CT 56.2% | 75–100% Pembro 36.5% | 90–100% Pembro + CT 58.3% | 90–100% Pembro 40.2% [Slide 3] Arm | Events n (%) | Median OS (95% CI), months | HR (95% CI) Two-sided P value Pembro + CT | 74 (46.3) | 50.2 (32.6–NC) | 0.77 (0.56–1.05) Pembro | 89 (55.3) | 29.9 (23.4–47.9) | NC Landmark labels: 12 mo: Pembro + CT 78.8% (72.7–85.4); Pembro 73.3% (66.8–80.5) 24 mo: Pembro + CT 64.4% (57.3–72.2); Pembro 56.2% (49.0–64.5) 36 mo: Pembro + CT 55.4% (47.9–63.9); Pembro 44.9% (37.5–53.7) 48 mo: Pembro + CT 50.9% (43.1–60.1); Pembro 38.9% (30.8–49.0) No. at Risk Pembro + CT 160 151 144 136 126 121 110 108 102 90 84 64 55 46 39 28 26 15 11 5 1 Pembro 161 145 134 126 118 108 103 97 87 74 68 57 45 39 31 22 19 10 4 2 0 Months since Randomization Figure 3 Overall survival. Kaplan–Meier estimates of overall survival in the Survival table Endpoint: Overall survival Arm | N | Median | Median CI Pembro + CT | 160 | 50.2 mo | 32.6–NC Pembro | 161 | 29.9 mo | 23.4–47.9 HR: 0.77 HR CI: 0.56–1.05 P: NC Landmark rates: 12 mo: Pembro + CT 78.8% (CI 72.7–85.4) | Pembro 73.3% (CI 66.8–80.5) 24 mo: Pembro + CT 64.4% (CI 57.3–72.2) | Pembro 56.2% (CI 49.0–64.5) 36 mo: Pembro + CT 55.4% (CI 47.9–63.9) | Pembro 44.9% (CI 37.5–53.7) 48 mo: Pembro + CT 50.9% (CI 43.1–60.1) | Pembro 38.9% (CI 30.8–49.0)

The PFS benefit was observed regardless of PD-L1 expression level PD-L1 50% - 74% Arm | Events n (%) | mPFS months (IC 95 %) | HR (IC 95 %) Pembro + CT | 32 (59.3) | 17.8 (9.3–NC) | 0.46 (0.29–0.74) Pembro | 44 (75.9) | 7.0 (3.6–10) Curve labels: Pembro + CT 57.2% / 45.4% / 39.7%; Pembro 32.8% / 21.9% / 15.3% No. at Risk: Pembro + CT 54 49 42 32 29 27 25 22 19 18 18 15 14 10 7 4 3 2 2 1; Pembro 58 34 30 21 15 14 13 12 10 8 8 8 7 7 3 1 1 0 0 0 PD-L1 75% - 100% Arm | Events n (%) | mPFS months (IC 95 %) | HR (IC 95 %) Pembro + CT | 51 (56.0) | 16.9 (11.1–NC) | 0.62 (0.43–0.91) Pembro | 65 (68.4) | 7.1 (5.1–11.2) Curve labels: Pembro + CT 56.2% / 46.9% / 41.9%; Pembro 36.5% / 29.0% / 26.4% No. at Risk: Pembro + CT 91 80 64 54 49 46 42 40 37 31 27 23 19 17 12 10 6 6 4 0; Pembro 95 61 50 39 31 30 27 24 22 21 18 16 15 12 9 9 4 2 2 0 PD-L1 90% - 100% Arm | Events n (%) | mPFS months (IC 95 %) | HR (IC 95 %) Pembro + CT | 26 (50.0) | 25.4 (11–NC) | 0.6 (0.36–1) Pembro | 39 (68.4) | 7.1 (4.9–18.8) Curve labels: Pembro + CT 58.3% / 52.0% / 49.4%; Pembro 40.2% / 30.3% / 28.2% No. at Risk: Pembro + CT 52 44 36 32 29 29 26 26 23 18 17 16 14 12 8 8 4 4 2 0 0 0; Pembro 57 39 29 24 22 21 18 15 14 14 11 10 10 7 6 6 3 2 2 0 0 0 Months since Randomization Chantal Decroisette Persée GFPC Survival table Endpoint: PFS by PD-L1 TPS subgroup Arm | N | Median | Median CI PD-L1 50-74%: Pembro + CT | 54 | 17.8 | 9.3–NC PD-L1 50-74%: Pembro | 58 | 7.0 | 3.6–10 PD-L1 75-100%: Pembro + CT | 91 | 16.9 | 11.1–NC PD-L1 75-100%: Pembro | 95 | 7.1 | 5.1–11.2 PD-L1 90-100%: Pembro + CT | 52 | 25.4 | 11–NC PD-L1 90-100%: Pembro | 57 | 7.1 | 4.9–18.8 HR: 50-74%: 0.46; 75-100%: 0.62; 90-100%: 0.6 HR CI: 0.29–0.74; 0.43–0.91; 0.36–1 Landmark rates: 12 mo (50-74%): Pembro + CT 57.2% | Pembro 32.8% 24 mo (50-74%): Pembro + CT 45.4% | Pembro 21.9% 36 mo (50-74%): Pembro + CT 39.7% | Pembro 15.3% 12 mo (75-100%): Pembro + CT 56.2% | Pembro 36.5% 24 mo (75-100%): Pembro + CT 46.9% | Pembro 29.0% 36 mo (75-100%): Pembro + CT 41.9% | Pembro 26.4% 12 mo (90-100%): Pembro + CT 58.3% | Pembro 40.2% 24 mo (90-100%): Pembro + CT 52.0% | Pembro 30.3% 36 mo (90-100%): Pembro + CT 49.4% | Pembro 28.2%

The PFS benefit was observed regardless of PD-L1 expression level PD-L1 50% - 74% Arm | Events n (%) | mPFS months (IC 95 %) | HR (IC 95 %) Pembro + CT | 32 (59.3) | 17.8 (9.3-NC) | 0.46 (0.29-0.74) Pembro | 44 (75.9) | 7.0 (3.6-10) | Landmarks: Pembro + CT 57.2%, 45.4%, 39.7%; Pembro 32.8%, 21.9%, 15.3% No. at Risk Pembro + CT: 54 49 42 32 29 27 25 22 19 18 18 15 14 10 7 4 3 2 2 1 No. at Risk Pembro: 58 34 30 21 15 14 13 12 10 8 8 8 7 7 3 1 1 0 0 0 PD-L1 75% - 100% Pembro + CT | 51 (56.0) | 16.9 (11.1-NC) | 0.62 (0.43-0.91) Pembro | 65 (68.4) | 7.1 (5.1-11.2) | Landmarks: Pembro + CT 56.2%, 46.9%, 41.9%; Pembro 36.5%, 29.0%, 26.4% No. at Risk Pembro + CT: 91 80 64 54 49 46 42 40 37 31 27 23 19 17 12 10 6 6 4 0 No. at Risk Pembro: 95 61 50 39 31 30 27 24 22 21 18 16 15 12 9 9 4 2 2 0 PD-L1 90% - 100% Pembro + CT | 26 (50.0) | 25.4 (11-NC) | 0.6 (0.36-1) Pembro | 39 (68.4) | 7.1 (4.9-18.8) | Landmarks: Pembro + CT 58.3%, 52.0%, 49.4%; Pembro 40.2%, 30.3%, 28.2% No. at Risk Pembro + CT: 52 44 36 32 29 29 26 26 23 18 17 16 14 12 8 8 4 4 2 0 0 No. at Risk Pembro: 57 39 29 24 22 21 18 15 14 14 11 10 10 7 6 6 3 2 2 0 0 Chantal Decroisette Persée GFPC

10:15 - 11:45 Proffered paper 1: NSCLC, metastatic | CHAIR: SURESH SENAN The PFS benefit was observed regardless of PD-L1 expression level PD-L1 50% - 74%: Pembro + CT 32 (59.3) | 17.8 (9.3–NC) | HR 0.46 (0.29–0.74); Pembro 44 (75.9) | 7.0 (3.6–10) Curve labels: 57.2% / 45.4% / 39.7% vs 32.8% / 21.9% / 15.3% PD-L1 75% - 100%: Pembro + CT 51 (56.0) | 16.9 (11.1–NC) | HR 0.62 (0.43–0.91); Pembro 65 (68.4) | 7.1 (5.1–11.2) Curve labels: 56.2% / 46.9% / 41.9% vs 36.5% / 29.0% / 26.4% PD-L1 90% - 100%: Pembro + CT 26 (50.0) | 25.4 (11–NC) | HR 0.6 (0.36–1); Pembro 39 (68.4) | 7.1 (4.9–18.8) Curve labels: 58.3% / 52.0% / 49.4% vs 40.2% / 30.3% / 28.2% Chantal Decroisette | Persée GFPC ALICANTE AUDITORIUM - HALL 6 Survival table Endpoint: PFS by PD-L1 TPS subgroup Arm | N | Median | Median CI PD-L1 50-74%: Pembro + CT | 54 | 17.8 | 9.3–NC PD-L1 50-74%: Pembro | 58 | 7.0 | 3.6–10 PD-L1 75-100%: Pembro + CT | 91 | 16.9 | 11.1–NC PD-L1 75-100%: Pembro | 95 | 7.1 | 5.1–11.2 PD-L1 90-100%: Pembro + CT | 52 | 25.4 | 11–NC PD-L1 90-100%: Pembro | 57 | 7.1 | 4.9–18.8 HR: 0.46; 0.62; 0.6 HR CI: 0.29–0.74; 0.43–0.91; 0.36–1

Primary endpoint: PFS by BCR (RECIST 1.1) Median follow-up was 38.7 (35.3-41.6) months Arm | Events n (%) | Median PFS (95% CI), months | HR (95% CI) Two-sided P value Pembro + CT | 93 (58.1) | 16.1 (11-34.1) | 0.56 (0.43-0.74) Pembro | 117 (72.7) | 6.9 (5.5-9.3) | P < 0.001 Curve labels: Pembro + CT 54.6% (47.3-63.1) at 12; 45.2% (37.9-53.9) at 24; 39.2% (31.8-48.4) at 36 Pembro 33.9% (27.1-42.5) at 12; 24.9% (18.7-33.1) at 24; 20.8% (15.0-28.9) at 36 Months since Randomization No. at Risk (0, 3, 6, ... 60) Pembro + CT | 160 | 141 | 113 | 90 | 82 | 77 | 71 | 66 | 60 | 52 | 48 | 41 | 34 | 28 | 20 | 15 | 10 | 9 | 6 | 1 | 0 Pembro | 161 | 99 | 84 | 63 | 47 | 44 | 40 | 36 | 32 | 29 | 26 | 24 | 22 | 19 | 12 | 10 | 5 | 2 | 2 | 0 | 0 PFS assessed by BCR was significantly longer in the pembro-CT group than in the pembro group, with a clinically meaningful benefit sustained over time. The PFS benefit was consistent with investigators assessment with a HR 0.59 (0.46-0.77). Chantal Decroisette Survival table Endpoint: PFS by blinded central review (RECIST 1.1) Arm | N | Median | Median CI Pembro + CT | 160 | 16.1 | 11-34.1 Pembro | 161 | 6.9 | 5.5-9.3 HR: 0.56 HR CI: 0.43-0.74 P: <0.001 (two-sided) Landmark rates: 12 months: Pembro + CT 54.6% (CI 47.3-63.1) | Pembro 33.9% (CI 27.1-42.5) 24 months: Pembro + CT 45.2% (CI 37.9-53.9) | Pembro 24.9% (CI 18.7-33.1) 36 months: Pembro + CT 39.2% (CI 31.8-48.4) | Pembro 20.8% (CI 15.0-28.9)

[Slide 2] Primary endpoint: PFS by BCR (RECIST 1.1) Median follow-up was 38.7 (35.3-41.6) months Arm | Events n (%) | Median PFS (95% CI), months | HR (95% CI) Two-sided P value Pembro + CT | 93 (58.1) | 16.1 (11–34.1) | 0.56 (0.43–0.74) Pembro | 117 (72.7) | 6.9 (5.5–9.3) | P < 0.001 Curve labels: Pembro + CT 54.6% / 45.2% / 39.2%; Pembro 33.9% / 24.9% / 20.8% No. at Risk: Pembro + CT 160 ...; Pembro 161 ... (remaining values too small to transcribe reliably) PFS assessed by BCR was significantly longer in the pembro-CT group than in the pembro group, with a clinically meaningful benefit sustained over time. The PFS benefit was consistent with investigators assessment with a HR 0.59 (0.46-0.77). Chantal Decroisette Persée GFPC Survival table Endpoint: PFS by BCR (RECIST 1.1) Arm | N | Median | Median CI Pembro + CT | 160 | 16.1 | 11–34.1 Pembro | 161 | 6.9 | 5.5–9.3 HR: 0.56 HR CI: 0.43–0.74 P: P < 0.001 (two-sided) Landmark rates: 12 mo: Pembro + CT 54.6% | Pembro 33.9% 24 mo: Pembro + CT 45.2% | Pembro 24.9% 36 mo: Pembro + CT 39.2% | Pembro 20.8% [Slide 4] Secondary endpoint : Overall survival (ITT) Arm | Events n (%) | Median OS (95% CI), months | HR (95% CI) Pembro + CT | 74 (46.3) | 50.2 (32.6–NC) | 0.77 (0.56–1.05) Pembro | 89 (55.3) | 29.9 (23.4–47.9) Curve labels: Pembro + CT 78.8% / 64.4% / 55.4% / 50.9%; Pembro 73.3% / 56.2% / 44.9% / 38.9% No. at Risk: Pembro + CT 160 ...; Pembro 161 ... (remaining values too small to transcribe reliably) A numerically longer median OS of 50.2 months in the pembro-CT group was observed compared with 29.9 months in the pembro group. The trial was not powered to demonstrate an overall survival difference. Chantal Decroisette Persée GFPC Survival table Endpoint: Overall survival (ITT) Arm | N | Median | Median CI Pembro + CT | 160 | 50.2 | 32.6–NC Pembro | 161 | 29.9 | 23.4–47.9 HR: 0.77 HR CI: 0.56–1.05 Landmark rates: 12 mo: Pembro + CT 78.8% | Pembro 73.3% 24 mo: Pembro + CT 64.4% | Pembro 56.2% 36 mo: Pembro + CT 55.4% | Pembro 44.9% 48 mo: Pembro + CT 50.9% | Pembro 38.9%

Secondary endpoint : Overall survival (ITT) Arm | Events n (%) | Median OS (95% CI), months | HR (95% CI) Pembro + CT | 74 (46.3) | 50.2 (32.6-NC) | 0.77 (0.56-1.05) Pembro | 89 (55.3) | 29.9 (23.4-47.9) Curve labels: Pembro + CT 78.8% (72.7-85.4) at 12; 64.4% (57.3-72.2) at 24; 55.4% (47.9-63.9) at 36; 50.9% (43.1-60.1) at 48 Pembro 73.3% (66.8-80.5) at 12; 56.2% (49.0-64.5) at 24; 44.9% (37.5-53.7) at 36; 38.9% (30.8-49.0) at 48 Months since Randomization No. at Risk (0, 3, 6, ... 60) Pembro + CT | 160 | 151 | 144 | 136 | 126 | 121 | 110 | 108 | 102 | 90 | 84 | 64 | 55 | 46 | 39 | 28 | 26 | 15 | 11 | 5 | 1 Pembro | 161 | 145 | 134 | 126 | 118 | 108 | 103 | 97 | 87 | 74 | 68 | 57 | 45 | 39 | 31 | 22 | 19 | 10 | 4 | 2 | 0 A numerically longer median OS of 50.2 months in the pembro-CT group was observed compared with 29.9 months in the pembro group. The trial was not powered to demonstrate an overall survival difference. Chantal Decroisette Survival table Endpoint: Overall survival (ITT) Arm | N | Median | Median CI Pembro + CT | 160 | 50.2 | 32.6-NC Pembro | 161 | 29.9 | 23.4-47.9 HR: 0.77 HR CI: 0.56-1.05 Landmark rates: 12 months: Pembro + CT 78.8% (CI 72.7-85.4) | Pembro 73.3% (CI 66.8-80.5) 24 months: Pembro + CT 64.4% (CI 57.3-72.2) | Pembro 56.2% (CI 49.0-64.5) 36 months: Pembro + CT 55.4% (CI 47.9-63.9) | Pembro 44.9% (CI 37.5-53.7) 48 months: Pembro + CT 50.9% (CI 43.1-60.1) | Pembro 38.9% (CI 30.8-49.0)

[Slide 1] Treatment Exposure | Pembrolizumab + Carboplatin + Paclitaxel (N = 32) | Pembrolizumab + Carboplatin + Pemetrexed (N = 113) | Pembrolizumab + Cisplatin + Pemetrexed (N = 15) | Pembrolizumab Monotherapy (N = 161) Median number of cycles (Q1–Q3) Pembrolizumab | 7.5 (5.0–28.0) | 13.0 (7.0–32.0) | 34.0 (26.0–35.0) | 10.0 (4.0–23.0) Carboplatin | 4.0 (3.0–4.0) | 4.0 (4.0–4.0) | NA | NA Cisplatin | NA | NA | 4.0 (3.0–4.0) | NA Pemetrexed | NA | 8.0 (4.0–14.0) | 10.0 (6.0–16.0) | NA Paclitaxel | 4.0 (3.0–4.0) | NA | NA | NA Duration of treatment, months Median (95% CI) | 10.5 (8.5–16.1) [pembrolizumab plus chemotherapy groups] | 6.3 (4.7–8.4) [pembrolizumab monotherapy] HR (95% CI) | 0.57 (0.45–0.71) CI, confidence interval; HR, hazard ratio; N, total number of patients; NA, not applicable; Q, quartile. Duration of treatment is presented for the pembrolizumab plus chemotherapy and pembrolizumab monotherapy groups. Chantal Decroisette [Slide 3] The PFS benefit was observed across all subgroups Subgroup | # Event/# Total Pembro+CT | Pembro | Hazard ratio Est. (95%CI) Overall | 93/160 | 117/161 | 0.56 (0.43 ; 0.74) Age <70 yr | 71/126 | 86/116 | 0.48 (0.35 ; 0.67) Age ≥70 yr | 22/34 | 31/45 | 0.83 (0.47 ; 1.46) Sex Male | 60/107 | 78/110 | 0.55 (0.39 ; 0.77) Sex Female | 33/53 | 39/51 | 0.58 (0.36 ; 0.93) ECOG PS 0 | 43/76 | 50/75 | 0.58 (0.38 ; 0.88) ECOG PS 1 | 50/84 | 67/86 | 0.52 (0.35 ; 0.75) Stage 3 | 6/11 | 5/10 | 0.99 (0.30 ; 3.27) Stage 4 | 87/149 | 112/151 | 0.55 (0.41 ; 0.73) Smoker Non-smoker | 3/4 | 3/3 | 0.27 (0.03 ; 2.70) Smoker Smoker | 31/56 | 44/64 | 0.52 (0.33 ; 0.84) Smoker Quit smoking | 59/99 | 70/94 | 0.60 (0.42 ; 0.85) Liver metastasis Yes | 18/23 | 24/26 | 0.63 (0.32 ; 1.25) Liver metastasis No | 75/137 | 93/135 | 0.56 (0.41 ; 0.76) Bone metastasis Yes | 36/61 | 43/49 | 0.42 (0.27 ; 0.68) Bone metastasis No | 57/99 | 74/112 | 0.62 (0.43 ; 0.88) Brain metastasis Yes | 26/39 | 31/36 | 0.54 (0.31 ; 0.93) Brain metastasis No | 67/121 | 86/125 | 0.57 (0.42 ; 0.79) Histology Squamous | 23/31 | 28/32 | 0.70 (0.40 ; 1.24) Histology Non-squamous | 70/129 | 89/129 | 0.53 (0.39 ; 0.72) Tumor cell PD-L1 Between 50% and 74% | 32/54 | 44/58 | 0.46 (0.29 ; 0.74) Tumor cell PD-L1 Between 75% and 100% | 51/91 | 65/95 | 0.62 (0.43 ; 0.91) Tumor cell PD-L1 Between 90% and 100% | 26/52 | 39/57 | 0.60 (0.36 ; 1.00) Number of metastatic sites ≤1 | 17/38 | 19/41 | 0.84 (0.44 ; 1.63) Number of metastatic sites >1 | 76/122 | 98/120 | 0.45 (0.33 ; 0.62) Number of metastatic sites ≤3 | 70/126 | 89/129 | 0.57 (0.41 ; 0.78) Number of metastatic sites >3 | 23/34 | 28/32 | 0.53 (0.30 ; 0.94) Pembro+CT Better <- -> Pembro Better Chantal Decroisette

[Slide 2] Potential explanations 1. Conventional imaging modalities currently leveraged to define the oligometastatic state cannot detect micro-metastatic disease 2. Spatial tumor heterogeneity is inherent in the oligometastic state and could lead to differential responses to radiation in the patient 3. Despite earlier data suggesting synergy, there has been more recent data suggesting a negative interaction with IO and radiation [Slide 3] My take 1. These results are not a fluke 2. Escalation is not for everyone 3. Future trials in this space will be informative [Slide 4] INSIGNA TRIAL DESIGN Stratification • PD-L1 status (tumor proportion score 1%-49% vs ≥ 50%) • Smoking status (ever vs never) RANDOMIZATION Arm A | 1st line treatment*: Pembrolizumab 200 mg IV over 30 min on day 1 | 2nd line treatment§: Pemetrexed 500 mg/m² IV over 10 min followed by Carboplatin AUC 5 IV over 30 min on day 1. Repeat for 4 cycles or until disease progression Arm B | 1st line treatment*: Pembrolizumab 200 mg IV over 30 min on day 1 | 2nd line treatment§: Pembrolizumab 200 mg IV over 30 min followed by Pemetrexed 500 mg/m² IV over 10 min followed by Carboplatin AUC 5 IV over 30 min on day 1. Repeat for 4 cycles or until disease progression Arm C | 1st line treatment: Induction (4 cycles)¶: Pembrolizumab 200 mg IV over 30 min followed by Pemetrexed 500 mg/m² IV over 10 min followed by Carboplatin AUC 5 IV over 30 min on day 1 → Maintenance**: Pembrolizumab 200 mg IV over 30 min followed by Pemetrexed 500 mg/m² IV over 10 min on day 1 | 2nd line treatment: Not specified Accrual goal = 846 patients. Cycle = 3 weeks (21 d). *Repeat until progression or maximum of 2 years. If maximum treatment duration is reached prior to PD, or treatment is discontinued for any other reason, patient will remain in observation until progression. If patient doesn't continue onto 2nd line treatment, he/she will proceed to long-term follow-up. §Following completion of 2nd line treatment, patient will proceed to long-term follow-up. **Repeat for 2 years of total treatment across induction and maintenance, or until disease progression. If after 2 years there is no progression, pemetrexed will continue per standard of care. Slide courtesy of Hoss Borghaei

[Slide 2] The PFS benefit was observed regardless of PD-L1 expression level PD-L1 50% - 74% Arm | Events n (%) | mPFS months (IC 95 %) | HR (IC 95 %) Pembro + CT | 32 (59.3) | 17.8 (9.3–NC) | 0.46 (0.29–0.74) Pembro | 44 (75.9) | 7.0 (3.6–10) | PD-L1 75% - 100% Arm | Events n (%) | mPFS months (IC 95 %) | HR (IC 95 %) Pembro + CT | 51 (56.0) | 16.9 (11.1–NC) | 0.62 (0.43–0.91) Pembro | 65 (68.4) | 7.1 (5.1–11.2) | PD-L1 90% - 100% Arm | Events n (%) | mPFS months (IC 95 %) | HR (IC 95 %) Pembro + CT | 26 (50.0) | 25.4 (11–NC) | 0.6 (0.36–1) Pembro | 39 (68.4) | 7.1 (4.9–18.8) | [KM curves with landmark labels and No. at Risk rows: illegible] Chantal Decroisette [Slide 3] Secondary endpoint : Overall survival (ITT) Arm | Events n (%) | Median OS (95% CI), months | HR (95% CI) Pembro + CT | 74 (46.3) | 50.2 (32.6–NC) | 0.77 (0.56–1.05) Pembro | 89 (55.3) | 29.9 (23.4–47.9) | Landmarks (Pembro + CT): 78.8%; 64.4%; 55.4%; 50.9% Landmarks (Pembro): 73.3%; 56.2%; 44.9%; 38.9% Months since Randomization No. at Risk: [small; consistent with higher-resolution copy 2114289340218323040_s4] A numerically longer median OS of 50.2 months in the pembro-CT group was observed compared with 29.9 months in the pembro group. The trial was not powered to demonstrate an overall survival difference. Chantal Decroisette Survival table Endpoint: Overall survival (ITT) Arm | N | Median | Median CI Pembro + CT | 160 | 50.2 months | 32.6–NC Pembro | 161 | 29.9 months | 23.4–47.9 HR: 0.77 HR CI: 0.56–1.05 Landmark rates: 12 months: Pembro + CT 78.8% | Pembro 73.3% 24 months: Pembro + CT 64.4% | Pembro 56.2% 36 months: Pembro + CT 55.4% | Pembro 44.9% 48 months: Pembro + CT 50.9% | Pembro 38.9% [Slide 4] Conclusion • PERSEE is the first positive randomized phase III trial comparing pembrolizumab plus chemotherapy with pembrolizumab alone as 1st-line treatment in advanced NSCLC with PD-L1 ≥50%, with a superiority design, and PFS as primary endpoint • Pembrolizumab plus chemotherapy significantly improved PFS, with a durable and clinically meaningful benefit (16.1 vs 6.9 months; HR 0.56, 95% CI 0.43-0.74; P<0.001) • The PFS benefit was associated with a higher ORR (77.5% vs 50.3%) and a substantial reduction in early disease progression (5.8% vs 26.5%) • OS was numerically improved with pembro-CT : median OS of 50.2 months vs 29.9 months with pembro (HR, 0.77; 95% CI, 0.56–1.05) • These efficacy gains were achieved at the cost of increased treatment-related toxicity, particularly grade ≥3 TRAEs (54% vs 19%), largely attributable to chemotherapy • QoL analyses and translational studies exploring immunosenescence and predictive biomarkers are ongoing PERSEE supports first line treatment with pembrolizumab-chemotherapy which should be routinely considered in patients eligible for chemotherapy Chantal Decroisette

Patient Demographic and Disease Characteristics (ITT Population) From December 22, 2020, to December 21, 2023, a total of 321 patients at 28 centers in France were randomized. | Pembrolizumab plus Chemotherapy (N = 160) | Pembrolizumab Monotherapy (N = 161) Age, years Median (range) | 63 (38-83) | 64 (39-86) <70 years | 126 (78.8) | 116 (72.0) Sex Male | 107 (66.9) | 110 (68.3) ECOG performance status 0 | 76 (47.5) | 75 (46.6) 1 | 84 (52.5) | 86 (53.4) Disease stage III | 11 (6.9) | 10 (6.2) IV | 149 (93.2) | 151 (93.7) Smoking status# Current smoker | 56 (35.2) | 64 (39.8) Former smoker | 99 (62.3) | 94 (58.4) Non-smoker | 4 (2.5) | 3 (1.9) Histologic features Non squamous | 129 (80.6) | 129 (80.1) Squamous | 31 (19.4) | 32 (19.9) Metastases Median metastasis sites No. (range) | 2 (0-8) | 2 (0-7) >3 metastatic sites | 34 (21.3) | 32 (19.9) Brain | 39 (24.4) | 36 (22.4) Liver | 23 (14.4) | 26 (16.1) Bone | 61 (38.1) | 49 (30.4) PD-L1 ≥50%* 50-74% | 54 (33.8) | 58 (36.0) 75-100% | 91 (56.9) | 95 (59.0) 90-100% | 52 (32.5) | 57 (35.4) Not evaluated | 15 (9.4) | 8 (5.0) NGS mutation, n/N (%) No | 54 / 139 (38.8) | 59 / 136 (43.4) Yes | 85 / 139 (61.2) | 77 / 136 (56.6) Missing | 21 | 25 Mutations identified, n/N (%) KRAS mutation | 58 / 139 (41.7) | 57 / 136 (41.9) Data are n (%) unless otherwise specified. ECOG, Eastern Cooperative Oncology Group; n, number of patients; N, total number of patients; PD-L1, Programmed Death-Ligand 1. # Smoking status missing in one patient in chemotherapy group * PD-L1 expression level locally measured "Not evaluated" indicates that the measurement was reported as ≥50% without an exact value Chantal Decroisette

[Slide 2] 10:15 - 11:45 Proffered paper 1: NSCLC, metastatic CHAIR: SURESH SENAN Primary endpoint: PFS by BCR (RECIST 1.1) Median follow-up was 38.7 (35.3-41.6) months Arm | Events n (%) | Median PFS (95% CI), months | HR (95% CI) Two-sided P value Pembro + CT | 93 (58.1) | 16.1 (11–34.1) | 0.56 (0.43–0.74) Pembro | 117 (72.7) | 6.9 (5.5–9.3) | P < 0.001 Percentage of Patients Months since Randomization No. at Risk PFS assessed by BCR was significantly longer in the pembro-CT group than in the pembro group, with a clinically meaningful benefit sustained over time. The PFS benefit was consistent with investigators assessment with a HR 0.59 (0.46-0.77). Chantal Decroisette Chantal Decroisette Pembrolizumab with or without platinum-based chemotherapy as first-line treatment in advanced NSCLC with PD-L1 ≥50%: PERSEE phase 3 trial (GFPC 01-2020) ALICANTE AUDITORIUM - HALL 6 Survival table Endpoint: PFS by blinded central review (RECIST 1.1) Pembro + CT | 160 | 16.1 | 11–34.1 Pembro | 161 | 6.9 | 5.5–9.3 HR: 0.56 HR CI: 0.43–0.74 P: <0.001 (two-sided) 12 mo | Pembro + CT: 54.6% (47.3–63.1) | Pembro: 33.9% (27.1–42.5) 24 mo | Pembro + CT: 45.2% (37.9–53.9) | Pembro: 24.9% (18.7–33.1) 36 mo | Pembro + CT: 39.2% (31.8–48.4) | Pembro: 20.8% (15.0–28.9) [Slide 3] 10:15 - 11:45 Proffered paper 1: NSCLC, metastatic CHAIR: SURESH SENAN The PFS benefit was observed regardless of PD-L1 expression level PD-L1 50% - 74% Arm | Events n (%) | mPFS months (IC 95 %) | HR (IC 95 %) Pembro + CT | 32 (59.3) | 17.8 (9.3–NC) | 0.46 (0.29–0.74) Pembro | 44 (75.9) | 7.0 (3.6–10) PD-L1 75% - 100% Pembro + CT | 51 (56.0) | 16.9 (11.1–NC) | 0.62 (0.43–0.91) Pembro | 65 (68.4) | 7.1 (5.1–11.2) PD-L1 90% - 100% Pembro + CT | 26 (50.0) | 25.4 (11–NC) | 0.6 (0.36–1) Pembro | 39 (68.4) | 7.1 (4.9–18.8) Percentage of Patients Months since Randomization No. at Risk Chantal Decroisette Chantal Decroisette Pembrolizumab with or without platinum-based chemotherapy as first-line treatment in advanced NSCLC with PD-L1 ≥50%: PERSEE phase 3 trial (GFPC 01-2020) ALICANTE AUDITORIUM - HALL 6 Survival table Endpoint: PFS by PD-L1 subgroup PD-L1 50-74%: Pembro + CT | 17.8 | 9.3–NC PD-L1 50-74%: Pembro | 7.0 | 3.6–10 PD-L1 75-100%: Pembro + CT | 16.9 | 11.1–NC PD-L1 75-100%: Pembro | 7.1 | 5.1–11.2 PD-L1 90-100%: Pembro + CT | 25.4 | 11–NC PD-L1 90-100%: Pembro | 7.1 | 4.9–18.8 [Slide 4] 10:15 - 11:45 Proffered paper 1: NSCLC, metastatic CHAIR: SURESH SENAN Secondary endpoint : Overall survival (ITT) Arm | Events n (%) | Median OS (95% CI), months | HR (95% CI) Pembro + CT | 74 (46.3) | 50.2 (32.6–NC) | 0.77 (0.56–1.05) Pembro | 89 (55.3) | 29.9 (23.4–47.9) Probability of Overall Survival Months since Randomization No. at Risk A numerically longer median OS of 50.2 months in the pembro-CT group was observed compared with 29.9 months in the pembro group. The trial was not powered to demonstrate an overall survival difference. Chantal Decroisette Chantal Decroisette Pembrolizumab with or without platinum-based chemotherapy as first-line treatment in advanced NSCLC with PD-L1 ≥50%: PERSEE phase 3 trial (GFPC 01-2020) ALICANTE AUDITORIUM - HALL 6 Survival table Endpoint: Overall survival (ITT) Pembro + CT | 160 | 50.2 | 32.6–NC Pembro | 161 | 29.9 | 23.4–47.9 HR: 0.77 HR CI: 0.56–1.05 12 mo | Pembro + CT: 78.8% (72.7–85.4) | Pembro: 73.3% (66.8–80.5) 24 mo | Pembro + CT: 64.4% (57.3–72.2) | Pembro: 56.2% (49.0–64.5) 36 mo | Pembro + CT: 55.4% (47.9–63.9) | Pembro: 44.9% (37.5–53.7) 48 mo | Pembro + CT: 50.9% (43.1–60.1) | Pembro: 38.9% (30.8–49.0)

ESMO ANNALS OF ONCOLOGY ORIGINAL ARTICLE · Articles in Press, October 25, 2026 Open Access Pembrolizumab with or without platinum-based chemotherapy as first-line treatment in advanced NSCLC expressing PD-L1 ≥50%: The PERSEE randomized phase III trial (GFPC 01-2020) Renaud Descourt 1,# · Alice Mogenet 2 · Maelys Consigny 3 · ... · Christos Chouaïd 7,29 · Laurent Greillier 2 · Chantal Decroisette 30,# ... Show more

[Slide 1] PFS2 is prolonged with Pembro-CT Arm | Events n (%) | Median PFS2 (95% CI), months | HR (95% CI) Two-sided P value Pembro + CT | 79 (49.3) | 39 (26.2–NR) | 0.68 (0.5-0.91) Pembro | 101 (62.7) | 15.9 (13.2–29.9) | Landmarks (Pembro + CT): 70.6% (63.8–78.0); 59.1% (51.9–67.3); 51.1% (43.7–59.7) Landmarks (Pembro): 60.6% (53.4–68.7); 43.3% (36.2–51.8); 37.0% (30.0–45.6) Months since Randomization No. at Risk Pembro + CT 160 150 140 126 111 106 99 97 92 80 72 58 49 42 37 28 25 15 11 5 1 Pembro 161 145 130 116 96 80 75 69 64 57 54 46 38 33 26 20 17 9 4 2 0 Chantal Decroisette Survival table Endpoint: PFS2 Arm | N | Median | Median CI Pembro + CT | 160 | 39 months | 26.2–NR Pembro | 161 | 15.9 months | 13.2–29.9 HR: 0.68 HR CI: 0.5-0.91 Landmark rates: 12 months: Pembro + CT 70.6% (CI 63.8–78.0) | Pembro 60.6% (CI 53.4–68.7) 24 months: Pembro + CT 59.1% (CI 51.9–67.3) | Pembro 43.3% (CI 36.2–51.8) 36 months: Pembro + CT 51.1% (CI 43.7–59.7) | Pembro 37.0% (CI 30.0–45.6) [Slide 2] Safety | Overall | Pembro + CT | Pembro Intention-to-Treat population | N = 321 | N = 160 | N = 161 General Any grade AE, Any cause | 320 (100) | 160 (100) | 160 (99) Grade ≥3 AE | 214 (67) | 118 (74) | 96 (60) Serious AE | 179 (56) | 95 (59) | 84 (52) Safety population | N = 317 | N = 158 | N = 159 Treatment-related AE Any grade | 268 (85) | 149 (94) | 119 (75) Grade ≥3 | 117 (37) | 86 (54) | 31 (19) Grade 5 | 6 (1.9) | 3 (1.9) | 3 (1.9) Immune related AE Any grade | 71 (22) | 39 (25) | 32 (20) Grade ≥3 | 41 (13) | 21 (13) | 20 (13) Grade 5 | 2 (0.6) | 0 (0) | 2 (1.3) Treatment action (All AEs) Leading to pembro discontinuation | 82 (26) | 44 (28) | 38 (24) Leading to pembro interruption | 117 (37) | 78 (49) | 39 (25) Leading to CT discontinuation | 76 (24) | 76 (48) | 0 (0) Leading to CT interruption | 72 (23) | 72 (46) | 0 (0) Leading to CT dose reduction | 72 (23) | 72 (46) | 0 (0) No unexpected toxicity was observed. More grade 3 TRAEs were reported in the pembro-CT group (54%) compared to the pembro group (19%). Immune-related AEs occurred at similar frequency in both study groups. 48% of patients discontinued chemotherapy due to adverse events. Chantal Decroisette [Slide 3] Pembro-CT resulted in higher ORR, longer DoR, and lower Early Progression Rate Pembrolizumab + chemotherapy / Pembrolizumab monotherapy Objective response rate (ORR): OR 3.48 (95% CI, 2.15-5.72); 124 (77.5%) vs 81 (50.3%) Early disease progression*: OR 0.17 (95% CI, 0.07-0.35); 9 (5.8%) vs 41 (26.5%) (* based on non missing values : pembrolizumab plus chemotherapy, n = 154; pembrolizumab monotherapy, n = 155) DoR Pembrolizumab + chemotherapy: 30.0 (13.1-NR) Pembrolizumab: 15.6 (9.8-32.8) Time (months) Number at risk Pembrolizumab: 81 73 60 49 42 38 37 33 29 26 25 22 19 17 10 7 3 2 0 0 0 Pembrolizumab + chemotherapy: 124 114 93 82 76 69 66 61 53 44 40 33 27 21 15 13 9 9 4 0 0 Chantal Decroisette Abbreviations :OR: odd ratio; DoR: duration of response

ESMO congress MADRID 2026 Pembrolizumab with or without platinum-base[d chemotherapy] as first-line treatment in advanced NSCLC with [...] PERSEE phase III trial (GFPC 01-2020) Decroisette C1, Descourt R2, Bernardi M3, Corre R4, Bylicki O5, Auliac JB6, Vergnenègre A7, Ricordel C8, Pinsol[...] L13, Guisier F14, Hauss PA15, Bizieux-Thaminy A16, Hominal S17, Consigny M18, Chouaid C5, Greillier L19 GFPC | Persée GFPC October 25th 2026

LBA66 - Sacituzumab tirumotecan (sac-TMT) plus pembrolizumab (P) versus chemotherapy (chemo) plus pembrolizumab (P) as first-line (1L) treatment for PD-L1-negative advanced non-squamous (nsq) NSCLC: Randomized phase 3 OptiTROP-Lung06 study Speakers Caicun Zhou (Shanghai, China) Lecture Time 10:53 - 11:03 Background Pembrolizumab plus chemotherapy is the standard 1L treatment for PD-L1-negative advanced nsq-NSCLC without actionable genomic alterations, but outcomes remain suboptimal. Sac-TMT, a TROP2-directed ADC, previously showed encouraging activity with PD-1 blockade in 1L PD-L1-positive NSCLC (Xiong et al., Lancet, 2026). We report the results from the preplanned interim PFS analysis of OptiTROP-Lung06 (NCT06711900). Methods Eligible patients (pts) had treatment-naïve, locally advanced or metastatic nsq-NSCLC without EGFR/ALK alterations and PD-L1 tumor proportion score (TPS) <1% using the 22C3 assay. Pts were randomized 1:1 to receive sac-TMT 4 mg/kg Q2W + P 400 mg Q6W or chemo + P. The primary endpoint was PFS per RECIST 1.1 assessed by blinded independent central review (BICR); OS was secondary. Results A total of 455 pts (median age 63 yrs; 72.1% male; 82.4% ECOG 1; 91.6% stage IV) were randomized to receive sac-TMT + P (n=229) or chemo + P (n=226). As of Jun 9, 2026, median follow-up was 9.7 mo. Sac-TMT + P showed a highly statistically significant and clinically meaningful improvement in PFS by BICR compared with chemo + P (median: 15.0 vs 5.7 mo; HR, 0.49; 95% CI 0.36-0.65; p<0.0001). OS data were immature, but a favorable trend was observed in the sac-TMT + P group (HR, 0.58; 95% CI 0.38-0.88). ORR by BICR was 50.2% vs 47.3%, and median DoR was 13.6 vs 6.8 mo (HR 0.47, 95% CI 0.29-0.77). Grade ≥3 TEAEs were less frequent with sac-TMT + P than with chemo+ P (47.6% vs 60.6%), including neutrophil count decreased (19.7% vs 28.8%), white blood cell count decreased (9.6% vs 16.4%), anemia (6.1% vs 20.4%), and platelet count decreased (3.9% vs 14.6%). Sac-TMT-associated AESIs were manageable; grade ≥3 stomatitis occurred in 2.2%. Conclusions Sac-TMT+ P significantly prolonged PFS versus chemo + P, with longer response duration and a manageable safety profile. To our knowledge, this is the first randomized phase 3 study showing superior PFS with a TROP2-directed ADC plus PD-1 blockade versus an immunotherapy-plus-chemo standard therapy in NSCLC, supporting sac-TMT +P as a potential 1L treatment option for this population.

Sacituzumab tirumotecan+pembrolizumab Chemotherapy+pembrolizumab 12-month labels: 77.6 (95% CI, 69.5–83.8); 66.4 (95% CI, 57.6–73.8) | No. of Patients with Event/No. of Patients (%) | Median Overall Survival (95% CI) mo Sacituzumab Tirumotecan+Pembrolizumab | 36/229 (15.7) | NE (NE–NE) Chemotherapy+Pembrolizumab | 55/226 (24.3) | NE (15.6–NE) Hazard ratio for death, 0.58 (95% CI, 0.38–0.88) No. at Risk (months 0-18) Sacituzumab tirumotecan+pembrolizumab 229 228 228 215 200 184 166 148 130 109 92 79 60 51 36 24 21 7 4 Chemotherapy+pembrolizumab 226 226 220 210 188 167 152 136 122 98 85 67 51 39 25 17 10 3 1 Survival table Endpoint: Overall survival Arm | N | Median | Median CI Sacituzumab tirumotecan + pembrolizumab | 229 | NE | NE–NE Chemotherapy + pembrolizumab | 226 | NE | 15.6–NE HR: 0.58 HR CI: 0.38–0.88 Landmark rates: 12 months: Sacituzumab tirumotecan + pembrolizumab 77.6% (CI 69.5–83.8) | Chemotherapy + pembrolizumab 66.4% (CI 57.6–73.8)

[Slide 2] OptiTROP-Lung06 Study Design Randomized, multicenter, open-label, phase 3 trial (NCT06711900) Key eligibility criteria • Locally advanced (stage IIIB/IIIC) or metastatic (stage IV) nsq-NSCLC • No prior systemic therapy for advanced disease • No sensitizing EGFR or ALK alteration • PD-L1 TPS< 1% (IHC 22C3, central lab) • ECOG PS score 0 or 1 Stratification factors • Sex (male vs. female) • ECOG PS score (0 vs. 1) • Disease stage (IIIB/IIIC vs. IV) R 1:1, n = 455 Sac-TMT 4 mg/kg Q2W + Pembro 400 mg Q6Wa (n = 229) Pemetrexed 500 mg/m² + Carboplatin AUC 5 or Cisplatin 75 mg/m² Q3Wb + Pembro 400 mg Q6Wa (n = 226) Treatment until disease progression or unacceptable toxicity Endpoints c — Primary: • PFS assessed by BICR; Secondary endpoints included: • OS • PFS assessed by investigator • ORR, DCR, DOR • Safety a Pembro was administered for a maximum of 18 cycles (6 weeks per cycle). b Platinum chemotherapy was administered for up to 4 doses Q3W (two 6-week cycles), followed by maintenance pemetrexed plus pembrolizumab. c Tumor response was assessed using RECIST version 1.1 every 6 weeks for the first 54 weeks after randomization, then every 12 weeks. Professor Caicun Zhou [Slide 3] PFS by BICR Sac-TMT + pembro significantly improved PFS vs. Chemo + pembro, with a 51% reduction in risk of disease progression or death. | Sac-TMT + Pembro (n = 229) | Chemo + Pembro (n = 226) PFS events, n (%) | 79 (34.5) | 126 (55.8) Median, mo (95% CI) | 15.0 (9.9–NE) | 5.7 (5.5–7.4) HR 0.49 (95% CI: 0.36–0.65) P < 0.0001a 12-month landmark (dashed line): Sac-TMT+Pembro 51.1%; Chemo+Pembro 22.8% Time in Months: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 At Risk Sac-TMT+Pembro | 229 224 197 173 162 141 107 88 77 60 45 41 28 15 15 12 2 2 2 Chemo+Pembro | 226 221 192 157 141 109 77 64 56 38 25 22 16 5 4 4 1 1 0 PFS by investigator assessment was consistent with the results of BICR: HR 0.52 (95% CI: 0.40–0.69) . a Updated efficacy boundary (corresponding to 205 actual PFS events by BICR): 0.0160 (2-sided). NE, not estimable. Professor Caicun Zhou Survival table Endpoint: PFS by BICR Arm | N | Median | Median CI Sac-TMT + Pembro | 229 | 15.0 | 9.9–NE Chemo + Pembro | 226 | 5.7 | 5.5–7.4 HR: 0.49 HR CI: 0.36–0.65 P: P < 0.0001 (efficacy boundary 0.0160 2-sided) Landmark rates: 12 months: Sac-TMT + Pembro 51.1% | Chemo + Pembro 22.8% [Slide 4] Descriptive Overall Survival at PFS IA A favorable trend in overall survival was observed with sac-TMT + pembro over chemo + pembro. | Sac-TMT + Pembro (n = 229) | Chemo + Pembro (n = 226) OS events, n (%) | 36 (15.7) | 55 (24.3) Median, mo (95% CI) | NR (NE–NE) | NR (15.6–NE) HR 0.58 (95% CI: 0.38–0.88) 12-month landmark (dashed line): Sac-TMT+Pembro 77.6%; Chemo+Pembro 66.4% Time in Months: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 At Risk Sac-TMT+Pembro | 229 228 228 215 200 184 166 148 130 109 92 79 60 51 36 24 21 7 4 Chemo+Pembro | 226 226 220 210 188 167 152 136 122 98 85 67 51 39 25 17 10 3 1 Median follow-up was 9.7 months. Professor Caicun Zhou Survival table Endpoint: Overall survival (descriptive, at PFS interim analysis) Arm | N | Median | Median CI Sac-TMT + Pembro | 229 | NR | NE–NE Chemo + Pembro | 226 | NR | 15.6–NE HR: 0.58 HR CI: 0.38–0.88 Landmark rates: 12 months: Sac-TMT + Pembro 77.6% | Chemo + Pembro 66.4%

[Slide 1] Phase 2 Study 2870-002/SKB264-II-06 (NCT05642780) Key Eligibility Criteria for UC Cohort • Aged ≥18 y • Histologically or cytologically confirmed locally advanced or metastatic urothelial carcinoma of the renal pelvis, ureter, bladder, or urethraa • Not eligible for cisplatin-containing chemotherapyb • No prior systemic therapy for advanced or unresectable urothelial carcinomac • Measurable lesion per RECIST v1.1 • ECOG PS 0 or 1 → Sac-TMT 4 mg/kg or 5 mg/kg IV Q2Wd + Pembrolizumab 400 mg IV Q6W; N = 40 Endpoints Primary • Safety, including AEs and AEs leading to treatment discontinuation • ORR per RECIST v1.1 by investigator Secondary • DCR per RECIST v1.1 by investigator • DOR per RECIST v1.1 by investigator • PFS per RECIST v1.1 by investigator • OS aParticipants with mixed histology were eligible provided that the urothelial component was >50% and the plasmacytoid component was <10% per local pathology assessment. Participants with non-urothelial cancer of the urinary tract were not eligible. bDefined as meeting ≥1 of the following criteria: creatinine clearance (calculated or measured) <60 mL/min but >30 mL/min, grade ≥2 audiometric hearing loss, grade ≥2 peripheral neuropathy, or NYHA Class III heart failure. cPlatinum-based therapy or nivolumab used in the adjuvant or neoadjuvant setting with recurrence >12 months since completion of therapy was permitted. Low-dose chemotherapy given concurrently with radiation to the primary tumor site was not considered systemic therapy. dParticipants received sac-TMT until PD or unacceptable toxicity and pembrolizumab until PD, unacceptable toxicity, or completion of 2 years of treatment. [Slide 2] ORR, DCR, and DOR by TROP2 and PD-L1 | Sac-TMT + pembro (N = 40) Confirmed ORR (95% CI), % | 68 (51–81) Subgroup, % (95% CI) [responders/N] TROP2 H-score <median | 72 (47–90) [13/18] TROP2 H-score ≥median | 68 (43–87) [13/19] PD-L1 CPS <10 | 72 (53–86) [23/32] PD-L1 CPS ≥10 | 50 (16–84) [4/8] Confirmed DCR (95% CI), % | 88 (73–96) Subgroup, % (95% CI) [responders/N] TROP2 H-score <median | 89 (65–99) [16/18] TROP2 H-score ≥median | 89 (67–99) [17/19] PD-L1 CPS <10 | 94 (79–99) [30/32] PD-L1 CPS ≥10 | 63 (25–92) [5/8] DOR, median (range), mo | 19.4 (2.0–25.8+) Subgroup, median, (range), mo [no. of responders] TROP2 H-score <median | NR (2.0–25.8+) [13] TROP2 H-score ≥median | 15.4 (3.6–22.2+) [13] PD-L1 CPS <10 | 19.4 (2.0–25.8+) [23] PD-L1 CPS ≥10 | 9.0 (3.9–22.1+) [4] Median study follow-up duration, 22.8 (range, 19.8–30.0) months. Median TROP2 H-score = 270. 95% CIs for ORR and DCR are based on the Clopper-Pearson method. Median DORs are based on the Kaplan-Meier method for censored data. Data cutoff, Jan 26, 2026. [Slide 3] PFS by TROP2 and PD-L1 [Left KM: PFS, % vs Time, mo] Median (95% CI), mo TROP2 H-score <median | 11.2 (3.7–NR) TROP2 H-score ≥median | 11.9 (5.6–21.1) No. at risk (0, 3, 6, 9, 12, 15, 18, 21, 24, 27, 30) <median | 18 16 11 9 7 7 7 6 3 1 0 ≥median | 19 17 12 10 9 8 6 4 0 0 0 [Right KM: PFS, % vs Time, mo] Median (95% CI), mo PD-L1 CPS <10 | 15.6 (7.2–NR) PD-L1 CPS ≥10 | 5.7 (0.5–NR) No. at risk (0, 3, 6, 9, 12, 15, 18, 21, 24, 27, 30) CPS <10 | 32 30 22 18 15 15 13 9 4 1 0 CPS ≥10 | 8 5 3 3 3 2 2 2 0 0 0 • In the overall cohort, median PFS was 11.2 (95% CI, 5.8–NR) months and median OS was NR (95% CI, 17.2 mo to NR) Median study follow-up duration, 22.8 (range, 19.8–30.0) months. Median PFS and OS are based on the Kaplan-Meier method for censored data. Data cutoff, Jan 26, 2026. Survival table Endpoint: PFS by TROP2 H-score and PD-L1 CPS (single-arm sac-TMT + pembro, N=40) Arm | N | Median | Median CI Overall cohort | 40 | 11.2 | 5.8-NR TROP2 H-score <median | 18 | 11.2 | 3.7-NR TROP2 H-score ≥median | 19 | 11.9 | 5.6-21.1 PD-L1 CPS <10 | 32 | 15.6 | 7.2-NR PD-L1 CPS ≥10 | 8 | 5.7 | 0.5-NR

ORR, Deep Response, and DOR (BICR) ORR was similar between groups, with deeper and more durable responses observed with sac-TMT + pembro. | Sac-TMT + Pembro (n = 229) | Chemo + Pembro (n = 226) ORR, % (95% CI) | 50.2 (43.6–56.9) | 47.3 (40.7–54.1) Deep response, % | 28.8 | 17.3 DCR, % (95% CI) | 90.4 (85.8–93.9) | 88.5 (83.6–92.3) Median DOR, mo (95% CI) | 13.6 (9.7–NE) | 6.8 (4.2–10.6) HR (95% CI) | 0.47 (0.29–0.77) 12-mo DOR rate, % (95% CI) | 61.7 (47.5–73.1) | 27.2 (11.2–46.0) [DOR KM: HR 0.47 (95% CI: 0.29–0.77); 12-month 61.7% vs 27.2%] At Risk (Time in Months 0–13) Sac-TMT+Pembro | 115 103 92 77 72 59 46 37 32 25 10 8 7 5 Chemo+Pembro | 107 97 80 57 53 43 32 26 22 12 6 4 2 2 ORR: the proportion of patients with a CR/PR; Deep response: CR or PR with ≥ 50% reduction from baseline in the sum of target-lesion diameters. Percentages are based on the ITT population. CR, complete response; PR, partial response. Professor Caicun Zhou

[Slide 1] Conclusions • Sac-TMT + pembro demonstrated a statistically significant and clinically meaningful improvement in PFS compared with chemo + pembro. • Sac-TMT + pembro showed a higher deep response rate and longer DOR, reflecting deeper and more durable responses. • OS data were immature, but a favorable trend was observed (HR 0.58; 95% CI, 0.38–0.88). • The safety profile was manageable and consistent with the known profiles of the individual agents, with no new safety signals identified. Median PFS by BICR Sac-TMT + Pembro 15.0 mo Chemo + Pembro 5.7 mo HR 0.49; 95% CI: 0.36–0.65; P < 0.0001 Median DOR by BICR Sac-TMT + Pembro 13.6 mo Chemo + Pembro 6.8 mo HR 0.47; 95% CI: 0.29–0.77 Results from OptiTROP-Lung06 study support sac-TMT + pembro as a potential new first-line treatment option for PD-L1-negative advanced nonsquamous NSCLC without EGFR or ALK alterations Professor Caicun Zhou [Slide 4] Summary of Safety Grade ≥3 TEAEs were less frequent with sac-TMT + pembro (47.6% vs. 60.6%), particularly hematologic events. Event, n (%) | Sac-TMT + Pembro (n = 229) | Chemo + Pembro (n = 226) TEAEs | 228 (99.6) | 225 (99.6) Grade ≥3 TEAEs | 109 (47.6) | 137 (60.6) Serious TEAEs | 82 (35.8) | 82 (36.3) TEAEs Leading to dose reductiona | 11 (4.8) | 45 (19.9) TEAEs Leading to discontinuation of any study drug | 13 (5.7) | 32 (14.2) TEAEs Leading to death | 5 (2.2) | 14 (6.2) TRAEs Leading to death | 1 (0.4) | 3 (1.3) Any grade TEAEs with incidence ≥ 20% in either treatment group (Sac-TMT+Pembro any grade | grade ≥3 || Chemo+Pembro grade ≥3 | any grade, %): Anaemia 83 | 6 || 20 | 85 Alopecia 62 | 0 || 0 | 14 WBC decreased 48 | 10 || 16 | 65 NEUT decreased 47 | 20 || 29 | 65 Stomatitis 45 | 2 || 0 | 7 Decreased appetite 40 | 0 || 0 | 45 Asthenia 31 | 2 || 2 | 34 Nausea 31 | 0 || 0 | 31 Hypoalbuminaemia 30 | 0 || 0 | 27 ALT increased 29 | 0 || 3 | 37 AST increased 26 | 1 || 1 | 38 Constipation 21 | 0 || 0 | 23 Rash 21 | 1 || 0 | 14 PLT decreased 19 | 4 || 15 | 51 LDH increased 17 | 0 || 0 | 25 a Only dose of sac-TMT and chemotherapy was allowed to be modified. Professor Caicun Zhou

[Slide 1] Phase 2 Study 2870-002/SKB264-II-06 (NCT05642780) Key Eligibility Criteria for UC Cohort • Aged ≥18 y • Histologically or cytologically confirmed locally advanced or metastatic urothelial carcinoma of the renal pelvis, ureter, bladder, or urethra(a) • Not eligible for cisplatin-containing chemotherapy(b) • No prior systemic therapy for advanced or unresectable urothelial carcinoma(c) • Measurable lesion per RECIST v1.1 • ECOG PS 0 or 1 Sac-TMT 4 mg/kg or 5 mg/kg IV Q2W(d) + Pembrolizumab 400 mg IV Q6W N = 40 Endpoints Primary • Safety, including AEs and AEs leading to treatment discontinuation • ORR per RECIST v1.1 by investigator Secondary • DCR per RECIST v1.1 by investigator • DOR per RECIST v1.1 by investigator • PFS per RECIST v1.1 by investigator • OS [Slide 2] ORR, DCR, and DOR by TROP2 and PD-L1 | Sac-TMT + pembro (N = 40) Confirmed ORR (95% CI), % | 68 (51–81) Subgroup, % (95% CI) [responders/N] TROP2 H-score <median | 72 (47–90) [13/18] TROP2 H-score ≥median | 68 (43–87) [13/19] PD-L1 CPS <10 | 72 (53–86) [23/32] PD-L1 CPS ≥10 | 50 (16–84) [4/8] Confirmed DCR (95% CI), % | 88 (73–96) Subgroup, % (95% CI) [responders/N] TROP2 H-score <median | 89 (65–99) [16/18] TROP2 H-score ≥median | 89 (67–99) [17/19] PD-L1 CPS <10 | 94 (79–99) [30/32] PD-L1 CPS ≥10 | 63 (25–92) [5/8] | Sac-TMT + pembro (N = 40) DOR, median (range), mo | 19.4 (2.0–25.8+) Subgroup, median, (range), mo [no. of responders] TROP2 H-score <median | NR (2.0–25.8+) [13] TROP2 H-score ≥median | 15.4 (3.6–22.2+) [13] PD-L1 CPS <10 | 19.4 (2.0–25.8+) [23] PD-L1 CPS ≥10 | 9.0 (3.9–22.1+) [4] [Slide 3] PFS by TROP2 and PD-L1 Left panel: Median (95% CI), mo TROP2 H-score <median | 11.2 (3.7–NR) TROP2 H-score ≥median | 11.9 (5.6–21.1) Time, mo: 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 | 24 | 27 | 30 No. at risk TROP2 <median | 18 | 16 | 11 | 9 | 7 | 7 | 7 | 6 | 3 | 1 | 0 TROP2 ≥median | 19 | 17 | 12 | 10 | 9 | 8 | 6 | 4 | 0 | 0 | 0 Right panel: Median (95% CI), mo PD-L1 CPS <10 | 15.6 (7.2–NR) PD-L1 CPS ≥10 | 5.7 (0.5–NR) Time, mo: 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 | 24 | 27 | 30 No. at risk PD-L1 CPS <10 | 32 | 30 | 22 | 18 | 15 | 15 | 13 | 9 | 4 | 1 | 0 PD-L1 CPS ≥10 | 8 | 5 | 3 | 3 | 3 | 2 | 2 | 2 | 0 | 0 | 0 Survival table Endpoint: PFS by biomarker subgroup (single-arm, sac-TMT + pembrolizumab) Arm | N | Median | Median CI TROP2 H-score <median | 18 | 11.2 mo | 3.7–NR TROP2 H-score ≥median | 19 | 11.9 mo | 5.6–21.1 PD-L1 CPS <10 | 32 | 15.6 mo | 7.2–NR PD-L1 CPS ≥10 | 8 | 5.7 mo | 0.5–NR

Baseline Characteristics Characteristic | Sac-TMT + Pembro (n = 229) | Chemo + Pembro (n = 226) Age | Median (range), years | 63 (29, 75) | 63 (37, 75) Age | ≥ 65 | 103 (45.0) | 104 (46.0) Sex | Male | 165 (72.1) | 163 (72.1) ECOG PS score | 1 | 189 (82.5) | 186 (82.3) Smoking history | Current or former smoker | 150 (65.5) | 142 (62.8) Smoking history | Never smoker | 79 (34.5) | 84 (37.2) Histology | Adenocarcinoma | 228 (99.6) | 226 (100) Histology | Large-cell carcinoma | 1 (0.4) | 0 Disease stage | IIIB/IIIC | 19 (8.3) | 19 (8.4) Disease stage | IV | 210 (91.7) | 207 (91.6) Metastases | ≥ 3 Distant metastatic sites | 52 (22.7) | 58 (25.7) Metastases | Brain | 5(2.2) | 6 (2.7) Metastases | Liver | 17(7.4) | 26 (11.5) Metastases | Bone | 75 (32.8) | 89 (39.4) Professor Caicun Zhou

Figure 1. Progression-free Survival as Assessed by Blinded Independent Central Review. No. of Patients with Event/No. of Patients (%) | Median Progression-free Survival (95% CI), mo Sacituzumab Tirumotecan+Pembrolizumab | 79/229 (34.5) | 15.0 (9.9–NE) Chemotherapy+Pembrolizumab | 126/226 (55.8) | 5.7 (5.5–7.4) Hazard ratio for disease progression or death, 0.49 (95% CI, 0.36–0.65) 12-month: Sacituzumab tirumotecan+pembrolizumab 51.1 (95% CI, 41.5–59.9); Chemotherapy+pembrolizumab 22.8 (95% CI, 15.3–31.3) No. at Risk Sacituzumab tirumotecan+pembrolizumab 229 224 197 173 162 141 107 88 77 60 45 41 28 15 15 12 2 2 2 Chemotherapy+pembrolizumab 226 221 192 157 141 109 77 64 56 38 25 22 16 5 4 4 1 1 0 Shown are the Kaplan–Meier estimates of progression-free survival as assessed by blinded independent central review in the intention-to-treat population (which included all the patients who had undergone randomization) as of the data-cutoff date of June 9, 2026. Progression-free survival was defined as the time from randomization to the first documented occurrence of disease progression or death from any cause, whichever occurred first. Tumor assessments were performed according to the Response Evaluation Criteria in Solid Tumors, version 1.1, by blinded independent central review. Confidence intervals have not been adjusted for multiplicity and should not be used in place of hypothesis testing. NE denotes not estimable. Survival table Endpoint: PFS by BICR Arm | N | Median | Median CI Sacituzumab tirumotecan + pembrolizumab | 229 | 15.0 | 9.9–NE Chemotherapy + pembrolizumab | 226 | 5.7 | 5.5–7.4 HR: 0.49 HR CI: 0.36–0.65 Landmark rates: 12 months: Sacituzumab tirumotecan + pembrolizumab 51.1% (CI 41.5–59.9) | Chemotherapy + pembrolizumab 22.8% (CI 15.3–31.3)

[Slide 1] OptiTROP-Lung06 Study Design Randomized, multicenter, open-label, phase 3 trial (NCT06711900) Key eligibility criteria • Locally advanced (stage IIIB/IIIC) or metastatic (stage IV) nsq-NSCLC • No prior systemic therapy for advanced disease • No sensitizing EGFR or ALK alteration • PD-L1 TPS< 1% (IHC 22C3, central lab) • ECOG PS score 0 or 1 R 1:1, n = 455 Sac-TMT 4 mg/kg Q2W + Pembro 400 mg Q6W(a) (n = 229) Pemetrexed 500 mg/m2 + Carboplatin AUC 5 or Cisplatin 75 mg/m2 Q3W(b) + Pembro 400 mg Q6W(a) (n = 226) Treatment until disease progression or unacceptable toxicity Endpoints(c) Primary • PFS assessed by BICR Secondary endpoints included • OS • PFS assessed by investigator • ORR, DCR, DOR • Safety Stratification factors • Sex (male vs. female) • ECOG PS score (0 vs. 1) • Disease stage (IIIB/IIIC vs. IV) a Pembro was administered for a maximum of 18 cycles (6 weeks per cycle). b Platinum chemotherapy was administered for up to 4 doses Q3W (two 6-week cycles), followed by maintenance pemetrexed plus pembrolizumab. c Tumor response was assessed using RECIST version 1.1 every 6 weeks for the first 54 weeks after randomization, then every 12 weeks. ALK, anaplastic lymphoma kinase; BICR, blinded independent central review; DCR, disease control rate; DOR, duration of response; ECOG, Eastern Cooperative Oncology Group; EGFR, epidermal growth factor receptor; ORR, objective response rate; Pembro, Pembrolizumab; R, randomization; RECIST, Response Evaluation Criteria in Solid Tumors; TPS, tumor proportion score. Professor Caicun Zhou [Livestream frame: 10:15 - 11:45 Proffered paper 1: NSCLC, metastatic; CHAIR: SURESH SENAN; speaker card: Caicun Zhou — Sacituzumab tirumotecan (sac-TMT) plus pembrolizumab (P) versus chemotherapy (chemo) plus pembrolizumab (P) as first-line (1L) treatment for PD-L1-negative advanced non-squamous (nsq) NSCLC: Randomized phase 3 OptiTROP-Lung06 study] [Slide 3] PFS by BICR Sac-TMT + pembro significantly improved PFS vs. Chemo + pembro, with a 51% reduction in risk of disease progression or death. | Sac-TMT + Pembro (n = 229) | Chemo + Pembro (n = 226) PFS events, n (%) | 79 (34.5) | 126 (55.8) Median, mo (95% CI) | 15.0 (9.9–NE) | 5.7 (5.5–7.4) HR 0.49 (95% CI: 0.36–0.65) P < 0.0001(a) 12-month labels on curve: 51.1% ; 22.8% Legend: Sac-TMT+Pembro; Chemo+Pembro At Risk Sac-TMT+Pembro 229 224 197 173 162 141 107 88 77 60 45 41 28 15 15 12 2 2 2 Chemo+Pembro 226 221 192 157 141 109 77 64 56 38 25 22 16 5 4 4 1 1 0 PFS by investigator assessment was consistent with the results of BICR: HR 0.52 (95% CI: 0.40–0.69). a Updated efficacy boundary (corresponding to 205 actual PFS events by BICR): 0.0160 (2-sided). NE, not estimable. Professor Caicun Zhou [Livestream frame: 10:15 - 11:45 Proffered paper 1: NSCLC, metastatic; CHAIR: SURESH SENAN; speaker card: Caicun Zhou — Sacituzumab tirumotecan (sac-TMT) plus pembrolizumab (P) versus chemotherapy (chemo) plus pembrolizumab (P) as first-line (1L) treatment for PD-L1-negative advanced non-squamous (nsq) NSCLC: Randomized phase 3 OptiTROP-Lung06 study] Survival table Endpoint: PFS by BICR Arm | N | Median | Median CI Sac-TMT + Pembro | 229 | 15.0 mo | 9.9–NE Chemo + Pembro | 226 | 5.7 mo | 5.5–7.4 HR: 0.49 HR CI: 0.36–0.65 P: P < 0.0001 Landmark rates: 12 mo (dashed line): Sac-TMT + Pembro 51.1% | Chemo + Pembro 22.8%

PFS (BICR) in Subgroups A generally consistent PFS benefit of sac-TMT + pembro over chemo + pembro was observed across subgroups. Subgroup | Number of Events/Patients Sac-TMT+Pembro | Chemo+Pembro | Hazard Ratio (95% CI) All subjects | 79/229 | 126/226 | 0.49 (0.36-0.65) Sex Male | 53/165 | 99/163 | 0.39 (0.28-0.55) Female | 26/64 | 27/63 | 0.88 (0.51-1.51) Age <65 years | 46/126 | 75/122 | 0.44 (0.30-0.64) ≥ 65 years | 33/103 | 51/104 | 0.57 (0.37-0.88) ECOG Performance Status 0 | 13/40 | 22/40 | 0.49 (0.24-1.00) 1 | 66/189 | 104/186 | 0.49 (0.36-0.66) Clinical stage at enrollment Stage IV | 75/210 | 120/207 | 0.49 (0.37-0.66) Smoking history Current or former smoker | 47/150 | 88/142 | 0.36 (0.25-0.51) Never smoker | 32/79 | 38/84 | 0.80 (0.50-1.29) Liver metastasis No | 73/212 | 108/200 | 0.52 (0.39-0.70) Number of distant metastatic sites <3 | 52/158 | 78/149 | 0.52 (0.36-0.74) ≥3 | 23/52 | 42/58 | 0.47 (0.28-0.78) TROP2 expression Low (H-score ≤200) | 26/77 | 47/87 | 0.49 (0.30-0.80) High (H-score >200) | 32/89 | 47/83 | 0.59 (0.37-0.92) Unknown | 21/63 | 32/56 | 0.38 (0.22-0.67) <--Sac-TMT+Pembro Better Chemo+Pembro Better--> HR along with the 95% CI were estimated using unstratified Cox Proportional Hazards model in each subgroup. Subgroup analyses will not be presented for any subgroup variable in which one or more levels account for less than 10% of the total analysis population. TROP-2 expression was evaluated in formalin-fixed, paraffin-embedded tumor specimens by immunohistochemistry. Professor Caicun Zhou

[Slide 1] Percentage of Patients Months Sacituzumab tirumotecan + pembrolizumab Chemotherapy + pembrolizumab 51.1 (95% CI, 41.5–59.9) 22.8 (95% CI, 15.3–31.3) No. of Patients with Event/No. of Patients (%) | Median Progression-free Survival (95% CI) mo Sacituzumab Tirumotecan + Pembrolizumab 79/229 (34.5) | 15.0 (9.9–NE) Chemotherapy + Pembrolizumab 126/226 (55.8) | 5.7 (5.5–7.4) Hazard ratio for disease progression or death, 0.49 (95% CI, 0.36–0.65) No. at Risk Survival table Endpoint: Progression-free survival Arm | N | Median | Median CI Sacituzumab tirumotecan + pembrolizumab | 229 | 15.0 mo | 9.9–NE Chemotherapy + pembrolizumab | 226 | 5.7 mo | 5.5–7.4 HR: 0.49 HR CI: 0.36–0.65 12-month PFS (printed at dashed line at 12 mo): 51.1% (95% CI 41.5–59.9) vs 22.8% (95% CI 15.3–31.3) No. at risk sac-TMT+pembro (0-18 mo, monthly): 229,224,197,173,162,141,107,88,77,60,45,41,28,15,15,12,2,2,2 No. at risk chemo+pembro: 226,221,192,157,141,109,77,64,56,38,25,22,16,5,4,4,1,1,0 [Slide 2] Percentage of Patients Months Sacituzumab tirumotecan + pembrolizumab Chemotherapy + pembrolizumab 77.6 (95% CI, 69.5–83.8) 66.4 (95% CI, 57.6–73.8) No. of Patients with Event/No. of Patients (%) | Median Overall Survival (95% CI) mo Sacituzumab Tirumotecan + Pembrolizumab 36/229 (15.7) | NE (NE–NE) Chemotherapy + Pembrolizumab 55/226 (24.3) | NE (15.6–NE) Hazard ratio for death, 0.58 (95% CI, 0.38–0.88) No. at Risk Survival table Endpoint: Overall survival Arm | N | Median | Median CI Sacituzumab tirumotecan + pembrolizumab | 229 | NE | NE–NE Chemotherapy + pembrolizumab | 226 | NE | 15.6–NE HR: 0.58 HR CI: 0.38–0.88 12-month OS (printed at dashed line at 12 mo): 77.6% (95% CI 69.5–83.8) vs 66.4% (95% CI 57.6–73.8) No. at risk sac-TMT+pembro (0-18 mo, monthly): 229,228,228,215,200,184,166,148,130,109,92,79,60,51,36,24,21,7,4 No. at risk chemo+pembro: 226,226,220,210,188,167,152,136,122,98,85,67,51,39,25,17,10,3,1

Summary of Treatment-Related(a) AEs | Sac-TMT 4 mg/kg + pembro (n = 26) | Sac-TMT 5 mg/kg + pembro (n = 14) | Sac-TMT + pembro (N = 40) Treatment duration, median (range), mo Sac-TMT | 11.8 (0.03–25.1) | 10.0 (0.5–29.1) | 10.9 (0.03–29.1) Pembrolizumab | 9.2 (0.03–22.8) | 9.3 (0.03–23.5) | 9.3 (0.03–23.5) Treatment-related AEs | 25 (96) | 14 (100) | 39 (98) Grade 3 or 4 | 17 (65) | 8 (57) | 25 (63) Serious | 3 (12) | 3 (21) | 6 (15) Led to any treatment discontinuation | 2 (8) | 4 (29) | 6 (15) Led to sac-TMT discontinuation | 1 (4)(b) | 2 (14)(c) | 3 (8) Led to pembrolizumab discontinuation | 1 (4)(d) | 2 (14)(e) | 3 (8) Led to death | 0 | 0 | 0 Treatment-Related AEs Occurring in ≥25% of Participants (any grade, %): Anemia 73 | Stomatitis 73 | Alopecia 43 | Hypoalbuminemia 43 | Decreased appetite 40 | Rash 40 | Neutrophil count decreased 35 | Blood creatinine increased 33 | WBC count decreased 33 | Hyperglycemia 30 | Lymphocyte count decreased 30 | Urinary tract infection 28 | Proteinuria 25 Legend: Grade 1 or 2 | Grade 3 or 4

[logos: SARON, Cancer Research UK, Cancer Trials Centre, UCL, RTTQA Radiotherapy Trials Quality Assurance] Stereotactic Ablative Radiotherapy (SABR) for Oligometastatic NSCLC Lung primary (+/- nodes) suitable for radical RT or SABR, with 1-5 synchronous oligometastatic disease ↓ 6-8 weeks induction systemic anti-cancer therapy (SACT) ↓ 140 patients randomised 1:1 Investigational arm: CFRT (Conventional fractionated radiotherapy) or SABR to primary (+/- nodes) & SABR / SRS (Stereotactic radiosurgery) to metastases Maximum of further 4-6 weeks of induction SACT + maintenance Control arm: Maximum of further 4-6 weeks of induction SACT + maintenance Thoracic SABR Safety Sub-Study (TSSS) Patients treated CFRT to primary (+/- nodes) & SABR to thoracic metastases [CT dose-wash images: sagittal, coronal and axial planes] Fig 1: Example of CFRT to primary & SABR to thoracic spine Report acute toxicity (<3months post radiotherapy) and dosimetric association

[Slide 1] Background and Rationale SARON Local consolidative therapy (LCT) with radiotherapy or surgery improves outcomes in patients with oligo-metastatic / oligo-persistent NSCLC treated with chemotherapy 1-2 LCT with radiotherapy improves outcomes in patients with synchronous oligo-metastatic, oligo-organ & poly-metastatic EGFR mutated NSCLC treated with TKI 3-5 LCT with radiotherapy or surgery is not associated with improved outcomes in patients with oligo-persistent NSCLC treated with chemotherapy & immunotherapy 6-7 Do patients with synchronous oligometastatic, actionable genomic alteration (AGA)-negative, NSCLC benefit from the addition of consolidation radiotherapy to chemotherapy and / or immunotherapy? 1Gomez et al Lancet Oncol 2016; 2Iyengar et al JAMA Oncol 2018; 3Wang et al JNCI 2023; 4Sun et al JCO 2024; 5Elamin et al ESMO 2025; 6Iyengar et al ASCO 2024; 7Altan et al WCLC 2026 Podium: ESMO congress MADRID SPAIN 23-27 OCTOBER 2026 [Slide 2] ESMO congress MADRID 2026 CANCER RESEARCH UK | Cancer Trials Centre | UCL | RTTQA Radiotherapy Trials Quality Assurance Primary results of SARON Stereotactic Ablative Radiotherapy for Oligometastatic Non-small cell lung cancer A randomised-controlled trial of consolidation radiotherapy for synchronous oligometastatic non-small cell lung cancer Fiona McDonald, Nicholas Counsell, Shahreen Ahmad, Merina Ahmed, Philip Atherton, Andrew Bates, Charles Comins, John Conibear, Corinne Faivre-Finn, Kevin Franks, Gerard G. Hanna, Stephen Harrow, Matthew Hatton, Ka Man Mak, Philip Mayles, Rushil Patel, Clive Peedell, Tony Pope, Martin Forster, David Landau Dr Fiona McDonald, Consultant Thoracic Oncologist Royal Marsden Hospital, London, UK October 2026 FPN: 2613O British Thoracic Oncology Group (BTOG) [Slide 3] Trial Design: Randomised Phase II NSCLC without driver mutation Primary tumour (+/- nodes) suitable for radical RT or SABR 1-5 synchronous metastases in up to 3 organs* treatable by SABR / SRS PS 0-1, PET-CT and brain CT/MRI staged N=200 Registration -> 6-8 weeks induction systemic anti-cancer therapy (SACT) -> PS 0-2, CT scan: CR/PR/SD -> N=134 R 1:1 PS >=3 or PD -> Withdrawal Investigational arm: Max of further 4-6 weeks of induction SACT -> Conventional RT or SABR to primary tumour (+/- nodes) + SABR / SRS to metastases Control arm: Max of further 4-6 weeks of induction SACT Follow-up: Follow-up for 3 years or until death +/- maintenance SACT Randomisation (R) using minimisation stratified by: histology, N-stage, number of metastases, brain metastases, SACT & recruiting site *Brain metastases can be included if at least one extra-cranial metastasis is also present RT = radiotherapy; SABR = stereotactic ablative radiotherapy; SRS = stereotactic radiosurgery; PS = performance status; CR = complete response; PR = partial response; SD = stable disease; PD = progressive disease [Slide 4] Overall Survival SARON mOS 28.0 months in SACT arm mOS 31.0 months in SACT+RT arm HR=0.99 (95%CI: 0.63-1.55), p=0.95 Time from Randomisation (Months) # at risk (0, 12, 24, 36, 48, 60, 72, 84) Control: 70 | 51 | 38 | 15 | 4 | 2 | 2 | 0 Investigational: 70 | 55 | 33 | 21 | 0 | 0 | 0 | 0 Survival table Endpoint: Overall survival SACT (control) | 70 | 28.0 months SACT+RT (investigational) | 70 | 31.0 months HR: 0.99 HR CI: 0.63-1.55 P: 0.95

[Slide 1] 10:15 - 11:45 Proffered paper 1: NSCLC, metastatic CHAIR: SURESH SENAN SARON Overall Survival mOS 28.0 months in SACT arm mOS 31.0 months in SACT+RT arm HR=0.99 (95%CI: 0.63–1.55), p=0.95 # at risk (months 0,12,24,36,48,60,72,84) Control: 70 | 51 | 38 | 15 | 4 | 2 | 2 | 0 Investigational: 70 | 55 | 33 | 21 | 0 | 0 | 0 | 0 ALICANTE AUDITORIUM - HALL 6 Survival table Endpoint: Overall survival SACT (control) | 70 | 28.0 SACT+RT (investigational) | 70 | 31.0 HR: 0.99 HR CI: 0.63–1.55 P: 0.95 [Slide 2] 10:15 - 11:45 Proffered paper 1: NSCLC, metastatic CHAIR: SURESH SENAN SARON Patterns of Disease Progression Progression at Sites of Initial Disease: 39.5% in SACT; 13.0% in SACT+RT; Gray's Test p<0.001 # at risk: Control: 70 | 44 | 27 | 17 | 13; Investigational: 70 | 56 | 39 | 30 | 22 Progression at New Sites of Disease: 32.3% in SACT; 36.2% in SACT+RT; Gray's Test p=0.28 # at risk: Control: 70 | 44 | 27 | 17 | 13; Investigational: 70 | 56 | 39 | 30 | 22 [Cumulative incidence curves; x-axis Time from Randomisation (Months) 0, 12, 24] ALICANTE AUDITORIUM - HALL 6 [Slide 3] 10:15 - 11:45 Proffered paper 1: NSCLC, metastatic CHAIR: SURESH SENAN SARON Conclusion • Patients with synchronous oligometastatic AGA-negative NSCLC receiving first-line SACT achieved better control of initial disease sites and longer PFS with the addition of consolidation RT compared to SACT alone with acceptable adverse events & similar QoL • However, this did not translate into OS benefit ALICANTE AUDITORIUM - HALL 6

[Slide 1] Why local therapy in advanced disease? Patterns of failure Residual primary and nodes are the commonest site of first progression, including on TKIs and immunotherapy Resistant clones and reseeding Ablating persistent disease removes drug-tolerant clones before they seed new metastases Symptoms from the primary Haemoptysis, obstruction, cough and pain: thoracic RT remains the most reliable palliation Immune modulation RT may prime anti-tumour immunity with checkpoint inhibitors (PEMBRO-RT pooled data), but be aware of the "dark side of radiotherapy"... 64% of patients with stage IV NSCLC progressing after first-line chemotherapy failed only at sites of original disease Rusthoven et al. Acta Oncologica, 2009; 48: 578-583 [Figure A/B: schematic of primary tumour, metastases and radiation therapy effect on reseeding; labels: Primary tumour, Metastases, Radiation therapy, (i), (ii)] McMahon RA, et al. J Thorac Oncol 2026;21:103643 [Slide 2] Radiotherapy to the primary in stage IV NSCLC: suggested clinical pathway All stage IV NSCLC: PET-CT, MRI brain, NGS (incl. fusions), PD-L1, lung function; thoracic MDT discussion before systemic therapy starts Symptomatic primary Palliative thoracic RT at any stage of the pathway • PS 2-4: 10 Gy/1, 17 Gy/2 or 20 Gy/5 • PS 0-1: consider 36-39 Gy/12-13. Don't delay systemic therapy [I, A] Synchronous oligometastatic (≤5 mets, ≤3 organs) Systemic therapy first; restage at about 3 months • No progression: MDT for LCT to primary, nodes and all mets. EGFR+: consider LCT on osimertinib. • Prefer a trial (SARON-type) [II, B] Polymetastatic, asymptomatic primary No routine thoracic RT outside a trial • Offer TOURIST-PRINCE, PRIME-LUNG or PREVENT [Trial] Oligoprogression (≤5 sites) SABR to progressing sites; continue current systemic therapy • Rebiopsy or ctDNA to exclude histological transformation or new drivers first [II, B] [Slide 3] Technical recommendations for radiotherapy to the primary Dose and fractionation • Primary + nodes: 60-66 Gy/30-33 or 55 Gy/20 (sequential) • Peripheral N0 primary: risk-adapted SABR (e.g. 54 Gy/3, 60 Gy/8 central) • Metastases: SABR per site; SRS for brain Planning • 4D-CT with PET fusion; target post-response residual disease • Plan-sum any prior thoracic RT • Lung V20 ≤30%, MLD ≤18-20 Gy (lower with ICI); protect PBT and oesophagus Systemic therapy interactions • ICI: generally continue • Chemo: sequential, not concurrent with radical dose • TKIs and anti-VEGF or bispecifics: usually peri-RT hold, avoid ultracentral SABR Follow-up and governance • Pneumonitis review at 6-12 weeks • CT every 3 months; MRI brain per risk • Prospective registry (e.g. OligoCare), ctDNA in research only [Slide 4] RT as local therapy in advanced lung cancer – take home messages Why: Most progression arises at known sites; LCT improves PFS in selected patients, OS benefit unproven with IO When: Consolidate after best response at about 3 months; SABR for oligoprogression; palliate symptoms at any time Who: Patients with good PS, PET/MRI-staged, ≤5 mets, responding, safely treatable. Beware early progressors, ILD, anti-VEGF + central disease Next: Recruit to TOURIST and await the results of SARON: they will tell us whether RT to the primary matters in the modern era

Progression-Free Survival SARON mPFS 8.9 months in SACT arm mPFS 13.9 months in SACT+RT arm HR=0.64 (95%CI: 0.43-0.95), p=0.03 Time from Randomisation (Months) # at risk Control: 70 | 43 | 26 | 16 | 12 | 9 | 6 | 0 | 0 Investigational: 70 | 55 | 38 | 29 | 21 | 14 | 11 | 0 | 0 Survival table Endpoint: Progression-free survival SACT (control) | 70 | 8.9 months SACT+RT (investigational) | 70 | 13.9 months HR: 0.64 HR CI: 0.43-0.95 P: 0.03

Patterns of Disease Progression SARON Progression at Sites of Initial Disease 39.5% in SACT 13.0% in SACT+RT Gray's Test p<0.001 Progression at New Sites of Disease 32.3% in SACT 36.2% in SACT+RT Gray's Test p=0.28 Y axis: Cumulative Incidence; X axis: Time from Randomisation (Months) 0, 12, 24 # at risk (both panels) Control: 70 | 44 | 27 | 17 | 13 Investigational: 70 | 56 | 39 | 30 | 22

[Slide 2] Demographics & Baseline Characteristics SARON | Control Arm (N=70) | Investigational Arm (N=70) Age years, median (range) | 68 (47-83) | 68 (45-85) Sex, n (%) Female | 27 (38.6) | 30 (42.9) Male | 43 (61.4) | 40 (57.1) Performance Status, n (%) 0 - Fully Active | 38 (54.3) | 31 (44.3) 1 - Restricted | 32 (45.7) | 39 (55.7) Disease Type*, n (%) Adenocarcinoma | 42 (60.0) | 40 (57.1) SCC | 20 (28.6) | 25 (35.7) NSCLC - NOS | 8 (11.4) | 4 (5.7) LCC | 0 (0.0) | 1 (1.4) *Squamous Cell Carcinoma (SCC); NSCLC - Not Otherwise Specified (NOS); Large Cell Carcinoma (LCC); | Control Arm (N=70) | Investigational Arm (N=70) T Stage, n (%) 1 | 11 (15.7) | 12 (17.1) 2 | 19 (27.1) | 23 (32.9) 3 | 17 (24.3) | 14 (20.0) 4 | 22 (31.4) | 21 (30.0) x | 1 (1.4) | 0 (0.0) N Stage, n (%) 0 | 22 (31.4) | 20 (28.6) 1 | 10 (14.3) | 14 (20.0) 2 | 32 (45.7) | 30 (42.9) 3 | 6 (8.6) | 6 (8.6) No. of metastases, n (%) 1 | 42 (60.0) | 46 (65.7) 2 | 17 (24.3) | 16 (22.9) 3 | 7 (10.0) | 6 (8.6) 4 | 3 (4.3) | 2 (2.9) 5 | 1 (1.4) | 0 (0.0) Brain metastases, n (%) | 4 (5.7) | 5 (7.1) [Slide 3] Treatment Details and Compliance SARON Induction SACT (both arms, n=140): | Control Arm (N=70) | Investigational Arm (N=70) | Total (N=140) Chemotherapy only, n (%) | 29 (41.4%) | 27 (38.6%) | 56 (40.0%) Chemo-immunotherapy, n (%) | 30 (42.9%) | 33 (47.1%) | 63 (45.0%) Immunotherapy only, n (%) | 11 (15.7%) | 10 (14.3%) | 21 (15.0%) Cycles received, median (range) | 4 (2-4) | 4 (2-5) | 4 (2-5) Radiotherapy (investigational arm only, n=70): 62 (88.6%) received radical radiotherapy to primary tumour (51 conventional and 11 SABR) 54 (77.1%) received SABR to at least 1 oligometastatic site 8 (11.4%) did not receive any radiotherapy: ineligible (n=5), patient/clinical decision (n=3)

Conclusion SARON - Patients with synchronous oligometastatic AGA-negative NSCLC receiving first-line SACT achieved better control of initial disease sites and longer PFS with the addition of consolidation RT compared to SACT alone with acceptable adverse events & similar QoL - However, this did not translate into OS benefit

Progression-Free Survival mPFS 8.9 months in SACT arm mPFS 13.9 months in SACT+RT arm HR=0.64 (95%CI: 0.43–0.95), p=0.03 Time from Randomisation (Months) # at risk Control: 70 43 26 16 12 9 6 0 0 Investigational: 70 55 38 29 21 14 11 0 0 Survival table Endpoint: Progression-free survival SACT (control) | 70 | 8.9 months SACT+RT (investigational) | 70 | 13.9 months HR: 0.64 HR CI: 0.43–0.95 P: 0.03

The genomes of metastasis diverge markedly from those of the primary tumor TRACER-X LUNG STUDY AND PEACE AUTOPSY PROGRAM • 500 longitudinally collected primary and metastatic tumor samples from 24 patients with NSCLC • 80 % of metastases contained a subclone that was not detected in any other metastasis • The number of subclones detected per metastasis increased with the number of metastasis regions sampled [Figure: per-patient sampling timelines (CRUKP IDs) vs time after primary surgery (days) and body map of sample sites; small print not transcribed] Hessey et al. Nature. May 2026

[Slide 2] PFS by BICR: Primary Endpoint Iza-bren showed statistically significant and clinically meaningful improvement vs PBC with doubling of mPFS. | Iza-bren (N = 216) | PBC (N = 206) Number of events | 128 | 164 Median, mo (95% CI) | 11.1 (9.2, 12.7) | 5.7 (5.5, 5.9) HR (95% CI) | 0.40 (0.31, 0.51)† P-value | <0.0001* 75.1% 58.2% 46.5% 42.9% 24.4% 14.4% Progression-free survival probability (%) Time (Months) Iza-bren PBC + Censored Subjects at risk PFS (BICR) was evaluated in the mITT population. The median duration of follow-up for PFS was 14.6 months. Similar results were observed for PFS (BICR) in the ITT population, with 129 events in the iza-bren arm and 164 events in the PBC arm. The hazard ratio (95% CI) was 0.40 (0.31, 0.51). †Based on a stratified Cox regression model. *One-sided, based on a stratified log-rank test. Li Zhang Data cut off: July 15, 2026 Survival table Endpoint: PFS by BICR (mITT) Arm | N | Median | Median CI Iza-bren | 216 | 11.1 | 9.2, 12.7 PBC (platinum-based chemotherapy) | 206 | 5.7 | 5.5, 5.9 HR: 0.40 HR CI: 0.31, 0.51 P: <0.0001 (one-sided, stratified log-rank) 6 mo: 75.1% vs 42.9% 9 mo: 58.2% vs 24.4% 12 mo: 46.5% vs 14.4% [Slide 3] OS: Key Secondary Endpoint OS data was not mature for both arms. At the interim analysis, a trend of OS benefit favoring iza-bren was observed. | Iza-bren (N = 216) | PBC (N = 206) Number of events (death) | 64 | 82 Median OS, mo (95% CI) | NR (19.3, NR) | 19.4 (16.4, 22.5) Median follow-up for OS, mo (95% CI) | 16.9 (15.9, 17.7) | 16.5 (15.1, 17.5) 12-mo OS rate (95% CI) | 83.0 (77.3, 87.5) | 72.9 (66.1, 78.5) 18-mo OS rate (95% CI) | 68.8 (60.9, 75.5) | 55.3 (46.8, 63.0) More patients received subsequent anti-cancer therapy in the PBC arm, including ADCs Percentage (%) ADC | 2.8% | 29.1% Targeted therapy | 25.9% | 30.1% Immune therapy | 17.6% | 31.1% Chemotherapy | 31.5% | 40.3% OS was evaluated in the mITT population. The overall median duration of follow-up for OS was 16.6 months. Similar results were observed for OS in the ITT population. Li Zhang Data cut off: July 15, 2026 [Slide 4] Safety Summary | Iza-bren (N = 216) | PBC (N = 206) TRAEs, n (%) | 216 (100) | 204 (99.0) ≥Grade 3 TRAEs, n (%) | 170 (78.7) | 128 (62.1) Treatment-related SAEs, n (%) | 92 (42.6) | 40 (19.4) TRAEs leading to death, n (%) | 1 (0.5) | 1 (0.5) TRAEs leading to treatment discontinuation, n (%) | 4 (1.9) | 4 (1.9) TRAEs leading to dose reduction, n (%) | 130 (60.2) | 48 (23.3) TRAEs leading to dose interruption, n (%) | 116 (53.7) | 49 (23.8) TRAE, treatment related adverse events. Safety data based on the safety analysis set. TRAEs leading to death: iza-bren arm, 1 patient died due to sepsis; in pbc arm, 1 patient died due to unknown cause. • Median (range) duration of treatment, mo: – Iza-bren: 10.3 (0.3, 22.7) – PBC: Pemetrexed 5.7 (0.2, 23.5); Cisplatin/Carboplatin 3.1 (0.2, 5.8) • Median (range) number of treatment cycles: – Iza-bren: 13 (1, 31) – PBC: Pemetrexed 8 (1, 33); Cisplatin/Carboplatin 4 (1, 7) • ≥Grade 3 TRAEs and treatment-related SAEs were higher in iza-bren group; the safety profile of iza-bren was manageable through standard supportive care including dose delay and dose reductions. • The rate of treatment discontinuation due to TRAEs was low in both groups (1.9%). Data cut off: July 15, 2026 Li Zhang

[Slide 3] ORR (BICR): Secondary Endpoint Iza-bren nearly doubled cORR by BICR versus PBC with improved duration of response. Iza-bren (N=216) | PBC (N=206) Best overall response, n (%) CR 1 (0.5) | 0 PR 129 (59.7) | 71 (34.5) Non-CR/Non-PD 1 (0.5) | 2 (1.0) SD 61 (28.2) | 102 (49.5) PD 9 (4.2) | 25 (12.1) NE 15 (6.9) | 6 (2.9) Confirmed ORR, % (95% CI) 60.2 (53.3, 66.8) | 34.5 (28.0, 41.4) Difference, % (95% CI) 25.6 (16.2, 34.6) Odds ratio (95% CI) 2.9 (1.9, 4.3) DCR, % (95% CI) 88.9 (83.9, 92.8) | 85.0 (79.3, 89.5) In the iza-bren arm, 91.2% of patients (197/216) had tumor shrinkage and the median (range) shrinkage (%) was -46.8 (-100, -0.5); in the PBC arm, 78.6% of patients (162/206) had tumor shrinkage and the median (range) shrinkage (%) was -33.5 (-94.4, -0.5). Iza-bren | PBC Median DoR, mo (95% CI) 10.4 (9.2, 12.7) | 5.7 (4.4, 7.1) HR (95% CI) 0.41 (0.29, 0.60) Duration of response rate (%) Time (Months) ORR, objective response rate (confirmed); DCR, disease control rate; DoR, duration of response; CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease; NE, not evaluable. Data cut off: July 15, 2026 Li Zhang [Slide 4] Conclusions • Iza-bren demonstrated a statistically significant and clinically meaningful improvement in PFS vs PBC in patients with EGFR-mutant non-squamous NSCLC after progression on a 3rd generation EGFR TKI • At this interim analysis, OS remained immature and a trend towards OS improvement with iza-bren was observed. • Iza-bren demonstrated an improved response rate with durable response vs. PBC. • The safety profile of iza-bren was manageable and consistent with previous studies. PFS by BICR Iza-bren 11.1 mo PBC 5.7 mo HR: 0.40, 95% CI: 0.31 - 0.51, P < 0.0001 OS Iza-bren Not reached PBC 19.4 mo cORR by BICR Iza-bren 60.2% PBC 34.5% Difference: 25.6%, 95% CI: 16.2 - 34.6 > These results support iza-bren as a promising new treatment option for EGFR-mutant NSCLC patients with EGFR TKI resistance. > Global registrational studies in EGFR-mutant NSCLC patients after progression on a third-generation EGFR TKI and in the first line setting are ongoing (IZABRIGHT-Lung01, NCT07100080; IZABRIGHT-Lung02, NCT 07680790). Li Zhang

LBA71 - PANKU-Lung01: Randomized phase III study of izalontamab brengitecan (iza-bren) in EGFR-mutant non-squamous non-small cell lung cancer (NSCLC) after progression on a third-generation EGFR TKI Speakers Li Zhang (Guangzhou, China) Lecture Time 09:14 - 09:24 Authors Li Zhang (Guangzhou, China), Yunpeng Yang (Guangzhou, China), Xiangjiao Meng (Jinan, China), Zhihuang Hu (Shanghai, China), Wei Zheng (Shenyang, China), Jiuwei Cui (Changchun, China), Yongzhong Luo (Changsha, China), Lin Wu (Changsha, China), Meijuan Huang (Chengdu, China), Haipeng Xu (Fuzhou, China), Jie Hu (Shanghai, China), Yanqiu Zhao (Zhengzhou, China), Wenxiu Yao (Chengdu, China), Sa Xiao (Chengdu, China), Hai Zhu (Redmond, United States of America), Yi Zhu (Chengdu, China) Abstract Background Patients (pts) with advanced/metastatic EGFR-mutant NSCLC who have progressed on an EGFR TKI have limited treatment options. Iza-bren is a first-in-class bispecific antibody-drug conjugate (ADC) targeting EGFR and HER3 conjugated to a potent topoisomerase I inhibitor (Ed-04) via tetrapeptide-based cleavable linker. Methods Eligible pts had locally advanced or metastatic non-squamous NSCLC with EGFR sensitizing mutant (exon 19 deletion or L858R), and progression on prior 3rd generation EGFR TKI. Pts were randomized 1:1 to receive iza-bren (2.5 mg/kg IV D1D8 Q3W) or platinum-based chemotherapy (PBC) (pemetrexed+cisplatin/carboplatin, Q3W). The primary endpoint was PFS assessed by blind independent central review (BICR); secondary endpoint included OS. A pre-planned interim analysis served as the final analysis for PFS and the interim analysis for OS. Results As of data cutoff (Jul 15, 2026), 422 pts were treated (iza-bren/PBC, 216/206). Median follow-up for PFS was 14.6 vs 15.2 mo (iza-bren/PBC). mPFS by BICR was 11.14 mo (95% CI, 9.17-12.68) with iza-bren vs 5.65 mo (95% CI, 5.45-5.95) with PBC (HR = 0.40; 95% CI, 0.31-0.51; p <0.0001). The OS data were immature; however, a favorable trend towards improved mOS was observed with iza-bren vs PBC (HR= 0.66; 95% CI, 0.47–0.91). The cORR by BICR was 60.2% vs 34.5%, and the mDOR was 10.45 vs 5.68 mo. Gr≥3 TRAEs were predominantly hematologic in nature. TRAEs were manageable with standard supportive care, and only 1.9% of pts in the iza-bren group discontinued treatment due to TRAEs. ILD was uncommon with iza-bren (adjudicated ILD, two [0.9%]; one Gr 3 and one Gr 1). No new safety signals were observed. Conclusions Iza-bren demonstrated a statistically significant and clinically meaningful improvement in PFS vs PBC in pts with EGFR-mutant non-squamous NSCLC after progression on a 3rd generation EGFR TKI, meeting the primary endpoint. A trend towards OS improvement was observed. The safety profile was manageable and consistent with previous studies. These results support iza-bren as a new standard of care in this population in China. A global study is ongoing (NCT07100080).

THE LANCET This journal Journals Publish Clinical Global health Multimedia Events About ARTICLES · Online first, October 26, 2026 Izalontamab brengitecan versus platinum-based chemotherapy in EGFR-mutated advanced non-squamous non-small-cell lung cancer after progression on third-generation EGFR tyrosine kinase inhibitors (PANKU-Lung01): a multicentre, randomised, open-label, phase 3 trial in China Shaodong Hong, MD a,* · Ting Zhou, MD a,* · Xue Hou, MD a,* · Prof Xiangjiao Meng, MD b,* · Zhihuang Hu, MD c,* · Prof Wei Zheng, MD d,* · et al. Show more Affiliations & Notes Article Info

Background Iza-bren, EGFR×HER3 bispecific ADC αEGFR Human EGFR Affinity: High DAR = 8 Cat B cleavable linker Ed-04 (top1i) wt Fc IgG1 αHER3 Human HER3 Affinity: Low • EGFR TKI is the standard therapy for advanced EGFR-mutant NSCLC. However, most patients inevitably develop resistance1. Following disease progression on 3rd generation TKIs, platinum-based chemotherapy remains a common treatment option2. Therapeutic options with improved efficacy are still needed. • Izalontamab brengitecan (iza-bren, BL-B01D1) is a first-in-class bispecific ADC targeting EGFRxHER3 bound to a novel topo I inhibitor (Ed-04) via a cleavable linker. • In previous phase I and II studies, iza-bren showed promising antitumor activity in patients with EGFR-mutant NSCLC after EGFR TKI therapy (cORR, 56.0%; mDOR, 13.7 months; mPFS, 12.5 months)3. Here we report the interim analysis results from a randomized phase III study evaluating the efficacy and safety of iza-bren vs PBC in patients with EGFR-mutant NSCLC after progression on 3rd generation EGFR TKI (PANKU-Lung01). Wt, wild type; Cat B, cathepsin B 1. Lung Cancer 194 (2024) 107895; 2. Lung Cancer 181 (2023) 107248; 3. Ann Oncol. 2026 Jun;37(6):813-824 Li Zhang

PFS by BICR: Primary Endpoint Iza-bren (N = 216) | PBC (N = 206) Number of events | 128 | 164 Median, mo (95% CI) | 11.1 (9.2, 12.7) | 5.7 (5.5, 5.9) HR (95% CI) | 0.40 (0.31, 0.51)† P-value | <0.0001* Progression-free survival probability (%) Time (Months) Iza-bren PBC Censored Subjects at risk Survival table Endpoint: PFS by BICR (primary) Arm | N | Median | Median CI Iza-bren | 216 | 11.1 mo | 9.2-12.7 PBC | 206 | 5.7 mo | 5.5-5.9 HR: 0.40 HR CI: 0.31-0.51 P: <0.0001 6 mo: Iza-bren 75.1% vs PBC 42.9% 9 mo: 58.2% vs 24.4% 12 mo: 46.5% vs 14.4%

TRAEs in ≥ 20% of Patients Preferred Term | Iza-bren Grade ≥3 | Iza-bren Grade 1–2 | PBC Grade 1–2 | PBC Grade ≥3 Anemia | 33.3 | 58.4 | 71.8 | 18.0 Thrombocytopenia | 38.4 | 44.5 | 48.0 | 20.9 Leukopenia | 42.1 | 39.8 | 53.9 | 19.9 Neutropenia | 46.3 | 31.5 | 36.4 | 37.4 Alopecia | | 77.3 | 14.6 | Decreased appetite | 1.9 | 61.1 | 44.6 | 1.0 Asthenia | 11.1 | 50.9 | 44.6 | 3.9 Nausea | 2.8 | 49.1 | 40.8 | ALT increased | 1.4 | 38.4 | 39.8 | 2.4 Constipation | 0.5 | 38.9 | 27.7 | AST increased | 1.9 | 32.8 | 42.2 | 0.5 Diarrhea | 3.2 | 30.1 | 5.8 | 0.5 Hypoalbuminemia | 0.5 | 31.9 | 16.0 | Weight decreased | | 32.4 | 7.8 | Stomatitis | 2.9 | 29.1 | 4.4 | Lymphocyte count decreased | 7.4 | 23.2 | 17.9 | 4.9 Vomiting | 1.9 | 26.8 | 20.9 | 0.5 Incidence (%) Iza-bren | Grade 1–2; Iza-bren | Grade ≥3; PBC | Grade 1–2; PBC | Grade ≥3 • Most common TRAEs were hematologic toxicities, which were effectively managed with standard supportive care including growth factor support, dose delay and reductions, which allowed majority of patients to continue iza-bren treatment with a low discontinuation rate. Primary G-CSF prophylaxis was not mandatory. • In the iza-bren arm, Grade ≥3 anemia resolved in a median of 8.0 days; Grade ≥3 neutropenia and thrombocytopenia generally resolved within 2.0 to 3.0 days. The incidence of febrile neutropenia was low (6.0%). • Non-hematologic AEs were mostly low grade; asthenia was the most common grade ≥ 3 non-hematologic AEs in both groups (11.1% with iza-bren and 3.9% with PBC). • Treatment-related ILD occurred in two patients (0.9%) in the iza-bren arm (one grade 1 and one grade 3). None in the PBC arm. Li Zhang Data cut off: July 15, 2026

[Slide 2] Baseline Demographics | Iza-bren (N = 216) | PBC (N = 206) Median (range) age, years | 59.5 (37.0, 78.0) | 61.0 (31.0, 79.0) Age group (years), n (%) < 65 | 155 (71.8) | 139 (67.5) ≥ 65 | 61 (28.2) | 67 (32.5) Sex, n (%) Male | 94 (43.5) | 92 (44.7) Female | 122 (56.5) | 114 (55.3) ECOG PS, n (%) 0 | 51 (23.6) | 47 (22.8) 1 | 165 (76.4) | 159 (77.2) Smoking status, n (%) Never | 153 (70.8) | 147 (71.4) Former/Current | 63 (29.2) | 59 (28.6) CNS metastases, n (%) Yes | 21 (9.7) | 21 (10.2) No | 195 (90.3) | 185 (89.8) Liver metastases, n (%) | 29 (13.4) | 26 (12.6) EGFR mutation type, n (%) EGFR Exon 19 deletion | 117 (54.2) | 122 (59.2) EGFR Exon 21 L858R | 99 (45.8) | 85 (41.3) EGFR T790M | 18 (8.3) | 20 (9.7) Other | 3 (1.4) | 2 (1.0) Metastatic status, n (%) Yes | 214 (99.1) | 205 (99.5) Number of metastatic organs < 3 | 88 (40.7) | 88 (42.7) Number of metastatic organs ≥ 3 | 126 (58.3) | 117 (56.8) No | 2 (0.9) | 1 (0.5) Number of treatment lines, n (%) 1 line | 156 (72.2) | 150 (72.8) 2 lines | 60 (27.8) | 55 (26.7) 3 lines | 0 | 1 (0.5) Prior anti-angiogenic therapy*, n (%) Yes | 21 (9.7) | 18 (8.7) No | 195 (90.3) | 188 (91.3) Prior TKI, n (%) 1G-TKI/2G-TKI and 3G-TKI | 48 (22.2) | 53 (25.7) 3G-TKI only | 168 (77.8) | 153 (74.3) 1G, first-generation; 2G, second-generation; 3G, third-generation. *: including Bevacizumab, Anlotinib, and Endostar. Data cut off: July 15, 2026 Li Zhang [Slide 4] Subgroup Analysis of PFS (BICR) A consistent PFS benefit of iza-bren over PBC was seen across all predefined subgroups. Category | Subgroup | Iza-bren Events/n | PBC Events/n | Hazard Ratio (95% CI) All Subjects | | 128/216 | 164/206 | 0.42 (0.33, 0.54) ECOG PS | 0 | 31/51 | 39/48 | 0.40 (0.25, 0.66) | 1 | 97/165 | 125/158 | 0.41 (0.31, 0.54) CNS metastases | Yes | 17/21 | 16/20 | 0.75 (0.37, 1.50) | No | 111/195 | 148/186 | 0.39 (0.30, 0.51) Prior anti-angiogenic therapy | Yes | 10/19 | 12/18 | 0.37 (0.14, 0.95) | No | 118/197 | 152/188 | 0.42 (0.33, 0.54) Scores of the QLQ-C30 | ≤ Median score | 64/113 | 86/107 | 0.48 (0.34, 0.67) | > Median score | 64/103 | 78/99 | 0.38 (0.27, 0.54) Age group | < 65 years | 91/155 | 113/139 | 0.37 (0.28, 0.49) | ≥ 65 years | 37/61 | 51/67 | 0.57 (0.37, 0.89) Sex | Male | 53/94 | 80/92 | 0.36 (0.25, 0.52) | Female | 75/122 | 84/114 | 0.48 (0.34, 0.66) Number of metastatic organs | < 3 | 41/90 | 70/89 | 0.35 (0.24, 0.52) | ≥ 3 | 87/126 | 94/117 | 0.46 (0.34, 0.63) Number of prior treatment lines | 1 line | 103/156 | 121/150 | 0.52 (0.40, 0.68) | 2 lines | 25/60 | 42/55 | 0.17 (0.10, 0.32) Prior TKI | 1G-TKI/2G-TKI and 3G-TKI | 24/48 | 40/53 | 0.25 (0.14, 0.43) | 3G-TKI only | 104/168 | 124/153 | 0.48 (0.36, 0.62) EGFR mutation type | EGFR Exon 19 deletion | 60/117 | 100/122 | 0.31 (0.22, 0.43) | EGFR 21 L858R mutation | 68/99 | 64/85 | 0.59 (0.41, 0.83) Smoking status | Never | 89/153 | 111/147 | 0.45 (0.34, 0.60) | Former/Current | 39/63 | 53/59 | 0.35 (0.23, 0.55) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 Iza-bren better | PBC better Data cut off: July 15, 2026 Li Zhang

MADRID 2026 ESMO congress PANKU-Lung01: Randomized Phase III Study of Izalontamab Brengitecan (iza-bren) vs Platinum-Based Chemotherapy (PBC) in Patients with EGFR-Mutant Non-squamous Non-Small Cell Lung Cancer (NSCLC) After Progression on a Third-Generation EGFR Tyrosine Kinase Inhibitor (TKI) Li Zhang, Yunpeng Yang, Xiangjiao Meng, Zhihuang Hu, Wei Zheng, Juwei Cui, Yongsheng Luo, Lin Wu, Meijuan Huang, Haipeng Xu, Jie Hu, Yanqiu Zhao, Wenxiu Yao, Sa Xiao, Hai Zhu, Yi Zhu Presented by Li Zhang, MD Sun Yat-sen University Cancer Center, Guangzhou, China 26 October 2026

[Slide 2] PACE-LUNG: investigator-initiated, multicenter, phase 2 study Pre-screening eligibility criteria • Stage IIIB-IV NSCLC • EGFR L858R or exon 19 deletion (EGFRm; local test) • Osimertinib monotherapy planned or started within 3 weeks of enrollment • No history of leptomeningeal disease Screening eligibility criteria • Liquid biopsy positivity (ctDNA positivity) after 3-4 weeks of osimertinib, defined as any EGFRm VAF >0% of cell-free DNA (Guardant 360®) • Received osimertinib for no longer than 10 weeks • Asymptomatic CNS metastases allowed • No disease progression • ECOG PS 0-2 Endpoints • Primary: PFS1 from start of chemo per RECIST 1.1 • Secondary: PFS0 from start of osimertinib, OS1, OS0, ORR1, ORR0, safety and tolerability (CTCAE 5.0), quality of life N = 400 patients expected to pre-screen N = 50 patients Osimertinib 80 mg QD → Osimertinib 80 mg QD → Osimertinib 80 mg QD + 4 cycles of platinum-pemetrexed(a) → Osimertinib 80 mg QD Optional liquid biopsy | Liquid biopsy after 21-28 days | ctDNA EGFRm positivity | Optional liquid biopsy | No pemetrexed maintenance Pre-Screening | Screening | Study treatment | Maintenance / follow-up 3-4 weeks 10 weeks max. (a) investigator choice of cisplatin 75 mg/m² or carboplatin AUC 5 mg/mL/min, in combination with pemetrexed 500 mg/m². Pemetrexed maintenance not allowed; ctDNA denotes circulating tumor DNA, OS overall survival, ORR objective response rate, PFS progression-free survival, RECIST Response Evaluation Criteria in Solid Tumors, VAF variant allele frequency [Slide 3] Patient disposition and liquid biopsy results ➢ Between 11-2021 and 08-2025, 255 patients were screened at 16 sites in Germany. Thereof, 50 patients were included in the trial. Patient disposition 255 patients entered pre-screening → LB2 negative (n=159); LB2 missing/NE(a) (n=24) 72 patients with week 3 ctDNA positivity (LB2) → Eligibility criteria (n=10), withdrawal of consent (n=7), progression (n=2), death (n=3) 50 patients included in PACE-LUNG trial (ITT) 50 patients started osimertinib-chemotherapy 45 patients completed 4 cycles of chemotherapy 27 patients alive at data cut-off August 24th 2026 Liquid biopsy results [Sankey diagram] LB1 pre-osimertinib: Missing/NE 91; Negative 32; Positive 132 LB2 osimertinib week 3: 24 (missing/NE); 159 (negative); 72 (positive) Flow values shown: 3, 7, 14, 59, 25, 75, 29, 43 ➢ In patients with evaluable LB1 and LB2, no patient with LB1-negativity had a positive LB2 ➢ Among LB2-evaluable patients, 31% were ctDNA positive after 3-4 weeks of osimertinib monotherapy (a) includes cases where LB2 sampling was performed too early/late (n=4), did not occur due to eligibility criteria (n=4), withdrawal of consent (n=3), and death (n=2) ITT denotes intention-to-treat, LB liquid biopsy, NE not evaluable [Slide 4] Progression-free survival PFS1 from start of chemotherapy Median PFS1 (95% CI): 16.3 (10.8-18.9) | PFS1 at 12 months: 58% (43-72) | PFS1 at 24 months: 22% (9-35) + Censored x-axis: Months from start of chemotherapy At Risk (0, 10, 20, 30, 40): 50 33 12 3 0 Median follow-up, 30.0 months (95% CI, 22.3-NC) PFS0 from start of osimertinib Median PFS0 (95% CI): 17.8 (13.1-20.5) | PFS0 at 12 months: 74% (61-87) | PFS0 at 24 months: 31% (16-45) + Censored x-axis: Months from start of osimertinib At Risk (0, 10, 20, 30, 40): 50 39 14 5 0 Median follow-up, 31.5 months (95% CI, 24.4-NC) CI denotes confidence interval, NC not calculable, PFS progression-free survival. Survival table Endpoint: PFS (single-arm): PFS1 from start of chemotherapy and PFS0 from start of osimertinib Arm | N | Median | Median CI PFS1 (from start of chemotherapy), osimertinib + platinum-pemetrexed | 50 | 16.3 mo | 10.8-18.9 PFS0 (from start of osimertinib), osimertinib + platinum-pemetrexed | 50 | 17.8 mo | 13.1-20.5 Landmark rates: 12 mo: PFS1 58% (CI 43-72) | PFS0 74% (CI 61-87) 24 mo: PFS1 22% (CI 9-35) | PFS0 31% (CI 16-45)

LBA73 - PACE-LUNG: A multicenter, single-arm, phase II clinical trial evaluating the efficacy of additional chemotherapy for patients with EGFRm NSCLC with the continued presence of plasma ctDNA EGFRm at week 3 after start of osimertinib first-line treatment (AIO-TRK-0422/ass) Speakers Martin Sebastian (Frankfurt am Main, Germany) Lecture Time 08:30 - 08:35 Abstract Background In patients (pts) with common EGFR-mutant (EGFRm) advanced/metastatic non-small cell lung cancer (NSCLC), persisting EGFRm positivity in circulating tumor DNA (ctDNA) after 3 weeks on osimertinib is associated with an unfavorable outcome with a median progression-free survival (PFS) of 11.3 months (mo) (FLAURA). Recently, the addition of chemotherapy (CTx) to osimertinib has significantly improved PFS and overall survival (OS) with added toxicity compared with osimertinib alone. As no predictive biomarker is available for treatment stratification in these pts, PACE-LUNG aims to explore feasibility and clinical utility of a biomarker-guided treatment intensification strategy in pts with EGFRm NSCLC. Methods PACE-LUNG is an academic multicenter, single-arm, investigator-initiated phase II trial. Pts with untreated advanced/metastatic NSCLC and persistent ctDNA EGFRm by Guardant360® after 3 weeks osimertinib monotherapy received treatment escalation with 4 cycles of additional CTx, followed by osimertinib maintenance. Primary endpoint is investigator-assessed PFS1 per RECIST 1.1 from start of CTx. Secondary endpoints include PFS0, measured from start of osimertinib, OS1, safety, and others. Results Between Jan 2022 and Sep 2025, 255 pts were screened; 74 (29%) had persistent ctDNA EGFRm after 3 weeks of osimertinib. Thereof, 50 pts received additional CTx. Median age was 63 yrs (range, 39-85), ECOG PS was 0 or 1 (94%) at CTx initiation. Median PFS1 was 16.3 mo (95% CI, 10.8-18.9) with 57.8% (42.0-70.6) and 21.7% (10.2-36.1) of pts progression-free and alive at 12 and 24 mo after start of CTx, respectively. Median PFS0 was 17.8 mo (12.9 – 20.5). Median OS1 was 29.2 mo (95% CI, 23.2-42.4) with 87.4%(74.1-94.2) and 65.4% (48.0-78.2) of pts alive at 12 and 24 mo, respectively. No new safety signals were observed. Conclusions PACE-LUNG demonstrates that ctDNA-guided treatment intensification is feasible and results in a PFS that compares favorably with historical data. This approach should be used in future trials to enable personalized therapy in pts with EGFRm NSCLC.

[Slide 2] Therapeutic landscape of first-line treatments for EGFR-mut NSCLC 2018 | Overall | Ex19Del | L858R Osimertinib | 18.9 m | 21.4 m | 14.4 m Lazertinib | 20.6 m | 20.7 m | 17.8 m Firmonertinib | 20.8 m | 21.1 m | 17.1 m 2026 FLAURA2 | 25.5 m | 27.9 m | 24.7 m MARIPOSA | 23.7 m | NR | 18.4 m RAMOSE | 24.8 m | 25.8 m | 20.5 m NORTHSTAR | 25.3 m | 39.8 m | 19.0 m Xiuning Le MD PhD MD Anderson Cancer Center [Slide 3] Combining FL2 and PACE-LUNG results Baseline Un-detectable ctDNA EGFR 20% total pts PFS osimertinib 30 mons => Osimertinib monotherapy patients adding chemo – PFS 33 mon Start with FL2 osi-chemo upfront no selection => median PFS 24.8 months Baseline Detectable ctDNA EGFR 80% total pts Start with osi mono 64% Total pts -> ctDNA cleared at 3 wks => Continue osi mono PFS 15.2 months 16% Total pts -> ctDNA NOT cleared at 3 wks => FLAME Lung Osimertinib mono-only PFS 12.7 months FL2 AACR ctDNA 2025 baseline non-detected plasma EGFRm Median PFS, months (95% CI) Osimertinib + CTx (n=65) 33.3 (23.8, NC) Osimertinib mono (n=48) 30.3 (25.0, NC) baseline-detected plasma EGFRm Median PFS, months (95% CI) Osimertinib + CTx (n=147) 24.8 (19.6, 27.9) Osimertinib mono (n=161) 13.9 (13.6, 16.6) [partially obscured] HR (95% CI) 0.60 (0.45, 0.80) FL2 ESMO ctDNA 2026 | Clearance | No clearance mPFS (95% CI), months Osi + plat–pem | 27.6 (21.0, NR) | 19.6 (10.1, NR) Osi mono | 15.2 (13.6, 19.3) | 11.1 (5.6, 16.5) HR (95% CI) | 0.51 (0.35, 0.75) | 0.62 (0.33, 1.14) Xiuning Le MD PhD MD Anderson Cancer Center [Slide 4] Primary Endpoint: Grade ≥2 Dermatology AEIs by Week 12 Main COCOON study(1) OR, 0.24 (95% CI, 0.13–0.45) P<0.0001 SoC DM (n=100): 75% (68% Grade 2; 7% Grade >2) COCOON DM (n=99): 42% (36% Grade 2; 6% Grade >2) ORR 71% vs 70% SC amivantamab + simplified dermatologic prophylaxis(a) OR, 0.31 (95% CI, 0.11–0.87) Nominal P=0.0264 Standard schedule (n=70): 39% (36% Grade 2; 3% Grade >2) Laz lead-in (n=34): 18% FL2 rash 1% G3 28% All G Legend: Grade 2; Grade >2 Participants (%) • Dermatologic prophylaxis has previously shown significant reductions in DAEIs by Week 12(1) • Simplified prophylaxis demonstrated a similar reduction in DAEIs as the full COCOON DM regimen(b) • Lazertinib lead-in prior to starting amivantamab + lazertinib at Cycle 2 further reduced early DAEIs ▪ An exploratory analysis of Week 16 data showed consistent findings (standard schedule, 50% [4% grade ≥3]; laz lead-in, 24% [none grade ≥3]) (a)All 104 participants had the opportunity to receive treatment for 12 weeks; however, some discontinued prior to Week 12. (b)In a preliminary analysis, investigator-assessed overall response rates were consistent with the COCOON DM cohort. DAEI, dermatologic adverse event of interest; DM, dermatologic management; laz, lazertinib; OR, odds ratio; SoC, standard of care. 1. Cho BC, et al. J Thorac Oncol. 2025;20(10):1517–1530. Scan the QR code. Copies of this slide deck obtained through this QR code are for personal use only and may not be reproduced without permission of the authors.

[Slide 1] Discussion ➢ PACE-LUNG demonstrates that ctDNA-guided treatment intensification is feasible and results in outcomes that compare favorably with historical data ➢ PACE-LUNG: mPFS0, 17.8 months (95% CI, 13.1-20.5) ➢ FLAURA: mPFS, 11.3 months (95% CI, 9.5-16.5), in the osimertinib / week 3 ctDNA+ cohort(1) ➢ FLAURA2: mPFS, 11.1 months (95% CI, 5.6-16.5), in the osimertinib without chemo / week 3 ctDNA+ cohort(2) ➢ The safety profile was consistent with previous findings ➢ Recently, upfront combination therapies have demonstrated superior outcomes in biomarker-unselected patients with EGFRm, advanced or metastatic NSCLC(3-5) ➢ In light of these studies, PACE-LUNG calls into question the putative predictive role of ctDNA EGFRm alone as biomarker for guiding treatment escalation in this setting ➢ Pending analyses of co-mutational data of PACE-LUNG may help to inform future ctDNA-based treatment strategies ➢ Ultimately, this may be an important step in further personalizing the treatment of patients with EGFRm NSCLC 1. Zhou et al. ASCO 2019, 2. Jänne et al. AACR 2024, 3. Jänne et al. NEJM 2025, 4. Yang et al. NEJM 2025, 5. Li et al. Lancet Oncology 2026 Martin Sebastian [Slide 2] Conclusions • On-treatment plasma EGFRm clearance was prognostic, but not predictive of PFS or OS benefit with osimertinib + platinum–pemetrexed at all timepoints using a single-gene assay • Plasma EGFRm clearance rates were comparable across treatment arms at each timepoint • Studies using ultra-sensitive multi-marker ctDNA assays are planned to investigate if early ctDNA dynamics beyond plasma EGFRm can serve as a predictive biomarker Efficacy of osimertinib + platinum–pemetrexed was superior to osimertinib monotherapy regardless of plasma EGFRm clearance status, reinforcing osimertinib + platinum–pemetrexed as a preferred first-line treatment option for EGFR-mutated advanced NSCLC ctDNA, circulating tumour DNA; EGFR, epidermal growth factor receptor; EGFRm, EGFR mutation; NSCLC, non-small cell lung cancer; OS, overall survival; PFS, progression-free survival David Planchard, MD, PhD

Overall survival OS1 from start of chemotherapy Median OS1 (95% CI): 29.2 (24.4-42.4) | OS1 at 12 months: 87% (78-97) | OS1 at 24 months: 66% (51-81) + Censored x-axis: Months from start of chemotherapy At Risk (0, 10, 20, 30, 40, 50): 50 42 25 12 5 0 Median follow-up, 30.2 months (95% CI, 22.4-36.5) OS0 from start of osimertinib Median OS0 (95% CI): 31.5 (27.0-44.4) | OS0 at 12 months: 92% (84-100) | OS0 at 24 months: 72% (59-86) + Censored x-axis: Months from start of osimertinib At Risk (0, 10, 20, 30, 40, 50): 50 46 27 13 5 0 Median follow-up, 32.4 months (95% CI, 24.5-38.1) CI denotes confidence interval, OS overall survival Survival table Endpoint: OS (single-arm): OS1 from start of chemotherapy and OS0 from start of osimertinib Arm | N | Median | Median CI OS1 (from start of chemotherapy) | 50 | 29.2 mo | 24.4-42.4 OS0 (from start of osimertinib) | 50 | 31.5 mo | 27.0-44.4 Landmark rates: 12 mo: OS1 87% (CI 78-97) | OS0 92% (CI 84-100) 24 mo: OS1 66% (CI 51-81) | OS0 72% (CI 59-86)

[Slide 1] 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic CHAIRS: PASI A. JÄNNE, XIUNING LE, EGBERT SMIT Combining FL2 and PACE-LUNG results Baseline Un-detectable ctDNA EGFR 20% total pts: PFS osimertinib 30 mons => Osimertinib monotherapy patients adding chemo - PFS 33 mon Baseline Detectable ctDNA EGFR 80% total pts: Start with FL2 osi-chemo upfront no selection => median PFS 24.8 months Start with osi mono 64% Total pts: ctDNA cleared at 3 wks => Continue osi mono [red-highlighted value: PFS [illegible] months] 16% Total pts: ctDNA NOT cleared at 3 wks FLAME-lung Osimertinib mono-only PFS 12.7 months PACE-Lung osi+chemo PFS 17.8 months FLAME-Lung Osi+cheomo PFS 23.1 months FL2 AACR ctDNA 2025 baseline non-detected plasma EGFRm Median PFS, months (95% CI) Osimertinib + CTx (n=65[?]) 33.3 (23.8, NC) Osimertinib mono (n=[illegible]) 30.3 (25.0, NC) baseline-detected plasma EGFRm Median PFS, months (95% CI) Osimertinib + CTx (n=147[?]) 24.8 (19.6, 27.9) Osimertinib mono (n=[illegible]) [illegible] HR (95% CI) [illegible - approx. 0.60 (0.45, 0.80)] FL2 ESMO ctDNA 2026 mPFS (95% CI), months | Clearance | No clearance Osi + plat-pem | 27.6 (21.0, NR) | 19.6 (10.1, NR) Osi mono | [red-highlighted, illegible] | 11.1 (5.6, 16.5) HR (95% CI) | 0.51 (0.35, 0.75) | 0.62 (0.33, 1.14) Xiuning Le MD PhD MD Anderson Cancer Center ALMERIA AUDITORIUM - HALL 9 [Slide 2] 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic CHAIRS: PASI A. JÄNNE, XIUNING LE, EGBERT SMIT Take home message [Non-detectable] • For patients with baseline non-detectable ctDNA, the prognosis is good, there might be an option for osimertinib monotherapy. Median PFS 30 month [Detectable] • For patients started on osimertinib monotherapy, but without ctDNA clearance of EGFRm at 3-4 weeks, the addition of platinum-chemotherapy is indicated. Median PFS 17.8 month • For patients who had a clearance of ctDNA EGFRm, the magnitude of benefit of osi-chemo is large, with 12 months absolute improvement of PFS, HR 0.51. • Therefore, all eligible patients should start with osi-chemo because on treatment ctDNA EGFRm clearance was NOT predictive of benefit with osi+chemo vs. osimertinib monotherapy. Median PFS 24 month Xiuning Le MD PhD MD Anderson Cancer Center ALMERIA AUDITORIUM - HALL 9

Overall survival OS1 from start of chemotherapy + Censored Median OS1 (95% CI) | OS1 at 12 months | OS1 at 24 months 29.2 (24.4-42.4) | 87% (78-97) | 66% (51-81) Months from start of chemotherapy 0 | 10 | 20 | 30 | 40 | 50 At Risk | 50 | 42 | 25 | 12 | 5 | 0 Median follow-up, 30.2 months (95% CI, 22.4-36.5) OS0 from start of osimertinib + Censored Median OS0 (95% CI) | OS0 at 12 months | OS0 at 24 months 31.5 (27.0-44.4) | 92% (84-100) | 72% (59-86) Months from start of osimertinib 0 | 10 | 20 | 30 | 40 | 50 At Risk | 50 | 46 | 27 | 13 | 5 | 0 Median follow-up, 32.4 months (95% CI, 24.5-38.1) CI denotes confidence interval, OS overall survival Martin Sebastian Survival table Endpoint: Overall survival (OS1 from start of chemotherapy; OS0 from start of osimertinib), single-arm Arm | N | Median | Median CI OS1 (from start of chemotherapy) | 50 | 29.2 | 24.4-42.4 OS0 (from start of osimertinib) | 50 | 31.5 | 27.0-44.4 Landmark rates: 12 months: OS1 87% (CI 78-97) | OS0 92% (CI 84-100) 24 months: OS1 66% (CI 51-81) | OS0 72% (CI 59-86)

08:30 - 10:00 Rapid oral 1: NSCLC, metastatic CHAIRS: PASI A. JÄNNE, XIUNING LE, EGBERT SMIT Progression-free survival PFS1 from start of chemotherapy Median PFS1 (95% CI): 16.3 (10.8-18.9) | PFS1 at 12 months: 58% (43-72) | PFS1 at 24 months: 22% (9-35) [x-axis: Months from start of chemotherapy, 0-40; At Risk row illegible] Median follow-up, 30.0 months (95% CI, 22.3-NC) PFS0 from start of osimertinib Median PFS0 (95% CI): 17.8 (13.1-20.5) | PFS0 at 12 months: 74% (61-87) | PFS0 at 24 months: 31% (16-45) [x-axis: Months from start of osimertinib, 0-40; At Risk row illegible] Median follow-up, 31.5 months (95% CI, 24.4-NC) CI denotes confidence interval, NC not calculable, PFS progression-free survival. ALMERIA AUDITORIUM - HALL 9 Survival table Endpoint: PFS1 (from start of chemotherapy, primary) and PFS0 (from start of osimertinib) — single-arm PFS1 (osimertinib + platinum-pemetrexed, from chemo start) | 16.3 | 10.8-18.9 PFS0 (from osimertinib start) | 17.8 | 13.1-20.5 Landmark rates: 12 mo: PFS1 58% (CI 43-72) | PFS0 74% (CI 61-87) 24 mo: PFS1 22% (CI 9-35) | PFS0 31% (CI 16-45)

[Slide 2] PFS per BICR Ruzaltatug rezetecan (n = 250) | Chemo (n = 248) Event, n (%) 108 (43.2) | 156 (62.9) Median PFS (95% CI), mo 8.3 (6.9-9.9) | 5.4 (4.4-5.6) HR* (95% CI) 0.47 (0.36-0.60) 1-sided P value† <0.0001 59.8% | 32.4% | Δ27.4% 46.3% | 16.7% | Δ29.6% PFS (%) Time since randomization (mo) No. at risk Ruzaltatug rezetecan 250 225 179 90 62 20 10 5 0 Chemo 248 200 141 44 26 7 4 2 0 • Consistent PFS benefit per investigator assessment: HR 0.46 (95% CI 0.36-0.60) * Stratified Cox proportional hazards model. † Stratified Log-rank test. Stratification factors were the same as those used for randomization. Survival table Endpoint: PFS per BICR Arm | N | Median | Median CI Ruzaltatug rezetecan | 250 | 8.3 mo | 6.9-9.9 Chemo | 248 | 5.4 mo | 4.4-5.6 HR: 0.47 HR CI: 0.36-0.60 P: <0.0001 (1-sided, stratified log-rank) 6 mo: 59.8% vs 32.4% (Δ27.4%) 9 mo: 46.3% vs 16.7% (Δ29.6%) [Slide 3] Tumor Response per BICR ORR Δ 25.7 (95% CI 17.0-33.9) 60.0 (150/250) | 34.3 (85/248) DCR Δ 9.7 (95% CI 4.0-15.5) 92.8 (232/250) | 83.1 (206/248) Rate (%) Ruzaltatug rezetecan | Chemo Ruzaltatug rezetecan (n = 250) | Chemo (n = 248) Median DoR (95% CI), mo 8.5 (5.6-13.1) | 4.4 (4.2-5.8) Ongoing response (%) 56.0% | 48.0% | 38.2% 35.1% | 23.4% Time since first response (mo) No. at risk Ruzaltatug rezetecan 150 114 81 37 25 5 3 0 Chemo 85 62 35 11 3 1 0 Survival table Endpoint: Duration of response per BICR Arm | N | Median | Median CI Ruzaltatug rezetecan | 150 | 8.5 mo | 5.6-13.1 Chemo | 85 | 4.4 mo | 4.2-5.8 6 mo: 56.0% vs 35.1% 9 mo: 48.0% vs 23.4% 12 mo: 38.2% (ruzaltatug rezetecan only) [Slide 4] Safety Summary Ruzaltatug rezetecan (n = 248) | Chemo (n = 239) Any TRAE 247 (99.6) | 236 (98.7) Grade ≥3 101 (40.7) | 93 (38.9) Serious 38 (15.3) | 36 (15.1) Leading to dose reduction 36 (14.5) | 24 (10.0) Leading to dose discontinuation 3 (1.2) | 3 (1.3) Leading to death 1 (0.4)* | 0 ILD 6 (2.4) | 0 Grade ≥3 1 (0.4)† | 0 Data are n (%). * Death of unknown cause. † Grade 3. Common TRAEs‡ Ruzaltatug rezetecan | Chemo Neutrophil count decreased 71.4 25.8 | 19.2 56.5 WBC count decreased 63.3 11.7 | 12.1 60.3 Anemia 60.9 6.0 | 10.9 76.6 Alopecia 59.7 0.0 | 0.0 12.6 Nausea 55.2 1.6 | 2.1 38.9 Decreased appetite 54.4 1.2 | 0.0 39.7 Asthenia 39.5 3.2 | 0.8 32.6 ALT increased 37.9 0.8 | 0.0 38.9 AST increased 33.9 1.2 | 0.0 38.1 GGT increased 25.8 2.0 | 2.1 18.4 Constipation 25.4 0.0 | 0.0 24.7 Platelet count decreased 23.0 2.0 | 9.2 41.0 Hypoalbuminemia 21.4 0.0 | 0.0 13.8 Vomiting 21.0 1.2 | 2.1 17.2 Grade 1-2 / Grade ≥3 Incidence (%) ‡ TRAEs of any grade occurring in ≥20% of patients and of grade ≥3 occurring in ≥5% in either group are listed. ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyltransferase; ILD, interstitial lung disease; TRAE, treatment-related adverse event; WBC, white blood cell.

LBA72 - Ruzaltatug rezetecan, a HER3-directed ADC, vs platinum-based chemotherapy (PBC) in EGFR-mutated advanced NSCLC progressing after EGFR-TKI: Interim analyses (IA) from the phase 3 HERON-Lung01 trial Speakers Chongrui Xu (Guangzhou, China) Lecture Time 09:24 - 09:34 Authors Yi-Long Wu (Guangzhou, China), Chongrui Xu (Guangzhou, China), Yan Zhang (Chengdu, China), Guinan Lin (Wuxi, China), Haiyong Wang (Jinan, China), Qiming Wang (Zhengzhou, China), Dingzhi Huang (Tianjin, China), Yan Yu (Harbin, China), Zhou Jin (Chengdu, China), Hua Zhong (Shanghai, China), Lin Wu (Changsha, China), Lingxiang Liu (Nanjing, China), Jianchun Duan (Beijing, China), Zhiyi He (Nanning, China), Yongzhong Luo (Changsha, China), Hongxu Liu (Shenyang, China), Bi Chen (Xuzhou, China), Jingyu Zhang (Shanghai, China), Fei Qiu (Beijing, China), Lulu Yang (Shanghai, China) Abstract Background Ruzaltatug rezetecan (SHR-A2009) is a novel ADC composed of a fully human anti-HER3 IgG1 mAb, cleavable peptide linker and DNA topoisomerase I inhibitor payload. In a previous phase 1 trial, SHR-A2009 showed promising antitumor activity and manageable safety in pretreated EGFR-mutated NSCLC, supporting pivotal investigation. Methods In this phase 3 study, patients (pts) with EGFR-mutated (Ex19del/L858R) advanced nonsquamous NSCLC progressing after EGFR–TKI were randomized 1:1 to receive SHR-A2009 (8 mg/kg, IV, Q3W) or PBC (pemetrexed 500 mg/m2 + carboplatin AUC 5 or cisplatin 75 mg/m2 Q3W for 4 cycles, followed by maintenance pemetrexed). Crossover to SHR-A2009 from PBC arm was allowed after disease progression. The primary endpoint was progression-free survival (PFS) per blinded independent central review (BICR). Results A total of 498 pts were randomized (SHR-A2009, n=250; PBC, n=248). At the prespecified PFS IA (data cutoff, Apr. 15, 2026), median follow-up was 7.7 mo. BICR-assessed PFS was significantly longer with SHR-A2009 vs PBC (median, 8.3 vs 5.4 mo; HR 0.47 [95% CI 0.36–0.60]; 1-sided P <0.0001). PFS benefits persisted across all prespecified subgroups, regardless of presence of brain metastases or EGFR mutation type. Overall survival was immature but favored SHR-A2009 (HR 0.70; 95% CI, 0.43–1.14), despite substantial post-progression crossover (19.4%) from PBC arm. Objective response rate (60.0% vs 34.3%), disease control rate (92.8% vs 83.1%) and median duration of response (8.5 vs 4.4 mo) per BICR were also improved with SHR-A2009. Grade ≥3 treatment-related adverse events (TRAEs) occurred in 40.7% with SHR-A2009 vs 38.9% with PBC, most commonly decreased neutrophil count (25.8% vs 19.2%), decreased WBC count (11.7% vs 12.1%) and anaemia (6.0% vs 10.9%). Interstitial lung disease occurred in 6 pts (2.4%; grade 1–2, n=5; grade 3, n=1) with SHR-A2009. TRAEs led to dose discontinuation in 1.2% with SHR-A2009 and 1.3% with PBC. Conclusions SHR-A2009 significantly improved PFS vs PBC, with a manageable safety profile, supporting it as a new standard-of-care option for advanced EGFR-mutated NSCLC progressing after EGFR-TKI.

[Slide 1] Study Design HERON-Lung01: an open-label, randomized, phase 3 trial (NCT06671379) Key eligibility criteria • Locally advanced/metastatic Nsq NSCLC • Confirmed EGFR mutation (L858R/Ex19del*) • Progression during/after EGFR–TKI – Prior 3rd gen TKI required for T790M tumors • No other prior systemic therapy permitted† • ECOG PS 0–1 R 1:1 Ruzaltatug rezetecan 8 mg/kg Q3W, IV Carboplatin AUC5/ Cisplatin 75 mg/m2 Q3W for 4 cycles + Pemetrexed 500 mg/m2 Q3W Crossover permitted at PD Ruzaltatug rezetecan 8 mg/kg Q3W, IV Primary endpoint • PFS per BICR Secondary endpoints • OS • PFS per INV • ORR per BICR and INV • DCR per BICR and INV • DoR per BICR and INV • Safety Stratification factors • Brain metastases (yes vs no) • EGFR mutation type (L858R vs Ex19del) • Prior first-line 3rd gen EGFR-TKI (yes vs no) * EGFR mutation status was based on tissue or blood testing. † Prior anti-angiogenic agents in combination with EGFR-TKI were permitted. [Slide 3] Tumor Response per BICR ORR Ruzaltatug rezetecan 60.0 (150/250) | Chemo 34.3 (85/248) Δ 25.7 (95% CI 17.0–33.9) DCR Ruzaltatug rezetecan 92.8 (232/250) | Chemo 83.1 (206/248) Δ 9.7 (95% CI 4.0–15.5) Rate (%) Ruzaltatug rezetecan (n = 250) | Chemo (n = 248) Median DoR (95% CI), mo | 8.5 (5.6–13.1) | 4.4 (4.2–5.8) Ongoing response (%) Time since first response (mo) No. at risk Survival table Endpoint: Duration of response per BICR Arm | N | Median | Median CI Ruzaltatug rezetecan | 150 | 8.5 mo | 5.6–13.1 Chemo | 85 | 4.4 mo | 4.2–5.8 6 mo: 56.0% vs 35.1% 9 mo: 48.0% vs 23.4% 12 mo: 38.2% (ruzaltatug rezetecan)

[Slide 1] RESISTANCE MECHANISMS OF ROS1 Primary resistance mechanisms ①Specific mutations in kinase domain: ROS1 L1951R, ALK L1196M ②Compound mutations: ALK L1196M and G1202R ③Concurrent mutations: ALK, MET amplification ④Histological phenotype transition: Primary resistance to TKI Secondary resistance mechanisms ROS1: ①Bypass pathway activation: KRAS, NRAS, EGFR, HER2, BRAF V600E, MET, KIT, PIK3CA activation; KRAS and MET amplification; NF1 loss-of-function mutation ②Secondary resistance mutations in the kinase domain: Crizotinib: G2032R, D2033N, L2026M, L1951R, S1986F/Y, L2155S; Entrectinib: F2004C/I, G2023R; Lorlatinib: G2032K/R, S1986F, L2000V, L2086F ③Downstream signaling pathway activation ④Histological phenotype transition: AD -> SCLC ALK: ①Bypass pathway activation: FGFR2, PIK3CA, MET, EGFR, KRAS, NARS, HER, BRAF, SRC, YAP and IGF-1R activation; RET and NRG1 fusion; MET and KIT amplification; MET hyperactivation; NF1 and NF2 mutations ②ALK amplification and secondary resistance mutations in the kinase domain: Crizotinib: L1196M, G1269A, C1156Y, G1202R, I1171T/N/S, E1210K, S1206C/Y, L1152P/R, V1180L, L1151T, F1174C, 1151Tins; Ceritinib, alectinib, and brigatinib: G1202R, F1174C/L, I1151Tins, L1152P/R, C1156Y/T, I1171N/T/S, V1180L, G1202del, L1196M, E1210K, S1206C/Y, D1203N; Lorlatinib: L1198F, G2023R, D1203N, G1123D, L1256F ④Histological phenotype transition: SCLC, SCC, SC, EMT phenotype, NEC Pathway nodes: PI3K -> AKT -> mTOR; RAS -> RAF -> MEK -> ERK; JAK -> STAT3; SRC ○ Secondary mutations in the ROS1 kinase domain • G2032R is the most common solvent front mutation identified in tumors with acquired resistance to ROS1 TKIs. • Other common secondary on-target resistance mutations: • Crizotinib: G2032R, D2033N, L2026M, L1951R, S1986F/Y, and L2155S • Entrectinib: F2004C/I and G2023R • Lorlatinib: G2032K/R, S1986F, L2000V, and L2086F. ○ Bypass signaling pathway activation via activation of KRAS, NRAS, EGFR, HER2, BRAF V600E, MET, KIT, and PIK3CA; amplification of KRAS and MET; and loss-of-function mutations in NF1, rendering cells resistant to ROS1 inhibition. ○ SCLC transformation is another mechanism of acquired resistance to ROS1 TKIs. Drilon A, et al. Nat Rev Clin Oncol 2021;18(1):35-55; Li C, et al. Signal Transduct Target Ther. 2026 Oct 19;11(1):454 Dr. Calvo de Juan, MD, PhD [Slide 2] WHEN? Three different situations SYNCHRONOUS OLIGOMETASTATIC DISEASE • Limited metastatic disease at diagnosis • Can local treatment after initial systemic therapy improves long-term disease control? OLIGOPERSISTENT DISEASE / OLIGORESIDUAL DISEASE • Limited number of residual lesions after systemic response • Does treating residual macroscopic disease add benefit beyond continuing effective systemic treatment? OLIGOPROGRESSIVE DISEASE / OLIGORECURRENT DISEASE • A small number of lesions progress while the rest remains controlled • Can local ablation control resistant sites and delay changing the systemic treatment? [small body-diagram illustrations, labels illegible] Dr. Calvo de Juan, MD, PhD [Slide 3] WHEN? Targeted therapy + local ablative therapy Trial | N | Max met sites | Treatment | Outcomes SINDAS, Wang XS 2023 (1), Ph 3 RCT | 133, EGFRm | ≤5, ≤2 per organ; Brain metastases excluded | 1st-gen EGFR TKI (erlotinib, gefitinib, icotinib) ± upfront SBRT (25-40Gy in 5 fractions) to all sites (including primary tumor) | POSITIVE; mPFS 20.2 vs 12.5 months (p<0.001); mOS 25.5 vs 17.4 months (p<0.001) NROG-002, Sun H 2025 (2), Ph 3 RCT | 118, EGFRm | No lesion limit; ≤3 organs | 1st-gen EGFR TKI ± concurrent thoracic RT (60 Gy) | POSITIVE; mPFS 17.1 vs 10.6 months; mOS 34.4 vs. 26.2 moths LUNG-SORT, Zhou Y 2024 (3), Ph 2 single-arm | 61, EGFRm | ≤5; ≤3 organs | 3rd-gen EGFR TKI (median 3.7 months, range 1-6) then consolidative SRT to all residual sites | mPFS 29.9 months; 1-yr PFS 91.6%; 2-yr PFS 53.4% NCT03667820, Sampath 2025 (4), Ph 2 single-arm | 42, EGFRm | No limit | Osimertinib 80mg daily + consolidative RT, primarily stereotactic RT at best response (8–10 weeks) | mPFS 32.6 months; mOS 45.7 months NORTHSTAR, Elamin YY 2025 (5), Ph 2 randomized | 119, EGFRm | No limit (~60% polymetastatic) | Induction osimertinib 6-12 weeks -> R 1:1: osimertinib + LCT (RT 59%, surgery 29%, both 12%); vs osimertinib alone | POSITIVE; mPFS 25.3 vs 17.5 months (p=0.025); Polymetastatic: 20.7 vs 15.9 (HR 0.73) BRIGHTSTAR, Elamin YY 2025 (6), Ph 2 single-arm | 34, ALK | No limit (82% polymetastatic, 41% BM) | Brigatinib induction 8 weeks -> LCT (RT 79%, surgery 9%) -> continued brigatinib | POSITIVE; mPFS 66 months; 5-yr PFS 51% (vs historical brigatinib mPFS 29 months) (1) Wang XS, et al. J Natl Cancer Inst. 2023 Jun 8;115(6):742-8; (2) Sun H, et al. J Clin Oncol. 2025 Feb;43(4):412-21; (3) Zhou Y, et al. EClinicalMedicine. 2024 Sep 26;76:102853; (4) Sampath S, et al. EClinicalMedicine. 2025 Aug 26;87:103435; (5) Elamin YY, et al. Ann Oncol. 2025;36(Suppl 2):LBA72 (ESMO 2025); (6) Elamin YY, et al. J Thorac Oncol.2025;20(Suppl 1):S23–S24 (WCLC 2025) Dr. Calvo de Juan, MD, PhD [Slide 4] TAKE HOME MESSAGES • Our challenge is not to treat as many lesions as possible... • ...it is to identify the right patient, at the right time, with a clear therapeutic objective • Oligometastasis and oligoprogression are different clinical problems • Local therapy is a strategy, not a default – It depend on the clinical scenario, disease biology and therapeutic objective • Medical oncologist defines the clinical objective... • ...the multidisciplinary team determines how best to achieve it – Integrate local and systemic treatment around the expected benefit for the individual patients Dr. Calvo de Juan, MD, PhD

OS Ruzaltatug rezetecan (n = 250) | Chemo (n = 248) Event, n (%) 29 (11.6) | 37 (14.9) HR* (95% CI) 0.70 (0.43-1.14) 93.8% | 88.6% 76.8% | 71.5% OS (%) Time since randomization (mo) No. at risk Ruzaltatug rezetecan 250 246 214 156 111 58 21 7 1 0 Chemo 248 242 209 136 95 52 18 5 0 • Subsequent systemic therapy: 28.0% in ruzaltatug rezetecan arm vs 44.8% in chemo arm (including 19.4% crossed over) * Stratified Cox proportional hazards model. Stratification factors were the same as those used for randomization. Survival table Endpoint: Overall survival Ruzaltatug rezetecan | 250 Chemo | 248 HR: 0.70 HR CI: 0.43-1.14 6 mo: 93.8% vs 88.6% 12 mo: 76.8% vs 71.5%

[Slide 2] Baseline Characteristics Ruzaltatug rezetecan (n = 250) | Chemo (n = 248) Median age (range), yr 60 (27-75) | 61 (35-75) Male 116 (46.4) | 106 (42.7) Current/former smoker 70 (28.0) | 80 (32.3) ECOG PS 1 192 (76.8) | 192 (77.4) Adenocarcinoma 250 (100.0) | 248 (100.0) Clinical stage IV 245 (98.0) | 241 (97.2) Brain metastases 48 (19.2) | 47 (19.0) Liver metastases 35 (14.0) | 35 (14.1) EGFR mutation subtype* Exon 19 deletion 154 (61.6) | 152 (61.3) L858R mutation 96 (38.4) | 96 (38.7) Prior 3rd gen EGFR-TKI 246 (98.4) | 239 (96.4) First-line therapy 194 (77.6) | 194 (78.2) Prior anti-angiogenic therapy 23 (9.2) | 26 (10.5) Data are n (%) or otherwise indicated. * EGFR mutation status was assessed using tissue or blood samples. [Slide 4] PFS per BICR in Subgroups No. of events/no. of patients: Ruzaltatug rezetecan | Chemo | HR* (95% CI) Age <65 yr 74/169 | 101/155 | 0.41 (0.30-0.55) Age ≥65 yr 34/81 | 55/93 | 0.54 (0.35-0.83) Sex Male 53/116 | 65/106 | 0.45 (0.31-0.66) Sex Female 55/134 | 91/142 | 0.45 (0.32-0.63) Smoking history Yes 37/70 | 51/80 | 0.49 (0.32-0.75) Smoking history No 71/180 | 105/168 | 0.44 (0.32-0.60) ECOG PS 0 22/57 | 32/56 | 0.37 (0.21-0.64) ECOG PS 1 86/192 | 124/192 | 0.49 (0.37-0.65) EGFR mutation Exon 19 deletion 72/154 | 97/152 | 0.48 (0.35-0.66) EGFR mutation L858R mutation 36/96 | 59/96 | 0.41 (0.27-0.62) Brain metastases Yes 28/48 | 33/47 | 0.61 (0.36-1.02) Brain metastases No 80/202 | 123/201 | 0.42 (0.32-0.56) Liver metastases Yes 20/35 | 26/35 | 0.33 (0.18-0.62) Liver metastases No 88/215 | 130/213 | 0.47 (0.36-0.62) Prior first-line 3rd gen EGFR-TKI Yes 90/194 | 125/194 | 0.50 (0.38-0.66) Prior first-line 3rd gen EGFR-TKI No 18/56 | 31/54 | 0.29 (0.15-0.54) Prior lines of systemic therapy 1 81/178 | 120/191 | 0.48 (0.36-0.64) Prior lines of systemic therapy ≥2 27/72 | 36/57 | 0.39 (0.23-0.66) 0.1 1 10 Favors ruzaltatug rezetecan ← → Favors chemo * Unstratified Cox proportional hazards model.

[Slide 3] Difference in clinical activities? PANKU-Lung01 PFS by BICR: Primary Endpoint Iza-bren (N = 216) | PBC (N = 206) Number of events 128 | 164 Median, mo (95% CI) 11.1 (9.2, 12.7) | 5.7 (5.5, 5.9) HR (95% CI) 0.40 (0.31, 0.51)† P-value <0.0001* Median PFS, 11.1 vs 5.7 months; HR, 0.40 (95% CI: 0.31, 0.51), p < 0.0001 OS: Key Secondary Endpoint Iza-bren (N = 216) | PBC (N = 206) Number of events (death) 64 | 82 Median OS, mo (95% CI) NR (19.3, NR) | 19.4 (16.4, 22.5) Median follow-up for OS, mo (95% CI) 16.9 (15.9, 17.7) | 16.5 (15.1, 17.5) 12-mo OS rate (95% CI) 83.0 (77.3, 87.5) | 72.9 (66.1, 78.5) 18-mo OS rate (95% CI) 68.8 (60.9, 75.5) | 55.3 (46.8, 63.0) 24-mo OS rate (95% CI) 50.6 (36.1, 63.5) | NR (NR, NR) HR (95% CI) 0.66 (0.47, 0.91)† P-value 0.0058* More patients received subsequent anti-cancer therapy in the PBC arm, including ADCs Iza-bren | PBC: ADC 2.8% | 29.1%; Targeted therapy 25.9% | 30.1%; Immune therapy 17.6% | 31.1%; Chemotherapy 31.5% | 40.3% A trend towards improved OS favoring iza-bren was observed at this interim analysis. HERON-Lung01 PFS per BICR Ruzaltatug rezetecan (n = 250) | Chemo (n = 248) Event, n (%) 108 (43.2) | 156 (62.9) Median PFS (95% CI), mo 8.3 (6.9–9.9) | 5.4 (4.4–5.6) HR* (95% CI) 0.47 (0.36–0.60) 1-sided P value† <0.0001 • Consistent PFS benefit per investigator assessment: HR 0.46 (95% CI 0.36–0.60) – Median PFS: 8.3 mo vs 5.4 mo; HR 0.47 (95% CI 0.36–0.60); 1-sided P <0.0001 OS Ruzaltatug rezetecan (n = 250) | Chemo (n = 248) Event, n (%) 29 (11.6) | 37 (14.9) HR* (95% CI) 0.70 (0.43–1.14) • Subsequent systemic therapy: 28.0% in ruzaltatug rezetecan arm vs 44.8% in chemo arm (including 19.4% crossed over) Survival table Endpoint: HERON-Lung01 PFS per BICR (panel also shows PANKU-Lung01 PFS/OS and HERON-Lung01 OS) Arm | N | Median | Median CI Ruzaltatug rezetecan | 250 | 8.3 mo | 6.9–9.9 Chemo | 248 | 5.4 mo | 4.4–5.6 HR: 0.47 HR CI: 0.36–0.60 P: <0.0001 (1-sided) HERON-Lung01 PFS printed curve labels: 59.8% vs 32.4% (Δ27.4%) and 46.3% vs 16.7% (Δ29.6%) at two vertical markers (approx. 6 and 9 mo) HERON-Lung01 PFS no. at risk (0-16 mo, 2-mo steps) ruzaltatug: 250,225,179,90,62,20,10,5,0; chemo: 248,200,141,44,26,7,4,2,0 HERON-Lung01 OS: events 29 (11.6%) vs 37 (14.9%), HR 0.70 (0.43–1.14); printed labels 93.8% vs 88.6% and 76.8% vs 71.5% (approx. 6 and 12 mo) PANKU-Lung01 OS: 64 vs 82 deaths, median NR (19.3, NR) vs 19.4 (16.4, 22.5), HR 0.66 (0.47, 0.91), P 0.0058; 12/18/24-mo OS 83.0/68.8/50.6% vs 72.9/55.3/NR [Slide 4] Tumor Response per BICR ORR Ruzaltatug rezetecan 60.0 (150/250) | Chemo 34.3 (85/248) Δ 25.7 (95% CI 17.0–33.9) DCR Ruzaltatug rezetecan 92.8 (232/250) | Chemo 83.1 (206/248) Δ 9.7 (95% CI 4.0–15.5) Rate (%) Ruzaltatug rezetecan (n = 250) | Chemo (n = 248) Median DoR (95% CI), mo 8.5 (5.6–13.1) | 4.4 (4.2–5.8) Ongoing response (%) Time since first response (mo) No. at risk Survival table Endpoint: Duration of response per BICR Arm | N | Median | Median CI Ruzaltatug rezetecan | 150 | 8.5 mo | 5.6–13.1 Chemo | 85 | 4.4 mo | 4.2–5.8 Printed curve labels ruzaltatug: 56.0%, 48.0%, 38.2% at vertical markers (approx. 6, 9, 12 mo); chemo: 35.1%, 23.4% (approx. 6, 9 mo) No. at risk (0-14 mo, 2-mo steps) ruzaltatug: 150,114,81,37,25,5,3,0; chemo: 85,62,35,11,3,1,0

[Slide 1] 08:30 - 10:00 Proffered paper 2: NSCLC, metastatic CHAIRS: PATRICK FORDE, ANTONIO PASSARO PFS by BICR: Primary Endpoint Iza-bren showed statistically significant and clinically meaningful improvement vs PBC with doubling of mPFS. Iza-bren (N = 216) | PBC (N = 206) Number of events 128 | 164 Median, mo (95% CI) 11.1 (9.2, 12.7) | 5.7 (5.5, 5.9) HR (95% CI) 0.40 (0.31, 0.51)† P-value <0.0001* Progression-free survival probability (%) Time (Months) Iza-bren PBC + Censored Subjects at risk PFS (BICR) was evaluated in the mITT population. The median duration of follow-up for PFS was 14.6 months. Similar results were observed for PFS (BICR) in the ITT population, with 129 events in the iza-bren arm and 164 events in the PBC arm. The hazard ratio (95% CI) was 0.40 (0.31, 0.51). †Based on a stratified Cox regression model. *One-sided, based on a stratified log-rank test. Data cut off: July 15, 2026 Li Zhang Li Zhang PANKU-Lung01: Randomized phase III study of izalontamab brengitecan (iza-bren) in EGFR-mutant non-squamous non-small cell lung cancer (NSCLC) after progression on a third-generation EGFR TKI ALICANTE AUDITORIUM - HALL 6 Survival table Endpoint: PFS by BICR (mITT) Arm | N | Median | Median CI Iza-bren | 216 | 11.1 mo | 9.2, 12.7 PBC | 206 | 5.7 mo | 5.5, 5.9 HR: 0.40 HR CI: 0.31, 0.51 P: <0.0001 (one-sided, stratified log-rank) Printed curve labels: iza-bren 75.1% / 58.2% / 46.5% vs PBC 42.9% / 24.4% / 14.4% at vertical markers ~6, 9 and 12 months (timepoints not explicitly labelled) Subjects at risk iza-bren (0,3,6,9,12,15,18,21,24 mo): 216,172,143,109,68,39,15,0,0 Subjects at risk PBC: 206,153,78,40,18,8,5,1,0 [Slide 4] 08:30 - 10:00 Proffered paper 2: NSCLC, metastatic CHAIRS: PATRICK FORDE, ANTONIO PASSARO PFS per BICR Ruzaltatug rezetecan (n = 250) | Chemo (n = 248) Event, n (%) 108 (43.2) | 156 (62.9) Median PFS (95% CI), mo 8.3 (6.9–9.9) | 5.4 (4.4–5.6) HR* (95% CI) 0.47 (0.36–0.60) 1-sided P value† <0.0001 PFS (%) Time since randomization (mo) No. at risk • Consistent PFS benefit per investigator assessment: HR 0.46 (95% CI 0.36–0.60) * Stratified Cox proportional hazards model. † Stratified Log-rank test. Stratification factors were the same as those used for randomization. Chongrui Xu Ruzaltatug rezetecan, a HER3-directed ADC, vs platinum-based chemotherapy (PBC) in EGFR-mutated advanced NSCLC progressing after EGFR-TKI: Interim analyses (IA) from the phase 3 HERON-Lung01 trial ALICANTE AUDITORIUM - HALL 6 Survival table Endpoint: PFS per BICR Arm | N | Median | Median CI Ruzaltatug rezetecan | 250 | 8.3 mo | 6.9–9.9 Chemo | 248 | 5.4 mo | 4.4–5.6 HR: 0.47 HR CI: 0.36–0.60 P: <0.0001 (1-sided, stratified log-rank) 6-mo (dotted line at 6): 59.8% vs 32.4% (Δ27.4%) 9-mo (dotted line at 9): 46.3% vs 16.7% (Δ29.6%) No. at risk ruzaltatug (0-16 mo, 2-mo steps): 250,225,179,90,62,20,10,5,0 No. at risk chemo: 248,200,141,44,26,7,4,2,0

[Slide 1] SANOVO Study Design Randomized, double blind, multi-center phase 3 study conducted across 74 centers in China [NCT05009836]. Key inclusion criteria •Untreated unresectable or metastatic NSCLC1 •EGFRm (19del or L858R) •Centrally confirmed MET OE2 ➢ IHC 3+(≥90% of tumor cells demonstrating strong staining) or ➢ IHC 2+ (H-score≥220 and <90% of tumor cells demonstrating strong staining) •Age ≥ 18 years •ECOG PS 0-1 N = 320 R 1:1 Stratified by: EGFRm type ex19del vs L858R MET status IHC 3+ vs IHC 2+ Brain metastases yes vs no Savo 300 mg/200 mg BID (by BW) + osimertinib 80 mg QD Placebo 300 mg/200 mg BID (by BW) + osimertinib 80 mg QD Until PD or intolerable toxicity3 Endpoints Primary endpoint: PFS by INV Secondary endpoints: PFS by IRC; ORR, DoR, DCR by INV and IRC; OS Safety 1 Patients treated chemotherapy in neoadjuvant/adjuvant setting were included if progression occurred ≥6 months after the last dose; 2 MET IHC kit (D1C2, Amoy Diagnostics, Xiamen, China); 3 Patients received treatment until objective progressive disease by RECIST 1.1 as assessed by investigator, unacceptable toxicity, withdrawn consent, or another discontinuation criterion was met. Crossover between treatment group was not allowed; patients were followed up for safety outcomes 28 days after the last dose of study treatment. BID, twice daily; BW, body weight; DCR, disease control rate; DoR, duration of response; ECOG PS, European Cooperative Oncology Group performance status; EGFRm, epidermal growth factor receptor-mutated; Ex19del, exon 19 deletion; IHC, immunohistochemistry; INV, investigator; IRC, independent review committee; NSCLC, non-small cell lung cancer; ORR, objective response rate; OS, overall survival; PD, progressive disease; PFS, progression-free survival; QD, once daily; RECIST, Response Evaluation Criteria in Solid Tumours. Savo, Savolitinib; Osi, Osimertinib; OE, overexpression. [Slide 2] PFS in MET IHC 3+ and ITT Populations by INV MET IHC 3+ population Median follow-up: 30.4 mo Savo+Osi N = 98 | Osi N = 98 mPFS (95% CI), mo 24.8 (19.4, 27.6) | 11.1 (8.3, 16.6) HR (95% CI) 0.45 (0.31, 0.66) P-value p<0.0001 Landmarks: 76.0% vs 47.0%; 53.7% vs 28.3%; 26.9% vs 14.4% ITT population Median follow-up: 38.6 mo Savo+Osi N = 162 | Osi N = 164 mPFS (95% CI), mo 24.9 (22.0, 27.6) | 13.8 (10.9, 16.6) HR (95% CI) 0.56 (0.43, 0.74) P-value p<0.0001 Landmarks: 74.6% vs 54.9%; 54.9% vs 30.3%; 25.5% vs 19.3% Probability of PFS (%) Time (Mo) Savo+Osi Osi + Censored The combination of Savo + Osi reduced the risk of disease progression or death by 55% and 44%, respectively, compared to Osi alone CI, confidence interval; IHC, immunohistochemistry; INV, investigator; ITT, intention-to-treat; PFS, progression-free survival; Savo, Savolitinib; Osi, Osimertinib; HR, hazard ratio. Survival table Endpoint: PFS by INV (primary; MET IHC 3+ and ITT) Arm | N | Median | Median CI Savo+Osi (MET IHC 3+) | 98 | 24.8 mo | 19.4, 27.6 Osi (MET IHC 3+) | 98 | 11.1 mo | 8.3, 16.6 Savo+Osi (ITT) | 162 | 24.9 mo | 22.0, 27.6 Osi (ITT) | 164 | 13.8 mo | 10.9, 16.6 HR: MET IHC 3+: 0.45; ITT: 0.56 HR CI: MET IHC 3+: 0.31, 0.66; ITT: 0.43, 0.74 P: p<0.0001 (both) MET IHC 3+ | 12 mo | Savo+Osi: 76.0% | Osi: 47.0% MET IHC 3+ | 24 mo | Savo+Osi: 53.7% | Osi: 28.3% MET IHC 3+ | 36 mo | Savo+Osi: 26.9% | Osi: 14.4% ITT | 12 mo | Savo+Osi: 74.6% | Osi: 54.9% ITT | 24 mo | Savo+Osi: 54.9% | Osi: 30.3% ITT | 36 mo | Savo+Osi: 25.5% | Osi: 19.3% [Slide 3] Overall Survival Median follow-up: 38.7 mo Savo+Osi N = 162 | Osi N = 164 Number of death, n (%)* 49 (30.2) | 83 (50.6) Median OS, mo (95% CI) 44.5 (41.3, NE) | 34.9 (26.3, 41.1) HR (95% CI) 0.48 (0.33, 0.68) Probability of OS (%) Time (Mo) Savo+Osi Osi + Censored Landmarks: 94.4% vs 81.0% (Δ 13.4%); 79.8% vs 65.4% (Δ 14.4%); 67.4% vs 47.4% (Δ 20.0%) Savo + Osi reduced the risk of death by 52% in the ITT population compared with Osi alone, achieving a clinically meaningful survival benefit *At the data cutoff date, the overall survival maturity was 40.5%. CI, confidence interval; IHC, immunohistochemistry; ITT, intention-to-treat; OS, overall survival; Savo, Savolitinib; Osi, Osimertinib; HR, hazard ratio; NE, not estimable Survival table Endpoint: OS (ITT) Savo+Osi | 162 | 44.5 mo | 41.3, NE Osi | 164 | 34.9 mo | 26.3, 41.1 HR: 0.48 HR CI: 0.33, 0.68 12 mo | Savo+Osi: 94.4% | Osi: 81.0% 24 mo | Savo+Osi: 79.8% | Osi: 65.4% 36 mo | Savo+Osi: 67.4% | Osi: 47.4% [Slide 4] Safety Profile and Most Common TEAEs Exposure and safety summary | Savo+Osi N = 162 | Osi N = 164 Total exposure, mo -Median (range) | 22.1 (0.4, 44.2) | 12.9 (0.4, 45.5) At least one TEAE, n (%) | 162 (100.0) | 163 (99.4) Gr. ≥ 3 TEAE | 125 (77.2) | 63 (38.4) Leading to dose modification | 119 (73.5) | 57 (34.8) Leading to dose discontinuation | 24 (14.8) | 7 (4.3) Serious TEAE | 89 (54.9) | 47 (28.7) Leading to death (excl. PD) | 7 (4.3) | 8 (4.8) The most common TEAEs (≥25%) [Savo+Osi all grades / Savo+Osi grade≥3 | Placebo+Osi grade≥3 / Placebo+Osi all grades, %] Peripheral oedema 65 / 7 | 7 Hypoalbuminaemia 63 / 1 | 22 Platelet count decreased 51 / 2 | 2 / 32 Anaemia 49 / 6 | 2 / 40 Blood creatinine increased 47 | 23 AST increased 45 / 7 | 29 White blood cell count decreased 45 / 4 | 1 / 33 Neutrophil count decreased 42 / 4 | 2 / 25 ALT increased 41 / 7 | 27 Hypocalcaemia 41 / 2 | 10 Rash 39 / 2 | 30 Diarrhoea 38 / 1 | 1 / 38 Decreased appetite 35 / 1 | 1 / 23 Mouth ulceration 30 / 2 | 1 / 16 Hypokalaemia 27 / 7 | 2 / 12 Lymphocyte count decreased 27 / 9 | 4 / 18 Nausea 25 | 16 Weight decreased 23 / 1 | 1 / 25 Patients (%) Savolitinib + Osimertinib, all grades; Placebo + Osimertinib, all grades; Savolitinib + Osimertinib, Grades≥3; Placebo + Osimertinib, Grades≥3 Overall, Savo + Osi demonstrated a tolerable and manageable safety profile consistent with the known profiles of both agents and their combination, and no new safety concerns identified. TEAE, treatment-emerged adverse event; gr, grade; PD, progression disease; Savo, Savolitinib; Osi, Osimertinib; ALT, Alanine aminotransferase; AST, Aspartate aminotransferase

LBA69 - Savolitinib or placebo combined with osimertinib in treatment-naïve advanced NSCLC with EGFR mutation and MET overexpression: Results from the phase 3 SANOVO study Speakers Yi-Long Wu (Guangzhou, China) Lecture Time 08:30 - 08:40 Authors Yi-Long Wu (Guangzhou, China), Qing Zhou (Guangzhou, China), Longhua Sun (Nanchang, China), Baogang Liu (Harbin, China), Qiming Wang (Zhengzhou, China), XueYing Zhang (Changchun, China), Jianhua Chen (Changsha, China), Xingya Li (Zhengzhou, China), Jingzhang Li (Liuzhou, China), Tongmei Zhang (Beijing, China), Yuanyuan Ji (Anyang, China), Yun Fan (Hangzhou, China), Xiaorong Dong (Wuhan, China), Tienan Yi (Xiangyang, China), XuHong Min (Hefei, China), Liyun Miao (Nanjing, China), Hongrui Niu (Xinxiang, China), Yu Yao (Xi'an, China), Jie Yu (Shanghai, China), Songhua Fan (Shanghai, China) Abstract Background EGFR-mutant (EGFRm) advanced NSCLC patients (pts) with de novo MET overexpression (MET-OE) were observed to have reduced efficacy with first-line EGFR-TKI monotherapy (Ma L, et al). Here we report Ph3 SANOVO study (NCT05009836) primary results, evaluating savolitinib (savo, a potent, CNS penetrated METi) + osimertinib (osi) vs. placebo + osi in first-line EGFRm and MET-OE advanced NSCLC in China. Methods Treatment-naive pts with locally advanced/metastatic NSCLC harboring EGFRm (exon 19 del/L858R) and MET-OE (IHC 2+ [H-score ≥220, <90% tumor cells with strong staining] or IHC 3+ [≥90% tumor cells with strong staining]) were randomized 1:1 in a double-blind manner to savo+osi or placebo+osi, stratified by EGFRm type (exon 19 del/L858R), brain metastases (yes/no), and MET-OE level (IHC 2+/3+). Savo/placebo (300 or 200 mg BID, weight-based) and osi (80 mg QD) were dosed until disease progression, intolerable toxicity or death, whichever occurred first. The primary endpoint, investigator (INV)-assessed PFS, was tested hierarchically (MET IHC 3+ first, then ITT if positive). DCO: 31 Jul, 2026. Results Of 326 pts randomized (savo+osi, n=162; placebo+osi, n=164), 98 per group had MET IHC 3+ status. Baseline characteristics were generally balanced. Savo+osi significantly prolonged INV-assessed PFS vs. placebo+osi in both MET IHC 3+ (24.8 vs. 11.1 mo; HR 0.45; 95% CI 0.31 to 0.66; P<0.0001) and ITT populations (24.9 vs. 13.8 mo; HR 0.56; 95% CI, 0.43 to 0.74; P<0.0001). Median OS in the ITT population was 44.5 vs. 34.9 mo (HR, 0.48; 95% CI, 0.33 to 0.68) at a median follow-up of 38.7 mo. Median treatment exposure was 22.1 mo with savo+osi vs. 12.9 mo with placebo+osi, while the incidence of AEs leading to discontinuation was 14.8% vs. 4.3%. Safety was consistent with known profiles of this combination. Conclusions Savo+osi demonstrated a statistically significant PFS improvement and clinically meaningful OS benefit over osi alone in first-line EGFRm and MET-OE advanced NSCLC, with manageable safety and no new safety signals. This dual oral regimen offers a potential new treatment option for this biomarker-defined population.

[Slide 1] SANOVO Study Design Randomized, double blind, multi-center phase 3 study conducted across 74 centers in China [NCT05009836]. Key inclusion criteria •Untreated unresectable or metastatic NSCLC1 •EGFRm (19del or L858R) •Centrally confirmed MET OE2 ➢ IHC 3+(≥90% of tumor cells demonstrating strong staining) or ➢ IHC 2+ (H-score≥220 and <90% of tumor cells demonstrating strong staining) •Age ≥ 18 years •ECOG PS 0-1 N = 320 R 1:1 Stratified by: EGFRm type ex19del vs L858R MET status IHC 3+ vs IHC 2+ Brain metastases yes vs no Savo 300 mg/200 mg BID (by BW) + osimertinib 80 mg QD Placebo 300 mg/200 mg BID (by BW) + osimertinib 80 mg QD Until PD or intolerable toxicity3 Endpoints Primary endpoint: PFS by INV Secondary endpoints: PFS by IRC; ORR, DoR, DCR by INV and IRC; OS Safety 1Patients treated chemotherapy in neoadjuvant/adjuvant setting were included if progression occurred ≥6 months after the last dose; 2MET IHC kit (D1C2, Amoy Diagnostics, Xiamen, China); 3Patients received treatment until objective progressive disease by RECIST 1.1 as assessed by investigator, unacceptable toxicity, withdrawn consent, or another discontinuation criterion was met. Crossover between treatment group was not allowed, patients were followed up for safety outcomes 28 days after the last dose of study treatment. [Slide 4] Tumour Response by INV in ITT Population Savo+Osi, N = 162 | Osi, N = 164 | OR (95% CI) ORR, % (95% CI) | 80.9 (74.0, 86.6) | 67.7 (59.9, 74.8) | 2.00 (1.17, 3.45) DCR, % (95% CI) | 97.5 (93.8, 99.3) | 88.4 (82.5, 92.9) | 5.02 (1.66, 21.33) Median DoR, mo (95% CI) | 23.5 (20.1, 26.4) | 15.2 (12.4, 18.00) | - Median best % change in target lesion size, % (range) | -54.0 (-100.0, -10.8) | -44.3 (-100.0, 74.2) | - Savo+Osi Osi Best % Change In Target Lesion Size From Baseline Partial response Stable disease Progressive disease

[Slide 2] Patient Disposition Enrollment: Aug 2022~Aug 2025 DCO: Jul 31, 2026 Pre-screened/Screened N = 1896* 1449 EGFRm pts successfully tested MET IHC Randomization N = 326 Savo + Osimertinib N =162 Treated N =162 (100%) End of treatment, N = 101 (62.3%) • Disease progression, n = 76 (46.9%) • Adverse events, n = 11 (6.8%) • Patient's decision, n = 7 (4.3%) • Death, n = 5 (3.1%) • New anti-tumor treatment, n = 1 (0.6%) • Investigator's decision n = 1 (0.6%) Ongoing treatment N = 61 (37.7%) Placebo + Osimertinib N =164 Treated N = 164 (100%) End of treatment, N = 124 (75.6%) • Disease progression, n = 106 (64.6%) • Patient's decision, n = 8 (4.9%) • Adverse events, n = 5 (3.0%) • Death, n = 4 (2.4%) • Investigator's decision n = 1 (0.6%) Ongoing treatment N = 40 (24.4%) Percentages calculated based on randomized patients. *All patients who provided informed consent. DCO, data cut-off date. [Slide 3] PFS in MET IHC 3+ and ITT Populations by IRC MET IHC 3+ population Median follow-up: 30.4 mo Savo+Osi N = 98 | Osi N = 98 mPFS (95% CI), mo 22.2 (19.4, NE) | 13.7 (8.3, 19.4) HR (95% CI) 0.48 (0.32, 0.72) Landmarks: 76.5% vs 53.5%; 49.9% vs 33.4%; 41.5% vs 23.2% ITT population Median follow-up: 33.1 mo Savo+Osi N = 162 | Osi N = 164 mPFS (95% CI), mo 27.9 (22.1, 35.8) | 16.4 (9.7, 16.6) HR (95% CI) 0.53 (0.39, 0.71) Landmarks: 79.2% vs 55.8%; 56.8% vs 35.7%; 36.7% vs 26.9% Probability of PFS (%) Time (Mo) Savo+Osi Osi + Censored IRC-assessed PFS benefit was consistent with investigator assessment CI, confidence interval; IHC, immunohistochemistry; ITT, intention-to-treat; IRC, independent review committee; PFS, progression-free survival; Savo, Savolitinib; Osi, Osimertinib; HR, hazard ratio; NE, not estimable. Survival table Endpoint: PFS by IRC (MET IHC 3+ and ITT) Arm | N | Median | Median CI Savo+Osi (MET IHC 3+) | 98 | 22.2 mo | 19.4, NE Osi (MET IHC 3+) | 98 | 13.7 mo | 8.3, 19.4 Savo+Osi (ITT) | 162 | 27.9 mo | 22.1, 35.8 Osi (ITT) | 164 | 16.4 mo | 9.7, 16.6 HR: MET IHC 3+: 0.48; ITT: 0.53 HR CI: MET IHC 3+: 0.32, 0.72; ITT: 0.39, 0.71 MET IHC 3+ | 12 mo | Savo+Osi: 76.5% | Osi: 53.5% MET IHC 3+ | 24 mo | Savo+Osi: 49.9% | Osi: 33.4% MET IHC 3+ | 36 mo | Savo+Osi: 41.5% | Osi: 23.2% ITT | 12 mo | Savo+Osi: 79.2% | Osi: 55.8% ITT | 24 mo | Savo+Osi: 56.8% | Osi: 35.7% ITT | 36 mo | Savo+Osi: 36.7% | Osi: 26.9%

[Slide 1] Questions What is the right MET definition in the 1L population? Is there benefit in the 2+ population? How does a MET selected population fare with FLAURA2 or MARIPOSA regimen? What resistance will emerge from SANOVO combination? [Slide 2] Phase 3 SANOVO: Savolitinib or Placebo Combined with Osimertinib in Treatment-Naïve Advanced NSCLC with EGFR Mutation and MET Overexpression Key inclusion criteria •Untreated unresectable or metastatic NSCLC¹ •EGFRm (19del or L858R) •Centrally confirmed MET OE² ➤ IHC 3+(≥90% of tumor cells demonstrating strong staining) or ➤ IHC 2+ (H-score≥220 and <90% of tumor cells demonstrating strong staining) •Age ≥ 18 years •ECOG PS 0-1 N = 320 R 1:1 Stratified by: EGFRm type ex19del vs L858R MET status IHC 3+ vs IHC 2+ Brain metastases yes vs no Savo 300 mg/200 mg BID (by BW) + osimertinib 80 mg QD Placebo 300 mg/200 mg BID (by BW) + osimertinib 80 mg QD Until PD or intolerable toxicity³ Endpoints Primary endpoint: PFS by INV Secondary endpoints: PFS by IRC; ORR, DoR, DCR by INV and IRC; OS Safety Wu et al ESMO 2026;

[Slide 2] Demographics and Baseline Characteristics Characteristics | MET IHC 3+ population: Savo+Osi N = 98 | Osi N = 98 | ITT population: Savo+Osi N = 162 | Osi N = 164 Age (y) Median (Min, Max) | 62.0 (33, 80) | 63.5 (34, 83) | 62.0 (33, 80) | 61.0 (34, 89) ≥ 65 | 45 (45.9) | 49 (50.0) | 73 (45.1) | 72 (43.9) Sex, n (%) M / F | 36 (36.7) / 62 (63.3) | 48 (49.0) / 50 (51.0) | 64 (39.5) / 98 (60.5) | 78 (47.6) / 86 (52.4) Smoking, n (%) Never / Current or Previous | 68 (69.4) / 30 (30.6) | 68 (69.4) / 30 (30.6) | 116 (71.6) / 46 (28.4) | 109 (66.5) / 54 (32.9) * Pathological diagnosis, n (%) Adenocarcinoma / Other | 97 (99.0) / 1 (1.0) # | 95 (96.9) / 3 (3.1)† | 161 (99.4) / 1 (0.6)# | 158 (96.3) / 6 (3.7) †‡ ECOG PS, n (%) 0 / 1 | 24 (24.5) / 74 (75.5) | 21 (21.4) / 77 (78.6) | 36 (22.2) / 126 (77.8) | 38 (23.2) / 126 (76.8) Current tumour stage, n (%) Stage IIIB or IIIC / Stage IV | 7 (7.1) / 91 (92.9) | 3 (3.1) / 95 (96.9) | 10 (6.2) / 152 (93.8) | 4 (2.4) / 160 (97.6) MET IHC status by central lab, n (%) MET IHC 2+ / 3+ | 0 / 98 (100.0) | 0 / 98 (100.0) | 64 (39.5) / 98 (60.5) | 66 (40.2) / 98 (59.8) EGFR mutation subtype, n (%) Exon 19 Deletion/ Exon 21 (L858R) | 50 (51.0) / 48 (49.0) | 49 (50.0) / 49 (50.0) | 85 (52.5) / 77 (47.5) | 85 (51.8) / 79 (48.2) Brain metastasis at baseline, n (%) Y / N | 37 (37.8) / 61 (62.2) | 38 (38.8) / 60 (61.2) | 63 (38.9) / 99 (61.1) | 66 (40.2) / 98 (59.8) *: One unknown was not included. # One case was poorly differentiated carcinoma. † Two cases were squamous cell carcinoma, one case was mixed type. ‡ One case was squamous cell carcinoma, two cases were poorly differentiated carcinoma. ECOG PS, Eastern Cooperative Oncology Group performance status; EGFR, epidermal growth factor receptor; IHC, immunohistochemistry; ITT, intention-to-treat; M/F, male/female; Y/N, yes/no. [Slide 4] PFS in ITT Population by INV Across Subgroups Subgroup | Savo + Osi Event/N | Osi Event/N | HR (95% CI) Overall | 101/162 | 124/164 | 0.56 (0.43, 0.74) Sex Male | 48/64 | 61/78 | 0.75 (0.52, 1.10) Female | 53/98 | 63/86 | 0.58 (0.40, 0.83) Age at randomization < 65 | 55/89 | 74/92 | 0.54 (0.38, 0.76) >= 65 | 46/73 | 50/72 | 0.77 (0.52, 1.16) Baseline ECOG performance status 0 | 22/36 | 31/38 | 0.99 (0.57, 1.72) 1 | 79/126 | 93/126 | 0.57 (0.42, 0.77) Smoking status Current/Previous smoker | 35/46 | 44/54 | 0.74 (0.48, 1.17) Nonsmoker (Never) | 66/116 | 80/109 | 0.61 (0.44, 0.85) Current tumor stage IIIB/IIIC | 6/10 | 4/4 | 0.62 (0.15, 2.50) IV | 95/152 | 120/160 | 0.64 (0.48, 0.83) Brain metastases per eCRF Yes | 42/63 | 55/66 | 0.42 (0.28, 0.64) No | 59/99 | 69/98 | 0.76 (0.54, 1.08) MET status per eCRF IHC 3+ | 54/98 | 74/98 | 0.52 (0.36, 0.73) IHC 2+ | 47/64 | 50/66 | 0.85 (0.57, 1.27) Type of EGFR mutations per eCRF Exon 19 Deletion | 51/85 | 63/85 | 0.73 (0.50, 1.05) (Exon 21)L858R | 50/77 | 61/79 | 0.52 (0.36, 0.76) Hazard Ratio ←Savolitinib + Osimertinib better Placebo + Osimertinib better→ Savo + Osi demonstrated consistent PFS benefit compared with Osi alone across the ITT subgroups For subgroups, the HRs (savo+osi versus osi) and 95% CI were calculated using an un-stratified Cox proportional hazards model. CI, confidence interval; IHC, immunohistochemistry; ITT, intention-to-treat; INV, investigator; PFS, progression-free survival; ECOG, European Cooperative Oncology Group;

08:30 - 10:00 Proffered paper 2: NSCLC, metastatic CHAIRS: PATRICK FORDE, ANTONIO PASSARO My takeaways on Abstract LBA69 SANOVO: With SANOVO, Wu et al show benefit of adding savolitinib to osimertinib in the 1L treatment of EGFR mt (L858R, exon 19 del) with MET overexpression, especially with MET IHC 3+ The osimertinib arm in SANOVO suggests a possible detrimental effect of baseline MET overexpression on outcomes; this is ameliorated by addition of Savolitinib Benefit appears higher in more highly MET + (IHC3+) population ALICANTE AUDITORIUM - HALL 6

[Slide 2] SigVie-002 study design SigVie-002(a) is a global, randomized, open-label phase 3 study Key eligibility • Advanced, unresectable, or metastatic nonsquamous NSCLC • Prior treatment: • PBC • If no known AGAs: • PD-(L)1 inhibitor required • If known AGA detected: • ≥1 targeted therapy required • PD-(L)1 inhibitor optional • No prior anti-microtubule agents • ECOG PS ≤1 • No active CNS lesions(b) Enrollment period: March 6, 2024-June 5, 2025 Randomized 1:1 N=703 SV (n=354) 1.8 mg/kg AiBW IV Q2W Docetaxel (n=349) 75 mg/m2 IV Q3W Stratified by: • ECOG PS (0 versus 1) • AGAs (known EGFR mutations or ALK rearrangements; present versus absent or unknown) • Regions (North America and Western Europe versus Asia Pacific versus rest of world) Endpoints Primary • OS Secondary(c) • PFS by BICR and investigator • ORR by BICR and investigator • Safety and tolerability Exploratory(d) • Efficacy in IB6-high(e) subgroup Data cutoff: May 18, 2026 aNCT06012435. bPatients with definitively treated brain metastases are allowed if (1) CNS metastases have been clinically stable ≥4 weeks prior to treatment initiation with no evidence of new or enlarged metastasis on baseline scans; (2) patient is on a stable does of ≤10 mg/day of prednisone or equivalent for ≥2 weeks; (3) treatment with corticosteroids >1 month prior to the screening visit unless allowed as a concomitant therapy, and anticonvulsants at a stable dose were allowed; and (4) no evidence of clinical and radiographic disease progression in the CNS for ≥21 days after definitive radiotherapy and/or surgery. cAdditional secondary endpoints included duration of response by BICR and investigator and quality of life in both treatment arms and pharmacokinetics and immunogenicity of SV. dAdditional exploratory endpoints were included but not presented here. eIB6-high was defined as ≥50% of tumor cells showing immunohistochemistry staining at ≥2+ intensity. [Slide 3] Primary and secondary endpoints Primary: OS in overall population | SV (n=354) | Docetaxel (n=349) OS, median (95% CI), months | 11.9 (10.1–14.0) | 11.5 (10.2–13.3) HR (95% CI) | 0.929 (0.767–1.126) Stratified log-rank 1-sided P value | 0.226 No. at risk SV 354 317 266 206 158 81 35 10 1 1 Docetaxel 349 294 257 201 154 72 29 9 5 0 Secondary: PFS by BICR in overall population | SV (n=354) | Docetaxel (n=349) PFS, median (95% CI), months | 4.7 (4.2–6.0) | 4.2 (4.0–6.0) HR (95% CI) | 0.844 (0.699–1.019) No. at risk SV 354 211 126 59 32 13 6 2 0 Docetaxel 349 175 103 41 23 9 1 0 0 Secondary: cORR by BICR in overall population | SV (n=354) | Docetaxel (n=349) cORR (95% CI), % | 16 (12–20) | 13 (10–17) DCR (95% CI), % | 69 (64–73) | 60 (55–65) DOR, median (95% CI),a months | 8.3 (6.2–NR) | 10.1 (6.2–14.5) In the overall population, SV did not show a statistically significant improvement for the primary endpoint of OS or secondary endpoints aCalculated out of responders. Survival table Endpoint: OS (primary) and PFS by BICR (secondary), overall population Arm | N | Median | Median CI SV | 354 | OS 11.9 mo; PFS 4.7 mo | OS 10.1–14.0; PFS 4.2–6.0 Docetaxel | 349 | OS 11.5 mo; PFS 4.2 mo | OS 10.2–13.3; PFS 4.0–6.0 HR: OS 0.929; PFS 0.844 HR CI: OS 0.767–1.126; PFS 0.699–1.019 P: OS stratified log-rank 1-sided P = 0.226 [Slide 4] Overall survival by subgroups Subgroup | Events/n SV | Events/n Docetaxel | HR (95% CI) Overall | 211/354 | 215/349 | 0.93 (0.77–1.13) Age group <65 years | 103/173 | 113/193 | 1.04 (0.80–1.36) Age group ≥65 years | 108/181 | 102/156 | 0.83 (0.63–1.08) Sex Male | 130/210 | 127/191 | 0.88 (0.69–1.13) Sex Female | 81/144 | 88/158 | 0.98 (0.73–1.33) Region(a) North America and Western Europe | 149/249 | 163/246 | 0.84 (0.67–1.05) Region(a) Asia Pacific | 30/58 | 27/57 | 1.25 (0.74–2.11) Region(a) Rest of world | 32/47 | 25/46 | 1.20 (0.71–2.02) ECOG performance status(a) 0 | 64/124 | 63/121 | 0.96 (0.68–1.37) ECOG performance status(a) 1 | 147/230 | 152/228 | 0.93 (0.74–1.16) Smoking history Former or current | 167/281 | 174/272 | 0.88 (0.71–1.09) Smoking history Never | 43/72 | 41/77 | 1.16 (0.76–1.79) EGFR or ALK alterations(a) Present | 52/84 | 41/81 | 1.32 (0.87–1.99) EGFR or ALK alterations(a) Absent or unknwon | 159/270 | 174/268 | 0.85 (0.68–1.05) AGA status AGA | 70/113 | 53/95 | 1.14 (0.80–1.63) AGA status Non-AGA | 141/241 | 162/254 | 0.86 (0.69–1.08) Line of systemic therapy 2L | 134/236 | 151/241 | 0.83 (0.66–1.05) Line of systemic therapy 3L+ | 77/117 | 63/105 | 1.19 (0.85–1.66) Axis: 0.5 – 1 – 5; ← Favors SV | Favors docetaxel → AGA presence by treatment line (No AGAs / ≥1 AGA) 2L: 10% 3L+: 72% aPer randomization and trial supply management.

PHASE 3 TRIALS WITH ADCs in 2L NOA-NSCLC TRIAL | ADC | ADC-TARGET | HISTOLOGY | POPULATION | PRIMARY | STATUS/ RESULTS TROPION-Lung01 | Dato Deruxtecan | TROP2 | NSCLC | Unselected (+/- AGAS) | Dual PFS/OS | POSITIVE PFS NEGATIVE OS ✗ EVOKE-01 | Saci Govitecan | TROP2 | NSCLC | Unselected (+/- AGAs) | OS | NEGATIVE OS ✗ CARMEN-LC03 | Tusa Ravtansine | CEACAM5 | NSQ-NSCLC | Selected CEACAM5 OE | Dual PFS/OS | NEGATIVE OS ✗ SigVie-002 (Be6A Lung-01) | Sigvotatug Vedotin SGN-B6A | IB6 | NSQ-NSCLC | Unselected (+/- AGAs) No previous taxanes | PFS/OS | NEGATIVE OS (press release June 26) ✗ PANKU-Lung02 | Iza-Bren | EGFR/HER3 | NSCLC | Unselected (EGFR WT) | PFS/OS | Active, not recruiting PADL1NK-0052 | Fetrastobart vedotin (Fetra-V) | PD-L1 | NSCLC | TPS ≥1% AGA permissive | OS | Recruiting HLX43-NSCLC302 | HLX43 (ADC) ± HLX07 (EGFR mAb) | PD-L1 | SQ-NSCLC | AGA permissive | Dual PFS/OS | Recruiting TeliMET NSCLC-01 | Teliso-V ABBV-399 | c-MET | NSQ-NSCLC | Selected c-Met OE | Co-primary PFS/OS | Recruiting TROPION-Lung17 | Dato Deruxtecan | TROP2 | NSQ-NSCLC | Selected TROP2 NMR + | Dual PFS/OS | Recruiting Noemi Reguart | ESMO Therapeutic Landscapes Webinar 2026

LBA68 - Phase 3 study of sigvotatug vedotin (SV) vs docetaxel (DTX) in previously treated nonsquamous non-small cell lung cancer (NSCLC) Speakers Solange Peters (Lausanne, Switzerland) Lecture Time 11:13 - 11:23 Abstract Background SV, a novel integrin beta-6 (IB6)-directed ADC, showed promising antitumor activity in patients (pts) with heavily pretreated advanced NSCLC in the phase 1 SGNB6A-001 study. Phase 3 SigVie-002 (NCT06012435) results are reported. Methods Pts with nonsquamous NSCLC were randomized to SV (1.8 mg/kg by AiBW IV Q2W) or DTX (75 mg/m2 IV Q3W). Eligible pts received prior platinum-based chemotherapy; prior PD-(L)-1 inhibitor was required by pts with no known actionable genomic alterations (AGAs); and ≥1 targeted therapy was required for known AGAs. Endpoints: primary, overall survival (OS); secondary, progression-free survival (PFS) and safety. Results As of May 18, 2026, 703 pts were assigned to SV (n=354) or DTX (n=349) with balanced baseline characteristics. Overall, ≥1 AGA, 30%; EGFR mutation, 19%. Median OS: 11.9 mo with SV and 11.5 mo with DTX (HR=0.929; one-sided P=0.226). Median PFS by BICR: 4.7 mo with SV and 4.2 mo with DTX (HR=0.844). Post-hoc analysis in second line (2L) subgroup (68% of overall; 90% non-AGAs) showed median OS of 13.6 mo with SV and 11.1 mo with DTX (HR=0.831), along with PFS benefit (HR=0.795). In contrast, in third line or later (3L+) subgroup (32% of overall; 72% AGAs), no similar benefit trends were observed. There was no clear relationship between IB6 expression level and efficacy. Any-grade and grade≥3 treatment-related adverse events (TRAEs) occurred in 91% and 33% with SV and 87% and 36% with DTX, respectively. TRAEs led to treatment discontinuation in 11% with SV and 9% with DTX. Table: LBA68 | SV | DTX | Hazard ratio (95% CI) Overall population, n | 354 | 349 | Median OS (95% CI), mo | 11.9 (10.1-14.0) | 11.5 (10.2-13.3) | 0.929 (0.767-1.126) one-sided p=0.226 Median PFS (95% CI), mo | 4.7 (4.2-6.0) | 4.2 (4.0-6.0) | 0.844 (0.699-1.019) 2L patients, n | 236 | 241 | Median OS (95% CI), mo | 13.6 (10.7-16.4) | 11.1 (9.6-13.3) | 0.831 (0.658-1.049) Median PFS (95% CI), mo | 5.9 (4.3-6.3) | 4.2 (4.0-6.0) | 0.795 (0.634-0.996) 3L+ patients, n | 117 | 105 | Median OS (95% CI), mo | 10.0 (8.2, 12.0) | 11.9 (8.9, 16.8) | 1.187 (0.850, 1.657) Median PFS (95% CI), mo | 4.2 (3.9, 5.0) | 4.2 (3.3, 6.0) | 1.024 (0.739, 1.419) Conclusions SV did not significantly improve OS over DTX in overall population. Notably, favorable OS and PFS trends were observed with SV in 2L subgroup, which consisted primarily of non-AGA pts. Taken together with phase 1 data showing manageable safety and promising preliminary efficacy in combination with pembrolizumab, these results support further development of SV in combination with pembrolizumab or PF-0864404 (PD1/VEGF bispecific) in earlier settings.

[Slide 3] Outcomes in the 2L treatment setting (post hoc analysis) OS No. at risk SV 236 210 179 144 117 63 24 8 1 0 Docetaxel 241 206 174 138 104 45 19 5 2 0 | SV (n=236) | Docetaxel (n=241) OS, median (95% CI), months | 13.6 (10.7–16.4) | 11.1 (9.6–13.3) Unstratified HR (95% CI) | 0.831 (0.658–1.049) PFS No. at risk SV 236 144 90 46 24 10 4 2 0 Docetaxel 241 121 74 30 17 6 1 0 0 | SV (n=236) | Docetaxel (n=241) PFS, median (95% CI), months | 5.9 (4.3–6.3) | 4.2 (4.0–6.0) Unstratified HR (95% CI) | 0.795 (0.634–0.996) In patients who received SV in the 2L, favorable OS and PFS trends were observed Survival table Endpoint: OS and PFS in 2L treatment setting (post hoc) Arm | N | Median | Median CI SV | 236 | OS 13.6 mo; PFS 5.9 mo | OS 10.7–16.4; PFS 4.3–6.3 Docetaxel | 241 | OS 11.1 mo; PFS 4.2 mo | OS 9.6–13.3; PFS 4.0–6.0 HR: OS 0.831; PFS 0.795 (unstratified) HR CI: OS 0.658–1.049; PFS 0.634–0.996 [Slide 4] Efficacy by IB6 expression status (exploratory analysis) • IB6 was retrospectively tested using mostly archival tissue samples with an immunohistochemistry assay and was classified as IB6-high and -low using the predetermined cutoff of 50% for expression – IB6-high: ≥50% positive tumor cells at ≥2+ staining intensity – IB6-low: <50% positivity of tumor cells at ≥2+ staining intensity, including tumors with 0% positivity of tumor cells • The 50% cutoff was defined based on preliminary data from the phase 1 study • Of the evaluable patients, 72% had IB6-high expression OS — SV | IB6-low (n=72) | IB6-high (n=180) OS, median (95% CI), months | 13.4 (8.8–19.6) | 12.1 (9.9–14.4) HR of SV vs docetaxel (95% CI) | 0.668 (0.424–1.054) | 0.913 (0.695–1.199) No. at risk: IB6-High 180 138 85 11 0 0; IB6-Low 72 56 32 13 0 OS — Docetaxel | IB6-low (n=67) | IB6-high (n=186) OS, median (95% CI), months | 9.0 (7.3–11.9) | 12.4 (10.7–14.1) No. at risk: IB6-High 186 143 87 16 2 0; IB6-Low 67 45 22 2 0 PFS — SV | IB6-low (n=72) | IB6-high (n=180) PFS, median (95% CI), months | 4.3 (2.6–6.2) | 5.6 (4.1–6.2) HR of SV vs docetaxel (95% CI) | 0.753 (0.476–1.192) | 0.890 (0.683–1.159) No. at risk: IB6-High 180 70 19 2 0; IB6-Low 72 22 5 2 0 PFS — Docetaxel | IB6-low (n=67) | IB6-high (n=186) PFS, median (95% CI), months | 4.0 (2.3–6.0) | 4.3 (4.0–6.1) No. at risk: IB6-High 186 58 15 1 0; IB6-Low 67 19 3 0 0 No clear trend was observed; this was consistent across different thresholds tested

Primary and secondary endpoints Primary: OS in overall population | SV (n=354) | Docetaxel (n=349) OS, median (95% CI), months | 11.9 (10.1–14.0) | 11.5 (10.2–13.3) HR (95% CI) | 0.929 (0.767–1.126) Stratified log-rank 1-sided P value | 0.226 No. at risk SV 354 317 266 206 158 81 35 10 1 1 Docetaxel 349 294 257 201 154 72 29 9 5 0 Secondary: PFS by BICR in overall population | SV (n=354) | Docetaxel (n=349) PFS, median (95% CI), months | 4.7 (4.2–6.0) | 4.2 (4.0–6.0) HR (95% CI) | 0.844 (0.699–1.019) No. at risk SV 354 211 126 59 32 13 6 2 0 Docetaxel 349 175 103 41 23 9 1 0 0 Secondary: cORR by BICR in overall population | SV (n=354) | Docetaxel (n=349) cORR (95% CI), % | 16 (12–20) | 13 (10–17) DCR (95% CI), % | 69 (64–73) | 60 (55–65) DOR, median (95% CI),a months | 8.3 (6.2–NR) | 10.1 (6.2–14.5) In the overall population, SV did not show a statistically significant improvement for the primary endpoint of OS or secondary endpoints aCalculated out of responders. [Livestream frame: CHAIR: SURESH SENAN; speaker card: Solange Peters — Phase 3 study of sigvotatug vedotin (SV) vs docetaxel (DTX) in previously treated nonsquamous non-small cell lung cancer (NSCLC)] Survival table Endpoint: OS (primary) and PFS by BICR (secondary), overall population Arm | N | Median | Median CI SV | 354 | OS 11.9 mo; PFS 4.7 mo | OS 10.1–14.0; PFS 4.2–6.0 Docetaxel | 349 | OS 11.5 mo; PFS 4.2 mo | OS 10.2–13.3; PFS 4.0–6.0 HR: OS 0.929; PFS 0.844 HR CI: OS 0.767–1.126; PFS 0.699–1.019 P: OS stratified log-rank 1-sided P = 0.226

Baseline characteristics Characteristic | SV (n=354) | Docetaxel (n=349) Age, median (range), years | 65 (25-85) | 63 (33-87) Male, n (%) | 210 (59) | 191 (55) Race, n (%) White, n (%) | 227 (64) | 229 (66) Race, n (%) Asian | 61 (17) | 59 (17) Race, n (%) Black or African American | 10 (3) | 1 (<1) Race, n (%) Other/multiple/unknown/not reportable | 56 (16) | 60 (17) Smoking status,a n (%) Former or current | 281 (79) | 272 (78) Smoking status,a n (%) Never | 72 (20) | 77 (22) ECOG PS of 1,b n (%) | 236 (67) | 227 (65) IB6 status,c n (%) IB6-high | 180 (51) | 186 (53) IB6 status,c n (%) IB6-low | 72 (20) | 67 (19) IB6 status,c n (%) Not evaluabled | 102 (29) | 96 (28) Metastatic disease, n (%) | 344 (97) | 333 (95) History of brain metastases,e n (%) | 54 (15) | 62 (18) AGAs, n (%) Any AGA presentf | 113 (32) | 95 (27) AGAs, n (%) EGFR mutation | 66 (19) | 68 (19) AGAs, n (%) ALK rearrangement | 4 (1) | 3 (1) AGAs, n (%) KRAS G12C mutation | 24 (7) | 16 (5) Prior lines of systemic therapy, n (%) 0 | 1 (<1) | 3 (1) Prior lines of systemic therapy, n (%) 1 | 236 (67) | 241 (69) Prior lines of systemic therapy, n (%) ≥2 | 117 (33) | 105 (30) • Baseline characteristics were well-balanced between arms • At the time of data cutoff, most patients (>90%) had discontinued treatment, with majority due to progressive disease (SV: 72%; docetaxel: 64%) • 15 patients in each arm had treatment ongoing • For patients with no known AGAs, all had received prior PBC and almost all had received a PD-(L)1 inhibitorg • For patients with ≥1 known AGA, almost all (99%) received prior PBC, about half (47%) received prior PD-(L)1 inhibitor, and 86% received prior targeted therapy aOne patient in the SV arm had missing information. bOne patient in each treatment arm had missing information. cBoth fresh and archived tumor samples were accepted for IB6 expression testing. IB6-high was defined as ≥50% of tumor cells showing immunohistochemistry staining at ≥2+ intensity. IB6-low was defined as <50% tumor cells showing immunohistochemistry staining at ≥2+ intensity. dNot evaluable included patients with no tissue sample submitted or the submitted tissue sample was not evaluable due to sample type of insufficient tumor content. eIncludes patients with a history of brain radiotherapy, surgery, and/or presence of baseline target or non-target lesions in the CNS per investigator assessment. fCategories are not mutually exclusive. Additional categories not listed here included BRAF V600 mutation, NTRK rearrangement, HER2 mutation, MET mutation, RET rearrangement, ROS1 mutation, and other AGA. gOne patient in each of the SV and docetaxel arms did not receive a PD-(L)1 inhibitor.

FPN: LBA68 Phase 3 study of sigvotatug vedotin vs docetaxel in previously treated nonsquamous non-small cell lung cancer Solange Peters,1 Noboru Yamamoto,2 William N. William,3 Michael L. Cheng,4 Laurent Greillier,5 Oscar Juan-Vidal,6 Enriqueta Felip,7 Izabela Chmielewska,8 Manuel Cobo-Dols,9 Federico Cappuzzo,10 Yi-Long Wu,11 David Planchard,12 Hidehito Horinouchi,13 Robert Jotte,14 Vishal Navani,15 Scott Knowles,16 Jingmin Liu,16 Eric Kim,16 Polly Cronin,17 Luis Paz-Ares18 1Oncology Department, Centre Hospitalier Universitaire Vaudois, Lausanne University Hospital, Lausanne, Switzerland; 2Department of Experimental Therapeutics, National Cancer Center Hospital, Tokyo, Japan; 3Departments of Thoracic Oncology and Head and Neck Oncology, Grupo Oncoclínicas, São Paulo, Brazil; 4Department of Medicine, University of California San Francisco, San Francisco, CA, USA; 5Multidisciplinary Oncology and Therapeutic Innovations Department, Assistance Publique – Hôpitaux de Marseille, Aix Marseille Univ, Marseille, France; 6Department of Medical Oncology, Hospital Universitari i Politècnic La Fe, Valencia, Spain; 7Oncology Department, Vall d'Hebron University Hospital, Vall d'Hebron Institute of Oncology, Barcelona, Spain; 8Department of Pneumonology, Oncology and Allergology, Medical University of Lublin and Institute of Genetics and Immunology, GENIM, Lublin, Poland; 9Medical Oncology, Hospital Quironsalud Málaga, Málaga, Spain; 10IRCCS Regina Elena National Cancer Institute, Rome, Italy; 11Guangdong Lung Cancer Institute, Guangdong Provincial People's Hospital and Guangdong Academy of Medical Sciences, Guangdong, China; 12Department of Medical Oncology, Gustave Roussy, Villejuif and Faculty of Medicine, Paris-Saclay University, Paris, France; 13Department of Thoracic Oncology, National Cancer Center Hospital, Tokyo, Japan; 14SCRI at Rocky Mountain Cancer Centers, Denver, CO, USA; 15Division of Medical Oncology, Arthur Child Cancer Centre, Calgary, AB, Canada; 16Late Stage Oncology Development, Pfizer, Bothell, WA, USA; 17Translational Medicine, Pfizer, Bothell, WA, USA; 18Department of Medical Oncology, Hospital Universitario 12 de Octubre, Madrid, Spain

[Slide 2] Enter DatoDxd ...and Sac-TMT.... Dato DxD: TROP2, exetecan, DAR 4 Deruxtecan; Cleavable Tetrapeptide-Based Linker; Payload Topoisomerase I Inhibitor (DXd) DatoDxd was approved by the US FDA based on a pooled analysis of EGFR+ NSCLC pts (n=114) from: • TROPION-LUNG 05 (single arm phase 2 +AGAs) • TROPION-LUNG 01 (RP3 all comer NSCLC) SacTMT: TROP2, tirumotecan, DAR 7.4 • Humanized IgG1 κ • MW: ~148 kD • Fc domain was remodeled by mutations in L237A, L238A, and G240A, without ADCC/CDC Profile: Unique PD-L1 antigen-binding epitope; Excellent pharmaceutical properties; Good pharmacokinetics and safety Structural schema of KL-A167 SacTMT approved by Chinese Regulatory authorities based on OptiTROP-Lung 04 Melissa L. Johnson MD, SCRI [Slide 3] TROP ADCs Taught Us More Lessons • ADCs are not always as effective as T-Dxd • ADCs are not really chemo-free • Identifying ADC efficacy is not always easy TROPION-Lung 01 Final Analysis Overall Survival: Subgroup Analyses No. of events/No. of patients | Dato-DXd | Docetaxel | HR Histology: Nonsquamous | 160/234 | 163/234 | 0.84 Squamous | 55/65 | 55/71 | 1.32 Actionable genomic alterationsa: Absent | 182/249 | 185/254 | 0.97 Present | 33/50 | 33/51 | 0.66 Favors Dato-DXd | Favors docetaxel Melissa L. Johnson MD, SCRI Data cutoff: March 1, 2024 [partially cut off]

LBA67 - Final OS analysis of sacituzumab tirumotecan (sac-TMT) versus platinum-based chemotherapy in EGFR-mutated non-small cell lung cancer (NSCLC) after progression on EGFR-TKI: Phase III OptiTROP-Lung04 study Speakers Wen Feng Fang (Guangzhou, China) Lecture Time 11:03 - 11:13 Abstract Background Sac-TMT is a TROP2-directed ADC with a unique bifunctional linker and a belotecan-derivative topoisomerase I inhibitor payload. In phase III OptiTROP-Lung04, sac-TMT significantly improved PFS and OS versus platinum-based chemotherapy in patients (pts) with EGFRm NSCLC after progression on EGFR-TKI at the prespecified interim OS analysis (Fang et al. NEJM 2025). Here, we report the prespecified final OS analysis with extended follow-up. Methods Eligible pts with locally advanced or metastatic EGFRm NSCLC who had progressed on EGFR-TKI therapy were randomized (1:1) to receive sac-TMT monotherapy (5 mg/kg Q2W) or pemetrexed plus platinum, followed by maintenance pemetrexed. OS was a key secondary endpoint. OS was adjusted for subsequent ADC therapy in the chemotherapy group using the rank-preserving structural failure time (RPSFT) model. Results A total of 376 pts were randomly assigned (188/group). As of 06 August, 2026, median follow-up was 32.3 months (mo). The statistically significant OS benefit observed at the prespecified interim analysis was sustained in the final analysis. Median OS was 25.9 mo (95% CI: 22.2-27.5) with sac-TMT vs 17.4 mo (95% CI: 15.7-20.3) with chemotherapy (HR 0.69, 95% CI: 0.54-0.89; nominal one-sided p=0.0021). The 24-mo and 30-mo OS rates were 54.1% vs 37.2% and 40.5% vs 28.4%, respectively. Subsequent ADC therapy was received by 22.6% of pts in the chemotherapy group. After RPSFT analysis adjusted for subsequent ADC therapy, median OS was 25.9 vs 16.6 mo (HR 0.57, 95% CI: 0.42-0.75). No new safety signals emerged with extended follow-up. Conclusions With an extended follow-up of 32.3 mo, sac-TMT continues to demonstrate clinically meaningful and sustained OS benefit compared with platinum-based chemotherapy in advanced EGFRm NSCLC after progression on EGFR-TKI therapy, including after adjustment for subsequent ADC therapy. These findings further support sac-TMT as an effective treatment option in this patient population.

[Slide 1] Adjusted Overall Survival* After adjustment for subsequent ADC therapy using pre-specified RPSFT model, sac-TMT significantly improved OS over chemotherapy with 43% lower risk of death. | Sac-TMT (n = 188) | Chemo (n = 188) OS events, n (%) | 114 (60.6) | 119 (63.3) Median, mo (95% CI) | 25.9 (22.2, 27.5) | 16.6 (15.1, 19.1) 24-mo OS rate, % (95% CI) | 54.1 (46.6, 60.9) | 32.1 (24.6, 39.8) Sac-TMT mOS: 25.9 mo Chemo mOS: 16.6 mo HR 0.57 (95% CI: 0.42–0.75) Legend: Sac-TMT; Chemotherapy No. at risk Sac-TMT 188 184 167 158 147 133 123 111 100 81 58 27 1 0 Chemotherapy 188 180 162 147 129 110 84 48 8 0 *An RPSFT model was used to account for the potential impact of subsequent ADC therapy on OS among patients in the chemotherapy group who received such treatment. Data cutoff: 06 August 2026 Professor Wenfeng Fang [Livestream frame: speaker card: Wen Feng Fang — Final OS analysis of sacituzumab tirumotecan (sac-TMT) versus platinum-based chemotherapy in EGFR-mutated non-small cell lung cancer (NSCLC) after progression on EGFR-TKI: Phase III OptiTROP-Lung04 study] Survival table Endpoint: Overall survival, RPSFT-adjusted for subsequent ADC therapy Arm | N | Median | Median CI Sac-TMT | 188 | 25.9 mo | 22.2, 27.5 Chemotherapy | 188 | 16.6 mo | 15.1, 19.1 HR: 0.57 HR CI: 0.42–0.75 Landmark rates: 24 mo: Sac-TMT 54.1% (CI 46.6, 60.9) | Chemotherapy 32.1% (CI 24.6, 39.8) [Slide 2] Conclusions • At the pre-specified final OS analysis (median follow-up: 32.3 months), sac-TMT demonstrated a sustained, clinically meaningful OS benefit over platinum-based chemotherapy. - Median OS was 25.9 vs 17.4 months (HR, 0.69, 95% CI, 0.54–0.89). The 24-month OS rate was 54.1% vs 37.2%, an absolute difference of 16.9 percentage points. • The pre-specified analysis accounting for subsequent ADC therapy supported the ITT result (HR, 0.57; 95% CI, 0.42–0.75). • No new safety signals emerged with extended follow-up. Common TRAEs in the sac-TMT group included hematologic toxicities and stomatitis. These findings reinforce sac-TMT as a treatment option with a demonstrated survival benefit in advanced EGFR-mutated non-squamous NSCLC after progression on EGFR-TKI therapy. Professor Wenfeng Fang [Livestream frame: speaker card: Wen Feng Fang — Final OS analysis of sacituzumab tirumotecan (sac-TMT) versus platinum-based chemotherapy in EGFR-mutated non-small cell lung cancer (NSCLC) after progression on EGFR-TKI: Phase III OptiTROP-Lung04 study]

[Slide 2] Overall Survival Sac-TMT demonstrated a sustained OS benefit over chemotherapy, with a 31.0% reduction in the risk of death, corresponding to an 8.5-month difference in median OS between the two groups. Curve labels: Sac-TMT mOS: 25.9 mo; Chemo mOS: 17.4 mo | Sac-TMT (n = 188) | Chemo (n = 188) OS events, n (%) | 114 (60.6) | 137 (72.9) Median, mo (95% CI) | 25.9 (22.2, 27.5) | 17.4 (15.7, 20.3) 24-mo OS rate, % (95% CI) | 54.1 (46.6, 60.9) | 37.2 (30.3, 44.2) 30-mo OS rate, % (95% CI) | 40.5 (33.4, 47.5) | 28.4 (22.0, 35.2) HR 0.69 (95% CI: 0.54–0.89) Nominal one-sided P = 0.0021 No. at risk (months 0-39, every 3) Sac-TMT 188 184 167 158 147 133 123 111 100 81 58 27 1 0 Chemotherapy 188 180 162 147 132 114 89 74 67 59 40 19 2 0 Median follow-up: 32.3 months Data cutoff: 06 August 2026. CI, confidence interval; HR, hazard ratio; mo, month. Professor Wenfeng Fang Survival table Endpoint: Overall survival (final analysis) Arm | N | Median | Median CI Sac-TMT | 188 | 25.9 mo | 22.2, 27.5 Chemo | 188 | 17.4 mo | 15.7, 20.3 HR: 0.69 HR CI: 0.54–0.89 P: nominal one-sided 0.0021 Landmark rates: 24 months: Sac-TMT 54.1% (CI 46.6, 60.9) | Chemo 37.2% (CI 30.3, 44.2) 30 months: Sac-TMT 40.5% (CI 33.4, 47.5) | Chemo 28.4% (CI 22.0, 35.2) [Slide 4] TRAEs in ≥ 20% of Patients TRAE | Sac-TMT Any Grade | Sac-TMT Grade ≥3 | Chemotherapy Grade ≥3 | Chemotherapy Any Grade (%) Anemia | 85.1 | 11.7 | 14.3 | 76.4 Alopecia* | 84.0 | 0 | 0 | 9.3 WBC decreased | 83.5 | 28.2 | 22.0 | 69.8 NEUT decreased | 76.1 | 40.4 | 33.0 | 69.2 Stomatitis | 62.2 | 4.8 | 0 | 3.8 Nausea | 47.9 | 0.5 | 1.1 | 47.3 Decreased appetite | 41.5 | 0 | 0 | 31.9 Fatigue | 38.3 | 3.7 | 2.2 | 40.1 Weight decreased | 27.7 | 0 | 0.5 | 15.4 PLT decreased | 27.1 | 2.1 | 16.5 | 46.7 Vomiting | 27.1 | 0 | 0.5 | 21.4 ALT increased | 25.5 | 0.5 | 1.1 | 34.6 Constipation | 21.3 | 0 | 0 | 17.0 AST increased | 19.7 | 0.5 | 1.1 | 34.6 • The most common TRAEs in both groups were hematologic toxicities, with a lower incidence of PLT decreased in the sac-TMT group. • Stomatitis occurring with sac-TMT group was mostly grade 1–2 and manageable, with grade 3 events reported in 4.8% of patients. *An evaluation of alopecia consistent with hair loss was based on a patient's self-reported experiences included in the Patient-Reported Outcome (PRO) questionnaires ALT, alanine aminotransferase; AST, aspartate aminotransferase; PLT, platelet; NEUT, neutrophil; WBC, white blood cell. Professor Wenfeng Fang

Subsequent Anticancer Therapy | Sac-TMT (n = 173)* | Chemo (n = 186)* At least one treatment, n (%) | 132 (76.3) | 160 (86.0) Chemotherapy | 85 (49.1) | 101 (54.3) Pemetrexed-based chemotherapy | 78 (45.1) | 25 (13.4) EGFR tyrosine kinase inhibitor | 75 (43.4) | 77 (41.4) Anti-angiogenic agent** | 61 (35.3) | 90 (48.4) Immunotherapy† | 41 (23.7) | 48 (25.8) Antibody-drug conjugate | 6 (3.5) | 42 (22.6) TROP2 ADC | 3 (1.7) | 16 (8.6) HER3 ADC | 0 | 15 (8.1) EGFR ADC | 0 | 4 (2.2) EGFR/HER3 ADC | 1 (0.6) | 1 (0.5) *Patients who discontinued treatment and those who did not receive any treatment after randomization. **Including anti-angiogenic agent with or without combined immunotherapy (excluding PD(L)-1/VEGF bispecific antibody). †Containing PD(L)-1 inhibitor and PD(L)-1/VEGF bispecific antibody. Data cutoff: 06 August 2026 Professor Wenfeng Fang

[Slide 2] Patient Demographics and Baseline Characteristics Characteristic | Sac-TMT (n = 188) | Chemo (n = 188) Median age (range), years | 60 (31-75) | 59 (33-75) ≥ 65 years, n (%) | 58 (30.9) | 51 (27.1) Male, n (%) | 66 (35.1) | 83 (44.1) ECOG PS 1, n (%) | 153 (81.4) | 145 (77.1) Smoking history, n (%) Never smoked | 145 (77.1) | 135 (71.8) Current or former smoker | 43 (22.9) | 53 (28.2) Clinical stage at enrollment, n (%) Stage IIIB/IIIC | 6 (3.2) | 3 (1.6) Stage IV | 182 (96.8) | 185 (98.4) ≥ 3 metastatic sites, n (%) | 128 (68.1) | 126 (67.0) Brain metastases, n (%) | 33 (17.6) | 36 (19.1) Liver metastases, n (%) | 25 (13.3) | 33 (17.6) EGFR mutation subtype*, n (%) Exon 19 deletion | 106 (56.4) | 118 (62.8) Exon 21 L858R | 84 (44.7) | 71 (37.8) T790M mutation status, n (%) Positive | 29 (15.4) | 36 (19.1) Negative | 48 (25.5) | 40 (21.3) Unknown | 111 (59.0) | 112 (59.6) Prior 3rd generation EGFR-TKI, n (%) 1st line | 118 (62.8) | 117 (62.2) 2nd line | 60 (31.9) | 60 (31.9) *Including overlapping mutations in some patients. Other EGFR mutations co-occurring with exon 21 L858R or exon 19 deletion were included in the subcategory "Other" (sac-TMT: 4.3%; chemotherapy: 3.7%). Professor Wenfeng Fang [Slide 4] Overall Survival: Subgroup Analyses A consistent OS benefit of sac-TMT over chemotherapy was observed across all subgroups. Subgroup | Events / Number Sac-TMT | Chemotherapy | HR (95% CI) All subjects | 114/188 | 137/188 | 0.69 (0.54 - 0.89) Sex: Male | 44/66 | 64/83 | 0.75 (0.51 - 1.11) Sex: Female | 70/122 | 73/105 | 0.66 (0.48 - 0.92) Age: <65 | 75/130 | 97/137 | 0.66 (0.49 - 0.89) Age: >=65 | 39/58 | 40/51 | 0.70 (0.45 - 1.10) Smoking history: Yes | 29/43 | 42/53 | 0.73 (0.45 - 1.17) Smoking history: No | 85/145 | 95/135 | 0.68 (0.50 - 0.91) ECOG: 0 | 19/35 | 27/43 | 0.73 (0.41 - 1.32) ECOG: 1 | 95/153 | 110/145 | 0.66 (0.50 - 0.87) Prior EGFR-TKI therapy: 3rd-generation EGFR-TKI in 1st-line | 69/118 | 88/117 | 0.64 (0.46 - 0.87) Prior EGFR-TKI therapy: 3rd-generation EGFR-TKI in 2nd-line | 40/60 | 41/60 | 0.81 (0.53 - 1.26) Brain metastases: Yes | 23/33 | 28/36 | 0.76 (0.43 - 1.32) Brain metastases: No | 91/155 | 109/152 | 0.66 (0.50 - 0.88) Liver metastases: Yes | 15/25 | 28/33 | 0.52 (0.28 - 0.99) Liver metastases: No | 99/163 | 109/155 | 0.72 (0.55 - 0.95) EGFR mutation type: L858R | 58/84 | 53/71 | 0.83 (0.57 - 1.20) EGFR mutation type: 19-Del | 56/106 | 85/118 | 0.55 (0.39 - 0.78) T790M mutation status: Negative | 30/48 | 29/40 | 0.76 (0.45 - 1.26) T790M mutation status: Positive | 20/29 | 24/36 | 0.82 (0.45 - 1.48) T790M mutation status: Unknown | 64/111 | 84/112 | 0.62 (0.45 - 0.86) Metastatic sites: <3 | 30/60 | 38/62 | 0.78 (0.48 - 1.26) Metastatic sites: >=3 | 84/128 | 99/126 | 0.62 (0.46 - 0.82) <---Sac-TMT Better Chemotherapy Better---> Data cutoff: 06 August 2026 Professor Wenfeng Fang

[Slide 2] Extended follow-up confirms sustained OS benefit of tarlatamab Overall survival (%) Months Tarlatamab Chemotherapy Number of patients at risk: Data cutoff date: 18 Feb 2026. Tarlatamab showed sustained survival benefit vs chemotherapy, with reduced mortality risk, persistent separation of the Kaplan–Meier curves, and higher landmark OS rate CI, confidence interval; FU, follow-up; HR, hazard ratio; OS, overall survival. Charles M. Rudin, MD, PhD | Tarlatamab in Previously Treated SCLC: Extended Follow-up From the Phase 3 DeLLphi-304 Trial 5 Survival table Endpoint: Overall survival (extended follow-up) Tarlatamab | 254 | 13.6 months Chemotherapy | 255 | 8.4 months HR: 0.62 HR CI: 0.51–0.76 18-month OS: tarlatamab 42% vs chemotherapy 24% 24-month OS: tarlatamab 30% vs chemotherapy 15% [Slide 3] Tarlatamab demonstrated sustained PFS benefit during extended follow-up Progression-free survival (%) Months Tarlatamab Chemotherapy Number of patients at risk: Data cutoff date: 18 Feb 2026. The PFS benefit observed with tarlatamab in the primary analysis was sustained during extended follow-up, with higher landmark PFS rates than chemotherapy CI, confidence interval; FU, follow-up; HR, hazard ratio; PFS, progression-free survival. Charles M. Rudin, MD, PhD | Tarlatamab in Previously Treated SCLC: Extended Follow-up From the Phase 3 DeLLphi-304 Trial 6 Survival table Endpoint: Progression-free survival (extended follow-up) Tarlatamab | 254 | 4.2 months Chemotherapy | 255 | 3.2 months HR: 0.69 HR CI: 0.57–0.84 12-month PFS: tarlatamab 21% vs chemotherapy 6% 18-month PFS: tarlatamab 16% vs chemotherapy 4% [Slide 4] DeLLphi-304 extended follow-up conclusions The 23-month long-term follow-up of the phase 3 DeLLphi-304 trial demonstrated durable clinical benefit with tarlatamab vs chemotherapy including: • Sustained survival benefit with 38% lower risk of death and 31% lower risk of disease progression or death • Landmark 2-year overall survival at 30% for tarlatamab vs 15% for chemotherapy • Durable responses with sustained disease control of ≥ 52 weeks at 24% for tarlatamab vs 4% for chemotherapy • Favorable safety profile with long-term tolerability and no new safety signals Extended follow-up confirms durable clinical benefit and favorable long-term tolerability, reinforcing tarlatamab as the standard of care after first-line platinum-based chemotherapy Charles M. Rudin, MD, PhD | Tarlatamab in Previously Treated SCLC: Extended Follow-up From the Phase 3 DeLLphi-304 Trial 10

[Slide 1] CHAIRS: EILEEN PARKES, TIMOTHY ANTHONY YAP INTRODUCTION | LANDSCAPE | CHALLENGES | TAKE HOME MESSAGES DLL3: parallel to your BiTE story in SCLC CD3 T Cell TCE Bispecific Antibody Cancer Cell Killing Cancer Cell Target Antigen DSB SSB Chelator Targeting molecule Linker Target protein α emitter Cancer cell Radiopharmaceutical Abscopal effect Primary targeted tumour Secondary untreated tumour The immune system activation Bispecific T-cell engager Tarlatamab First-in-class DLL3-directed BiTE. FDA approved in extensive-stage SCLC after platinum progression (DeLLphi-301). Phase III DeLLphi-304 met OS endpoint. Radioligand ABD-147 (225Ac) · MP0712 (212Pb) ■ ABD-147 (Abdera): antibody + 225Ac · phase I active ■ MP0712 (Molecular Partners/Orano): DARPin + 212Pb · phase I first-in-human dosing (ASCO/SNMMI 2026) ■ Anti-DLL3 IgG4 antibodies for 89Zr imaging + 177Lu therapy (PNAS 2026) Maria Rodríguez-Ruiz Updates on radioligands [Slide 2] CHAIRS: EILEEN PARKES, TIMOTHY ANTHONY YAP INTRODUCTION | LANDSCAPE | CHALLENGES | TAKE HOME MESSAGES FAP: the stroma as target LuMIERE [177Lu]Lu-FAP-2286 First prospective trial of a FAP radioligand Design Phase 1/2 dose escalation, 3.7–7.4 GBq Patients 11 pretreated; colorectal, pancreatic, others Result 1 partial response, 1 stable disease (interim) Tumor cell Cancer-associated fibroblast Inflammatory or immune cell Extracellular matrix Cytokine Peptidase Exosomes or microvesicles FRONTIER [177Lu]Lu-PNT6555 Phase 1 with whole-body dosimetry Design Dose escalation, 4–12 GBq, up to 6 cycles Patients 10 treated: pancreatic, colorectal, sarcoma... Result Mean tumor dose 0.16 Gy/GBq (range 0.02–0.45) FAP is promising, but PET uptake is not enough Maria Rodríguez-Ruiz Updates on radioligands [Slide 3] CHAIRS: EILEEN PARKES, TIMOTHY ANTHONY YAP INTRODUCTION | LANDSCAPE | CHALLENGES | TAKE HOME MESSAGES 4 “Authorization for access to new therapeutic radiopharmaceuticals.” The issue of authorization for the use of 225Ac by the... CSN CONSEJO DE SEGURIDAD NUCLEAR • For therapeutic radiopharmaceuticals, authorization is not granted by radionuclide, but by radiopharmaceutical. • Biodistribution data must be provided (Circular 2/2024; CSN/C/DPR/TGE/24/28) SOLUTION FOR CLINICAL TRIALS: • If pharmaceutical companies submit biodistribution data to the CSN, each center may follow an approved protocol.”(art. 113 RD1217/2024) Maria Rodríguez-Ruiz Updates on radioligands [Slide 4] CHAIRS: EILEEN PARKES, TIMOTHY ANTHONY YAP INTRODUCTION | LANDSCAPE | CHALLENGES | TAKE HOME MESSAGES 1 Radioligand ≈ ADC in architecture Change the payload — crossfire and theranostic pair appear. The ceiling moves. 2 Beyond PSMA and SSTR, the field is broader than it looks FAP, DLL3, hCA IX, hematologic, pediatric — at least ten target families with active trials. 3 Manufacturing and regulation are the ceiling Immuno combinations work by design, not by rationale. The girentuximab CRL is a structural warning Maria Rodríguez-Ruiz Updates on radioligands

[Slide 2] Progression-free survival was significantly longer with tarlatamab vs chemotherapy Legend: Tarlatamab / Chemotherapy | Tarlatamab (n = 254) | Chemotherapy (n = 255) Median PFS, months | 4.2 | 3.7 HR (Tarlatamab/Chemotherapy) (95% CI) | 0.71 (0.59, 0.86) RMST p-value (2-sided) | p = 0.002* Curve annotations: Tarlatamab 31% (6 mo), 20% (12 mo); Chemotherapy 23% (6 mo), 4% (12 mo) Y axis: Progression-free Survival, %; X axis: Time from randomization (months) 0-18 Number of patients at risk (0, 3, 6, 9, 12, 15, 18 months): Tarlatamab: 254 147 78 37 18 2 0 Chemotherapy: 255 137 56 15 3 0 Median follow-up time: 11.0 months for the tarlatamab and the chemotherapy group. *The restricted mean PFS time in the tarlatamab and the chemotherapy group was 5.3 months and 4.3 months at 12 months respectively, resulting in statistically significant improvement of the tarlatamab group over the chemotherapy group. HR: hazard ratio; PFS, progression-free survival. [2025 ASCO Annual Meeting logo] #ASCO25 PRESENTED BY: Charles M. Rudin, MD, PhD Survival table Endpoint: PFS Tarlatamab | 254 | 4.2 months Chemotherapy | 255 | 3.7 months HR: 0.71 HR CI: 0.59, 0.86 P: 0.002 (RMST, 2-sided) Landmark rates: 6 months: Tarlatamab 31% | Chemotherapy 23% 12 months: Tarlatamab 20% | Chemotherapy 4% [Slide 3] Tarlatamab was associated with more frequent and more durable responses | Tarlatamab (n = 254) | Chemotherapy (n = 255) Best overall response*†, n (%) Complete response | 3 (1) | 0 (0) Partial response | 86 (34) | 52 (20) Stable disease | 84 (33) | 112 (44) Progressive disease | 56 (22) | 50 (20) Not evaluable/no post-baseline scan | 25 (10) | 41 (16) Objective response rate‡, % (95% CI) | 35 (29–41) | 20 (16–26) Median duration of response, months | 6.9 | 5.5 Median time to objective response, months | 1.5 | 1.4 Ongoing response at data cutoff, n§ (%) | 42 (47) | 8 (15) Duration of response Legend: Tarlatamab / Chemotherapy | Tarlatamab (n = 254) | Chemotherapy (n = 255) Median DOR, months | 6.9 | 5.5 Curve annotations: Tarlatamab 56% (6 mo), 41% (12 mo); Chemotherapy 29% (6 mo), 13% (12 mo) Y axis: Responders Without Progression or Death (%); X axis: Time from initial response (months) 0-18 Number of patients at risk (0, 3, 6, 9, 12, 15, 18 months): Tarlatamab: 89 70 41 22 12 2 0 Chemotherapy: 52 40 14 2 1 0 *Assessment of disease response was based on RECIST 1.1 guidelines. Confirmation of complete response and partial response was required no fewer than 4 weeks after initial documentation of complete response or partial response. †Investigator-assessed response in the intention-to-treat analysis set; ‡Odds ratios and p value not shown as the difference in ORR between the 2 arms was not formally tested. §Percentage of total number of responders. DOR, duration of response; IRECIST, Response Evaluation Criteria in Solid Tumors. [2025 ASCO Annual Meeting logo] #ASCO25 PRESENTED BY: Charles M. Rudin, MD, PhD Survival table Endpoint: Duration of response (responders) Tarlatamab | 89 | 6.9 months Chemotherapy | 52 | 5.5 months Landmark rates: 6 months: Tarlatamab 56% | Chemotherapy 29% 12 months: Tarlatamab 41% | Chemotherapy 13% [Slide 4] DeLLphi-304 met its primary endpoint with tarlatamab demonstrating superior overall survival over chemotherapy Legend: Tarlatamab / Chemotherapy | Tarlatamab (n = 254) | Chemotherapy (n = 255) Median OS, months | 13.6 | 8.3 HR (Tarlatamab/Chemotherapy) (95% CI) | 0.60 (0.47, 0.77) p-value (2-sided) | p < 0.001 Curve annotations: Tarlatamab 76% (6 mo), 53% (12 mo); Chemotherapy 62% (6 mo), 37% (12 mo) Y axis: Overall Survival (%); X axis: Time from randomization (months) 0-21 Number of patients at risk (0, 3, 6, 9, 12, 15, 18, 21 months): Tarlatamab: 254 220 192 131 60 17 0 Chemotherapy: 255 210 156 97 42 9 2 0 Median follow-up time: 11.2 months for the tarlatamab group and 11.7 months for the chemotherapy group. p-value was calculated using a stratified log-rank test. HR, hazard ratio; OS, overall survival. [2025 ASCO Annual Meeting logo] #ASCO25 PRESENTED BY: Charles M. Rudin, MD, PhD Survival table Endpoint: Overall survival Tarlatamab | 254 | 13.6 months Chemotherapy | 255 | 8.3 months HR: 0.60 HR CI: 0.47, 0.77 P: <0.001 (2-sided, stratified log-rank) Landmark rates: 6 months: Tarlatamab 76% | Chemotherapy 62% 12 months: Tarlatamab 53% | Chemotherapy 37%

[Slide 1] Randomized, controlled, phase 3 DeLLphi-304 study Key inclusion criteria: • Age ≥ 18 years • SCLC that progressed during or after 1L platinum-based therapy • ECOG PS 0–1 • Treated and untreated asymptomatic brain metastases N = 509 R (1:1) Tarlatamab (n = 254) Chemotherapy (n = 255)* Randomization stratification: • Prior anti-PD-(L)1 therapy (yes vs no) • Chemotherapy-free interval (< 90 days vs ≥ 90 day to < 180 days vs ≥ 180 days) • Presence of brain metastasis (yes vs no) • Chemotherapy (topotecan/amrubicin vs lurbinectedin) • Primary endpoint: OS • Secondary endpoints: PFS, ORR, DC, DOR, PROs, and safety *Amrubicin (n = 23; Japan), Lurbinectedin (n = 47; United States, Canada, Australia, Singapore, and Republic of Korea), Topotecan (n = 185; all countries except Japan) in 21-day cycle. 1L, first-line; DC, disease control; DOR, duration of response; ECOG-PS, Eastern Cooperative Oncology Group performance status; PD-(L)1, programmed death-(ligand) 1; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; PRO, patient-reported outcomes; R, randomization; RECIST, Response Evaluation Criteria in Solid Tumors; SCLC, small cell lung cancer. Mountzios G, et al. N Engl J Med. 2025;393:349-361. [Slide 2] Previously reported primary analysis: tarlatamab improved OS vs chemotherapy • Prior to DeLLphi-304 results, median real-world overall survival in the 2L setting was poor (approximately 5-6 months)1,2 DeLLphi-304 Overall Survival (Primary Analysis)3 Median OS: 13.6 vs 8.3 months HR 0.60; P < 0.001 Tarlatamab / Chemotherapy Overall survival (%) / Time from randomization (months) mFU for tarlatamab: 11.2 months (95% CI, 10.4–12.1) mFU for chemotherapy: 11.7 months (95% CI, 10.6–12.3) We now report results from the extended follow-up of DeLLphi-304 (median follow-up, 23 months) assessing the durability of clinical benefit and long-term safety of tarlatamab vs chemotherapy 2L, second-line; CI, confidence interval; HR, hazard ratio; mFU, median follow-up; OS, overall survival. 1. Shaw J, et al. Oncologist. 2024;29:1079-1089; 2. Aggarwal H, et al. Front Oncol. 2026;16:1846362; 3. Mountzios G, et al. N Engl J Med. 2025;393:349-361. Survival table Endpoint: OS (primary analysis, previously reported) Tarlatamab | 13.6 months Chemotherapy | 8.3 months HR: 0.60 P: < 0.001

[Slide 1] INTRODUCTION LANDSCAPE CHALLENGES TAKE HOME MESSAGES Beyond prostate and NETs: the target-tumor matrix Target | Malignancy | Lead radioligand · trial | Development phase FAP ★ | Multiple solid: sarcoma, breast, thyroid, pancreatic | 177Lu-FAP-2286 (LuMIERE) | Phase I/II DLL3 ★ | Small cell lung cancer · NEPC | 225Ac-ABD-147 · 212Pb-MP0712 | Phase I hCA IX | Clear cell renal cell carcinoma | 177Lu-girentuximab (STARLITE-2) | Phase II CXCR4 | Refractory lymphoma · AML | 177Lu/90Y-pentixather (PENTILULA) | Phase I/II HER2 | Breast · gastric | 131I-GMIB-anti-HER2-VHH1 | Phase I GRPR | Breast · prostate (PSMA-low) | 177Lu-NeoB · 177Lu-RM2 | Phase I/II CD20/CD22 | Non-Hodgkin lymphoma | 227Th-anti-CD22 · Bexxar legacy | Phase I MC1R | Metastatic melanoma | 212Pb/203Pb-VMT01 | Phase I/II NET | Neuroblastoma · pheochromocytoma · paraganglioma | 131I-MIBG (established) | Approved LAT1 | Glioblastoma | 131I-iodofalan | Phase III selected targets based on data published at AACR, ASCO, ASTRO, ESMO, ESTRO meetings between 2024 and 2026. [Slide 2] INTRODUCTION LANDSCAPE CHALLENGES TAKE HOME MESSAGES DLL3: parallel to your BiTE story in SCLC CD3 T Cell TCE Bispecific Antibody Cancer Cell Killing Cancer Cell Target Antigen Bispecific T-cell engager Tarlatamab First-in-class DLL3-directed BiTE. FDA approved in extensive-stage SCLC after platinum progression (DeLLphi-301). Phase III DeLLphi-304 met OS endpoint. Radioligand ABD-147 (225Ac) · MP0712 (212Pb) ABD-147 (Abdera): antibody + 225Ac · phase I active MP0712 (Molecular Partners/Orano): DARPin + 212Pb · phase I first-in-human dosing (ASCO/SNMMI 2026) Anti-DLL3 IgG4 antibodies for 89Zr imaging + 177Lu therapy (PNAS 2026) [Slide 3] INTRODUCTION LANDSCAPE CHALLENGES TAKE HOME MESSAGES Availability of Therapeutic Radiopharmaceuticals and Logistical Challenges Very Limited Isotope Production ACTINIO-225 GeneraTor Th-229 (1,7 Ci/año) GOAL PanTera 2028: (>100 Ci/año): irradiar con protones Ra-226 PLOMO-212 • Limited Number of GeneratorsTh-228 RADIOFARMACIA: SÍNTESIS Y QA Control of production impurities (Ac-227) • Optimization of chelators/linkers • Dedicated laboratories for handling alpha emissions • Complex and time-consuming QA HOSPITAL Limited stability: rapid transport required • Logistics to hospitals → final preparation, dispensing and administration → imaging, dosimetry, and response monitoring

[Slide 1] Methods [COCOON logo] COCOON 1L ami + laz in EGFR+ NSCLC SC cohorts (Non-randomized cohorts) Key eligibility criteria - Locally advanced or metastatic NSCLC - Treatment-naive for advanced disease - Documented EGFR Ex19del or L858R - ECOG PS score of 0 or 1 Simplified COCOON prophylaxis - Oral tetracycline (24 weeks) - Ceramide-based moisturizer(a) - Oral zinc gluconate(b) Standard dosing schedule (n=70) SC amivantamab (Weekly x 4 on C1 then Q4W) Oral lazertinib (Daily in 28-day cycles) Cycle 1 | Cycle 2 | Cycle 3+ Lazertinib lead-in dosing schedule (n=34) SC amivantamab (Week 1, Week 3 on C2 then Q4W) Oral lazertinib (Daily in 28-day cycles) Cycle 1 | Cycle 2 | Cycle 3+ - Primary endpoint was incidence of grade >=2 dermatologic AEs in the first 12 weeks after initiation of anticancer treatment; no prespecified formal comparisons between cohorts - Additional site-directed interventions were evaluated for concomitant use when clinically indicated - Preferred terms(c) for dermatologic AEs included rash, dermatitis acneiform, pruritus, skin fissures, acne, folliculitis, erythema, eczema, rash maculopapular, skin exfoliation, skin lesion, skin irritation, dermatitis, rash erythematous, rash macular, rash papular, rash pruritic, rash pustular, dermatitis contact, dermatitis exfoliative generalized, drug eruption, dyshidrotic eczema, eczema asteatotic, and paronychia COCOON (ClinicalTrials.gov Identifier: NCT06120140). (a)La Roche-Posay Lipikar AP+M moisturizer was used in COCOON. (b)70 mg once daily for up to 52 weeks. (c)Severity per NCI CTCAE v5.0. C, Cycle; D, Day; Q, every; SC, subcutaneous; W, week. Scan the QR code. Copies of this slide deck obtained through this QR code are for personal use only and may not be reproduced without permission of the authors. [Slide 2] Primary Endpoint: Grade >=2 Dermatologic AEs by Week 12 [COCOON logo] COCOON 1L ami + laz in EGFR+ NSCLC Main COCOON study(1) (IV amivantamab + lazertinib) OR, 0.24 (95% CI, 0.13-0.45) P<0.0001 SoC DM (n=100): 75% (Grade 2: 68%; Grade >2: 7%) COCOON DM (n=99): 42% (Grade 2: 36%; Grade >2: 6%) SC amivantamab + simplified dermatologic prophylaxis OR, 0.31 (95% CI, 0.11-0.87) Nominal P=0.0264 Standard schedule (n=70): 39% (Grade 2: 36%; Grade >2: 3%) Laz lead-in (n=34): 18% Week 16(b) (exploratory endpoint) Laz lead-in (n=34): 24% Legend: Grade 2 / Grade >2 Y-axis: Participants with dermatologic AEs (%) - Dermatologic prophylaxis has previously been shown to reduce dermatologic AEs by Week 12(1) - Simplified prophylaxis is associated with a similar reduction in dermatologic AEs as the full COCOON DM regimen - Lazertinib lead-in prior to starting amivantamab + lazertinib at Cycle 2 further reduced early dermatologic AEs(a) - An exploratory analysis of Week 16 data showed consistent findings(b) (a)In a preliminary analysis, investigator-assessed overall response rates were consistent between standard and lazertinib lead-in cohorts (71% vs 70%). (b)At Week 16, participants had up to 12 weeks of amivantamab exposure and 16 weeks of lazertinib exposure. DM, dermatologic management; IV, intravenous; laz, lazertinib; OR, odds ratio; SC, subcutaneous; SoC, standard of care. 1. Cho BC, et al. J Thorac Oncol. 2025;20(10):1517-1530. [Slide 3] Location of Grade >=2 Dermatologic AEs by Cohort [COCOON logo] COCOON 1L ami + laz in EGFR+ NSCLC Legend: Face | Body | Scalp | Paronychia Main COCOON study(1) (IV amivantamab + lazertinib) SoC DM (n=100): Face 48% | Body 44% | Scalp 26% | Paronychia 23% COCOON DM (n=99): Face 15% | Body 20% | Scalp 10% | Paronychia 21% SC amivantamab + simplified dermatologic prophylaxis Standard schedule (n=70): Face 14% | Body 24% | Scalp 6% | Paronychia 20% Laz lead-in (n=34): Face 9% | Body 9% | Scalp 3% | Paronychia 12% Similar observed pattern | Consistently lower Y-axis: Participants with dermatologic AEs (%) - In the first 12 weeks, the incidence of grade >=2 dermatologic AEs across body locations was similar for COCOON DM and simplified dermatologic prophylaxis - Rates were consistently lower for the lazertinib lead-in cohort DM, dermatologic management; laz, lazertinib; SC, subcutaneous; SoC, standard of care. 1. Cho BC, et al. J Thorac Oncol. 2025;20(10):1517-1530. [Slide 4] Dose Modifications [COCOON logo] COCOON 1L ami + laz in EGFR+ NSCLC Amivantamab dose modifications in the first 12 weeks, % | Main COCOON study (n=199)(1) (IV amivantamab + lazertinib): SoC DM (n=100) | COCOON DM (n=99) | SC amivantamab + simplified dermatologic prophylaxis (n=104): Standard schedule (n=70) | Laz lead-in (n=34) Dose modifications due to dermatologic AEs Dose interruptions | 21 | 8 | 1 | 0 Dose reductions | 9 | 2 | 3 | 3 Discontinuations | 2 | 2 | 1 | 0 Dose modifications due to AEs Dose interruptions | 45 | 42 | 21 | 26 Dose reductions | 16 | 12 | 3 | 6 Discontinuations | 12 | 9 | 6 | 12 - Both SC cohorts showed lower rates of dose modifications due to dermatologic AEs and any AEs in the first 12 weeks DM, dermatologic management; laz, lazertinib; SC, subcutaneous; SoC, standard of care. 1. Cho BC, et al. J Thorac Oncol. 2025;20(10):1517-1530.

Simplified COCOON Regimen for Derm Prophylaxis and Management1 COCOON 1L ami + laz in EGFR+ NSCLC Presented at ESMO 2026 PROPHYLAXIS Start on Day 1 of treatment to reduce the risk of developing dermatologic AEs 1 START PROPHYLAXIS ON DAY 1 OF AMIVANTAMAB + LAZERTINIB ORAL TETRACYCLINE Doxycycline or minocycline 100 mg by mouth twice daily Weeks 1–12 CERAMIDE-BASED MOISTURIZER Apply to face and body At least once daily Weeks 1+ CONCOMITANT / EARLY INTERVENTION Begin site-directed treatment at first onset of grade 1 dermatologic AEs, as clinically indicated 2 IF A DERMATOLOGIC AE EMERGES FACE / BODY Corticosteroids2 Tailor to location and severity Topical ruxolitinib cream or tacrolimusa SCALPb Clobetasol 0.05% shampoo2 3 times weekly Propranolol 10 mg By mouth once daily until resolution PARONYCHIAb Timolol 0.5% solution2 Apply twice daily Chlorhexidine 4% solution2 Apply once daily OTHER CONSIDERATIONS Escalate care based on severity/persistence, consider early dermatology referral, and see amivantamab label for dose modifications Oral zinc gluconate and extended tetracycline treatment were also evaluated in COCOON, but clinical benefit has not been established. aInvestigational treatment under evaluation in COCOON. bCombination regimen under investigation in COCOON. 1. Figure adapted from Cho BC, et al. J Thorac Oncol. 2025;20(10):1517–1530. 2. Nikolaou V, et al. Lung Cancer. 2026;213:108950. 3. RYBREVANT® (amivantamab-vmjw) injection for intravenous use [package insert]. Janssen Biotech, Inc.; 2025.

[Corner logo] COCOON 1L ami + laz in EGFR+ NSCLC Conclusions • Simplified dermatologic prophylaxis along with 1L SC amivantamab + lazertinib achieved a similar reduction in grade ≥2 dermatologic AEs as the full COCOON regimen, supporting a streamlined approach that may reduce treatment burden while maintaining patient benefit • Lazertinib lead-in for 1 cycle prior to 1L SC amivantamab + lazertinib further reduced early dermatologic AEs, suggesting that flexible treatment initiation may also lead to an improved safety profile • Outcomes in the COCOON study will continue to be evaluated as data mature Overall, these findings support a simplified and flexible regimen to help patients remain on treatment, potentially maximizing the long-term survival benefit of 1L amivantamab + lazertinib 1L, first-line; SC, subcutaneous.

[Slide 1] 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic CHAIRS: PASI A. JÄNNE, XIUNING LE, EGBERT SMIT Simplified COCOON Regimen for Derm Prophylaxis and Management(1) [COCOON logo] COCOON 1L ami + laz in EGFR+ NSCLC Presented at ESMO 2026 PROPHYLAXIS: Start on Day 1 of treatment to reduce the risk of developing dermatologic AEs 1 START PROPHYLAXIS ON DAY 1 OF AMIVANTAMAB + LAZERTINIB ORAL TETRACYCLINE: Doxycycline or minocycline, 100 mg by mouth twice daily - Weeks 1-12 CERAMIDE-BASED MOISTURIZER: Apply to face and body, At least once daily - Weeks 1+ CONCOMITANT / EARLY INTERVENTION: Begin site-directed treatment at first onset of grade 1 dermatologic AEs, as clinically indicated 2 IF A DERMATOLOGIC AE EMERGES FACE / BODY: Corticosteroids(2), Tailor to location and severity; Topical ruxolitinib cream or tacrolimus(a) SCALP(b): Clobetasol 0.05% shampoo(2), 3 times weekly; Propranolol 10 mg, By mouth once daily until resolution PARONYCHIA(b): Timolol 0.5% solution(2), Apply twice daily; Chlorhexidine 4% solution(2), Apply once daily OTHER CONSIDERATIONS: Escalate care based on severity/persistence, consider early dermatology referral, and see amivantamab label for dose modifications Oral zinc gluconate and extended tetracycline treatment were also evaluated in COCOON, but clinical benefit has not been established. (a)Investigational treatment under evaluation in COCOON. (b)Combination regimen under investigation in COCOON. 1. Figure adapted from Cho BC, et al. J Thorac Oncol. 2025;20(10):1517-1530. 2. Nikolaou V, et al. Lung Cancer. 2026;213:108950. 3. RYBREVANT® (amivantamab-vmjw) injection for intravenous use [package insert]. Janssen Biotech, Inc.; 2025 [Speaker video panel] Byoung Chul Cho - Improving prevention of moderate to severe dermatologic adverse events in EGFR-mutant advanced NSCLC treated with first-line (1L) subcutaneous (SC) amivantamab + lazertinib: First results from a new COCOON cohort ALMERIA AUDITORIUM - HALL 9 [Slide 2] 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic CHAIRS: PASI A. JÄNNE, XIUNING LE, EGBERT SMIT Conclusions [COCOON logo] COCOON 1L ami + laz in EGFR+ NSCLC • Simplified dermatologic prophylaxis along with 1L SC amivantamab + lazertinib achieved a similar reduction in grade ≥2 dermatologic AEs as the full COCOON regimen, supporting a streamlined approach that may reduce treatment burden while maintaining patient benefit • Lazertinib lead-in for 1 cycle prior to 1L SC amivantamab + lazertinib further reduced early dermatologic AEs, suggesting that flexible treatment initiation may also lead to an improved safety profile • Outcomes in the COCOON study will continue to be evaluated as data mature Overall, these findings support a simplified and flexible regimen to help patients remain on treatment, potentially maximizing the long-term survival benefit of 1L amivantamab + lazertinib 1L, first-line; SC, subcutaneous [Speaker video panel] Byoung Chul Cho - Improving prevention of moderate to severe dermatologic adverse events in EGFR-mutant advanced NSCLC treated with first-line (1L) subcutaneous (SC) amivantamab + lazertinib: First results from a new COCOON cohort ALMERIA AUDITORIUM - HALL 9 [Slide 3] 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic CHAIRS: PASI A. JÄNNE, XIUNING LE, EGBERT SMIT Primary Endpoint: Grade ≥2 Dermatologic AEs by Week 12 [COCOON logo] COCOON 1L ami + laz in EGFR+ NSCLC Y-axis: Participants with dermatologic AEs (%) Main COCOON study(1) (IV amivantamab + lazertinib) OR, 0.24 (95% CI, 0.13-0.45) P<0.0001 SoC DM (n=100): 75% total (68% grade 2; 7% grade >2) COCOON DM (n=99): 42% total (36% grade 2; 6% grade >2) SC amivantamab + simplified dermatologic prophylaxis OR, 0.31 (95% CI, 0.11-0.87) Nominal P=0.0264 Standard schedule (n=70): 39% total (36% grade 2; 3% grade >2) Laz lead-in (n=34): 18% Legend: Grade 2; Grade >2 • Dermatologic prophylaxis has previously been shown to reduce dermatologic AEs by Week 12(1) • Simplified prophylaxis is associated with a similar reduction in dermatologic AEs as the full COCOON DM regimen • Lazertinib lead-in prior to starting amivantamab + lazertinib at Cycle 2 further reduced early dermatologic AEs(a) (a)In a preliminary analysis, investigator-assessed overall response rates were consistent between standard and lazertinib lead-in cohorts (71% vs 70%) DM, dermatologic management; IV, intravenous; laz, lazertinib; OR, odds ratio; SC, subcutaneous; SoC, standard of care. 1. Cho BC, et al. J Thorac Oncol. 2025;20(10):1517-1530. [Speaker video panel] Byoung Chul Cho - Improving prevention of moderate to severe dermatologic adverse events in EGFR-mutant advanced NSCLC treated with first-line (1L) subcutaneous (SC) amivantamab + lazertinib: First results from a new COCOON cohort ALMERIA AUDITORIUM - HALL 9 [Slide 4] 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic CHAIRS: PASI A. JÄNNE, XIUNING LE, EGBERT SMIT Demographic and Baseline Disease Characteristics [COCOON logo] COCOON 1L ami + laz in EGFR+ NSCLC Characteristic, n (%) | Main COCOON study (n=199)(1) (IV amivantamab + lazertinib): SoC DM (n=100) | COCOON DM (n=99) | SC amivantamab + simplified dermatologic prophylaxis (n=104): Standard schedule (n=70) Median (range) follow-up, 8.1 (1.0-9.7) months | Laz lead-in (n=34) Median (range) follow-up, 5.5 (1.6-6.0) months Median age (range), y | 62.5 (28-83) | 63 (34-80) | 63 (35-87) | 66 (36-81) Female | 57 (57) | 61 (62) | 46 (66) | 21 (62) Race(a) Asian | 65 (65) | 66 (67) | 47 (67) | 28 (82) White | 32 (32) | 32 (32) | 22 (31) | 5 (15) Black or African American | 1 (1) | 0 | 0 | 1 (3) Multiple | 2 (2) | 0 | 1 (1) | 0 ECOG PS score of 1 | 55 (55) | 59 (60) | 43 (61) | 27 (79) History of brain metastases | 43 (43) | 32 (32) | 21 (30) | 16 (47) History of smoking | 33 (33) | 34 (34) | 13 (19) | 10 (29) Histologic type(b) Adenocarcinoma | 97 (97) | 94 (95) | 70 (100) | 33 (97) Squamous cell carcinoma | 2 (2) | 2 (2) | 0 | 1 (3) Large cell carcinoma | 0 | 1 (1) | 0 | 0 Mutation type Ex19del | 54 (54) | 50 (51) | 41 (59) | 14 (41) L858R | 46 (46) | 49 (49) | 29 (41) | 20 (59) Clinical cutoff: September 1, 2026. (a)One participant in the COCOON DM cohort self-reported their race as American Indian or Alaska Native. (b)For 2 participants in the COCOON DM cohort and 1 participant in the SoC DM cohort, histologic type was reported as "Other". DM, dermatologic management; IV, intravenous; laz, lazertinib; SC, subcutaneous; SoC, standard of care 1. Cho BC, et al. J Thorac Oncol. 2025;20(10):1517-1530 [Speaker video panel] Byoung Chul Cho - Improving prevention of moderate to severe dermatologic adverse events in EGFR-mutant advanced NSCLC treated with first-line (1L) subcutaneous (SC) amivantamab + lazertinib: First results from a new COCOON cohort ALMERIA AUDITORIUM - HALL 9

08:30 - 10:00 Rapid oral 1: NSCLC, metastatic CHAIRS: PASI A. JÄNNE, XIUNING LE, EGBERT SMIT Background [COCOON logo] COCOON 1L ami + laz in EGFR+ NSCLC • In MARIPOSA, IV amivantamab + lazertinib significantly improved OS versus osimertinib in 1L common EGFR-mutant advanced NSCLC (HR, 0.75; P=0.005), but was associated with dermatologic AEs, including rash and paronychia, commonly occurring early in treatment and leading to discontinuations(1) • In the phase 2 COCOON study (NCT06120140), enhanced dermatologic management(a) with 1L IV amivantamab + lazertinib significantly reduced grade ≥2 dermatologic AEs during the first 12 weeks versus standard dermatologic care in participants with EGFR-mutant advanced NSCLC(2) • We further evaluated a simplified dermatologic prophylaxis regimen in combination with SC amivantamab, which was previously shown to improve overall treatment experience and help patients stay on treatment longer(3-5) Here we report the first results from 2 new COCOON cohorts using a simplified dermatologic prophylaxis regimen in participants receiving 1L SC amivantamab(b) + lazertinib (a)Oral tetracycline for 12 weeks, ceramide-based moisturizer, topical clindamycin, chlorhexidine (b)Co-formulated with recombinant human hyaluronidase PH20 1L, first-line; HR, hazard ratio; IV, intravenous; SC, subcutaneous 1. Yang JCH, et al. N Engl J Med. 2025;393(17):1681-1693. 2. Cho BC, et al. J Thorac Oncol. 2025;20(10):1517-1530. 3. Leighl NB, et al. J Clin Oncol. 2024;42:3593-3605. 4. Alexander MA, et al. Eur J Cancer. 2025;227:115624. 5. Halmos B. Presented at World Conference on Lung Cancer (WCLC) 2026, September 12-15, 2026; Seoul, Republic of Korea. Scan the QR code. Copies of this slide deck obtained through this QR code are for personal use only and may not be reproduced without permission of the authors.

[Slide 1] ASTRIDE study design Between 2022 and 2025, a total of 201 patients were enrolled at 70 study sites across the United States. Key eligibility criteria: • Age ≥18 years; • Performance status 0/1; • Histologically or cytologically confirmed ES-SCLC; • No prior systemic therapy for ES-SCLC (Patients who have received one cycle of chemotherapy for newly diagnosed ES-SCLC are eligible); • Measurable disease as assessed by BICR per RECIST v1.1. N = 201 R 1:1 IV, Q3W, up to 4 cycles Serplulimab (300 mg, D1) + carboplatin (AUC 5, D1)^a + etoposide (100 mg/m², D1–3) Atezolizumab (1200 mg, D1) + carboplatin (AUC 5, D1)^a + etoposide (100 mg/m², D1–3) IV, Q3W^b Serplulimab (300 mg, D1) Atezolizumab (1200 mg, D1) Descriptive Statistical Analysis: Primary endpoint: Overall survival Secondary endpoints: PFS, ORR, DOR, safety, pharmacokinetics, immunogenicity, quality of life, exploratory biomarker analysis. ^a Up to 750 mg. ^b Until disease progression, intolerable toxicity, discontinuation decided by the participant or the investigator, death, withdrawal of informed consent, pregnancy, noncompliance with protocol or procedure requirements, administrative reasons, or other protocol-specified reasons, whichever occurred first. AUC, area under the curve; BICR, blinded independent central review; D, day; DOR, duration of response; ES-SCLC, extensive-stage small-cell lung cancer; IV, intravenous; ORR, objective response rate; PFS, progression-free survival; Q3W, every 3 weeks; R, randomization; RECIST, Response Evaluation Criteria in Solid Tumors. Presented by Dr. David R. Gandara [Slide 2] Efficacy: OS and subgroup analysis Median OS (95% CI), months | Events, n (%) Serplulimab group 12.4 (9.0–17.9) 50 (49.5) Atezolizumab group 10.2 (8.4–13.2) 58 (58.0) Hazard ratio (95% CI), 0.82 (0.56–1.20) 1-year OS rate (95% CI) 52.4% (40.8%–62.7%) 41.4% (30.2%–52.3%) 2-year OS rate (95% CI) 32.2% (20.1%–44.9%) 17.6% (8.1%–30.0%) Overall survival (%) Time since randomization (months) No. at risk (No. censored) Subgroup | Serplulimab (Events/Patients) | Atezolizumab (Events/Patients) | HR for death (95% CI) Age <65 years 18/39 19/37 1.03 (0.54–1.97) ≥65 years 32/62 39/63 0.70 (0.44–1.13) Sex Female 26/55 26/47 0.79 (0.46–1.36) Male 24/46 32/53 0.87 (0.51–1.49) Performance status 0 12/30 8/23 0.92 (0.37–2.29) 1 38/70 49/76 0.89 (0.58–1.36) ≥2 0/1 1/1 NE Race American Indian or Alaska native 3/3 0/0 NE Asian American 1/2 1/4 NE Black or African American 0/2 4/8 NE White 46/94 53/88 0.78 (0.52–1.16) Smoking history Current 20/35 17/28 0.78 (0.41–1.51) Former 28/62 41/69 0.71 (0.44–1.16) Never 2/4 0/3 NE Liver metastasis Yes 33/50 27/40 0.85 (0.51–1.42) No 17/51 31/60 0.60 (0.33–1.09) Brain metastasis Yes 4/15 7/14 0.67 (0.20–2.31) No 46/86 51/86 0.83 (0.55–1.24) 1st cycle chemotherapy Yes 9/19 12/24 1.16 (0.49–2.75) No 41/82 46/76 0.77 (0.50–1.17) Favors serplulimab Favors atezolizumab CI, confidence interval; HR, hazard ratio; NE, not evaluable; No., number; OS, overall survival. Presented by Dr. David R. Gandara Survival table Endpoint: Overall survival (primary, descriptive) Arm | N | Median | Median CI Serplulimab + carboplatin/etoposide | 101 | 12.4 | 9.0–17.9 Atezolizumab + carboplatin/etoposide | 100 | 10.2 | 8.4–13.2 HR: 0.82 HR CI: 0.56–1.20 1-year OS 52.4% (40.8–62.7) vs 41.4% (30.2–52.3) 2-year OS 32.2% (20.1–44.9) vs 17.6% (8.1–30.0) [Slide 3] Efficacy: Secondary endpoints assessed by BICR Serplulimab (n = 99)^a | Atezolizumab (n = 100) PFS^b, median (95% CI), months 5.4 (4.5–5.8) | 4.9 (4.2–5.6) HR (95% CI) 0.95 (0.67–1.35) Best overall response^c, n (%) Complete response 1 (1.0) | 1 (1.0) Partial response 64 (64.6) | 66 (66.0) Stable disease 25 (25.3) | 21 (21.0) Progressive disease 3 (3.0) | 6 (6.0) Not evaluable^d 6 (6.1) | 6 (6.0) ORR^c (95% CI), % 65.7 (55.4–74.9) | 67.0 (56.9–76.1) DOR^c, median (95% CI), months 4.2 (4.1–5.6) | 4.0 (3.1–4.5) HR (95% CI) 0.95 (0.62–1.47) ^a Patients who did not receive study treatment were excluded from the secondary endpoint analyses. ^b According to BICR assessments per RECIST v1.1. ^c Based on confirmed tumor responses according to BICR assessments per RECIST v1.1. ^d These patients did not have post-treatment tumor assessment, except for one patient in the atezolizumab group with a non-evaluable target lesion at post-treatment tumor assessment. BICR, blinded independent central review; CI, confidence interval; DOR, duration of response;; HR, hazard ratio; ORR, objective response rate; PFS, progression-free survival. Presented by Dr. David R. Gandara [Slide 4] Safety and tolerability Serplulimab (n = 99) | Atezolizumab (n = 100) TEAE 97 (98.0) | 99 (99.0) Grade ≥3 73 (73.7) | 73 (73.0) Serious 44 (44.4) | 46 (46.0) Leading to Tx interruption 54 (54.5) | 51 (51.0) Leading to Tx discontinuation 16 (16.2) | 12 (12.0) Leading to death 9 (9.1) | 8 (8.0) TEAEs related to serplulimab or atezolizumab 63 (63.6) | 68 (68.0) Grade ≥3 20 (20.2) | 25 (25.0) Serious 9 (9.1) | 7 (7.0) Leading to Tx interruption 22 (22.2) | 22 (22.0) Leading to Tx discontinuation 5 (5.1) | 4 (4.0) Leading to death^a 2 (2.0) | 2 (2.0) Adverse event of special interest 23 (23.2) | 23 (23.0) Infusion-related reaction 4 (4.0) | 3 (3.0) Immune-related adverse event 21 (21.2) | 23 (23.0) Most common TEAEs (incidence ≥15% in either group) Serplulimab Atezolizumab Anemia Nausea Fatigue Neutrophil count decreased Alopecia Diarrhea Vomiting Constipation Hypokalemia Platelet count decreased Arthralgia Dyspnea Decreased appetite Headache Atezolizumab, Grade 1–2 Atezolizumab, Grade 3–4 Serplulimab, Grade 1–2 Serplulimab, Grade 3–4 Incidence (%) ^a One patient in the serplulimab had respiratory failure and septic shock, and the other one had respiratory failure. One patient in the atezolizumab group had colitis and septic shock, and the other one had myocarditis. TEAE, treatment-emergent adverse event; Tx, treatment Presented by Dr. David R. Gandara

Conclusions • In the ASTRIDE trial, compared with atezolizumab, serplulimab was associated with ➢ numerically longer median OS ✓ ITT population: 12.4 versus 10.2 months, HR 0.82; 95% CI, 0.56–1.20 ✓ No 1st cycle chemotherapy: 13.4 versus 10.8 months, HR 0.77; 95% CI, 0.50–1.17 ➢ similar PFS, ORR, DOR outcomes ➢ similar safety profiles Serplulimab-chemotherapy compared favorably with atezolizumab-chemotherapy, supporting the benefit of serplulimab and applicability of the ASTRUM-005 findings to the USA patients with treatment-naïve ES-SCLC. CI, confidence interval. DOR, duration of response; ES-SCLC, extensive-stage small cell lung cancer; HR, hazard ratio; ORR, objective response rate; OS, overall survival; PFS, progression-free survival. Presented by Dr. David R. Gandara

Baseline characteristics and subsequent cancer therapy Data cutoff date: February 18, 2026; Median follow-up duration: 14.1 months ITT population: 201 patients; Safety set: 199 patients Serplulimab (n = 101) | Atezolizumab (n = 100) Age, median (range), years | 68 (35–91) | 67 (40–82) <65 years, n (%) | 39 (38.6) | 37 (37.0) ≥65 years, n (%) | 62 (61.4) | 63 (63.0) Sex, n (%) Female | 55 (54.5) | 47 (47.0) Male | 46 (45.5) | 53 (53.0) Performance status, n (%) 0 | 30 (29.7) | 23 (23.0) 1 | 70 (69.3) | 76 (76.0) ≥2a | 1 (1.0) | 1 (1.0) Race, n (%) American Indian or Alaska native | 3 (3.0) | 0 Asian | 2 (2.0) | 4 (4.0) Black or African American | 2 (2.0) | 8 (8.0) White | 94 (93.1) | 88 (88.0) Sum of longest diameter of target lesions, median (range), mm | 109.0 (12.0–262.0) | 108.0 (17.0–273.0) Smoking history, n (%) Current | 35 (34.7) | 28 (28.0) Former | 62 (61.4) | 69 (69.0) Never | 4 (4.0) | 3 (3.0) Metastatic disease, n (%) | 98 (97.0) | 98 (98.0) Liver metastasis | 50 (49.5) | 40 (40.0) Bone metastasis | 25 (24.8) | 28 (28.0) Brain metastasis | 15 (14.9) | 14 (14.0) Had 1st cycle of chemotherapy, n (%) | 19 (18.8) | 24 (24.0) Subsequent cancer therapy, n (%) | 38 (37.6) | 45 (45.0) Chemotherapy | 27 (26.7) | 24 (24.0) Lurbinectedin | 22 (21.8) | 16 (16.0) Immunotherapy | 16 (15.8) | 20 (20.0) Tarlatamab | 13 (12.9) | 15 (15.0) Radiotherapy | 21 (20.8) | 21 (21.0) Targeted therapy | 1 (1.0) | 1 (1.0) a The most recent performance status evaluations prior to the first dose of study treatment are presented. One patient in each group had performance status 1 at screening and 2 before the first dose of study treatment. ES-SCLC, extensive-stage small cell lung cancer. Presented by Dr. David R. Gandara

128P: Multi-LLM–Assisted Trial Selection for Network Meta-Analysis in First-Line Extensive-Stage Small Cell Lung Cancer Sameena Parveen, Darko Medin, Mark Gramling, Bruno Larvol, Judith Pérez-Granado The Larvol Group LLC, San Francisco/CA/USA Correspondence: judith.perez@larvol.com MADRID SPAIN 23-27 OCTOBER 2026 INTRODUCTION Eligibility-based trial selection may retain clinically distinct populations, limiting valid indirect comparisons in 1L ES-SCLC. Clinical comparability therefore requires assessment beyond standard eligibility criteria. We developed a multi-LLM consensus approach to identify clinically comparable trials for OS and PFS network meta-analysis. CONCLUSIONS • Strong cross-model agreement enabled consistent trial selection and comparable OS/PFS NMA networks in 1L ES-SCLC. • Separating eligibility from comparability flagged a responder-enriched trial as unsuitable for indirect comparison. • Heterogeneity: not detected for OS, high for PFS; Inconsistency: not assessed no closed loops. • Multi-LLM ranking was supported; PFS heterogeneity points to unmeasured effect modifiers, supporting future work on toxicity and patient-level effect modifiers. METHODS 1L ES-SCLC RCTs - OS/PFS HRs + 95% CIs available - Identified in LARVOL CLIN MULTI-LLM TRIAL ASSESSMENT: GPT-5.5 | Claude Opus 4.7 | Grok 4.3 Independent assessment in parallel ELIGIBILITY ASSESSMENT - Disease, Stage - LOT, Randomised design, Endpoint - OUTPUT: Evidence-linked inclusion decision CLINICAL COMPARABILITY & TRANSITIVITY AUDIT - Responder enrichment, Crossover, effect modifiers - OUTPUT: Comparable trials retained for NMA REGIMEN HARMONIZATION - Common vocabulary RANDOM-EFFECTS NMA - OS and PFS separately HETEROGENEITY & INCONSISTENCY ASSESSMENT EFFECT ESTIMATES & RANKING - HR + 95% CI P-scores RESULTS Random-effects NMA: Overall Survival [Network diagram - small node labels not transcribed] Endpoint | OS | PFS Eligible trials | 21–22 | 19–20 Clinically comparable | 21 | 19 NMA trials (regimens) | 17 (17) | 17 (17) Top-ranked regimen | Atligotatug + nivolumab + chemo | Anlotinib + chemo Heterogeneity | I² = 0% | I² =79.5% P-score ranking (OS): 1 | 0.906 | Fucosyl GM1 inhibitor + PD-1 inhibitor + Chemo | Nivolumab + Chemo 2 | 0.842 | PD-1 inhibitor + Chemo | Serplulimab + Chemo 3 | 0.812 | Multi-target TKI + PD-1 inhibitor + Chemo | Benmelstobart + Anlotinib + Chemo 4 | 0.693 | PD-1 inhibitor + CDK4/6 inhibitor + Chemo | Atezolizumab + Trilaciclib + Chemo 5 | 0.669 | PD-1 inhibitor + Chemo | Nivolumab + Chemo 6 | 0.649 | PD-L1 inhibitor + Chemo | Atezolizumab + Chemo 7 | 0.579 | PD-L1 inhibitor + Chemo | Adebrelimab + Chemo 8 | 0.554 | TIGIT inhibitor + PD-L1 inhibitor + Chemo | Atezolizumab + Tiragolumab + Chemo 9 | 0.511 | PD-1 inhibitor + Chemo | Pembrolizumab + Chemo 10 | 0.467 | PD-1 inhibitor + Chemo | Tislelizumab + Chemo 11 | 0.424 | PD-L1 inhibitor + Chemo | Socazolimab + Chemo 12 | 0.424 | PD-1 inhibitor + Chemo | Toripalimab + Chemo 13 | 0.382 | PD-L1 inhibitor + CTLA-4 inhibitor + Chemo | Durvalumab + Tremelimumab + Chemo 14 | 0.282 | TIGIT inhibitor + Chemo | Vibostolimab + Chemo 15 | 0.177 | CTLA-4 inhibitor + Chemo | Ipilimumab + Chemo 16 | 0.094 | Chemo | Chemo 17 | 0.036 | CDK inhibitor + Chemo | Ronociclib + Chemo P-score scale: 1.00 (Higher relative ranking) 0.75 0.50 0.25 0.00 (Lower relative ranking)

Background • The global phase 3 ASTRUM-005 trial demonstrated survival benefits with serplulimab-chemotherapy for treatment-naïve extensive-stage small cell lung cancer (ES-SCLC)1. • Confirmatory data for consistency of ASTRUM-005 findings in USA patients was required. • Moreover, there is no prospective data directly comparing different immune checkpoint inhibitors in this setting. Here we present the randomised ASTRIDE bridging trial, comparing serplulimab-chemotherapy versus atezolizumab-chemotherapy in USA patients with ES-SCLC. USA, United States of America. 1. JAMA. 2022 Sep 27;328(12):1223-1232. Presented by Dr. David R. Gandara MADRID 2026 ESMO congress

[Slide 2] Less = more ? A - Continuation (N=47) Not treated (N=2) - Disease evolution (n=1) - Toxicity (n=1) A – Continuation - treated (N=45) Study withdrawn (N=39) - Disease evolution (n=21, 53.8%) - 2-year treatment (n=12, 30.8%) - Toxicity (n=4, 10.[illegible]%) - Second cancer (n=1, 2.6%) - Death (n=1, 2.6%) Treatment ongoing (N=6) Subsequent treatment (N=22, 56.4%) B - De-escalation (N=46) Not treated (N=5) - Disease evolution (n=5) B – De-escalation - treated (N=41) Study withdrawn (N=30) - Disease evolution (n=28, 93.3%) - Death (n=1, 3.3%) - 2-year treatment (n=1, 3.3%) Treatment ongoing (N=11) Subsequent treatment (N=28, 93.3%) Continuation: 12 patients completed 2 years De-escalation: only 1 patient completed 2 years Lower progression-free survival with de-escalation? Less progression with continuation? Subsequent treatment 56.4% vs 93.3% 4 toxicity withdrawals with continuation — per protocol Mihaela Aldea, MD PhD 2618RO – Toffart AC, ESMO 2026 [Slide 3] Re-treatment with pembrolizumab – STOP & GO ? Keynote 189: median PFS 8.8 mo A Progression-free Survival Hazard ratio for disease progression or death, 0.52 (95% CI, 0.43–0.64) P<0.001 Between stopping at 6 months and the expected median progression-free survival, pembrolizumab is still present in the blood (half-life 23 days) Pembrolizumab combination / Placebo combination No. at Risk (0, 3, 6, 9, 12, 15, 18, 21 months) Pembrolizumab combination | 410 | 322 | 256 | 149 | 60 | 17 | 5 | 0 Placebo combination | 206 | 141 | 80 | 40 | 16 | 3 | 1 | 0 Should all patients who stop be rechallenged with ICI at progression? Does stopping ICI have any detrimental effect for patients who had CR or PR? Or for those with progression after ICI clearance ~ 4 mo ? Mihaela Aldea, MD PhD Gandhi L, NEJM 2018; Patil, Cancer Res Stat and Treat 2019 Survival table Endpoint: PFS (KEYNOTE-189, discussant reproduction) Pembrolizumab combination | 410 Placebo combination | 206 HR: 0.52 HR CI: 0.43–0.64 P: P<0.001 [Slide 4] Pragmatic or tailored de-escalation? FIXED duration 2 years the BOLD 6 months IFCT-2103 DIAL trial the CAUTIOUS 2 years + PET/CT guided stop Frost N, Ann Oncol 2026 PET/CT at 2 years (n = 126) Complete metabolic response (62%) → stop ICB Residual metabolic activity (38%) → biopsy ± local ablative treatment 28% of re-biopsies = second primary lung cancers (immune-cold) STOP & GO Guided by deep molecular responses Minimal residual disease (MRD) in leukemia [MRD-over-time line plots; axis labels not transcribed] Niemannn C, Blood Advances 2025 Are 2 years of ICI needed for all patients? Are 6 months enough for all patients? Mihaela Aldea, MD PhD

[Slide 2] KRAS combos: response by PD-L1 Chemo-pembro (KEYNOTE-189): Unselected by KRAS mut <1% (n=127): 32%, 6.1 (CI NR) | 1-49% (n=128): 49%, 9.0 (CI NR) | >=50% (n=132): 62%, 11.3 (CI NR) Adagrasib + pembro <1% (n=39): 36%, 8.2 (2.7-13.6) | 1-49% (n=27): 41%, 13.5 (2.3-NE) | >=50% (n=38): 61%, 27.7 (9.6-NE) Olomorasib + pembro 0-49% (n=25): 60%, 13.2 (3.6-NR) | >=50% (n=20): 90%, NR (12.0-NR) Calderasib + pembro 1-49% (n=58): 55%, 25.1 (10.5-NR) | >=50% (n=64): 84%, 34.4 (21.8-NR) ORR ↑ with PD-L1 expression for chemo-pembro and KRASi-pembro Olomorasib + chemo-pembro <1% (n=26): 54%, 9.7 (5.9-15.9) | 1-49% (n=31): 68%, 12.1 (5.9-NR) | >=50% (n=18): 67%, NR (7.7-NR) ORR plateau Waterfall: Maximum Change in Tumor Size (%); Ongoing; Received 1 cycle of SOC prior to enrollment; PD-L1 unknown; PD-L1 <1%; PD-L1 1-49%; PD-L1 >=50% Chance finding, or does chemo limits KRASi + pembro efficacy in PD-L1 high ? Gadgeel JCO 2020; Gandhi NEJM 2018; Burns JTO 2026, Negrao ESMO 2026. Gray ESMO 2026, Jänne ASCO 2025 Mihaela Aldea, MD PhD [Slide 3] Can patients stay on combos? | Monotherapy: Elisrasib (n=43) | Olomorasib‡ (n=195) | Calderasib‡ (n=271) | Divarasib‡ (n=137) | Adagrasib (n=116) | + Pembrolizumab: Calderasib (n=136) | Elisrasib (n=52) | Olomorasib (n=93) | Divarasib (n=78) | Adagrasib (n=149) | + Chemo-pembro: Olomorasib (n=77) | Calderasib (n=46) KRASi dose reduction | 0% | 6% | NR | 14% | 52% | NR | 23% | 20% | 53% | 48% | 19% | NR KRASi discontinuation | 0% | 1% | 1% | 3% | 7% | 4% | 2% | 9%¶ | 13%§ | 13% | 8%† | 7% ICB discontinuation | — | — | — | — | — | 24% | 10% | 23%¶ | 13%§ | 24% | NR | 30% Grade >=3 TRAE | 7% | 7% | 8% | 12% | 45% | 29% | 33% | 33% | 65% | 68% | 55% | 72% ‡All tumor types. Discontinuation of †all or §any study treatment. ¶Incl. 6.5% stopping both drugs. NR, not reported. KRAS OFF inhibitor combos lead to frequent drug reduction or discontinuation. Impact? Negrao M, ESMO 2026; Gray J,, ESMO 2026; Spira AI, ASCO 2025; Lu S, ASCO 2026; Skoulidis F, ASCO 2026; Clarke JM, WCLC 2023; Jänne PA, ASCO 2025, JCO 2026, NEJM 2022; Murciano-Goroff YR, Nat Commun 2026; Sacher AG, NEJM 2023; Burns TF, J Thorac Oncol 2026 [Slide 4] Unanswered questions in the 1L Combination with optimal drug exposure: e.g. EGFR: FLAURA2 -> Overall survival benefit Combination with sub-optimal drug exposure: e.g. KRAS combos with high rates of drug reduction or early discontinuation [scale: PD-L1 low / PD-L1 high] Sequence of drugs with optimal drug exposure: ICB-chemo -> KRASi or even KRASi -> ICB-chemo [scale: PD-L1 low / PD-L1 high] Impact on survival? - Can KRASi resensitize cold tumors to ICI? - PD-L1 is not enough: translational studies on co-mutations and TME are needed - How to design adaptive trials by predicting which drug component matters most, to guide de-escalation for toxicity or re-escalation if ICB is discontinued ? One size does not fit all in KRAS-mutant NSCLC Mihaela Aldea, MD PhD

IFCT Conclusion of phase II Shortening maintenance immunotherapy in stage IV NSCLC patients with 6-month disease control, after chemo-immunotherapy induction ■ was reassuring since associated with similar rates of 18-month, and 24-month OS, as compared with immunotherapy continuation ■ leading to numerically lower frequency of TRAEs Based on this analysis, in December 2025, the Independent Data Monitoring Committee recommended to pursue the trial phase 3 part of which accrual is currently continuing (493/1250 patients) ESMO 2026

IFCT Patients' characteristics | | Included (N=197) | Randomized (N=93) | Arm A - Continuation (N=47) | p-value | Arm B - De-escalation (N=46) Sex | Male | 134 (68.0) | 62 (66.7) | 32 (68.1) | 0.77 | 30 (65.2) Age (year) | Median [IQR] | 63.3 [58.4-68.9] | 65.5 [59.0-69.1] | 65.4 [59.1-68.4] | 0.73 | 66.4 [58.4-70.5] Histology | Adenocarcinoma | 139 (70.6) | 75 (80.6) | 37 (78.7) | 1.00 | 38 (82.6) Histology | Squamous cell carcinoma | 34 (17.3) | 13 (14.0) | 7 (14.9) | | 6 (13.0) TNM at inclusion | IVA | 67 (34.0) | 34 (36.6) | 14 (29.8) | 0.17 | 20 (43.5) TNM at inclusion | IVB | 130 (66.0) | 59 (63.4) | 33 (70.2) | | 26 (56.5) PS | 0 | 79 (40.1) | 45 (48.4) | 24 (51.1) | 0.60 | 21 (45.7) PS | 1 | 118 (59.9) | 48 (51.6) | 23 (48.9) | | 25 (54.3) Brain metastasis | Yes | 35 (17.8) | 17 (18.3) | 9 (19.1) | 0.83 | 8 (17.4) PD-L1 expression | Highly positive (≥ 50%) | 47 (23.9) | 24 (25.8) | 13 (27.7) | 0.91 | 11 (23.9) PD-L1 expression | Positive (≥ 1 et < 50%) | 67 (34.0) | 33 (35.5) | 16 (34.0) | | 17 (37.0) PD-L1 expression | Negative (<1%) | 83 (42.1) | 36 (38.7) | 18 (38.3) | | 18 (39.1) 6-month response status | Complete or partial response | NA | 72 (77.4) | 35 (74.5) | 0.49 | 37 (80.4) 6-month response status | Stable disease | NA | 21 (22.6) | 12 (25.5) | | 9 (19.6) ESMO 2026

IFCT Intergroupe Francophone de Cancérologie Thoracique IFCT-2103 DIAL Analysis of the phase 2 part of IFCT-2103 DIAL phase 2-3 randomized trial, comparing continuation pembrolizumab up to 2 years versus observation, in metastatic Non-Small Cell Lung Cancer patients experiencing disease control at 6 months A-C Toffart, C. Mascaux, B. Godbert, C. Guguen, J. Cadranel, M. Dewolf, D. Debieuvre, L. Bigay, A. Lemoine, E. Pons-Tostivint, J. Madelaine, L. Moreau, E. Giroux-Leprieur, T. Chatellier, C. Dayen, E. Amour, A. Langlais, F. Morin, A. Cortot, G. Zalcman Sponsor/Funding: IFCT NCT 05255302 25OCT2026

[Slide 2] A Practical Classification BISPECIFIC BISPECIFIC BISPECIFIC BISPECIFIC TRISPECIFIC TETRASPECIFIC PD-1 Bispecific Receptor Ligand Bispecific ADC Bispecific ICI Non-Bridging Bispecific T-cell Engager Trispecific Cell Bridging Tetraspecific Cell Bridging [Slide 3] Bispecific ICI Non-Bridging Tumour Microenvironment (PD-1 / VEGFR) Pumitamig + Chemo (Rosetta-Lung02)1 1L NSCLC | Phase 2/3 | Global PD-L1 | <1% | 1-49% | >50% ORR | 48% | 78% | 100% Setting | Population | Regimen Rosetta Lung-01 | 1L ES-SCLC | Extensive Stage | P + EP chemo Rosetta Lung-02 | 1L Stage 4 NSCLC | Sq & Non-Sq | P + chemo Rosetta Lung-107 | 2L NSCLC | Sq & Non-Sq | P + Doce Rosetta Lung-201 | Stage III NSCLC | Unresectable | P Rosetta Lung-202 | 1L NSCLC | PD-L1 >50% | P Pumitamig + B7-H3 ADC (BNT-324-01)2 Ph1 | Dose optimisation & signal seeking | 7 cohorts (NSCLC, SCLC, AGA+, AGA-) SCLC • ORR (95% CI) by line of therapy*: • 1L (n=13): 92.3% (64.0–99.8) • 2L (n=21): 76.2% (52.8–91.8) • 3L+ (n=21): 52.4% (29.8–74.3) • ORR (95% CI) prior DLL3-targeting agents: • DLL3 TCE (n=8) 87.5% (47.4–99.7) • DLL3 TCE + DLL3 ADCs (n=10): 70.0% (34.8–93.3) • ORR (95% CI) prior IO (n=48): 60.4% (45.3–74.2) 2L+ SCLC Percent change from baseline (%) 1Peters et al., ASCO Annual Meeting (2026), Chicago, USA; 2Sun et al., IASLC WCLC (2026), Seoul, Korea [Slide 4] T-cell Engagers: Classification Lessons from Haematology What Did We Learn? • Effective • Practicalities of use • Individually dosed • Management of TRAEs • Combinations (chemotherapy, TKIs) Challenges in Solid Tumours • More heterogenous • TAAs harder to identify • TCEs need higher exposure Immune Cell-Engagers T-cell x membrane TAA T-cell x HLA restricted TAA NK cell engagers Macrophage engagers CD3xDLL3; CD28xPSMA; CD3xTROP2; Vδ2-PSMA CD3xHLA*0201-GP100; CD3xHLA*0201- PRAME NKG2DxCD16xHER-2; CD16xIL-15xCD33 CD47xFCγ; CD4xFAPxHSA; CD40x4-1BB T-cell Tumor cell HLA NK cell CD16 CD47 FCγR Macrophage Fontana et al., Lancet Regional Health – Europe (2026);64: 101599

[Slide 2] Global phase 2 in post-chemoimmunotherapy mNSCLC ROSETTA LUNG-107 Key inclusion criteria • 2L stage IV NSCLC after PD on chemoimmunotherapy^a • Measurable disease per RECIST v1.1 • ECOG PS 0 or 1 • PD-L1 all-comers Key exclusion criteria • Prior anti-VEGF mAb or PD-(L)1/VEGF bsAb • Significant risk of hemorrhage Part 1: Safety run-in Part 1A N = 3: Pumitamig 1400 mg Q3W (IV) + Docetaxel 75 mg/m2 Q3W (IV) Part 1B N = 3: Pumitamig 2000 mg Q3W (IV) + Docetaxel 75 mg/m2 Q3W (IV) IRC safety evaluation Part 2: Dose expansion N = 53^b: Pumitamig 1400 mg Q3W (IV)^c + Docetaxel 75 mg/m2 Q3W (IV) Key endpoints Primary: • DLTs (Part 1), safety (TEAEs, SAEs, AESIs), ORR (per investigator) Secondary: • DCR, DOR, depth of response, TTR, PFS per RECIST v1.1, OS, PK, ADA prevalence Data cutoff: July 10, 2026 Median follow-up: 12.2 weeks (range, 0.1–41.3)^d Clinical trials identifier: NCT06841055 Enrollment (final): N = 59^b Location: Global (Australia, Republic of Korea, Spain, Türkiye, United Kingdom, United States) Status: Active, fully enrolled ^aPatients with AGAs were required to have received 1L immunotherapy in combination with targeted therapy to be eligible. ^bSix patients were enrolled after data cutoff. ^cPumitamig was administered at 1400 mg as recommended by the IRC based on the totality of evidence from Part 1 and the pumitamig lung program; patients were treated until disease progression, intolerable toxicity, patient withdrawal, death, trial termination, or a 2-year limit (whichever occurs first). ^dThe median follow-up reflects a pause in enrolment for a pre-specified futility analysis conducted after the first 22 patients were enrolled at the pumitamig 1400 mg dose level, with longer follow-up among these patients. [Slide 3] Safety summary and TRAEs in the overall population ROSETTA LUNG-107 Safety summary | Pumitamig 1400 mg + docetaxel (N = 50) Any TEAE, n (%) | 47 (94) Grade ≥3 | 28 (56) Any TRAE, n (%)^a | 26 (52) Grade ≥3 | 9 (18) Serious TEAE (%) | 18 (36) Serious TRAE, n (%)^a | 3 (6) TRAE leading to treatment interruption, n (%)^a | 9 (18) TRAE leading to discontinuation of treatment, n (%)^a | 4 (8) TRAE leading to death, n (%)^a | 0 VEGF-related events, n (%) | 13 (26) Grade ≥3 | 2 (4)^b Hemorrhage events, n (%) | 8 (16)^c Grade ≥3 | 0 irAE, n (%) | 11 (22) Grade ≥3 | 5 (10)^d TRAEs occurring in ≥5% of patients^a (Patients, % (n); Any grade / Grade ≥3) Stomatitis | 10% (5) Infusion-related reaction | 10% (5) / 4% (2) AST increased | 8% (4) / 4% (2) Diarrhea | 8% (4) Epistaxis | 6% (3) Asthenia | 6% (3) / 2% (1) Anemia | 6% (3) ALT increased | 6% (3) / 2% (1) The safety profile was manageable. All bleeding events were grade 1, and there were no grade ≥3 hypertension/proteinuria events. The rate of discontinuation was low, and no DLTs were reported ^aTRAEs related to pumitamig. ^bBoth events were grade 3: colovesical fistula (n = 1) and pulmonary embolism (n = 1). ^cAll events were grade 1: epistaxis (n = 5), hemoptysis (n = 3), hematochezia (n = 1), hematuria (n = 1), hemorrhoidal hemorrhage (n = 1), and rectal hemorrhage (n = 1). ^dGrade 3 hepatitis (n = 2), grade 3 pruritus (n = 1), grade 4 colitis (n = 1), and grade 4 pneumonitis (n = 1); no grade 5 irAEs were reported. [Slide 4] Efficacy: BOR in the response-evaluable population^a ROSETTA LUNG-107 [Waterfall plot: Depth of response (%), bars colored by cBOR (PR, SD, PD, NE) with PD-L1 TPS strip (<1%, 1%–49%, ≥50%)] [Spider plot: Percent change in tumor size from baseline vs Overall treatment duration (weeks); NE, PD, SD, PR, On Treatment] Encouraging responses were observed across PD-L1 levels and appeared to be durable ^aResponse-evaluable patients had ≥2 post-baseline tumor assessments or 1 assessment followed by PD, death, or study discontinuation.

LBA76 - A global phase 2, open-label trial of pumitamig (PD-L1 × VEGF-A bsAb) + docetaxel post-chemoimmunotherapy in mNSCLC: First data from ROSETTA Lung-107 Speakers Byoung Chul Cho (Seoul, Republic of Korea) Lecture Time 08:30 - 08:35 Abstract Background Pumitamig (BNT327/BMS986545) is an investigational bsAb designed to target both PD-L1 and VEGF-A within the tumor and TME with a single molecule. This dual blockade is hypothesized to boost antitumor immunity and improve clinical outcomes. We report initial efficacy and safety of pumitamig + docetaxel post-chemoimmunotherapy (CIT) in patients with mNSCLC. Methods ROSETTA Lung-107 is a global, open-label, single-arm phase 2 trial in patients with stage IV NSCLC who received 1L CIT, had measurable disease (RECIST v1.1), and ECOG PS ≤1. In Part 1 (safety run-in/dose escalation), patients received pumitamig 1400 (n=3) or 2000 mg (n=3; not shown) + docetaxel Q3W. In Part 2 (dose expansion), patients received pumitamig 1400 mg + docetaxel Q3W. Primary endpoints were DLTs (Part 1 only), safety, and ORR. Secondary endpoints included DCR and PFS. Results At data cutoff (July 10, 2026), 50 patients had received pumitamig 1400 mg + docetaxel in Part 1 and 2, and 33 remained on treatment. Median follow-up was 12.2 wk (0.1–41.3). Median age was 67.0 y (34–83), 70.0% were male, 26.0% had squamous histology, and 66.0% had ECOG PS 1. Encouraging responses were observed in response-evaluable patients (n=30) across PD-L1 levels (Table); median TTR was 1.4 mo (1.2–6.1). Median PFS was 6.9 mo (4.2–NR). In the safety analysis set (n=50), any-grade pumitamig-related TRAEs occurred in 52.0% of patients (Gr ≥3 18.0%). Pumitamig-related TRAEs led to discontinuation in 8.0%. Immune-mediated AEs occurred in 30.0% (Gr ≥3 14.0%) and VEGF-related events in 26.0% (Gr ≥3 4.0%). Hemorrhage/bleeding events occurred in 16.0% (all Gr 1/2). No DLTs were reported. Table: LBA76 | Overall (N=30) | PD-L1 TPS <1% (n=7) | PD-L1 TPS 1–49% (n=18) | PD-L1 TPS ≥50% (n=5) Confirmed ORR, n (%) | 11 (36.7) | 3 (42.9) | 6 (33.3) | 2 (40.0) DCR, n (%) | 27 (90.0) | 6 (85.7) | 17 (94.4) | 4 (80.0) BOR, n (%) | 11 (36.7) | 3 (42.9) | 6 (33.3) | PR | 16 (53.3) | 3 (42.9) | 11 (61.1) | 2 (40.0) SD | 2 (6.7) | 1 (14.3) | 0 | 2 (40.0) PD | 1 (3.3) | 0 | 1 (5.6) | 1 (20.0) NE | | | | 0 Unconfirmed ORR, n (%) | 13 (43.3) | 3 (42.9) | 8 (44.4) | 2 (40.0) BOR, n (%) | | 3 (42.9) | 8 (44.4) | 2 (40.0) PR | 13 (43.3) | 3 (42.9) | 9 (50.0) | 2 (40.0) SD | 14 (46.7) | 1 (14.3) | 0 | 1 (20.0) PD | 2 (6.7) | 0 | 1 (5.6) | 0 NE | 1 (3.3) | | | Patients received pumitamig 1400 mg + docetaxel 75 mg/m2 Q3W. Response-evaluable patients had ≥2 post-baseline tumor assessments or 1 assessment followed by PD, death, or study discontinuation. Conclusions Pumitamig + docetaxel showed encouraging antitumor activity with a manageable safety profile in patients with mNSCLC post-CIT. These data support further development of pumitamig in this setting and will inform pivotal study design.

[Slide 1] Global phase 2 in post-chemoimmunotherapy mNSCLC ROSETTA LUNG-107 Key inclusion criteria • 2L stage IV NSCLC after PD on chemoimmunotherapy^a • Measurable disease per RECIST v1.1 • ECOG PS 0 or 1 • PD-L1 all-comers Key exclusion criteria • Prior anti-VEGF mAb or PD-(L)1/VEGF bsAb • Significant risk of hemorrhage Part 1: Safety run-in Part 1A N = 3: Pumitamig 1400 mg Q3W (IV) + Docetaxel 75 mg/m2 Q3W (IV) Part 1B N = 3: Pumitamig 2000 mg Q3W (IV) + Docetaxel 75 mg/m2 Q3W (IV) IRC safety evaluation Part 2: Dose expansion N = 53^b: Pumitamig 1400 mg Q3W (IV)^c + Docetaxel 75 mg/m2 Q3W (IV) Key endpoints Primary: • DLTs (Part 1), safety (TEAEs, SAEs, AESIs), ORR (per investigator) Secondary: • DCR, DOR, depth of response, TTR, PFS per RECIST v1.1, OS, PK, ADA prevalence Clinical trials identifier: NCT06841055 Enrollment (final): N = 59^b Location: Global (Australia, Republic of Korea, Spain, Türkiye, United Kingdom, United States) Status: Active, fully enrolled Data cutoff: July 10, 2026 Median follow-up: 12.2 weeks (range, 0.1–41.3)^d ^aPatients with AGAs were required to have received 1L immunotherapy in combination with [illegible — cut off] [Slide 2] Efficacy: BOR in the response-evaluable population^a ROSETTA LUNG-107 | Pumitamig 1400 mg + docetaxel^a (N = 30) | TPS <1% (n = 7) | TPS 1%–49% (n = 18) | TPS ≥50% (n = 5) cORR, n (%) [95% CI] | 11 (36.7) [19.9–56.1] | 3 (42.9) [9.9–81.6] | 6 (33.3) [13.3–59.0] | 2 (40.0) [5.3–85.3] uORR, n (%) | 13 (43.3) | 3 (42.9) | 8 (44.4) | 2 (40.0) cDCR, n (%) | 27 (90.0) | 6 (85.7) | 17 (94.4) | 4 (80.0) cBOR, n (%) PR | 11 (36.7) | 3 (42.9) | 6 (33.3) | 2 (40.0) SD | 16 (53.3) | 3 (42.9) | 11 (61.1) | 2 (40.0) PD | 2 (6.7) | 1 (14.3) | 0 | 1 (20.0) NE | 1 (3.3) | 0 | 1 (5.6) | 0 • Encouraging responses were observed across histologies ▪ SQ NSCLC (n = 7): cORR and uORR were 57.1%, with cDCR of 100% ▪ NSQ NSCLC (n = 23): cORR was 30.4% (uORR 39.1%), with cDCR of 87.0% • Early mPFS (ITT) was 6.9 months (95% CI 4.2–NR)^b ▪ The 6-month PFS rate was 63.3% (95% CI 39.6–79.8) Encouraging responses were observed across PD-L1 levels and histologies, with early PFS data showing a median of 6.9 months ^aResponse-evaluable patients had ≥2 post-baseline tumor assessments or 1 assessment followed by PD, death, or study discontinuation. ^bMedian follow-up: 12.2 weeks (range, 0.1–41.3). [Slide 3] Study Design A Phase 1/2 study (NCT06741644) is ongoing to evaluate CS2009 as monotherapy or in combination with chemotherapy in patients with advanced non-small cell lung cancer (NSCLC) across first-line and later-line settings. Key Eligibility Criteria: • Histologically or cytologically confirmed stage IIIB/IIIC or IV NSCLC without known actionable oncogenic alteration • ECOG PS 0–1 • At least one measurable lesion per RECIST v1.1 Endpoints: Safety, ORR per RECIST v1.1 Safety was assessed in all treated patients. Efficacy was assessed in patients with at least one post-baseline tumor assessment per RECIST v1.1. Group 1 (≥2L NSCLC monotherapy) Heavily pre-treated later-line NSCLC → CS2009 10 / 20 / 30 / 45 mg/kg, IV, Q3W Group 2 (2/3L NSCLC combination) NSCLC pre-treated with platinum-based chemotherapy and anti-PD-(L)1 inhibitor → CS2009 20 / 30 mg/kg + docetaxel, IV, Q3W Group 3 (1L NSCLC monotherapy) Treatment-naïve NSCLC, PD-L1 TPS ≥1% → CS2009 20 / 30 mg/kg, IV, Q3W Group 4 (1L SQ NSCLC combination) Treatment-naïve squamous NSCLC → CS2009 20 / 30 mg/kg + paclitaxel + carboplatin, IV, Q3W, up to 4 cycles, followed by maintenance of CS2009 Group 5 (1L NSQ NSCLC combination) Treatment-naïve non-squamous NSCLC → CS2009 20 / 30 mg/kg + pemetrexed + carboplatin, IV, Q3W, up to 4 cycles, followed by maintenance of pemetrexed + CS2009 Treatment continued until progressive disease, unacceptable toxicity, consent withdrawal or death Note: Demographic and safety data for Group 5 are presented. Efficacy data remain immature due to short follow-up and are not presented. Dr. Anwen Xiong

LBA76 ROSETTA Lung-107 A global phase 2, open-label trial of pumitamig (PD-L1 × VEGF-A bsAb) + docetaxel post-chemoimmunotherapy in mNSCLC: first data from ROSETTA Lung-107 Byoung Chul Cho,1 Anja Williams, Ozgur Ozyilkan, Adnan Khattak, Gyeong-Won Lee, Oscar Juan-Vidal, Se-Hoon Lee, Peter Dickinson, Paul Shaw, Uchechukwu Dimkpa, Angus Lee, Ulug Gunaydin, Vikesh Patel, Aurora O'Brate, Ilhan Celik, Özlem Türeci, Ugur Sahin, Ernest Nadal 1Division of Medical Oncology, Yonsei Cancer Center, Yonsei University of Medicine, Seoul, Republic of Korea Byoung Chul Cho October 25, 2026 Full list of author affiliations can be seen on slide 13

[Slide 1] NeoCOAST-2: EFS, ctDNA, and updated safety data • ctDNA clearance after neoadjuvant chemoimmunotherapy is associated with pCR/MPR and survival outcomes.1–3 – In AEGEAN, ctDNA clearance before surgery was associated with pCR and longer EFS.4–6 • The NeoCOAST-2 primary analysis demonstrated promising pCR/MPR rates and manageable safety.7 Key eligibility criteria • Resectable Stage IIA–IIIB NSCLC (AJCC 8th edition) • EGFR/ALK wild-type • ECOG PS 0 or 1 • Previously untreated R* Neoadjuvant therapy (every 3 weeks, four cycles) Arm 1 Durvalumab + oleclumab + CT (N=76) Arm 2 Durvalumab + monalizumab + CT (N=72) Arm 4 Durvalumab + Dato-DXd + Pt (N=54) Surgery† Adjuvant therapy for ≤1 year Durvalumab + oleclumab Durvalumab + monalizumab Durvalumab Safety and efficacy follow-up Outcomes • First report of EFS‡ • ctDNA clearance • EFS by ctDNA clearance and pathological response • Updated safety data Data cutoff for ctDNA/safety: 14 Oct 2025 Data cutoff for EFS: 25 March 2026 ctDNA assessment: Baseline | C3D1 (pre-dose) | Pre-surgery | Post-surgery → Analysed using the GRAIL Portrait™ Assay8§ ctDNA clearance at C3D1 was defined as ctDNA detected at baseline and undetected at C3D1 [footnotes and references: small print, largely illegible] [Slide 2] Conclusions • In NeoCOAST-2, novel perioperative combination regimens demonstrated prolonged EFS, with 24-month EFS rates ranging from 70.5% to 77.9% across arms. • Neoadjuvant ctDNA clearance was associated with improved EFS in all arms, supporting previous analyses1–6 and highlighting neoadjuvant ctDNA clearance as a potential prognostic biomarker. – Patients with both ctDNA clearance at C3D1 and pCR had the longest EFS. • No new safety signals were identified during adjuvant treatment when compared with the primary analysis.7 • The findings available to date, including pCR rates reported in the primary analysis7 and the EFS and ctDNA data reported here, warrant further evaluation of combinations with Dato-DXd in resectable NSCLC. – In this analysis, patients in Arm 4 had the numerically highest ctDNA clearance rate. A manuscript containing additional EFS, ctDNA, and safety data from these analyses has been published simultaneously in Nature (McGrath L, et al. Nature 2026) and can be accessed via the QR code. 1. Chaudhuri AA, et al. Cancer Discov 2017;7:1394–403; 2. Moding EJ, et al. Nat Cancer 2020;1:176–83; 3. Reck M, et al. Ann Oncol 2024;35:Abstract LBA49; 4. Gale D, et al. Cancer Res 2024;84(7_suppl):Abstract CT238; 5. Heymach J, et al. Ann Oncol 2025;36:Abstract LBA70; 6. Cascone T, et al. Nature 2026; doi: 10.1038/s41586-026-10925-6; 7. Cascone T, et al. Nat Med 2025;31:2788–96.

NeoCOAST-2: EFS, ctDNA, and updated safety data - ctDNA clearance after neoadjuvant chemoimmunotherapy is associated with pCR/MPR and survival outcomes.1-3 - In AEGEAN, ctDNA clearance before surgery was associated with pCR and longer EFS.4-6 - The NeoCOAST-2 primary analysis demonstrated promising pCR/MPR rates and manageable safety.7 Key eligibility criteria - Resectable Stage IIA-IIIB NSCLC (AJCC 8th edition) - EGFR/ALK wild-type - ECOG PS 0 or 1 - Previously untreated R* Neoadjuvant therapy (every 3 weeks, four cycles) Arm 1 Durvalumab + oleclumab + CT (N=76) Arm 2 Durvalumab + monalizumab + CT (N=72) Arm 4 Durvalumab + Dato-DXd + Pt (N=54) Surgery† Adjuvant therapy for <=1 year Durvalumab + oleclumab Durvalumab + monalizumab Durvalumab Safety and efficacy follow-up Outcomes First report of EFS‡ ctDNA clearance EFS by ctDNA clearance and pathological response Updated safety data Data cutoff for ctDNA/safety: 14 Oct 2025 Data cutoff for EFS: 25 March 2026 ctDNA assessment: Baseline | C3D1 (pre-dose) | Pre-surgery | Post-surgery -> Analysed using the GRAIL PortraitTM Assay8§ ctDNA clearance at C3D1 was defined as ctDNA detected at baseline and undetected at C3D1 *Stratified by PD-L1 status (<1% versus >=1%). †Within 40 days of the last dose of neoadjuvant therapy. ‡After the opportunity to complete >=24 months of follow-up in all patients, including the opportunity to complete adjuvant therapy. §ctDNA was analysed using the tumour-agnostic methylation-based GRAIL Portrait Assay (GRAIL, Inc, Sunnyvale, CA) with a proprietary NSCLC classifier for ctDNA detection. AJCC, American Joint Committee on Cancer; ALK, anaplastic lymphoma kinase; C3D1, Cycle 3 Day 1; CT, platinum-doublet chemotherapy; ctDNA, circulating tumour DNA; Dato-DXd, datopotamab deruxtecan; ECOG PS, Eastern Cooperative Oncology Group performance status; EFS, event-free survival; EGFR, epidermal growth factor receptor; MPR, major pathological response; pCR, pathological complete response; PD-L1, programmed cell death ligand-1; Pt, single-agent platinum chemotherapy 1. Chaudhuri AA, et al. Cancer Discov 2017;7:1394-403; 2. Moding EJ, et al. Nat Cancer 2020;1:176-83; 3. Cascone T, et a. Nature 2026; doi: 10.1038/s41586-026-10925-6; 4. Reck M, et al. Ann Oncol 2024;35:Abstract LBA49; 5. Gale D, et al. Cancer Res 2024;84(7_suppl):Abstract CT238; 6. Heymach J, et al. Ann Oncol 2025;36:Abstract LBA70; 7. Cascone T, et al. Nat Med 2025;31:2788-96; 8. Klein E, et al. Ann Oncol 2021;32:1167-77.

24-month EFS rates were numerically higher in patients with versus without pCR Arm 1: Durvalumab + oleclumab + CT 24-month EFS rate: 84.6% (95% CI: 51.2-95.9) [pCR] 24-month EFS rate: 67.0% (95% CI: 52.6-77.9) [non-pCR] Arm 2: Durvalumab + monalizumab + CT 24-month EFS rate: 94.4% (95% CI: 66.6-99.2) [pCR] 24-month EFS rate: 69.1% (95% CI: 54.0-80.1) [non-pCR] Arm 4: Durvalumab + Dato-DXd + Pt 24-month EFS rate: 92.9% (95% CI: 59.1-99.0) [pCR] 24-month EFS rate: 69.4% (95% CI: 48.8-83.0) [non-pCR] Legend: pCR / Non-pCR Time from randomisation (months) Number of patients at risk [illegible - digits rendered vertically/stacked] Lung cancer-specific EFS events at 24 months in patients with pCR: 1/15 (6.7%) in Arm 1; 0/18 in Arm 2; 0/19 in Arm 4. Data cutoff: 25 March 2026. Analysis conducted according to investigator assessment in the mITT population. pCR (no viable tumour cells in the resected tumour specimen and sampled lymph nodes at surgery) was determined by blinded independent pathology review. EFS events, n/N (%): 20/74 (27.0) in Arm 1; 17/70 (24.3) in Arm 2; 11/54 (20.4) in Arm 4. A lung-cancer specific EFS event is any EFS event or death specifically related to lung cancer. Lung cancer-specific EFS event rates in patients without a pCR at 24 months were 15/59 (25.4%) patients in Arm 1, 11/52 (21.1%) patients in Arm 2, and 7/35 (20.0%) patients in Arm 4. Median follow-up in censored patients (range), months: 26.0 (3.0-37.0) in Arm 1; 25.8 (1.0-42.2) in Arm 2; 23.8 (2.8-29.0) in Arm 4. Survival table Endpoint: 24-month EFS rate by pCR status (investigator, mITT) Arm 1 Durvalumab + oleclumab + CT - pCR Arm 1 - non-pCR Arm 2 Durvalumab + monalizumab + CT - pCR Arm 2 - non-pCR Arm 4 Durvalumab + Dato-DXd + Pt - pCR Arm 4 - non-pCR Landmark rates: 24 mo: Arm 1 pCR 84.6% (CI 51.2-95.9) | Arm 1 non-pCR 67.0% (CI 52.6-77.9) | Arm 2 pCR 94.4% (CI 66.6-99.2) | Arm 2 non-pCR 69.1% (CI 54.0-80.1) | Arm 4 pCR 92.9% (CI 59.1-99.0) | Arm 4 non-pCR 69.4% (CI 48.8-83.0)

pCR and MPR were associated with higher ctDNA clearance rates at neoadjuvant C3D1 Evaluable patients with detected ctDNA at baseline: 59/72 (81.9%) in Arm 1; 56/69 (81.2%) in Arm 2; 45/54 (83.3%) in Arm 4. Clearance rate (%) by pCR status Arm 1 | pCR 69.2% | No pCR 45 [% partially obscured] Arm 2 | pCR 60% | No pCR 52.5% Arm 4 | pCR 81.3% | No pCR 66.7% Clearance rate (%) by MPR status Arm 1 | MPR 70.4% | No MPR 30.8% Arm 2 | MPR 66.7% | No MPR 40 [% partially obscured] Arm 4 | MPR 75% | No MPR 66.7% Arm 1: Durvalumab + oleclumab + CT (n=27/53 with ctDNA clearance at C3D1) Arm 2: Durvalumab + monalizumab + CT (n=30/55 with ctDNA clearance at C3D1) Arm 4: Durvalumab + Dato-DXd + Pt (n=29/40 with ctDNA clearance at C3D1) Lack of ctDNA clearance at C3D1 identified patients unlikely to experience a pCR. - The Negative Predictive Value of ctDNA clearance for pCR ranged from 72.7% to 84.6%. - The Positive Predictive Value of ctDNA clearance for pCR ranged from 30.0% to 44.8%. Data cutoff: 14 October 2025. The C3D1 ctDNA clearance-evaluable population comprised all randomised patients who: (a) had confirmed NSCLC histology; (b) received at least one dose of any study treatment; and (c) had an evaluable sample from the C3D1 assessment visit (result either detected [non-clearance] or undetected [clearance, if baseline sample was evaluable and ctDNA detected]) at the data cutoff. Non-clearance was defined as detected ctDNA at C3D1 irrespective of the baseline ctDNA result. Patients with undetected ctDNA at C3D1 with undetected or missing baseline samples were considered non-evaluable for clearance at C3D1. Error bars represent 95% CIs.

Both ctDNA clearance at neoadjuvant C3D1 and pCR/MPR were associated with prolonged EFS in a pooled analysis* EFS according to ctDNA clearance at C3D1 and pCR status Legend: no PCR, ctDNA clearance; no pCR, no ctDNA clearance; pCR, ctDNA clearance; pCR, no ctDNA clearance Number of patients at risk (0,5,10,15,20,25,30,35,40 mo) no PCR, ctDNA clearance | 55 | 54 | 49 | 46 | 37 | 29 | 16 | 6 | 0 no pCR, no ctDNA clearance | 49 | 37 | 30 | 29 | 23 | 20 | 12 | 5 | 1 pCR, ctDNA clearance | 31 | 30 | 28 | 26 | 24 | 17 | 5 | 0 | 0 pCR, no ctDNA clearance | 13 | 10 | 10 | 9 | 9 | 8 | 6 | 2 | 1 EFS according to ctDNA clearance at C3D1 and MPR status Legend: no MPR, ctDNA clearance; no MPR, no ctDNA clearance; MPR, ctDNA clearance; MPR, no ctDNA clearance Number of patients at risk (0,5,10,15,20,25,30,35,40 mo) no MPR, ctDNA clearance | 26 | 25 | 21 | 20 | 16 | 11 | 4 | 2 | 0 no MPR, no ctDNA clearance | 37 | 26 | 20 | 19 | 15 | 13 | 7 | 3 | 0 MPR, ctDNA clearance | 60 | 59 | 56 | 52 | 45 | 35 | 17 | 4 | 0 MPR, no ctDNA clearance | 25 | 21 | 20 | 19 | 17 | 15 | 11 | 4 | 2 The HR for experiencing an EFS event in the pooled analysis* was: - 0.10 (95% CI: 0.02-0.42) for patients with ctDNA clearance at C3D1 and pCR versus patients with neither. - 0.10 (95% CI: 0.04-0.28) for patients with ctDNA clearance at C3D1 and MPR versus patients with neither. *Pooled analyses were conducted across arms to provide adequate patient numbers for subgroup analyses, and because of the similar outcomes observed across arms. Data cutoff: 14 October 2025 for ctDNA data; 25 March 2026 for EFS data. Analysis conducted according to investigator assessment in the C3D1 ctDNA clearance-evaluable population. HRs calculated using Cox proportional hazards regression. EFS events, n/N (%): 20/74 (27.0) in Arm 1; 17/70 (24.3) in Arm 2; 11/54 (20.4) in Arm 4. Median follow-up in censored patients (range), months: 26.0 (3.0-37.0) in Arm 1; 25.8 (1.0-42.2) in Arm 2; 23.8 (2.8-29.0) in Arm 4. Survival table Endpoint: EFS by ctDNA clearance at C3D1 and pCR / MPR status (pooled across arms) no pCR, ctDNA clearance | 55 no pCR, no ctDNA clearance | 49 pCR, ctDNA clearance | 31 pCR, no ctDNA clearance | 13 no MPR, ctDNA clearance | 26 no MPR, no ctDNA clearance | 37 MPR, ctDNA clearance | 60 MPR, no ctDNA clearance | 25 HR: 0.10 (ctDNA clearance + pCR vs neither); 0.10 (ctDNA clearance + MPR vs neither) HR CI: 0.02-0.42 (pCR); 0.04-0.28 (MPR)

[Slide 1] Study Design: HS-10241-301 is a randomized, open-label, multicenter, phase 3 study. Key Eligibility Criteria • Aged ≥18 years • Advanced NSCLC with EGFRm+ and c-MET+ • ECOG PS 0–1 • ≥1 measurable target lesion (RECIST v1.1) • Objective disease progression(PD) or intolerance after EGFR-TKIs c-MET+ definition: MET copy number ≥5 or MET/CEP7 ratio ≥2 (FISH); confirmed by central lab Treatment R 1:1 TKI-combo group (N=156): Dalmelitinib 300 mg BID + Aumolertinib 110 mg QD Chemo group (N=158): (cisplatin 75 mg/m2 or carboplatin AUC 5) + pemetrexed 500 mg/m2 for 4-6 cycles, then pemetrexed as maintenance treatment Treatment until PD, intolerable toxicity or any other reason for discontinuation Conditionally switched to TKI-combo after BICR-assessed PD Random stratification factors: ✓Prior 3rd-generation EGFR TKIs (yes or no) ✓Baseline brain metastasis (yes or no) ✓MET/CEP7 ratio (≥2 or <2) Endpoints Primary endpoint: • BICR-PFS Secondary endpoints: • BICR-ORR • BICR-DCR • BICR-DoR • OS • Adverse events The data cut-off date was December 10, 2025. The study analysis was generally conducted based on FAS, except for the safety analysis in SS. • Full Analysis Set (FAS): all randomized subjects. • Safety Analysis Set (SS): all randomized subjects who receive ≥1 dose of study drugs. • Final analysis was pre-planned at 197 events of PFS. • Pre-planned OS sensitivity analyses adjusting for treatment switching will be performed when >50% OS rate is observed. Prof. Ziming Li Abbreviation: AUC, Area Under Curve; BICR, blinded independent central review; BID, twice a day; CEP, C-terminally encoded peptide; DCR, disease control rate; DoR, duration of response; ECOG PS, Eastern Cooperative Oncology Group Performance Status; EGFR, Epidermal Growth Factor Receptor; MET, Mesenchymal-Epithelial Transition factor; NSCLC, non small cell lung cancer; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; QD, once a day; TKI, Tyrosine Kinase Inhibitor [Slide 2] Final analysis met the primary endpoint- PFS assessed by BICR • TKI-combo treatment significantly improved PFS over chemotherapy with 40% lower risk of disease progression or death (mPFS 6.2 vs. 4.4 months, HR=0.601, P <0.001). TKI-combo N=156 | Chemo N=158 Median follow-up time, months 11.4 | 10.7 Events, n (%) 92 (59.0) | 105 (66.5) Median PFS, months 6.2 | 4.4 Hazard Ratio (95% CI) 0.601 (0.448, 0.805) P value <0.001 Probability of progression-free survival (%) Months TKI-combo Chemo Prof. Ziming Li Data cutoff December 10, 2025 Abbreviation: BICR, blinded independent central review; CI, confidence interval; PFS, progression-free survival; TKI, Tyrosine Kinase Inhibitor Survival table Endpoint: PFS by BICR (primary) TKI-combo (dalmelitinib + aumolertinib) | 156 | 6.2 mo Chemo | 158 | 4.4 mo HR: 0.601 HR CI: 0.448, 0.805 P: <0.001 [Slide 3] Secondary endpoints • Clinical benefit was consistent across secondary efficacy endpoints, including ORR and DoR. • OS data was still evolving with overall event rate of 22.9%; 82 (51.9%) patients in the chemo group received subsequent MET inhibitors (including 74 patients receiving study crossover treatment and 8 patients receiving other MET inhibitors). TKI-combo group (N=156) | Chemo group (N=158) mOS, months (95% CI) NR (17.5, NE) | 19.6 (14.6, NE) Unstratified HR (95% CI) 0.840 (0.524, 1.346) Probability of overall survival(%) Months TKI-combo Chemo Data cutoff December 10, 2025;Median follow-up was:11.2 months; Secondary endpoints by BICR | TKI-combo group (N=156) | Chemo group (N=158) ORR, % (95% CI) | 59.6 (51.5, 67.4) | 38.6 (31.0, 46.7) mDoR, months (95% CI) | 9.7 (7.0, 14.0) | 4.4 (3.0, 8.4) DCR, % (95% CI) | 84.6 (78.0, 89.9) | 74.1 (66.5, 80.7) Prof. Ziming Li Abbreviation: BICR, blind independent central review; DCR, disease control rate; DoR, duration of response; HR, hazard ratio; m, median; NE, not estimatable; NR, not reached; ORR, objective response rate; OS, overall survival; TKI, Tyrosine Kinase Inhibitor Survival table Endpoint: OS (secondary, immature) Arm | N | Median | Median CI TKI-combo | 156 | NR | 17.5, NE Chemo | 158 | 19.6 mo | 14.6, NE HR: 0.840 (unstratified) HR CI: 0.524, 1.346 [Slide 4] Safety Summary • Similar incidence and severity of TRAEs between two groups, with different safety profile. • Common TRAEs aligned with the expected MET/EGFR-TKI safety profile, mainly elevated liver enzymes, and most were grade 1-2. After hepatoprotective therapy and dose modification or discontinuation, most events resolved to baseline or normal, supporting a tolerable and reversible hepatic safety profile. | TKI-combo group (N=156) | Chemo group (N=145) Any TEAEs | 154 (98.7) | 142 (97.9) Any TRAEs | 154 (98.7) | 139 (95.9) Grade ≥3 TRAE | 73 (46.8) | 65 (44.8) TRAEs Leading to Dose Reduction | 43 (27.6) | 32 (22.1) TRAEs Leading to Discontinuation | 11 (7.1) | 4 (2.8) TRAEs Leading to Death | 1 (0.6) | 2 (1.4) Prof. Ziming Li Abbreviation: EGFR, Epidermal Growth Factor Receptor; MET, Mesenchymal-Epithelial Transition factor; TEAE, treatment-emergent adverse event; TKI, Tyrosine Kinase Inhibitor; TRAE, treatment related adverse event

TRAEs in ≥15% of Patients in Either Group Legend: TKI-combo any grade | TKI-combo grade≥3 | Chemo any grade | Chemo grade≥3 (Patients %) AE | TKI-combo any grade | TKI-combo grade≥3 | Chemo grade≥3 | Chemo any grade Any TRAEs | 98.7 | 46.8 | 44.8 | 95.9 Alanine aminotransferase increased | 62.8 | 12.8 | 3.4 | 38.6 Aspartate aminotransferase increased | 60.3 | 9.6 | 1.4 | 39.3 Gamma-glutamyl transferase increased | 44.2 | 15.4 | 1.4 | 15.9 Creatine phosphokinase increased | 41.0 | 5.8 | – | – Hypoalbuminemia | 36.5 | – | – | 11.7 Anemia | 34.0 | 1.9 | 17.2 | 78.6 Nausea | 29.5 | 0.6 | 0.7 | 30.3 Platelet count decreased | 25.6 | 1.9 | 16.6 | 40.0 Oedema peripheral | 25.6 | 1.9 | – | 4.8 Decreased appetite | 25.6 | 0.6 | 2.1 | 25.5 Blood alkaline phosphatase increased | 23.1 | – | 0.7 | 9.0 Blood creatinine increased | 22.4 | 0.6 | – | 10.3 Vomiting | 21.2 | 1.3 | 0.7 | 20.0 Hypocalcaemia | 19.9 | 0.6 | – | 4.1 Blood lactate dehydrogenase increased | 17.9 | – | – | 13.8 Hypokalaemia | 16.7 | 0.6 | 0.7 | 6.2 Asthenia | 15.4 | 1.3 | 2.8 | 24.1 Hypercholesterolaemia | 14.1 | – | – | 17.2 White blood cell count decreased | 11.5 | – | 17.9 | 60.0 Neutrophil count decreased | 8.3 | 1.3 | 23.4 | 53.8 Hypertriglyceridaemia | 3.8 | 0.6 | – | 15.2 Prof. Ziming Li Abbreviation: TEAE, treatment-emergent adverse event; TKI, Tyrosine Kinase Inhibitor; TRAE, treatment related adverse event

My takeaways on Abstract 2615O Dalmelitinib/Aumolertinib: EGFR + MET inhibition is a proven treatment strategy in MET+ acquired resistance With Dalmelitinib + Aumolertinib, Lu et al add to the Phase III data supporting this strategy · Osimertinib + Savolitinib (SACHI, SAFFRON) · Aumolertinib + Dalmelitinib MET at time of progression is an actionable biomarker & requires testing

[Slide 1] Conclusion In this randomized, open-label, multicenter, phase 3 study, • Dalmelitinib plus Aumolertinib demonstrated superior clinical benefit over platinum-based chemotherapy in patients with EGFR mutation-positive and MET-amplified (GCN ≥5 or MET/CEP7 ≥2) NSCLC that progressed on EGFR-TKI therapy. -mPFS 6.2 vs. 4.4 months, HR=0.601, P<0.001; ORR 59.6% vs. 38.6% • The safety profile of this combination was manageable with no new safety signals identified. -TRAEs were predominantly grade 1-2. -Common grade ≥3 TRAEs (incidence ≥10%) included increased gamma-glutamyltransferase and alanine aminotransferase, and most of them were tolerable and reversible. This study underscores Dalmelitinib plus Aumolertinib as a promising new chemo-free treatment option for NSCLC patients with EGFR-mutation and MET-amplification (GCN ≥5 or MET/CEP7 ≥2) after EGFR-TKI failure. Prof. Ziming Li [Slide 3] Subgroup Analysis of PFS by BICR The direction of PFS benefit was consistent across all pre-defined subgroups, irrespective of prior EGFR-TKI generation and MET gene copy number. Subgroup | No. of events/total TKI-combo | Chemo | Unstratified Hazard Ratio (95% CI) FAS | 92/156 | 105/158 | 0.568 (0.426, 0.759) Sex Male | 29/50 | 43/56 | 0.445 (0.271, 0.733) Female | 63/106 | 62/102 | 0.655 (0.458, 0.938) Age <65 | 59/89 | 60/86 | 0.711 (0.492, 1.028) ≥65 | 33/67 | 45/72 | 0.414 (0.260, 0.659) Smoking status Current or former smoker | 21/36 | 35/49 | 0.474 (0.270, 0.834) Never smoker | 71/120 | 70/109 | 0.596 (0.425, 0.838) EGFR mutation subtype Ex19del | 47/76 | 57/74 | 0.516 (0.346, 0.769) L858R | 45/80 | 48/84 | 0.635 (0.418, 0.964) Brain metastases Yes | 28/35 | 31/38 | 0.647 (0.384, 1.089) No | 64/121 | 74/120 | 0.545 (0.386, 0.771) Prior 3rd-Generation EGFR TKIs Yes | 88/148 | 97/148 | 0.607 (0.451, 0.815) No | 4/8 | 8/10 | 0.180 (0.037, 0.861) MET/CEP7 ratio ≥2 | 37/65 | 42/64 | 0.457 (0.287, 0.728) <2 | 55/91 | 63/94 | 0.657 (0.453, 0.953) MET copy number ≥10 | 32/53 | 34/52 | 0.485 (0.290, 0.811) ≥5 to <10 | 60/103 | 71/106 | 0.618 (0.435, 0.878) ← TKI-combo favors Chemo → Prof. Ziming Li Abbreviation: BICR, blind independent central review; CEP, C-terminally encoded peptide; CI, confidence interval; EGFR, Epidermal Growth Factor Receptor; GCN, gene copy number; HR, hazard ratio; IRC, independent review committee; m, median; MET, Mesenchymal-Epithelial Transition factor; PFS, progression-free survival; TKI, Tyrosine Kinase Inhibitor

[Slide 2] Study Design ▪ Phase 2, non-comparative, randomized, 3 parallel arms study ▪ Primary estimand: 1-Year Survival Rate in all randomized patients KEY ELIGIBILITY CRITERIA ➤ NSCLC relapsed after ≥ 4 cycles of 1L chemoimmunotherapy ➤ HLA-A2 positive ➤ EGFR, ALK, ROS Wild Type ➤ Patients with primary resistance to immunotherapy are excluded R ARM A (n=36) Docetaxel IV q3w X 6 cycles + OSE2101 SC q3w X 6 cycles → OSE2101 q6w from C7 until Year 1, then q12w until PD, unacceptable toxicity or patient refusal ARM B (n=35) Nivolumab IV q3w X 6 cycles + OSE2101 SC q3w X 6 cycles → Nivo q3w + OSE2101 q6w from C7 until Year 1, then q12w until PD, unacceptable toxicity or patient refusal ARM C (n=35) Docetaxel IV q3w X 6 cycles → Up to 6 cycles or until PD, unacceptable toxicity or patient refusal Docetaxel was administred at 75mg/m2 in ARM C and 60 mg/m2 in ARM A STATISTICAL HYPOTHESIS 1. To test in each experimental arm the 1-year OS rate using 5% one-sided exact binomial test whether 1-year OS exceeded the predefined 20% threshold: if OS Rate ≤ 20%: treatment does not warrant further investigation If OS Rate ≥40%: treatment has sufficient activity 2. If primary analysis positive, Bayesian analyses to assess the posterior probability to improve 1-year OS rate of • Arm A vs Arm C • Arm B vs Arm C F. Cappuzzo, Rome, Italy [Slide 3] Overall Survival Median OS in months [90% CI] | A (OSE2101+Docetaxel) 11.9 [8.9;15.3] | B (OSE2101+Nivolumab) 11.3 [6.1;13.3] | C (Docetaxel) 6.4 [5.2;9.0] Legend: A (N= 36), B (N= 35), C (N= 35) 1-year OS 47% / 48% ; 31% Months since randomization Number of subjects at risk (0, 6, 12, 18, 24, 30 months) A 36 27 17 9 3 1 B 35 22 16 9 7 3 C 35 17 10 9 6 2 Subsequent systemic anticancer treatment was administrated in 66% in Arm A OSE2101+Docetaxel, 53% in Arm B OSE2101+Nivolumab and 41% in Arm C Docetaxel Median FU 27 months; Snapshot 01OCT2026 – Randomized population F. Cappuzzo, Rome, Italy Survival table Endpoint: Overall survival Arm | N | Median | Median CI A (OSE2101+Docetaxel) | 36 | 11.9 mo | 8.9;15.3 (90% CI) B (OSE2101+Nivolumab) | 35 | 11.3 mo | 6.1;13.3 (90% CI) C (Docetaxel) | 35 | 6.4 mo | 5.2;9.0 (90% CI) Landmark rates: 12 months: A (OSE2101+Docetaxel) 47% | B (OSE2101+Nivolumab) 48% | C (Docetaxel) 31% [Slide 4] Conclusion ▪ 1-year OS rate was improved with the IO-IO combination OSE2101 + NIVOLUMAB, and similarly with the IO-CT combination OSE2101 + DOCETAXEL in metastatic NSCLC patients progressing after first-line chemo-immunotherapy and no primary resistance to immunotherapy ▪ Both combinations showed a manageable safety profile without increasing toxicity ▪ OSE2101- based combinations showed improved survival compared with docetaxel, supporting further evaluation of vaccine-based immunotherapy strategies in this setting ▪ A phase 3 registrational study of OSE2101 monotherapy compared to docetaxel (ARTEMIA) is currently recruiting NSCLC patients with secondary resistance to immunotherapy F. Cappuzzo, Rome, Italy FoRT Fondazione Ricerca Traslazionale

[Slide 1] Treatment Options in the Post-"Chemo-ICI" Setting Resistance post-ICI and a platinum-based doublet (sequential or combined) Docetaxel | Taxanes + antiangiogenic drugs | Targeted therapies | Antibodies drug-conjugated | Immunotherapy-based strategies Targeted therapies: KRASG12C; MET exon 14; HER2 mutation; NRG1 fusion; MTAP deletion; ... Antibodies drug-conjugated: Biomarker agnostic; Biomarker driven Immunotherapy-based strategies: Checkpoints inhibition; Bispecific antibodies; Targeting TME (VEGF); T-cells engagers; Vaccines; Adoptive cell therapies Docetaxel: TAX 317; TAX 320 Taxanes + antiangiogenic drugs: REVEL; LUME-1; ULTIMATE Targeted therapies: KRASG12C inhibitors; MET TKIs; HER2 TKIs; Zenocutuzumab; PRMT5 inhibitors; ... ADCs: TROP2; HER2; HER3; MET; HER2; ... Immunotherapy: Multiple approaches and trials ... Maurice Pérol [Slide 2] How to Interpret the Combi-TED Trial Results? Strengths - First randomized signal supporting OSE2101-based combinations after chemo-immunotherapy failure - Strong biological rationale - Consistent OS improvement across both experimental arms - Favorable safety profile (toxicity driven by docetaxel exposure) Key limitations - Small phase II sample size with wide CI - Underperforming control arm may have exaggerated the observed OS benefit - 1-year OS rates in the experimental arms within the range reported with docetaxel in recent studies - No improvement in ORR/PFS despite OS signal - Results remain hypothesis-generating and require phase III confirmation Maurice Pérol [Slide 3] Limitations and Challenges of TIL Therapy in NSCLC - Manufacturing time - Need for surgical excision of a lesion >= 1.5 cm - 4-6 weeks may be too long for rapidly progressing patients - Toxicity in NSCLC patients - Cyclophosphamide/fludarabine conditioning (cytopenia, infection risk) - High-dose IL-2 support (capillary leak syndrome ...) - Patient selection: still unknown - Which genomic subsets benefit most - Whether oncogene-driven tumours respond similarly - Which TIL phenotypes are optimal - T-cell fitness and persistence - Prolonged ex vivo culture may produce terminally differentiated or exhausted T cells - Failure of TIL therapy associated with inadequate persistence of neoantigen-specific T cells - Cost and accessibility Figure: Cumulative frequency of Infused TIL clonotypes (%); Durable benefit / No benefit; TIL infusion; Wk -9, Wk -5, Wk -1, Wk 0, Wk 1, Wk 2, Wk 4, Wk 8-10, Wk 14-17, Wk 20-35 Wang et al., Nat Cancer 2025 Maurice Pérol [Slide 4] Phase I study of TIL Therapy in Combination with Chemoimmunotherapy in >=2nd Line Advanced NSCLC Patients [check] Feasible and better tolerated than conventional TIL approaches [check] Potentially more scalable for routine clinical practice [warning] Antitumor activity remains modest and requires further validation [warning] Translational analyses are needed to clarify the contribution of TIL therapy [?] Can a simplified TIL platform expand access without sacrificing efficacy? [rocket] The answer may come too late given the emergence of next-generation engineered TILs Maurice Pérol

[Slide 2] Patient Demographics & Characteristics | | ARM A DOCE + OSE2101 N=36 | ARM B NIVO + OSE2101 N=35 | ARM C DOCE N=35 Age (years) | Median (range) | 67 (44 - 79) | 67 (38 – 81) | 63 (50 – 83) Gender (N, %) | Male | 21 (58%) | 21 (60%) | 30 (86%) Race (N,%) | White/Caucasian | 34 (97%) | 34 (97%) | 34 (97%) HLA-A2 status (N,%) | Positive | 36 (100%) | 35 (100%) | 35 (100%) Histology (N, %) | Adenocarcinoma | 27 (77%) | 28 (80%) | 26 (77%) | Squamous | 8 (23%) | 7 (20%) | 8 (23%) Smoking Status (N, %) | Never Smoked | 5/35 (14%) | 3/35 (9%) | 2/34 (6%) ECOG PS (N, %) | Grade 0 | 19 (53%) | 17 (48.5%) | 17 (49%) | Grade 1 | 17 (47%) | 17 (48.5%) | 18 (51%) | Grade 2 | 0 (0%) | 1 (3%) | 0 (0%) PDL-1 status | Positive (≥1%) | 21/35 (60%) | 19/33 (58%) | 19/34 (56%) Snapshot 01OCT2026 – Randomized Population F. Cappuzzo, Rome, Italy [Slide 3] Primary estimand 1-Year Survival Rate | ARM A DOCE + OSE2101 N=36 | ARM B NIVO + OSE2101 N=35 | ARM C DOCE N=35 1-Year OS Rate (n, %) | 17 (47%) | 16 (46%) | 10 (29%) 90% CI # | [33%;62%] | [31%;61%] | [16%;44%] One-sided p-value* | <.001* | <.001* | 0.15* 4 patients withdrew their consent before the 1-year timepoint and therefore have insufficient follow-up * one-sided exact binomial test evaluating the following hypotheses at a 5% one-sided significance level: H0: P ≤ 20% # 90% CI corresponding to 5% one-sided Bayesian analyses to assess the posterior probability to improve 1-year OS rate between experimental arms vs Arm C ARM A DOCE + OSE2101 vs ARM C | ARM B NIVO + OSE2101 vs ARM C 94.6% | 92.8% Based uninformative prior Beta(1, 1) and Markov Chains Monte Carlo (MCMC) algorithm If the predictive probability that the 1-year survival rate difference is > 95% then the arm of interest can be considered as promising. Snapshot 01OCT2026 – Randomized population F. Cappuzzo, Rome, Italy

PFS and Best Response | A(OSE2101+Docetaxel) | B(OSE2101+Nivolumab) | C(Docetaxel) Median PFS in months [90%CI] | 4.8 [2.4;5.7] | 2.7 [2.1;4.1] | 2.3 [1.9;4.1] Legend: A (N= 36) | B (N= 35) | C (N= 35) Number of subjects at risk (Months since randomization 0, 3, 6, 9) A 36 21 9 4 B 35 13 8 6 C 35 13 5 2 Best Overall Response | A OSE2101 + Docetaxel) (N=36) | B OSE2101 + Nivolumab (N=35) | C Docetaxel (N=35) n* | 30 | 28 | 20 Partial Response | 6 (20%) | 0 | 2 (10%) Stable Disease | 15 (50%) | 13 (46%) | 7 (35%) Progressive Disease | 9 (30%) | 15 (54%) | 11 (55%) *Patients with RECIST 1.1 evaluation F. Cappuzzo, Rome, Italy Snapshot 01OCT2026 – Randomized population Survival table Endpoint: Progression-free survival (randomized population) Arm | N | Median | Median CI A (OSE2101+Docetaxel) | 36 | 4.8 | 90% CI 2.4;5.7 B (OSE2101+Nivolumab) | 35 | 2.7 | 90% CI 2.1;4.1 C (Docetaxel) | 35 | 2.3 | 90% CI 1.9;4.1

[Slide 1] Background • In the phase 3 KEYNOTE-671 study, perioperative pem[illegible] improved EFS, OS, pCR, and MPR vs neoadjuvant che[illegible] NSCLC1,2 EFS, IA11 Median follow-up time: 25.2 mo Data cutoff: July 29, 2022 Pembro arm Placebo arm HR, 0.58 (95% CI, 0.46–0.72; P < 0.001) • Survival benefits of perioperative pembrolizumab we[illegible] • We report an exploratory biomarker analysis of the [illegible] (TcellinfGEP)a and tumor mutational burden (TMB) w[illegible] TcellinfGEP, TMB, and KRAS mutation status aTcellinfGEP is an 18-gene IFNγ-responsive expression signature that measures IFNγ-driven [illegible] 1Wakelee H et al. N Engl J Med 2023;389:491–503. 2Spicer JD et al. Lancet 2024;404:1240–[illegible] Invest. 2017;127(8):2930-2940. [Slide 2] EFS by TcellinfGEP Status (n = 387) TcellinfGEP Low (<1st Tertile) TcellinfGEP High (≥1st Tertile) Event-Free Survival, % Time, mo No. at risk Pembro Placebo Median (range) time from randomization to data cutoff (July 3, 2025) was 63.6 (42.6–85.8) mo for the RNAseq population and 62.4 (42.6–85.8) mo for the WES population. Survival table Endpoint: EFS by TcellinfGEP status (n = 387), exploratory Arm | N | Median | Median CI TcellinfGEP Low: Pembro | 74 | 29.3 mo | 14.3–69.5 TcellinfGEP Low: Placebo | 55 | 18.2 mo | 10.8–22.2 TcellinfGEP High: Pembro | 124 | NR | 39.4–NR TcellinfGEP High: Placebo | 134 | 20.6 mo | 14.5–34.1 HR: Low: 0.71; High: 0.53 HR CI: Low: 0.46–1.11; High: 0.38–0.75 [Slide 3] OS by TcellinfGEP Status (n = 387) TcellinfGEP Low (<1st Tertile) TcellinfGEP High (≥1st Tertile) Overall Survival, % Time, mo No. at risk Pembro Placebo aOS data are immature; HRs for subgroups are exploratory/underpowered and should be interpreted with caution. Median (range) time from randomization to data cutoff (July 3, 2025) was 63.6 (42.6–85.8) mo for the RNAseq population and 62.4 (42.6–85.8) mo for the WES population. Survival table Endpoint: OS by TcellinfGEP status (n = 387), exploratory; OS data immature Arm | N | Median | Median CI TcellinfGEP Low: Pembro | 74 | NR | 47.4–NR TcellinfGEP Low: Placebo | 55 | 41.6 mo | 28.8–NR TcellinfGEP High: Pembro | 124 | NR | 72.3–NR TcellinfGEP High: Placebo | 134 | NR | 53.7–NR HR: Low: 0.59; High: 0.82 HR CI: Low: 0.36–0.97; High: 0.55–1.22 [Slide 4] Conclusions • The benefits of perioperative pembrolizumab plus neoadjuvant chemotherapy vs neoadjuvant chemotherapy were generally consistent regardless of immunotherapy-relevant biomarkers • For both treatment arms, positive associations were observed between baseline TcellinfGEP and TMB and certain efficacy outcomes – TcellinfGEP was positively associated with EFS and MPR for pembrolizumab and with OS for placebo – TMB was positively associated with EFS, OS, pCR, and MPR for pembrolizumab and with pCR and MPR for placebo • EFS and OS benefit was generally observed with pembrolizumab vs placebo regardless of biomarker subgroup, including TcellinfGEP high vs low, TMB high vs low, and KRAS mutation vs wild-type • As these findings were exploratory with an ~50% biomarker-evaluable subset and regional imbalance in the RNAseq population, results should be interpreted with caution • These findings support the use of perioperative pembrolizumab plus neoadjuvant chemotherapy as a standard of care for patients with resectable, early-stage NSCLC

EFS by TMB Status (n = 459) TMB Low (<175 mut/exome) Pts w/ Event | Median (95% CI), mo | HR (95% CI) Pembro (n = 132): 53.0% | 39.4 (18.4–NR) | 0.63 (0.47–0.86) Placebo (n = 134): 73.1% | 17.9 (13.8–21.1) TMB High (≥175 mut/exome) Pembro (n = 97): 38.1% | 69.5 (63.7–NR) | 0.49 (0.32–0.73) Placebo (n = 96): 63.5% | 25.1 (14.5–38.2) Event-Free Survival, % Time, mo No. at risk Median (range) time from randomization to data cutoff (July 3, 2025) was 63.6 (42.6–85.8) mo for the RNAseq population and 62.4 (42.6–85.8) mo for the WES population. Survival table Endpoint: EFS by TMB status Arm | N | Median | Median CI TMB Low - Pembro | 132 | 39.4 | 18.4–NR TMB Low - Placebo | 134 | 17.9 | 13.8–21.1 TMB High - Pembro | 97 | 69.5 | 63.7–NR TMB High - Placebo | 96 | 25.1 | 14.5–38.2 HR: Low 0.63; High 0.49 HR CI: Low 0.47–0.86; High 0.32–0.73

#7975 Neoadjuvant Sacituzumab Govitecan Plus Pembrolizumab in Resectable Stage II-III Non-Small Cell Lung Cancer: Phase II CITADEL Study Molly SC Li, Joyce WY Chan, Olivia H Chen, Alvin HK Cheung, Aliss TC Chang, Kevin KS Mok, Florence ST Mok, Jonathan SF Nyaw, S Chan, Teddy HY Wong, Victor WK Lee, Fiona MY Lim, KK Li, Gary JW Liao, James Chow, Ivan WC Wong, Innes YP Wan, Rainbow WH Lau, Calvin SH Ng, Tony SK Mok MADRID 2026 ESMO congress CU Medicine Faculty of Medicine The Chinese University of Hong Kong Background • Perioperative chemoimmunotherapy improves outcomes in resectable non-small cell lung cancer (NSCLC) without EGFR/ALK alterations, with reported pathological complete response between 17–41%. • The combination of sacituzumab govitecan and pembrolizumab has demonstrated activity in advanced NSCLC, supporting neoadjuvant evaluation. Study design Investigator-initiated, open-label, multicentre, single-arm phase II trial NCT06055465 | Enrolment: 22 May 2024 – 17 April 2026 Population: Resectable stage II–III NSCLC (AJCC 8th edition) EGFR / ALK negative; potentially resectable Neoadjuvant treatment: Sacituzumab govitecan 10 mg/kg, days 1 and 8 + pembrolizumab 200 mg, day 1 Every 3 weeks × 4 cycles. Tumour imaging after neoadjuvant therapy Surgery: Resection 4–8 weeks after treatment Adjuvant treatment: Pembrolizumab 200 mg, day 1 Every 3 weeks × 13 cycles Endpoints Primary endpoint • Pathological complete response (pCR) in the intention-to-treat (ITT) population Secondary endpoints • pCR in resected patients • Major pathological response (MPR): ITT and resected patients • Resection rate after neoadjuvant treatment; safety • Objective response rate (ORR) • 12-month and 24-month event-free survival (EFS) and overall survival (OS) Statistical considerations Hypothesized pCR: 38% (20 percentage-point improvement over KEYNOTE-671) Sample size: 37; dropout: 5%; one-sided α = 0.05; power: 80% Data cutoff: 14 August 2026 Median follow-up: 11.4 months (range, 2.6–21.4) Table 1. Baseline characteristics (N=36 *) Age, median (range), years 69 (46–78) Sex Female 7 (19.4%); Male 29 (80.6%) Smoking Never-smoker 12 (33.3%); Ever-smoker 24 (66.7%) ECOG performance status 0: 30 (83.3%); 1: 6 (16.7%) Histology Adenocarcinoma 25 (69.4%); Squamous cell carcinoma 6 (16.7%); Lymphoepithelioma-like carcinoma 3 (8.3%); NSCLC, not otherwise specified 1 (2.8%); Adenosquamous carcinoma 1 (2.8%) Clinical nodal stage cN0 11 (30.6%); cN1 8 (22.2%); cN2 17 (47.2%) Clinical stage Stage II 8 (22.2%); Stage IIIA 16 (44.4%); Stage IIIB 12 (33.3%) Driver alteration KRAS mutation 10 (27.8%); HER2 exon 20 insertion 2 (5.6%); None 19 (52.8%); Unknown 5 (13.9%) PD-L1 TPS <1% 8 (22.2%); 1–49% 13 (36.1%); ≥ 50% 13 (36.1%); Unknown 2 (5.6%) * 37 patients enrolled, 1 patient was excluded after enrolment due to a finding of benign histology. Table 2. Treatment delivery Neoadjuvant treatment N = 36: 4 cycles 33 (91.7%); 3 cycles 3 (8.3%) Surgery N = 36: Resection performed 30 (83.3%); No surgery 6 (16.7%); Disease progression 3 (8.3%); Inadequate lung function 2 (5.6%); Declined surgery 1 (2.8%) Adjuvant treatment N = 30 resected: Started 29 (96.7%); Not yet started 1 (3.3%) Status among patients who started N = 29: Completed 10 (34.5%); Ongoing 12 (41.4%); Discontinued 7 (24.1%); Disease progression 5 (17.2%); Adverse event 1 (3.4%); Death unrelated to treatment 1 (3.4%) Figure 1. Pathological response pCR (ITT) 11.1% 4/36 95% CI 3.1–26.1% MPR (ITT) 33.3% 12/36 95% CI 18.6–51.0% pCR (Resected) 13.3% 4/30 95% CI 3.8–30.7% MPR (Resected) 40.0% 12/30 95% CI 22.7–59.4% MPR threshold (≤ 10% residual viable tumour) • Radiological response rate was 63.9% (23/36). • Nodal downstaging and nodal clearance were observed in 81.0% (17/21) and 76.2% (16/21) of cases, respectively. • In exploratory analysis, MPR was more frequently observed among ever-smokers (55.0%, 11/20 versus 10.0%, 1/10 among never-smokers; p=0.024) and tumours with PD-L1 TPS ≥ 50% (72.7%, 8/11 versus 21.1%, 4/19 with PD-L1 TPS <50% or unknown; p=0.009). • Sex, histology, stage, and presence of KRAS mutation were not significantly associated with MPR. Figure 2. Adverse events Neoadjuvant phase (N = 36) | Adjuvant phase (N = 29) Overall safety (N = 36): Any-grade TEAEs 100.0%; Grade ≥ 3 TEAEs 41.7%; Any-grade TRAEs 100.0%; Grade ≥ 3 TRAEs 27.8%; SAEs 36.1% No grade 4-5 TRAEs recorded Conclusion • The combination of neoadjuvant sacituzumab govitecan and pembrolizumab did not improve pCR compared with standard perioperative immunochemotherapy. • No additional safety signal was identified. This is an investigator-initiated trial supported by Gilead and Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA

[Slide 2] Background and Methods ■ Patients with KRAS G12C-mutant NSCLC generally have high-risk clinical characteristics and more comorbidities.1 ■ There are no approved targeted therapies for the 1L treatment of KRAS G12C-mutant NSCLC. Toxicity has been a persistent challenge for some KRAS G12C inhibitor-immunotherapy combinations, limiting their development in the 1L setting.2 ■ Olomorasib, a potent, selective next-generation KRAS G12C inhibitor with intracranial activity, can be combined with immunotherapy with or without chemotherapy and previously showed promising 1L efficacy and durability across PD-L1 levels.1,3-6 ■ Here, with an additional 9 months of follow-up, we report updated results in 1L patients from the integrated analysis of LOXO-RAS-20001 and the safety lead-in for SUNRAY-01. Olomorasib + pembrolizumab + pemetrexed + platinum(a) LOXO-RAS-20001: Cohort B9: 1L NSCLC (n=37) SUNRAY-01: Safety Lead-in: 1L NSCLC (n=40) Eligibility ▪ Confirmed stage IIIB-IIIC or stage IV non-squamous KRAS G12C-mutant NSCLC ▪ No prior therapy for advanced or metastatic NSCLC – Patients may have received up to one cycle of SOC prior to enrollment (defined as one cycle of either pemetrexed-platinum +/- pembrolizumab, or pembrolizumab monotherapy) ▪ PD-L1 expression 0-100%(b) ▪ ECOG PS of 0 or 1 ▪ Small (≤1.5 cm)(c), asymptomatic untreated or treated stable brain metastases Key Objectives ▪ Safety, tolerability, and pharmacokinetics ▪ ORR, DOR, DCR, and PFS per RECIST v1.1 References: 1. Peters S, et al. Future Oncol. 2026;22(9):1073–1082; 2. Li BT, et al. J Thorac Oncol 2022; 17(9, Suppl.): S10–S11. 3. Burns TF, et al. J Thorac Oncol. 2026;21(4):103528; 4. Murciano-Goroff YR, et al. Nat Commun. 2026;17(1):3834; 5. Johnson ML, et al. J Thorac Oncol. 2025;20(10 suppl 1):S59 [abstr MA02.06]; 6. Negrao MV, et al. J Thorac Oncol. 2025;20(10 suppl 1):S26–S27 [abstr OA08.02]. (a)Doses: Olomorasib 50 or 100 mg BID; Pembrolizumab 200 mg IV Q3W; Pemetrexed 500 mg/m2, Q3W, Platinum - cisplatin 75 mg/m2 or carboplatin AUC 5 mg/mL/min, Q3W (max dose = 750 mg). (b)PD-L1 status was assessed using various clones to determine PD-L1 expression locally, and the antibody clone used for testing was at the discretion of the site. (c)Small defined as ≤1.5 cm in SUNRAY-01 only. Abbreviations: 1L, first-line; BID, twice daily; DCR, disease control rate; DOR, duration of response; ECOG PS, Eastern Cooperative Oncology Group performance status; IV, intravenous; NSCLC, non-small cell lung cancer; ORR, overall response rate; PD-L1, programmed death-ligand 1; PFS, progression-free survival; Q3W, every three weeks; RECIST, Response Evaluation Criteria in Solid Tumors; SOC, standard of care. LOXO-RAS-20001: NCT04956640, SUNRAY-01: NCT06119581. Dr. Marcelo V. Negrao [Slide 3] Efficacy of 1L Olomorasib + Chemoimmunotherapy Olomorasib + chemoimmunotherapy Outcomes | All patients PD-L1 0-100% (N=77)(a) | PD-L1 <1% (n=26) | PD-L1 1-49% (n=31) | PD-L1 ≥50% (n=18) cORR(b), % (n/N) | 62 (48/77) | 54 (14/26) | 68 (21/31) | 67 (12/18) BOR, n (%) | cPR | 48 (62) | 14 (54) | 21 (68) | 12 (67) SD | 21 (27) | 8 (31) | 8 (26) | 5 (28) PD | 3 (4) | 2 (8) | 0 | 0 NE | 5 (6) | 2 (8) | 2 (6) | 1 (6) DCR, % (n/N) | 90 (69/77) | 85 (22/26) | 94 (29/31) | 94 (17/18) mDOR(b), mo (95% CI) | 14.5 (8.5–NE) | 10.5 (5.4–NE) | NE (8.5–NE) | NE (6.1–NE) 12-mo DOR rate, % (95% CI) | 55.7 (39.5–69.1) | 49.0 (21.6–71.7) | 60.3 (34.0–78.9) | 54.5 (22.9–78.0) mPFS(b), mo (95% CI) | 11.6 (9.5–21.5) | 9.7 (5.9–15.9) | 12.1 (5.9–NE) | NE (7.7–NE) Median follow-up time: 18.7 months (IQR: 16.8–20.3) Median treatment duration: 9.7 months (IQR: 3.8–18.7) [Waterfall plot: Maximum Change in Tumor Size (%); legend: Ongoing; Received 1 cycle of SOC prior to enrollment; PD-L1 unknown; PD-L1 <1%; PD-L1 1-49%; PD-L1 ≥50%] Data cutoff: 05 Mar 2026. Efficacy analyses done in efficacy evaluable patients who were those with ≥1 post-baseline response assessment or had discontinued treatment before the first post-baseline response assessment. Patients with not evaluable response values were not included in the plot. Median time to response was 1.4 months (95% CI: 1.3–2.9). (a)PD-L1 status was unknown for 2 patients. (b)ORR, DOR, and PFS were investigator assessed. Abbreviations: BOR, best overall response; cORR, confirmed overall response rate; cPR, confirmed partial response; DCR, disease control rate; IQR, interquartile range; mDOR, median duration of response; mPFS, median progression-free survival; NE, not evaluable; PD, progressive disease; PD-L1, programmed death-ligand 1; SD, stable disease. Dr. Marcelo V. Negrao [Slide 4] Safety of 1L Olomorasib + Chemoimmunotherapy Olomorasib + chemoimmunotherapy (N=77)(a) — TRAEs (≥10%) Parameter, n (%) | Any grade | Grade 1 | Grade 2 | Grade 3 | Grade 4 Any TRAE | 74 (96.1) | 5 (6.5) | 27 (35.1) | 31 (40.3) | 11 (14.3) Nausea | 34 (44.2) | 16 (20.8) | 15 (19.5) | 3 (3.9) | 0 Anemia | 31 (40.3) | 4 (5.2) | 14 (18.2) | 12 (15.6) | 1 (1.3) Fatigue | 31 (40.3) | 16 (20.8) | 14 (18.2) | 1 (1.3) | 0 AST increased | 25 (32.5) | 11 (14.3) | 6 (7.8) | 7 (9.1) | 1 (1.3) Diarrhea | 25 (32.5) | 12 (15.6) | 9 (11.7) | 4 (5.2) | 0 Neutropenia | 24 (31.2) | 3 (3.9) | 5 (6.5) | 8 (10.4) | 8 (10.4) ALT increased | 22 (28.6) | 7 (9.1) | 6 (7.8) | 9 (11.7) | 0 Leukopenia | 15 (19.5) | 6 (7.8) | 1 (1.3) | 5 (6.5) | 3 (3.9) Blood creatinine increased | 14 (18.2) | 9 (11.7) | 5 (6.5) | 0 | 0 Thrombocytopenia | 14 (18.2) | 4 (5.2) | 1 (1.3) | 4 (5.2) | 5 (6.5) Vomiting | 14 (18.2) | 8 (10.4) | 5 (6.5) | 1 (1.3) | 0 Decreased appetite | 11 (14.3) | 5 (6.5) | 5 (6.5) | 1 (1.3) | 0 Constipation | 10 (13.0) | 6 (7.8) | 4 (5.2) | 0 | 0 Pruritus | 10 (13.0) | 4 (5.2) | 6 (7.8) | 0 | 0 Rash | 9 (11.7) | 5 (6.5) | 3 (3.9) | 1 (1.3) | 0 Stomatitis | 9 (11.7) | 7 (9.1) | 1 (1.3) | 1 (1.3) | 0 Pneumonitis/ILD | 8 (10.4) | 0 | 5 (6.5) | 3 (3.9) | 0 ➢ The safety profile remains consistent with an additional 9 months of follow-up time. ➢ TRAEs led to olomorasib(b) dose reductions in 15 (19.5%) patients. ➢ TRAEs led to permanent discontinuation of the treatment regimen(c) in 6 (7.8%) patients. Data cutoff: 05 Mar 2026. Neutropenia=neutrophil count decreased/neutropenia; Leukopenia=white blood cell count decreased/leukopenia; Thrombocytopenia=platelet count decreased/thrombocytopenia. (a)Safety population consists of patients who received at least one dose of study treatment. No grade 5 TRAEs. (b)Reasons for olomorasib dose reductions: ALT/AST increased (n=10), anemia (n=4), neutropenia/neutrophil count decreased (n=3), and biliary sepsis, blood creatinine increased, diarrhea, fatigue, hypokalemia, iron deficiency anemia, lipase increased, nausea, renal failure, vomiting, and white blood cell count decreased (n=1 each). (c)Reasons for discontinuation of the treatment regimen (olomorasib + chemoimmunotherapy): pancytopenia (n=2), pneumonitis (n=1), interstitial pneumonia (n=1), neutropenia followed by diarrhea and pneumonitis (n=1) and anemia followed by declining kidney function (n=1). Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; ILD, interstitial lung disease; TRAEs, treatment-related adverse events. Dr. Marcelo V. Negrao

[Slide 2] Conclusions ■ Olomorasib + chemoimmunotherapy demonstrated durable efficacy in patients with PD-L1 expression 0–100%, including in higher-risk patients with PD-L1 expression 0–49%. ➢ With a median follow-up of 18.7 months (IQR: 16.8–20.3), the ORR was 62% in the overall population, with a median DOR of 14.5 months and median PFS of 11.6 months. ■ The safety profile of olomorasib + chemoimmunotherapy was consistent with previous reports and with the known profiles of each drug. ➢ Safety was monitorable and manageable with supportive care, dose adjustments, and/or corticosteroids. ➢ No prophylactic medications for olomorasib related toxicities are required by the protocol. Part B of the ongoing Phase 3 SUNRAY-01 trial (NCT06119581) evaluating chemoimmunotherapy with olomorasib or placebo as first-line treatment in patients with advanced KRAS G12C-mutant NSCLC, irrespective of PD-L1 expression, has completed enrollment and is expected to read out in the first half of 2027. Abbreviations: DOR, duration of response; IQR, interquartile range; NSCLC, non-small cell lung cancer; ORR, overall response rate; PD-L1, programmed death ligand 1; PFS, progression-free survival. Dr. Marcelo V. Negrao [Slide 3] Hepatic Events Over Time in the Safety Population Patients (N=77) [Swimmer plot: Overall Treatment Duration (Months), 0–30; axis ticks not transcribed] ▶ Ongoing # Received 1 cycle of SOC prior to enrollment Hepatic Event Grade 2 Hepatic Event Grade 3 Hepatic Event Grade 4 ➢ Median time to onset of Grade ≥3 hepatic events was 43 days (range, 13–145). ➢ Median duration of Grade ≥3 hepatic events was 4 days (range, 1–37). ➢ All Grade ≥3 hepatic events resolved to Grade 1 or baseline after dose modifications and/or corticosteroids. ➢ No confirmed Hy's Law. Hepatic events included ALT increased, AST increased, hepatitis, immune hepatitis, autoimmune hepatitis, and hepatotoxicity. Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; SOC, standard of care. Dr. Marcelo V. Negrao

2683P First-line Olomorasib + Pembrolizumab in KRAS G12C-Mutant NSCLC: Updated Results from LOXO-RAS-20001 and SUNRAY-01 Trials Melissa L. Johnson, L. Leon-Mateos, Tadaaki Yamada, Neeharika Makani, Bryan Chan, Wen-Tsung Huang, Dionisios Spyratos, Jaromir Roubec, Niels Reinmuth, Adrian Sacher, Alexander I. Spira, Nimit Singhal, Shinji Takeuchi, Yonina R. Murciano-Goroff, Justin Han, Aaron A. Fink, Alyson Merced, Melinda D. Willard, Nicolas Fasnacht, Carla Visseren-Grul, Solange Peters Study sponsored by Eli Lilly and Company OBJECTIVE To assess the efficacy and safety of 1L olomorasib + pembrolizumab in KRAS G12C-mutant advanced NSCLC. – N=85 patients – Integrated analysis of the dose optimization cohorts of LOXO-RAS-20001 and SUNRAY-01 – Median follow-up: 20.7 months (IQR: 18.0-23.0) CONCLUSIONS ❖ Olomorasib + pembrolizumab demonstrates promising efficacy across PD-L1 levels. – With a median follow-up of 20.7 months, median PFS and DOR were not yet reached in the overall population and in patients with PD-L1 ≥50%. – The 18-month PFS rate is 55% in the overall patient population (65% in patients with PD-L1 ≥50%). ❖ TRAEs were primarily low-grade and hepatic events were generally transient and manageable demonstrating a safety profile consistent with prior reports and the known profiles of the individual drugs and highlights the combinability of this regimen. ❖ These updated findings further support the rationale for the ongoing Phase 3 trials evaluating olomorasib in combination with immunotherapy in KRAS G12C-mutant NSCLC PD-L1 0-100%: SUNRAY-01 (NCT06119581) and SUNRAY-02 (NCT06890598). BACKGROUND ▪ There are no approved targeted therapies in the 1L setting for patients with KRAS G12C-mutant NSCLC. – Current 1L treatment for this population consists of immunotherapy +/- platinum-based chemotherapy, guided by PD-L1 expression and clinical characteristics. – Toxicity has been a persistent challenge for some KRAS G12C inhibitor-immunotherapy-based combinations, limiting their development in the 1L setting. ▪ Olomorasib is a potent, selective next-generation KRAS G12C inhibitor with: – Durable 1L efficacy in combination with pembrolizumab across all PD-L1 levels. – Tolerable safety profile with toxicities that can be readily managed. – Intracranial activity. ▪ Now with over 20 months of follow-up, we report updated results in 1L patients receiving olomorasib + pembrolizumab from the integrated analysis of LOXO-RAS-20001 and dose optimization for SUNRAY-01. STUDY DESIGN AND METHODS Olomorasib + pembrolizumab LOXO-RAS-20001: Part G 1L NSCLC (n=43) SUNRAY-01: Dose Optimization 1L NSCLC (n=42) Eligibility ▪ Confirmed Stage IIIB-IIIC or Stage IV non-squamous KRAS G12C-mutant NSCLC ▪ No prior therapy for advanced or metastatic NSCLC – Patients may have received up to one cycle of SOC pembrolizumab prior to enrollment (21-day cycle of pembrolizumab 200 mg) ▪ PD-L1 expression 0-100% ▪ ECOG PS of 0-1 ▪ Small (≤1.5 cm), asymptomatic untreated or treated stable brain metastases Key Objectives ▪ Safety, tolerability, and pharmacokinetics ▪ ORR, DOR, DCR, and PFS per RECIST v1.1 aDoses: Olomorasib 50 or 100 mg BID; Pembrolizumab 200 mg IV Q3W. RESULTS Table 1. Demographics and Baseline Characteristics | Olomorasib + pembrolizumab (N=85) Age, median, years (range) | 69 (44-88) Sex, Male / Female | 41 (48.2) / 44 (51.8) Race, White / Asian / Black / Not reported | 48 (56.5) / 29 (34.1) / 2 (2.4) / 6 (7.1) ECOG PS 0 / 1 | 39 (45.9) / 46 (54.1) Former or current smoker / Never smoker / Unknown | 80 (94.1) / 4 (4.7) / 1 (1.2) PD-L1 expression <1% | 13 (15.3); 1-49% | 18 (21.2); ≥50% | 54 (63.5) Metastases at baseline: Brain 16 (18.8); Liver 11 (12.9); Bone 29 (34.1) Received 1 cycle of SOC (pembrolizumab) prior to enrollment | 14 (16.5) Data cutoff date of 5 March 2026. Includes n=42 treated at 50 mg BID and n=43 treated at 100 mg BID of olomorasib. Table 2. Response and Time-to-Event Endpoints in the Efficacy Evaluable Population Outcomes | All Patients PD-L1 0-100% (N=84) | PD-L1 ≥50% (n=54) ORR, % (n/N) | 72.6 (61/84) | 77.8 (42/54) BOR, n (%): CR 3 (3.6) | 0 (0.0); PR 58 (69.0) | 42 (77.8); SD 15 (17.9) | 7 (13.0); PD 5 (6.0) | 3 (5.6); NE 3 (3.6) | 2 (3.7) DCR, % (n/N) | 90.5 (76/84) | 90.7 (49/54) Median DOR, months (95% CI) | NE (16.6-NE) | NE (NE-NE) Median PFS, months (95% CI) | NE (13.4-NE) | NE (15.9-NE) PFS rate, % (95% CI): 6-month 76.8 (66.1-84.6) | 81.0 (67.5-89.3); 12-month 65.0 (53.4-74.4) | 68.9 (54.3-79.7); 18-month 54.9 (43.0-65.3) | 64.5 (49.6-76.0) Data cutoff date of 05 Mar 2026. Median follow-up time was 20.7 months (IQR: 18.0-23.0). Median time to response: 1.4 months (IQR: 1.4-1.7). Efficacy evaluable patients (n=84) are those who had at least one post-baseline response assessment or had discontinued treatment before the first post-baseline response assessment. ORR, DOR, and PFS were investigator assessed. Confirmed responses. Figure 1. Best Change in Tumor Burden by PD-L1 Level in the Efficacy Evaluable Population Figure 2. Treatment Duration by PD-L1 Level in the Efficacy Evaluable Population ▪ Median follow-up: 20.7 months (IQR: 18.0-23.0). ▪ Median duration of treatment: 13.6 months (IQR: 5.6-21.2). Table 3. TRAEs by Grade with ≥10% Incidence in the Safety Population | Olomorasib + pembrolizumab (N=85) | Any Grade | Grade 1 | Grade 2 | Grade 3 | Grade 4 Any TRAE | 77 (90.6) | 11 (12.9) | 29 (34.1) | 31 (36.5) | 6 (7.1) Diarrhea | 28 (32.9) | 9 (10.6) | 12 (14.1) | 7 (8.2) | - ALT increased | 23 (27.1) | 6 (7.1) | 3 (3.5) | 13 (15.3) | 1 (1.2) AST increased | 21 (24.7) | 4 (4.7) | 5 (5.9) | 11 (12.9) | 1 (1.2) Nausea | 15 (17.6) | 9 (10.6) | 6 (7.1) | - | - Fatigue | 14 (16.5) | 6 (7.1) | 8 (9.4) | - | - Pruritus | 13 (15.3) | 9 (10.6) | 4 (4.7) | - | - Vomiting | 12 (14.1) | 6 (7.1) | 5 (5.9) | 1 (1.2) | - Decreased appetite | 9 (10.6) | 5 (5.9) | 4 (4.7) | - | - Rash | 9 (10.6) | 6 (7.1) | 2 (2.4) | 1 (1.2) | - Other TRAEs of interest: Pneumonitis/ILD | 3 (3.5) | 1 (1.2) | - | 1 (1.2) | 1 (1.2) ▪ The safety profile remains consistent with an additional 9 months of follow-up time. ▪ TRAEs led to olomorasib dose reductions in 27 (31.8%) patients. ▪ TRAEs led to permanent discontinuation of the treatment regimen in 10 (11.8%) patients. Data cutoff date of 05 Mar 2026. No Grade 5 TRAEs. Figure 3. Hepatic Events Over Time in the Safety Population ▪ Median time to onset of Grade ≥3 hepatic events was 61 days (range, 23-302). ▪ Median duration of Grade ≥3 hepatic events was 7 days (range, 3-49). ▪ All Grade ≥3 hepatic events resolved to Grade 1 or baseline after dose modifications and/or corticosteroids. ▪ No confirmed Hy's Law. Hepatic events include ALT increased, AST increased, hepatitis, immune hepatitis, autoimmune hepatitis, and hepatotoxicity. European Society for Medical Oncology (ESMO) Congress Madrid, Spain; 23-27 October 2026 Melissa.johnson@scri.com

LBA64 - Toripalimab plus chemotherapy as perioperative treatment for resectable stage II or III non-small cell lung cancer (NSCLC): Final analysis of the phase 3 Neotorch study Speakers Shun Lu (Shanghai, China) Lecture Time 15:23 - 15:33 Authors Shun Lu (Shanghai, China), Lin Wu (Changsha, China), Wenxiang Wang (Changsha, China), Longhua Sun (Nanchang, China), Peng Zhang (Shanghai, China), Wentao V. Fang (Shanghai, China), Zheng Yan (Zhengzhou, China), Qixun Chen (Hangzhou, China), Lin Yang (, China), Lijie Tan (Shanghai, China), Jiandong Mei (Chengdu, China), Jinlu Shan (Chongqing, China), Shidong Xu (Harbin, China), Xiaohong Sun (Urumqi Municipality, China), Nong Yang (Changsha, China), Xiaohua Hu (Nanning, China), Guohua Yu (Weifang, China, Kuiwen District), Dongliang Yu (Nanchang, China), Yuping Chen (Shantou, China), Ligang Xing (Jinan, China) Abstract Background In the interim analysis of the Neotorch study, perioperative toripalimab plus chemotherapy significantly improved the event-free survival (EFS) in patients with resectable stage III NSCLC. Here, we report the final analysis in patients with Stage II or III disease. Methods Eligible adult patients with histologically confirmed resectable Stage II or III NSCLC (excluding patients with EGFR or ALK alterations for non-squamous NSCLC) were randomized 1: 1 to receive 240 mg toripalimab or placebo once every 3 weeks (Q3W) plus platinum-based chemotherapy for 3 cycles before surgery, 1 cycle after surgery, followed by toripalimab or placebo monotherapy for up to 13 cycles. The primary endpoints were EFS assessed by investigators (INV-EFS) and major pathological response rate by a blinded independent pathologic review (BIPR) in Stage III and Stage II or III populations. Secondary endpoints included overall survival (OS), pathologic complete response (pCR) rate, and safety. Results A total of 501 patients (404 Stage III, 97 Stage II) were randomized to toripalimab (n=250) or placebo (n=251) arms. By April 18, 2026, the median follow-up was 48.2 months. Baseline characteristics were well balanced between the 2 arms. The INV-EFS was significantly improved in the toripalimab arm (hazard ratio [HR]: 0.52 ; 95% CI: 0.391, 0.677; p <0.0001). The OS was also significantly prolonged in the toripalimab arm (HR: 0.66; 95% CI: 0.486, 0.894; p =0.0072). The BIPR-assessed MPR and pCR rates were much higher in the toripalimab arm (Table). The safety profile was consistent with the interim analysis, with no new safety signals identified (Table). Table: LBA64 Efficacy and safety summary | Toripalimab arm (n=250) | Placebo arm (n=251) INV-EFS, median (95% CI), m | NR (51.3, NE) | 20.2 (14.3, 24.8) HR (95% CI) | 0.52 (0.391, 0.677) p | <0.0001 BIPR-MPR rate, (95% CI), % | 47.2 (40.9, 53.6) | 9.2 (5.9, 13.4) p | <0.0001 OS, median (95% CI), m | NR (NE, NE) | NR (58.6, NE) HR (95% CI) | 0.66 (0.486, 0.894) p | 0.0072 BIPR-pCR, (95% CI), % | 25.6 (20.3, 31.5) | 1.2 (0.2, 3.5) p | <0.0001 Common adverse events, n (%) Anaemia | 166 (66.4) | 163 (64.9) Neutropenia | 145 (58.0) | 134 (53.4) Leukopenia | 142 (56.8) | 127 (50.6) Alanine aminotransferase increased | 83 (33.2) | 65 (25.9) Thrombocytopenia | 76 (30.4) | 58 (23.1) Aspartate aminotransferase increased | 76 (30.4) | 60 (23.9) Decreased appetite | 75 (30.0) | 38 (15.1) NR, not reached; NE, not estimable. Conclusions The addition of toripalimab to perioperative chemotherapy showed significant and clinically meaningful improvements in EFS, MPR rate, and OS for resectable Stage II or III NSCLC patients with a manageable safety profile.

[Slide 2] EFS in Stage II and III Patients EFS by investigator Median follow-up: 48.2 months Toripalimab Placebo Median(95% CI): NR(51.3,NE) 20.2(14.3,24.8) HR (95% CI) 0.52 (0.391,0.677) P-value <0.0001 EFS by IRC Median follow-up: 48.2 months Toripalimab Placebo Median(95% CI): NR(NE,NE) 22.6 (15.1, 45.8) HR (95% CI) 0.53(0.395, 0.701) P-value <0.0001 Event-free survival,% of patients Follow-up,mo No. at risk Toripalimab Placebo NR, not reached; NE: not estimable; HR; Hazard ratio; CI: confidence interval Data cutoff date: April 18, 2026 Shun Lu Survival table Endpoint: EFS (stage II and III), by investigator and by IRC Arm | N | Median | Median CI Toripalimab (investigator-assessed) | 250 | NR | 51.3, NE Placebo (investigator-assessed) | 251 | 20.2 | 14.3, 24.8 Toripalimab (IRC) | 250 | NR | NE, NE Placebo (IRC) | 251 | 22.6 | 15.1, 45.8 HR: 0.52 (investigator); 0.53 (IRC) HR CI: 0.391, 0.677 (investigator); 0.395, 0.701 (IRC) P: <0.0001 (both) Investigator: 24-mo 71.5 vs 45.5; 36-mo 63.1 vs 40.6 (toripalimab vs placebo, printed labels) IRC: 24-mo 70.1 vs 48.4; 36-mo 64.0 vs 44.5 (toripalimab vs placebo, printed labels) [Slide 3] Pathological response in Stage II and III Patients MPR by BIPR Difference=38% (95% CI: 31, 45) P < 0.0001 Toripalimab + chemo 47.2 Placebo + chemo 9.2 pCR by BIPR Difference=24% (95% CI: 19, 30) P < 0.0001 Toripalimab + chemo 25.6 Placebo + chemo 1.2 HR; Hazard ratio; CI: confidence interval Data cutoff date: April 18, 2026 Shun Lu [Slide 4] Overall Survival in Stage II and III Patients Median follow-up: 48.2 months; maturity: 34.3% Toripalimab Placebo Overall survival,% of patients Follow-up,mo No. at risk(censor) N 250 251 Events, n(%) 73(29.2) 99(39.4) Median OS (95% CI), mo NR(NE, NE) NR(58.6, NE) HR (95% CI) 0.66(0.49, 0.89) P-value 0.0072 NR, not reached; NE, not estimable; HR; Hazard ratio; CI: confidence interval Data cutoff date: April 18, 2026 Shun Lu Survival table Endpoint: OS (stage II and III) Arm | N | Median | Median CI Toripalimab | 250 | NR | NE, NE Placebo | 251 | NR | 58.6, NE HR: 0.66 HR CI: 0.49, 0.89 P: 0.0072 24-mo 84% vs 78% (toripalimab vs placebo, printed labels) 36-mo 78% vs 67% (toripalimab vs placebo, printed labels)

[Slide 1] Neotorch Study Design Neoadjuvant* | Adjuvant | Maintenance - Newly diagnosed resectable stage II-III NSCLC - EGFR/ALK wild type - Biopsy tissue available for biomarker analysis - Measurable lesion(s) 1:1 (N=501) Toripalimab 240mg + Platinum-based chemotherapy Q3W 3 cycles Placebo + Platinum-based chemotherapy Q3W 3 cycles SURGERY Toripalimab 240mg + Platinum-based chemotherapy Q3W 1 cycle Placebo + Platinum-based chemotherapy Q3W 1 cycle Toripalimab 240mg Q3W up to 13 cycles Placebo Q3W up to 13 cycles Follow Up Primary endpoints: - EFS by Investigator (stage III) - EFS by Investigator (stage II-III) - MPR by BIPR (stage III) - MPR by BIPR (stage II-III) Secondary Endpoints: - Overall survival - pCR by BIPR/site pathologist for stage III and stage II-III - EFS by IRC for stage III and stage II-III - Disease-free survival - Safety and feasibility of surgery Stratification factors: - II vs IIIA vs IIIB - Lobectomy vs pneumonectomy - Non-squamous vs squamous - PD-L1 TC expression: ≥ 1% vs < 1% or non-evaluable EFS: event-free survival; MPR: major pathological response; pCR: pathologic complete response; BIPR: blinded independent pathologic review; IRC: independent review committee * 3 cycles of neoadjuvant chemotherapy with 4 cycles of peri-operative chemotherapy in total were required with in Neotorch study, meanwhile, surgeons were allowed to determine the most appropriate timing for surgery based on the patient's condition. Shun Lu [Slide 3] EFS in Different PD-L1 Expression Subgroups PD-L1 < 1% (n=129) PD-L1 1-49% (n=164) PD-L1 ≥ 50% (n=157) Toripalimab | Placebo Median EFS 51.3 vs. 20.2 months HR = 0.72 (95% CI 0.436-1.191) Median EFS 40.9 vs. 15.4 months HR = 0.41 (95% CI 0.264-0.631) Median EFS NR vs. NR months HR = 0.47 (95% CI 0.262-0.812) Event-free survival,% of patients Follow-up,mo No. at risk NR, not reached; HR; Hazard ratio; CI: confidence interval Data cutoff date: April 18, 2026 Shun Lu Survival table Endpoint: EFS by PD-L1 subgroup Toripalimab, PD-L1 <1% | 63 | 51.3 mo Placebo, PD-L1 <1% | 66 | 20.2 mo Toripalimab, PD-L1 1-49% | 85 | 40.9 mo Placebo, PD-L1 1-49% | 79 | 15.4 mo Toripalimab, PD-L1 ≥50% | 75 | NR Placebo, PD-L1 ≥50% | 82 | NR PD-L1 <1%: HR 0.72 (95% CI 0.436-1.191) PD-L1 1-49%: HR 0.41 (95% CI 0.264-0.631) PD-L1 ≥50%: HR 0.47 (95% CI 0.262-0.812)

[Slide 1] PEARLS Trial: Adjuvant Pembrolizumab vs. Placebo AEs were monitored through 30 days after last study dose | Pembro Arm (n = 580) | Placebo Arm (n = 581) Any AE 556 (95.9) | 529 (91.0) Grade 3–5 197 (34.0) | 149 (25.6) Led to death 11 (1.9) | 6 (1.0) Led to treatment discontinuation 116 (20.0) | 34 (5.9) Treatment-related AEs 436 (75.2) | 305 (52.5) Grade 3–5 89 (15.3) | 25 (4.3) Led to death 4 (0.7)b | 0 Led to treatment discontinuation 99 (17.1) | 19 (3.3) [row label obscured] [obscured].9) | 80 (13.8) [row label obscured] [obscured]9) | 11 (1.9) [row label obscured] [obscured]) | 0 [row label obscured] [obscured].3) | 9 (1.5) • What about long term AEs, irAES? • High discontinuation rates → Do all patients need 1-year of ADJ IO? • Can we preserve the DFS benefit while reducing the treatment exposure? • How do we select fit patients for ADJ IO? INITIALLY REGISTERED N=1955 → RANDOMISED N=1177 (60%) → 580 RECEIVED ≥1 DOSE ADJ IO → COMPLETED ADJ IO n=300 (52%) / DISCONTINUED ADJ IO n=280. Most common reason AEs 131/280 (47%) O'Brien – ESMO 2026 (Table modified) * O'Brien – Lancet Oncol 2022 [Slide 2] From LACE Pooled analysis to PEARLS trial A Overall Survival (%) — Chemotherapy / No chemotherapy — Time From Randomization (years) Deaths / person years by period: Years 0-3 | Years 4-5 | Years ≥ 6 Control 966 / 5,155 | 239 / 1,668 | 49 / 720 Chemotherapy 857 / 5,181 | 203 / 1,817 | 76 / 790 B Disease-Free Survival (%) — Chemotherapy / No chemotherapy Events / person years by period: Years 0-3 | Years 4-5 | Years ≥ 6 Control 1,222 / 4,341 | 163 / 1,396 | 35 / 610 Chemotherapy 1,047 / 4,627 | 159 / 1,606 | 59 / 708 Chemotherapy Arm in PEARLS trial ↑ 5y OS ~ 5y DFS PEARLS Overall Survival, % — 6 Years: 68.5% (95% CI, 64.5–72.1) | 66.4% (95% CI, 62.4–70.1) No. at risk Pembro 590 551 513 475 441 406 361 206 90 25 0 Placebo 587 556 510 459 428 399 354 202 81 27 0 PEARLS Disease-Free Survival, % — 6 Years: 44.0% (95% CI, 39.7–48.2) | 39.2% (95% CI, 35.0–43.2) Is a 5% DFS benefit clinically meaningful if there is no OS benefit, considering the 31% reduction in lung cancer deaths? Pignon – JCO 2008 * NSCLC Meta analysis – Lancet 2010 (5y OS adjuvant CT 64%) * O'Brien – ESMO 2026 [Slide 3] Heterogeneous results with ADJ IO IMpower010 trial BSC vs. Atezolizumab — sII-IIIAPDL1≥1% (US flag), PDL1≥50% (EU flag) — DFS sII-IIIA PDL1 ≥1%: 37,3 vs 68,5 PEARLS/KEYNOTE-091 trial Placebo vs. Pembrolizumab — All PDL1, post-ADJ CT — DFS ITT: 43,3 vs 58,7; DFS PDL1 ≥50%: 47,6 vs 67 BR.31 trial Placebo vs. Durvalumab — DFS EGFR/ALK- PDL1 ≥25%: 60,2 vs 69,9 ALCHEMIST (EA5142) trial Placebo vs. Nivolumab — DFS ITT: 68,8 vs 71,3; DFS PDL1 ≥50%: 78,5 vs 89,8 Primary endpoint IMpower010 HR: 0.70 (0.55-0.91) PEARLS DFS ITT HR: 0.85 (0.73-0.98)* PEARLS PDL1 ≥50% DFS IA3 HR: 0.83 (0.59-1.16) DFS Final HR: 0.87 (0.65-1.17)* BR.31 HR: 0.93 (0.71-1.25) ALCHEMIST ITT HR: 0.97 (0.81-1.17); PDL1 ≥50% HR: 0.86 (0.59-1.25) Overall Survival IMpower010 87.1 mo. vs. NE HR: 0.77 (0.56-1.06) Exploratory PEARLS NR. vs. NR* HR: 0.98 (0.81-1.18) p=0.41 PEARLS PDL1 ≥50%: Not Reported BR.31 NR vs NR HR: 0.98 (0.69-1.42) ALCHEMIST: Not mature Remon &Peters – Lancet Resp med 2026, modified (*HR according to last data presented by O'Brien at ESMO 2026. Note the median DFS in high PD-L1 is not based on final analysis as not reported )

[Slide 1] PEARLS/KEYNOTE-091 Study Design (NCT02504372) Randomized, triple-blind, Phase 3 Trial Eligibility for Registration • Confirmed stage IB (T ≥4 cm), II, or IIIA NSCLC per AJCC v7 • Complete surgical resection with negative margins (R0) • Provision of tumor tissue for PD-L1 testing PD-L1 testing (centrally using PD-L1 IHC 22C3 pharmDx) Eligibility for Randomization • No evidence of disease • ECOG PS 0 or 1 • Adjuvant chemotherapya • Considered for stage IB (T ≥4 cm) disease • Strongly recommended for stage II and IIIA disease • Limited to ≤4 cycles N~1180 R 1:1 Pembrolizumab 200 mg IV Q3W for ≤18 administrations (~1 y) Placebo IV Q3W for ≤18 administrations (~1 y) Randomization Stratification Factors • Disease stage (IB vs II vs IIIA) • PD-L1 TPS (<1% vs 1%–49% vs ≥50%) • Receipt of adjuvant chemotherapy (yes or no) • Geographic region (Asia vs Eastern Europe vs Western Europe vs rest of world) Dual primary endpoints • DFS in the ITT population • DFS in the PD-L1 TPS ≥50% population Secondary endpoints • DFS in the PD-L1 TPS ≥1% population • OS in ITT, PD-L1 TPS ≥50%, and PD-L1 TPS ≥1% populations • Lung cancer–specific survival (LCSS) in the ITT population • Safety Exploratory endpoints • Health-related quality of life aParticipants began adjuvant chemotherapy ≤12 weeks of surgery and were randomized and received study treatment ≥3–≤12 weeks from last chemotherapy dose; participants who did not receive adjuvant chemotherapy were randomized and received study treatment ≤12 wk of surgery. [Slide 2] Final Analysis: Disease-Free Survivala,b After 7 Years of Follow-Up ITT Populationc Received Adjuvant Chemo Pts w/ Event | Median (95% CI), mo | HRd (95% CI) Pembro | Placebo 6 Years Disease-Free Survival, % Time, mo No. at risk aPer RECIST v1.1 by investigator assessment. bDFS HR in the PD-L1 TPS ≥50% population was 0.87 (95% CI, 0.65–1.17). cAmong DFS events, new malignancy events occurred in 55 participants (9.3%) in the pembrolizumab arm and 65 (11.1%) in the placebo arm. dBased on the multivariate Cox regression model. Median time from randomization to data cutoff (January 21, 2026) was 87.7 (range, 68.6–120.1) mo. Survival table Endpoint: DFS (investigator, RECIST v1.1) Arm | N | Median | Median CI Pembro (ITT) | 590 | 58.7 mo | 47.0–70.4 Placebo (ITT) | 587 | 43.3 mo | 35.0–55.6 Pembro (received adjuvant chemo) | 506 | 58.7 mo | 47.0–70.5 Placebo (received adjuvant chemo) | 504 | 39.1 mo | 31.3–51.7 HR: ITT 0.85; adjuvant chemo 0.80 HR CI: ITT 0.73–0.98; adjuvant chemo 0.68–0.94 ITT 6-yr DFS: pembro 44.0% (95% CI 39.7–48.2) vs placebo 39.2% (95% CI 35.0–43.2) Adjuvant chemo 6-yr DFS: pembro 43.6% (95% CI 38.9–48.1) vs placebo 38.1% (95% CI 33.7–42.5) PD-L1 TPS ≥50% DFS HR 0.87 (95% CI 0.65–1.17) (footnote) [Slide 3] Final Analysis: Overall Survival ITT Population Received Adjuvant Chemo 6 Years Pts w/ Event | Median (95% CI), mo | HRa (95% CI) Pembro | Placebo Overall Survival, % Time, mo No. at risk 12.8% of participants (pembro arm) vs 29.3% (chemo arm) with PD received subsequent IO therapy aBased on the multivariate Cox regression model. bOne-sided P value based on the Wald Test in the multivariate Cox regression model. Median time from randomization to data cutoff (January 21, 2026) was 87.7 (range, 68.6–120.1) mo. Survival table Endpoint: OS Arm | N | Median | Median CI Pembro (ITT) | 590 | NR | 100.5–NR Placebo (ITT) | 587 | NR | 110.4–NR Pembro (received adjuvant chemo) | 506 | NR | 100.5–NR Placebo (received adjuvant chemo) | 504 | NR | 103.6–NR HR: ITT 0.98; adjuvant chemo 0.91 HR CI: ITT 0.81–1.18; adjuvant chemo 0.74–1.12 P: ITT one-sided P = 0.41826 ITT 6-yr OS: pembro 68.5% (95% CI 64.5–72.1) vs placebo 66.4% (95% CI 62.4–70.1) Adjuvant chemo 6-yr OS: pembro 68.9% (95% CI 64.6–72.8) vs placebo 65.4% (95% CI 61.0–69.4) [Slide 4] Final Analysis: Lung Cancer–Specific Survival ITT Population Received Adjuvant Chemo 6 Years Pts w/ Event | Median (95% CI), mo | HRa (95% CI) Pembro | Placebo Disease-Free Survival, % Time, mo No. at risk Non–lung cancer causes of death | CV disease | COVID-19 | Other malignancy | Treatment-related AEs | Otherb Pembro, n (%) | 12c (2.0) | 8 (1.4) | 9 (1.5) | 4 (0.7) | 71 (12.1) Placebo, n (%) | 3 (0.5) | 1 (0.2) | 13 (2.2) | 0 | 67 (11.4) aBased on the multivariate Cox regression model. bOther includes other causes, other chronic disease, and unknown causes. cIncludes 1 death that was also associated with COVID-19. Median time from randomization to data cutoff (January 21, 2026) was 87.7 (range, 68.6–120.1) mo. Survival table Endpoint: Lung cancer-specific survival (LCSS) Arm | N | Median | Median CI Pembro (ITT) | 590 | NR | NR–NR Placebo (ITT) | 587 | NR | NR–NR Pembro (received adjuvant chemo) | 506 | NR | NR–NR Placebo (received adjuvant chemo) | 504 | NR | NR–NR HR: ITT 0.81; adjuvant chemo 0.69 HR CI: ITT 0.63–1.03; adjuvant chemo 0.53–0.91 ITT 6-yr LCSS: pembro 81.3% (95% CI 77.7–84.4) vs placebo 76.9% (95% CI 73.1–80.2) Adjuvant chemo 6-yr LCSS: pembro 81.6% (95% CI 77.7–84.9) vs placebo 75.0% (95% CI 70.8–78.7)

[Slide 1] • Antibody-drug conjugates (ADCs) represent one of the fastest-growing classes of targeted cancer therapeutics • More than 400 ADCs currently in active global clinical development • Bispecifics have emerged as an exciting class of drugs Three Complementary Bispecific-Based Strategies in Developmental Therapeutics 999O | Elfetabart drozuntecan (B7-H3 ADC) + Pumitamig (PD-L1 x VEGF-A bsAb) | Elfe-D Anti-B7-H3 ADC | Pumitamig PD-L1 x VEGF-A bispecific | B7-H3 | PD-L1 | VEGF-A | Drozutecan payload (TOP1 inh) DAR = 6 1000O | HS-20122 (EGFR x c-MET bispecific ADC) | EGFR binding Fab | c-MET binding Fab | Cleavable linker | Topoisomerase I inhibitor (HS-9265) 1001O | IBI3001 (EGFR x B7-H3 bispecific ADC) | EGFR | B7-H3 | Exatecan payload Complementary approaches to address tumor heterogeneity, resistance, and the evolving tumor microenvironment Dr. Funda Meric-Bernstam Images adapted from presentations. [Slide 2] Selected VEGF × PD-1/PD-L1 Bispecific Antibodies Over 30 Bispecific antibodies co-targeting VEGF and the PD-1/PD-L1 axis are being developed Drug Candidate | Target Axis | Sponsor / Developer | Clinical Phase & Trial Acronyms Ivonescimab (AK112 / SMT112) | PD-1 × VEGF | Akeso / Summit Therapeutics | Approved (China) / Phase 3 • HARMONi (EGFRm NSCLC) • HARMONi-2 (1L NSCLC) Pumitamig (BNT327 / PM8002) | PD-L1 × VEGF | Biotheus / BioNTech | Phase 3 • Global registrational trials across solid tumors. MK-2010 | PD-1 × VEGF | Merck & Co. | Phase 1/2 • Ongoing basket evaluation in advanced tumors. IMM2510 | PD-L1 × VEGF | ImmuneOnco Biomed | Phase 1/2 • Focus on GI and lung malignancies. PF-08634404 | PD-1 × VEGF | Pfizer | Phase 1/2 • Dose-escalation and safety expansion. HB0025 | PD-L1 × VEGF | Huabio / Hualan Bio | Phase 1/2 • Solid tumor basket trial. JS207 | PD-1 × VEGF | Junshi Biosciences | Phase 1 • Advanced solid tumors. HLX37 | PD-L1 × VEGF | Henlius | Phase 1 • First-in-human trial. Dr. Funda Meric-Bernstam [Slide 3] HS-20122 : Efficacy in Patients with at least One Post-Baseline Tumor Assessment Best Overall Response in Various Tumor Types (N=177) | ORR: 44.1% | DCR: 91.0% EGFRm NSCLC Best Overall Response (N=106) | ORR: 55.7% | DCR: 92.5% | 6-mo PFS rate: 76.7% Best Overall Response in 6mg/kg Q3W (N=31) | ORR: 61.3% | 6-mo PFS rate: 75.3% | 95% CI (45.8, 90.2) Best Overall Response in 5mg/kg Q2W (N=40) | ORR: 45.0% | 6-mo PFS rate: 45.6% | 95% CI (1.0, 87.5) Legend: PR | SD | PD | NE HS-20122 with 44.1% ORR Most mature data in EGFRm NSCLC More pts treated with 6.mg/kg q3 wk than 5 mg/kg q 2wk but 6g/kg has: • Numerically higher ORR in EGFRm NSCLC (61.3% vs 45%) • Lower dose interruptions for AEs (17.8% vs 29.5%) in overall population Responses also reported in other tumor types: • 4 of 6 nasopharyngeal cancers • 2 of 6 HNSCC • 1 of 7 CRC Dr. Funda Meric-Bernstam [Slide 4] EGFR bispecifics: Summary Safety: • Both EGFR/MET and EGFR/B7H3 agents with manageable safety, mainly heme tox as expected by payload • Limited ILD, limited rash Efficacy: HS-20122, an EGFR/c-Met ADC, larger study 215 pts, 177 efficacy evaluable • Preliminary encouraging activity. In the 6 mg/kg Q3W dose level, - More mature lung data: In EGFR TKI pre-treated NSCLC, ORR was 61.3% with 6-mo PFS rate 75.3% Not known which mechanisms of resistance were present after EGFR TKIs - Limited experience with other tumors types but responses observed: squamous NSCLC, HNC, CRC IBI3001: bispecific EGFR and B7H3 ADC : smaller study: 57 patient safety cohort and 53 efficacy cohort • Less mature data but preliminary activity: ORR was 30.2% • Responses seen in EGFRm NSCLC : 5 (31.3%) of 16, as well as in SCLC, PDAC, HNSCC, ESCC Dr. Funda Meric-Bernstam

[Slide 2] My Agenda Immunotherapy 1.0 PD-1 and CTLA-4 single target antibodies Toxicity 1.0 Single target toxicity Immunotherapy 2.0 Bispecific antibodies Toxicity 2.0 Multi-targets toxicity Immunotherapy 3.0 T-cells engagers Toxicity 3.0 «T-cells storm» toxicity ROBERTO FERRARA [Slide 3] How to improve anti-CTLA-4 antibodies Fc: the NEGLECTED HALF of the antibodies SCORPION antibody: Enhanced Fc region → increased depletion of CTLA-4+ cells (namely T-regs) Periphery · normal tissue T cell PD-1 only dissociates Only one arm can bind low avidity → it lets go peripheral T cells are spared Tumour microenvironment PD-1 CTLA-4 TIL PD-1 + CTLA-4 Both arms bind at once high avidity → it stays the drug is retained in the tumour Fv Fab Fc Bispecific CTLA-4 / PD-(L)1 antibodies CRAB antibody: 2 claws - 2 targets higher avidity in the microenvironment → low peripheral effect and toxicity IgG1 anti-CTLA-4: Fc engagement can deplete Tregs by ADCC Ipilimumab Treg FcγR NK cell / macrophage Treg killed CTLA-4 1 binds CTLA-4 on the Treg 2 Fc engages FcγR 3 Treg killed (ADCC / ADCP) [Slide 4] The anti-TIGIT species: a different story SCORPION anti-TIGIT • Tiragolumab (IgG1 human, Fc active) CRAB anti-TIGIT/PD-1 • Rilvegostomig Bispecific TIGIT/PD1 (IgG1 humanized – reduced Fc fuction) TRAEs G3/G4 20% 8% Rilvegostomig Immune-mediated rash, any grade 42% 8% Tiragolumab Infusion reactions any grade 12% 6% 0% 10% 20% 30% 40% 50% Tiragolumab + atezolizumab (SKYSCRAPER-01) — Fc-active Rilvegostomig (ARTEMIDE-01) — Fc-silent bispecific Tiragolumab →CRS-like tox: rash and infusion reactions Rilvegostomig → low rate grade 3-4 tox Peters JCO 2026, Jonhson ASCO 2022, Rohrberg Clinical Cancer Res 2026 ROBERTO FERRARA

Poster no. 2577 ESMO 2026 Tambotatug pelitecan (YL201): Efficacy and safety in locally advanced/metastatic NSCLC from a phase 1 study Stephen V. Liu, Valentina Gambardella, Bernard Doger de Spéville, Siqing Fu, Deborah Doroshow, Shiraj Sen, Ulka Vaishampayan, Steve Chin, Xian Zhang, Ying Sun, Barbara Gitlitz, Mark Yan, Anjali Rohatgi BACKGROUND • Tambotatug pelitecan (Tam-Peli [YL201]) is an optimized antibody-drug conjugate (ADC) that consists of a human anti-B7-homolog 3 (B7-H3) antibody linked to a novel topoisomerase I inhibitor payload via the tumor microenvironment activable linker-payload (TMALIN®) platform (Figure 1). Figure 1. Tam-Peli mechanism of action Target: B7-H3 • Highly expressed across various solid tumors, on tumor cells, and in the tumor microenvironment • Minimally expressed in healthy tissue Payload • Novel topoisomerase I inhibitor payload • Eight molecules of cytotoxic payload per antibody Linker • Highly hydrophilic and plasma-stable linker to minimize premature payload release Dual release of payload TMALIN® technology 1. Intracellular release of payload 2. Extracellular cleavage releasing payload directly into the tumor microenvironment • Tam-Peli showed promising antitumor activity in a combined analysis from two phase 1 studies (INT-101-01 [NCT05434234] and CN-101-01 [NCT06057922]) in patients with previously treated solid tumors, including previously treated non-small cell lung cancer (NSCLC). • Tam-Peli showed a significant progression-free survival (PFS) and overall survival benefit over chemotherapy in previously treated, extensive stage small cell lung cancer in the ongoing phase 3 TAISHAN-302 study, which is conducted in China. • We present safety and efficacy data of Tam-Peli in patients with locally advanced/metastatic non-squamous NSCLC from the dose-expansion phase of the global phase 1 INT-101-01 study. METHODS Figure 2. Study design for the global INT-101-01 study Key eligibility for patients with NSCLC: • Locally advanced or metastatic non-squamous NSCLC • No actionable genomic alterations • At least one line of prior therapy • ECOG PS 0–1 Dose escalation (3+3)*: Patients with previously treated NSCLC (N=19) 2.0 mg/kg (n=5) R Randomization stratified by brain metastasis Dose expansion: Patients with previously treated NSCLC 1.6 mg/kg Q3W (n=20); Patients with previously treated NSCLC 2.0 mg/kg Q3W (n=20) • Primary endpoints included objective response rate (ORR) per RECIST v1.1 and safety per CTCAE v5.0. • Secondary endpoints included disease control rate (DCR), duration of response (DOR), and PFS. RESULTS Patients • As of 10 April 2026, 40 patients with actionable genomic alteration-negative NSCLC were enrolled, of which 37 were included in the efficacy analysis set – Baseline characteristics are presented in Table 1. • Prior therapies included anti-programmed death-ligand 1 therapy (100%; n=40), platinum-based chemotherapy (100%; n=40), other chemotherapy (55%; n=22), and ADC therapy (5%; n=2) – Prior radiotherapy had been received by 47.5% (n=19) of patients. Table 1. Baseline characteristics (1.6 mg/kg n=20 | 2.0 mg/kg n=20 | All patients N=40) Median age, years | 61.5 | 64.5 | 63.0 Distant metastasis: Brain | 4 (20.0) | 3 (15.0) | 7 (17.5) Prior lines of therapy, Median, range | 3 (1–5) | 2 (1–7) | 2 (1–7) Efficacy • In the efficacy-analysis set (n=37), Tam-Peli monotherapy demonstrated promising clinical activity (Table 2; Figure 3) with a confirmed ORR of 37.8% and a DCR of 91.9%. • Clinical activity was observed in both dose cohorts, but appeared greater in the 2.0 mg/kg cohort than the 1.6 mg/kg cohort – In the 2.0 mg/kg cohort, confirmed ORR was 44.4%, DCR was 100%, and median DOR was 6.8 months – In the 1.6 mg/kg cohort, confirmed ORR was 31.6%, DCR was 84.2%, and median DOR was 4.2 months. • PFS was relatively immature but, after a median follow-up of 5.6 months, the median PFS was 5.7 months in all patients, 8.0 months in the 2.0 mg/kg cohort, and 5.5 months in the 1.6 mg/kg cohort. Table 2. Summary of response in the efficacy-analysis set* (1.6 mg/kg n=19 | 2.0 mg/kg n=18 | All patients n=37) Unconfirmed ORR, n (%) [95% CI] | 7 (36.8) [16.3–61.6] | 11 (61.1) [35.7–82.7] | 18 (48.6) [31.9–65.6] Confirmed ORR, n (%) [95% CI] | 6 (31.6) [12.6–56.6] | 8 (44.4) [21.5–69.2] | 14 (37.8) [22.5–55.2] PR | 6 (31.6) | 8 (44.4) | 14 (37.8) SD | 10 (52.6) | 10 (55.6) | 20 (54.1) PD | 2 (10.5) | 0 | 2 (5.4) NE | 1 (5.3) | 0 | 1 (2.7) DCR, % (95% CI) | 84.2 (60.4–96.6) | 100 (81.5–100) | 91.9 (78.1–98.3) Median DOR, months (95% CI) | 4.2 (4.1–NE) | 6.8 (2.9–NE) | 6.8 (2.9–NE) Median PFS, months (95% CI) | 5.5 (2.7–NE) | 8.0 (4.2–NE) | 5.7 (4.8–NE) *The efficacy-analysis set included patients who had received at least one dose of study drug. Figure 3. PFS Dose level | 1.6 mg/kg | 2.0 mg/kg Median PFS | 5.5 (2.7–NE) | 8.0 (4.2–NE) 3 Months (%) | 73.1 (46.8–87.9) | 89.2 (63.1–97.2) 6 Months (%) | 30.0 (6.6–58.7) | 58.1 (30.7–77.9) 9 Months (%) | - (NE–NE) | 29.1 (1.8–68.2) 12 Months (%) | - (NE–NE) | 0.0 (NE–NE) No. at risk 1.6 mg/kg 20 13 1 0; 2.0 mg/kg 20 16 6 1 0 Safety • The median duration of treatment was 5.6 months in all patients, 6.3 months in the 2.0 mg/kg cohort, and 4.2 months in the 1.6 mg/kg cohort. • The safety profile was generally similar between both dose cohorts (Table 3). • Overall, 36 patients (90.0%) experienced treatment-related adverse events (TRAEs) – Grade ≥3 TRAEs were reported in 14 patients (35.0%) – Two patients had a Grade 5 adverse event (AE), neither were treatment related ▪ Pneumothorax (n=1) in the 1.6 mg/kg cohort and brain edema (n=1) in the 2.0 mg/kg cohort – TRAEs led to dose interruption in 11 patients (27.5%), to dose reduction in six patients (15.0%), and to treatment discontinuation in one patient (2.5%). • Interstitial lung disease was reported in one patient (2.0 mg/kg dose cohort) which was Grade 3 (resolved) and considered treatment related by the investigator. Table 3. Safety summary (1.6 mg/kg n=20 | 2.0 mg/kg n=20 | All patients N=40) Median duration of treatment, months (range) | 4.2 (0.7–10.7) | 6.3 (0.7–10.3) | 5.6 (0.7–10.7) Median number of treatment cycles received (range) | 5.5 (1–14) | 9.0 (1–13) | 8.0 (1–14) TEAE | 20 (100) | 19 (95.0) | 39 (97.5) TRAE | 19 (95.0) | 17 (85.0) | 36 (90.0) Grade ≥3 TEAE | 13 (65.0) | 10 (50.0) | 23 (57.5) Grade ≥3 TRAE | 6 (30.0) | 8 (40.0) | 14 (35.0) Grade 5 TEAE | 1 (5.0) | 1 (5.0) | 2 (5.0) Grade 5 TRAE | 0 | 0 | 0 SAE | 10 (50.0) | 6 (30.0) | 16 (40.0) TRAE leading to Dose interruption | 5 (25.0) | 6 (30.0) | 11 (27.5) Dose reduction | 3 (15.0) | 3 (15.0) | 6 (15.0) Treatment discontinuation | 0 | 1 (5.0) | 1 (2.5) Most common TRAEs (≥10% in all patients) Fatigue | 9 (45.0) | 8 (40.0) | 17 (42.5) Nausea | 7 (35.0) | 8 (40.0) | 15 (37.5) Neutropenia | 7 (35.0) | 7 (35.0) | 14 (35.0) Alopecia | 6 (30.0) | 8 (40.0) | 14 (35.0) Anemia | 5 (25.0) | 6 (30.0) | 11 (27.5) Leukopenia | 5 (25.0) | 4 (20.0) | 9 (22.5) Diarrhea | 4 (20.0) | 5 (25.0) | 9 (22.5) Vomiting | 5 (25.0) | 3 (15.0) | 8 (20.0) Decreased appetite | 4 (20.0) | 1 (5.0) | 5 (12.5) Thrombocytopenia | 3 (15.0) | 1 (5.0) | 4 (10.0) Infusion-related reaction | 3 (15.0) | 1 (5.0) | 4 (10.0) CONCLUSIONS Tam-Peli showed encouraging antitumor activity in heavily pretreated, advanced NSCLC, with a favorable and manageable safety profile and only one case of interstitial lung disease (resolved). Clinical activity appeared greater at the 2.0 mg/kg dose than the 1.6 mg/kg dose, and the safety profile was generally similar. A dose of 2.0 mg/kg was selected for the ongoing TAISHAN-302 phase 3 study in relapsed small cell lung cancer, and for continued investigation of Tam-Peli in NSCLC. Acknowledgments: This study was sponsored by MediLink Therapeutics and F. Hoffmann-La Roche Ltd.

[Slide 1] Progression-free Survival in the Full Analysis Population No. of Events/ No. of Patients Median Progression-free Survival (95% CI) mo Osimertinib–Savolitinib 117/171 7.6 (5.6–9.7) Platinum–Pemetrexed 113/167 5.4 (4.2–5.8) Hazard ratio for disease progression or death, 0.55 (95% CI, 0.42–0.72) P<0.001 Percentage of Patients Months since Randomization Osimertinib–savolitinib Platinum–pemetrexed 36 10 ORIGINAL ARTICLE MEETING OF THE EUROPEAN SOCIETY FOR MEDICAL ONCOLOGY Osimertinib plus Savolitinib in Advanced NSCLC S. Lu and Others In patients with EGFR-mutated and MET-altered non–small-cell lung cancer that had progressed during osimertinib therapy, osimertinib plus savolitinib (a MET tyrosine kinase inhibitor) led to longer survival than platinum–pemetrexed. Survival table Endpoint: Progression-free survival (full analysis population) Arm | N | Median | Median CI Osimertinib–savolitinib | 171 | 7.6 mo | 5.6–9.7 Platinum–pemetrexed | 167 | 5.4 mo | 4.2–5.8 HR: 0.55 HR CI: 0.42–0.72 P: <0.001 12-month annotation printed on curve: 36 (osimertinib–savolitinib) vs 10 (platinum–pemetrexed), percent [Slide 2] Percentage of Patients without an Event Months Divarasib Control 65.7 47.8 43.9 24.0 ORIGINAL ARTICLE MEETING OF THE EUROPEAN SOCIETY FOR MEDICAL ONCOLOGY Divarasib for KRAS G12C–Mutated NSCLC F. Skoulidis and Others In patients with previously treated advanced non–small-cell lung cancer with a KRAS G12C mutation, median progression-free survival was 9.8 months with the KRAS G12C inhibitor divarasib and 5.7 months with sotorasib or adagrasib. Survival table Endpoint: Progression-free survival (figure axis: percentage of patients without an event) Divarasib | 9.8 mo Control (sotorasib or adagrasib) | 5.7 mo 6-month annotation printed on curve: 65.7 (divarasib) vs 47.8 (control), percent 12-month annotation printed on curve: 43.9 (divarasib) vs 24.0 (control), percent [Slide 3] Percentage of Patients Months Sacituzumab tirumotecan + pembrolizumab 51.1 (95% CI, 41.5–59.9) Chemotherapy + pembrolizumab 22.8 (95% CI, 15.3–31.3) ORIGINAL ARTICLE MEETING OF THE EUROPEAN SOCIETY FOR MEDICAL ONCOLOGY Sacituzumab Tirumotecan–Pembrolizumab for NSCLC F. Zhou and Others Among patients with PD-L1–negative advanced nonsquamous non–small-cell lung cancer, progression-free survival was longer with sacituzumab tirumotecan plus pembrolizumab than with chemotherapy plus pembrolizumab (15.0 vs. 5.7 months). Survival table Endpoint: Progression-free survival Sacituzumab tirumotecan + pembrolizumab | 15.0 mo Chemotherapy + pembrolizumab | 5.7 mo 12-month annotation printed on curve: 51.1% (95% CI, 41.5–59.9) vs 22.8% (95% CI, 15.3–31.3) [Slide 4] Percentage of Patients Months Tarlatamab–durvalumab Durvalumab 91.5 79.4 74.6 53.3 ORIGINAL ARTICLE MEETING OF THE EUROPEAN SOCIETY FOR MEDICAL ONCOLOGY Tarlatamab–Durvalumab in Small-Cell Lung Cancer M. Pérol and Others In a prespecified interim analysis of a phase 3 trial involving patients with extensive-stage SCLC, the median overall survival was 21.1 months with tarlatamab–durvalumab maintenance therapy and 12.7 months with durvalumab alone. Survival table Endpoint: Overall survival (prespecified interim analysis; figure axis unlabeled beyond percentage of patients) Tarlatamab–durvalumab | 21.1 mo Durvalumab | 12.7 mo 6-month annotation printed on curve: 91.5 (tarlatamab–durvalumab) vs 74.6 (durvalumab), percent 12-month annotation printed on curve: 79.4 vs 53.3, percent

[Slide 2] Progression-free Survival as Assessed by Blinded Independent Central Review Percentage of Patients (0-100) 58.5 | Anbenitamab plus HB1801 34.4 | THP Hazard ratio for disease progression or death, 0.46 (95% CI, 0.37 to 0.56) P<0.001 Months since Randomization (0, 3, 6, 9, 12, 15, 18, 21, 24, 27, 30, 33) ORIGINAL ARTICLE MEETING OF THE EUROPEAN SOCIETY FOR MEDICAL ONCOLOGY Anbenitamab plus HB1801 for Advanced Breast Cancer J. Wang and Others Survival table Endpoint: PFS by BICR Anbenitamab plus HB1801 THP HR: 0.46 HR CI: 0.37 to 0.56 P: P<0.001 Landmark rates: 24 months: Anbenitamab plus HB1801 58.5 | THP 34.4 [Slide 3] Percentage of Patients (0-100) 73.1 | 63.5 | Durvalumab-EV 50.3 | 45.1 | Control Months since Randomization (0, 3, 6, 9, 12, 15, 18, 21, 24, 27, 30, 33, 36, 39, 42, 45) ORIGINAL ARTICLE MEETING OF THE EUROPEAN SOCIETY FOR MEDICAL ONCOLOGY Durvalumab Regimens in Invasive Bladder Cancer T. Powles and Others Survival table Endpoint: Not printed (likely EFS; curve title not shown) Durvalumab-EV Control Landmark rates: 12 months: Durvalumab-EV 73.1 | Control 50.3 24 months: Durvalumab-EV 63.5 | Control 45.1

THE NEW ENGLAND JOURNAL of MEDICINE ORIGINAL ARTICLE Tarlatamab–Durvalumab Maintenance for Extensive-Stage Small-Cell Lung Cancer THE NEW ENGLAND JOURNAL of MEDICINE ORIGINAL ARTICLE Sacituzumab Tirumotecan plus Pembrolizumab in PD-L1–Negative Advanced NSCLC Primary results from the Krascendo 1 study are available online now at the New England Journal of Medicine Divarasib as Compared with Sotorasib or Adagrasib in KRAS G12C–Mutated NSCLC THE NEW ENGLAND JOURNAL of MEDICINE ORIGINAL ARTICLE Osimertinib–Savolitinib in Advanced NSCLC after Progression during EGFR-TKI Therapy Publication THE LANCET Izalontamab brengitecan versus platinum-based chemotherapy in EGFR-mutated advanced non-squamous non-small-cell lung cancer after progression on third-generation EGFR tyrosine kinase inhibitors (PANKU-Lung01): a multicentre, randomised, open-label, phase 3 trial in China https://doi.org/10.1016/S0140-6736(26)02026-X THE LANCET Ruzaltatug rezetecan versus chemotherapy for EGFR-mutated advanced non-small-cell lung cancer progressing after EGFR tyrosine kinase inhibitor therapy (HERON-Lung01): interim analysis of a randomised, open-label, phase 3 trial in China thelancet.com/journals/lancet/article/PIIS0140-6736(26)01899-4/fulltext Pembrolizumab with or without platinum-based chemotherapy as first-line treatment in advanced NSCLC expressing PD-L1 ≥50%: The PERSEE randomized phase III trial (GFPC 01-2020) DOI: 10.1016/j.annonc.2026.10.108 Simultaneous publication Extended follow-up of tarlatamab in patients with small-cell lung cancer after platinum-based chemotherapy in the randomized phase III DeLLphi-304 study --- The article is simultaneously published in Annals of Oncology DOI: 10.1016/j.annonc.2026.10.094 Conclusions • In NeoCOAST-2, novel perioperative combination regimens demonstrated prolonged EFS, with 24-month EFS rates ranging from 70.5% to 77.9% across arms. • Neoadjuvant ctDNA clearance was associated with improved EFS in all arms, supporting previous analyses and highlighting neoadjuvant ctDNA clearance as a potential prognostic biomarker. – Patients with both ctDNA clearance at CT and pCR had the longest EFS. • No new safety signals were identified during adjuvant treatment when compared with the primary analysis. • The findings available to date, including pCR rates reported in the primary analysis and the EFS and ctDNA data reported here, warrant further evaluation of combinations with Dato-DXd in resectable NSCLC. – In this analysis, patients in Arm 4 had the numerically highest ctDNA clearance rate. A manuscript containing additional EFS, ctDNA, and safety data from these analyses has been published simultaneously in Nature (McGrath L, et al. Nature 2026) and can be accessed via the QR code. nature medicine Neoadjuvant durvalumab plus chemotherapy in resectable or borderline resectable non-small-cell lung cancer: the phase 2 MDT-BRIDGE trial Dalmelitinib plus Aumolertinib in Advanced NSCLC Patients With EGFR-TKI-Resistance and MET-Amplification: A Randomized Phase 3 Trial Primary Endpoint-PFS by BICR HR 0.60 40% reduction in risk of disease progression or death Relevance Journal of Clinical Oncology

[Slide 1] Median follow-up: 48.2 months Median(95% CI): HR (95% CI) P-value Toripalimab NR(51.3,NE) 0.52 (0.391,0.677) <0.0001 Similar to all tria[cut off] Event-free survival,% of patients Toripalimab Placebo 71.5 63.1 45.5 40.6 PD-L1 expression evaluated using a validated JS[cut off] • EFS PD-L1 1-49% (n=164), HR 0.41 (0.26-0.6[cut off] • EFS PD-L1 ≥ 50% (n=157), HR 0.47 (0.26-0.8[cut off] Lu – ESMO 2026 * Lu – JAMA 2024 (EFS interim ana[cut off] MADRID 2026 ESMO congress MADRID SPAIN 23-27 OCTOBER 2026 Survival table Endpoint: EFS (NEOTORCH, toripalimab vs placebo) Toripalimab | NR | 51.3–NE Placebo | [illegible] | [illegible] HR: 0.52 HR CI: 0.391–0.677 P: <0.0001 24-month EFS: 71.5% vs 45.5% 36-month EFS: 63.1% vs 40.6% [Slide 2] NEOTORCH in the current treatment landscape Neoadjuvant | Perioperative pCR CheckMate 816 — 24% vs. 2.2% — EFS HR: 0.65 (0.44-0.96); OS HR: 0.72 (0.52-0.99) — bars 64 / 45, 65 / 55 — 2y EFS / 5y OS — All PDL1 (US flag), PDL1 ≥1% (EU flag) — CT + Nivo → S / CT → Surgery (S) KEYNOTE 671 — 18% vs. 4.0% — EFS HR: 0.59 (0.46-0.72); OS HR: 0.72 (0.56-0.93) — bars 61 / 42, 65 / 54 — 2y EFS / 5y OS — All PDL1, All PDL1 — CT + Pemb → S → P / CT → S → Placebo AEGEAN — 17.2% vs. 4.3% — EFS HR: 0.69 (0.55-0.88); OS HR: 0.89 (0.70-1.14) — bars 63 / 52 — 2y EFS / 5y OS — All PDL1, All PDL1 — CT +Durv → S → D / CT → S → Placebo CheckMate 77T — 25.3% vs. 4.7% — EFS HR: 0.61 (0.46-0.80); OS HR: 0.85 (0.58-1.25) — bars 67 / 44 — 2y EFS / 5y OS — All PDL1, PDL1 ≥1% — CT+Nivo → S → Nivo / CT → S → Placebo RATIONALE 315 — 41% vs. 5.7% — EFS HR: 0.58 (0.43-0.79); OS HR: 0.65 (0.45-0.93) — bars 67 / 52, 72 / 62 — 2y EFS / 4y OS — All PDL1 — CT + Tisle → S → Ti / CT → S → Placebo NEOTORCH — 25.6% vs. 1.2% — EFS HR: 0.52 (0.39-0.67); OS HR: 0.66 (0.48-0.89) — bars 72 / 46 — 2y EFS / 5y OS — CT+T → S → CT + T / CT → S → Placebo ADEBRELIMAB — 31.1% vs.7.6% — EFS HR: 0.52 (0.38-0.72); OS HR: 0.57 (0.36-0.88) — bars 75 / 56 — 2y EFS / 5y OS — CT + Ade → S → A / CT → S → Placebo IMpower030 — 31% vs.8.7% — EFS HR: 0.77 (0.58-1.02); OS HR: 0.77 (0.57-1.05) — bars 63 / 57, 66 / 60 — 2y EFS / 5y OS — CT+Atz → S → Atz / CT → S → BSC Immunotherapy arm (different colors according the drug tested / control arm. Overall survival highlighted refers to final OS of the trial, for other trials is still immature and not reported. CT: chemotherapy. S: Surgery Flags indicates FDA / EMA approvals. Nivo: nivolumab. Pem (P): pembrolizumab. Durv (D): durvalumab. Tisle (Ti): Tislelizuamb. T: toripalimab. Ade (A): adebrelimab. Atz: Atezolizumab.pCR: pathological complete rate. BSc: Best Supportive Care Remon– Lancet Resp Med 2026, modified [Slide 3] Similar EFS and OS across neoadjuvant and peri-op trials CheckMate 816: Nivolumab, 9 weeks NIVO + chemo (n = 179) Chemo (n = 179) Median OS, mo NR 73.7 HR (95% CI); P value 0.72 (0.523-0.998); 0.0479 (51%) anti-PD(L)-1 at relapse 5y OS: 65% vs. 55% 2y EFS: 64% vs. 45% pCR: 24% vs. 2.2% KEYNOTE 671: Pembrolizumab, 52 weeks Pts w/ Event / Median (95% CI), mo / HR (95% CI) Pembro 35.8% NR (NR–NR) 0.74 (0.59–0.92) Placebo 45.5% 70.7 (53.7–NR) 5 Years (65%) anti-PD(L)-1 at relapse 5y OS: 64.6% vs. 53.6% 2y EFS: 61% vs. 43% pCR: 18% vs. 4.0% RATIONALE 315: Tislelizumab, 56 weeks ITT Analysis Set Tislelizumab / Placebo Events, n (%) 52 (23.0) 70 (30.8) Median OS, months (95% CI) NR (NE, NE) NR (NE, NE) (19%) anti-PD(L)-1 at relapse 4y OS: 72% vs. 62% 2y EFS: 67% vs. 52% pCR: 41% vs. 5.7% NEOTORCH: Toripalimab, 51 weeks Toripalimab Placebo Median(95% CI): NR ( NE, NE) NR (58.6, NE) HR (95% CI) 0.66(0.49, 0.89) (48%) anti-PD(L)-1 at relapse 3y OS: 78% vs. 67% 2y EFS: 72% vs. 46% pCR: 25.6% vs. 1.2% Forde – NEJM 2025 * Wakelee – Annals of Oncol 2026 * Wang – Annals of Oncol 2025 * Lu - ESMO 2026 (cross trial comparisons should be undertaken with caution) [Slide 4] What is the added value of ADJ therapy? Received adjuvant treatment | Did not receive adjuvant treatment CheckMate 77T Nivo arm, ADJ Nivo: YES Placebo arm, ADJ Placebo: YES HR among those with adjuvant treatment, 0.43 (95% CI, 0.28–0.67) Cascone – NEJM 2024 (From surgery) Nivo arm, ADJ Nivo: NO Placebo arm, ADJ Placebo: NO HR among those without adjuvant treatment, 0.44 (95% CI, 0.22–0.87) KEYNOTE 671 Pembro arm, ADJ Pembro: YES Placebo arm, ADJ Placebo: YES HR 0·54 (95% CI 0·41–0·70) Spicer – Lancet 2024 (From surgery) Pembro arm, ADJ Pembro: NO Placebo arm, ADJ Placebo: NO AEGEAN Durva arm, ADJ Durva: YES — 88.4%, 79.2%, 72.8% Placebo arm, ADJ Placebo: YES — 80.5%, 68.2%, 61.6% Heymach – WCLC 2024 (From Randomization) Durva arm, ADJ Durva: NO — 35.9%, 28.7%, 28.7% Placebo arm, ADJ Placebo: NO — 27.5%, 23.2%, 16.5%

[Slide 2] KRASi and the de-addiction landscape EGFR mutated Osimertinib, 1st line | ORR 80% ALK+ Lorlatinib 1st line | ORR 90%, ROS1+ Taletrectinib 1st line | ORR 88% BRAF V600 Encorafenib+ binimetinib 1st line | ORR 75% KRAS G12C Sotorasib 2nd line | ORR 28% KRAS G12C Adagrasib 2nd line | ORR 32% KRASCENDO-1 KRAS G12C Sotorasib/Adagrasib 2nd line | ORR 32% KRAS G12C Divarasib 2nd line | ORR 58% [waterfall plots; axis labels and small legends not transcribed] Benjamin BESSE Soria NEJM 2018, Perol JCO 2025, Riely JCO 2023, Solomon Lancet Oncol 2018, De Langen Lancet 2023, Barlesi Lancet 2025, Skoulidis ESMO 2026 [Slide 3] Not all KRAS G12Ci are IO friendly ! Liver toxicity when IO is added ALT / AST, any grade % · dark = grade ≥3 Immune-mediated toxicity in addition to KRAS-induced toxicity Chemo increased the liver tox but grade 3 is rather stable Monotherapy (ALT / AST, any grade; grade ≥3 where shown): Olomorasib | 10 | 9 Calderasib | 11 (3) | 12 (3) Divarasib | 15 (6) | 14 (5) Elisrasib | 16 | 14 Adagrasib | 28 (4) | 25 (3) + Pembrolizumab: Olomorasib | 25 (12) | 23 (10) Calderasib | 32 (9) | 35 (7) Adagrasib | 40 (11) | 36 (14) Elisrasib | 48 (8) | 40 (6) Divarasib | 53 (23) | 49 (18) + Chemo-pembro: Olomorasib | 29 (12) | 33 (10) Calderasib | 44 (11) | 44 (9) Half-life: Adagrasib ~23 h > divarasib ~17.6 h > sotorasib ~5 h > olomorasib ~3 h. COURTESY OF M.ALDEA *Treatment-related AEs. Mono: calderasib/olomorasib all tumors, elisrasib 1L, others 2L+ NSCLC. Calderasib/elisrasib read from figures. Gray ESMO26; Murciano-Goroff NatCommun26; Burns JTO26; Negrao ESMO26; Sacher JCO25; Skoulidis ASCO26; Jänne NEJM22, ASCO25; Lu ASCO26; de Langen Lancet23; Shi NatMed25 Benjamin BESSE IO : immunotherapy [Slide 4] KRAS G12Ci First line PD-L1 ≥50% — IO backbone without chemotherapy KRYSTAL-7 | NCT04613596 | Adagrasib + pembrolizumab vs pembrolizumab | PRIMARY PFS, OS | TIMING Dec 2020 – Oct 2029 KANDLELIT-004 | NCT06345729 | Calderasib + pembrolizumab vs placebo + pembrolizumab | PRIMARY PFS, OS | TIMING May 2024 – Feb 2031 SUNRAY-01 | NCT06119581 | Olomorasib + pembrolizumab vs placebo + pembrolizumab | PRIMARY PFS | TIMING Dec 2023 – Oct 2029 Is IO alone the optimal control arm in PD-L1-high disease? ALL COMERS — IO ± chemotherapy backbone SUNRAY-01 | NCT06119581 | Olomorasib + pembrolizumab + chemo vs pembrolizumab + chemo | PRIMARY PFS | TIMING Dec 2023 – Oct 2029 KANDLELIT-007 | NCT07190248 | Calderasib + SC pembrolizumab vs SC pembrolizumab + chemo | PRIMARY PFS (PD-L1 ≥1%) | TIMING Oct 2025 – Aug 2032 KRASCENDO-2 | NCT06793215 | Divarasib + pembrolizumab vs pembrolizumab + chemo | PRIMARY PFS, OS | TIMING Oct 2025 – Oct 2030 KRYSTAL-4 | NCT06875310 | Adagrasib + pembrolizumab + chemo vs pembrolizumab + chemo | PRIMARY PFS, OS | TIMING Apr 2025 – Apr 2032 PD-L1 <1% — chemotherapy backbone CodeBreaK 202 | NCT05920356 | Sotorasib + chemo vs pembrolizumab + chemo | PFS, OS • Nov 2023 – Jun 2031 IO = immunotherapy; CT = chemotherapy; SC = subcutaneous; 1L = first line. Available at: https://clinicaltrials.gov Benjamin BESSE

[Slide 1] CHAIRS: PATRICK FORDE, ANTONIO PASSARO Study Design HERON-Lung01: an open-label, randomized, phase 3 trial (NCT06671379) Key eligibility criteria • Locally advanced/metastatic Nsq NSCLC • Confirmed EGFR mutation (L858R/Ex19del*) • Progression during/after EGFR-TKI – Prior 3rd gen TKI required for T790M tumors • No other prior systemic therapy permitted† • ECOG PS 0–1 Stratification factors • Brain metastases (yes vs no) • EGFR mutation type (L858R vs Ex19del) • Prior first-line 3rd gen EGFR-TKI (yes vs no) R 1:1 Ruzaltatug rezetecan 8 mg/kg Q3W, IV Carboplatin AUC5/ Cisplatin 75 mg/m2 Q3W for 4 cycles + Pemetrexed 500 mg/m2 Q3W Crossover permitted at PD Ruzaltatug rezetecan 8 mg/kg Q3W, IV Primary endpoint • PFS per BICR Secondary endpoints • OS • PFS per INV • ORR per BICR and INV • DCR per BICR and INV • DoR per BICR and INV • Safety *EGFR mutation status was based on tissue or blood testing. †Prior anti-angiogenic agents in combination with EGFR-TKI were permitted. BICR, blinded independent central review; DCR, disease control rate; DoR, duration of response; ECOG PS, Eastern Cooperative Oncology Group performance status; Gen, generation; INV, investigator; Nsq, non-squamous; ORR, objective response rate; OS, overall survival; PD, progressive disease; PFS, progression-free survival Chongrui Xu Ruzaltatug rezetecan, a HER3-directed ADC, vs platinum-based chemotherapy (PBC) in EGFR-mutated advanced NSCLC progressing after EGFR-TKI: Interim analyses (IA) from the phase 3 HERON-Lung01 trial [Slide 2] CHAIRS: GIUSEPPE CURIGLIANO, YELENA JANJIGIAN SAFFRON study design Key inclusion criteria • Aged ≥18 years (≥19 years in the Republic of Korea) • EGFRm (Ex19del, L858R, and/or T790M) locally advanced or metastatic NSCLC • MET overexpression (IHC3+≥90%) and/or amplification (FISH10+)* centrally confirmed with post-progression tissue biopsy • Progression on 1L or 2L osimertinib as the most recent anticancer treatment† • ECOG PS 0 / 1 • Stable CNS metastases were allowed N = 324 R 1:1 Stratified by: Race (Asian / non-Asian) Baseline brain metastases (Yes / no) Line of treatment (2L / 3L) Osimertinib 80 mg QD + savolitinib 300 mg BID‡ Treatment beyond PD allowed per investigator discretion¶ Pemetrexed 500 mg/m2 + carboplatin AUC5 / cisplatin 75 mg/m2 (Q3W for 4 cycles) Maintenance pemetrexed 500 mg/m2 (Q3W)‡ Investigator-assessed PD (per RECIST v1.1) or discontinuation Endpoints Primary: PFS by BICR per RECIST v1.1 in the FAS Key secondary / exploratory: • PFS by investigator, OS, ORR, DoR, DCR and tumour shrinkage in the FAS • PFS in patients in the MET-overexpressed (IHC3+≥90%) or MET-amplified (FISH10+) analysis sets • Safety in the SAF *MET IHC and FISH cut-offs were identified in preliminary analyses of the Phase II SAVANNAH study.1 Shun Lu Osimertinib (osi) + savolitinib (savo) vs platinum–pemetrexed (plat–pem) in EGFRm MET-overexpressed (OE) and/or -amplified (AMP) advanced NSCLC post-osi: SAFFRON phase (Ph) 3 primary results [Slide 3] 10:15 - 11:45 Proffered paper 1: NSCLC, metastatic CHAIR: SURESH SENAN OptiTROP-Lung04 Study Design Randomized, multicenter, open-label, phase 3 trial (NCT05870319) Key Eligibility • ECOG score 0 or 1 • Nsq-NSCLC (stage IIIB/IIIC or stage IV) • EGFR-sensitive mutations • Progression after 3rd gen TKI therapy or progression after 1st or 2nd gen TKIs with negative T790M R 1:1 N=376 Sac-TMT 5 mg/kg IV, Q2W (n=188) • Pemetrexed 500 mg/m2 + Carboplatin AUC 5 or Cisplatin 75 mg/m2 Q3W for up to 4 cycles • Pemetrexed 500 mg/m2 maintenance, Q3W (n=188) Treatment until disease progression, intolerable toxicity, or patient request to discontinue treatment. Primary endpoints* • PFS assessed by BICR Secondary endpoints* • OS (key secondary endpoint) • PFS assessed by investigator • ORR, DCR, DOR, etc. • Safety Stratification factors: 1. Prior EGFR-TKI therapy (3rd gen TKI in 1st line vs in 2nd line vs no 3rd gen TKI) 2. Brain metastases (yes vs no) Pre-specified final analysis for OS: • approximately 250 OS events have occurred *Tumor response was assessed using RECIST version 1.1. BICR, blinded independent central review; DCR, disease control rate; DOR, duration of response; ECOG, Eastern Cooperative Oncology Group; IA, interim analysis; ORR, objective response rate; R, randomization; RECIST, Response Evaluation Criteria in Solid Tumors. Professor Wenfeng Fang Wen Feng Fang Final OS analysis of sacituzumab tirumotecan (sac-TMT) versus platinum-based chemotherapy in EGFR-mutated non-small cell lung cancer (NSCLC) after progression on EGFR-TKI: Phase III OptiTROP-Lung04 study ALICANTE AUDITORIUM - HALL 6 [Slide 4] CHAIRS: PATRICK FORDE, ANTONIO PASSARO PANKU-Lung01 Study Design PANKU-Lung01 study is an open-label, multicenter, randomized, phase III study Key eligibility criteria • Locally advanced or metastatic non-squamous NSCLC • Documented EGFR sensitizing mutant (exon 19 deletion or L858R) • Progressed on prior 3rd-generation EGFR TKI therapy (up to 2 prior lines of EGFR TKI) • Treated/stable CNS metastasis allowed • Measurable lesion per RECIST v1.1 • ECOG PS 0-1 R 1:1 N=432 Iza-bren 2.5 mg/kg D1D8 Q3W PBC Pemetrexed + Cisplatin/Carboplatin Q3W Stratified by • ECOG PS (0 vs 1) • CNS metastases (Yes vs No) • Prior anti-angiogenic therapy (Yes vs No) Treatment until Disease progression per RECIST v1.1 or intolerable toxicity Primary endpoint: PFS (BICR) Key secondary endpoint: OS HPFS α=0.025 → HOS α=0 Other secondary endpoints • PFS (INV) • ORR (BICR & INV) • DCR (BICR & INV) • DOR (BICR & INV) • Safety, PK, Immunogenicity Note: Iza-bren was dosed with 10% compensation per protocol; pemetrexed 500 mg/m2 plus cisplatin 75 mg/m2 or pemetrexed 500 mg/m2 plus carboplatin (AUC 5) Q3W. ECOG PS, Eastern Cooperative Oncology Group performance status; Q3W, every 3 weeks; AUC, area under the curve; CNS, central nervous system PFS, progression-free survival; BICR, blinded independent central review; OS, overall survival; INV, investigator assessed; ORR, objective response rate (confirmed); DCR, disease control rate; DOR, duration of response. Li Zhang PANKU-Lung01: Randomized phase III study of izalontamab brengitecan (iza-bren) in EGFR-mutant non-squamous non-small cell lung cancer (NSCLC) after progression on a third-generation EGFR TKI

Randomized trials of the resistance "agnostic" approach in molecularly unselected patients MARIPOSA-2(1-3) mOS (months): 17.7 | 15.3 mPFS (months): 6.3 | 4.2 ORR: 64% | 36% Legend: Amivantamab +ChT (n=131) | ChT (n=263) OptiTROP-Lung04(4,5) mOS (months): NR | 17.4 -> HR 0.60 (0.44-0.82) P<0.001 mPFS (months): 8.3 | 4.3 ORR: 60.6% | 43.1% Legend: Sac-TMT +ChT (n=188) | ChT (n=188) HARMONi(6) mOS (months): 16.8 | 14 -> HR 0.76 (0.61-0.95) P<0.01 mPFS (months): 6.8 | 4.4 Legend: Ivonescimab +ChT (n=219) | Placebo +ChT (n=219) COMPEL(7) mOS (months): 15.9 | 9.8 mPFS (months): 8.4 | 4.4 Legend: Osimertinib +ChT (n=49) | Placebo +ChT (n=49) 1. Popat S et al. ESMO. 2024; 2. Passaro A et al. ESMO.2023; 3. Passaro et al. Ann Oncol. 2024; 4. Zhang L et al. ESMO Congress. 2025; 5. Fang W et al. NEJM. 2026;394:13-26; 6. Passaro A. et al. WCLC. 2026; 7. Pasello G et al. WCLC. 2025; Natasha B. Leighl

[Slide 1] NEOTORCH vs. other Chinese and global Trials Strategy | Neoadjuvant | Perioperative Trial | CheckMate-816 | KEYNOTE-671 | AEGEAN | CheckMate-77T | ADEBRELIMAB | RATIONALE-315 | NEOTORCH | IMpower 030 Nº of patients | 358 | 797 | 740 | 461 | 501 | 453 | 501 | 440 IO drug | Nivolumab | Pembrolizumab | Durvalumab* | Nivolumab | Adebrelimab* | Tislelizumab | Toripalimab | Atezolizumab* Stage III | 7th TNM 64% | 8th TNM 70% | 8th TNM 71% | 8th TNM 64% | 8th TNM 63% | 8th TNM 58% (IIIA) | 8th TNM 81% | 8th TNM 59% PDL1 negative | 43% | 33% | 33% | 40% | 45% | 39% | 26% | 33% Non-Sq | 49% | 57% | 51% | 49% | 21% | 20% | 22% | 55% Cisplatin | 80% | 100% | 27% | 21% | NR | 54% | NR | NR Surgery cancelled (IO arm) | 17% | 18% | 19% | 22% | 11% | 16% | 16% | 11% Surgery cancelled (control arm) | 21% | 21% | 20% | 24% | 17% | 24% | 25% | 10% Pneumonectomy (% in IO arm) | 21% | 12% | 9% | 9% | NR | 8% | 9% | 9% pCR CTIO vs CT | 24% vs. 2.2% | 18% vs. 4.0% | 17.2% vs. 4.3% | 25.3% vs. 4.7% | 31% vs. 7.6% | 41% vs. 5.7% | 25.6% vs. 1.2% | 31% vs. 8.7% Remon – Lancet Resp Med 2026, modified (*: Anti-PD-L1). Note: In the AEGEAN trial upfront pneumonectomies were an ineligibility criterion. (NEOTORCH: Mediastinal lymph node: PET only 85%, 12% biopsy) MADRID 2026 ESMO congress [Slide 3] NEOTORCH in the current treatment landscape Neoadjuvant | Perioperative pCR CheckMate 816 2y EFS / 5y OS: pCR 24% vs. 2.2%; EFS HR: 0.65 (0.44-0.96); OS HR: 0.72 (0.52-0.99); bars 64, 45, 65, 55; All PDL1 / PDL1 ≥1%; CT + Nivo → S / CT → Surgery (S) KEYNOTE 671 2y EFS / 5y OS: pCR 18% vs. 4.0%; EFS HR: 0.59 (0.46-0.72); OS HR: 0.72 (0.56-0.93); bars 61, 42, 65, 54; All PDL1 / All PDL1; CT + Pemb → S → P / CT → S → Placebo AEGEAN 2y EFS / 5y OS: pCR 17.2% vs. 4.3%; EFS HR: 0.69 (0.55-0.88); OS HR: 0.89 (0.70-1.14); bars 63, 52; All PDL1 / All PDL1; CT +Durv → S → D / CT → S → Placebo CheckMate 77T 2y EFS / 5y OS: pCR 25.3% vs. 4.7%; EFS HR: 0.61 (0.46-0.80); OS HR: 0.85 (0.58-1.25); bars 67, 44; All PDL1 / PDL1 ≥1%; CT+Nivo → S → Nivo / CT → S → Placebo RATIONALE 315 2y EFS / 4y OS: pCR 41% vs. 5.7%; EFS HR: 0.58 (0.43-0.79); OS HR: 0.65 (0.45-0.93); bars 67, 52, 72, 62; All PDL1; CT + Tisle → S → Ti / CT → S → Placebo NEOTORCH 2y EFS / 5y OS: pCR 25.6% vs. 1.2%; EFS HR: 0.52 (0.39-0.67); OS HR: 0.66 (0.48-0.89); bars 72, 46; CT+T → S → CT+T / CT → S → Placebo ADEBRELIMAB 2y EFS / 5y OS: pCR 31.1% vs.7.6%; EFS HR: 0.52 (0.38-0.72); OS HR: 0.57 (0.36-0.88); bars 75, 56; CT + Ade → S → A / CT → S → Placebo IMpower030 2y EFS / 5y OS: pCR 31% vs.8.7%; EFS HR: 0.77 (0.58-1.02); OS HR: 0.77 (0.57-1.05); bars 63, 57, 66, 60; CT+Atz → S → Atz / CT → S → BSC In Chinese trials compared to global trials: • Better HR for EFS / OS. Is higher benefit it related to more squamous? • No clear evidence of increase EFS benefit in PD-L1 ≥50% vs. PD-L1 1–49%. Remon– Lancet Resp Med 2026, modified

The rules for my clinic next week • For a patient with driver-negative oligometastatic NSCLC: I will no longer routinely refer for local consolidative RT/SABR solely because a patient meets oligometastatic criteria • For a patient with PD-L1 ≥50% advanced NSCLC: I will have a much lower threshold to discuss pembrolizumab + chemotherapy • For everyone: The future of treatment intensification for advanced NSCLC will be predicated on enrichment strategies grounded in better understanding of tumor biology that will go beyond genomic alterations, PD-L1 cutoffs or number of sites of disease

Conclusions 1 Will perioperative toripalimab change our SoC? (NEOTORCH trial) Toripalimab is a new drug in the perioperative arsenal. Whether data can be generalized is unknown, as well as whether this "new me too" will help to reduce economic impact of treatment 2 Are there predictive biomarkers for perioperative CTIO (KEYNOTE 671 trial) Efficacy of perioperative pembrolizumab occurs regardless of T-cell signature, TMB, KRASm status. We need new treatments to increase pCR rates and biomarkers to predict pCR 3 Will adjuvant pembrolizumab remain a SoC? (PEARLS trial) Adjuvant pembrolizumab improves DFS without OS benefit. The role of sequential adjuvant immunotherapy remains unclear (with or without preoperative therapy)

[Slide 1] What do we mean when we say MET+ ? Gene Copy Number Polysomy Centromere MET GCN4 CEP7 MET to CEP7 ratio 1 Amplification Amplification GCN4 MET to CEP7 ratio 4 Protein Overexpression c-MET overexpression Guo et al Nat Rev Clin Onc 2020; Bradley et al Nat Rev Clin Onc 2017; Tsao et al NPJ Prec Onc 2025 [Slide 2] Phase III trials of EGFR + MET in acquired resistance | SACHI | SAFFRON | Dalmelitinib/Aumolertinib EGFR + MET | Osimertinib + Savolitinib | Osimertinib + Savolitinib | Aumolertinib + Dalmelitinib Enrollment | China | Global | China Biomarker testing | Post-progression tissue biopsy | Post-progression tissue biopsy | Post-progression tissue biopsy RR | 63% v 36% | 51% v 27% | 59.6% v 38.6% PFS | 8.2 mo v 4.5 mo HR 0.34 p < 0.0001 | 7.6 mo v 5.4 mo HR 0.55 (0.42, 0.72) p <0.001 | 6.2 mo v 4.4 mo HR 0.60 (0.45,0.81) p < 0.001 OS | 22.9 mo v 17.7 mo HR 0.84 (0.55-1.29) | 17.8 mo v 12.9 mo HR 0.68 (0.50,0.91) p = 0.010 | NR v 19.6 mo HR 0.84 (0.52, 1.35) In chemo arm: Received subsequent MET Rx | 52% | 13% | 51.9% Prior exposure to 3rd gen EGFR TKI | 35% | 100% | 95% MET definition | GCN ≥ 5 or MET:CEP7 ≥ 2 If prior 3rd gen EGFR TKI: GCN ≥ 10 | FISH ≥ 10 and/or MET IHC 3+ ≥ 90% | MET GCN ≥ 5 or MET:CEP7 ≥ 2 % with GCN ≥ 10 | 71% | 72% | 33% Lu et al Lancet 2026; Lu et al ESMO 2026; Lu et al ESMO 2026 [Slide 3] SANOVO: IHC3+ population Λ: Median PFS FLAURA Osimertinib Λ 18.9 38.6 Gefitinib/Erlotinib Λ 10.2 31.8 FLAURA2 Osimertinib + Chemotherapy Λ 25.5 47.5 Osimertinib Λ 16.7 37.6 MARIPOSA Amivantamab-Lazertinib Λ 23.7 N.R. Osimertinib Λ 16.6 36.7 SANOVO Osimertinib-Savolitinib Λ 27.9 44.5 Osimertinib Λ 16.4 34.9 0 10 20 30 40 50 Median PFS in 3+. Osimertinib + Savolitinib 22.2 mo Osimertinib 13.7 mo [Slide 4] Differential Adverse Event Profiles SANOVO Peripheral Edema Hypoalbuminemia Platelet count decreased Anemia MARIPOSA Paronychia Rash/Dermatitis acneiform Diarrhea Stomatitis FLAURA2 Anemia Diarrhea Nausea Neutropenia Wu et al ESMO 2026; Janne et al NEJM 2026; Yang et al NEJM 2025

[Slide 1] EUROPEAN SOCIETY FOR MEDICAL ONCOLOGY 2026 CONGRESS ORIGINAL ARTICLE | OCT 25, 2026 | NEJM.ORG Sacituzumab Tirumotecan plus Pembrolizumab in PD-L1–Negative Advanced NSCLC (OptiTROP-Lung06) Sacituzumab tirumotecan + pembrolizumab: 51.1 (95% CI, 41.5–59.9) Chemotherapy + pembrolizumab: 22.8 (95% CI, 15.3–31.3) Months 0–18 Figure 1. Progression-free Survival as Assessed by Blinded Independent Central Review. The NEW ENGLAND JOURNAL of MEDICINE Survival table Endpoint: Progression-free survival by BICR Sacituzumab tirumotecan + pembrolizumab Chemotherapy + pembrolizumab Landmark rates: 12 months: Sacituzumab tirumotecan + pembrolizumab 51.1% (CI 41.5–59.9) | Chemotherapy + pembrolizumab 22.8% (CI 15.3–31.3) [Slide 4] The NEW ENGLAND JOURNAL of MEDICINE Progression-free Survival in the Full Analysis Population No. of Events/No. of Patients | Median Progression-free Survival (95% CI), mo Osimertinib–Savolitinib | 117/171 | 7.6 (5.6–9.7) Platinum–Pemetrexed | 113/167 | 5.4 (4.2–5.8) Hazard ratio for disease progression or death, 0.55 (95% CI, 0.42–0.72) P<0.001 36 (Osimertinib–savolitinib, 12 months); 10 (Platinum–pemetrexed, 12 months) Months since Randomization 0–38 ORIGINAL ARTICLE MEETING OF THE EUROPEAN SOCIETY FOR MEDICAL ONCOLOGY Osimertinib plus Savolitinib in Advanced NSCLC S. Lu and Others In patients with EGFR-mutated and MET-altered non–small-cell lung cancer that had progressed during osimertinib therapy, osimertinib plus savolitinib (a MET tyrosine kinase inhibitor) led to longer survival than platinum–pemetrexed. Survival table Endpoint: Progression-free survival (full analysis population) Arm | N | Median | Median CI Osimertinib–savolitinib | 171 | 7.6 | 5.6–9.7 Platinum–pemetrexed | 167 | 5.4 | 4.2–5.8 HR: 0.55 HR CI: 0.42–0.72 P: <0.001 Landmark rates: 12 months: Osimertinib–savolitinib 36% | Platinum–pemetrexed 10%

[Slide 1] Questions How do the EGFR + MET TKI combinations compare to amivantamab-chemotherapy (MARIPOSA-2) regimen? biomarker selected v non-selected oral v IV regimens AE profiles Will there be differential ability of MET TKIs, Abs and Bispecific Abs, ADCs, to address MET acquired resistance and different cutoffs for activity? What is the best definition of MET positivity? How do we increase tumor biopsy at time of progression? [Slide 3] Is platinum chemotherapy the right comparator? MARIPOSA-2 n | Median progression-free survival, months (95% CI) Amivantamab–chemotherapy | 131 | 6.3 (5.6-8.4) Amivantamab–lazertinib–chemotherapy | 263 | 8.3 (6.8-9.1) Chemotherapy | 263 | 4.2 (4.0-4.4) Amivantamab–chemotherapy versus chemotherapy: Hazard ratio for disease progression or death, 0.48 (95% CI 0.36-0.64) P<0.001 Amivantamab–lazertinib–chemotherapy versus chemotherapy: Hazard ratio for disease progression or death, 0.44 (95% CI 0.35-0.56) P<0.001 Patients who are progression-free (%) Months • Amivantamab-chemotherapy improved PFS over platinum chemotherapy after progression on osimertinib • No selection by MET status Passaro et al Annals Onc 2024 Survival table Endpoint: PFS Arm | N | Median | Median CI Amivantamab–chemotherapy | 131 | 6.3 | 5.6-8.4 Amivantamab–lazertinib–chemotherapy | 263 | 8.3 | 6.8-9.1 Chemotherapy | 263 | 4.2 | 4.0-4.4 HR: 0.48 (amivantamab-chemotherapy vs chemotherapy); 0.44 (amivantamab-lazertinib-chemotherapy vs chemotherapy) HR CI: 0.36-0.64; 0.35-0.56 P: P<0.001; P<0.001

[Slide 1] CHAIR: SURESH SENAN ADCs Rise in NSCLC • Antibody, linker, payload • Smart-Bomb! • "Chemo-Free" • Biomarker selection should be easy → antigen target dependent The Story Began with HER2 ADC Trastuzumab Deruxtecan for HER2 mutant NSCLC HER2+ (genomic selection) 2018: DESTINY-Lung01 begins - Phase II; HER2-mutant and IHC-selected cohorts 2020–21: Early activity establishes proof of concept First U.S. approval in NSCLC: Accelerated approval for previously treated HER2-mutant NSCLC; 5.4 mg/kg 2023: DESTINY-Lung02 confirms dose/benefit - Randomized dose finding Sep 2026: DESTINY-Lung04 positive in 1L - PFS 14.3 vs 8.3 mo; HR 0.63 vs pembrolizumab + chemotherapy Melissa L. Johnson MD, SCRI Melissa Johnson - Invited Discussant LBA66, LBA67 and LBA68 [Slide 2] CHAIR: SURESH SENAN ADCs and ICIs play complimentary roles in the TME Tip efficacy/safety towards tumor Identify biomarkers that will aid in patient selection (maybe) Manage irAEs [Schematic: CTL-mediated apoptosis; Antigen presentation; Dendritic cell migration; Mature dendritic cells; Naive T cell; Activated T cell; T cell migration; Prolonged CTL activity; Treg; MDSC; Tumor cell; Immune attack; TCR; MHC I; PD-1; PD-L1/2; T cell inhibitory signals blocked; Dendritic cell; CD8+ T cell; DAMPs MHC-I Various cytokines; Granzymes and perforin; Cancer cell with EMP; TGF-β] Yin et al. Antibody Therapeutics 2025, 8 (3), 262-274. Melissa L. Johnson MD, SCRI Melissa Johnson - Invited Discussant LBA66, LBA67 and LBA68 [Slide 3] Conclusions: Do ADCs have what it takes to BEAT chemotherapy?? • Maybe not in an all-comers trial .... (Sig-Vie-002) • Opti TROP-Lung 04 shows it can be done! ➢ Find the right patient population ➢ ADC toxicity shouldn't be underestimated ➢ Dose attenuation, even pulsatile administration will be part of ADC adoption into NSCLC paradigms • Opti TROP-Lung 06 suggests synergy with checkpoint inhibitors is also within reach. ➢ Likely the primary exposure to IO will be best chance for success Melissa L. Johnson, MD, SCRI Melissa Johnson - Invited Discussant LBA66, LBA67 and LBA68

[Slide 2] C. Acitivity in CNS? HER3 expression in BM: (Tomasich Clin Cancer Res 2023;29:3225–36) Primary: 8.5% | 31.9% | 59.6% BM: 48.9% | 23.4% | 27.7% n=47 High Low Negative Patritumab deruxtecan (HER3-DXd) in patients with active brain metastases of non-small-cell lung cancer (TUXEDO-3): a single-arm, phase 2 trial (Fuereder Lancet Oncol 2025; 26: 1454–66) ORR 30% (n=20, RANO-BM) HERTHENA-Lung01 (HER3-Dxd): icORR 33.3% (n=10/30) (Yu et al. J Clin Oncol 2023, 41:5363-5375.) Figure 2: Best percentage of change in tumour burden for intracranial lesions according to RANO-BM criteria Subgroup Analysis of PFS by BICR – PANKU-Lung01 Iza-bren Events/n | PBC Events/n CNS metastases Yes 17/21 | 16/20 0.75 (0.37, 1.50) No 111/195 | 148/186 0.39 (0.30, 0.51) Iza-bren better ← → PBC better PFS per BICR in Subgroups – HERON-Lung01 No. of events/no. of patients Ruzaltatug rezetecan | Chemo Brain metastases Yes 28/48 | 33/47 0.61 (0.36-1.02) No 80/202 | 123/201 0.42 (0.32-0.56) Favors ruzaltatug rezetecan ← → Favors chemo [Slide 3] E. Positioning in the treatment algorithm? First line Stage IV mNSCLC with EGFR-activating Mutation Before Systemic Progression • Osimertinib ± CT • Lazertinib-amivantamab • EGFR TKIs • Erlotinib-anti-VEGF • Gefitinib - CT After progression EGFR-activating Mutation After Systemic Progression with Third-generation TKI • Amivantamab + CT (mPFS 6.3 m vs. 4.2; HR 0.48 (95% CI 0.36-0.64)(MARIPOSA-2)1 • CT • Atezolizumab-bev-carbo-pac Ongoing or recent studies after progression: • HARMONi (ivonescimab + chemotherapy)(NCT06396065) • Fourth-generation EGFR inhibitor programs • Datopotamab-deruxtecan studies • Amivantamab, lazertinib, and bevacizumab (ETOP 18-21 AMAZE-lung)1 Need to define the optimal treatment sequences and combinations 1. Passaro et al. Annals of Oncology, 2023; 35, 77-90. 2. Soo et al. Lancet Respir Med. 2026 Oct 5:S2213-2600(26)00203-1. [Slide 4] F. Translatability of results? PANKU-Lung01: • These results support iza-bren as a new standard of care in this population in China. • Validation in other populations: Global registrational studies in EGFRm NSCLC patients after progression on a third-generation EGFR TKI (IZABRIGHT-Lung01, NCT07100080) and in the first line setting are ongoing (IZABRIGHT-Lung02, NCT 07680790). HERON-Lung01: • Findings support ruzaltatug rezetecan as a new standard-of-care option for EGFRm NSCLC progressing after a third-generation EGFR–TKI • Phase 3 study of ruzaltatug rezetecan + 3rd gen EGFR–TKI aumolertinib for 1L EGFRm NSCLC is underway (NCT07183189)(in China) Simon Ekman

Sac-TMT 5 mg/kg OptiTROP-Lung 04 (vs. carbo/pemetrexed) | Sac-TMT (n = 188) | Chemo (n = 182) Median duration of exposure, months (range) | 9.6 (0.5, 35.8) | 4.9 (0.7, 34.1) TRAEs, n (%) Any grades | 188 (100) | 179 (98.4) Grade ≥3 | 110 (58.5) | 98 (53.8) Serious TRAEs | 19 (10.1) | 33 (18.1) Leading to dose reduction | 57 (30.3) | 41 (22.5) Leading to dose interruption | 75 (39.9) | 60 (33.0) Leading to discontinuation | 0 | 1 (0.5) Leading to death | 1 (0.5)* | 1 (0.5)* Sac-TMT 4 mg/kg OptiTROP-Lung 06 (vs. carbo/pemetrexed/pembro) Event, n (%) | Sac-TMT + Pembro (n = 229) | Chemo + Pembro (n = 226) TEAEs | 228 (99.6) | 225 (99.6) Grade ≥3 TEAEs | 109 (47.6) | 137 (60.6) Serious TEAEs | 82 (35.8) | 82 (36.3) Leading to dose reductiona | 11 (4.8) | 45 (19.9) Leading to any dose discont. | 13 (5.7) | 32 (14.2) TEAEs Leading to death | 5 (2.2) | 14 (6.2) TRAEs Leading to death | 1 (0.4) | 3 (1.3) Melissa L. Johnson MD, SCRI

[Slide 2] KRASi (OFF): highly competitive field (Phase I/II data) NEXT GENERATION INHIBITORS? Drug | Response Rate (%) | Progression Free Survival (mo.) | Overall Survival (mo.) FULZERASIB. N=116 | 49 | 9.7 | GARSORASIB. N=189 | 48 | 9.1 | 14.2 GLECIRASIB. N=119 | 48 | 8.2 | 13.6 ELISRASIB. N=68 (600 mg) | 56 | 11.8 | SOSIMERASIB. N=145 | 52 | 7.2 | OLOMORASIB. N=76 | 33 | 6.5 | CALDERASIB. N=24 | 38 | 8.3 | DIVARASIB. N=44 (400 mg) | 59 | 15.3 | ADAGRASIB. N=116 | 43 | 6.9 | 14.1 SOTORASIB. N=126 | 41 | 6.3 | 12.5 Remon Cancer Treat Rev 2026 Benjamin BESSE [Slide 3] KRAS G12Ci First line ORR | Median PFS, months (95% CI) L2 — Monotherapy 2L+ NSCLC, at RP2D Divarasib¹ n=44 | 59% | 15.3 (12.3–26.1) Elisrasib² n=68 | 56% | 11.8 Glecirasib³ n=117 | 48% | 8.2 (5.5–13.1) Adagrasib⁴ n=112 | 43% | 6.5 (4.7–8.4) Olomorasib*⁵ n=38 | 42% | 8.2 (4.3–15.6) Calderasib⁶ n=103 | 35% | 8.3 (6.0–16.4) Sotorasib⁷ n=171 | 28% | 5.6 (4.3–7.8) L1 — + Pembrolizumab 1L NSCLC Elisrasib + pembro‡⁸ n=48 | 81% | NR (8.4–NR) Olomorasib + pembro¹⁰ n=48 | 74% | NR (12.0–NR) Divarasib + pembro†⁹ n=59 | 73% | 19.3 (12.4–NR) Calderasib + pembro†⁸ n=122 | 71% | 28.8 (20.6–NR) Adagrasib + pembro¹³ n=149 | 44% | 11.0 (5.8–14.0) + Chemo-pembro 1L NSCLC Calderasib + chemo-pembro⁶ n=48 | 65% | 17.6 (10.5–26.3) Olomorasib + chemo-pembro¹¹ n=77 | 62% | 11.6 (9.5–21.5) Chemo-pembro (KN-189)§¹² n=410 | 48% | 9.0 (8.1–9.9) COURTESY OF M.ALDEA *After prior KRAS G12Ci. †PD-L1 ≥1%. ‡Incl. unconfirmed. § Not KRAS-selected. 1 Sacher JCO25; 2 Cho AACR26; 3 Shi NatMed25; 4 Jänne NEJM22; 5 Murciano-Goroff NatCommun26; 6 Gray ESMO26; 7 de Langen Lancet23; 8 Lu ASCO26; 9 Skoulidis ASCO26; 10 Burns JTO26; 11 Negrao ESMO26; 12 Gadgeel JCO20; 13 Jänne ASCO25 (KRYSTAL-7) Benjamin BESSE

[Slide 1] CHAIRS: PATRICK FORDE, ANTONIO PASSARO Treatment algorithm in EGFR-TKI resistant NSCLC EGFR-activating Mutation After Systemic Progression with Third-generation TKI Stage IV mNSCLC with EGFR-activating mutation progressing after a third-generation TKI Genomic analysis by NGS (tissue or cfDNA followed by tissue, if no target is found with cfDNA) [III, C] Resistance mechanism identified No resistance mechanism identified Clinical trial [III, B] or refer to no resistance mechanism Systemic progression Amivantamab-vmjw-platinum-based doublet ChT [I, A; MCBS 3] Platinum-based doublet ChT [III, A] Atezolizumab-bevacizumab-paclitaxel-carboplatin [III, B; MCBS 3] Datopotamab deruxtecan (a) [III, B; MCBS 2] T-DXd [III, B] for patients with HER2+ (IHC 3+) (b) • Amivantamab + CT (MARIPOSA-2) • CT • Atezolizumab-bev-carbo-pac (IMpower150) ESMO Living Guideline Oncogene-Addicted Metastatic NSCLC v1.3 - Feb 2026 EGFR EXON 19 DELETION OR L858R MUTATION SUBSEQUENT THERAPY Progression on Osimertinib or Lazertinib ± Amivantamab-vmjw Asymptomatic • Consider definitive local therapy (eg, SABR or surgery) for limited lesions • Continue Osimertinib or Lazertinib ± Amivantamab-vmjw Symptomatic Brain • Consider definitive local therapy (eg, SRS) • Continue Osimertinib or Lazertinib ± Amivantamab-vmjw • NCCN Guidelines for CNS Cancers Systemic Limited progression • Consider definitive local therapy (eg, SABR or surgery) • Continue Osimertinib or Lazertinib ± Amivantamab-vmjw or • Therapy for multiple lesions, below Multiple lesions Carboplatin/Pemetrexed + Amivantamab-vmjw (if not previously given) (nonsquamous) (category 1) (preferred) or Lazertinib + Amivantamab-vmjw (if not previously given) (useful in certain circumstances) or Systemic therapy (NSCL-K 1 of 6) • Amivantamab + CT (MARIPOSA-2) • CT • Lazertinib+Amivantamab (MARIPOSA) NCCN Guidelines Version 9.2026 Non-Small Cell Lung Cancer Resistance mechanisms to first-line osimertinib Acquired EGFR mutations; Acquired amplifications; Acquired oncogenic fusions; Acquired MAPK-PI3K mutations; Acquired cell cycle gene alterations; Unknown; Transformations (SCLC, SCC) 15%; 6–10%; 8–17%; 1–8%; 13–14%; 10%; 40–50% C797X; L718Q; EGFRamp; G724S; S768I; METamp (7–15%); HER2amp (1–2%); SPTBN1-ALK; RET fusions; BRAF fusions; BRAFV600E (3%); PI3KCA (7%); KRAS (3–4%); HER2 (1%); CCND1amp; CCND2amp; CCNE1amp; CDK4amp; CDK6amp Leonetti et al. British Journal of Cancer volume 121, pages725–737(2019) Simon Ekman Invited Discussant LBA71 and LBA72 [Slide 2] CHAIRS: PATRICK FORDE, ANTONIO PASSARO HER3 targeting overcomes heterogeneity in EGFR-TKI resistance mechanisms – a mutation-agnostic therapeutic approach HERTHENA-Lung01 Phase II Trial of Patritumab Deruxtecan (HER3-DXd) Confirmed best overall response (BICR) CR PR SD PD NE + Ongoing treatment Best Change in Sum of Diameters (BICR) From Baseline (%) EGFR-activating mutations: Ex19del; L858R EGFR TKI resistance mechanisms: EGFR-dependent: Anya; C797X EGFR-independent: Any; MET amp Unidentified HER3 expressionb H-score 0-300 Simon Ekman Yu et al. J Clin Oncol 2023, 41:5363-5375. Simon Ekman Invited Discussant LBA71 and LBA72 [Slide 3] CHAIRS: PATRICK FORDE, ANTONIO PASSARO E. Toxicity profiles? Safety Summary PANKU-Lung01 Iza-bren (N = 216) | PBC (N = 206) TRAEs, n (%) 216 (100) | 204 (99.0) ≥Grade 3 TRAEs, n (%) 170 (78.7) | 128 (62.1) Treatment-related SAEs, n (%) 92 (42.6) | 40 (19.4) TRAEs leading to death, n (%) 1 (0.5) | 1 (0.5) TRAEs leading to treatment discontinuation, n (%) 4 (1.9) | 4 (1.9) TRAEs leading to dose reduction, n (%) 130 (60.2) | 48 (23.3) TRAEs leading to dose interruption, n (%) 116 (53.7) | 49 (23.8) • Adjudicated treatment-related ILD occurred in two patients (0.9%) in the iza-bren arm (one grade 1 and one grade 3). None in the PBC arm. HERON-Lung01 Ruzaltatug rezetecan (n = 248) | Chemo (n = 239) Any TRAE 247 (99.6) | 236 (98.7) Grade ≥3 101 (40.7) | 93 (38.9) Serious 38 (15.3) | 36 (15.1) Leading to dose reduction 36 (14.5) | 24 (10.0) Leading to dose discontinuation 3 (1.2) | 3 (1.3) Leading to death 1 (0.4)* | 0 ILD 6 (2.4) | 0 Grade ≥3 1 (0.4)† | 0 Data are n (%). * Death of unknown cause. † Grade 3. Simon Ekman Invited Discussant LBA71 and LBA72

NEOTORCH trial, results EFS by investigator Median follow-up: 48.2 months Similar to all trials (~60%) Toripalimab / Placebo Overall Survival Similar to all trials (~70%) Follow-up, mo No. at risk (censor) PD-L1 expression evaluated using a validated JS311 IHC: • EFS PD-L1 1-49% (n=164), HR 0.41 (0.26-0.63) • EFS PD-L1 ≥ 50% (n=157), HR 0.47 (0.26-0.82) Lu – ESMO 2026 * Lu – JAMA 2024 (EFS interim analyses HR 0.40; 0.28-0.57; P<0.001. IHC: immunohistochemistry) * Cascone –NEJM 2024 Survival table Endpoint: EFS by investigator (Neotorch, final analysis; median follow-up 48.2 months) Toripalimab | NR | 51.3, NE Placebo | 20.2 | 14.3, 24.8 HR: 0.52 HR CI: 0.391, 0.677 P: <0.0001 EFS toripalimab 71.5% (24 mo), 63.1% (36 mo); placebo 45.5% at 24 mo, placebo 36-mo label [illegible]

08:30 - 10:00 Proffered paper 2: NSCLC, metastatic CHAIRS: PATRICK FORDE, ANTONIO PASSARO Questions What is the right MET definition in the 1L population? Is there benefit in the 2+ population? How does a MET selected population fare with FLAURA2 or MARIPOSA regimen? What resistance will emerge from SANOVO combination? ALICANTE AUDITORIUM - HALL 6

[Slide 2] Efficacy in context – 2L+ Docetaxel Combo Study / regimen | Patients | ORR | Median PFS ROSETTA Lung-107 Pumitamig + docetaxel | Post-chemo-IO, phase II | 36.7%* | 6.9 mo DRUN Docetaxel + ramucirumab | Post-ICI, phase II | 33.3% | 4.9 mo Retrospective docetaxel + ramucirumab | Post-ICI, n=55 | 42% | 5.8 mo REVEL Docetaxel + ramucirumab | Post-platinum, pre-ICI era | ~23% | 4.5 mo REVEL control Docetaxel alone | Post-platinum, pre-ICI era | ~14% | 3.0 mo RC 148 10 mg + Docetaxel | Post- Chemo-IO, N 21 | 33% | 5,8 mo RC 148 20 mg + Docetaxel | Post- Chemo-IO, N=21 | 67% | 8,5 mo CS2009 + Docetaxel | Post-Chemo-IO. N=18 | 47% | ?? Luis Paz-Ares – ESMO 2026 J Yu et al., WLCC 2026 [Slide 3] CS2009 – 1L Activity N=55 Monotherapy (in PDL1+) | All Evaluable Patients | 20 mg/kg | 30 mg/kg | Squamous | Non-squamous TPS ≥1%: ORR, % (n) | 66.7% (32/48) | 58.3% (14/24) | 75.0% (18/24) | 69.0% (20/29) | 63.2% (12/19) TPS ≥1%: DCR, % (n) | 93.8% (45/48) | 95.8% (23/24) | 91.7% (22/24) | 96.6% (28/29) | 89.5% (17/19) TPS ≥50%: ORR, % (n) | 87.5% (21/24) | 76.9% (10/13) | 100% (11/11) | 91.7% (11/12) | 83.3% (10/12) TPS ≥50%: DCR, % (n) | 95.8% (23/24) | 92.3% (12/13) | 100% (11/11) | 100% (12/12) | 91.7% (11/12) Median DoR and median PFS were not reached due to short follow-up. N=27 CS2009 + Cb/Paclitaxel (in SCC) | All Evaluable Patients | 20 mg/kg | 30 mg/kg | PD-L1 TPS <1% | PD-L1 TPS 1-49%* | PD-L1 TPS ≥50% ORR, % (n) | 80.8% (21/26) | 76.9% (10/13) | 84.6% (11/13) | 100% (8/8) | 66.7% (10/15) | 100% (2/2) DCR, % (n) | 100% (26/26) | 100% (13/13) | 100% (13/13) | 100% (8/8) | 100% (15/15) | 100% (2/2) [footnote, small: *... TPS ... Note: One additional patient with unknown PD-L1 status achieved a PR - partially [illegible]] Median DoR and median PFS were not reached due to short follow-up. [Waterfall and spider plots - axis values illegible] Luis Paz-Ares – ESMO 2026 [Slide 4] CS2009 – 1L Activity N=55 Monotherapy (in PDL1+) | All Evaluable Patients | 20 mg/kg | 30 mg/kg | Squamous | Non-squamous TPS ≥1%: ORR, % (n) | 66.7% (32/48) | 58.3% (14/24) | 75.0% (18/24) | 69.0% (20/29) | 63.2% (12/19) TPS ≥1%: DCR, % (n) | 93.8% (45/48) | 95.8% (23/24) | 91.7% (22/24) | 96.6% (28/29) | 89.5% (17/19) TPS ≥50%: ORR, % (n) | 87.5% (21/24) | 76.9% (10/13) | 100% (11/11) | 91.7% (11/12) | 83.3% (10/12) TPS ≥50%: DCR, % (n) | 95.8% (23/24) | 92.3% (12/13) | 100% (11/11) | 100% (12/12) | 91.7% (11/12) Median DoR and median PFS were not reached due to short follow-up. Luis Paz-Ares – ESMO 2026 Harmoni 2 trial - Ivonescimab Primary endpoint: PFS per IRRC | Ivonescimab (n = [illegible]) | Pembrolizumab (n = [illegible]) mPFS, mos (95% CI) | 11.14 (7.33, NE) | 5.82 (5.03, 8.21) Stratified HR (95% CI) | 0.51 (0.38, 0.69) p-value | <0.0001 Median Follow-up: [illegible] months ORR, DCR and DoR per IRRC ORR, % (95% CI) | 50.0 (42.8, 57.2) | 38.5 (31.7, 45.6) DCR, % (95% CI) | 89.9 (84.8, 93.7) | 70.5 (63.7, 76.7) Median DoR, mos (95% CI) | NR (NE, NE) | NR (8.38, NE) Bar chart: Ivonescimab 89.9% DCR / [illegible]% ORR; Pembrolizumab 70.5% DCR / 38.5% ORR ORR and DCR were higher with ivonescimab vs. pembrolizumab. A Xiong et al., Lancet 2025

Are the results from SAFFRON clinically meaningful? YES Progression-free survival (BICR) Median PFS, months (95% CI) Osimertinib + savolitinib 7.6 (5.6, 9.7) Platinum–pemetrexed 5.4 (4.2, 5.8) HR 0.55 (95% CI 0.42-0.72); p<0.01 36% 10% Probability of Progression-free Survival Months since Randomisation Overall survival Median OS, months (95% CI) Osimertinib + savolitinib 17.8 (15.6, 23.7) Platinum–pemetrexed 12.9 (10.6, 16.8) HR 0.68; 95% CI, 0.50-0.91; p=0.010* 70% 53% Overall Survival Subsequent therapy Osi+Savo – 60%; 41% ChT Chemo – 66%; 13% MET+EGFR TKI PFS in control arm similar to molecularly unselected post-osimertinib trials (MARIPOSA2, OptiTROP Lung04, COMPEL) OS in control arm may be lower but molecular selection for worse prognosis No. at Risk Median follow-up PFS: osimertinib+savolitinib 5.6 months, platinum–pemetrexed 3.5 months. Overall maturity: 68%. Median follow-up (OS): osimertinib+savolitinib 14.2 months, platinum–pemetrexed 10.2 months. Overall maturity: 53%. *Boundary to declare statistical significance p<0.03047. Objective response rate higher with osimertinib + savolitinib (51% v. 27%), odd ratio 2.9; Median duration of response 9.7 months v. 5.7 months Lu S et al. ESMO Congress. 2026 LBA6; Passaro A et al. Ann Oncol 2023; Popat S et al. Ann Oncol 2024; Fang W et al NEJM 2026 Natasha B. Leighl Survival table Endpoint: Progression-free survival (BICR) Arm | N | Median | Median CI Osimertinib + savolitinib | 171 | 7.6 mo | 5.6, 9.7 Platinum–pemetrexed | 167 | 5.4 mo | 4.2, 5.8 HR: 0.55 HR CI: 0.42-0.72 P: <0.01 12-mo PFS: 36% vs 10% (annotated on slide)

Primary endpoint: OS • Ivonescimab with chemotherapy significantly improved OS 12-month rate: 59.7% (Ivonescimab+chemo) | 50.7% (Durvalumab+chemo) 18-month rate: 44.0% (Ivonescimab+chemo) | 32.5% (Durvalumab+chemo) | Ivonescimab + chemo | Durvalumab + chemo mOS (95% CI), months | 14.9 (12.9, NE) | 12.1 (11.0, 13.4) Stratified HR (95% CI) | 0.75 (0.61, 0.92) P-value | 0.0031 (one-side) mOS, median overall survival; HR, hazard ratio; CI, confidence interval; NE, not estimable. Time (months) Number of Risk Ivonescimab+chemo: 340 337 330 322 314 304 292 281 266 253 238 221 180 146 101 66 42 32 22 9 1 0 Durvalumab+chemo: 342 336 329 316 309 297 275 259 240 217 199 186 151 119 77 46 33 24 14 3 1 0 Survival table Endpoint: Overall survival (primary endpoint) Arm | N | Median | Median CI Ivonescimab + chemo | 340 | 14.9 | 12.9, NE Durvalumab + chemo | 342 | 12.1 | 11.0, 13.4 HR: 0.75 (stratified) HR CI: 0.61, 0.92 P: 0.0031 (one-side) Landmark rates: 12 months: Ivonescimab + chemo 59.7% | Durvalumab + chemo 50.7% 18 months: Ivonescimab + chemo 44.0% | Durvalumab + chemo 32.5%

Figure 5. Intracranial Response in a Patient With SQ NSCLC With Untreated Brain Metastases Baseline: D=20mm → Week 36: D=0mm In a 72-year-old male with SQ NSCLC and untreated brain metastases, treatment with BGB-58067 200 mg resulted in a PR with 100% decrease in the diameter of brain target lesion. Abbreviations: D, lesion diameter; NSCLC, non-small cell lung cancer; SQ, squamous.

First-in-Human Trial of BGB-58067, an MTA-Cooperative PRMT5 Inhibitor, in MTAP-Deficient Advanced Tumors FPN: 1043P Jordi Rodon, Leiguang Ye, Adnan Nagrial, ... Caicun Zhou (BeOne Medicines and collaborating centres) CONCLUSIONS • BGB-58067, an MTA-cooperative PRMT5 inhibitor, demonstrated promising monotherapy activity in heavily pretreated patients with MTAP-deficient solid tumors, including NSCLC • In patients with EGFR wild-type NSCLC, BGB-58067 demonstrated encouraging and durable clinical activity with an ORR of 38.5% and a median PFS of 8.1 months • The low incidence of CNS progression in 2L+ NSCLC, together with an intracranial objective response rate of 30.8% observed in patients with brain metastases, shows clinically meaningful CNS activity of BGB-58067 • BGB-58067 demonstrated a well-tolerated and manageable safety profile • BGB-58067 demonstrated a favorable PK profile across the evaluated dose ranges, and robust tumor cell target engagement as suggested by SDMA depletion observed in paired biopsy • These data support ongoing exploration of combination cohorts in 1L NSCLC and PDAC with standard-of-care treatments, and with RAS inhibitors in PDAC RESULTS (excerpt) • As of the data cutoff date of September 12, 2026, 182 patients were treated; 43 patients remain on treatment • In the overall cohort (excluding patients with glioma), confirmed ORR (95% CI) was 21.3% (14.8-29.0) • In the NSCLC cohort, confirmed ORR (95% CI) was 25.3% (16.0-36.7) • Intracranial ORR (95% CI) in patients with BM in the NSCLC cohort was 30.8% (9.1-61.4) Presented at the ESMO Congress, Madrid, Spain, October 23-27, 2026

Progress in Classical EGFR-mutated NSCLC (ex19del, L858R) Regulatory approval: USA (grey) | China (red) | EU (navy) Selected milestones: Platinum doublet | [star] | 1st gen EGFR TKI | 2nd gen EGFR TKI | 3rd gen EGFR TKI | Anti-VEGF | Adjuvant EGFR TKI | 1L combinations | Consolidation post CRT | TROP2 ADC, VEGF-PD1 bsAb Timeline: 2000 2004 2009 2012 2013 2014 2016 2018 2019 2021 2022 2023 2024 2025 2026 2004 - EGFR mutation discovered 2005 – EGFR T790M resistance mutation described 2007 – MET amplification as resistance mechanism 2015 – EGFR C797S as resistance mechanism Stage I-III: Osimertinib (adjuvant); Icotinib (adjuvant); Osimertinib (consolidation st3); Aumolertinib (adjuvant); Aumolertinib (consolidation st3) Stage IV 1L: Platinum doublet; Gefitinib2; Erlotinib3; Osimertinib; Dacomitinib; Afatinib (uncommon mut); Erlotinib + Ramucirumab; Aumolertinib; Furmonertinib; Befotertinib; Osi+ChT; Ami+Lazer; Rezivertinib; Zorifertinib (CNS); Limertinib Stage IV 2L+: Erlotinib1 (Unselected EGFR); Afatinib6,7; Icotinib; Osimertinib (T790M); Atezolizumab/Bevacizumab/Paclitaxel/Carboplatin; Befotertinib(T790M); Rilertinib(T790M); Rezivertinib(T790M); Ami +ChT (Post Osi); Sac-TMT (post osi); Ivonescimab +ChT (post 3G TKI); Limertinib (T790M) Median PFS and OS in EGFRm NSCLC (selected trials) | mPFS (months) | mOS (months) Platform doublet | ~6 | ~23 G1/G2 EGFR TKI (e.g. Gefitinib, Erlotinib, Icotinib, Afatinib, Dacomitinib) | 9.5 | ~30 G3 EGFR TKI (e.g. Osimertinib, Aumonertinib, Furmonertinib,) | 18.9 | ~37 G3 EGFR TKI Combos (e.g Osimertinib+ChT, Amivantamab+ Lazertinib (Ami+Lazer), Aumolertinib+ ChT) | 25.5 | 48+ Shepherd FA et al. NEJM. 2005; Paez, Janne et al Science 2004; Lynch et al NEJM 2004; Engelman et al Nature 2007; Mok TS et al. NEJM. 2009; 3. Rosell R et al. NEJM. 2012; 4. Yang JCH et al. Lancet Oncol. 2015; 5 Sequist LV et al. J Clin Onol. 20[illegible] Natasha B. Leighl

[Slide 2] Time to deterioration in pulmonary symptoms (NSCLC-SAQ) Cough | T-DXd (n=227) | Pembro + chemo (n=227) Median TTD, months (95% CI) | 26.2 (17.1, NC) | 11.0 (7.4, NC) Hazard ratio (95% CI) | 0.66 (0.44, 0.99) Chest pain | T-DXd (n=227) | Pembro + chemo (n=227) Median TTD, months (95% CI) | NC (11.5, NC) | 14.3 (8.3, NC) Hazard ratio (95% CI) | 0.80 (0.54, 1.18) Dyspnea | T-DXd (n=227) | Pembro + chemo (n=227) Median TTD, months (95% CI) | 10.1 (6.5, NC) | 5.1 (3.2, 11.0) Hazard ratio (95% CI) | 0.77 (0.54, 1.09) Probability of no sustained deterioration; Time from randomization (months) No. at risk (0–36 months, every 3) Cough — T-DXd | 227 106 73 58 41 34 25 17 11 7 4 3 0 Cough — Pembro + chemo | 227 92 54 38 21 12 10 7 6 4 3 2 0 Chest pain — T-DXd | 227 98 70 55 37 30 22 15 10 8 7 4 0 Chest pain — Pembro + chemo | 227 91 56 40 25 14 10 7 6 2 2 1 0 Dyspnea — T-DXd | 227 85 62 50 36 29 23 18 11 9 7 5 0 Dyspnea — Pembro + chemo | 227 66 36 21 16 9 5 5 3 3 1 1 0 T-DXd monotherapy delayed time to, and reduced risk of, deterioration in cough versus pembro + chemo, with similar trends observed for chest pain and dyspnea Assessed in the full analysis set. Compliance rates for the NSCLC-SAQ: 67.8% T-DXd vs 71.9% pembro + chemo (baseline); 76.1% T-DXd vs 69.0% pembro + chemo (overall). Sustained deterioration was defined as a meaningful worsening at two or more consecutive post-baseline timepoints relative to baseline (regardless of whether the patient withdrew from treatment or received another anticancer therapy prior to deterioration). Patients with no sustained deterioration were censored at the time of their last PRO assessment with an evaluable score or at Cycle 1 Day 1 NC, not calculable; TTD, time to deterioration Silvia Novello, MD, PhD Survival table Endpoint: Time to deterioration (NSCLC-SAQ) — cough [chest pain and dyspnea panels in notes] Arm | N | Median | Median CI T-DXd (cough) | 227 | 26.2 months | 17.1, NC Pembro + chemo (cough) | 227 | 11.0 months | 7.4, NC HR: 0.66 HR CI: 0.44, 0.99 [Slide 3] Physical functioning (EORTC QLQ-C30 physical functioning scale) T-DXd (n=226) Pembro + chemo (n=220) Proportion of patients reporting 'maintained or improved' physical function (%) Week 0–58, EOT Proportions of patients with 'maintained/improved' physical function were similar across arms through Week 18, and higher with T-DXd vs pembro + chemo at most timepoints from Week 20 onward, including at EOT Timepoints after Week 58 through Week 94 have been excluded owing to low sample size (<10% of patients in the safety analysis set, either arm) Assessed in the safety analysis set. Compliance rates for EORTC QLQ-C30: 65.9% T-DXd vs 69.5% pembro + chemo (baseline); 68.7% T-DXd vs 58.8% pembro + chemo (overall). Overall compliance rates for EORTC IL19: 76.9% T-DXd vs 71.6% pembro + chemo. Baseline physical functioning was recorded prior to first dose of study treatment at Week 1. Change in physical function was calculated relative to baseline; therefore, baseline datapoints are not shown. 'Improvement' was defined as an increase of ≥13.33 points from baseline, 'deterioration' was defined as a decrease of ≥13.33 points from baseline, and 'maintained' was defined as no improvement or deterioration. No data were collected at Week 6 EOT, end of treatment Silvia Novello, MD, PhD

Subsequent Anti-tumor Therapy • The rate of subsequent anti-tumor therapy was 14.1% higher in the durvalumab plus chemo arm than ivonescimab plus chemo arm Category, n(%) | Ivonescimab plus chemo (N=340) | Durvalumab plus chemo (N=342) Any anti-tumor therapy | 108 (31.8) | 157 (45.9) Chemotherapy | 82 (24.1) | 96 (28.1) Targeted therapy | 49 (14.4) | 86 (25.1) Immunotherapy | 49 (14.4) | 72 (21.1) Antibody-Drug Conjugate | 9 (2.6) | 16 (4.7) Traditional chinese medicine | 2 (0.6) | 24 (7.0) Investigational anti-tumor drug | 1 (0.3) | 0 Others | 1 (0.3) | 2 (0.6) Jian Zhou, MD, PhD

[Slide 2] Modes of action of BsAbs currently in clinical development a, Signal inhibition by receptor blockade of two different receptors. b, Signal inhibition by inhibition of a soluble ligand and a receptor. c, Signal inhibition by dual blockade of two different ligands. d, Drug delivery by BiAb drug conjugates. e, Targeted blockade of the 'don't eat me signal'→phagocytosis. f, Blockade of the 'don't eat me signal' in combination with ICI. g, Effector cell engagement through CD3 of TCR on T cells or CD16 on NK cells. h, Tumor cell binding mediated immune cell (co-)stimulation. i, PD-L1 binding-mediated immune cell (co-)stimulation. j, Dual immune cell co-stimulation. k, Tumor-targeted PD-L1 pathway inhibition. l, Tumor-targeted PD-1 pathway inhibition. m, Checkpoint binding-mediated immune cell co-stimulation. n, Dual immune checkpoint inhibition (ICI). o, ICI and soluble signaling inhibition by ligand blockage. TAAs of solid tumours 5T4, B7-H3, B7-H4, CD96, CDH17, CEA, CLDN6, CLDN18.2, cMET, DLL3, DLL4, EGFR, EGFRvIII, ENPP3, EpCAM, FRα, gp100, GPC3, GUCY2C, HER2, HER3, HLA-G, KLK2, LGR5, MAGEA4/8, mesothelin, MUC16, PDL1, PRAME, PSMA, ROR1, STEAP1, TMEFF2, and others TAAs of haematological malignancies BCMA, CD19, CD20, CD22, CD30, CD33, CD38, CD123, FcRH5, FLT3, GPRC5D, ROR1, and others Cancer cell T cell Macrophage T cell/natural killer cell T cell APC/dendritic cell Tina Cascone, MD, PhD Adapted from Klein C et al., Nat Rev Drug Discov 2024 [Slide 4] PD-1 vs. PD-L1 × VEGF-A: Biological distinction and therapeutic implications Biological distinction | PD-1×VEGF-A | PD-L1×VEGF-A | Therapeutic implication Checkpoint coverage | Blocks PD-1 inhibition by both PD-L1 and PD-L2.1 Potential benefit if PD-L2 drives suppression.1 | Blocks PD-L1–PD-1, preserves PD-L2–PD-11 | PD-1 targeting could be advantageous when PD-L2 contributes to immune suppression.1 VEGF-A blockade does not directly close this checkpoint axis.2 Cellular binding site | Binds PD-1-expressing immune cells, including activated/ exhausted T cells.3 VEGF-enhanced binding: ivonescimab data3 | Binds PD-L1-expressing tumor cells and immune (myeloid) cells.1,2 | Checkpoint arm can influence where the molecule is retained and where VEGF-A is neutralized.2 Preferential tumor localization must be demonstrated for each construct.† CD80/CD28/CTLA-4 network | Does not directly disrupt PD-L1-CD80 complexes.4 Preserves this interaction on APCs.4 | May disrupt these complexes, depending on the antibody.4 No inherent CD28 costimulation gain.4,5 | Does not unlock CD28-costimulation: CD80 can already engage CD28 within the complex. Disruption can instead expose CD80 to CTLA-4-mediated removal, potentially reducing available costimulation.4,5† Preserved inhibitory signaling | Removes PD-L2-mediated PD-1 signaling as well | Leaves PD-L2-mediated PD-1 signaling available | Preserving PD-L2 could maintain some physiological immune restraint, but a safety advantage for the bispecific cannot be assumed. Tina Cascone, MD, PhD 1. Yearley et al. Clin Cancer Res. 2017; 2. Miao et al. Cancer Res. 2025;85(suppl):6061. 3. Zhong et al, iScience 2025; 4. Zhao et al. Immunity. 2019; 5. Sugiura et al. Science. 2019. † Mechanistic inference, not proven comparative benefit.

The NEW ENGLAND JOURNAL of MEDICINE [PFS Kaplan–Meier figure; y-axis Percentage of Patients; x-axis Months 0–18] Sacituzumab tirumotecan + pembrolizumab: 51.1 (95% CI, 41.5–59.9) at 12 months Chemotherapy + pembrolizumab: 22.8 (95% CI, 15.3–31.3) at 12 months ORIGINAL ARTICLE MEETING OF THE EUROPEAN SOCIETY FOR MEDICAL ONCOLOGY Sacituzumab Tirumotecan–Pembrolizumab for NSCLC F. Zhou and Others Survival table Endpoint: Progression-free survival (figure label not shown; matches presented PFS by BICR) Sacituzumab tirumotecan + pembrolizumab Chemotherapy + pembrolizumab Landmark rates: 12 months: Sacituzumab tirumotecan + pembrolizumab 51.1% (CI 41.5–59.9) | Chemotherapy + pembrolizumab 22.8% (CI 15.3–31.3)

3815P SHR-4849 (IDE849), an ADC targeting delta-like ligand 3 (DLL3), in relapsed small cell lung carcinoma (SCLC) and other neuroendocrine carcinomas (NECs): long-term results from the phase 1 study Haifeng Liu, Liang Zhang, Linlin Wang, Yan Zhang, Longhua Sun, Jinhua Wen, Wenzhong Su, Xin Wang, Baogang Liu, Yunjian Huang, Yan Zhang, Peng Chen, Qiming Wang, Rong Qiao, Shengxiang Ren, Yang Wei, Zhengbo Song, Wenqing Zhu, Xiaoxia Lu, Lulu Yang BACKGROUND • Delta-like ligand 3 (DLL3), a protein that inhibits Notch signaling, is an attractive therapeutic target in small cell lung cancer (SCLC) and other neuroendocrine carcinomas (NECs) due to high expression in tumors of neuroendocrine lineage. • SHR-4849 (IDE849*) is an antibody-drug conjugate comprising a humanized anti-DLL3 IgG1 monoclonal antibody, linked to a DNA topoisomerase I inhibitor payload by a cleavable linker. • We initiated a first-in-human study of SHR-4849 in relapsed or metastatic SCLC and other NECs. SHR-4849 at doses ranging from 0.8 to 4.2 mg/kg showed a tolerable and manageable safety profile, favorable pharmacokinetic profile, and promising anti-tumor activity. METHODS • Study design: a multicenter, phase 1 study (NCT06443489). • Key eligibility criteria: – Histologically or cytologically confirmed relapsed or metastatic SCLC and other DLL3-expressing NECs. DLL3 positivity was defined as H-score >0. – Had progressive or recurrent disease after standard treatment(s). – Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1. – Stable CNS metastases following local therapy. • Treatment: Eligible patients received intravenous SHR-4849 at 0.8, 1.8, 2.4, 3.0, 3.5, and 4.2 mg/kg every 3 weeks. • This report presents the updated efficacy and safety data. – Data cutoff: Aug 21, 2026. – Median follow-up duration: 12.1 months (IQR, 8.2–16.0). RESULTS Table 1. Patient characteristics (SCLC N=103; Other NECs N=19; Total N=122) Data are n (%) or median (range). One (0.8%) patient had previously received T-cell-engaging therapy targeting DLL3. Table 2. Safety summary | Total (N=122) Any TRAE | 121 (99.2) Grade ≥3 | 73 (59.8) Leading to dose reduction | 23 (18.9) Leading to treatment discontinuation | 3 (2.5) Serious | 26 (21.3) Leading to death | 1 (0.8) Patients • A total of 122 patients were enrolled, including 103 (84.4%) with SCLC and 19 (15.6%) with other NECs (Table 1). Safety • The most common grade ≥3 treatment-related adverse events (TRAEs) were hematological toxicities (Figure 1). • Occurrence of grade ≥3 TRAEs and serious TRAEs increased with dose; the rates in the 2.4, 3.0, and 3.5 mg/kg cohorts were: – Grade ≥3 TRAEs: 33.3%, 70.3%, and 76.2%; – Serious TRAEs: 15.2%, 24.3%, and 26.2%. • Treatment-related interstitial lung disease occurred in 9 (7.4%) patients (grade 1–2: 8 [6.6%]; grade ≥3: 1 [0.8%]). • Only 2.5% of patients discontinued SHR-4849 due to TRAEs (Table 2). • One patient died due to TRAE (pneumonia). Figure 1. TRAEs occurring in ≥15% of patients Figure 2. Best percentage change in target lesions in SCLC patients Figure 3. PFS in SCLC: PFS events, n (%): 86 (84.3); Median PFS (95% CI): 6.8 months (5.5–8.1) Figure 4. OS in SCLC: 59.5% (95% CI, 49.1–68.5); OS events, n (%): 55 (53.9); Median OS (95% CI): 14.7 months (11.9–18.3) Table 3. Antitumor activity SCLC (≥1.8 mg/kg): All (n=101) | 2L Setting (n=51) | 3L Setting (n=34) | With brain metastases (n=26) | Other NECs (n=19) ORR, n (%; 95% CI): 69 (68.3%; 58.3–77.2) | 37 (72.5%; 58.3–84.1) | 21 (61.8%; 43.6–77.8) | 20 (76.9%; 56.4–91.0) | 10 (52.6%; 28.9–75.6) Confirmed ORR, n (%; 95% CI): 55 (54.5%; 44.2–64.4) | 31 (60.8%; 46.1–74.2) | 17 (50.0%; 32.4–67.6) | 17 (65.4%; 44.3–82.8) | 7 (36.8%; 16.3–61.6) DCR, n (%; 95% CI): 94 (93.1%; 86.2–97.2) | 50 (98.0%; 89.6–100.0) | 32 (94.1%; 80.3–99.3) | 26 (100.0%; 86.8–100.0) | 16 (84.2%; 60.4–96.6) ORR, objective response rate. DCR, disease control rate. Efficacy in SCLC (Table 3; Figures 2-5) • In patients receiving SHR-4849 at ≥1.8 mg/kg: Confirmed ORR was 54.5%, and DCR was 93.1%; Median PFS was 6.8 months; Median OS was 14.7 months, and 12-month OS rate was 59.5%. • In patients in the 2L setting receiving SHR-4849 at ≥1.8 mg/kg: Confirmed ORR was 60.8%, and DCR was 98.0%; Median PFS was 6.9 months (95% CI, 5.5–8.3); Median OS was 17.4 months (95% CI, 9.9–not reached), and 12-month OS rate was 60.0 (95% CI, 45.1–72.1). • In patients with brain metastases receiving SHR-4849 at ≥1.8 mg/kg: Confirmed ORR was 65.4%, and DCR was 100.0%. Efficacy in other NECs (Table 3; Figures 5-6) • Confirmed ORR was 36.8%, and DCR was 84.2%; • Median PFS was 7.0 months; • 9-month OS rate was 82.4% (95% CI, 54.7–93.9). Figure 5. The best percentage change in target lesions in other NECs Figure 6. PFS in other NECs: PFS events, n (%): 11 (57.9); Median PFS (95% CI): 7.0 months (4.2–15.2) CONCLUSIONS • SHR-4849 (IDE849) has promising efficacy in patients with relapsed SCLC and other NECs, with clinically meaningful ORR and DCR and encouraging PFS and OS results. • The safety profile is favorable, with few treatment discontinuations. • Registrational studies are planned in SCLC and other NECs. Note: *SHR-4849 is being developed as IDE849 by IDEAYA Biosciences, worldwide outside of greater China. Contact: Haifeng Liu (HFLIU2025@163.com)

[Photo of presenter standing beside poster; poster board 4404P] Decoding tumor gene expression from circulating chromatin at single-gene resolution Authors (partially obscured): ... Gunsagar S. Gulati, Damien Vasseur, Sahabeddin Sotudian, Karl Semaan, Marc Eid, Ji-Heui Seo, Mohamed-Amine Bani, Christophe Massard, Antoine Italiano, Antoine Hollebecque, Jean-Charles Soria, Fabrice André, Joaquim Bellmunt, Harshabad Singh, Andrew J. Aguirre, Brian M. Wolpin, Toni K. Choueiri, Sylvan C. Baca, Matthew L. Freedman 1 Dana-Farber Cancer Institute, Boston, MA; 2 Gustave Roussy, Villejuif, France; † Contributed equally as first authors INTRODUCTION [partially obscured by presenter; illegible] RESULTS Figure 1: Independent validation of tumor gene expression [box plot; Plasma epigenomic approach] Figure 2: Pancreatic cancer subtyping with PurIST [box plots incl. APEX] Figure 3: NECTIN4 expression and response to EV [waterfall plots] Figure 4: DLL3 expression across SCLC subtypes • In 34 held-out samples across 18 cancer types, plasma-inferred expression correlated with matched tumor RNA-seq (ρ = 0.79). SCLC signature detection reached 0.1% ctDNA. • APEX-based PurIST probabilities separated basal and classical pancreatic tumors (P = 0.010), with 80% overall classification accuracy. • In EV-treated patients, high versus low NECTIN4 expression was associated with 58% versus 0% ORR and longer PFS and OS. • Plasma-inferred DLL3 expression recapitulated SCLC subtype-associated patterns in independent tumor RNA-seq. • Across 417 patients, 43% had exceptionally high expression of targets for FDA-approved expression-based therapies. CONCLUSION • APEX enables noninvasive inference of tumor gene expression across cancer types. • Plasma-inferred expression supports transcriptional subtyping and identifies candidate biomarkers associated with treatment outcomes. • Functional liquid biopsy could help monitor tumor biology and guide precision oncology. ESMO Congress 2026, Madrid, Spain

[Slide 1] Opportunities For Liquid Biopsy Use Across The Lung Cancer Journey [Diagram: blood vessel with circulating analytes; ctDNA levels over time curve] Genomic changes; RNA; Methylation; CTC; Fragment analysis; TEP; Exosome APPLICATIONS: Screening, Diagnosis | Treatment monitoring | MRD detection, molecular genotyping | Detecting resistance CTDNA LEVELS; MRD detection limit; Early stage; Late stage; Acquired resistance TIME: Neoadjuvant monitoring | Adjuvant therapy | Predictive biomarkers | Response monitoring Genotyping assays High sensitivity assays – "MRD" tumour informed or tissue-free Modified from Garcia… Leighl. Br J Cancer 2022 Natasha B. Leighl [Slide 2] What do the guidelines recommend? Indication for ctDNA use | ESMO^1 2022 | NCCN^2 2026 | IASLC^3 2021 | ASCO^4-5 2026 Initial genotyping: Validated, sensitive ctDNA assays can guide targeted therapy selection for patients with advanced cancer | ✓ | ✓ | ✓ | ✓ Plasma-first/concurrent testing: ctDNA can be used when tissue is insufficient/unavailable or rapid molecular results are needed | ✓ | ✓ | ✓ | ✓ Negative results: Negative plasma findings warrant reflex tissue testing; sensitivity is lower for fusions and copy-number alterations | ✓ | ✓ | ✓ | ✓ Acquired resistance: ctDNA can identify resistance mechanisms at disease progression following targeted therapy | ✓ | ✓ | ✓ | ✓ Treatment monitoring: Routine serial ctDNA monitoring to guide treatment changes is NOT RECOMMENDED outside clinical trials | ✗ | ✗ | ✗ | ✗ Molecular residual disease: Routine ctDNA MRD-guided treatment decisions or surveillance are NOT RECOMMENDED outside clinical trials | ✗ | ✗ | ✗ | ✗ ESMO, European Society of Medical Oncology; NCCN, National Comprehensive Cancer Network; IASLC, International Association of the Study of Lung Cancer; ASCO, American Society of Clinical Oncology; PCO, Provisional Clinical Opinion; ctDNA, circulating tumour DNA; MRD, molecular residual disease 1. Pascual J et al. Ann Oncol. 2022;33(8):750-768; 2. National Comprehensive Cancer Network V6.2026; 3. Rolfo C et al. J Thorac Oncol. 2021;16(10):1647-1662; 4. Reuss J et al. J Clin Oncol. 2026;44:e15-e55 (ASCO Living Guidelines, ver. 2026.3.4); 5. Lockwood CM et al. JCO Oncol Pract. 2026;22:1468-1489 Natasha B. Leighl [Slide 3] Pitfalls Of Plasma cfDNA NGS Plasma cfDNA Testing vs. Tissue Testing for Biomarkers in NSCLC Biomarker class | Examples | Plasma cfDNA performance | Clinical takeaway Driver mutations | EGFR, KRAS G12C, BRAF V600E, MET exon 14 skipping, HER2 (ERBB2) | Comparable clinical outcomes. Similar outcomes when mutations are identified in plasma or tissue. | Plasma results can guide targeted therapy when positive. Fusion genes | ALK, ROS1, RET, NTRK1/2/3, NRG1 | Lower sensitivity than tissue. Especially with DNA-based assay. | A negative plasma result requires tissue testing. Gene amplification / copy number | MET amplification, HER2 amplification | Variable and generally lower sensitivity/specificity | Tissue confirmation often needed with current assays Protein expression | PD-L1, HER2 (IHC), MET (IHC) | Not assessable by plasma cfDNA. Protein expression cannot be measured in plasma cfDNA. | Tissue required (for IHC or other protein-based assays). False negatives • Lower sensitivity than tissue False positives Clonal hematopoiesis in ~25% Other • Knowing when to reflex to tissue – informative vs uninformative negative result • Unable to determine tissue of origin, cancer diagnosis, pathologic subtyping Natasha B. Leighl [Slide 4] PM ACCELERATE: Can We Adopt A Pre-diagnostic Plasma NGS, Tissue-sparing Approach? Week 1: Patients with suspected advanced lung cancer based on imaging (multidisciplinary assessment) Plan diagnostic tissue biopsy, complete staging, and draw liquid biopsy (plasma NGS) 18% insufficient tissue for NGS 8% no testing, 10% single gene Week 2: Plasma NGS informative: actionable driver (eg, RET fusion) | Plasma NGS informative: nonactionable driver (eg, KRAS G12D mutation) | Plasma NGS uninformative: (ctDNA not detected, not tumor-derived or nondriver) Week 3: Diagnostic tissue sampling; Cancer confirmed, pathologic subtyping as appropriate Expedite tissue diagnosis, start targeted therapy if indicated(a) (tissue NGS optional, not required) 51% ESCAT Tier 1 27% Week 4: Initiate PD-L1; Initiate tissue NGS testing (per standard of care) if plasma ctDNA uninformative or nondriver alteration identified Start immunotherapy with or without chemotherapy based on pathologic subtype, PD-L1 (tissue NGS pending(b)) 22% (KRAS, STK11, PIK3CA, TP53) (*1/20 FN) 18% cfDNA not detected 8% tissue AGA Week 5: Tissue NGS results Start treatment if clinically indicated Start treatment based on pathologic subtype, PD-L1, and tissue NGS results Week 6 ✓ Detects more actionable alterations than tissue NGS alone (83% vs 71%) ✓ Higher costs (~$4000 CAD), higher QALYs ✓ Shorter mean time to treatment (15 v 23+ days) García-Pardo M…Leighl NB. JAMA Netw Open. 2023; Zhu J… Leighl NB WCLC 2026 Natasha B. Leighl

Different toxicity profiles, not different degrees of toxicity Trastuzumab deruxtecan DESTINY-Breast06 | Sacituzumab govitecan TROPiCS-02 | Datopotamab deruxtecan TROPION-Breast01 Nausea, any grade 65.9% | Neutropenia, G3–4 51% | Nausea, any grade 51.1% Alopecia, any grade 45.4% | Nausea, any grade 55% | Alopecia, any grade 36.4% Fatigue, any grade 45.4% | Alopecia, any grade 46% | Dry eye, any grade 21.7% Neutropenia, G3–4 20.7% | Diarrhoea, G3–4 9% | Stomatitis, G3–4 6.4% ILD / pneumonitis 11.3% (G5 0.7%) | ILD / pneumonitis Not reported | ILD / pneumonitis 3.3% (one G5) Watch: interstitial lung disease | Watch: myelosuppression and diarrhoea | Watch: stomatitis and ocular surface toxicity Comorbidity should drive the choice: lung disease argues against T-DXd, low marrow reserve against SG, ocular surface disease against Dato-DXd. Yeon Hee Park Rates as compiled in Kang S, Kim SB. Ther Adv Med Oncol 2025;17:1–32. Cross-trial comparison — populations and reporting conventions differ.

[Slide 1] MDT-BRIDGE Study Design Global, Non-randomized, Phase 2 Study Neoadjuvant period A | Neoadjuvant period B | Adjuvant/ consolidation period Baseline MDT resectability assessment (resectable or borderline resectable) Durvalumab + investigator's choice of platinum-based CT Q3W for 2 cycles MDT reassessment Restaging/ optional pathologic confirmation Resectable -> Durvalumab + investigator's choice of platinum-based CT Q3W for 1-2 cycles -> Surgery Unresectable -> CRT Durvalumab monotherapy Q4W for 12 cycles C1D1 (Baseline) | C2D1 | End of C2 | Pre-CRT (after 2 cycles of D + CT) | Pre-surgery (after 3-4 cycles of D + CT)* | Additional post-surgery ctDNA analysis is planned Tissue collection timepoints Plasma collection timepoints for neoadjuvant ctDNA analysis Key inclusion criteria and study requirements • Aged ≥18 years and previously untreated • Pathologically confirmed, resectable or borderline resectable stage IIB-IIIB NSCLC (AJCC 8th edition1) • EGFR/ALK wild type (per local test) • WHO/ECOG PS 0-1 • At least 1 target lesion not previously irradiated • Pre-operative RT not allowed • Primary endpoint: Resection rate, the proportion of all patients (FAS) who underwent definitive surgery • Key secondary endpoints: Resection rate by baseline MDT resectability; ORR; pCR; EFS; PFS; safety; and ctDNA clearance from baseline (pre-treatment) to pre-surgery/pre-CRT • ctDNA clearance from baseline to all neoadjuvant timepoints was assessed in evaluable patients using the tumor-informed Foresight CLARITY™ custom assay (LoD95: ≤1 ppm variant allele fraction [VAF]) - After whole genome sequencing (WGS) of tumor tissue and paired germline blood samples, a personalized set of up to 5000 tumor-specific targets, including phased variants, was used to analyze ctDNA *Excluding patients who switched to CRT, 42/112 (37.5%) deemed resectable at MDT reassessment completed 4 cycles. AJCC, American Joint Committee on Cancer; CXDX, cycle X day X; D, durvalumab; ECOG, Eastern Cooperative Oncology Group; EFS, event-free survival; FAS, full analysis set (i.e., all patients who received ≥1 dose of study intervention); LoD95, 95% limit of detection; ORR, objective response rate; pCR, pathological complete response; PFS, progression-free survival; PS, performance status; ppm, parts per million; QXW, once every X weeks; RT, radiotherapy; WHO, World Health Organization. 1. Amin MB, et al. AJCC Cancer Staging Manual. 8th ed. NY, NY: Springer; 2017. [Slide 2] ctDNA Detection and Dynamics During Neoadjuvant Treatment (ctDNA BEP)* • 99.1% of patients with biomarker evaluable samples were ctDNA+ at baseline • VAF levels decreased markedly from baseline after 1-2 cycles of neoadjuvant durvalumab + CT - ctDNA was detected as low as 0.000046% VAF (0.46 ppm VAF) at any neoadjuvant timepoint Neoadjuvant period A | Neoadjuvant period B† VAF, % | VAF, parts per million ctDNA detection: Negative; Positive C1D1 (Baseline) | C2D1 | End of C2 | Pre-surgery/pre-CRT ctDNA detected, n/N‡ (%) | 105/106 (99.1) | 96/104 (92.3) | 56/79 (70.9) | 57/91 (62.6) Median VAF, % (ppm)¶ | 0.3339 (3,339) | 0.0152 (152) | 0.00779 (77.9) | 0.00901 (90.1) Mean, % | 2.648 | 0.432 | 0.332 | 0.885 Range, % | 0.00265-33.8 | 0.000109-15.9 | 0.00006-7.59 | 0.000046-26.4 DCO, 12 January 2026. *Analyses are based on patients with evaluable plasma samples at each respective timepoint; denominators may, therefore, vary. †This period included both resectable and unresectable patients, for whom there was variability in the timings of sample collection at the pre-surgery and pre-CRT timepoints and in the number of neoadjuvant treatment cycles (i.e., resectable patients received an additional 1-2 cycles of neoadjuvant durvalumab + CT). ‡At each timepoint, n patients with ctDNA detected among N patients with biomarker evaluable samples. ¶Based solely on patients with ctDNA detected. ppm, parts per million; VAF, variant allele fraction. [Slide 3] Association of Neoadjuvant ctDNA Clearance with pCR Status (ctDNA BEP) • As early as C2D1, higher rates of neoadjuvant ctDNA clearance were observed in patients with pCR vs non-pCR • The positive predictive value (PPV) of ctDNA clearance for pCR was highest at C2D1 (83.3%), potentially enabling early identification of responders after only 1 cycle of durvalumab + CT - The negative predictive value (NPV) was highest at pre-surgery (85.0%) ctDNA clearance from baseline* Clearance rate, % C2D1: pCR No 2.1% (1/48); pCR Yes 19.2% (5/26) End of C2: pCR No 14.6% (6/41); pCR Yes 61.9% (13/21) Pre-surgery: pCR No 20.9% (9/43); pCR Yes 76.0% (19/25) Predictive value of ctDNA clearance at different timepoints for pCR† Timepoint | PPV (95% CI) | NPV (95% CI) C2D1 | 83.3% (43.7-97.0) | 69.1% (57.4-78.8) End of C2 | 68.4% (46.0-84.6) | 81.4% (67.4-90.3) Pre-surgery | 67.9% (49.3-82.1) | 85.0% (70.9-92.9) DCO, 12 January 2026. Error bars represent 95% CIs calculated using the Wilson score method. *Analyses are based on patients with evaluable plasma samples at each respective timepoint, denominators may, therefore, vary. †95% CIs for PPV and NPV calculated using the Wilson score method. PPV, positive predictive value (i.e., the likelihood of ctDNA- patients achieving pCR). NPV, negative predictive value (i.e., the likelihood of ctDNA+ patients not achieving pCR); [Slide 4] Association of ctDNA Clearance at Pre-surgery/pre-CRT with Preliminary EFS and PFS (ctDNA BEP)* • Among resected patients, none with pre-surgery ctDNA clearance had an EFS event, while 12-month EFS in patients without clearance was 88.9% • Among CRT-treated patients, only one with pre-CRT ctDNA clearance had a PFS event (after 12 months), while 12-month PFS in patients without clearance was 77.5%, although patient numbers were small EFS (resected patients) ctDNA clearance | Non-ctDNA clearance No. events / no. patients (%) | 0/28 (0) | 5/40 (12.5) 100% 88.9% (95% CI, 72.8-95.7) Probability of EFS Time from first dose (months) No. at risk Clearance 28 28 26 18 10 2 Non-clearance 40 40 35 25 11 0 PFS (CRT-treated patients)† ctDNA clearance | Non-ctDNA clearance No. events / no. patients (%) | 1/4 (25.0) | 5/16 (31.3) 100% 77.5% (95% CI, 44.8-92.3) Probability of PFS Time from first dose (months) No. at risk Clearance 4 4 3 3 2 0 Non-clearance 16 14 11 7 6 1 DCO, 12 January 2026. *Exploratory analyses of endpoints with limited maturity and small patient numbers; 12-month rates calculated by Kaplan-Meier method, with 95% CIs calculated by complementary log-log method. †PFS was defined as the time from the first dose of any study treatment until PD, assessed by the investigator per RECIST version 1.1, or death (by any cause in the absence of progression) regardless of whether the patient withdrew from therapy or received another anticancer therapy prior to progression. Survival table Endpoint: EFS (resected patients) and PFS (CRT-treated patients) by pre-surgery/pre-CRT ctDNA clearance EFS - ctDNA clearance | 28 EFS - Non-ctDNA clearance | 40 PFS - ctDNA clearance | 4 PFS - Non-ctDNA clearance | 16 12-month EFS: ctDNA clearance 100%; non-clearance 88.9% (95% CI, 72.8-95.7) 12-month PFS: ctDNA clearance 100%; non-clearance 77.5% (95% CI, 44.8-92.3)

[Slide 2] B7-H3 ADCs Ifinatamab deruxtecan (I-DXd)1 | Risvutatug Rezetecan (Ris-Rez)3 | Tambotatug Pelitecan (Tam-Peli)2 DAR ≈ 4 | DAR ≈ 4 | DAR ≈ 8 Linker: Cleavable tetrapeptide | Linker: Cleavable tetrapeptide | Linker: Cleavable tumour microenvironment-activatable (TMALIN®) Payload: Deruxtecan (DXd) | Payload: Rezetecan (HS-9265) | Payload: YL0010014 (C24) Humanised B7-H3 IgG1 mAb | Fully human B7-H3 IgG1 mAb | Fully human B7-H3 IgG1 mAb • Payload-class: TOP1 inhibitor | • Payload-class: TOP1 inhibitor | • Payload-class: TOP1 inhibitor • Payload with short systemic half-life | • Payload half-life not reported | • Payload with long systemic half-life • Bystander antitumour effect | • Plausible bystander antitumour effect | • Bystander antitumour effect Same target (B7-H3), same payload class, but different ADC engineering (DAR, linker, payload type) Noemi Reguart, MD, PhD | ESMO 2026 1.- Rudin C et al. WCLC 2024; 2.- Zhang L et al. WCLC 2026; 3.- Wang J et al. WCLC 2026 [Slide 3] PHASE 3 TRIALS in RELAPSED SCLC Target | ADC | Study Name | Design | Previous Tarlatamab | NCT ID B7-H3 | Tam-Peli | TAISHAN-302 | YL201 vs topotecan | - | NCT06612151 B7-H3 | Ris-Rez | ARTEMIS-008 | HS-20093 vs topotecan | - | NCT06498479 B7-H3 | MHB088C | MHB088C-P-301 | MHB088C vs inv. choice | - | NCT06954246 B7-H3 | Ris-Rez | EMBOLD | GSK5764227 vs topotecan | Allowed | NCT07099898 B7-H3 | I-DXd | IDeate-Lung02 | I-DxD vs inv. choice | - | NCT06203210 TROP2 | SG | EVOKE-SCLC-04 | SG vs topotecan | Allowed | NCT06801834 EGFR-HER3 | Iza-bren | PANKU-Lung03 | BL-B01D1 vs topotecan | - | NCT06500026 DLL3 | Zoci | DLLEVATE | ZL-1310 vs. inv. choice | Allowed | NCT07218146 SEZ6 | ABBV-706 | M23-384 | ABBV-706 vs. inv. choice | Required | NCT07365241 Increasing Accommodation of Prior Tarlatamab Exposure Noemi Reguart, MD, PhD | ESMO 2026 [Slide 4] WHAT'S NEXT? ADC COMBINATIONS 1L Therapy (Induction | Maintenance) → 2L Therapy Today: PE/IO | IO → Tarlatamab; ADC Tomorrow (DeLLphi 305): PE/IO | TCE + IO → ADC ? The Future: PE/IO + TCE → ADC ± bispecific ADC + IO ± CP → TCE ADC + TCE → ? Noemi Reguart, MD, PhD | ESMO 2026 1.- Adapted from Reguart N. ASCO 2026; PE, platinum-etoposide; TCE, T-Cell Engager; ADC, Antibody-Drug Conjugate; IO, immunotherapy

In patients with baseline-detected plasma EGFRm, rate of EGFRm clearance was similar between treatment arms at Weeks 1, 2, 3, and 6 - 421 / 557 (76%) patients enrolled in FLAURA2 had an evaluable baseline plasma EGFRm ddPCR result (ctDNA analysis set)1 - Of those patients included in the ctDNA analysis set, plasma EGFRm at baseline was detected in 147 / 212 (69%) patients in the osi + plat-pem arm and 161 / 209 (77%) patients in the osi mono arm1 Timepoint | Arm | n | EGFRm cleared | EGFRm not cleared Week 1 | Osi + plat-pem | n=104 | 38% (95% CI 29, 47) | 62% (95% CI 53, 71) Week 1 | Osi mono | n=131 | 42% (95% CI 34, 51) | 58% (95% CI 49, 66) Week 2 | Osi + plat-pem | n=113 | 64% (95% CI 55, 72) | 36% (95% CI 28, 45) Week 2 | Osi mono | n=133 | 64% (95% CI 55, 72) | 36% (95% CI 28, 45) Week 3 | Osi + plat-pem | n=109 | 81% (95% CI 72, 88) | 19% (95% CI 12, 28) Week 3 | Osi mono | n=140 | 71% (95% CI 63, 79) | 29% (95% CI 21, 37) Week 6 | Osi + plat-pem | n=105 | 88% (95% CI 80, 93) | 12% (95% CI 7, 20) Week 6 | Osi mono | n=137 | 89% (95% CI 83, 94) | 11% (95% CI 6, 17) Y-axis: Proportion of patients (%) Legend: Osi + plat-pem: EGFRm not cleared / EGFRm cleared; Osi mono: EGFRm not cleared / EGFRm cleared

2805eP - First-line BRAF/MEK Inhibition versus Immune Checkpoint Inhibitor–Based Therapy in BRAF V600E-Mutant Metastatic NSCLC: A Meta-Analysis Yakup Ergun, Rohit Gosain, Rahul Gosain Department of Medical Oncology, Bower Hospital, Diyarbakır, Türkiye, Roswell Park Comprehensive Cancer Center, Buffalo, USA, Wilmot Cancer Institute, University of Rochester, Rochester, NY, USA BACKGROUND The optimal first-line treatment for patients with BRAF V600E-mutant metastatic NSCLC remains unclear. While current ESMO and NCCN guidelines prioritize BRAF/MEK inhibition, no randomized trials have directly compared targeted therapy with immune checkpoint inhibitor (ICI)–based regimens. We performed a meta-analysis of retrospective studies to compare efficacy outcomes between these strategies. METHODS A systematic literature search was conducted in PubMed, Embase, and the Cochrane Library. Retrospective studies comparing first-line BRAF/MEK inhibitor combinations versus ICI-based therapies (monotherapy or chemoimmunotherapy) were included. The primary endpoints were progression-free survival (PFS) and overall survival (OS), while objective response rate (ORR) was evaluated as a secondary endpoint. Effect sizes were pooled as hazard ratios (HRs) for survival outcomes and odds ratios (ORs) for ORR with 95% confidence intervals (CIs) using fixed- or random-effects models based on heterogeneity. Heterogeneity was assessed using the I² statistic. RESULTS Four retrospective studies including 589 patients (BRAF/MEK: n=384; ICI-based: n=205) were analyzed. No statistically significant difference in PFS was observed between the two strategies (HR 1.22; 95% CI 0.89–1.68; p=0.22; I²=0%); however, a numerical trend favored BRAF/MEK inhibition. Similarly, OS did not differ significantly (HR 1.01; 95% CI 0.59–1.73; p=0.97), although substantial heterogeneity was present (I²=69%). In contrast, ORR was significantly higher with BRAF/MEK inhibition compared to ICI-based therapy (OR 0.60; 95% CI 0.37–0.97; p=0.04; I²=0%). OS forest plot Study or Subgroup | log[Hazard Ratio] | SE | Weight | Hazard Ratio IV, Random, 95% CI Di Federico et al. | -0.4559 | 0.2223 | 37.2% | 0.63 [0.41, 0.98] Johnson BE et al. | 0.555 | 0.3661 | 26.3% | 1.74 [0.85, 3.57] Wiesweg et al. | 0.0957 | 0.2308 | 36.5% | 1.10 [0.70, 1.73] Total (95% CI) | | | 100.0% | 1.01 [0.59, 1.73] Heterogeneity: Tau² = 0.15; Chi² = 6.44, df = 2 (P = 0.04); I² = 69% Test for overall effect: Z = 0.04 (P = 0.97) IO-Based | BRAF/MEKi PFS forest plot Study or Subgroup | log[Hazard Ratio] | SE | Weight | Hazard Ratio IV, Fixed, 95% CI Di Federico et al. | 0.1661 | 0.1677 | 94.2% | 1.18 [0.85, 1.64] Jia et al. | 0.7747 | 0.6735 | 5.8% | 2.17 [0.58, 8.12] Total (95% CI) | | | 100.0% | 1.22 [0.89, 1.68] Heterogeneity: Chi² = 0.77, df = 1 (P = 0.38); I² = 0% Test for overall effect: Z = 1.24 (P = 0.22) IO-Based | BRAF/MEKi ORR forest plot Study or Subgroup | IO-Based Events | Total | BRAF/MEKi Events | Total | Weight | Odds Ratio M-H, Fixed, 95% CI Di Federico et al. | 40 | 82 | 121 | 193 | 85.6% | 0.57 [0.34, 0.96] Jia et al. | 9 | 20 | 13 | 26 | 14.4% | 0.82 [0.25, 2.63] Total (95% CI) | | 102 | | 219 | 100.0% | 0.60 [0.37, 0.97] Total events | 49 | | 134 Heterogeneity: Chi² = 0.32, df = 1 (P = 0.57); I² = 0% Test for overall effect: Z = 2.08 (P = 0.04) BRAF/MEKi | IO-Based CONCLUSIONS In this meta-analysis of retrospective data, BRAF/MEK inhibition was associated with significantly higher response rates, while no statistically significant differences were observed in survival outcomes. A consistent numerical trend in PFS favoring targeted therapy suggests a potential efficacy signal that may not reach statistical significance due to the limited number of studies and patients. The substantial heterogeneity observed in OS further limits definitive interpretation. Overall, the current evidence base remains limited, and larger, well-designed studies with more homogeneous patient populations are required to better define the optimal first-line strategy in this molecular subgroup. REFERENCES 1. Wiesweg M, et al. J Thorac Oncol. 2025;20(9). 2. Di Federico A, et al. Lancet Oncol. 2025;26(10). 3. Jia B, et al. EBioMedicine. 2025;114. 4. Johnson BE, et al. JTO Clin Res Rep. 2022;3(5). CONTACT Yakup ERGUN, MD E-mail: dr.yakupergun@gmail.com X account: @dr_yakupergun No conflicts of interest

[Slide 1] MADRID 2026 ESMO congress Pre-treatment Tumor Immune-Repertoire Features associate with ICI-related pneumonitis in NSCLC Predicting ICI-related pneumonitis using immunogenomic and clinical features derived from real-world data Authors: Jarushka Naidoo1, Florent Guinot2, Rajat Mohindra3, Adrian B.Roth3, Svetlana Lyalina4, George Quartey4, Samuel Torres Florez3, Ramanandan Prabhakaran3, Cameron MacPherson2, G Scott Chandler3 Author affiliations: Beaumont Hospital and RCSI University of Health Sciences, Dublin, Ireland1 Institut Roche2, Roche3, Genentech4 [Slide 2] Key Results 1: Pretreatment TCR diversity associated with development of ICI pneumonitis The association model estimates risk; stability analyses ask whether the same feature is repeatedly selected UNIVARIABLE ANALYSIS Adjusted Cox Model TCR: TCR_richness; TCR_shannon_entropy; TCR_inverse_simpson; TCR_clonality_score; TCR_top_clone_frequency; TCR_gini_index; TCR_cdr3_length_mean TCR Shannon entropy HR 1.42 95% CI 1.13–1.79 raw p = 0.002 BH q = 0.067 Nominal association; it did not cross the BH-adjusted 0.05 threshold. Dashed line = HR 1 (no association). Points to the right indicate higher hazard. STABILITY ANALYSIS Multivariable analyses Ranked EFP Scores Features ranked by Expected False Positives (labeled if EFP ≤ 10) Rank (1 = lowest EFP) Feature Category: BCR heavy, IGHA, IGHD, IGHE, IGHG, IGHM, IGK, IGL, TCR TCR_shannon_entropy 0.3381015 | IGHM_richness 0.5508525 | IGK_shannon_entropy 2.6057871 TCR Shannon entropy Most stable feature (lowest EFP value) Risk marker: higher pretreatment TCR Shannon entropy = a broader, more even T-cell clonotype distribution. Mechanistic hypothesis: broader T-cell engagement may create greater opportunity for off-target lung inflammation; causality is not established.

[Slide 2] From safety descriptions to patient experience Years on treatment make daily function and chronic symptom burden clinically consequential. How early-phase trials describe toxicity [Pie chart A: No ISJETs 12%; Acceptable 2%; Favorable 7%; Multiple ISJETs 10%; Manageable 14%; Comparable 13%; Tolerable 42%] ISJET: Investigator-assigned Subjective or Judgmental Efficacy and Toxicity 100 early-phase abstracts, ASCO 2024 88% used subjective toxicity language PROs measure symptoms and treatment burden directly. ESMO-MCBS: evidence before QoL credit Primary or secondary endpoint Validated instrument (overall/global QoL) Meaningful effect size Transparent missing-data analysis PROs reveal how chronic toxicity affects daily life, beyond clinician-graded adverse events. Phillips, Watson & Camidge. JTO 2025;20:589-596, Fig. 2A. Oosting et al. Ann Oncol 2023;34:431-439; ESMO QoL checklist. [Slide 3] DESTINY-Lung04: better control, better daily life? Overall treatment tolerability [Stacked bar chart, Weekly assessments BL, 1-12: Proportion of patients bothered by treatment side effects (%)] T-DXd n=226 Pembro + chemo n=220 During the first 12 weeks, the proportion of patient-weeks with high side-effect bother was 13.4% in both arms* (score of 'very much' or 'quite a bit') Legend: Not at all; A little bit; Somewhat; Quite a bit; Very much TTD in cough T-DXd (n=227) Pembro + chemo (n=227) Median TTD, months (95% CI) 26.2 (17.1, NC) 11.0 (7.4, NC) Hazard ratio (95% CI) 0.66 (0.44, 0.99) [Kaplan-Meier curves: Probability of no sustained deterioration vs Time from randomization (months) 0-36] No. at risk (months 0,3,...,36) T-DXd: 227 106 73 58 41 34 25 17 11 7 4 3 0 Pembro + chemo: 227 92 54 38 21 12 10 7 6 4 3 2 0 Overall treatment tolerability was similar in both arms Novello et al. ESMO 2026; LBA74. Survival table Endpoint: Time to deterioration (TTD) in cough Arm | N | Median | Median CI T-DXd | 227 | 26.2 mo | 17.1-NC Pembro + chemo | 227 | 11.0 mo | 7.4-NC HR: 0.66 HR CI: 0.44-0.99 [Slide 4] Mechanisms of Acquired Resistance to Kinase Inhibitors in Lung Cancer Mutations in the drug target - Impact drug binding Bypass Signaling - Oncogene Amplification; Oncogene Rearrangement Mutations in Downstream Effectors - PI3K -> AKT; RAS -> RAF -> MEK -> ERK; [Cell Cycle diagram: G1, G0, G1/S Checkpoint, S, G2, G2/M Checkpoint, M, M/G1 Checkpoint] State Transformation - [histology image] Small Cell Lung cancer; Squamous Cell lung cancer Cohen at al. Nat Rev Drug Discov 2017

[Slide 2] Temab-A: A Novel Anti-c-Met Protein-Targeting Antibody-Drug Conjugate - Increased c-Met protein (also known as MET protein) is common in solid tumors, including NSCLC, and is associated with poor clinical outcomes1-3 - Telisotuzumab adizutecan (Temab-A; ABBV-400) is an antibody-drug conjugate (ADC) comprising the c-Met-targeting antibody telisotuzumab conjugated to a novel potent topoisomerase 1 inhibitor payload, adizutecan - In an ongoing first-in-human Phase 1 study (NCT05029882), Temab-A monotherapy showed promising efficacy with a confirmed objective response rate (ORR) of 50% and 63% in heavily pretreated patients with advanced epidermal growth factor receptor wildtype (EGFRwt) and mutated (EGFRmut) non-squamous, non-small cell lung cancer (NSCLC), respectively4,5 Temab-A Mechanism of Action1 Validated c-Met-targeted antibody (telisotuzumab) Potent Top1i payload (adizutecan) Stable linker - Minimizes free-payload toxicity Tumor cell; c-Met protein; Top1i inhibits DNA replication; Leading to apoptosis Antitumor effect of Temab-A does not rely on c-Met signaling, unlike c-Met protein-targeting TKIs and mAbs Herein, we present updated overall survival (OS) data for those 2 patient cohorts with a median follow-up of 30.1 months mAb, monoclonal antibody; TKI, tyrosine kinase inhibitor; Top1i, topoisomerase 1 inhibitor. 1. Moosavi F, et al. Crit Rev Clin Lab Sci. 2019;56:533-566. 2. Mer AH, et al. Drug Deliv Transl Res. 2024;14:2963-2988. 3. Motwani M, et al. J Thorac Oncol. 2021;16:S1169-S1170. 4. De Miguel M, et al. Ann Oncol. 2024;35(Suppl 2):S805-S806. 5. Camidge DR, et al. J Clin Oncol. 2025;43:8512. [Slide 3] In the Overall EGFRwt Cohort Unselected for c-Met, Median PFS Was 6.8 Months and Median OS Was 13.8 Months in Patients Who Received Temab-A - At the June 02, 2026, data cutoff, median follow-up was 30.1 months (range, 0.4-35.5) - ORR was 50% (24/48) (95% CI, 35.2-64.8) - Median DOR was 5.6 months (95% CI, 4.2-11.3) Progression-Free Survival PFS EGFRwt Overall (n=48) Events, n (%) 40 (83) Median PFS, months (95% CI) 6.8 (5.4-7.3) Number at risk (Overall, months 0-33): 48 45 41 37 34 32 24 18 15 13 11 10 9 8 5 5 5 5 5 4 5 5 5 4 5 4 4 3 2 2 2 1 1 0 Overall Survival OS EGFRwt Overall (n=48) Events, n (%) 33 (69) Median OS, months (95% CI) 13.8 (9.0-25.2) Number at risk (Overall, months 0-36): 48 46 46 44 41 39 36 35 31 30 26 25 24 24 23 23 20 19 18 17 16 15 14 13 13 13 12 9 9 7 7 4 3 3 2 1 0 + Censored Data cutoff date: 02 June 2026. DOR, duration of response; EGFRwt, epidermal growth factor receptor wildtype; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; Temab-A, telisotuzumab adizutecan. Survival table Endpoint: PFS and OS, EGFRwt overall cohort (Temab-A) Arm | N | Median | Median CI EGFRwt overall - PFS | 48 | 6.8 months | 5.4-7.3 EGFRwt overall - OS | 48 | 13.8 months | 9.0-25.2 [Slide 4] In the EGFRmut Cohort Unselected for c-Met, Median PFS Was 10.9 Months and Median OS Was 16.2 Months in Patients Who Received Temab-A - At the June 02, 2026, data cutoff, median follow-up was 29.5 months (range, 0.4-34.8) - ORR was 63% (26/41) (95% CI, 46.9-77.9) o Median DOR was 9.7 months (95% CI, 8.3-13.5) Progression-Free Survival PFS EGFRmut Overall (n=41) Events, n (%) 39 (95) Median PFS, months (95% CI) 10.9 (8.3-12.3) Number at risk (Overall, months 0-33): 41 40 40 37 36 34 31 29 29 26 20 16 15 10 9 7 4 4 4 3 3 3 3 3 2 2 2 1 1 0 0 0 0 0 Overall Survival OS EGFRmut Overall (n=41) Events, n (%) 31 (76) Median OS, months (95% CI) 16.2 (11.6-19.9) Number at risk (Overall, months 0-36): 41 40 40 39 38 38 37 36 35 34 30 27 26 25 25 24 21 18 16 15 13 11 10 10 9 9 9 8 6 6 3 1 1 1 1 0 0 + Censored Data cutoff date: 02 June 2026. DOR, duration of response; EGFRmut, epidermal growth factor receptor mutated; OS, overall survival; PFS, progression-free survival; Temab-A, telisotuzumab adizutecan. Survival table Endpoint: PFS and OS, EGFRmut overall cohort (Temab-A) Arm | N | Median | Median CI EGFRmut overall - PFS | 41 | 10.9 months | 8.3-12.3 EGFRmut overall - OS | 41 | 16.2 months | 11.6-19.9

[Slide 2] DESTINY-Lung04 trial design A randomized, open-label, multicenter, Phase 3 trial (NCT05048797)1,2 Patients • Unresectable, locally advanced, or metastatic nonsquamous NSCLC • Treatment naïve for advanced disease • HER2 exon 19 or 20 mutation by central/local test • ECOG PS 0 or 1 • Asymptomatic/stable/treated brain metastases permitted • ≥1 measurable lesion per RECIST 1.1 R 1:1, N=454 T-DXd* n=227 Pembro† + platinum chemo‡ + pemetrexed§ n=227 Endpoints Primary • PFS (BICR) Secondary • OS • ORR and DOR (BICR) • PFS2 (investigator) • Safety and tolerability • PROs Exploratory • PROs Stratification factors • Presence/history of brain metastases (yes vs no) • Smoking history status (ever vs never) We report PRO endpoints at primary analysis (DCO: June 9, 2026) PFS, ORR, and DOR all per RECIST 1.1. Patients with objective radiological CNS disease progression per RECIST 1.1 could continue with their assigned trial treatment if, in the investigator's opinion, they continued to receive clinical benefit and were without any discontinuation criteria. Treatment crossover was not permitted *5.4 mg/kg IV Q3W until disease progression, unacceptable toxicity, or other discontinuation criteria; †200 mg IV Q3W until disease progression, unacceptable toxicity, or other discontinuation criteria; ‡investigator choice of cisplatin (75 mg/m2) or carboplatin (AUC 5) IV Q3W for up to four cycles; §500 mg/m2 IV Q3W until disease progression, unacceptable toxicity, or other discontinuation criteria 1. Rotow J, et al. Oral presentation at WCLC 2026 (PL03.08); 2. NCT05048797. Updated July 10, 2026. Available from: https://clinicaltrials.gov/study/NCT05048797 (Accessed September 15, 2026) [Slide 3] Overall treatment tolerability (PGI-TT) "In the last 7 days, how bothered were you by the side effects of your cancer treatment?" [Stacked bar chart by week: Weekly assessments (BL, weeks 1-12); Q3W assessments (weeks 15-69); EOT] T-DXd n=226 Pembro + chemo n=220 Y axis: Proportion of patients bothered by treatment side effects (%) Legend: Not at all | A little bit | Somewhat | Quite a bit | Very much During the first 12 weeks, the proportion of patient-weeks with high side-effect bother* was 13.4% in both arms (score of 'very much' or 'quite a bit') Proportions of patients reporting overall and high side-effect bother were similar between arms throughout treatment Timepoints after Week 69 through Week 105 have been excluded owing to low sample size (<10% of patients in the safety analysis set, either arm) Assessed in the safety analysis set. Compliance rates: 65.9% T-DXd vs 70.5% pembro + chemo (baseline); 75.6% T-DXd vs 72.9% pembro + chemo (overall). The PGI-TT assessed how a patient perceived the overall bother of cancer treatment-related side effects over the past week *Calculated as the number of weeks with high side-effect bother divided by the number of total weeks on treatment across all patients (restricted to first 12 weeks of treatment and excluding missing visits) BL, baseline [Slide 4] Symptoms related to AEs (EORTC QLQ-C30 / EORTC IL / PRO-CTCAE) T-DXd vs pembro + chemo (measured across the duration of treatment) Generally better with T-DXd (8/15) • CoughE • DyspneaE • General painE • Chest painE • Watery eyesP • HeadacheP • Arm/leg swellingP • RashP Similar between arms (2/15) • FatigueE • ConstipationE Generally worse with T-DXd (5/15) • Hair lossP • Appetite lossE • NauseaE • DiarrheaE • VomitingE E: assessed via EORTC forms P: assessed via PRO-CTCAE For the majority of the 15 symptoms assessed, patient-reported symptom experience during treatment was generally better with T-DXd than with pembro + chemo Symptoms related to AEs were assessed in the safety analysis set (T-DXd n=226; pembro + chemo n=220). Analyses were descriptive only. Compliance rates for PRO-CTCAE: 62.6% T-DXd vs 69.5% pembro + chemo (baseline); 72.1% T-DXd vs 71.3% pembro + chemo (overall). Compliance rates for EORTC IL123: 66.8% T-DXd vs 71.5% pembro + chemo (baseline); 69.4% T-DXd vs 60.6% pembro + chemo (overall). Overall compliance rates for EORTC IL129: 77.7% T-DXd vs 75.6% pembro + chemo

2668P Phase II Randomized Trial of Carboplatin, Pemetrexed, and Bevacizumab with and without Atezolizumab in Stage IV Non-Squamous Non-Small Cell Lung Cancer (NSCLC) Patients Who Harbor a Sensitizing EGFR Mutation (EGFRm) or Have Never Smoked (NS) Joseph Treat, J. Nicholas Bodor, Jyoti D. Patel, Hossein Borghaei, Bruna Pellini, Benjamin P. Levy, Chao H. Huang, Jonathan E. Dowell, Joel W. Neal, Sonam Puri, Jorge J. Nieva, Christian A. Thomas, Mark A. Socinski, Kathleen C. Kerrigan, Divya M Gupta, Jinhong Cui, Sydney Whorral, Eric A. Ross, Margie L. Clapper, Heather A. Wakelee BACKGROUND • The emergence of immunotherapy (IO) has changed the therapy landscape for NSCLC. However, patients who have never smoked (NS) or have EGFR mutations (EGFRm) derive little benefit from PD-1(-L1) inhibitors. Developing a strategy that overcomes IO resistance for these patients remains a critical need. • Subgroup analyses from IMpower150 showed superior PFS when the PD-L1 inhibitor atezolizumab was added to carboplatin, paclitaxel, and bevacizumab as compared to carboplatin, paclitaxel, and bevacizumab alone in patients with EGFRm NSCLC, however no OS benefit was observed. • Subsequent prospective trials evaluating IO combined with anti-VEGF and chemotherapy in EGFRm NSCLC have produced conflicting results. Positive findings with these combinations have largely been in trials where the comparator arm includes only platinum-based chemotherapy bringing into question the relative contribution of the IO and the anti-VEGF components to the benefit seen in these studies. • This randomized, phase II, multi-center trial (NCT03786692) evaluated the efficacy of combining atezolizumab with carboplatin, pemetrexed, and bevacizumab (ACPB) vs. carboplatin, pemetrexed and bevacizumab alone (ACP) in patients with stage IV non-squamous NSCLC who possess sensitizing EGFRm or NS. STUDY DESIGN Key Eligibility Criteria • Stage IV non-squamous NSCLC • Must have sensitizing EGFRm (exon 19 deletion or L858R) or have NS with wildtype NSCLC (no EGFR, ALK, or ROS1 alterations). • If EGFRm, must have progression or intolerance to ≥ 1 TKIs. • Chemotherapy and immunotherapy naive Stratification Factors • EGFR status (i.e. EGFRm vs. NS wildtype) • Prior receipt of anti-VEGF therapy Randomization – 2:1 n = 70 Arm ACPB: Atezolizumab (1200 mg) + Carboplatin (AUC 5) + Pemetrexed (500 mg/m2) + Bevacizumab (15 mg/m2) Q3 weeks for 4 cycles -> Maintenance: Atezolizumab + Pemetrexed + Bevacizumab n = 35 Arm CPB: Carboplatin (AUC 5) + Pemetrexed (500 mg/m2) + Bevacizumab (15 mg/m2) Q3 weeks for 4 cycles -> Maintenance: Pemetrexed + Bevacizumab Primary Endpoint • Progression-Free Survival (PFS) Secondary Endpoints • Overall Survival (OS) • Overall Response Rate (ORR) • Safety/Adverse Events (AEs) • PFS, OS, and ORR in subset of patients with EGFRm Statistical Considerations: • Planned enrollment was 117 patients, however the trial was closed prior to reaching this goal due to poor accrual in a setting of rapidly changing front-line therapy options and competing trials in EGFRm NSCLC. • Comparison of the primary endpoint PFS by arm in the ITT population (N=105) was performed using the log-rank test with a one-sided significance level of 10%. Secondary endpoints included OS, ORR, and safety in the ITT, and evaluation of efficacy endpoints in the EGFRm subset (N=90) using a two-sided significance level of 5%. • The first patient was enrolled in August 2019. Data cutoff date was October 20, 2025. Median follow-up time was 42.2 months. RESULTS Table 1. Baseline Characteristics of ITT study population (N = 105) Fig 1. PFS in ITT population by Arm* (N = 105) Arm | Median PFS (95% CI) | HR | One-Sided P-value ACPB | 7.7 months (6.8 – 10.8) | 0.79 | 0.16 CPB | 6.8 months (4.8 – 10.8) Number at risk: Arm=CPB 35 20 9 3 2 0 0; Arm=ACPB 70 45 22 8 5 3 1 (0, 5, 10, 15, 20, 25, 30 Months) *ORR (95% CI) was 46% (33% – 58%) with ACPB vs. 34% (19% – 53%) with CPB (P-value = 0.29). Fig 2. Subgroup Analysis of PFS in ITT Fig 3. OS in ITT population by Arm (N = 105) Arm | Median OS (95% CI) | HR | P-value ACPB | 17.3 months (15.3 – 23.3) | 0.77 | 0.30 CPB | 13.8 months (10.2 – 23.8) Fig 4. Subgroup Analysis of OS in ITT Fig 5. PFS in EGFRm subset by Arm (N = 90) Arm | Median PFS (95% CI) | HR | P-value ACPB | 7.6 months (6.8 – 9.8) | 0.86 | 0.53 CPB | 6.9 months (5.1 – 11.6) Number at risk: Arm=CPB 30 19 9 3 2 0 0; Arm=ACPB 60 38 18 7 4 2 0 Fig 6. Subgroup Analysis of PFS in EGFRm Fig 7. OS in EGFRm subset by Arm (N = 90) Arm | Median OS (95% CI) | HR | P-value ACPB | 16.3 months (13.6 – 22.8) | 0.93 | 0.78 CPB | 13.8 months (10.1 – 23.9) Fig 8. Subgroup Analysis of OS in EGFRm Table 2. Summary of Treatment-Related and Serious AEs in ITT * Patients – n (%) | ACPB (n = 68) | CPB (n = 33) Treatment-related Grade ≥ 1 AEs | 68 (100.0%) | 31 (93.9%) Treatment-related Grade 3 – 5 AEs | 35 (51.5%) | 17 (51.5%) Serious AEs | 27 (39.7%) | 13 (39.4%) *Final complete AE data will be available in a future manuscript publication. CONCLUSIONS • In this randomized Phase II trial, adding atezolizumab to carboplatin, pemetrexed, bevacizumab did not significantly improve PFS or OS in patients with Stage IV NSCLC with EGFRm or NS, though trends favored ACPB. No new safety signals were observed. • Subgroup analyses suggest females may obtain greater benefit from ACPB as compared to CPB. The small subset of patients with NS wildtype tumors derived a significant OS benefit from ACPB. • Correlative studies are ongoing using a biobank of specimens obtained from enrolled patients. The impact of tumor-based immune and stromal makers using multiplex immunofluorescence on clinical outcomes with ACPB is being evaluated. Corresponding author: Joseph Treat, MD (Joseph.Treat2@fccc.edu) This study was funded by the National Comprehensive Cancer Network (NCCN) Oncology Research Program from general research support provided by Genentech, Inc.

[Slide 2] EOT ctDNA analysis confirmed previous findings(1) and identified new putative lorlatinib resistance mechanisms - Data cutoff for this analysis: October 31, 2025 - Baseline ALK rearrangements were detected by ctDNA in 71 of 120 (59%) patients - Paired baseline and post-progression EOT ctDNA samples (n=17) revealed no new ALK resistance mutations at EOT - Changes in genomic alteration status at EOT showed that resistance to lorlatinib may be driven by bypass mechanisms(a) and other cancer-related pathways; inflammatory-response pathway alterations were observed at EOT in 6 of 17 (35%) patients(b) Resistance mechanism | Genes(c) | Patients with altered genes, n (%)(d) DNA damage repair | ATM, BRCA2, CHEK2, XRCC2 | 6/17 (35%) Inflammatory-response | AXL, IL1R1, LYN, MET, MYC, and ROS1 | 6/17 (35%) RTK/RAS/MAPK | FLT1/VEGFR1, KRAS, MAP2K1, MET, NTRK1 | 6/17 (35%) Oncogene amplification | MET, MYC, NOTCH2 | 5/17 (29%) Tumor suppressor | BRCA2, FAT1, LRP5, RB1 | 5/17 (29%) Angiogenesis | FLT1/VEGFR1 | 3/17 (18%) Epigenetic regulation | HDAC6 | 2/17 (12%) ALK, anaplastic lymphoma kinase; ctDNA, circulating tumor DNA; EOT, end of treatment; MAPK, mitogen-activated protein kinase; RAS, rat sarcoma; RTK, receptor tyrosine kinase. (a)No dominant resistance pathway identified. (b)Indicating this pathway as a top potential resistance mechanism in the Hallmark pathway analysis. (c)The gene list for inflammatory-response is the result of Hallmark pathway analysis including all altered genes detected in >=1 patient, whereas the other resistance mechanism pathway genes were detected in >=2 patients. (d)A patient could have more than 1 gene altered in each pathway. 1. Solomon BJ, et al. J Clin Oncol. 2024;42:3400-3409. [Slide 3] Baseline ctDNA genetic alterations in EP and LTR with lorlatinib [Oncoprint: EPs (n=22) and LTRs (n=30); per-gene alteration frequencies] Gene: EPs % | LTRs % ALK: 59% | 57% EML4: 59% | 53% TP53: 50% | 17% ATM: 23% | 10% ZFHX3: 18% | 3% CHEK2: 18% | 0% NOTCH3: 14% | 7% NTRK3: 14% | 0% ARID1A: 14% | 10% NF1: 14% | 0% KRAS: 14% | 0% MYC: 14% | 0% CIC: 9% | 3% SPEN: 9% | 7% APC: 9% | 3% MTOR: 9% | 3% PREX2: 9% | 7% KDM5C: 9% | 3% PCDH15: 9% | 3% ROS1: 9% | 3% TET1: 9% | 3% ARID2: 9% | 3% ATRX: 9% | 3% PDGFRA: 9% | 3% DICER1: 9% | 3% BRIP1: 9% | 3% IGR: 9% | 13% ROBO2: 5% | 10% CDC73: 5% | 7% BRCA1: 5% | 7% PBRM1: 5% | 7% DNMT3B: 0% | 10% SETD2: 0% | 10% - EPs had a higher median number of altered genes (10 vs 6) and a higher median blood-based tumor mutation burden (7.7 vs 5.1) compared with LTRs - TP53 mutations were observed in 50% of samples from EPs compared with 17% in LTRs - Inflammatory response was the predominant EP-associated pathway, observed in 32% of EPs vs 7% of LTRs - Altered genes included IRF1, MET, MYC, ROS1, and TAPBP Legend: Alterations: SNV, Indel, CNA focal + amp, Fusion; Treatment: Lorlatinib; PFS status: Early, Late EPs are patients who had PFS events within 12 months of the start of the study. LTRs are patients who were alive with no documented progression at 84 months. Reprinted from AT Shaw, et al. Ann Oncol. 2026;37(9):1242-1252. Copyright (c) 2026 The Authors. Published by Elsevier Ltd on behalf of European Society for Medical Oncology. EP, early progressor; LTR, long-term responder; PFS, progression-free survival. [Slide 4] 7-year PFS by baseline TP53 mutation status - Lorlatinib showed prolonged PFS vs crizotinib regardless of baseline TP53 mutation status - In the lorlatinib group, longer PFS was observed in the TP53 mutation-negative vs mutation-positive subgroup; in the crizotinib group, PFS was short regardless of baseline TP53 status Table columns: Lorlatinib TP53 mut (-) n=73 | Lorlatinib TP53 mut (+) n=43 | Crizotinib TP53 mut (-) n=65 | Crizotinib TP53 mut (+) n=44 PFS, median (95% CI), months: NR (NR-NR) | 51.6 (16.4-NR) | 10.9 (9.0-12.8) | 7.2 (5.6-9.3) 7-year PFS (95% CI), %: 65.0 (54.5-77.4) | 40.8 (27.3-61.1) | 2.3 (0.3-15.7) | 6.9 (1.9-24.5) HR (95% CI), mut (-) vs mut (+): Lorlatinib 0.48 (0.268-0.857) | Crizotinib 0.78 (0.513-1.186) Y axis: Probability of PFS; X axis: Months (0-96) No. at risk Lorlatinib TP53 mut (-): 73 64 59 56 53 50 47 47 45 43 43 42 41 40 25 10 2 Lorlatinib TP53 mut (+): 43 33 29 23 21 19 17 16 16 14 13 12 11 11 5 5 0 Crizotinib TP53 mut (-): 65 45 23 14 10 8 5 5 5 5 5 4 4 3 1 0 0 Crizotinib TP53 mut (+): 44 25 11 7 4 4 3 3 3 3 2 2 2 1 1 0 0 HR, hazard ratio; mut, mutation; NR, not reached; PFS, progression-free survival. Survival table Endpoint: PFS by baseline TP53 mutation status (CROWN ctDNA analysis) Arm | N | Median | Median CI Lorlatinib TP53 mut (-) | 73 | NR | NR-NR Lorlatinib TP53 mut (+) | 43 | 51.6 | 16.4-NR Crizotinib TP53 mut (-) | 65 | 10.9 | 9.0-12.8 Crizotinib TP53 mut (+) | 44 | 7.2 | 5.6-9.3 Landmark rates: 7 years: Lorlatinib TP53 mut (-) 65.0% (CI 54.5-77.4) | Lorlatinib TP53 mut (+) 40.8% (CI 27.3-61.1) | Crizotinib TP53 mut (-) 2.3% (CI 0.3-15.7) | Crizotinib TP53 mut (+) 6.9% (CI 1.9-24.5)

[Slide 1] Study design Phase II non-comparative, randomized, open-labelled study Phase III non-inferiority, randomized, open-labelled study • Stage IV NSCLC • Any PD-L1 • ECOG PS 0 or 1 • 18 – 75 years • W/o addictive mutation • Chemo and ICI-naive Phase II: N=220 (Phase III: N=1250) Chemotherapy x 4* + Pembrolizumab 200mg Q3W At 6 month Patients with PR/SD Randomization 1:1 Phase II: N=88 (Phase III: N=525) Arm A Control: Continuation: Pembrolizumab (2 years max) + pemetrexed for non-SCC Arm B De-escalation: Observation for SCC Or pemetrexed for non-SCC Phase II: Primary endpoint: 18-month OS from inclusion (Phase III: Primary endpoint: OS) Secondary endpoints • Safety • Quality of life (EQ-5D-5L, EORTC QLQ-C30/QLQ-LC13) • 6-month PFS from randomization Stratification factors: • PS (0 vs. 1) • Histology (squamous cell carcinoma vs non squamous) • PD-L1 (PD-L1 < 1% vs 49%≥PD-L1 ≥ 1% vs PD-L1>49%) • Sex • CR/PR vs ST at randomization (6 months) * platinum-based doublet for at least 4 cycles, either paclitaxel-carboplatin for patient with SCC, or 4 cycles of pemetrexed + platinum salt followed by pemetrexed maintenance until randomization ESMO 2026 [Slide 2] Overall Survival On the first 84 eligible randomized patients (42 per arm), ITT post-randomisation Median follow-up of 27.9 months (95%CI: [25.0; 31.3]) for both arms | Arm A - Continuation (n=42) | Arm B - De-escalation (n=42) 18-month OS from inclusion | 83.3% (95%CI [68.2-91.7]) | 88.1% (95%CI [73.7-94.9]) 24-month OS from inclusion | 78.1% (95%CI [62.1-88.0]) | 77.2% (95%CI [60.6-87.5]) ESMO 2026 [Slide 3] Adverse events after randomization | Arm A - Continuation (n=45): Any grade N (%) | Grade 3 N (%) | Grade 4 N (%) | Grade 5 N (%) | Arm B - De-escalation (n=41): Any grade N (%) | Grade 3 N (%) | Grade 4 N (%) | Grade 5 N (%) Any adverse event | 44 (97.8) | 17 (37.8) | 1 (2.2) | 1 (2.2) | 41 (100.0) | 14 (34.1) | 3 (7.3) | 1 (2.4) Treatment-related adverse event | 43 (95.6) | 14 (31.1) | 0 (0) | 0 (0) | 33 (80.5) | 9 (22.0) | 2 (4.9) | 0 (0) Adverse event related to pembrolizumab | 39 (86.7) | 7 (15.6) | 0 (0) | 0 (0) | 21 (51.2) | 3 (7.3) | 0 (0) | 0 (0) ESMO 2026 [Slide 4] Flow Chart Included (N=197) From May 2022 to March 2024 Not treated (N=5) - Investigator's reason (n=5) Treated (N=192) Not randomized (N=99): Progression (n=68); Death (n=14); Toxicity (n=8); Patient's decision (n=3); Intercurrent event (n=2); Other (n=4) Randomized (N=93, 47%) A - Continuation (N=47) | B - De-escalation (N=46) Not treated (N=2): Disease evolution (n=1); Toxicity (n=1) | Not treated (N=5): Disease evolution (n=5) A – Continuation - treated (N=45) | B – De-escalation - treated (N=41) Study withdrawn (N=39): Disease evolution (n=21, 53.8%); 2-year treatment (n=12, 30.8%); Toxicity (n=4, 10.2%); Second cancer (n=1, 2.6%); Death (n=1, 2.6%) Study withdrawn (N=30): Disease evolution (n=28, 93.3%); Death (n=1, 3.3%); 2-year treatment (n=1, 3.3%) Treatment ongoing (N=6) | Treatment ongoing (N=11) Subsequent treatment (N=22, 56.4%) | Subsequent treatment (N=28, 93.3%) randomized population; safety population

[Slide 1] Integrating longitudinal imaging with molecular profiling and ctDNA to study metastatic growth in TRACERx and PEACE Charles Swanton | Mariam Jamal-Ha[njani] CANCER RESEARCH UK | TRACERx Timeline: Baseline → Follow up → Recurrence → Radiotherapy → Progression → Chemotherapy → 2nd Progression → Immunotherapy → Death; ctDNA sampling Surgery → #Days from Surgery 151 patients with relapse of disease Longitudinal tissue sequencing and ctDNA TRACERx 2014-2025: >1300 recruited patients; >800 eligible patients with early-stage NSCLC PEACE 2014-2025: >420 pan-cancer patients recruited; >280 autopsies performed Bailey et al. 2021., Hessey et al. 2[...] [Slide 2] Metastatic volume does not always correlate with cancer-related death Multiple Metastases | Single Metastasis Tumour Volume Before Death (mm³) Lesion Type: Abdominal Lymph Node; Adrenal; Bone; Brain; Cardiac; Extrathoracic Soft Tissue; Intrathoracic Soft Tissue; Kidney; Liver; Lung; Mediastinal Lymph Node; Pleura; Regional Lymph Node; Spleen; Total Volume 88 patients with a documented "lung cancer-related" death Median cumulative tumour volume of lesions per patient on the last scan before death = 52,473 mm³ [...]ng Kin Liu [Slide 3] Affected organ site if associated with worse survival Hazard Ratio for Death (forest plot, point estimates printed): Mediastinal Lymph Node | 2.5 Pleura | 2.5 Lung | 3.7 Brain | 2.7 Extrathoracic Soft Tissue | 2.8 Abdominal Lymph Node | 3.7 Adrenal | 3.7 Liver | 3.8 Bone | 4.6 151 patients, 706 metastases [presenter footer: ...n Liu] [Slide 4] Summary and conclusions 1. Metastatic volume of disease does not always correlate with cancer-related death 2. Extrathoracic metastases are associated with treatment resistance, higher levels of chromosomal instability and worse outcomes 3. Fast growing metastases are enriched in proliferation and downregulation in immune evasion pathways 4. The anatomical location of metastases influences ctDNA detection Metastatic NSCLC is a dynamic, heterogeneous disease in which lesion growth and treatment response vary by site and tumour biology, influencing patient outcomes Liu

LBA74 - First-line (1L) trastuzumab deruxtecan (T-DXd) in patients (pts) with HER2-mutant (HER2m) NSCLC: Patient-reported outcomes (PROs) from DESTINY-Lung04 (DL-04) Speakers Silvia Novello (Orbassano, Italy) Lecture Time 09:00 - 09:05 Abstract Background In DL-04, 1L T-DXd significantly improved PFS vs pembrolizumab (P) + chemotherapy (C) in pts with advanced/metastatic HER2m NSCLC. Safety was generally consistent with the known profiles of each drug. Here, we report PROs from DL-04, focusing on pulmonary symptoms and pt-reported tolerability while on treatment (Tx). Methods DL-04 (NCT05048797) is a global, open-label, phase 3 study of pts with Tx-naïve HER2m (exon 19/20) NSCLC randomized 1:1 to T-DXd 5.4 mg/kg or P 200 mg + C (cisplatin 75 mg/m2 or carboplatin AUC 5 + pemetrexed 500 mg/m2), all IV Q3W. Secondary PRO endpoints were time to deterioration (TTD) in pulmonary symptoms while on Tx (cough, dyspnea, chest pain) via NSCLC-SAQ and pt-reported tolerability on Tx (including physical function [PF] via EORTC QLQ-C30 PF subscale, overall side-effect bother via Patient Global Impression of Treatment Tolerability [PGI-TT], and symptomatic AEs via EORTC and PRO-CTCAE item libraries). High side-effect bother (score of 4 or 5 via PGI-TT, first 12 weeks only) was exploratory. Results As of June 9, 2026, 454 pts were randomized. Median (range) Tx duration was 12.3 (0.7–44.8) months [mo] with T-DXd vs 7.1 (0.7–49.1) mo with P+C. T-DXd vs P+C delayed median (95% CI) TTD in cough (26.2 [17.1, NC] vs 11.0 [7.4, NC] mo; HR [95% CI] 0.66 [0.44, 0.99]), with favorable trends in dyspnea (10.1 [6.5, NC] vs 5.1 [3.2, 11.0] mo; HR 0.77 [0.54, 1.09]) and chest pain (NC [11.5, NC] vs 14.3 [8.3, NC] mo; HR 0.80 [0.54, 1.18]). The proportion of pts with maintained/improved PF was similar across arms until Week 18 and generally higher with T-DXd from Week 20, including at end of Tx (EOT, 66.7% T-DXd vs 57.7% P+C). Overall side-effect bother was similar across arms for most timepoints. The proportion of pt-weeks with high side-effect bother was 13.4% in each arm. Symptomatic AEs will be reported at the congress. Conclusions TTD in pt-reported pulmonary symptoms favored T-DXd vs P+C. Tx tolerability (PF and overall/high side-effect bother) was generally comparable between arms, with PF generally higher with T-DXd from Week 20 and at EOT. These results, alongside efficacy and safety data from the primary analysis, support T-DXd as a new 1L Tx option for HER2m NSCLC.

Long-term real-world outcomes with lorlatinib in ALK-positive NSCLC: a Kuwait Cancer Control Center experience Abdulaziz AlJassim1, Fahad G. Alenizi1, Talal O. Alsalih1, Ahmed M. Elsheashaey2, Mariam Al Mutairi2, Amir A. Ahmed3, Muath A. Alnassar1 1Medical Oncology Department, Kuwait Cancer Control Center, Ministry of Health, Kuwait; 2Department of Pharmacy, Kuwait Cancer Control Center, Ministry of Health, Kuwait; 3Molecular Genetics Laboratory, Kuwait Cancer Control Center, Ministry of Health, Kuwait. Ministry of Health, State of Kuwait | MADRID 2026 ESMO congress | 2811eP INTRODUCTION • Lorlatinib, a third-generation ALK/ROS1 tyrosine kinase inhibitor (TKI), has demonstrated substantial systemic and intracranial efficacy and durable clinical benefit in ALK-positive NSCLC, including in the phase III CROWN study.1 • Real-world evidence describing long-term lorlatinib outcomes in patients from the Arabian Gulf Co-operation Council (GCC) countries remains limited. OBJECTIVE To evaluate real-world systemic efficacy, treatment durability, intracranial activity and survival status among patients with advanced ALK-positive NSCLC treated with lorlatinib at Kuwait Cancer Control Center. METHODS • Retrospective single-centre review of consecutive patients with advanced ALK-positive NSCLC treated with lorlatinib at Kuwait Cancer Control Center between 2016 and 2026. • ALK positivity was confirmed by next-generation sequencing (NGS), fluorescence in situ hybridisation (FISH), or immunohistochemistry (IHC) according to diagnostic era. • Outcomes included objective response rate (ORR) and disease control rate (DCR) per RECIST 1.1, treatment duration, intracranial activity, survival status and adverse events. • Ongoing patients were censored at data cut-off (07 May 2026). RESULTS 1 Patient characteristics (N = 26) Characteristic | n (%) Median age, years (range) | 47 (18–79) Female | 17 (65%) Never-smokers | 21 (81%) Brain metastases at diagnosis | 11 (42%) Treatment line with lorlatinib First-line | 10 (38%) Second-line | 12 (46%) Third-line | 4 (15%) ALK fusion variants (characterised in 14/26, 54%) EML4–ALK V1 | 7 (50%) EML4–ALK V3 | 6 (43%) EML4–ALK V2 | 1 (7%) 2 Systemic efficacy and treatment durability Lorlatinib demonstrated high response rates and durable disease control in routine clinical practice. 83% Objective response rate (ORR) (19/23) response-evaluable patients 96% Disease control rate (DCR) (22/23) response-evaluable patients 62% Patients remaining on treatment (16/26) at data cut-off Exceptional long-term treatment benefit >85 Longest ongoing exposure, months 3 Patients with ongoing treatment >60 months 17/26 (65%) Alive at data cut-off 3 Intracranial activity Intracranial disease control was documented in evaluable patients with baseline brain metastases. 11/26 (42%) had brain metastases at diagnosis Intracranial disease control was achieved in all evaluable patients with baseline brain metastases. Dose reduction occurred in 14 patients, most commonly due to dyslipidaemia, neuropsychiatric toxicity, oedema, or weight gain.2 CONCLUSIONS 01 Lorlatinib demonstrated robust systemic activity in this real-world Kuwaiti cohort, consistent with expected efficacy. 02 Durable disease control was observed across treatment lines, with a substantial proportion of patients remaining on treatment and exceptional long-term benefit in selected patients. 03 Meaningful intracranial disease control was documented in evaluable patients with baseline brain metastases. 04 These findings contribute regional real-world evidence supporting the effectiveness of lorlatinib in routine clinical practice.

[Slide 1] Two Ways to Redirect T cells to DLL3 BiTE/T cell Engager [Diagram: Neutrophil, Macrophage, Lymphocyte, IL-6, IL-10, TNF-α, IFN-γ, TCE, TCR, TAA, CD3, T cell proliferation, T cell, Granzymes and perforins, Tumor cell, Tumor cell death] | BiTE/TCE Recognition | DLL3 × CD3 Product | Off-the-shelf T-cell source | Endogenous Activation | Drug-dependent Persistence | Requires repeated dosing Logistics | Relatively simple Current SCLC status | Established therapy Recognition | DLL3 × CD3 CAR-T Cell [Diagram: CAR-T cell, SCLC cell, Chimeric Antigen Receptor (CAR): Antigen-recognition domain, Hinge domain, Transmembrane domain, Co-stimulatory domain (CD28 or 4-1BB), Activation domain (CD3ζ); Target antigen; Tumor cell lysis] | CAR-T Recognition | CAR recognizes DLL3 Product | Personalized cellular therapy T-cell source | Engineered/autologous Activation | Living/expanding therapy Persistence | Potential long persistence Logistics | Apheresis → manufacturing → lymphodepletion Current SCLC status | Investigational Recognition | CAR recognizes DLL3 [Slide 2] Other TCEs in Clinical Development in SCLC Agent | Tarlatamab | Obrixtamig (BI 764532) | MK-6070 (HPN328) | Alvetamig (ZG009) Structure | Half-life extended BiTE | Bispecific antibody | Trispecific Ab | Biparatopic Ab [Structure diagrams: Tarlatamab: Anti-CD3, Anti-DLL3, Fc domain (to extend half-life); Obrixtamig: Anti-DLL3, Anti-CD3, Engineered Fc domain; MK-6070: Anti-CD3ε, Anti-albumin (to extend half-life), Anti-DLL3; Alvetamig: Anti-DLL3, Anti-DLL3, Anti-CD3, Effector-less Fc (HLE)] ORR | 40% | 25% | 31% | 60% DOR | 9.7m | 8.5m | NA | NA OS | 15.2m | NA | NA | NA CRS (any) | 51% | 59% | 59% | 42% CRS ≥G3 | <1% | 2% | 3% | 0% ICANS (any) | 8% | NR | 7% | 0% ICANS ≥G3 | 0% | NR | 0% | 0% [Slide 3] Clinical Development of CAR-T cell in SCLC Target | CAR Construct (Co-stimulation) | Phase | Study (Sponsor) | Key Findings DLL3 | scFv – 4-1BB / CD3ζ | Phase I | NCT03090659 (Bellicum) | ORR ~23% Manageable CRS/NT DLL3 | scFv – CD28 / CD3ζ | Phase I | NCT04149947 (Adaptimmune) | ORR ~40% Durable responses B7-H3 | scFv – 4-1BB / CD3ζ | Phase I | NCT05706238 (Iovance) | Early data : responses observed CDH6 | scFv – 4-1BB / CD3ζ | Phase I | NCT05380999 (Carisma) | Early data : responses observed NKG2D ligands (MICA/B) | scFv – 4-1BB / CD3ζ | Phase I | NCT-4472624 (Bellicum) | Disease control in some patients [Slide 4] TCE vs CAR-T cell Therapy in SCLC: Competition vs Complementary? • BiTE and CAR-T should not necessarily be viewed as competitors but rather complementary T cell re-directing platforms • BiTE is ready for practice vs CAR-T therapy is ready for clinical testing • Rather how can we use different T cell re-directing platforms at the right time for the right patient and in the right sequence | BiTE/TCE | CAR-T Availability | Immediate | Manufacturing required Rapidly progressive SCLC | Advantage | Major challenge Repeat dosing | Required | Usually one infusion T cell engineering | No | Yes Expansion/persistence | Limited | Potential advantage TME resistance | Limited | Potential advantage Scalability | High | Low Current Evidence | Phase III/Approved | Phase I

[Slide 1] Session banner: 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic | CHAIRS: PASI A. JÄNNE, XIUNING LE, EGBERT SMIT Speaker card: Pasi A. Jänne — Invited Discussant LBA74 and 2636RO ALMERIA AUDITORIUM - HALL 9 First line metastatic HER2 mutant NSCLC Two oral TKIse now have US first-line indications; T-DXd has randomized first-line evidence. ZONGERTINIB: Oral HER2 TKI | First-line activity | Direct PRO data SEVABERTINIB: Oral HER2 TKI | First-line activity | Toxicity over time T-DXd: HER2-directed ADC | Randomized PFS benefit | PRO abstract to discuss The choice now requires efficacy, patient experience, safety, and a sequencing plan. FDA: zongertinib, 26 February 2026; sevabertinib, 9 September 2026. DESTINY-Lung04: WCLC 2026, PL03.08. [Slide 2] Session banner: 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic | CHAIRS: PASI A. JÄNNE, XIUNING LE, EGBERT SMIT Speaker card: Pasi A. Jänne — Invited Discussant LBA74 and 2636RO ALMERIA AUDITORIUM - HALL 9 Zongertinib: what patients actually reported Beamion LUNG-1 • First-line PROs (n=71) • FDA 1L approval: 26 February 2026 Figure 7. Baseline summary and responder analysis of EORTC QLQ-C30 physical functioning (n = 71) [Baseline box plot; Post-baseline stacked bars by visit C1D8, C1D15, C2D1 (Day 22), C3D1 (Day 43), C5D1 (Day 85), C7D1 (Day 127), C9D1 (Day 169)] Responder %: 63.4% | 63.4% | 62.0% | 64.8% | 70.4% | 60.6% | 53.5% Deteriorated %: 14.1% | 18.3% | 22.5% | 19.7% | 11.3% | 11.3% | 7.0% Stable %: 4.2% | 4.2% | 1.4% | – | 2.8% | 1.4% | 2.8% No PRO Data %: 18.3% | 14.1% | 14.1% | 15.5% | 15.5% | 26.8% | 36.6% Responder status: Responder | Stable | Deteriorated | No PRO Data Figure 3. Extent troubled with side effects of treatment (n = 71) [Alluvial stacked bars Baseline to C9D1; Response: Not at All | A Little | Quite a Bit | Very Much | No PRO Data — small values illegible] Response includes improvement OR maintained favorable baseline; grey = missing PRO data. Maintained function and low side-effect impact; single-arm evidence. Sabari et al. ASCO 2026, 8616: supp[obscured] [Phone overlay: esmocongress.esmo.org – To exit full screen, drag from the top and touch the back button] [Slide 3] Session banner: 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic | CHAIRS: PASI A. JÄNNE, XIUNING LE, EGBERT SMIT Speaker card: Pasi A. Jänne — Invited Discussant LBA74 and 2636RO ALMERIA AUDITORIUM - HALL 9 Mechanisms of Acquired Resistance to Kinase Inhibitors in Lung Cancer Mutations in the drug target — Impact drug binding Bypass Signaling — Oncogene Amplification; Oncogene Rearrangement Mutations in Downstream Effectors — PI3K → AKT; RAS → RAF → MEK → ERK; [Cell Cycle diagram] State Transformation — Small Cell Lung cancer; Squamous Cell lung cancer [histology image] Cohen at al. Nat Rev Drug Discov 2017 [Slide 4] Session banner: 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic | CHAIRS: PASI A. JÄNNE, XIUNING LE, EGBERT SMIT Speaker card: Pasi A. Jänne — Invited Discussant LBA74 and 2636RO ALMERIA AUDITORIUM - HALL 9 CROWN: baseline biomarkers and early ctDNA clearance EML4::ALK variant [KM plot, Probability of PFS vs Months 0–96] Variant 3 vs 1/2: HR 1.09 (95% CI 0.463–2.579) Teal: lorlatinib. Pink: crizotinib. Dashed: variants 1/2.;Solid: variant 3 TP53 status [KM plot, Probability of PFS vs Months 0–96] Mutation− vs mutation+: HR 0.48 (95% CI 0.268–0.857) Teal: lorlatinib. Pink: crizotinib. Dashed: TP53 mutation−. Baseline shedding (methylation-based TF) [KM plot with inset table: ND n=18 / D n=100; PFS, months, median (95% CI); HR (95% CI), ND vs D — inset values small] Undetected vs detected: HR 0.17 (95% CI 0.04–0.69) Teal: undetected. Pink: detected. Week 4 ctDNA clearance (TF) [KM plot, p = 0.013; inset legend small/illegible] Cleared vs persistent: HR 0.48 (95% CI 0.26–0.86) Teal: cleared. Pink: persistent. Lorlatinib-arm PFS comparisons. Solomon et al. ESMO 2026, 2636RO. TF, tumor fraction. Survival table Endpoint: PFS (lorlatinib arm biomarker comparisons) HR: 0.48 HR CI: 0.268–0.857

2667P Histologic transformation as an emerging resistance mechanism in NSCLC: insights from a multicenter clinicopathological and molecular study Sara Torresan, Laura Masfarré, Clara Salvá, ... BEST POSTER NSCLC, METASTATIC Table 1. Patients' characteristics. Characteristics N=137 % Race: Asian 4 2.9; Black 3 2.2; White 93 67.9; Not reported 37 27 CNS metastasis: At diagnosis 43 31.2; At HT 54 39.4 BoR after HT (n=122): CR 5 4.1; PR 38 31.1; SD 32 26.2; PD 31 25.4; NA 16 13.1 Figure 1. Genomic landscape of the cohort. Oncoplot showing recurrent and clinically relevant somatic mutations for each patient. Columns represent individual patients and rows represent genes. Figure 2. Transformation histologies frequencies A: Median OS: 36.6 months B: Median OS: 8.8 months Figure 3. Overall survival of patients with histologic transformation. Kaplan-Meier curves of overall survival (A) from initial diagnosis and (B) from the date of histologic transformation. Numbers at risk are shown below each plot. Figure 4. Clinical course of patients with histologic transformation (HT) and rare or no actionable genomic alterations. Swimmer plot showing the timeline of each patient from initial diagnosis. Each bar represents one patient. Markers indicate key timepoints Among pts with available matched data (before and after HT, 98 patients), all retained their original driver mutations except for 4 (2 EGFR, 2 RET) (Figure 1). After HT, 11.7% of patients received TT, while 75.9% received chemotherapy, alone (39.4%), combined with TT (17.5%), or with IT (19%). Median time from diagnosis to HT was 23.7 months. Median overall survival (mOS) from initial diagnosis was 31.7 months while mOS after HT was 9.5 months (Figure 3). Conclusions To the best of our knowledge, our study represents the largest reported cohort of HT, extending beyond EGFR-mutated NSCLC and SCLC transformations. HT remains an underrecognized clinical entity that may occur more frequently than previously assumed and demands timely recognition and tailored therapeutic strategies. Ongoing molecular analyses aim to identify potential biomarkers of HT. Disclosure statement: S. Torresan has no DOI. Corresponding author: pedrorocha@vhio.net

[Slide 1] FLAURA2 study design¹ Scan QR code to view slides and supplementary material [QR code] Patients with EGFR-mutated advanced NSCLC (N=557) Key inclusion criteria: • Aged ≥18 years (Japan: ≥20 years) • Pathologically confirmed non-squamous NSCLC • Ex19del / L858R (local / central test) • WHO PS 0 / 1 • No prior systemic therapy for advanced NSCLC • Stable CNS metastases were allowed • Brain scan at baseline (MRI / CT; mandatory) R 1:1* Osimertinib 80 mg (QD) + pemetrexed 500 mg/m² + carboplatin AUC5 or cisplatin 75 mg/m² (Q3W for 4 cycles of plat-based Tx) → Maintenance osimertinib 80 mg (QD) + pemetrexed 500 mg/m² (Q3W)† Osi 80 mg (QD)† • Primary endpoint: PFS (by investigator) per RECIST 1.1 • Key secondary endpoint: OS • Data cut-off: 03 April 2023 (PFS) and 12 June 2025 (OS) Exploratory endpoint: Correlation of plasma EGFRm clearance at Weeks 1, 2, 3, and 6 with PFS and OS to understand if monitoring ctDNA dynamics at earlier timepoints can be used to predict outcomes Plasma sample collection: Baseline, Week 1, Week 2, Week 3, Week 6 Baseline and Weeks 1, 2, 3, and 6 plasma samples‡ from both treatment arms analysed by ddPCR (Biodesix) FLAURA2 (NCT04035486) is a global, open-label, Phase III study. We acknowledge the potential presence of immortal time bias of analysing PFS and OS (measured from randomisation) by post-baseline subgroup as patients would need to remain progression-free or alive until Week 1, 2, 3 and 6 to be assessed in that subgroup. 1. Planchard D, et al. N Engl J Med 2023;389:1935–1948. *Patients were stratified by race (Asian Chinese / Asian non-Chinese / non-Asian), EGFRm test (local / central) and WHO PS (0 / 1). †Treatment continued until disease progression or another discontinuation criterion was met; treatment beyond progression was permitted if the investigator judged the patient had continued clinical benefit. ‡Plasma samples were collected at baseline (C1D1), weekly for Cycle 1 (C1DX), every 3 weeks (CXD1) until Cycle 7, Day 1 (C7D1), every 6 weeks until Cycle 17, every 12 weeks until disease progression, treatment discontinuation or primary PFS analysis. Post-primary PFS, further samples were collected at progression/discontinuation only. AUC, area under the curve; CNS, central nervous system; CT, computed tomography; ctDNA, circulating tumour DNA; ddPCR, droplet digital polymerase chain reaction; EGFR, epidermal growth factor receptor; EGFRm, EGFR mutation; Ex19del, exon 19 deletion; MRI, magnetic resonance imagery; NSCLC, non-small cell lung cancer; OS, overall survival; PFS, progression-free survival; plat, platinum; Q3W, every 3 weeks; QD, once daily; R, randomised; RECIST, Response Evaluation Criteria in Solid Tumors; Tx, treatment; WHO PS, World Health Organization performance status. David Planchard, MD, PhD [Slide 2] In patients with baseline-detected plasma EGFRm, rate of EGFRm clearance was similar between treatment arms at Weeks 1, 2, 3, and 6 • 421 / 557 (76%) patients enrolled in FLAURA2 had an evaluable baseline plasma EGFRm ddPCR result (ctDNA analysis set)1 • Of those patients included in the ctDNA analysis set, plasma EGFRm at baseline was detected in 147 / 212 (69%) patients in the osi + plat–pem arm and 161 / 209 (77%) patients in the osi mono arm1 [Stacked bar chart, Proportion of patients (%); top = Non-detected plasma EGFRm (clearance), bottom = Detected plasma EGFRm] Week 1: Osi + plat–pem n=104: Non-detected 38% (95% CI 29, 47); Detected 62% (95% CI 53, 71) | Osi mono n=131: Non-detected 42% (95% CI 34, 51); Detected 58% (95% CI 49, 66) Week 2: Osi + plat–pem n=113: Non-detected 64% (95% CI 55, 72); Detected 36% (95% CI 28, 45) | Osi mono n=133: Non-detected 64% (95% CI 55, 72); Detected 36% (95% CI 28, 45) Week 3: Osi + plat–pem n=109: Non-detected 81% (95% CI 72, 88); Detected 19% (95% CI 12, 28) | Osi mono n=140: Non-detected 71% (95% CI 63, 79); Detected 29% (95% CI 21, 37) Week 6: Osi + plat–pem n=105: Non-detected 88% (95% CI 80, 93); Detected 12% (95% CI 7, 20) | Osi mono n=137: Non-detected 89% (95% CI 83, 94); Detected 11% (95% CI 6, 17) Legend: Osi + plat–pem: Detected plasma EGFRm / Non-detected plasma EGFRm; Osi mono: Detected plasma EGFRm / Non-detected plasma EGFRm Patients were eligible for clearance analysis if they had baseline-detected plasma EGFRm and a valid plasma EGFRm result for the analysis week shown (Week 1, 2, 3, or 6). Clearance was defined as undetected plasma EGFRm (Ex19del, L858R) at Week 1, 2, 3, or 6, where it was detected at baseline. 1. Jänne PA, et al. Cancer Res 2024;84(Suppl. 7): Abstract CT107. CI, confidence interval; ctDNA, circulating tumour DNA; ddPCR, droplet digital polymerase chain reaction; EGFR, epidermal growth factor receptor; EGFRm, EGFR mutation; Ex19del, exon 19 deletion; mono, monotherapy; osi, osimertinib; pem, pemetrexed; plat, platinum. David Planchard, MD, PhD [Slide 3] PFS improved with osi + plat–pem versus osi mono irrespective of Week 1 and 2 plasma EGFRm clearance status [Four KM panels: Probability of PFS vs Time from randomisation (months)] Plasma EGFRm clearance at Week 1 — No. at risk (0,6,12,18,24,30,36): Osi + plat–pem 39, 32, 26, 21, 13, 4, 0; Osi mono 55, 46, 37, 22, 14, 7, 0 No Plasma EGFRm clearance at Week 1 — No. at risk: Osi + plat–pem 65, 55, 49, 38, 15, 8, 0; Osi mono 76, 57, 42, 22, 11, 4, 0 Week 1 table — mPFS (95% CI), months: Clearance | No clearance Osi + plat–pem: NR (18.3, NR) | 24.8 (16.8, 30.6) Osi mono: 16.6 (13.6, 21.0) | 13.6 (11.1, 16.4) HR (95% CI): 0.58 (0.33, 1.02) | 0.53 (0.35, 0.80) Plasma EGFRm clearance at Week 2 — No. at risk: Osi + plat–pem 72, 63, 55, 42, 21, 6, 0; Osi mono 85, 74, 56, 31, 19, 9, 0 No Plasma EGFRm clearance at Week 2 — No. at risk: Osi + plat–pem 41, 35, 27, 22, 10, 6, 0; Osi mono 48, 30, 26, 14, 7, 4, 0 Week 2 table — mPFS (95% CI), months: Clearance | No clearance Osi + plat–pem: 27.6 (19.7, NR) | 22.1 (11.8, 30.6) Osi mono: 13.9 (13.6, 19.4) | 13.8 (5.9, 16.8) HR (95% CI): 0.53 (0.35, 0.81) | 0.57 (0.34, 0.96) Data cut-off: 03 April 2023 (PFS). HRs were calculated using an unstratified Cox proportional hazards model. Clearance was defined as undetected plasma EGFRm (Ex19del, L858R) at Week 1, 2, 3, or 6, where it was detected at baseline. CI, confidence interval; EGFR, epidermal growth factor receptor; EGFRm, EGFR mutation; Ex19del, Exon 19 deletion; HR, hazard ratio; m, median; mono, monotherapy; NR, not reached; osi, osimertinib; pem, pemetrexed; PFS, progression-free survival; plat, platinum. David Planchard, MD, PhD Survival table Endpoint: PFS by Week 1 and Week 2 plasma EGFRm clearance status (FLAURA2, DCO 03 April 2023) Arm | N | Median | Median CI Wk1 clearance: Osi + plat–pem | 39 | NR | 18.3, NR Wk1 clearance: Osi mono | 55 | 16.6 | 13.6, 21.0 Wk1 no clearance: Osi + plat–pem | 65 | 24.8 | 16.8, 30.6 Wk1 no clearance: Osi mono | 76 | 13.6 | 11.1, 16.4 Wk2 clearance: Osi + plat–pem | 72 | 27.6 | 19.7, NR Wk2 clearance: Osi mono | 85 | 13.9 | 13.6, 19.4 Wk2 no clearance: Osi + plat–pem | 41 | 22.1 | 11.8, 30.6 Wk2 no clearance: Osi mono | 48 | 13.8 | 5.9, 16.8 [Slide 4] OS improved with osi + plat–pem versus osi mono irrespective of plasma EGFRm clearance status [Forest plot] n (Osi + plat–pem / Osi mono) | OS HR (95% CI) Week 1 cleared: 39 / 55 | 0.79 (0.45, 1.37) Week 1 not cleared: 65 / 76 | 0.69 (0.47, 1.02) Week 2 cleared: 72 / 85 | 0.69 (0.46, 1.03) Week 2 not cleared: 41 / 48 | 0.78 (0.48, 1.26) Week 3 cleared: 88 / 100 | 0.84 (0.58, 1.21) Week 3 not cleared: 21 / 40 | 0.65 (0.36, 1.20) Week 6 cleared: 92 / 122 | 0.73 (0.52, 1.04) Week 6 not cleared: 13 / 15 | 0.50 (0.22, 1.13) Axis: 0.125, 0.25, 0.50, 1, 2 — Favours osi + plat–pem / Favours osi mono Data cut-off: 12 June 2025 (OS). HRs were calculated using an unstratified Cox proportional hazards model. Clearance was defined as undetected plasma EGFRm (Ex19del, L858R) at Week 1, 2, 3, or 6, where it was detected at baseline. CI, confidence interval; EGFR, epidermal growth factor receptor; EGFRm, EGFR mutation; Ex19del, exon 19 deletion; HR, hazard ratio; mono, monotherapy; OS, overall survival; osi, osimertinib; pem, pemetrexed; plat, platinum. David Planchard, MD, PhD

[Slide 1] Common aspects of SCLC Associated with heavy tobacco exposure Centrally located primary tumor High-grade neuroendocrine features Strong metastatic predilection 2/3 of cases metastatic at diagnosis Exceptionally proliferative Ki67 typically 80-100% Necrotic cores Lethal [Figure: CD56 and ASCL1 IHC panels (50 µm); body diagram: Brain, Lymph nodes, Primary tumour, Liver, Bone, Lung, Adrenal glands, Blood vessel, RBC, Circulating tumour cells] Rudin et al., Nat Rev Dis Primers 2021 [Slide 2] What we call subsets might be better described as admixed cell states NE-high, 80% SCLC ASCL1 "Classic" • DLL3 • SYP • CHGA NEUROD1 "Variant" • MYC POU2F3 • Tuft cell origin Triple negative • YAP • Inflamed NE-low, 20% SCLC HighNE → LowNE • NOTCH • REST [Slide 3] Summary: paths to SCLC in the never-smoker RB1+/TP53+ ("atypical") SCLC Progenitors related to carcinoids Chromothripsis → TSG deletions | Rearrangements Oncogene amplifications: MDM2; CCND1, CDK4 • p53/pRb functional co-inactivation via MDM2, CDK4, and CCND1 amplifications • Molecular alterations of carcinoid tumors • ASCL1-high Two pathogenetic pathways for nsSCLC p53 inactivation pRb inactivation Cell-cycle dysregulation Small cell phenotype RB1− /TP53− SCLC Progenitors related to NSCLC APOBEC-mediated mutagenesis TP53 mutation RB1 mutation • TP53/RB1 genomic co-inactivation • Molecular alterations of NSCLC, e.g. EGFR driver mutations • APOBEC mutational signature • Enriched in YAP1-high Rekhtman et al., Cancer Discov 2025 [Slide 4] Summary: SCLC state diversity Distinct cell states of SCLC tumors can be defined by key developmental transcription factors Neuroendocrine high: ASCL1, NEUROD1 Neuroendocrine low: POU2F3, YAP1/Inflamed These states may have differential vulnerabilities – but not yet clearly clinically demonstrated Constraining plasticity might be key to get such targeted therapies to work There is a cohort of SCLC patients, small but real, who have durable benefit from ICB Absence of antigen presentation is implicated as a primary contributor to ICB failure in SCLC Epigenetic targeting (LSD1, EZH2) might reverse this MHC-independent T cell engagement can bypass this SCLC can occur in never-smokers Lineage plasticity with histologic transformation from LUAD Chromothripsis with extrachromosomal oncogene amplification from carcinoid

[Slide 1] MADRID 2026 ESMO congress 3182P REAL-WORLD OUTCOMES ACCORDING TO CHEMOTHERAPY SEQUENCING IN ADVANCED PANCREATIC DUCTAL ADENOCARCINOMA Key Survival Outcomes PFS-1L from start of 1L to PD or death 7.7 vs 6.7 months (5FU-based vs GEM-based) HR 1.13 (95% CI 0.91 – 1.4) p=0.28 PFS-2L from start of 2L to PD or death 4.0 vs 4.1 months (5FU-based vs GEM-based) HR 0.99 (95% CI 0.79 – 1.23) p=0.91 PFS-2 from start of 1L to PD on 2L or death 13.3 vs 12.2 months (5FU→GEM vs GEM→5FU) HR 0.98 (95% CI 0.79 – 1.22) p=0.87 OS from start of 1L to death 17.4 vs 15.3 months (5FU→GEM vs GEM→5FU) HR 0.95 (95% CI 0.76 – 1.18) p=0.63 CONCLUSIONS No significant differences in PFS or OS were observed between 5FU→GEM and GEM→5FU sequencing among patients who received at least two lines of chemotherapy. [Slide 2] MADRID 2026 ESMO congress 3182P REAL-WORLD OUTCOMES ACCORDING TO CHEMOTHERAPY SEQUENCING IN ADVANCED PANCREATIC DUCTAL ADENOCARCINOMA Kreina Sharela Vega-Cano1, Natalia Soledad Tissera1, Florian Castet2,3, Victor Navarro4, Jose Maria Ucha1, Gloria Castillo5, Marcos Daniel Bortz6, Aina de Torner1, Eduardo Terán1, Daniel Acosta1, Jorge Hernando1, Sergio Perez-Fernandez1, Alejandro Garcia-Alvarez1, Alexandre Sierra1, Sandra Martinez-Badal5, Susana Aguilar7, Elizabeth Pando8, Tian Tian1, Teresa Macarulla2,3, Jaume Capdevila1 Vall d'Hebron Instituto de Oncología BACKGROUND & OBJECTIVE 5FU- and gemcitabine-based regimens are the main chemotherapy backbones for advanced PDAC. Both are frequently used sequentially, but the impact of treatment order on outcomes remains unclear. We compared clinical outcomes between 5FU→GEM and GEM→5FU sequences in pts with advanced PDAC receiving ≥2 systemic treatment lines. METHODS Retrospective single-center study conducted at Vall d'Hebron University Hospital. Chemotherapy backbones were classified as: 5FU-based: FOLFIRINOX, NALIRIFOX, nal-IRI/5FU, FOLFOX, FOLFIRI GEM-based: gemcitabine/nab-paclitaxel, gemcitabine/cisplatin, GEMOX, gemcitabine monotherapy. Endpoints: PFS1L, PFS2L, PFS2, and OS. PFS2 was defined from 1L initiation to progression during 2L or death. PATIENT FLOW Figure 1. Patient selection and chemotherapy sequencing groups 819 advanced PDAC records screened (2012-2025) 446 excluded for ≥ 1 of the following: < 2 systemic treatment lines in the advanced setting. No exposure to both 5FU- and GEM-based ChT backbones. Germline BRCA1/2 or PALB2 pathogenic variant. Insufficient clinical or treatment data 373 patients met eligibility criteria 7 excluded: no 1L-2L cross-backbone sequence 366 patients included in the sequencing analysis 5FU-Based ChT → GEM-based ChT n = 156 (43%) 1L: 5FU-based ChT 2L: GEM-based ChT GEM-Based ChT → 5FU-Based ChT n = 210 (57%) 1L: GEM-based ChT 2L: 5FU-based ChT RESULTS Figure 2. Progression-free survival 2 by chemotherapy sequence 5FU-based ChT → GEM-based ChT (n=152) GEM-based ChT → 5FU-based ChT (n=198) Median PFS2 (months) 13.3 vs 12.2 HR 0.98 (95% CI 0.79–1.22) p = 0.863 No. at risk 5FU-based ChT → GEM-based ChT 152 138 83 39 19 GEM-based ChT → 5FU-based ChT 198 173 99 52 25 Figure 3. Overall survival by chemotherapy sequence 5FU-based ChT → GEM-based ChT (n=154) GEM-based ChT → 5FU-based ChT (n=209) Median OS (months) 17.45 vs 15.31 HR 0.95 (95% CI 0.76–1.18) p = 0.630 No. at risk 5FU-based ChT → GEM-based ChT 154 149 112 69 42 21 10 6 GEM-based ChT → 5FU-based ChT 209 199 137 82 49 31 14 7 Key Survival Outcomes PFS-1L from start of 1L to PD or death 7.7 vs 6.7 months (5FU-based vs GEM-based) HR 1.13 (95% CI 0.91 – 1.4) p=0.28 PFS-2L from start of 2L to PD or death 4.0 vs 4.1 months (5FU-based vs GEM-based) HR 0.99 (95% CI 0.79 – 1.23) p=0.91 PFS-2 from start of 1L to PD on 2L or death 13.3 vs 12.2 months (5FU→GEM vs GEM→5FU) HR 0.98 (95% CI 0.79 – 1.22) p=0.87 OS from start of 1L to death 17.4 vs 15.3 months (5FU→GEM vs GEM→5FU) HR 0.95 (95% CI 0.76 – 1.18) p=0.63 Figure 4. Genomic alterations by treatment sequence KRAS 81.7% vs 89.0% p=0.086; TP53 66.1% vs 68.9% p=0.628; SMAD4 17.9% vs 14.7% p=0.472; CDKN2A 16.1% vs 16.7% p=0.905 Figure 5. KRAS mutation subtype distribution (5FU→GEM vs GEM→5FU) Global p = 0.193 G12D 52.4% vs 40.1%; G12V 27.2% vs 32.1%; G12R 10.7% vs 11.7%; Q61 7.8% vs 12.4%; G12C 0% vs 2.9%; Other 1.9% vs 0.7% Table 1. Baseline clinical characteristics according to chemotherapy sequence Baseline characteristic | 5FU→GEM | GEM→5FU | p-value Age, median (IQR), years | 59 (51–65) | 67 (60–73) | <0.001 Female, n (%) | 68 (42.0%) | 100 (47.2%) | 0.32 Charlson Comorbidity Index, median (IQR) | 7 (6–8) | 9 (7–10) | <0.001 Pancreatic head primary, n (%) | 84 (51.9%) | 118 (55.7%) | 0.46 Biliary stent, n (%)† | 49 (32.2%) | 54 (26.2%) | 0.21 Metastatic at diagnosis, n (%) | 96 (59.3%) | 134 (63.2%) | 0.44 Primary tumor surgery, n (%) | 24 (14.8%) | 51 (24.1%) | 0.027 Liver metastases at diagnosis, n(%) | 65 (40.1%) | 92 (43.4%) | 0.52 Lung metastases at diagnosis, n(%) | 12 (7.4%) | 21 (9.9%) | 0.40 ECOG PS 0–1, n (%)† | 132 (99.2%) | 172 (94.0%) | 0.016 NLR ≥3, n (%)† | 45 (45.5%) | 77 (47.8%) | 0.71 CEA >5 ng/mL, n (%)† | 35 (44.9%) | 36 (33.6%) | 0.12 CA19-9 > 37 U/mL, n (%)† | 85 (83.3%) | 126 (78.3%) | 0.31 † Percentages calculated among patients with available data. NLR, neutrophil-to-lymphocyte ratio. CONCLUSIONS No significant differences in PFS or OS were observed between 5FU→GEM and GEM→5FU sequencing among patients who received at least two lines of chemotherapy. Although patients selected for the 5FU→GEM sequence were younger, had fewer comorbidities, and had better ECOG PS, this did not translate into a clear survival advantage for the 5FU-first strategy. PFS2 provides a pragmatic endpoint for evaluating treatment sequencing strategies and supports individualized sequencing based on clinical fitness and disease burden. The emergence of effective RAS-directed therapies is likely to redefine the role and timing of chemotherapy backbones in future treatment sequences. Disclosure: Kreina S. Vega Cano declares no conflicts of interest related to this study.

Lorlatinib showed unprecedented PFS, with median PFS not reached after 7 years in CROWN (NCT03052608) Table: Lorlatinib (n=149) | Crizotinib (n=147) Events, n: 62 | 119 PFS, months, median (95% CI): NR (68.5-NR) | 9.1 (7.4-10.9) HR (95% CI): 0.19 (0.13-0.26) Curve annotations (lorlatinib): 70%, 65%, 63%, 60%, 57%, 55%; (crizotinib): 15%, 10%, 10%, 8%, 7%, 3% Y axis: PFS, %; X axis: Months 0-100 No. at risk (censored) Lorlatinib: 149 126 118 111 103 96 93 89 87 81 81 79 78 75 73 71 69 68 66 65 62 38 28 9 2 0 (0) (8) (9) (10)(12)(14)(14)(17)(17)(19)(19)(20)(20)(21)(23)(23)(25)(25)(25)(25)(28)(51)(60)(78)(85)(87) Crizotinib: 147 107 70 42 30 19 16 16 11 10 9 9 9 8 8 7 6 6 6 5 2 2 0 0 0 0 (0) (15)(17)(20)(22)(23)(23)(23)(23)(24)(25)(25)(25)(25)(25)(25)(25)(25)(25)(25)(26)(26)(28) Reprinted from AT Shaw, et al. Ann Oncol. 2026;37(9):1242-1252. Copyright (c) 2026 The Authors. Published by Elsevier Ltd on behalf of European Society for Medical Oncology. Understanding lorlatinib resistance mechanisms and identifying patients at risk of early progression could help further improve outcomes for patients with advanced ALK-positive NSCLC - We analyzed ctDNA from patients in the lorlatinib group (n=149)(a) using the Guardant Health Infinity assay(b) to identify potential resistance mechanisms and predictors of early progression and long-term response to lorlatinib - Samples were collected at baseline (n=120) and EOT following progression (n=18) - EPs were defined as patients with a PFS event within 12 months; LTRs were defined as patients who were progression free at 84 months - Analyses included EOT resistance mechanisms, comparisons of baseline genomic alterations in EPs vs LTRs, efficacy by baseline biomarkers (EML4::ALK variant and TP53 status), and longitudinal methylation-based TF ALK, anaplastic lymphoma kinase; ctDNA, circulating tumor DNA; EOT, end of treatment; EP, early progressor; HR, hazard ratio; LTR, long-term responder; NSCLC, non-small cell lung cancer; PFS, progression-free survival; TF, tumor fraction. (a)Patients enrolled in China were excluded from ctDNA analyses. (b)The Guardant Infinity ctDNA next-generation sequencing assay assesses genomic alterations across 755 genes and uses a non-destructive methylation-based technology to quantify TF. Survival table Endpoint: PFS (CROWN, 7-year update reprinted from Shaw et al. Ann Oncol 2026) Arm | N | Median | Median CI Lorlatinib | 149 | NR | 68.5-NR Crizotinib | 147 | 9.1 | 7.4-10.9 HR: 0.19 HR CI: 0.13-0.26

[Slide 2] ADCs in PD-L1 > 50%: Awaiting results TROPION LUNG 08 Patient population (N = 740) • Stage IIIB, IIIC, or IV NSCLC • No actionable genomic alterations (documented EGFR/ALK/ROS1 negative) • No previous systemic therapy for adv/met NSCLC • ECOG PS 0 or 1 • Central PD-L1 TPS ≥50% R 1:1 → Dato-DXd 6 mg/kg IV Q3W + Pembrolizumab 200 mg IV Q3W | Pembrolizumab 200 mg IV Q3W Treatment until: • Occurrence of progressive disease by BICR, unacceptable toxicity, or other discontinuation criteria • Completion of 35 cycles of pembrolizumab (no maximum limit for Dato-DXd) Stratified by: • Histology (squamous vs nonsquamous) • Geography (East Asia vs rest of world) • Smoking status (former/current vs never) • ECOG PS (0 vs 1) TROFUSE-007 study Patient Population (N = 614) • Stage IV metastatic NSCLC • High PD-L1 expression (TPS ≥ 50%) • No actionable genomic alterations (EGFR/ALK negative) • Previously untreated (1st-Line) status • ECOG Performance Status 0 or 1 R 1:1 → Sacituzumab Tirumotecan (sac-TMT) Investigational TROP2 ADC + Pembrolizumab 200 mg IV Q3W | Pembrolizumab 200 mg IV Q3W (Standard-of-Care Monotherapy) Treatment Horizon & Endpoints Duration: • Treat until disease progression (BICR), unacceptable toxicity, or withdrawal. • Pembrolizumab max 35 cycles. Primary Endpoint: • Overall Survival (OS) Secondary Endpoints: • PFS, ORR, DOR, and Safety profiles. Stratified by: • Histology (Squamous vs. Nonsquamous) • Geography (East Asia vs. Rest of World) • Smoking status (Former/Current vs. Never) • ECOG Performance Status (0 vs. 1) [Slide 3] BR.36 Phase II adaptive trial Patients with metastatic PD-L1+ lung cancer after 6 weeks of immunotherapy → SCREEN → ctDNA+ and no progression on imaging → RANDOMIZE → Continue Immunotherapy | Add Chemotherapy Primary Endpoints: Progression-free & Overall survival Individuals receiving pembrolizumab and showing molecular progression on liquid biopsy can be rapidly and accurately identified and treatment can be optimized. Adapted from Anagnostou et al., Cancer Research 2019; Anagnostou et al., Nature Medicine 2023 (BR.36 clinical trial), Niknafs et al., Clin Cancer Res, 2026

[Slide 1] How to use surgery in advanced lung cancer? Surgical Principles for Lung Cancer Resection Goal R0 is mandatory If R1/R2 is likely → prefer SBRT or CRT Lobectomy vs Sublobar Resection Prefer lobectomy + systematic nodal dissection over sublobar if oncologically safe - Limited pulmonary reserve - Oligo lung meastases Pneumonectomy Avoidance Avoid pneumonectomy if possible • Favor sleeve / bronchoplastic / angioplastic procedures especially after immunotherapy (IO) Surgical Approach Minimally invasive surgery (VATS / RATS) is acceptable in experienced hands • Convert to open thoracotomy if safety concerns arise Lymph Node Management Perform systematic mediestinal lymphadenectomy even if nodes appear absent ("look gone") Residual micro-disease is common Yi-Long Wu @ Guandong Lung Cancer Institute, China [Slide 2] How to use surgery in advanced lung cancer? Surgical principles: For oligomets Adrenal solitary met → adrenalectomy (open/lap) often reasonable Brain solitary met → neurosurgery/SRS first if symptomatic; then thoracic primary Bone met → stabilisation/kyphoplasty + RT; surgery if mechanical instability/path fracture Contralateral lung nodule → treat as second primary if meets criteria, not automatic stage IV Yi-Long Wu @ Guandong Lung Cancer Institute, China [Slide 3] How to use surgery in advanced lung cancer? When not to operate Preoperative multidisciplinary assessment identifies patients who may require MDT discussion or treatment strategy adjustment. High-Risk Factors Metastatic & Progressive Disease Risk: Widely metastatic (multi-organ, M1c); Uncontrolled systemic progression on treatment Pleural Disease Limitations: • Diffuse pleural seeding or malignant effusion • Limited/localised pleural disease after MDT discussion is sdebated and not standard Disease Burden & Dissemination Risk: N3 with persistent dissemination risk and no major response Performance Status & Organ Reserve: ECOG ≥ 2 / frailty / insufficient pulmonary reserve Active Immune-Related Conditions: Active immune-related pneumonitis ≥ grade 2, hepatitis, myocarditis, etc. Driver-Mutant & Local Control: Driver-mutant disease where operation would delay superior TKI benefit unless local control is clearly additive. Key takeaway: More high-risk factors increase the need for MDT discussion and caution with surgical timing. Yi-Long Wu @ Guandong Lung Cancer Institute, China [Slide 4] Take home message Why use surgery in advanced lung cancer? Some patients can get survival benefit from surgery When use surgery in advanced lung cancer? Under effective system treatment How to use surgery in advanced lung cancer? R0 always How to select patient to surgery in advanced lung cancer Patients with special clinical phenotype or molecular based

Progression-free survival by BICR: MET subgroups MET-overexpressed analysis set Median PFS, months (95% CI) Osimertinib + savolitinib | 8.5 (5.7, 11.1) Platinum–pemetrexed | 5.6 (4.4, 6.7) HR (95% CI); p-value | 0.51 (0.37, 0.71); p<0.001 12-month labels on curve: 40% | 11% No. at risk Osimertinib + savolitinib: 124 120 106 94 92 76 64 59 55 50 40 36 29 26 22 18 17 16 16 10 10 6 6 5 4 1 1 1 1 1 1 1 1 1 1 1 1 1 0 | 127 / 77 Platinum–pemetrexed: 118 95 77 66 63 49 33 26 23 16 13 10 7 6 6 3 3 2 2 1 1 1 0 [remaining time points 0] | 118 / 80 MET-amplified analysis set* Median PFS, months (95% CI) Osimertinib + savolitinib | 7.6 (5.6, 9.7) Platinum–pemetrexed | 4.2 (3.1, 5.5) HR (95% CI) | 0.48 (0.35, 0.66) 12-month labels on curve: 33% | 8% No. at risk Osimertinib + savolitinib: 129 126 108 93 91 76 63 58 54 49 39 35 28 24 20 19 19 17 16 11 11 6 6 4 4 2 1 1 1 1 1 1 1 1 1 1 1 1 0 | 129 / 89 Platinum–pemetrexed: 116 96 72 52 42 32 23 19 16 12 9 5 4 3 3 1 1 0 [remaining time points 0] | 116 / 79 (No. of patients / no. of events) PFS by BICR in the MET-overexpressed and MET-amplified analysis sets was generally consistent with the full analysis set Data cut-off: 06 July 2026. *No formal statistical testing was carried out for the MET-amplified analysis set. BICR, blinded independent central review; CI, confidence interval; HR, hazard ratio; PFS, progression-free survival. Survival table Endpoint: PFS by BICR (MET-overexpressed and MET-amplified analysis sets) Arm | N | Median | Median CI Osimertinib + savolitinib (MET-overexpressed) | 127 | 8.5 | 5.7, 11.1 Platinum–pemetrexed (MET-overexpressed) | 118 | 5.6 | 4.4, 6.7 Osimertinib + savolitinib (MET-amplified) | 129 | 7.6 | 5.6, 9.7 Platinum–pemetrexed (MET-amplified) | 116 | 4.2 | 3.1, 5.5 HR: MET-overexpressed 0.51; MET-amplified 0.48 HR CI: 0.37, 0.71; 0.35, 0.66 P: MET-overexpressed p<0.001; MET-amplified no formal statistical testing Landmark rates: 12 months: Osimertinib + savolitinib (MET-overexpressed) 40% | Platinum–pemetrexed (MET-overexpressed) 11% | Osimertinib + savolitinib (MET-amplified) 33% | Platinum–pemetrexed (MET-amplified) 8%

[Slide 2] Background & Trial Design • First-line treatment for advanced NSCLC without targetable genomic alterations is based mainly on immunotherapy alone or combined with chemotherapy, but further improvement in outcomes is needed. TROP2-targeted ADCs may enhance PD-1/L1 blockade through cytotoxicity and immune activation. • BAT8008, a TROP2-directed ADC with an exatecan payload (DAR ~6), combined with BAT1308, an anti-PD-1 antibody, has shown manageable safety and encouraging preliminary antitumor activity across explored dose levels in previously treated NSCLC. Trial design BAT-8008+1308-002 Part B Main inclusion criteria • Pathologically confirmed stage IIIB–IV NSCLC • No prior systemic treatment • No EGFR mutation or ALK rearrangement • ECOG PS 0–1 • Archived tumor tissue or a new biopsy required Combination therapy BAT8008, 2.4 mg/kg IV Q2W + BAT1308, 200 mg IV Q2W Up to 2 years PFS & OS follow-up Primary endpoints: Safety and investigator-assessed efficacy (ORR, DOR, DCR, PFS, etc.) Secondary endpoints: PK parameters, ADA/NAb, and endpoints by TROP2 expression level [Slide 3] Baseline Characteristics & Safety • 104 patients recruited in China, and no new safety signals were observed. • Anemia and stomatitis were the most common TRAEs, with few treatment-related discontinuations. Characteristic | Total (N=104) Age – Median (range), years | 63.5 (39, 76) Histology – Squamous cell carcinoma | 55 (52.9%) Non-squamous cell carcinoma | 49 (47.1%) PD-L1 TPS – ≥50% | 32 (30.8%) 1%–49% | 30 (28.8%) <1% | 38 (36.5%) Not tested | 4 (3.8%) ECOG score – 1 | 92 (88.5%) TROP2 score1* – High expression | 27 (26.0%) Low expression | 24 (23.1%) Not tested | 53 (51.0%) Clinical stage – IV | 79 (76.0%) Safety summary | Total (N=104) TEAEs, n (%) | 104 (100.0) Treatment-related | 104 (100.0) Grade ≥3 TEAEs, n (%) | 69 (66.3) Treatment-related | 63 (60.6) Serious TEAEs, n (%) | 51 (49.0) Treatment-related | 41 (39.4) TEAEs leading to discontinuation of any study drug, n (%) | 6 (5.8) Treatment-related | 6 (5.8) Leading to discontinuation of BAT8008 | 4 (3.8) Leading to discontinuation of BAT1308 | 4 (3.8) TRAEs associated with death, n (%)2* | 1 (1.0) Figure 1. TRAEs reported in ≥15% of patients (Grade ≥3 / total %) Anemia 17.3 / 78.8 Stomatitis 16.3 / 65.4 White blood cell count decreased 18.3 / 55.8 Neutrophil count decreased 23.1 / 50.0 Alanine aminotransferase increased 2.9 / 31.7 Decreased appetite 1.0 / 30.8 Hypoalbuminemia – / 27.9 Aspartate aminotransferase increased 3.8 / 26.0 Asthenia 3.8 / 23.1 Constipation – / 18.3 Nausea – / 18.3 Platelet count decreased 6.7 / 17.3 Pneumonia 9.6 / 15.4 Legend: Grade 1/2; Grade ≥3. x-axis: Patients with TRAE (%) Date cutoff: Sep 22, 2026 1* TROP2 score high expression: H-score≥200 2* The only TRAE leading to death was cerebral infarction [Slide 4] Efficacy • At a median follow-up of 8.9 months (range, 7.9–9.5), the overall ORR was 61.2%, with higher response rates observed with increasing PD-L1 expression: 47.2% in TPS <1%, 58.6% in TPS 1%–49%, and 83.3% in TPS ≥50%. • Tumor shrinkage and objective responses were observed across PD-L1 subgroups, histologic subtypes, and TROP2 expression categories. Figure 2. Best Percentage Change in Sum of Diameters of Target Lesions [Waterfall plot] PD-L1 <1% (N=34) | PD-L1 1%–49% (N=29) | PD-L1 ≥50% (N=29) | PD-L1 unknown (N=3) y-axis: Best change in target-lesion sum from baseline (%) Histology: Squamous; Non-squamous TROP2: High expression; Low expression; Unknown Date cutoff: Sep 22, 2026

2650P Dose modifications to manage adverse events in 1L EGFR-mutated advanced NSCLC: Post hoc, exploratory analyses from the phase 3 MARIPOSA trial Shirish M Gadgeel*, Byoung Chul Cho, Nicolas Girard, Enriqueta Felip, Yichuan Xia, George Wang, Simon Todd, Seema Sethi, Andy L Johnson, Maria del Rosario Garcia Campelo Key Takeaway Amivantamab dose modifications for grade ≥2 AEs are an effective approach to mitigate rash/dermatitis acneiform, paronychia, and peripheral edema Conclusions A high rate of AE improvement (approximately 80%) was found in participants who had their dose modified with grade ≥2 key AEs (rash/dermatitis acneiform, paronychia, peripheral edema) After dose modification, most AEs improved in <1 month, with grade ≥3 AEs improving in approximately 2 weeks SC amivantamab and prophylactic measures, which were not included in MARIPOSA, have now been shown to further mitigate the incidence and severity of key AEs, and may enable patients to stay on therapy longer Introduction • In MARIPOSA (ClinicalTrials.gov Identifier: NCT04487080), first-line (1L) intravenous (IV) amivantamab + lazertinib significantly improved overall survival versus osimertinib (hazard ratio, 0.75; P=0.005) in participants with common (exon 19 deletion/exon 21 L858R substitution) epidermal growth factor receptor (EGFR)–mutated advanced non-small cell lung cancer (NSCLC) • Key adverse events (AEs) first occurred during the first 4 months after randomization in MARIPOSA – Dose modifications, including dose interruptions and dose reductions, assisted AE management while maintaining efficacy (Figure 1) • Notably, MARIPOSA was conducted prior to the availability of subcutaneous (SC) amivantamab and prospective data on the effectiveness of prophylactic AE strategies • The COPERNICUS study, which evaluated SC amivantamab + lazertinib in 1L EGFR-mutated NSCLC, showed that the SC formulation with dermatologic and venous thromboembolism prophylaxis demonstrated a meaningful reduction in the incidence and severity of AEs of special interest compared with MARIPOSA (Figure 2) • Here, we report the impact of first dose modifications among key clinically observable EGFR- and MET-related AEs that were the most common reasons for IV amivantamab dose modifications in MARIPOSA Figure 1: PFS by dose modification status within 4 months HR, 1.12 (95% CI, 0.80–1.58) No. at risk: Without dose modifications for amivantamab 154 154 143 131 107 72 28 6 0; With dose modifications for amivantamab 194 194 174 148 126 75 30 2 0 (0–32 months) Figure 2: Incidence of selected AESIs during the first 4 months of treatment in COPERNICUS and MARIPOSA (COPERNICUS grade 1–2 / grade ≥3 | MARIPOSA grade 1–2 / grade ≥3) Paronychia: COPERNICUS 35% (grade ≥3 1%) | MARIPOSA 50% (grade ≥3 4%) Rash: COPERNICUS 25% (grade ≥3 3%) | MARIPOSA 55% (grade ≥3 7%) VTE: COPERNICUS 3% (grade ≥3 <1%) | MARIPOSA 23% (grade ≥3 6%) ARR/IRR: COPERNICUS 3% (grade ≥3 <1%) | MARIPOSA 55% (grade ≥3 5%) The term ARR was used in COPERNICUS, and the term IRR was used in MARIPOSA. Methods • Participants with EGFR-mutated advanced NSCLC were randomized 2:2:1 to receive amivantamab + lazertinib (n=429), osimertinib monotherapy (n=429), or lazertinib monotherapy (n=216) – The safety analysis set (amivantamab + lazertinib: n=421) included randomized participants who received ≥1 dose of any study treatment • This post hoc, exploratory analysis of the amivantamab + lazertinib arm included per-protocol amivantamab dose modifications recommended for treatment-related grade ≥2 AEs – For grade 1 or 2 dermatologic AEs, supportive care was initiated and dose reduction was considered if there was no improvement after 2 weeks; for grade 3, amivantamab treatment was interrupted – For other grade 3 AEs, amivantamab treatment was withheld until recovery to grade ≤1 or baseline – Amivantamab was to be modified prior to modifying lazertinib in participants who experienced treatment-related grade ≥2 AEs • Improvement was defined as resolution of the AE, reduction in grade (eg, grade 3 to 2, grade 2 to 1, or grade 3 to 1), or both Results Types of dose modifications • The majority of participants (363/421; 86%) experienced amivantamab dose modifications (interruptions and/or reductions) – 345 participants had dose interruptions, and 274 participants had dose reductions • Median time to dose modification was 85 days (range, 2–925) • By 6 months, the median relative dose intensity among treated participants, defined as the percentage of the planned total dose received, was 83% AEs leading to first dose modifications • The most common amivantamab-associated AEs leading to first dose modifications were paronychia (27%), rash (26%), dermatitis acneiform (13%), and peripheral edema (5%) • Among common AEs reported (rash/dermatitis acneiform, paronychia, peripheral edema), 86% to 92% were grade 1 or 2 at first onset • Dose modifications for AEs typically occurred at grade 2 or 3 (Figure 3) – Interruption was the most common first modification (82% of participants) • Most participants who dose modified with grade ≥2 rash/dermatitis acneiform, paronychia, or peripheral edema experienced AE improvement (Figure 4) Figure 3: Grade at time of first dose modification (Grade 3 / Grade 2 / Grade 1) Rash/dermatitis acneiform (n=152): 32% / 51% / 18%; Grade 2–3: 83% Paronychia (n=113): 20% / 65% / 14%; Grade 2–3: 85% Peripheral edema (n=21): 14% / 67% / 19%; Grade 2–3: 81% Note: Totals may not sum to 100% due to rounding. Figure 4: Percentage of participants with improvement in grade ≥2 rash/dermatitis acneiform, paronychia, and peripheral edema after dose modification Rash/dermatitis acneiform (n=125) 80%; Paronychia (n=97) 77%; Peripheral edema (n=17) 82% Time to AE improvement after first dose modification for grade ≥2 AEs • After dose modification, the median time to improvement among those who improved was <1 month for both grade ≥2 and grade ≥3 AEs (Table 1) Table 1: Median time to improvement since dose modification for grade ≥2 AEs (Median time to AE improvement (days): Grade ≥2 AEs | Grade ≥3 AEs) Rash/dermatitis acneiform 16.5 | 15.0 Paronychia 22.0 | 15.5 Peripheral edema 27.5 | 15.0 This study was funded by Janssen Research & Development, LLC, a Johnson & Johnson company. Presented by SM Gadgeel at the European Society for Medical Oncology (ESMO) Congress; October 23–27, 2026; Madrid, Spain.

[Slide 1] We report a subgroup analysis of the DAREON®-7 trial by DLL3 expression status DAREON®-7 (NCT06132113): Phase I dose-escalation/expansion trial of 1L obrixtamig + SoC Patient population • Locally advanced/metastatic extrapulmonary NEC, LCNEC-L, or NEC of unknown primary site • DLL3-positive (central testing) • No prior treatment with DLL3-targeted T-cell engagers or cell therapies Primary endpoint: DLTs Secondary endpoint: Part A: AEs and DLTs; Part B: ORR and DoR* Further endpoint: PFS* DLL3 expression status Categorized by the proportion of tumor cells with moderate-to-strong cytoplasmic and/or membranous staining with an investigational antibody for DLL3 (SP347; Roche Diagnostics) • DLL3-positive: >0% – DLL3low: <50% – DLL3high: ≥50% Part A: Dose escalation (BLRM with overdose control; N=25) Obrixtamig (target dose levels: 10, 30, 60 mg)† + SoC‡ ↓ RP2D Part B: Dose expansion (N=19) Obrixtamig 60 mg† + SoC‡ Treatment schedule Induction: Cycle 1: SoC‡ | Cycle 2–6: Obrixtamig + SoC‡ → Maintenance: Cycle 7 onwards:§ Obrixtamig Q3W *RECIST v1.1 (investigator-assessed); †Patients received obrixtamig as step-up dosing followed by target dose; ‡Part A: carboplatin + etoposide, Part B: investigator's choice of platinum (carboplatin / cisplatin) + etoposide; §Treatment continues until disease progression or another reason requiring termination, for up to 36 months 1L, first-line; AE, adverse event; BLRM, Bayesian logistic regression model; DLL3, delta-like ligand 3; DLT, dose-limiting toxicity; DoR, duration of response; LCNEC-L, large cell neuroendocrine carcinoma of the lung; NEC, neuroendocrine carcinoma; ORR, objective response rate; PFS, progression-free survival; Q3W, every 3 weeks; RECIST v1.1, Response Evaluation Criteria in Solid Tumours version 1.1; RP2D, recommended Phase II dose; SoC, standard of care Jaume Capdevila [Slide 2] Key findings and conclusions DLL3 expression status: Overall, 44 patients with DLL3-positive NEC received 1L obrixtamig + SoC. Of these patients: 54.5% had DLL3high tumors; 45.5% had DLL3low tumors Safety: A favorable safety profile was maintained regardless of DLL3 expression status Efficacy: Encouraging efficacy was shown across both DLL3 subgroups, with numerically higher response rates and durable responses in the DLL3high subgroup; longer follow up is needed Conclusions: These findings provide a strong rationale for evaluating 1L obrixtamig + SoC in patients with DLL3-positive extrapulmonary NEC in the ongoing Phase III DAREON®-NEC-1 trial (NCT07544654) 1L, first-line; DLL3, delta-like ligand 3; NEC, neuroendocrine carcinoma; SoC, standard of care Jaume Capdevila [Slide 4] The safety profile of obrixtamig + SoC was aligned with known profiles of individual components across both DLL3 subgroups n (%) | All N=44: Any grade / Grade ≥3* | DLL3high n=24: Any grade / Grade ≥3 | DLL3low n=20: Any grade / Grade ≥3 TEAEs | 44 (100) / 41 (93.2) | 24 (100) / 22 (91.7) | 20 (100) / 19 (95.0) Obrixtamig-related AEs | 39 (88.6) / 8 (18.2) | 23 (95.8) / 4 (16.7) | 16 (80.0) / 4 (20.0) Most common obrixtamig-related AEs† CRS | 32 (72.7) / 0 | 22 (91.7) / 0 | 10 (50.0) / 0 Chemotherapy-related AEs | 44 (100) / 35 (79.5) | 24 (100) / 17 (70.8) | 20 (100) / 18 (90.0) Most common chemotherapy-related AEs† Neutropenia | 30 (68.2) / 24 (54.5) | 14 (58.3) / 10 (41.7) | 16 (80.0) / 14 (70.0) Anemia | 25 (56.8) / 19 (43.2) | 15 (62.5) / 12 (50.0) | 10 (50.0) / 7 (35.0) Dysgeusia | 23 (52.3) / 0 | 13 (54.2) / 0 | 10 (50.0) / 0 All cases of CRS were grade 1 or 2 [Bar chart: CRS by grade (Grade 1 / Grade 2), Patients (%) 0–100 for All, DLL3high, DLL3low] Obrixtamig-related ICANS and potential ICANS-like neurotoxicity were infrequent (n=2; 4.5%) and grade 1‡ *One grade 5 AE occurred (malignant neoplasm progression), this was not related to obrixtamig; †Any-grade AEs in ≥50% patients in the overall population; ‡Both cases were ICANS (DLL3high: n=1; DLL3low: n=1) AE, adverse event; CRS, cytokine release syndrome; DLL3, delta-like ligand 3; ICANS, immune effector cell-associated neurotoxicity syndrome; SoC, standard of care; TEAE, treatment-emergent AE Jaume Capdevila

[Slide 1] Conclusions •Median alignment was 68% with stronger design tan reporting and substantial gaps between planned and published assessments •Limited inclusion of patients with poor PS •Frequent reliance on surrogate endpoints •Major gaps in patient-centred outcomes, particularly HRQoL and long-term toxicity reporting Priority: More inclusive trials and transparent reporting of clinically meaningful, patient centred outcomes. Javier David Benitez-Fuentes | 3323RO [Slide 2] RAPID ORAL 3323RO Applying the Common Sense Oncology (CSO) checklist for phase III trials supporting European Medicines Agency (EMA) lung cancer approvals, 2020–2025 Javier David Benitez-Fuentes¹, Alba Moratiel Pellitero², Elisa Cabrera Acosta³, Paloma Alejandra Flores Navarro⁴, Ana Callejo Perez⁵, Rodrigo Lastra del Prado⁶ [affiliations: illegible]

Main causes and prevalence of drug-related ILD NO-DRUG RELATED Progression of the underlying oncologic disease Infective pneumonia (ie, bacterial, mycotic, and viral, including SARS-CoV-2) Radiotherapy-induced pneumonitis Other drug-induced lung toxic effect (ie, cancer drug or concomitant medication) Smoking-related ILD Cardiogenic pulmonary edema Connective tissue disorders (eg, scleroderma, lupus, rheumatoid arthritis, Sjögren syndrome, dermatomyositis/polymyositis, ankylosing spondylitis, and mixed connective tissue disease) Occupational/environmental (eg, pneumoconiosis, anthracosis, asbestosis, silicosis, and hypersensitivity pneumonitis) Vasculitis Sarcoidosis Pulmonary hemorrhage syndromes Idiopathic interstitial fibrosis Alveolar proteinosis Amyloidosis Bronchiolitis obliterans Chronic eosinophilic pneumonia Lymphocytic infiltrative lung disease DRUG RELATED DRUG | PREVALENCE BLEOMYCIN | 6.8-21% TAXANES | 4.6% ANTHRACYCLINES | RARE IRINOTECAN | RARE GEMCITABINE | 1-20% mTOR-inh | 10.4% ICI | 3.6% for PD-1 inh and 1.1% for PD-L1 inh EGFR-targeting agents | 1.2-1.6% METHOTREXATE (low-dose) | 0.3-2.1% CDK4/6inh | 0.3-2.1% T-DM1 | 0.5% (2.6% with concomitant RT) New drugs (phase I trials) | 5.3% Skeoch, et al 2018; Tarantino et al, 2021; Raschi et al, BCRT 2021; Trapani et al, 2024

Preoperative ctDNA+ associated with a higher likelihood of pathologic upstaging in clinical stage I NSCLC: The lung DETECT study • Multicenter prospective cohort study in clinical stage I NSCLC, preoperative ctDNA detection using a tumor-informed assay • Patients underwent guideline-concordant staging with PET / CT, CT-guided biopsy, and/or EBUS • Among 153 patients with clinical stage I, 34 were ctDNA-positive preoperatively • 14 of 153 patients had occult LN metastases discovered. Preoperative ctDNA was detected in 8 of the 14 LN+ patients Entered screening (n = 231) → Excluded (n = 78) Insufficient tissue for ctDNA panel creation (n = 48) Enrolled NSCLC Clinical stage I (n = 153) Pathologic Stage I NSCLC (n = 117) → Preoperative ctDNA positive (n = 16) Pathological stage II, III, IV NSCLC (n = 36) → Upstaged node negative patients (n = 22) → Preoperative ctDNA positive* (n = 10); Upstaged node positive patients (n = 14) → Preoperative ctDNA positive (n = 8) Clinical Stage I NSCLC → Tumor-informed preoperative ctDNA detected → Higher detection rate of occult LN metastasis Preoperative ctDNA is associated with occult LN metastasis in clinical stage I NSCLC. • ctDNA as a complementary tool to refine staging and guide decisions regarding invasive mediastinal staging or perioperative treatment in clinical stage I Coyne JTCVS Open 2026

[Slide 1] Greater Treatment Benefits Were Observed in Patients With Increased Expression (3+, ≥10%) of c-Met Protein in the EGFRwt Cohort • In patients receiving Temab-A in the EGFRwt population, responses were observed regardless of c-Met cutoffs – Enrichment of benefit was observed above the 3+, ≥10% c-Met cutoff, with a median OS of 25.2 months – 43% (20/46) of EGFRwt patients with available IHC samples had c-Met 3+, ≥10% Legend: Overall | 3+, ≥10% | 3+, <10% ORR, % (n/N): Overall 50% (24/48); 3+, ≥10% 60% (12/20); 3+, <10% 42% (11/26) PFS, Median (95% CI): Overall 6.8; 3+, ≥10% 7.2; 3+, <10% 5.4 (months) OS, Median (95% CI): Overall 13.8; 3+, ≥10% 25.2 (7.5, NE); 3+, <10% 9.7 (months) DOR, Median (95% CI): Overall 5.6; 3+, ≥10% 5.6; 3+, <10% 5.4 (2.7, NE) (months) Data cutoff date: 02 June 2026 CI, confidence interval; DOR, duration of response; EGFRwt, epidermal growth factor receptor wild type; IHC, immunohistochemistry; NE, not estimable; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; Temab-A, telisotuzumab adizutecan. [Slide 2] Consistent Treatment Benefit Was Observed in Patients Who Received Temab-A in the EGFRmut Cohort • In the EGFRmut population, enrichment in Temab-A benefit was not observed based on c-Met cutoff; however, responses were observed as compared to the overall population even at a c-Met cutoff of 3+, ≥25% – 49% (19/39) of EGFRmut patients with available IHC samples had c-Met 3+, ≥25% Legend: Overall | 3+, ≥25% | 3+, <25% ORR, % (n/N): Overall 63% (26/41); 3+, ≥25% 68% (13/19); 3+, <25% 60% (12/20) PFS, Median (95% CI): Overall 10.9; 3+, ≥25% 12.3; 3+, <25% 9.5 (months) OS, Median (95% CI): Overall 16.2; 3+, ≥25% 16.2; 3+, <25% 16.2 (months) DOR, Median (95% CI): Overall 9.7; 3+, ≥25% 10.9; 3+, <25% 9.0 (months) Data cutoff date: 02 June 2026 CI, confidence interval; DOR, duration of response; EGFRmut, epidermal growth factor receptor mutated; IHC, immunohistochemistry; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; Temab-A, telisotuzumab adizutecan. [Slide 3] ORR's of ADC's in resistant EGFR mut NSCLC Drug | ORR | Phase | Median OS | Outcome Dato-Dxd | 50% | III subset | 15.6 mo | HR OS 0.66 (TL-01) Patritumab-Dxd | 35% | III | 16 mo | OS negative Sac-TMT | 41% | III | NR | HR OS 0.66 p<0.001 Temab-A | 63% | II | 16.2 mo | Egbert Smit

TUMOR-TARGETING BsAbs Zenocutuzumab: Disrupting NRG1-driven HER2/HER3 signalling ST3204, CD74–NRG1, lung cancer PDX A Vehicle Zeno, 2.5 mg/kg Zeno, 8 mg/kg Zeno, 25 mg/kg Tumor volume (mm3) Treatment time (days) AUC Change in tumor volume (%) LUAD-0061AS3, SLC3A2–NRG1, lung cancer PDX B P < 0.0001 P < 0.0001 P < 0.0001 Dock and block HER2 binding positions HER3 blockade against NRG1-driven signalling. Lung PDXs CD74-NRG1 and SLC3A2-NRG1 models show suppression of tumor growth. Biological selection The oncogenic ligand-receptor circuit provides the rationale for selecting NRG1-fusion cancers. Tina Cascone, MD, PhD Schram et al. Cancer Discovery 2022

[Slide 1] LBL-024-CN001 Study Design Here, we report results from the Pivotal Phase IIb study. Phase I (N=64) 25 mg/kg n=16 15 mg/kg n=12 10 mg/kg n=12 6 mg/kg n=7 3.2 mg/kg n=13 0.8 mg/kg n=3 0.2 mg/kg n=1 "3+3" Rule; ATD RP2D 15 mg/kg Opamtistomig, IV, Q3W Key Eligibility Criteria: - Adults (18+ years) - Measurable advanced malignant tumors, for which prior standard therapy failed - ECOG of 0-1, and adequate organ function Phase IIa (N=111) Arm A (NEC) Previously treated, IO-naïve Neuroendocrine Carcinoma Arm B (BTC) Previously treated, IO-naïve Biliary Tract Carcinoma Arm C (NSCLC) Cohort C1: Failed previous PD-(L)1 or platinum-based chemotherapy NSCLC, with no driver mutation; Cohort C2: Failed previous TKI or platinum-based chemotherapy, PD-(L)1 naïve NSCLC, with a positive driver mutation Arm D (HCC, ESCC) Previously IO-treated Hepatocellular Carcinoma, Esophageal Squamous Cell Carcinoma, etc Phase IIb (Pivotal Study, N=96) RP2D 15 mg/kg → 3L+ EP-NEC Key Eligibility Criteria: - Central Lab confirmed EP-NEC prior to enrollment Predefined success criterion: Lower bound of 95% CI >13% (IRC-assessed confirmed ORR per RECIST 1.1) Phase I Endpoints Primary: Safety and tolerability, RP2D Secondary: PK, immunogenicity, efficacy per RECIST1.1 Phase IIa Endpoints Primary: ORR per RECIST1.1 Secondary: Safety, immunogenicity, DOR, PFS Phase IIb Endpoints Primary Endpoint: ORR by IRC per RECIST1.1 Secondary Endpoints: ORR by INV, DoR, DCR, TTR, PFS, OS, Safety Ming Lu [Slide 2] Efficacy Opamtistomig shows robust anti-tumor activity As of August 16, 2026, data cutoff, the median follow-up duration was 16.3 months. In the FAS, IRC-assessed confirmed ORR was 26.0% (95% CI: 17.6%–36.0%), with the lower bound exceeding the prespecified 13% efficacy threshold. Parameters | IRC-assessed per RECIST 1.1 Total (N=96) | INV-assessed per RECIST 1.1 Total (N=96) BOR, n (%) CR | 3 (3.1) | 0 PR | 22 (22.9) | 28 (29.2) SD | 9 (9.4) | 10 (10.4) PD | 52 (54.2) | 45 (46.9) NE | 0 | 3 (3.1) ND | 10 (10.4) | 10 (10.4) ORR, n (%) [95% CI] | 25 (26.0) [17.6-36.0] | 28 (29.2) [20.3-39.3] DCR, n (%) [95% CI] | 34 (35.4) [25.9-45.8] | 38 (39.6) [29.7-50.1] Best Overall Response: CR, PR, SD, PD BOR, best overall response; CR, complete response; PR, partial response; SD, stable disease; PD, Progressive Disease; NE, not evaluable; ND, not assessed; ORR, Objective Response Rate; DCR, disease-control rate; FAS: Full analysis set; IRC: Independent Review Committee; INV: Investigator Cutoff date: August 16, 2026. Ming Lu [Slide 3] Efficacy Overall survival (FAS) As of August 16, 2026, mOS was 8.7 months with a median follow-up duration of 16.3 months. In patients with BOR as PD (n=52), mOS was 6.1 months. Overall Survival (%) Time (Month) Median(95% CI) 15mg/kg 8.7 (6.1, 11.7) No. at Risk (No. Censored) No. of Events Parameters | Total (N=96) OS events Deaths, n (%) | 62 (64.6) Censored, n (%) | 34 (35.4) mOS, months (95% CI) | 8.7 (6.1-11.7) 12-m OS rate, % (95% CI) | 39.4 (29.5-49.1) mFu, months (95% CI) | 16.3 (13.5-19.0) mOS, median overall survival; mFu, median follow up. Ming Lu Survival table Endpoint: OS (FAS) Arm | N | Median | Median CI Opamtistomig 15 mg/kg | 96 | 8.7 mo | 6.1-11.7 12-mo OS rate 39.4% (29.5-49.1) [Slide 4] Safety LBL-024 was well tolerated with a manageable safety profile TEAEs occurred in 94.8% of patients - Grade ≥3 TRAEs: 27 patients (28.1%) - Treatment-related death: 4 patients (4.2%)* The most common TRAEs were anemia, increased AST, increased ALT. Liver enzymes abnormalities were predominantly grade 1-2 and transient. TRAEs in ≥15% of patients, Safety analysis set (n=96) Preferred Terms, n (%) | Any grade | ≥Grade 3 Anemia | 39 (40.6) | 6 (6.3) AST increased | 31 (32.3) | 2 (2.1) ALT increased | 30 (31.3) | 0 White blood cell count decreased | 20 (20.8) | 4 (4.2) Gamma-glutamyl transferase increased | 18 (18.8) | 4 (4.2) Platelet count decreased | 18 (18.8) | 3 (3.1) Proteinuria | 18 (18.8) | 0 Blood creatine phosphokinase increased | 17 (17.7) | 2 (2.1) Lymphocyte count decreased | 16 (16.7) | 3 (3.1) Hypothyroidism | 16 (16.7) | 0 Hypercholesterolemia | 16 (16.7) | 0 Hypertriglyceridemia | 16 (16.7) | 1 (1.0) Neutrophil count decreased | 15 (15.6) | 3 (3.1) Hyponatremia | 15 (15.6) | 1 (1.0) Decreased appetite | 15 (15.6) | 0 TEAE, treatment-emergent adverse event; TRAE, treatment-related adverse event; AST, Aspartate aminotransferase; ALT, Alanine aminotransferase *One patient experienced infectious pneumonia, one had sepsis, and one had septic shock, and one died of an unspecified cause. Cutoff date: August 16, 2026 Ming Lu

[Slide 2] Monotherapy or combination therapy? Factors to consider Monotherapy Tumour related factors: Low volume disease; Baseline ctDNA-ve; Exon 19 del Patient related factors: Older; Co-morbidities; Poor functional status Combination Tumour related factors: CNS mets; High tumour burden; TP53mt; ctDNA+ve; Liver mets Patient related factors: Good PS; Adequate end organ function; Patient preference; Social support Many pts in RWE setting have high risk features but not all are suitable for combination U.S. medical oncology databases Age 65+ years 66% Brain metastases 36% TP53mt 63% EGFR L858R 48% Liver metastases 15% ECOG PS 2 or more 17% ctDNA+ve 85% *from MARIPOSA Ross Soo, MB BS, PhD, FRACP Felip Annals Oncology 2024, Sabari JTO 2025 [Slide 3] Intensification based on detection of ctDNA Can we start with osimertinib alone and add chemotherapy only if ctDNA remains positive at 3 weeks? FLAME study Patients with locally advanced / metastatic EGFRm NSCLC who receive 3-weeks first-line osimertinib Key inclusion criteria: - Aged ≥ 18 years - History of EGFRm(19del/21 L858R) in plasma ctDNA prior to 1L osimertinib - Persistence ctDNA EGFRm at 3 weeks after 1L osimertinib - No disease progression by RECIST 1.1 - Asymptomatic and stable brain metastatic allowed - ECOG PS 0-1 N=80 R 1:1 Stratified by: Brain metastasis status (yes/no); EGFR mutation type (19Del/21 L858R) Osimertinib 80mg QD + pemetrexed 500 mg/m² + carboplatin AUC5 (Q3W for 4-6 cycles) Maintenance osimertinib 80 mg (QD) + pemetrexed 500 mg/m2 (Q3W) Osimertinib 80mg QD Follow-up: RECIST 1.1 assessment at 6, 12 and 18 weeks, then every 12 weeks until RECIST 1.1 defined radiological disease progression or other withdrawal criteria were met - Primary endpoint: PFS by investigator assessment per RECIST 1.1 - Secondary endpoints: 18-month OS rate, ORR, DCR, DoR, Depth of response, safety, resistance mechanisms - Exploratory endpoints: The dynamic change of molecular alterations by plasma analysis, QoL Primary endpoint: Progression-Free Survival N=80 | Chemo+ Osimertinib (n=40) | Osimertinib (n=40) ORR | 50% | 35% DOR | 15.6m | 10.5m PFS | 23.1m | 12.7m OS | NE | 30.5m - Persistent ctDNA after 3w of osimertinib identifies a high-risk population which benefits from the addition of chemotherapy - Not a biomarker predictive study (ctDNA-ve population not studied) - Open label design (observer bias) - Small sample size - <20% screened were randomized - Arbitrary 3-week cutoff Ross Soo, MB BS, PhD, FRACP Wang ASCO 2026 [Slide 4] After osimertinib: more therapeutic options Disease progression after 1L osimertinib Oligo-progression → Local therapy (RT/ surgery) Systemic progression Tumour biopsy → Histological transformation → Histology-specific chemotherapy Empiric approach: Chemotherapy; Chemo+ ICI+ anti-VEGF; Chemo+ amivantamab; Chemo+ osimertinib; ADCs; BsAb+ chemo Precision oncology approach: Osimertinib+ MET TKI; 4th gen EGFR TKI; Osimertinib+ matching AGA therapy Ross Soo, MB BS, PhD, FRACP

[Slide 1] Time horizon of software tasks different LLMs can complete 50% of the time METR Y-axis: Task duration (for humans) where logistic regression of our data predicts the AI has a 50% chance of succeeding: 4 sec, 36 sec, 6 min, 1 hour, 10 hours Task examples: Answer question; Count words in passage; Find fact on web; Train classifier; Train adversarially robust image model; Implement complex protocol from multiple RFCs Models labeled: GPT-2, GPT-3, GPT-3.5, GPT-4, GPT-4o, o1-preview, o1, o3, GPT-5, Claude Opus 4.6 X-axis: LLM release date 2020 2021 2022 2023 2024 2025 2026 Time Horizon 1.1 (Current) | Log Scale | Linear Scale | 50% Success | 80% Success [Slide 2] Maximum efficiency of heat engines and turbines, 1700-2025 Bold dark line = best-in-class available at any point in time. Coloured lines = individual technologies for context. Newcomen (1712) 0.5 % Watt + condenser (1788) — 2.5 % Cornish engine (1850) — 12 % Diesel engine (1893) — 26 % Combined-cycle turbine (1984) — 4[?] GE 9HA / Siemens HL (2018) — 64 % Legend: Best-in-class envelope; Steam engine; Gasoline ICE; Diesel engine; Gas turbine; Combined cycle gas turbine US horses & mules per 1 000 inhabitants, 1850-2012 USDA Census of Agriculture (horses + mules on farms). Peak ≈ 1915: 26 M equines for 100 M people ≈ 264 per 1 000 Mass production of Ford Model T (1908) and farm tractor (1917) −97 % in 50 years [Photos: horse-drawn plough → old tractor → modern tractor; horse-drawn coach → steam locomotive → high-speed train] [Slide 3] Where are we? 4,609 clinical studies of language models, January 2022 to September 2025 Most agent studies are here. Exam questions 1,704 studies Simulated cases 1,857 studies Real patient data 1,048 studies Randomised trials 19 Nat Med 2026 Chen SF et al. JAMA 2025 Bedi S et al. [Slide 4] We are now in a Transition Period Gare Montparnasse, Paris, 22 October 1895 1895 | Conclusions for Agents today Running late, the express entered the station too fast | Don't deploy ahead of evidence The air brake failed, train broke through the wall | Safeguards needed The guard was busy with paperwork and did not apply the handbrake | Don't overload the human A woman in the street below was killed | Harm to patients? Photo: Studio Lévy & fils, 1895 (public domain)

[Slide 1] 10:15 - 11:45 Proffered paper: NETs and endocrine tumours CHAIRS: JAYDIRA DEL RIVERO, ENRIQUE GRANDE LIBRETTO-531 Overall Survival Overall Survival Probability Selpercatinib Cab or Van 87.1% 66.8% No. Patients | No. Events | Median Follow-up Months (95% CI) Selpercatinib | 193 | 23 | 45.4 (43.3, 47.4) Cab or Van | 98 | 28 | 47.0 (43.7, 50.7) HR (95% CI) 0.338 (0.194, 0.587) P-value <0.0001 Time from randomization (months) No. at Risk With a median follow-up of 45 months, overall survival favored selpercatinib over the cabozantinib or vandetanib group Hazard ratios and 95% confidence intervals were calculated with a Cox proportional-hazards model, with stratification according to trial group (selpercatinib vs. control). Two-sided P values were calculated with a log-rank test, with stratification according to trial group. Tick marks indicate censored data. Patients in the control group were permitted to cross over to selpercatinib following BICR-confirmed disease progression. Jaume Capdevila, MD, PhD MURCIA AUDITORIUM - HALL 3 Jaume Capdevila Updated results of randomized phase 3 study of selpercatinib vs cabozantinib or vandetanib (C/V) in advanced, kinase inhibitor-naïve, RET-mutant medullary thyroid cancer (MTC) Survival table Endpoint: Overall survival Selpercatinib | 193 Cabozantinib or vandetanib | 98 HR: 0.338 HR CI: 0.194, 0.587 P: <0.0001 48 months (dashed line on slide): 87.1% vs 66.8% [Slide 2] 10:15 - 11:45 Proffered paper: NETs and endocrine tumours CHAIRS: JAYDIRA DEL RIVERO, ENRIQUE GRANDE LIBRETTO-531 Investigator-assessed PFS Progression-Free Survival Probability Selpercatinib Cab or Van 67.4% 20.5% No. Patients | No. Events | Median Follow-up Months (95% CI) Selpercatinib | 193 | 53 | 41.8 (38.7, 45.1) Cab or Van | 98 | 58 | 35.9 (19.4, 46.7) HR (95% CI) 0.205 (0.139, 0.304) P-value <0.0001 Time from randomization (months) No. at Risk PFS by investigator assessment remained significantly improved with selpercatinib Hazard ratios and 95% confidence intervals were calculated with a Cox proportional-hazards model, with stratification according to trial group (selpercatinib vs. control). Two-sided P values were calculated with a log-rank test, with stratification according to trial group. Tick marks indicate censored data. Jaume Capdevila, MD, PhD MURCIA AUDITORIUM - HALL 3 Survival table Endpoint: Investigator-assessed PFS Selpercatinib | 193 Cabozantinib or vandetanib | 98 HR: 0.205 HR CI: 0.139, 0.304 P: <0.0001 48 months (dashed line on slide): 67.4% vs 20.5% [Slide 3] 10:15 - 11:45 Proffered paper: NETs and endocrine tumours CHAIRS: JAYDIRA DEL RIVERO, ENRIQUE GRANDE LIBRETTO-531 Investigator-assessed PFS2 Progression-Free Survival Probability Selpercatinib Cab or Van 85.9% 54.6% No. Patients | No. Events | Median Follow-up Months (95% CI) Selpercatinib | 193 | 23 | 43.3 (40.4, 46.5) Cab or Van | 98 | 31 | 42.1 (34.4, 47.1) HR (95% CI) 0.228 (0.131, 0.395) P-value <0.0001 Time from randomization (months) No. at Risk PFS2 by investigator assessment favored selpercatinib over control group Progression-free survival 2 is defined as the time from randomization to disease progression (radiographic or symptomatic progression as determined by the investigator) on the next line of treatment or death from any cause in the absence of observed disease progression. Hazard ratios and 95% confidence intervals were calculated with a Cox proportional-hazards model, with stratification according to trial group (selpercatinib vs. control). Two-sided P values were calculated with a log-rank test, with stratification according to trial group. Tick marks indicate censored data. Jaume Capdevila, MD, PhD MURCIA AUDITORIUM - HALL 3 Survival table Endpoint: Investigator-assessed PFS2 Selpercatinib | 193 Cabozantinib or vandetanib | 98 HR: 0.228 HR CI: 0.131, 0.395 P: <0.0001 48 months (dashed line on slide): 85.9% vs 54.6% [Slide 4] 10:15 - 11:45 Proffered paper: NETs and endocrine tumours CHAIRS: JAYDIRA DEL RIVERO, ENRIQUE GRANDE LIBRETTO-531 Safety TEAEs occurring in ≥20% of patients in either study arm Selpercatinib (n=193) [Grade ≥3 | Grade <3] || Cabozantinib or Vandetanib (n=97) [Grade <3 | Grade ≥3] Hypertension 21% | 29% || 26% | 16% Dry mouth 1% | 38% || 9% | 1% Diarrhoea 6% | 27% || 53% | 12% Alanine aminotransferase increased 11% | 18% || 34% | 3% Aspartate aminotransferase increased 5% | 23% || 37% | 3% Oedema peripheral 1% | 27% || 7% Headache 1% | 25% || 26% Fatigue 4% | 21% || 18% | 5% Constipation 1% | 22% || 15% Erectile dysfunction* 22% Rash 2% | 17% || 21% | 2% Asthenia 1% | 16% || 23% | 5% Hypocalcaemia 2% | 15% || 22% | 7% Nausea 1% | 14% || 31% | 5% Decreased appetite 1% | 14% || 28% | 5% Vomiting 12% || 24% | 2% Weight decreased 1% | 9% || 33% | 6% Mucosal inflammation 1% | 7% || 12% | 12% Stomatitis 5% || 19% | 2% Palmar-plantar erythrodysaesthesia syndrome 4% || 36% | 8% Proteinuria 1% | 5% || 26% Grade ≥3 Grade <3 Grade <3 Grade ≥3 Dose reductions occurred in 52% and 82% of patients treated with selpercatinib and cabozantinib or vandetanib, respectively Subjects may be counted in more than one category. MedDRA Version MedDRA28.1 CTCAE Version CTCAE V5. One control arm patient was excluded from the safety population as they were randomized but did not receive treatment. *Denominator adjusted because gender-specific event for males. N=115 (selpercatinib), N=67 (cabozantinib or vandetanib), N=182 (Total). TEAE, treatment-emergent adverse event. Jaume Capdevila, MD, PhD MURCIA AUDITORIUM - HALL 3

[Slide 1] MEDICAL ONCOLOGY Local therapy complements systemic therapy, it does not replace it. Select patients who respond after 6 to 12 months. [Embedded slide photo] WHO: Selection matters more than the modality Likely to benefit - PS 0-1, staged with PET/CT and MRI brain - ≤5 metastases in ≤3 organs (ESTRO/EORTC); mediastinal nodes not counted - No progression after about 3 months of systemic therapy - Every residual site treatable to ablative dose - Driver-positive disease with durable TKI response (EGFR) - Symptomatic primary, whatever the disease burden Caution or avoid - Early progression or short response to systemic therapy - ILD, poor lung function or prior thoracic RT (high MLD) - Ultracentral disease with anti-VEGF agents or bispecifics - Suspected germline TP53 or other radiosensitivity syndrome - Bulky nodal disease needing large fields with ICI - Patient priorities favour minimal treatment burden #ESMO26 Madrid | Local therapy in advanced lung cancer [Slide 2] RADIATION ONCOLOGY SINDAS showed a survival benefit with SBRT in EGFR+ disease. [Embedded slide photo] SINDAS: Ph III in Synchronous OMD in EGFR+ Metastatic NSCLC with EGFRm; Up to 5 sites of Extracranial Disease; Max 2 in any Organ; No Brain Metastases -> Non-PD Enrol Randomise (1:1) -> First-generation TKI -> PD; TKI & SBRT (25-40 Gy 5# to primary +/- LNs & metastases) -> PD Primary Endpoint: Progression Free Survival First-generation TKI (gefitinib, erlotinib, or icotinib) Wang et al JNCI 2023 Patient Characteristics: TKI only (n = 65) vs TKI + RT (n = 68) [small table; e.g. Number of metastases 1-2: 38 (58.5) vs 32 (47.1); 3-4: 23 (35.4) vs 30 (44.1); 5: 4 (6.2) vs 6 (8.8) - remaining rows small] mPFS 20 vs 12 mths in favour of LCT p=0.001; HR 0.618 (95%CI 0.394-0.969) P<0.001 mOS 25 vs 17 mths in favour of LCT p=0.001; HR 0.682 (95%CI 0.456-1.001) P<0.001 6% Grade 3/4 Pneumonitis in RT & TKI arm; No Grade 5 events #ESMO26 Madrid | Local therapy in advanced lung cancer [Slide 3] RADIATION ONCOLOGY NRG-LU002 showed no PFS or OS benefit with immunotherapy. [Embedded slide photo] NRG-LU002: Local consolidative therapy plus maintenance in advanced NSCLC Phase 2/3 RCT Results - PFS Median Follow-up: All patients - 21.9 mo; Surviving patients - 29.4 mo Of 185/215 patients treated with IO-containing systemic therapy regimens, the PFS HR was 0.90 (95% CI: 0.61, 1.32). + Censored p=0.6638 (logrank) HR (95% CI)=0.93 (0.65, 1.31) No. at risk (0,6,12,18,24,30,36): LCT+ST 134 90 65 52 41 28 17; ST 80 47 33 29 18 14 10 Fail/Total | 1yr PFS Rate (95% CI) | 2yr PFS Rate (95% CI) ST 49/80 | 48.0% (35.9, 59.0) | 35.9% (24.8, 47.2) LCT+ST 89/134 | 51.5% (42.5, 59.8) | 40.1% (31.5, 48.6) Results - OS Median Follow-up: All patients - 21.9 mo; Surviving patients - 29.4 mo Of 185/215 patients treated with IO-containing systemic therapy regimens, the OS HR remained 1.05 (95% CI: 0.68, 1.63). + Censored p=0.8210 (logrank) HR (95% CI)=1.05 (0.70, 1.56) No. at risk: LCT+ST 134 112 97 80 58 40 25; ST 81 65 53 45 31 24 18 Fail/Total | 1yr OS Rate (95% CI) | 2yr OS Rate (95% CI) ST 37/81 | 75.8% (64.0, 84.2) | 58.1% (45.6, 68.8) LCT+ST 70/134 | 76.5% (68.1, 82.9) | 54.1% (44.9, 62.5) Yi-Long Wu @ Guandong Lung Cancer Institute, China Iyengar P, ASCO 2[cut off] #ESMO26 Madrid | Local therapy in advanced lung cancer [Slide 4] SURGERY Feasible and safe in carefully selected oligometastatic NSCLC, with potential survival benefit. [Embedded slide photo] When use surgery in advanced lung cancer? Oligometastatic disease classification Oligometastatic disease state | Definition Synchronous | Limited number of metastases identified within 6 mo of diagnosis Metachronous | Oligometastatic disease manifesting >6 mo after resection or ablation of primary tumor (with primary site controlled) Induced^5 | Polymetastatic disease reduced to an oligometastatic state due to profound response to systemic therapy Oligoprogressive* | Disease progression in only a few sites while on active systemic therapy Oligopersistent | Oligometastatic disease that remains stable or experiences a partial response *These oligoprogressive sites of disease are thought to arise from the emergence of resistant clones no longer responsive to systemic therapy. In carefully selected patients, local treatment of oligometastatic NSCLC with surgical resection or ablative therapies can provide a substantial survival benefit. Isbell JM et al. J Thorac Cardiovasc Surg 2022 [Small KM figure: Proportion Surviving vs Time (Months) 0-60; Replicated from Deboever, N. with permission; Number at risk PFS 52 19 14 12 12 8; OS 52 43 32 27 25 18] Surgery has been shown to be feasible, safe, and a valuable component of local therapy in appropriately selected patients with oligometastatic NSCLC. Antonoff MB et al. J Thorac Cardiovasc Surg 2024 Yi-Long Wu @ Guandong Lung Cancer Institute, China #ESMO26 Madrid | Local therapy in advanced lung cancer

[Slide 2] 7-year PFS by baseline EML4::ALK fusion variant(a) - Lorlatinib showed durable PFS across EML4::ALK variants 1/2 and 3, with improved outcomes vs crizotinib in both subgroups. In patients with variant 2 (n=6), the ORR was 100% and median PFS was NR Table: Lorlatinib | Crizotinib Columns: Lorlatinib Variants 1/2 n=25 | Lorlatinib Variant 3 n=22 | Crizotinib Variants 1/2 n=27 | Crizotinib Variant 3 n=25 PFS, median (95% CI), months: NR (26.0-NR) | 60.0 (33.3-NR) | 7.4 (3.7-9.3) | 7.2 (5.4-9.3) 7-year PFS (95% CI), %: 52.7 (35.7-77.7) | 42.6 (24.1-75.4) | 5.1 (0.8-33.7) | 0 (NA-NA)(b) HR (95% CI), variant 3 vs variants 1/2: Lorlatinib 1.09 (0.463-2.579) | Crizotinib 1.27 (0.697-2.318) Y axis: Probability of PFS; X axis: Months (0-96) No. at risk Lorlatinib (variants 1/2): 25 21 20 17 16 14 12 12 12 12 12 12 12 12 7 3 1 Lorlatinib (variant 3): 22 19 18 17 14 13 12 11 10 8 7 7 6 6 4 2 0 Crizotinib (variants 1/2): 27 15 5 4 2 2 1 1 1 1 1 1 1 1 1 0 0 Crizotinib (variant 3): 25 11 5 2 0 0 0 0 0 0 0 0 0 0 0 0 0 HR, hazard ratio; NR, not reached; ORR, objective response rate; PFS, progression-free survival. (a)The variants 1/2 subgroup combines patients with baseline EML4::ALK fusion variants 1 and 2. (b)No patients remained at risk at year 7. Survival table Endpoint: PFS by baseline EML4::ALK fusion variant (CROWN ctDNA analysis) Arm | N | Median | Median CI Lorlatinib variants 1/2 | 25 | NR | 26.0-NR Lorlatinib variant 3 | 22 | 60.0 | 33.3-NR Crizotinib variants 1/2 | 27 | 7.4 | 3.7-9.3 Crizotinib variant 3 | 25 | 7.2 | 5.4-9.3 Landmark rates: 7 years: Lorlatinib variants 1/2 52.7% (CI 35.7-77.7) | Lorlatinib variant 3 42.6% (CI 24.1-75.4) | Crizotinib variants 1/2 5.1% (CI 0.8-33.7) | Crizotinib variant 3 0% (CI NA-NA) [Slide 3] 7-year PFS by baseline methylation-based TF detection in the lorlatinib group - The Guardant Infinity ctDNA next-generation sequencing assay assesses genomic alterations across 755 genes and uses a non-destructive methylation-based technology to quantify tumor fraction - Baseline methylation-based TF level was prognostic for PFS, with longer PFS in patients with TF ND vs D at baseline(a) Baseline TF Table: ND n=18 | D n=100 PFS, median (95% CI), months: NR (NR-NR) | 80.3 (34.6-NR) HR (95% CI), ND vs D: 0.17 (0.04-0.69) Y axis: Probability of PFS; X axis: Months (0-96) No. at risk ND: 18 17 16 16 16 16 15 7 0 D: 100 73 59 48 45 40 37 22 2 D, detected; HR, hazard ratio; ND, not detected; NR, not reached; PFS, progression-free survival; TF, tumor fraction. (a)TF is defined as the ratio of observed tumor-specific methylation signals to methylation levels across constitutively methylated and unmethylated regions. ND means TF=0. Survival table Endpoint: PFS by baseline methylation-based tumor fraction (lorlatinib group, CROWN) Arm | N | Median | Median CI TF not detected (ND) | 18 | NR | NR-NR TF detected (D) | 100 | 80.3 | 34.6-NR HR: 0.17 HR CI: 0.04-0.69 [Slide 4] Conclusions - With median PFS still not reached after 7 years, lorlatinib delivered unprecedented durability; understanding resistance mechanisms and identifying patients at risk of early progression could further improve outcomes for patients with advanced ALK-positive NSCLC - No new ALK resistance mutations emerged at EOT. Resistance may instead involve bypass and other cancer-related pathways, notably inflammatory-response alterations - Baseline genomic complexity and inflammatory response, may help identify patients at risk of early progression - Lorlatinib showed prolonged PFS vs crizotinib regardless of TP53 mutation status and EML4::ALK variants, including in patients with poor prognostic features (TP53 mutations, variant 3) - In the lorlatinib group, methylation-based TF was prognostic at baseline and on treatment; TF clearance as early as week 4 was associated with longer PFS - Further research is needed to identify patients at risk of early progression and inform earlier therapeutic intervention strategies ALK, anaplastic lymphoma kinase; EOT, end of treatment; NSCLC, non-small cell lung cancer; PFS, progression-free survival; TF, tumor fraction.

[Slide 1] Poster no. 1017 | ESMO 2026 Phase I first-in-human (FIH) results of YL211, a c-MET targeting antibody-drug conjugate (ADC) in patients (pts) with advanced solid tumors Shiraj Sen, Yan Zhang, Yina Wang, Albiruni Ryan Abdul Razak, David Sommerhalder, Cesar A. Perez, Sophia Frentzas, Kanwal Raghav, Manish R. Patel, Rui-Hua Xu, Yuping Li, Jennifer Segar, Scott Laurie, Antonio Jimeno, Meng Yang, Dominik G. Kraus, Linda Mahjoubi, Jiaqiang Cai, Tongtong Xue, Michael Cecchini BACKGROUND - c-MET is a receptor tyrosine kinase that is frequently overexpressed in several solid tumors, including 30-50% of non-small cell lung cancers (NSCLC). - YL211 is an optimized ADC composed of a novel c-MET targeting monoclonal antibody linked to a potent topoisomerase I inhibitor payload via the tumor microenvironment activable linker-payload (TMALIN) platform (Figure 1) - Pre-clinical data showed the linker is stable in the circulation and payload release is efficient within tumors, potentially minimizing off-target toxicity. The novel topoisomerase I inhibitor payload also showed greater cellular potency than deruxtecan (payload used in two approved ADCs in NSCLC). - Here we present safety and preliminary efficacy results from a phase I study (NCT06384352) of YL211 in patients with advanced solid tumors. Figure 1. Schematic representation of YL211, a c-MET targeted ADC Target: c-MET - Frequently overexpressed in NSCLC Payload - Novel topoisomerase 1 inhibitor payload - Eight molecules of cytotoxic payload per antibody Linker - Highly hydrophilic and plasma-stable linker to minimize premature payload release Dual release of payload: TMALIN (Tumor Microenvironment Activable LINker) technology - 1. Intracellular release of payload 2. Extracellular cleavage releasing payload directly in the tumor microenvironment The potential for ADCs based on the TMALIN platform has been demonstrated in two Chinese, phase III studies of Tam-Peli, a B7-H3 targeted ADC METHODS - Adult patients (>=18 years) were required to have advanced solid tumors (c-MET all-comer) refractory to standard therapies, an Eastern Cooperative Oncology Group performance status (ECOG PS) of 0 or 1, and adequate organ and bone marrow function. - In the dose escalation phase, enrolled patients received YL211 at doses of 1.0-3.0 mg/kg, administered intravenously every 3 weeks (Q3W), to establish the maximum tolerated dose (MTD) and recommended dose for expansion (RDE) - During this period, prophylactic G-CSF was restricted, permitted only for high-risk patients at 2.6 mg/kg but allowed as treatment for toxicities or as secondary prophylaxis in subsequent cycles (per local guidelines). - In the backfill cohorts, patients with locally advanced unresectable or metastatic NSCLC (squamous and non-squamous histology; c-MET all-comer) who had received <=2 lines of prior cytotoxic systemic therapy were enrolled and received YL211 at the RDE. - Endpoints can be found in Table 1. Table 1. Study endpoints | Dose escalation phase | Backfill cohorts Primary objectives | Safety, tolerability, MTD, RDE | Safety, efficacy (ORR per RECIST v1.1), RDE Secondary objectives | Pharmacokinetics, anti-drug immune response RESULTS Patient characteristics - At data cut-off (01 July 2026; median follow-up: 7.7 months), 27 patients were enrolled across the dose escalation cohorts and 40 patients in the backfill cohorts. - Baseline patient characteristics can be found in Table 2. Table 2. Patient baseline characteristics n (%) unless stated | Dose escalation phase (n=27) | NSCLC backfill cohorts (n=40) | All patients (N=67) Median age, years (Q1, Q3) | 64 (53, 68) | 67 (59, 75) | 64 (59, 73) Sex Male | 18 (66.7) | 24 (60.0) | 42 (62.7) Sex Female | 9 (33.3) | 16 (40.0) | 25 (37.3) Race Asian | 1 (3.7) | 19 (47.5) | 20 (29.9) Black | 1 (3.7) | 2 (5.0) | 3 (4.5) White | 23 (85.2) | 18 (45.0) | 41 (61.2) Other | 2 (7.4) | 1 (2.5) | 3 (4.5) ECOG PS 0 | 15 (55.6) | 12 (30.0) | 27 (40.3) ECOG PS 1 | 12 (44.4) | 28 (70.0) | 40 (59.7) Prior lines of therapy 1 prior line | 4 (14.8) | 17 (42.5) | 21 (31.3) 2 prior lines | 11 (40.7) | 20 (50.0) | 31 (46.3) >=3 prior lines | 12 (44.4) | 3 (7.5) | 15 (22.5) Tumor type NSCLC | 0 | 40 (100) | 40 (59.7) Squamous | 0 | 11 (27.5) | 11 (16.4) Non-squamous | 0 | 29 (72.5) | 29 (43.3) mCRC | 16 (59.3) | 0 | 16 (23.9) Other | 11 (40.7)* | 0 | 11 (16.4) *Pancreatic ductal adenocarcinoma (PDAC), n=5 (18.5%); head and neck squamous cell carcinoma (HNSCC), n=3 (11.1%); biliary tract cancer (BTC), n=1 (3.7%); advanced gastroesophageal junction adenocarcinoma, n=1 (3.7%); unknown (presumed BTC), n=1 (3.7%). mCRC, metastatic colorectal cancer; Q, quartile. Table 3. Safety summary n (%) | 2.0 mg/kg (n=15) | 2.6 mg/kg (n=40) | Overall* (N=67) All TEAEs | 15 (100) | 38 (95.0) | 65 (97.0) Treatment-related | 14 (93.3) | 38 (95.0) | 64 (95.5) Grade >=3 TEAEs | 10 (66.7) | 26 (65.0) | 43 (64.2) Treatment-related | 9 (60.0) | 22 (55.0) | 36 (53.7) Grade 5 TEAEs | 2 (13.3)† | 0 | 3 (4.5)‡ Treatment-related | 0 | 0 | 1 (1.5)§ Serious TEAEs | 6 (40.0) | 14 (35.0) | 23 (34.3) Treatment-related | 1 (6.7) | 7 (17.5) | 11 (16.4) TEAEs leading to treatment discontinuation | 0 | 3 (7.5) | 6 (9.0) Treatment-related | 0 | 3 (7.5) | 6 (9.0) TEAEs leading to dose reduction | 2 (13.3) | 11 (27.5) | 14 (20.9) Treatment-related | 2 (13.3) | 8 (20.0) | 11 (16.4) *Also includes patients enrolled at 1.0 mg/kg, 2.8 mg/kg, and 3.0 mg/kg. †Neoplasm progression and cerebrovascular accident (n=1, each). ‡Sepsis, neoplasm progression, and cerebrovascular accident (n=1, each). §Sepsis (2.8 mg/kg). TEAE, treatment-emergent adverse event. MTD and RDE in the dose escalation phase - In the dose escalation phase, dose-limiting toxicities (DLTs) were observed in 4/27 patients: - 2.6 mg/kg: Grade 4 neutrophil count decreased (n=1) - 2.8 mg/kg: Grade 4 neutrophil count decreased, Grade 4 febrile neutropenia, Grade 4 platelet count decreased (n=1; all in one patient) - 3.0 mg/kg: Grade 4 infusion-related reaction (n=1) and Grade 3 diarrhea (n=1). - The MTD was identified as 2.6 mg/kg and the RDE were 2.0 and 2.6 mg/kg. Safety and tolerability - In the safety-evaluable population (n=67), TEAEs were reported in 97.0% of patients, with treatment-related adverse (TRAEs) occurring in 95.5% of patients (Table 3). - The most common TRAEs were nausea (59.7%), neutropenia (49.3%), and anemia (44.8%) (Table 4) - Gastrointestinal toxicities (nausea, vomiting, and diarrhea) were mainly low grade - The most common Grade >=3 TRAE was neutropenia (29.9%). - Hematological toxicities were dose-dependent, with the incidence of Grade 3-4 neutropenia and other cytopenias increasing at higher doses - Thrombocytopenia was manageable, with clinical bleeding remaining rare, low grade, and not associated with major hemorrhagic events or coagulopathy - Febrile neutropenia occurred in only one (1.5%) patient (Grade 4) at a dose of 2.8 mg/kg; no cases were reported at the RDE. - One instance of interstitial lung disease (Grade 3) and two instances of pneumonitis (Grade 1 and Grade 2) have been observed to date. - Six (9.0%) patients had TEAEs that resulted in treatment discontinuation. Table 4. Common (>=10% of patients) TRAEs n (%) | Overall (N=67) Any grade | Grade >=3 Nausea | 40 (59.7) | 1 (1.5) Neutropenia* | 33 (49.3) | 20 (29.9) Anemia | 30 (44.8) | 14 (20.9) Decreased appetite | 24 (35.8) | 1 (1.5) Fatigue | 24 (35.8) | 1 (1.5) Leukopenia | 24 (35.8) | 9 (13.4) Alopecia | 22 (32.8) | 0 Vomiting | 20 (29.9) | 2 (3.0) Myalgia | 17 (25.4) | 0 Thrombocytopenia† | 15 (22.4) | 7 (10.4) Lymphopenia | 13 (19.4) | 10 (14.9) Stomatitis | 10 (14.9) | 1 (1.5) Diarrhea | 10 (14.9) | 3 (4.5) Weight decreased | 9 (13.4) | 0 Hypokalemia | 8 (11.9) | 0 Constipation | 7 (10.4) | 0 Hyponatremia | 7 (10.4) | 1 (1.5) *Grouped term includes neutrophil count decreased, neutropenia, and febrile neutropenia. †Grouped term includes platelet count decreased and thrombocytopenia. Clinical activity - Across the intention-to-treat (ITT) population in the dose escalation cohorts (n=27), the unconfirmed ORR (uORR) was 11.1%, confirmed ORR (cORR) was 3.7%, and the disease control rate (DCR) was 74.1% (Figure 2A) - In patients with mCRC (n=16), uORR was 18.8%, cORR was 6.3%, and the DCR was 81.3%. - In the ITT population in the NSCLC backfill cohorts (n=40), uORR was 40.0% (n=16), cORR was 30.0% (n=12), and the DCR was 85.0% (n=34) (Figure 2B) - For 2.0 mg/kg (n=11), the uORR was 27.3%, cORR was 18.2%, and DCR was 63.6% - For 2.6 mg/kg (n=29), the uORR was 44.8%, cORR was 34.5%, and the DCR was 93.1%. Figure 2A. Best change in target lesion size by dose level and best response across all tumor types in the dose escalation phase Figure 2B. Best change in target lesion size by dose level and best response in patients in the NSCLC backfill cohorts Best response: PR | SD | PD; Treatment ongoing PD, progressive disease; PR, partial response; SD, stable disease. CONCLUSIONS The MTD during dose escalation was 2.6 mg/kg based on the observed DLTs. YL211-related toxicity was manageable, well tolerated, and as expected for a c-MET-targeted topoisomerase I inhibitor ADC. Early efficacy was observed across all dose levels and across all solid tumors during dose escalation. In addition, antitumor activity was demonstrated in the backfill cohorts at both tested dose levels in patients with squamous and non-squamous NSCLC. SUMMARY YL211 is an optimized ADC composed of a c-MET targeting monoclonal antibody linked to a potent topoisomerase I inhibitor payload (DAR 8) via the novel TMALIN platform The TMALIN platform provides a highly plasma-stable linker to minimize premature payload release. The dual-cleavage linker allows for internalization-independent payload release directly in the tumor microenvironment, increasing the bystander effect In this FIH phase I study, YL211 was well tolerated and demonstrated promising clinical activity across multiple tumor types These data highlight the therapeutic potential of YL211 and support its ongoing clinical development across solid tumors, especially in NSCLC Acknowledgments: MediLink Therapeutics and F. Hoffmann-La Roche Ltd would like to thank all the patients, their families, the study investigators, and clinical site staff for their involvement in this study. This study was sponsored by MediLink Therapeutics. Third party medical writing assistance, under the direction of the authors, was provided by Rachel Dobb, PhD, and Rick Burgon, MRes, of Ashfield MedComms and was funded by F. Hoffmann-La Roche. [Slide 2] Poster no. 1017 | ESMO 2026 Phase I first-in-human (FIH) results of YL211, a c-MET targeting antibody-drug conjugate (ADC) in patients (pts) with advanced solid tumors Shiraj Sen, Yan Zhang, Yina Wang, Albiruni Ryan Abdul Razak, David Sommerhalder, Cesar A. Perez, Sophia Frentzas, Kanwal Raghav, Manish R. Patel, Rui-Hua Xu, Yuping Li, Jennifer Segar, Scott Laurie, Antonio Jimeno, Meng Yang, Dominik G. Kraus, Linda Mahjoubi, Jiaqiang Cai, Tongtong Xue, Michael Cecchini BACKGROUND - c-MET is a receptor tyrosine kinase that is frequently overexpressed in several solid tumors, including 30-50% of non-small cell lung cancers (NSCLC). - YL211 is an optimized ADC composed of a novel c-MET targeting monoclonal antibody linked to a potent topoisomerase I inhibitor payload via the tumor microenvironment activable linker-payload (TMALIN) platform (Figure 1) - Pre-clinical data showed the linker is stable in the circulation and payload release is efficient within tumors, potentially minimizing off-target toxicity. The novel topoisomerase I inhibitor payload also showed greater cellular potency than deruxtecan (payload used in two approved ADCs in NSCLC). - Here we present safety and preliminary efficacy results from a phase I study (NCT06384352) of YL211 in patients with advanced solid tumors. Figure 1. Schematic representation of YL211, a c-MET targeted ADC Target: c-MET - Frequently overexpressed in NSCLC Payload - Novel topoisomerase 1 inhibitor payload - Eight molecules of cytotoxic payload per antibody Linker - Highly hydrophilic and plasma-stable linker to minimize premature payload release Dual release of payload: TMALIN (Tumor Microenvironment Activable LINker) technology - 1. Intracellular release of payload 2. Extracellular cleavage releasing payload directly in the tumor microenvironment The potential for ADCs based on the TMALIN platform has been demonstrated in two Chinese, phase III studies of Tam-Peli, a B7-H3 targeted ADC METHODS - Adult patients (>=18 years) were required to have advanced solid tumors (c-MET all-comer) refractory to standard therapies, an Eastern Cooperative Oncology Group performance status (ECOG PS) of 0 or 1, and adequate organ and bone marrow function. - In the dose escalation phase, enrolled patients received YL211 at doses of 1.0-3.0 mg/kg, administered intravenously every 3 weeks (Q3W), to establish the maximum tolerated dose (MTD) and recommended dose for expansion (RDE) - During this period, prophylactic G-CSF was restricted, permitted only for high-risk patients at 2.6 mg/kg but allowed as treatment for toxicities or as secondary prophylaxis in subsequent cycles (per local guidelines). - In the backfill cohorts, patients with locally advanced unresectable or metastatic NSCLC (squamous and non-squamous histology; c-MET all-comer) who had received <=2 lines of prior cytotoxic systemic therapy were enrolled and received YL211 at the RDE. - Endpoints can be found in Table 1. Table 1. Study endpoints | Dose escalation phase | Backfill cohorts Primary objectives | Safety, tolerability, MTD, RDE | Safety, efficacy (ORR per RECIST v1.1), RDE Secondary objectives | Pharmacokinetics, anti-drug immune response RESULTS Patient characteristics - At data cut-off (01 July 2026; median follow-up: 7.7 months), 27 patients were enrolled across the dose escalation cohorts and 40 patients in the backfill cohorts. - Baseline patient characteristics can be found in Table 2. Table 2. Patient baseline characteristics n (%) unless stated | Dose escalation phase (n=27) | NSCLC backfill cohorts (n=40) | All patients (N=67) Median age, years (Q1, Q3) | 64 (53, 68) | 67 (59, 75) | 64 (59, 73) Sex Male | 18 (66.7) | 24 (60.0) | 42 (62.7) Sex Female | 9 (33.3) | 16 (40.0) | 25 (37.3) Race Asian | 1 (3.7) | 19 (47.5) | 20 (29.9) Black | 1 (3.7) | 2 (5.0) | 3 (4.5) White | 23 (85.2) | 18 (45.0) | 41 (61.2) Other | 2 (7.4) | 1 (2.5) | 3 (4.5) ECOG PS 0 | 15 (55.6) | 12 (30.0) | 27 (40.3) ECOG PS 1 | 12 (44.4) | 28 (70.0) | 40 (59.7) Prior lines of therapy 1 prior line | 4 (14.8) | 17 (42.5) | 21 (31.3) 2 prior lines | 11 (40.7) | 20 (50.0) | 31 (46.3) >=3 prior lines | 12 (44.4) | 3 (7.5) | 15 (22.5) Tumor type NSCLC | 0 | 40 (100) | 40 (59.7) Squamous | 0 | 11 (27.5) | 11 (16.4) Non-squamous | 0 | 29 (72.5) | 29 (43.3) mCRC | 16 (59.3) | 0 | 16 (23.9) Other | 11 (40.7)* | 0 | 11 (16.4) *Pancreatic ductal adenocarcinoma (PDAC), n=5 (18.5%); head and neck squamous cell carcinoma (HNSCC), n=3 (11.1%); biliary tract cancer (BTC), n=1 (3.7%); advanced gastroesophageal junction adenocarcinoma, n=1 (3.7%); unknown (presumed BTC), n=1 (3.7%). mCRC, metastatic colorectal cancer; Q, quartile. Table 3. Safety summary n (%) | 2.0 mg/kg (n=15) | 2.6 mg/kg (n=40) | Overall* (N=67) All TEAEs | 15 (100) | 38 (95.0) | 65 (97.0) Treatment-related | 14 (93.3) | 38 (95.0) | 64 (95.5) Grade >=3 TEAEs | 10 (66.7) | 26 (65.0) | 43 (64.2) Treatment-related | 9 (60.0) | 22 (55.0) | 36 (53.7) Grade 5 TEAEs | 2 (13.3)† | 0 | 3 (4.5)‡ Treatment-related | 0 | 0 | 1 (1.5)§ Serious TEAEs | 6 (40.0) | 14 (35.0) | 23 (34.3) Treatment-related | 1 (6.7) | 7 (17.5) | 11 (16.4) TEAEs leading to treatment discontinuation | 0 | 3 (7.5) | 6 (9.0) Treatment-related | 0 | 3 (7.5) | 6 (9.0) TEAEs leading to dose reduction | 2 (13.3) | 11 (27.5) | 14 (20.9) Treatment-related | 2 (13.3) | 8 (20.0) | 11 (16.4) *Also includes patients enrolled at 1.0 mg/kg, 2.8 mg/kg, and 3.0 mg/kg. †Neoplasm progression and cerebrovascular accident (n=1, each). ‡Sepsis, neoplasm progression, and cerebrovascular accident (n=1, each). §Sepsis (2.8 mg/kg). TEAE, treatment-emergent adverse event. MTD and RDE in the dose escalation phase - In the dose escalation phase, dose-limiting toxicities (DLTs) were observed in 4/27 patients: - 2.6 mg/kg: Grade 4 neutrophil count decreased (n=1) - 2.8 mg/kg: Grade 4 neutrophil count decreased, Grade 4 febrile neutropenia, Grade 4 platelet count decreased (n=1; all in one patient) - 3.0 mg/kg: Grade 4 infusion-related reaction (n=1) and Grade 3 diarrhea (n=1). - The MTD was identified as 2.6 mg/kg and the RDE were 2.0 and 2.6 mg/kg. Safety and tolerability - In the safety-evaluable population (n=67), TEAEs were reported in 97.0% of patients, with treatment-related adverse (TRAEs) occurring in 95.5% of patients (Table 3). - The most common TRAEs were nausea (59.7%), neutropenia (49.3%), and anemia (44.8%) (Table 4) - Gastrointestinal toxicities (nausea, vomiting, and diarrhea) were mainly low grade - The most common Grade >=3 TRAE was neutropenia (29.9%). - Hematological toxicities were dose-dependent, with the incidence of Grade 3-4 neutropenia and other cytopenias increasing at higher doses - Thrombocytopenia was manageable, with clinical bleeding remaining rare, low grade, and not associated with major hemorrhagic events or coagulopathy - Febrile neutropenia occurred in only one (1.5%) patient (Grade 4) at a dose of 2.8 mg/kg; no cases were reported at the RDE. - One instance of interstitial lung disease (Grade 3) and two instances of pneumonitis (Grade 1 and Grade 2) have been observed to date. - Six (9.0%) patients had TEAEs that resulted in treatment discontinuation. Table 4. Common (>=10% of patients) TRAEs n (%) | Overall (N=67) Any grade | Grade >=3 Nausea | 40 (59.7) | 1 (1.5) Neutropenia* | 33 (49.3) | 20 (29.9) Anemia | 30 (44.8) | 14 (20.9) Decreased appetite | 24 (35.8) | 1 (1.5) Fatigue | 24 (35.8) | 1 (1.5) Leukopenia | 24 (35.8) | 9 (13.4) Alopecia | 22 (32.8) | 0 Vomiting | 20 (29.9) | 2 (3.0) Myalgia | 17 (25.4) | 0 Thrombocytopenia† | 15 (22.4) | 7 (10.4) Lymphopenia | 13 (19.4) | 10 (14.9) Stomatitis | 10 (14.9) | 1 (1.5) Diarrhea | 10 (14.9) | 3 (4.5) Weight decreased | 9 (13.4) | 0 Hypokalemia | 8 (11.9) | 0 Constipation | 7 (10.4) | 0 Hyponatremia | 7 (10.4) | 1 (1.5) *Grouped term includes neutrophil count decreased, neutropenia, and febrile neutropenia. †Grouped term includes platelet count decreased and thrombocytopenia. Clinical activity - Across the intention-to-treat (ITT) population in the dose escalation cohorts (n=27), the unconfirmed ORR (uORR) was 11.1%, confirmed ORR (cORR) was 3.7%, and the disease control rate (DCR) was 74.1% (Figure 2A) - In patients with mCRC (n=16), uORR was 18.8%, cORR was 6.3%, and the DCR was 81.3%. - In the ITT population in the NSCLC backfill cohorts (n=40), uORR was 40.0% (n=16), cORR was 30.0% (n=12), and the DCR was 85.0% (n=34) (Figure 2B) - For 2.0 mg/kg (n=11), the uORR was 27.3%, cORR was 18.2%, and DCR was 63.6% - For 2.6 mg/kg (n=29), the uORR was 44.8%, cORR was 34.5%, and the DCR was 93.1%. Figure 2A. Best change in target lesion size by dose level and best response across all tumor types in the dose escalation phase Figure 2B. Best change in target lesion size by dose level and best response in patients in the NSCLC backfill cohorts Best response: PR | SD | PD; Treatment ongoing PD, progressive disease; PR, partial response; SD, stable disease. CONCLUSIONS The MTD during dose escalation was 2.6 mg/kg based on the observed DLTs. YL211-related toxicity was manageable, well tolerated, and as expected for a c-MET-targeted topoisomerase I inhibitor ADC. Early efficacy was observed across all dose levels and across all solid tumors during dose escalation. In addition, antitumor activity was demonstrated in the backfill cohorts at both tested dose levels in patients with squamous and non-squamous NSCLC. SUMMARY YL211 is an optimized ADC composed of a c-MET targeting monoclonal antibody linked to a potent topoisomerase I inhibitor payload (DAR 8) via the novel TMALIN platform The TMALIN platform provides a highly plasma-stable linker to minimize premature payload release. The dual-cleavage linker allows for internalization-independent payload release directly in the tumor microenvironment, increasing the bystander effect In this FIH phase I study, YL211 was well tolerated and demonstrated promising clinical activity across multiple tumor types These data highlight the therapeutic potential of YL211 and support its ongoing clinical development across solid tumors, especially in NSCLC Acknowledgments: MediLink Therapeutics and F. Hoffmann-La Roche Ltd would like to thank all the patients, their families, the study investigators, and clinical site staff for their involvement in this study. This study was sponsored by MediLink Therapeutics. Third party medical writing assistance, under the direction of the authors, was provided by Rachel Dobb, PhD, and Rick Burgon, MRes, of Ashfield MedComms and was funded by F. Hoffmann-La Roche.

Post-surgical Molecular Residual Disease and Pathological Complete Response after Perioperative Nivolumab in Resectable Stage III NSCLC: Exploratory Analysis from NADIM II FNP: 2566P P. Mediavilla Medel, E. Nadal, C. Aguado de la Rosa, A. Martinez-Marti, R. Bernabe Caro, J. Bosch-Barrera, M. Lázaro Quintela, V. Calvo de Juan, M.A. Insa Mollá, S. Ponce Aix, N. Reguart, J. De Castro, B. Massuti Sureda, I. Macia Vidueira, F. Hernando, L. Romero Vielva, R. Serna-Blasco, A. Romero, M. Provencio Pulla BACKGROUND AND OBJECTIVE Pathological complete response (pCR) is associated with excellent long-term outcomes after neoadjuvant immunochemotherapy in resectable non-small-cell lung cancer (NSCLC). However, patients without pCR (non-pCR) remain heterogeneous, and pathological response alone may not fully define relapse risk. Circulating tumor DNA–based (ctDNA) molecular residual disease (MRD) assessment after surgery may refine postoperative risk stratification. We explored the relationship between pCR, post-surgical MRD, and long-term survival in patients enrolled in NADIM II trial. METHODS Resectable stage IIIA-IIIB NSCLC EXPERIMENTAL ARM: Nivolumab + Paclitaxel + Carboplatin (n=57) -> Nivolumab CONTROL ARM: Paclitaxel + Carboplatin (n=29) -> Observation GUARDANT REVEAL: Somatic mutation; Methylation profile MRD positivo / MRD negativo Pathological Response: pCR 0%; Non-pCR <10%, ≥10% CONCLUSIONS • pCR was uniformly associated with post-surgical MRD negativity and excellent 5-year survival. • Among patients non-pCR, post-surgical MRD status refined prognosis: MRD negativity identified patients with favorable long-term outcomes, whereas detectable MRD defined a high-risk subgroup with early recurrence. • These findings support integrating pathological response and post-surgical MRD assessment to guide future response-adapted strategies, including postoperative de-escalation and intensification. RESULTS Figure 1. ctDNA and MRD status by pCR. Among resected patients with evaluable post-surgical MRD, all patients with pCR were MRD negative. In contrast, among patients without pCR, MRD was detected in 6 of 39 patients (15.4%), whereas 33 of 39 (84.6%) were MRD negative. Post-surgery MRD (n = 58) Figure 2. Kaplan-Meier curves for OS and PFS by pCR among resected patients in both arms. HR: [illegible] Patients with pCR had excellent long-term outcomes, with 5-year OS of 95.7% and 5-year PFS of 91.3%. Figure 3. Kaplan-Meier curves for lung cancer-specific OS and PFS by pCR and post-surgical MRD status. pCR - Not Detected; Non-pCR - Not Detected; Non-pCR - Detected p < 0.001 Among patients non-pCR, those with undetectable MRD had median OS and PFS lung cancer-specific not reached; whereas those with detectable MRD had median OS and PFS lung cancer-specific of 22.3 and 10.4 months, respectively. FUNDING: This study is funded by Bristol-Myers Squibb. Presented at the ESMO Congress 2026; October 23-27, 2026; Madrid, Spain

10:15 - 11:30 Borderline resectable, non-oncogene driven stage III NSCLC CHAIRS: ELENA PRISCIANDARO, NIELS REINMUTH Borderline Resectable Stage 3 NSCLC Radio-Oncology Perspective How many patients with stage 3 NSCLC don't receive surgery after neoadjuvant CT-IO? Is CRT/RT feasible, safe, and effective following neoadjuvant CT-IO? Are there any situations where CRT should be preferred due to the high risk of not receiving any curative-intent local therapy? What are the implications for RO in delivering curative-intent thoracic RT after neoadjuvant CT-IO? CADIZ AUDITORIUM - NCC

BsAbs targeting ICs: Characteristics and properties Bispecific Antigen A Binding Site Antigen B Binding Site -Target two different antigens -Ability to treat more complex diseases -Allows more effective tumor cell killing c Tumour cell CTLA4 PD-1 T cell LAG3 TIGIT TIM3 Antibody CTLA4 PD-L1 T cell Tumour cell > The second IC can be located on the same T cell, tumor cell, or AP cell. > Combination of ICs may improve efficacy, boost immune responses and overcome resistance. > Immune-related toxicities may not overlap. > Can restore activity of TILs and suppress Tregs, preventing thus T-cell exhaustion. Tina Cascone, MD, PhD Goebeler et al. Nat Rev. Clin Oncol 2024

[Slide 1] KANDLELIT-001: An ongoing open-label phase 1 study of calderasib as monotherapy or as part of combination therapy (arms with KRAS G12C-mutated metastatic NSCLC) Key Eligibility Criteria (all arms) - Age >= 18 years - Histologic or blood-based confirmation of KRAS G12C mutation - ECOG PS 0 or 1 - Measurable disease per RECIST v1.1 Arms 1 + 3 (N=271 all solid tumors; n=144 NSCLC): Calderasib Monotherapy 25-800 mg/d PO(a) - Locally advanced unresectable or metastatic solid tumor - >=1 prior systemic therapy for advanced disease Arm 2 (N=136): Calderasib 25-400 mg/d PO(a) + Pembrolizumab(b) - Metastatic NSCLC with PD-L1 TPS >=1% - No indication for EGFR-, ALK-, or ROS1-directed therapy as primary treatment - No prior therapy for metastatic disease Arm 4 (N=46): Calderasib 25-400 mg/d PO(a) + Pembrolizumab(b) + Chemotherapy(c) - Metastatic NSCLC with any TPS(d) - No indication for EGFR-, ALK-, or ROS1-directed therapy as primary treatment - No prior therapy for metastatic disease END POINTS Primary: Incidence of DLTs, AEs, and AEs that led to discontinuation Secondary: ORR and DOR per RECIST v1.1 by investigator assessment; pharmacokinetics Exploratory: PFS per RECIST v1.1 by investigator assessment; KRAS G12C VAF and MSAF in ctDNA (a)Calderasib: PO QD or BID for the total daily dose indicated. (b)Pembrolizumab: 200mg IV Q3W for up to 35 cycles. (c)Chemotherapy: carboplatin area under the concentration-time curve 5 mg/mL/min administered IV Q3W for 4 cycles and pemetrexed 500 mg/m2 IV Q3W. (d)PD-L1 testing was not required for enrollment, but archival or fresh tumor samples were submitted for each participant. MSAF, maximum somatic allele frequency; VAF, variant allele fraction. [Slide 2] Summary of adverse events Treatment-related(a) AEs with incidence >=10% Arms 1+3 Calderasib Monotherapy All Solid Tumors N=271: AST up 12, ALT up 11, Nausea 10, Diarrhea 10 Arm 2 Calderasib + Pembro 1L NSCLC with TPS >=1% N=136: AST up 35, ALT up 32, Pruritus 27, Diarrhea 18, Rash 15, Anemia 14, Nausea 13, ALP up 13 Arm 4 Calderasib + Pembro + Chemo 1L NSCLC; any TPS N=46: Anemia 48, ALT up 43, AST up 43, Nausea 39, Neutrophil down 37, Fatigue 35, WBC count down 30, Platelet count down 24, Appetite down 24, Lymphocyte count down 22, Pruritus 22, Neutropenia 17, Creatinine up 15, Rash 15, Diarrhea 13, Thrombocytopenia 13, GGT up 13, Constipation 13 | Arms 1+3 (N=271) | Arm 2 (N=136) | Arm 4 (N=46) All AEs Any grade AE, n (%) | 251 (93) | 131 (96) | 46 (100) Grade 3 | 82 (30) | 53 (39) | 29 (63) Grade 4 | 5 (2) | 10 (7) | 6 (13) Grade 5 | 11 (4)(b) | 6 (4)(c) | 5 (11)(d) Led to discontinuation of calderasib | 13 (5) | 13 (10) | 7 (15) Led to discontinuation of pembrolizumab | N/A | 38 (28) | 17 (37) Treatment-related(a) AEs Any grade AE, n (%) | 164 (61) | 121 (89) | 45 (98) Grade 3 | 20 (7) | 34 (25) | 27 (59) Grade 4 | 1 (<1) | 5 (4) | 5 (11) Grade 5 | 0 (0)(b) | 1 (1)(c) | 1 (2)(d) Led to discontinuation of calderasib | 2 (1) | 6 (4) | 3 (7) Led to discontinuation of pembrolizumab | N/A | 33 (24) | 14 (30) | Arms 1+3 | Arm 2 | Arm 4 DLT-evaluable population(e) | N=107 | N=34 | N=46 Any DLT, n (%) | 1 (1) | 2 (6) | 5 (11) ALT increased (grade 3) | 0 | 1 (3)(f) | 2 (4)(f) AST increased (grade 3) | 0 | 1 (3)(f) | 1 (2)(f) Colitis (grade 3) | 0 | 0 | 1 (2) Diarrhea (grade 3) | 0 | 0 | 1 (2) ECG QT prolonged (grade 3) | 1 (1) | 0 | 0 Nausea (grade 3) | 0 | 0 | 1 (2) Immune-mediated nephritis (grade 3) | 0 | 1 (3) | 0 (a)Determined by the investigator to be related to treatment. (b)Includes intestinal ischemia (n=2), pneumonia (n=2), and cardiorespiratory arrest, cardiopulmonary failure, cerebrovascular accident, chronic obstructive pulmonary disease, condition aggravated, intestinal obstruction, and pain (n=1 each). None of these were treatment related. (c)Includes myocardial infarction (n=2), and bronchitis viral, cerebral hemorrhage, cerebrovascular accident, myelosuppression, platelet count decreased, and pneumonia fungal (n=1 each). One participant had treatment-related grade 5 myelosuppression and platelet count decreased. (d)Includes acute respiratory distress syndrome, brain injury, immune-mediated myositis, respiratory tract infection, ventricular tachycardia (n=1 each). One participant had treatment-related grade 5 immune-mediated myositis. (e)Per protocol, includes participants in the DLT-evaluation period during dose escalation. (f)One participant in arm 2 and one participant in arm 4 experienced grade 3 increases in both ALT and AST. ALP, alkaline phosphatase; ALT, alanine transaminase; AST, aspartate aminotransferase; GGT, gamma-glutamyltransferase; N/A, not applicable. Data cutoff date: 22 July 2026 [Slide 3] Progression-free survival(a) Participants with NSCLC (Arms 1 + 3, 2, and 4) 12-mo rate: 71% / 66% / 40% 6-mo rate: 86% / 82% / 62% No. at risk (0,5,10,15,20,25,30,35,40,45 mo) Arm 2 | 122 | 95 | 69 | 53 | 40 | 26 | 11 | 7 | 3 | 2 Arm 4 | 46 | 37 | 30 | 21 | 10 | 3 | 0 | 0 | 0 | 0 Arms 1 + 3 NSCLC | 103 | 43 | 15 | 9 | 5 | 3 | 1 | 1 | 0 | 0 | Pts With Event | Median PFS (95% CI) Arm 2 Calderasib + pembro (1L NSCLC with TPS >=1%) | 45/122 (37%) | 28.8 (20.6-NR) Arm 4 Calderasib + pembro + chemo (1L NSCLC; any TPS) | 24/46 (52%) | 17.6 (10.5-26.3) Arms 1 + 3 Calderasib monotherapy (advanced NSCLC) | 49/103 (48%) | 8.3 (6.0-16.4) Participants in Arm 2 by PD-L1 TPS 12-mo rate: 76% / 65% 6-mo rate: 90% / 80% No. at risk (0,5,10,15,20,25,30,35,40,45 mo) TPS >=50% | 64 | 54 | 43 | 36 | 28 | 17 | 8 | 4 | 0 | 0 TPS 1%-49% | 58 | 41 | 26 | 17 | 12 | 9 | 3 | 3 | 3 | 2 | Pts With Event | Median PFS (95% CI) Arm 2 Calderasib + pembro (1L NSCLC with TPS >=50%) | 22/64 (34%) | 34.4 (21.8-NR) Arm 2 Calderasib + pembro (1L NSCLC with TPS 1%-49%) | 23/58 (40%) | 25.1 (10.5-NR) (a)By investigator review per RECIST v1.1; includes all participants with >=1 postbaseline imaging assessment and who were first dosed >=14 weeks before the date of data cutoff. Data cutoff date: 22 July 2026 Survival table Endpoint: PFS (investigator, RECIST v1.1) Arm | N | Median | Median CI Arm 2 Calderasib + pembro (1L NSCLC with TPS >=1%) | 122 | 28.8 | 20.6-NR Arm 4 Calderasib + pembro + chemo (1L NSCLC; any TPS) | 46 | 17.6 | 10.5-26.3 Arms 1 + 3 Calderasib monotherapy (advanced NSCLC) | 103 | 8.3 | 6.0-16.4 Arm 2 Calderasib + pembro (1L NSCLC with TPS >=50%) | 64 | 34.4 | 21.8-NR Arm 2 Calderasib + pembro (1L NSCLC with TPS 1%-49%) | 58 | 25.1 | 10.5-NR Landmark rates: 6 mo: Arm 2 86% | Arm 4 82% | Arms 1 + 3 NSCLC 62% | Arm 2 TPS >=50% 90% | Arm 2 TPS 1%-49% 80% 12 mo: Arm 2 71% | Arm 4 66% | Arms 1 + 3 NSCLC 40% | Arm 2 TPS >=50% 76% | Arm 2 TPS 1%-49% 65% [Slide 4] Best change from baseline in target lesion size(a) Arms 1 + 3: Calderasib Monotherapy | NSCLC Only (N=103) ORR, % (95% CI) | 35 (26-45) Median TTR (range), months | 1.4 (1.2-8.4) Median DOR (range), months | 16.8 (1.2+ to 36.1+) DCR, % (95% CI) | 78 (68-85) Arm 4: Calderasib + Pembro + Chemo | Total (N=46) ORR, % (95% CI) | 65 (50-79) Median TTR (range), months | 1.5 (1.2-18.1) Median DOR (range), months | 19.4 (1.9+ to 27.4+) DCR, % (95% CI) | 89 (76-96) Arms 2: Calderasib + Pembro (TPS >=1%) | Total (N=122) | TPS 1%-49% (n=58) | TPS >=50% (n=64) ORR, % (95% CI) | 71 (62-78) | 55 (42-68) | 84 (73-92) Median TTR (range), months | 1.4 (1.2-10.3) | 1.4 (1.2-4.2) | 1.4 (1.2-10.3) Median DOR (range), months | 33.0 (1.2+ to 45.3+) | 26.2 (1.4+ to 45.3+) | NR (1.2+ to 36.2+) DCR, % (95% CI) | 93 (87-97) | 91 (81-97) | 94 (85-98) (a)By investigator review per RECIST v1.1; waterfall plots include all participants with >=1 postbaseline imaging assessment. ORR and DCR included participants who were first dosed >=14 weeks before the date of data cutoff. TTR and DOR included participants with >=1 confirmed complete or partial response. DCR, disease control rate; DOR, duration of response; NR, not reached; ORR, objective response rate; TTR, time to response. Data cutoff date: 22 July 2026

[Slide 1] The Evolution of Oncology Drug Development: Past, Present, and Future a Past: Drugs -> Which patients respond best? Current and future: Determine molecular profile of the patient's tumour -> Determine which drugs are most appropriate b Past - Cytotoxic chemotherapies: PK, PD, Efficacy, Toxicity MTD Current and future - Molecular targeted therapies: Efficacy, Predictive biomarkers, PD, PK, Toxicity MTD, Intermediate end-point biomarkers, Other molecular biomarkers [illustrations of pills and patient figures; bubble diagrams] Christian Rolfo, MD, PhD, MBA, Dr.hc. [Slide 2] Can ctDNA Integration Overcome the Limitations of RECIST? Hypothesis testing using intermediate end points of clinical response Surrogate Endpoints in Oncology Trials Columns: Endpoint | Events | Measures | Surveillance True clinical endpoint: Overall Survival | Deaths | Death certificates | Patient contact Traditional surrogate endpoint: Disease-free Survival | Deaths Disease Status | Imaging, blood test, BM biopsy, etc. (RECIST, IMWG) | Routine examinations (SOC, standard processes) Novel Surrogate Endpoint: MRD | MRD Status | PSA, MRD, ctDNA, etc. | Study specific Treatment specific Outside of routine examination FDA-AACR Workshop. 2025 Christian Rolfo, MD, PhD, MBA, Dr.hc. [Slide 3] Project Optiums- FDA Traditional Paradigm Pre-clinical -> Early stage -> Maximum Tolerated Dose -> Late stage -> Benefit / Risk Project Optimus - informed Nonclinical Pharmacology -> Dose Escalation -> Dose Optimization -> Dose Confirmation -> Benefit / Risk MIDD & ToE -> Benefit/Risk vs. Standard-of-Care Optimus, prioritizes dose selection based on the totality of evidence rather than just maximum tolerated dose. Gao et al, CPT Pharmacometrics Syst Pharmacol. 2024;13:691–709. Christian Rolfo, MD, PhD, MBA, Dr.hc. [Slide 4] My elevator pitch "The future of early drug development is not just about developing new therapies faster, but about making smarter decisions earlier. Liquid biopsy can turn molecular insights into more efficient trials and better outcomes for our patients." [illustration of two people in an elevator] Christian Rolfo, MD, PhD, MBA, Dr.hc.

FPN: 2575P Phase II Study of Izalontamab Brengitecan (Iza-bren) plus Osimertinib in Patients with EGFR Mutant Locally Advanced or Metastatic Non-Small Cell Lung Cancer (NSCLC) Fei Zhou, Haiyong Wang, Qiming Wang, Qitao Yu, Ruiling Ning, Xingya Li, Yan Yu, Yinghua Ji, Jianhua Shi, Longhua Sun, Bo Jiang, Bihui Li, Zhiyong He, Dongqing Lv, Guifang Zhang, Sa Xiao, Hai Zhu, Yi Zhu, Caicun Zhou Background • Izalontamab brengitecan (iza-bren, BL-B01D1) is a first-in-class ADC comprised of an EGFR x HER3 bispecific antibody conjugated to a novel topo-I inhibitor payload (Ed-04) via a stable tetrapeptide-based cleavable linker. • Previous results of this study showed iza-bren plus osimertinib (osi) had shown promising early clinical activity in first-line locally advanced or metastatic EGFR mutant (EGFRmt) NSCLC patients (WCLC 2025) • Clinical trial information: NCT05880706. αEGFR (Human EGFR Affinity: High); DAR = 8 (Cat B cleavable linker, Ed-04 (TOPI inhibitor)); wt Fc IgG1; αHER3 (Human EGFR Affinity: Low) Here we present the updated safety/efficacy results from a phase II study evaluating iza-bren in combination with osimertinib in first-line locally advanced or metastatic EGFRmt NSCLC patients after about 2 years of follow-up. Objectives Primary Objectives • To assess the safety and tolerability of iza-bren monotherapy, and iza-bren plus osimertinib combination therapy in locally advanced or metastatic EGFRmt NSCLC patients and further explore the optimal combination dose. • To evaluate the efficacy of iza-bren monotherapy, and iza-bren plus osimertinib combination therapy in locally advanced or metastatic EGFRmt NSCLC patients. Secondary Objectives • To characterize the pharmacokinetics (PK) of the investigational drug iza-bren. • To assess the immunogenicity of iza-bren. • To explore potential drug–drug interactions (DDI). Methods This non-randomized study includes three cohorts. • Patients who progressed on prior EGFR-TKI therapy were enrolled in Cohort A and received iza-bren at 2.5 or 2.75 mg/kg D1D8 Q3W. • Treatment-naïve patients were enrolled in Cohort B or Cohort C and received iza-bren in combination with osimertinib (80 mg QD). In Cohort B, iza-bren was given at 2.2, 2.5, or 2.75 mg/kg D1D8 Q3W; in Cohort C, iza-bren was given at 4.0 or 4.5 mg/kg on D1 Q3W. Endpoints • The primary endpoints include objective response rate (ORR) and recommended phase 2 dose (RP2D) (for combination treatment). • The secondary endpoints include progression-free survival (PFS), disease control rate (DCR), duration of response (DOR), and safety. • The exploratory endpoints include overall survival (OS), PK, immunogenicity, and DDI. Study Design Key Eligibility Criteria • Locally advanced or metastatic NSCLC • With EGFR-sensitizing mutations • ECOG PS 0-1 • Measurable disease per RECIST v1.1 • Treatment-naïve Treatment (Cohort B + Cohort C): D1D8 Q3W + 80 mg osimertinib QD: 2.2 mg/kg (N=41); 2.5 mg/kg (N=40); 2.75 mg/kg (N=42). D1 Q3W + 80 mg osimertinib QD: 4.0 mg/kg (N=18); 4.5 mg/kg (N=13) Here we present results from Cohort B and Cohort C. Endpoints • Primary: RP2D, ORR • Secondary: PFS, DCR, DOR, Safety • Exploratory: PK, Nab, DDI, OS, Biomarker Enrollment • As of September 21, 2026, 154 treatment-naive patients with EGFRmt NSCLC were enrolled in Cohort B and Cohort C and treated with iza-bren plus osimertinib. Only cohort B data were presented in this analysis. (Table 1). Table 1. Patient Characteristics (D1D8 Q3W: Total (N=123) | 2.2 mg/kg (N=41) | 2.5 mg/kg (N=40) | 2.75 mg/kg (N=42)) Median (Q1, Q3) age, years 60.0 (54.0, 66.0) | 59.0 (55.0, 66.0) | 59.0 (53.5, 64.5) | 60.0 (55.0, 66.0) Male, n (%) 55 (44.7) | 19 (46.3) | 13 (32.5) | 23 (54.8) Current or former smoker, n (%) 36 (29.3) | 15 (36.6) | 8 (20.0) | 13 (31.0) Median (Q1, Q3) BMI, kg/m² 23.1 (20.4, 25.9) | 23.8 (20.3, 25.9) | 22.4 (20.1, 25.3) | 23.3 (20.8, 25.7) Median (Q1, Q3) baseline SOD, mm 52.6 (34.4, 85.0) | 59.0 (34.7, 75.0) | 48.7 (33.8, 87.8) | 52.0 (36.0, 81.0) ECOG-PS Score of 1, n (%) 95 (77.2) | 33 (80.5) | 32 (80.0) | 30 (71.4) Median (Q1, Q3) # of metastatic organs 3.0 (2.0, 4.0) | 2.0 (2.0, 4.0) | 3.0 (1.5, 4.0) | 3.0 (2.0, 4.0) Baseline brain metastasis, n (%) 36 (29.3) | 16 (39.0) | 8 (20.0) | 12 (28.6) EGFR exon19del mutation, n (%) 72 (58.5) | 23 (56.1) | 26 (65.0) | 23 (54.8) EGFR L858R mutation, n (%) 51 (41.5) | 18 (43.9) | 14 (35.0) | 19 (45.2) #: Asymptomatic and stable brain metastases. Safety • The median treatment duration for all D1D8Q3W patients was 18.2 months, 26.9 months for patients in 2.5mg/kg dose level. • The most frequent hematologic TRAEs (all Grade) were anemia (95.1%), leukopenia (91.1%), neutropenia (91.1%), and thrombocytopenia (79.7%); the most frequent non-hematologic TRAEs were nausea (72.4%) and stomatitis (69.1%). • Grade 3 and above TRAEs, predominantly hematologic in nature, were effectively managed with standard supportive measures, including dose reductions and growth factor support. • TRAE leading to treatment discontinuation rate was 14.6% in D1D8 Q3W regimen, and 12.5% in 2.5mg/kg dose level. • One death due to unknown cause, possibly related iza-bren in 2.5mg/kg, was reported. Two cases of ILD were reported (1.3%; one Grade 2, one Grade 3 in 2.2mg/kg dose level only). • No new safety signals were identified. Table 2. Treatment-related adverse events (>30% in D1D8Q3W cohort) (D1D8 Q3W (N=123) All grade | Grade≥3; 2.5 mg/kg D1D8 Q3W (N=40) All grade | Grade≥3) Any treatment-related AE 123 (100) | 102 (82.9) | 40 (100) | 33 (82.5) Anaemia 117 (95.1) | 24 (19.5) | 36 (90.0) | 10 (25.0) Leukopenia 112 (91.1) | 39 (31.7) | 36 (90.0) | 18 (45.0) Neutropenia 112 (91.1) | 63 (51.2) | 36 (90.0) | 26 (65.0) Thrombocytopenia 98 (79.7) | 34 (27.6) | 32 (80.0) | 12 (30.0) Nausea 89 (72.4) | 11 (8.9) | 31 (77.5) | 1 (2.5) Stomatitis 85 (69.1) | 21 (17.1) | 27 (67.5) | 6 (15.0) Weight decreased 70 (56.9) | 5 (4.1) | 29 (72.5) | 2 (5.0) Decreased appetite 69 (56.1) | 3 (2.4) | 24 (60.0) | 2 (5.0) Asthenia 68 (55.3) | 11 (8.9) | 20 (50.0) | 3 (7.5) Diarrhoea 63 (51.2) | 3 (2.4) | 22 (55.0) | 2 (5.0) Aspartate aminotransferase increased 62 (50.4) | 1 (0.8) | 17 (42.5) | 0 Vomiting 61 (49.6) | 7 (5.7) | 16 (40.0) | 1 (2.5) Alanine aminotransferase increased 60 (48.8) | 1 (0.8) | 18 (45.0) | 0 Hypokalaemia 49 (39.8) | 10 (8.1) | 16 (40.0) | 4 (10.0) Rash 47 (38.2) | 1 (0.8) | 20 (50.0) | 1 (2.5) Hypoalbuminaemia 46 (37.4) | 0 (0.0) | 13 (32.5) | 0 Alopecia 38 (30.9) | 0 (0.0) | 12 (30.0) | 0 Efficacy • Efficacy analysis was conducted based on all patients who received at least one dose of study drug. The median follow-up was 29.2 mo. • In the cohort of iza-bren 2.5 mg/kg D1D8Q3W, ORR was 100%, cORR was 97.5%, median DOR and PFS were not reached. The 24-mo PFS rate was 87.0%, while 30-mo PFS rate was not reached. The median (Q1, Q3) duration of iza-bren treatment was 26.9 (7.5, 27.9) mo. Table 3. Summary of Efficacy Results (D1D8 Q3W: Total (N=123) | 2.2 mg/kg (N=41) | 2.5 mg/kg (N=40) | 2.75 mg/kg (N=42)) Best overall response, n (%): PR 107 (87.0) | 33 (80.5) | 40 (100) | 34 (81.0); Confirmed 104 (84.6) | 32 (78.0) | 39 (97.5) | 33 (78.6); PR pending confirmation 0 | 0 | 0 | 0; SD 12 (9.8) | 6 (14.6) | 0 | 6 (14.3); PD 0 | 0 | 0 | 0; NE 4 (3.3) | 2 (4.9) | 0 | 2 (4.8) ORR, % (95% CI) 87.0 (79.7, 92.4) | 80.5 (65.1, 91.2) | 100 (91.2, 100.0) | 81.0 (65.9, 91.4) Confirmed ORR, % (95% CI) 84.6 (76.9, 90.4) | 78.0 (62.4, 89.4) | 97.5 (86.8, 99.9) | 78.6 (63.2, 89.7) DCR, % (95% CI) 96.7 (91.9, 99.1) | 95.1 (83.5, 99.4) | 100 (91.2, 100.0) | 95.2 (83.8, 99.4) Median DoR, mo, (95% CI) 30.6 (30.6, NR) | NR (14.1, NR) | NR (NR, NR) | 30.6 (23.8, NR) Median PFS, mo, (95% CI) 32.0 (31.5, NR) | NR (21.1, NR) | NR (NR, NR) | 31.8 (25.1, NR) Median PFS (exponential model simulation), mo 51.1 | 40.6 | 108.1 | 38.9 Median PFS (weibull model simulation), mo 40.7 | 36.2 | 66.6 | 33.2 24-month PFS rate, % (95% CI) 75.5 (66.3, 82.5) | 65.3 (46.9, 78.6) | 87.0 (71.6, 94.4) | 73.0 (55.5, 84.5) 27-month PFS rate, % (95% CI) 67.7 (57.3, 76.1) | 62.2 (43.8, 76.1) | 83.8 (67.3, 92.4) | 59.5 (40.8, 74.0) 30-month PFS rate, % (95% CI) 67.7 (57.3, 76.1) | 62.2 (43.8, 76.1) | NR (NR, NR) | 59.5 (40.8, 74.0) Median OS, mo, (95% CI) NR (NR, NR) | NR (28.9, NR) | NR (NR, NR) | NR (NR, NR) 24-month OS rate, % (95% CI) 86.6 (79.1, 91.6) | 76.7 (60.0, 87.1) | 95.0 (81.5, 98.7) | 87.9 (73.3, 94.8) 27-month OS rate, % (95% CI) 81.0 (72.5, 87.1) | 68.8 (51.6, 80.9) | 91.5 (75.4, 97.2) | 82.3 (66.4, 91.2) 30-month OS rate, % (95% CI) 79.4 (70.4, 85.9) | 65.8 (48.3, 78.5) | NR (NR, NR) | 82.3 (66.4, 91.2) Figure 1. Progression Free Survival (2.5mg/kg D1D8 Q3W) Figure 2. Overall Survival (2.5mg/kg D1D8 Q3W) Conclusions • With updated 2-year follow-up, iza-bren plus osimertinib continued to demonstrate promising and durable anti-tumor activity with a manageable safety profile in first-line EGFRmt NSCLC patients, supporting further development of the combination in this population. • The 2.5 mg/kg D1D8 Q3W was chosen as the RP3D for combination treatment. • Phase III study is ongoing in China (NCT06838273). Global registrational study in firs line setting is ongoing (BMS IZABREN-LUNG02, NCT 07680790) Declaration of interest: Prof. Fei Zhou has no interest to declare.

2625P Early Clinical Evidence of VRN110755: A Novel Brain-Penetrant EGFR TKI Active Across Common, Uncommon, and Resistance EGFR Mutations (REACH-EGFR) T.-Y. Yang, M.-J. Ahn, K.H. Lee, M.H. Hong, H.R. Kim, C.-C. Lin, B.Y. Shim, C.H. Chiu, Y.J. Kim, S. Sukumaran, J.-Y. Hung, Y.-H. Luo, S. Arulananda, M. Li, D. Yu, S.Y. Zhang VORONOI Introduction Abstract • VRN110755 (VRN11) is an orally available, highly selective EGFR tyrosine kinase inhibitor (TKI) designed to target a broad spectrum of EGFR mutations, including C797S. • Based on safety and clinical activity, 320 mg was selected as the lower dose for randomized dose optimization against a higher dose (TBD). • As of the October 9, 2026 data cut-off, safety is reported for all treated patients, focusing on the ≥ 160 mg subset, the lowest exposure attainable at 320 mg after two permitted reductions. REACH-EGFR: Ongoing Phase Ia study • Primary Endpoints: DLTs · AEs / SAEs • Secondary Endpoints: PK · ctDNA · ORR · DoR · DCR · PFS · intracranial ORR Dose levels: 10 mg, 20 mg, 40 mg, 80 mg, 160 mg, 240 mg, 320 mg, 400 mg, 480 mg, 560 mg (+ Backfill), 640 mg (On-going) Potent Anti-Tumor Activity starting from 40 mg (-51% Baseline to C5D1; Tumor disappearance C2D1 to C5D1) Demographics (Enrolled patients 10-560 mg, N=82*) Age, years: Median (range) 61 (45–87); <65 / ≥65 49 (59.8) / 33 (40.2) Sex: Female / Male 53 (64.6) / 29 (35.4) Race: Asian 80 (97.6); White 1 (1.2); Others 1 (1.2) ECOG PS 0 / 1 28 (34.1) / 54 (65.9) EGFR mutation at baseline: Del19 33 (40.2); L858R 24 (29.3); T790M 9 (11.0); C797X 15 (18.3); Uncommon EGFR 10 (12.2); EGFR amplification 8 (9.8); EGFR not detected / Not reported 6 (7.3) / 13 (15.9) Brain metastases at baseline: Yes 39 (47.6); Measurable / Non-measurable 13 (15.9) / 26 (31.7); No 43 (52.4) Prior systemic regimens: Median (range) 3 (1–10); 1: 17 (20.7); 2: 18 (22.0); ≥3: 46 (56.1); Not reported 1 (1.2) Safety and Long-term tolerability Treatment-related adverse events (Patients 10-560 mg, N=82): G1 | G2 | ≥G3 | Total Any 28 (34) | 20 (24) | 2 (2) | 50 (61) EGFR-related: Diarrhea 11 (13) | 3 (4) | 0 (0) | 14 (17); Pruritus 9 (11) | 4 (5) | 0 (0) | 13 (16); Rash maculo-papular 9 (11) | 2 (2) | 0 (0) | 11 (13); Rash 5 (6) | 4 (5) | 0 (0) | 9 (11); Dermatitis acneiform 5 (6) | 0 (0) | 0 (0) | 5 (6); Dry skin 4 (5) | 1 (1) | 0 (0) | 5 (6); Paronychia 3 (4) | 2 (2) | 0 (0) | 5 (6); Stomatitis 5 (6) | 0 (0) | 0 (0) | 5 (6) Non-EGFR-related: Nausea 9 (11) | 1 (1) | 0 (0) | 10 (12); AST increased 7 (9) | 1 (1) | 0 (0) | 8 (10); ALT increased 7 (9) | 0 (0) | 0 (0) | 7 (9) Special interests: Lung-related 0 (0) | 0 (0) | 1 (1) | 1 (1); QT prolongation 3 (4) | 0 (0) | 0 (0) | 3 (4) • Most TRAEs were mild to moderate with known EGFR-related AEs (diarrhea and skin); no DLT during dose escalation, and no drug-related discontinuation was reported. • Total 90% of treatment days were TRAE-free or G1 only, with no accumulation of high grade TRAEs in patients at ≥ 160 mg. Preclinical Evidences Target Engagement Ratio (TER): Systemic TER (VRN11 320mg | Osimertinib 80mg | VRN11 ÷ Osimertinib): Common 56 | 14 | ~4x; + C797S 22 | <1 | 140x; + uncommon 20 | 9 | 2x; Uncommon 5 | 2 | 3x. CNS TER: Common 112 | 3 | ~40x Brain Permeability in Human: Kp,uu,CSF VRN11 | Osimertinib: Mouse 2.3 | 0.8; Monkey 2.7 | NA; Human 2.0 | 0.2 • Estimated target engagement at 320 mg exceeded osimertinib 80 mg by ~4-fold for common and 2–3-fold for uncommon EGFR mutations, and ~40-fold in the CNS. Intracranial activity: PC-9 Luc Intracranial Model • VRN11 10 mg/kg outperformed osimertinib 25 mg/kg • Continued decline through day 28, while osimertinib plateaued Clinical Results Efficacy (1) Intracranial activity; C797S-negative, ≥ 160 mg (n=20) 6-mo iPFS 79.8%; 12-mo iPFS 49.9%; median iPFS: 9.1 mo (95% CI, 3.7-NR); median Follow-up (range): 13.2 mo (1.2-19.4) No. at risk 20 12 9 7 5 1 1 1 1 0 Prior Lines, median (range) 3 (1-9); Prior 3G EGFR TKI 17 (85.0%) Intracranial Best overall response, n (%): CR 4 (20.0); PR 0 (0.0); SD 5 (25.0); Non-CR/Non-PD 10 (50.0); PD 1 (5.0); NE 0 (0.0); Confirmed iORR 4 (20.0); iDCR 19 (95.0) In T790M-negative patients treated with osimertinib 80 mg, median iPFS was 3.5 months and the 6-mo iPFS rate was 31%. Case Report: intracranial CR within 4 weeks • 71-year-old female · 400 mg · ECOG 1 · 3 prior lines including osimertinib and lazertinib • Genomic alteration at entry: EGFR L858R, PIK3CA, two TP53 • Week 4: Intracranial CR and ~95% decrease in ctDNA VAF • Week 8: Confirmed systemic PR Efficacy (2) C797S-positive, ≥ 160 mg (n=8) 6-mo PFS 66.7%; median PFS: 11.0 mo (95% CI, 4.7-11.0); median Follow-up (range): 10.8 mo (1.2-15.5) No. at risk 8 6 6 4 3 1 0 ORR 87.5% (7/8) Speed of response: VRN11 ≥ 160 mg (C797S-positive) Depth of Response -45.9%, Time to Best Response 26 days; Osimertinib 80 mg (T790M-positive) -43.7%, 54 days • Driven by strong target engagement, deep response reached earlier with VRN11 than with osimertinib (26 days vs 54 days) in a later line setting (C797S-positive vs T790M-positive). Case Report: C797S after osimertinib +savolitinib • 63-year-old male · 480 mg · ECOG 0 · 6 prior lines • Resistance shifted from MET amp to C797S • EGFR VAF (entry): L858R 25.74%, C797S 25.36%, L718V 1.75%, L792V 0.42% • All 4 target lesions decreased: 2 lung, 2 lymph node (-43.8% Baseline to C2D1) Ongoing Phase 1/2 Design Efficacy (3) C797S-negative, ≥ 320 mg (n=17) 6-mo PFS 52.5%; median PFS: 7.3 mo (95% CI, 1.9-NR); median Follow-up (range): 7.2 mo (1.2-14.6) No. at risk 17 8 6 3 2 1 0 Prior lines, median (range) 3 (1–9); Prior lines ≥3 10 (58.8%); Prior 3G EGFR TKI 14 (82.4%); Prior 1/2G EGFR TKI 11 (64.7%); Prior chemotherapy 13 (76.5%) • For reference, osimertinib in T790M-negative disease showed 2.8 months (AURA phase 1; dose-ranging 20–240 mg, after prior EGFR TKI) and 4.1 months (WJOG12819L; 80 mg, after 1st/2nd-generation EGFR TKI and platinum); cross-study comparisons, different populations—both references enrolled after 1st/2nd-generation TKI only. Dose selection: Low dose 320mg; High dose TBD • Clinical activity was observed from ≥ 160 mg across all three settings — intracranial disease, C797S-positive disease, and C797S-negative disease after prior TKI. • 320 mg was selected as the low dose because two permitted reductions from it still land at 160 mg, keeping every patient within the active range; no DLT occurred during dose escalation, so a higher dose is being evaluated against it. Expansion cohorts: 1L common EGFR Treatment-naïve Enrolling Cohort B; 1L atypical EGFR Treatment-naïve Enrolling Cohort D; C797X Common + C797X Enrolling Cohort A; CNS metastases Lesion ≥ 5 mm and/or LM Cohort E; Combination Post-3G TKI; BOIN design Cohort G Conclusions • Well tolerated: TRAEs mostly G1–2 (any 61%, ≥ G3 2%); no DLT or treatment-related discontinuation. • C797S+, ≥ 160 mg: 7/8 with > 40% target lesion reduction; median time to best response 26 days (osimertinib, T790M+: 54 days); mPFS 11.0 mo. • C797S−, CNS lesions, ≥ 160 mg (n=20; 85% post-3G TKI): 6-/12-mo iPFS 79.8%/49.9% (osimertinib, T790M−, post-1-2G TKI: 31% at 6 mo). • C797S−, ≥ 320 mg (n=17; 82% post-3G TKI): mPFS 7.3 mo (osimertinib, post-1-2G TKI: 2.8–4.1 mo). • Activity, observed almost entirely after 3G TKI progression, is consistent with estimated target engagement ~4× (systemic) and ~40× (CNS; Kp,uu,CSF 2.0 vs 0.22) that of osimertinib 80 mg. • Cohorts B and D open the first line. ESMO Congress 2026, October 23 – 27 2026, Madrid, Spain Copyright ©2026 Voronoi, Inc. All rights reserved

LBA65 - Surgical conversion and survival outcomes after induction chemoimmunotherapy in potentially resectable or unresectable stage III NSCLC: Preliminary results from the prospective phase II study (CTONG 1804 part 3) Speakers Yi Pan (Guangzhou, China) Lecture Time 10:40 - 10:45 Authors Yi Pan (Guangzhou, China), Li-Bo Tang (Guangzhou, China), Xuening Yang (Guangzhou, China), Hui Lin (Guangzhou, China), Hong-Hong Yan (Guangzhou, China), Yi-Chen Zhang (Guangzhou, China), Ziyi Xu (Guangzhou, China), Zhiyong Chen (Guangzhou, China), Eileen Ke (Guangzhou, China), Chang Lu (Guangzhou, China), Rui Fu (Guangzhou, China), Haiyu Zhou (Guangzhou, China), Xiao S. Ben (Guangzhou, China), Jingjing Chen (Guangzhou, China), Hua-Jen Chen (Guangzhou, China), Riqiang Liao (Guangzhou, China), Song Dong (Guangzhou, China), Wenzhao Zhong (Guangzhou, China), Qing Zhou (Guangzhou, China), Yi-Long Wu (Guangzhou, China) Abstract Background Because 25–30% of Patients (Pts) with unresectable (UR) stage III NSCLC cannot receive consolidation immunotherapy (IO) after chemoradiotherapy (CRT), surgical conversion after induction chemoimmunotherapy has been explored. However, whether surgical conversion improves survival remains unclear. Methods CTONG 1804 Part 3 is a prospective, single-arm phase II study evaluating surgical conversion after induction chemoimmunotherapy in stage III NSCLC. Eligible Pts were treatment naïve, EGFR/ALK wild-type, and MDT-assessed as potentially resectable or UR. Pts received ≥3 cycles of PD-1/PD-L1 inhibitor plus platinum-doublet chemotherapy, followed by MDT reassessment. Resectable Pts underwent surgery and UR Pts received definitive CRT, followed by IO for up to 1 year. Dual primary endpoints were surgical resection rate and 18-month (mo) EFS in surgical Pts. Secondary endpoints included pathological response, OS, EFS. Results Between June 2022 and August 2025, 215 Pts were enrolled (median age, 63 years; 89.8% male; stage IIIA/IIIB/IIIC, 19.5%/54.5%/26.0%). MDT classified 196 Pts (91.2%) as UR and 19 (8.8%) as potentially resectable; 155 had EORTC-defined stage IIIB–IIIC UR-NSCLC. At 31.3-mo median follow-up, R0 resection was achieved in 98.1% of surgical Pts, with 84.8% undergoing minimally invasive surgery. Outcomes are summarized in the table. Table: LBA65 Outcomes in the overall population and stage IIIB–IIIC UR-NSCLC Outcomes | Overall n=215: Surgery | CRT | Consent withdrawal* | Progression during induction | Death during induction † | Overall | UR stage IIIB-IIIC n=155‡: Surgery | CRT Pts, n(%) | 105(48.8) | 60(27.9) | 27(12.6) | 18(8.4) | 5(2.3) | 215(100) | 73(47.1) | 41(26.5) ORR, % | 82.9 | 68.3 | 51.9 | 0 | 40.0 | 67.9 | 79.5 | 70.7 pCR, % | 48.6 | — | — | — | — | — | 42.5 | — MPR, % | 64.8 | — | — | — | — | — | 63.0 | — 12-mo EFS(%) | 84.6 | 66.7 | 69.8 | — | — | 68.6 | 82.0 | 68.3 Median EFS, mo | NR | 20.5 | 15.2 | 2.4 | 4.3 | 29.2 | NR | 21.0 Pattern in Pts with progression, n (%) Total | 23 | 30 | 15 | 18 | — | 86 | 19 | 21 Local | 9(39.1) | 16(53.3) | 8(53.3) | 9(50.0) | — | 42(48.8) | 8(42.1) | 10(47.6) Distant | 10(43.5) | 8(26.7) | 2(13.3) | 4(22.2) | — | 24(27.9) | 9(47.4) | 7(33.3) Both | 4(17.4) | 6(20.0) | — | 4(22.2) | — | 14(16.3) | 2(10.5) | 4(19.0) Unknown | — | — | 5 (33.3) | 1(5.6) | — | 6(7.0) | — | — *No local therapy. †Deaths: 2 treatment-related pneumonitis; 3 unrelated (COVID-19, n=2; cholecystitis, n=1). ‡41/155 Pts (26.5%) received no local therapy. NR, not reached. Conclusions Induction chemoimmunotherapy enabled surgical conversion with encouraging pathological response and survival in potentially resectable or stage III UR-NSCLC. Durable disease control was also observed in selected Pts without local therapy.

Efficacy by Histology • The efficacy spider plot demonstrates that the majority of patients exhibited target lesion shrinkage at their first evaluation, and maintained tumor control across subsequent assessments. • Patients across different histology types all demonstrated a PFS benefit. Median PFS was 12.42 (6.67-NE) monthes, with 7.72 months in patients with squamous NSCLC and was not reached in patients with non-squamous NSCLC. Figure 3. Depth and durability of response by histology [Spider plot] Histology: Squamous (N=50); Non-squamous (N=45). y-axis: Change in target-lesion sum from baseline (%); x-axis: Time from first dose (months) Figure 4. Progression-free survival by histology Median PFS (months) Population | N | mPFS | 95% CI All-comer | 98 | 12.42 | 6.67–NE Squamous | 52 | 7.72 | 5.26–NE Non-squamous | 46 | NR | 7.89–NE Number at risk (0, 3, 6, 9, 12, 15 months) All-comer: 98 68 35 21 8 0 Squamous: 52 38 15 10 3 0 Non-squamous: 46 30 20 11 5 0 Date cutoff: Sep 22, 2026 Survival table Endpoint: PFS by histology (BAT8008 + BAT1308, 1L NSCLC, single arm) Arm | N | Median | Median CI All-comer | 98 | 12.42 months | 6.67-NE Squamous | 52 | 7.72 months | 5.26-NE Non-squamous | 46 | NR | 7.89-NE

Young Women's Needs and Preferences during the Lung Cancer Journey in the UK: the HER-LUNG Study ESMO 2026 FPN: 3504eP BACKGROUND METHODS 15 women 4 MSPs 18–55 years Semi-structured interviews with 15 women diagnosed with lung cancer and 4 main support persons (1 male, 3 female). Data were analysed using codebook thematic analysis. KEY MESSAGE Recognition, life-stage burden and self-advocacy were interconnected. Younger women and their families often carried substantial practical, emotional and navigational work across the lung cancer journey. RESULTS — THREE INTERCONNECTED THEMES 1 Delayed Recognition and Response 2 Burden of Holding Life Together 3 The Value of Self-Advocacy "The first GP said, 'What do you want me to do about it?' Not very helpful." "I ran the house... now have to ask for help and I hate asking." CONCLUSIONS & CLINICAL IMPLICATIONS Earlier recognition Communication Biomarkers Psychosocial care Equity

PFS improved with osi + plat–pem versus osi mono irrespective of Week 3 and 6 plasma EGFRm clearance status Plasma EGFRm clearance at Week 3 — No. at risk (0,6,12,18,24,30,36 mo): Osi + plat–pem 98 79 69 53 29 12 0; Osi mono 100 90 65 37 19 10 0 No Plasma EGFRm clearance at Week 3 — No. at risk: Osi + plat–pem 21 16 12 9 4 2 0; Osi mono 40 22 18 10 6 3 0 Week 3 | Clearance | No clearance mPFS (95% CI), months Osi + plat–pem | 27.6 (21.0, NR) | 19.6 (10.1, NR) Osi mono | 15.2 (13.6, 19.3) | 11.1 (5.6, 16.5) HR (95% CI) | 0.51 (0.35, 0.75) | 0.62 (0.33, 1.14) Plasma EGFRm clearance at Week 6 — No. at risk: Osi + plat–pem 92 83 70 53 28 14 0; Osi mono 122 110 83 48 25 12 0 No Plasma EGFRm clearance at Week 6 — No. at risk: Osi + plat–pem 13 10 6 5 1 0 0; Osi mono 15 3 3 1 0 0 0 Week 6 | Clearance | No clearance mPFS (95% CI), months Osi + plat–pem | 27.6 (22.0, NR) | 11.8 (5.1, 22.1) Osi mono | 16.4 (13.8, 19.4) | 5.4 (1.6, 5.7) HR (95% CI) | 0.54 (0.38, 0.77) | 0.39 (0.16, 0.94) Data cut-off: 03 April 2023 (PFS) HRs were calculated using an unstratified Cox proportional hazards model. Clearance was defined as undetected plasma EGFRm (Ex19del, L858R) at Week 1, 2, 3, or 6, where it was detected at baseline. CI, confidence interval; EGFR, epidermal growth factor receptor; EGFRm, EGFR mutation; Ex19del, Exon 19 deletion; HR, hazard ratio; m, median; mono, monotherapy; NR, not reached; osi, osimertinib; pem, pemetrexed; PFS, progression-free survival; plat, platinum. David Planchard, MD, PhD Survival table Endpoint: PFS by Week 3 / Week 6 plasma EGFRm clearance status (FLAURA2 exploratory ctDNA analysis) Arm | N | Median | Median CI Wk3 clearance — Osi + plat–pem | 98 | 27.6 | 21.0, NR Wk3 clearance — Osi mono | 100 | 15.2 | 13.6, 19.3 Wk3 no clearance — Osi + plat–pem | 21 | 19.6 | 10.1, NR Wk3 no clearance — Osi mono | 40 | 11.1 | 5.6, 16.5 Wk6 clearance — Osi + plat–pem | 92 | 27.6 | 22.0, NR Wk6 clearance — Osi mono | 122 | 16.4 | 13.8, 19.4 Wk6 no clearance — Osi + plat–pem | 13 | 11.8 | 5.1, 22.1 Wk6 no clearance — Osi mono | 15 | 5.4 | 1.6, 5.7 HR: Wk3 clearance 0.51; Wk3 no clearance 0.62; Wk6 clearance 0.54; Wk6 no clearance 0.39 HR CI: 0.35, 0.75; 0.33, 1.14; 0.38, 0.77; 0.16, 0.94

[Slide 1] OTHER KRAS-OFF INHIBITORS + IO More Feasible, BUT NOT Toxicity-Free | Adagrasib + PEMBRO1 (Ph 2, KRYSTAL-7) | Divarasib + PEMBRO2 (Ph 1b/2, KRASCENDO 170) | Olomorasib +PEMBRO3 (Ph 1-2, LOXO- RAS-20001) | Calderasib + PEMBRO4 (Ph 1, KANDLELIT -001) N | 149 | 78 | 93 | 107 Grade ≥3 TRAES | 70% | 65% | 33% | 33% Diarrhea Grade ≥3 | 3% | 18% | 12% | 0% Hepatic toxicity ALT Grade ≥3 | 11% | 23% | 12% | 8% AST Grade ≥3 | 14% | 18% | 10% | ≈ 7% KRASi dose reductions | 48% | 53% | 20% | - Tested in Phase III trials | KRYSTAL-7 ADA + PEMBRO vs PEMBRO | Krascendo-2 DIVA + PEMBRO vs PEMBRO + ChT | SUNRAY-01 OLO + PEMBRO vs plb + PEMBRO | KANDLELIT-004 CALD + PEMBRO vs plb + PEMBRO Noemi Reguart, MD, PhD | ESMO 2026 1.- Garassino MC. et al. ELCC 2025; 2.- Skoulidis F et al. ASCO 2026; 3.- Burns T et al. JTO 2025; 4.Sacher A. et al. ELCC 2026 [Slide 2] TOXICITY 2 PREDICTABLE = PROACTIVE MANAGEMENT SKIN/NAILS (EGFR, HER2, RAS-ON) | GI/ORAL (EGFR, HER2, RAS-ON/OFF) | METABOLIC/OEDEMA (MET, ALK, ROS, NTRK) | LIVER (ALK, KRAS-OFF) | CNS/CARDIAC (EGFR, ALK, ROS1, NTRK) | LUNG/ILD (EGFR, ALK, KRAS-OFF) ANTICIPATE Emollients, Sun protection, Nail care | Diet, Hydration, Oral hygiene | Lifestyle, Lipids, Weight, Electrolytes, Renal function | LFTs (AST/ALT), Drug interactions | Mood, Cognition, ECG (QTc) | Monitor symptoms + imaging DRUG-SPECIFIC PROPHYLAXIS Doxycycline/ Minocycline P.O. INTERVENE EARLY Topical corticosteroids (e.g. betamethasone), Antiseptic soaks (e.g. chlorhexidine) | Early antidiarrheals, Antiemetic therapy, Steroid mouthwash ± lidocaine | Lipid-lowering therapy, Compression, Limb elevation, Physiotherapy, Diuretics if appropriate HOLD EARLY if clinically significant Corticosteroids if indicated LUNG/ILD: HOLD EARLY if suspected -> DISCONTINUE if confirmed (any grade) DOSE MODIFICATION WHEN APPROPIATE Noemi Reguart, MD, PhD | ESMO 2026 [Slide 3] TOXICITY 3 DOSE MODIFICATION ≠ TREATMENT FAILURE LORLATINIB: Long-term efficacy with vs without dose reduction within 26 weeks (CROWN) [KM curve: PFS, % vs Months 0-100; legend: With dose reduction / Without dose reduction] AFATINIB: impact of dose adjustment on efficacy (LUX-Lung 3) [KM curve: Estimated PFS probability vs Time (months) 0-27; inset table with median PFS, HR (95% CI), P-value - [illegible small print]] DACOMITINIB: impact of dose adjustment on efficacy (ARCHER 1050) [KM curve: PFS by IRC (%) vs Time (months) 0-36; inset table Patients n, Events n, Median (95% CI) PFS - [illegible small print]] DOSE OPTIMIZATION CAN BE A WIN – THE GOAL IS PRESERVING TREATMENT CONTINUITY Noemi Reguart, MD, PhD | ESMO 2026 1.- Mok T et al. ASCO 2026; 2.- Yang J et al. Annals of Oncol 2016; 3.- Corral J et al. Future Oncology 2019 [Slide 4] TAKE-HOME MESSAGES What oncologists should know about managing toxicities of novel therapies? KNOW THE TARGET Different targets, drugs and mechanisms bring distinct toxicity profiles. EFFICACY ≠ TOLERABILITY Balance therapeutic benefit against toxicity. CONTEXT MATTERS Combinations and sequencing can reshape toxicity and pose new challenges. ACT EARLY & PROACTIVELY Anticipate, intervene early and optimize dosing to enable treatment continuity. Noemi Reguart, MD, PhD | ESMO 2026

Hypothesis Four design choices that SETHY and NCT07068542 leave open 1. Immunotherapy partner Evidence: ATC escapes while target shrink; PD-1 alone 19%, lenvatinib + pembrolizumab 36-52%; DTC combination data point the same way Design: ADC + PD-1, 7.5 mg/kg as in SURE-02, with primary G-CSF 2. Lineage and grade as selection criteria Evidence: PTC 13.5 vs FTC 3.9 mo; PDTC 5.4 mo; differentiated high-grade carcinoma not enrolled Design: BRAF V600E or RET/NTRK fusion in, RAS out; a high-grade cohort; classical PTC on a PFS endpoint 3. Biomarkers on the mandatory biopsy Evidence: H-score non-discriminative; the deepest PTC response had an H-score < 100 Design: central TROP2 with Ki-67, TOP1 and SLFN11 pre-specified; confirmed responses; 6-weekly imaging for ATC 4. Neoadjuvant intent: R status as endpoint impractical recruitment ? Evidence: NCT07068542 adds an R0/R1 endpoint; dabrafenib-trametinib already bridges ATC to surgery Design: locally advanced ATC and high-grade DTC, ADC + PD-1 to resection; R0/R1 and pathological response A Simon design with a confirmed-response endpoint needs a cohort that can respond: lineage and grade first, then antigen. Garcia-Alvarez ESMO 2026 (4209O); NCT07068542 protocol v4.0 and SAP v1.0 (Dec 2025); SURE-01 (Necchi JCO 2026); SURE-02 (NCT registry); Hamidi Thyroid 2026; Dierks ESMO 2022; Toda Cancers 2022; Zoppoli PNAS 2012; Bible ATA ATC guideline 2021. Tiago Lopes Nunes da Silva, MD - IPO Lisboa

[Slide 1] Studies on conversion surgery strategies in initially deemed unresectable or borderline resectable, stage III NSCLC Trial | Author Year | Eligibility | N | Neoadjuvant tx | Treatment | N(%) Conversion rate | R0 | MPR | pCR | EFS | OS Retrospective studies NA | Deng Ann Surg2022 | Unresectable stage IIIB | 51 | IO+CTx | Surgery | 31(61%) | N/A | N/A | N/A | Median 27.5m | N/A | | | | | RT | N/A | | | | N/A | N/A NA | Wang, Clin Lung Cancer 2024 | Unresectable stage III | 113 | IO+CTx (various) x 2-4 | Surgery | 79(70%) | 99% | 56% | 32% | 72%@2y | 85%@2y | | | | | RT | NA | | | | N/A | N/A NA | Qi Scientific Rep 2025 | potentially resectable but unsuited for R0 resection | 148 | IO+CTx (various) x 2-4 | Surgery | 102(69%) | 100% | 65% | 41% | 78%@2y | 87%@2y | | | | | RT | 40(27%) | | | | N/A | N/A NA | Ricciuti JAMA Oncol 2025 | T4 and/or N2-N3 (T4 33%, N3 5%) | 112 | IO+CTx | Surgery | 84(75%) | 95% | 42% | 29% | Median 53m | N/A | | | | | RT | 28(25%) | | | | 15m | N/A NA | Wang Transl Lung Cancer Res 2025 | Resectable III | 171 | Various IO+CTx | Surgery | 147(86%) | 100% | 52% | 28% | 66%@3y | 89%@3y | | Unresectable | 469 | | | 278(59%) | 99.6% | 60% | 36% | 75%@3y | 91%@3y Prospective studies TRAILBLAZER NCT04580498 | Zhou Cancer Cell 2024 | Unresectable stage III | 107 | SHR-1701 (+CT ) | Surgery | 27(25%) | 100% | 44% | 26% | 74%@18m | N/A | | | | | RT | 59(55%) | | | | 57%@18m | N/A PeriSCOPE ChiCTR2000040673 | Yu 2025 | High probability of radical resection, IIIB | 30 | Sintilimab+CBDCA+ pemetrexed(adeno) or +Pac/nabPAC(non adeno) x2 | Surgery | 27(90%) | 96% | 44% | 22% | 43%@3y | 70%@3y | | | | | RT | NA | | | | NA | NA PANDA-1 | Liu WCLC 2026 | Unresectable stage III (T4, multiple/bulky N2,3 | 55 | Tislelizumab+ nabPAC+Ptx 2-4 | Surgery | 43(78%) | 95% | 79% | 49% | N/A | N/A | | | | | RT | 5(9%) | | | | N/A | N/A MDT-Bridge NCT05925530 | Reck WCLC 2026 | 92 Resectable and 50 borderline stage IIIB-C | 142 | Durvalumab+CT | Surgery | 106(75%) | 96% | N/A | 23% | 89%@1y | N/A | | | | | RT | 25(18%) | | | | N/A | N/A Tetsuya Mitsudomi, MD, PhD [Slide 2] Can radiotherapy safely wait until after upfront immunochemotherapy? Real-world multicenter retrospective study with PS matching Li et al., Lung Cancer 2026 PFS (PSM adjusted PFS) Median Progression-free Survival, mo (95% CI) ADJ : 16.3 (13.4 to 20.4) NEO : 29.3 (22.0 to NR) HR (95% CI), 0.53 (0.37 to 0.74) Log-rank P < .001 Neo IO+CT->CRT (N=154) CRT->IO(N=101) OS (PSM adjusted OS) Median Overall Survival, mo (95% CI) ADJ : 41.1 (26.5 to NR) NEO : NR (NR to NR) HR (95% CI), 0.47 (0.28 to 0.78) Log-rank P = .003 Prospective clinical trials that tested neoadjuvant immuno(chemo)therapy followed by cCRT for unresectable NSCLC Trial | Author | Phase | N | Stage, etc. | Treatment | 1Y PFS(%) | mPFS months | 1Y OS(%) | mOS months Neoadjuvant immunotherapy followed by radiation (ICI -> CT RT -> ICI) SPRINT NCT 03523702 | Ohri JCO 2023 | 2 | 25 | II/III PD-L1>=50% | Pembro x3 ->RT(55Gy) ->Pembro (1Y) | 76 | 26 | 92 | NR AFT-16 NCT03102242 | Ross JAMA Oncol 2024 | 2 | 62 | III | Atezo x4 ->cCRT(60Gy)->Atezo(1Y) | 68.9 / 76.2 (from end of RT) | 30 | 87 / 95.5(from end of RT) | NR Neoadjuvant immunochemotherapy followed by radiation (ICI CT -> CT RT -> ICI) APOLO NCT04776447 | Provencio WCLC 2024 | 2 | 38 | IIIA-IIIC | Atezo+PC x3 ->cCRT(66Gy) ->Atezo (1Y) | 68.4 / 68.9 (from end of RT) | 20.8 | 86.8 | NR PACIFIC-BRAZIL/ LACOG2218 NCT04230408 | William WCLC 2024 | 2 | 49 | III | Durva+PC x2->cCRT(60Gy) +Durva ->Durva (x12) | 68.1 / 66.4 (from consolidation ) | NR | 81.2 | NR GASTO-1091/ CA209-7AL NCT04085250 | Liu ASCO 2024 | Rnd 2 | ITT 264 | III | Nivo+CisDoc ->cCRT(54-64Gy) | Rnd N=172 | Nivo (1Y) | 72.6 | NR | 83.3 | NR | | | | | | | Obs | 52.5 | 12.2 | 69.6 | NR InTRist/ GCOG0074 NCT05888402 | Wang ASCO 2025 | Rnd 2 | 50 | III (T>=5cm or N>=2cm) | Rnd | Toripa+Platinum doublet x2 -> cCRT(60-66Gy) ->T(1Y) | 85.6 | NR | | | | | | Platinum doublet x2 | 54.5 | 12.4 cCRT followed by maintenance IO (for comparison) (CT RT -> ICI) PACIFIC NCT02125461 | Spigel JCO 2022 | 3 | 713 | III non PD after cCRT | Rnd(2:1) | D (1Y) | 55.7 | 16.9 | 83.1 | 47.5 | | | | | | Placebo | 34.5 | 5.6 | 74.6 | 29.1 PC: paclitaxel + carboplatin, CisDoc: cisplatin + docetaxel, cCRT: concurrent chemoradiation. NR: not reached, Rnd 2: randomized phase 2, Rnd: randomization [Slide 3] Current and potential future strategy for localized NSCLC without AGA Current Standard of Care Stage II-III NSCLC Without AGA -> MDT review Unresectable -> CT RT -> ICI Resectable -> CT ICI -> Surg -> ICI Potential Future Strategy To be validated in prospective trials Stage II-III NSCLC Without AGA -> CT ICI -> MDT review Unresectable -> CT RT -> ICI Resectable -> Surg -> ICI Tetsuya Mitsudomi, MD, PhD [Slide 4] A sense of balance is essential in clinical decision-making Favorable outcome <-> Unfavorable outcome Resectability: <=T3 / Single N2 / Lobectomy / oExperienced surgical team <-> T4 / multiple N2, N3 / Pneumonectomy / broncho/angioplasty / Inexperienced team Operability: Good PS / Young / No comorbidity / Fit / Autoimmune disease/ILD absent <-> poor PS / Elderly / comorbidity present / Unfit / present Biomarker: PD-L1 high / Actionable driver (EGFR/ALK) absent <-> PD-L1 low or absent Multidisciplinary Team (MDT): Medical Oncologist, Radiation Oncologist, Pathologist, Physiotherapist, Nurse, Pharmacist, Radiologist, Pulmonologist, Thoracic Surgeon Patient Preference

313P A large-scale concordance comparing digital and manual PD-L1 scoring using surgical specimens of non-small cell lung cancer in the LC-SCRUM-Advantage/MRD study. Yoshitaka Zenke, Kiyotaka Yoh, Kotaro Nomura, Takafumi Hashimoto, Toshio Kasugai, Shingo Miyamoto, Akira Yura, Tomohiro Maniwa, Hiroshi Saijo, Mayu Kawakami, Tetsuya Sakai, Hiroki Izumi, Hibiki Udagawa, Eri Sugiyama, Shingo Matsumoto, Soohyun Hwang, Tetsuro Taki, Keiju Aokage, Koichi Goto Background - Programmed death ligand 1 (PD-L1) expression is a useful biomarker for immune checkpoint inhibitors in advanced lung cancer. However, pathologists' manual evaluation of PD-L1 expression has practical limitations, including observer bias. - We previously compared programmed death-ligand 1 (PD-L1) tumor proportion score (TPS) using biopsy specimens between pathologists and an artificial intelligence (AI). - In this study, we conducted a detailed analysis of the concordance between a pathologist and AI analyzers in the evaluation of PD-L1 TPS using surgical specimens. Methods - This multicenter prospective observational study (LC-SCRUM-Advantage/MRD) included patients with stage I to III non-small cell lung cancer (NSCLC) between August 2023 and November 2025; a total of 1,000 lung cancer patients were enrolled. - PD-L1 Tumor Proportion Score (TPS) was assessed in lung surgical specimens, by using a PD-L1 IHC 22C3 pharmDx and scanned at ×40 magnification using a whole-slide images scanner (Hamamatsu Photonics). - The results of PD-L1 TPS were evaluated independently by a lung pathologist trained in IHC assessment of PD-L1 expression. - We examined an AI-powered PD-L1 analyzer, namely Lunit SCOPE PD-L1 (Figure 1). - Manual PD-L1 assessment requires at least 100 viable tumor cells, whereas AI evaluated a mean of approximately 289,000 tumor cells per case. Table 1: Characteristics of Patients Characteristics | n | (%) No. of patients | 1000 | Age | Median (range) | 60 (27-94) Sex | Male | 591 | 59 Sex | Female | 409 | 41 Histology | Adeno (Ad) | 751 | 75 Histology | Squamous (Sq) | 182 | 18 Histology | Others | 67 | 7 Pathological stage | I | 662 | 66 Pathological stage | II | 199 | 20 Pathological stage | III | 139 | 14 Smoking status | Ever | 700 | 70 Smoking status | Never | 300 | 30 Surgical procedure | Lobectomy | 741 | 74 Surgical procedure | Sublobar resection | 256 | 26 Neoadjuvant therapy | Yes | 15 | 2 Neoadjuvant therapy | No | 985 | 98 Table 2. Concordance rate of AI- and pathologist-assessed PD-L1 TPS in NSCLC. N (%) | Pathologist TPS (N=993): < 1% | 1-49 % | ≥ 50% AI-TPS < 1% | 216 (51%) | 10 (3%) | 0 (0%) AI-TPS 1-49 % | 209 (49%) | 359 (95%) | 37 (20%) AI-TPS ≥ 50% | 2 (0.4%) | 8 (2%) | 152 (80%) Overall concordance in 3 groups: 73% Table 3. Concordance rate of AI- and pathologist PD-L1 TPS in Ad. N (%) | Pathologist TPS (N=751): < 1% | 1-49 % | ≥ 50% AI-TPS < 1% | 198 (54%) | 9 (3%) | 0 (0%) AI-TPS 1-49 % | 164 (45%) | 261 (95%) | 27 (25%) AI-TPS ≥ 50% | 2 (0.5%) | 6 (2%) | 79 (75%) Overall concordance in 3 groups: 72% Table 4. Concordance rate of AI- and pathologist PD-L1 TPS in Sq. N (%) | Pathologist TPS (N=182): < 1% | 1-49 % | ≥ 50% AI-TPS < 1% | 7 (16%) | 0 (0%) | 0 (0%) AI-TPS 1-49 % | 35 (83%) | 79 (99%) | 7 (12%) AI-TPS ≥ 50% | 0 (0%) | 1 (1%) | 51 (88%) Overall concordance in 3 groups: 75% Figure 5. AI PD-L1 TPS classification as TPS <1% by a pathologist: NSCLC (n = 427) N = 420 (98%); Adeno (n = 364) N = 358 (98%); Squamous (n = 42) N = 41 (98%) SUMMARY - The concordance rate in PD-L1 TPS using surgical specimens between a pathologist and an AI analyzer was lower in PD-L1 <1% than in the ≥50/1-49%, and it was even lower in Sq than in Adeno. - Discordance was largely confined to the low PD-L1 range, with 98% of pathologist-defined TPS <1% tumors showing AI TPS <10%. - Further clinical studies should evaluate whether these discordant results affect clinical outcomes. Conclusion: Our study findings suggest that an AI TPS cutoff of <10% may better correspond to pathologist-defined PD-L1 negativity (<1%).

[Slide 1] AI in biomedical research and healthcare nature reviews cancer | Review article A guide to artificial intelligence for cancer researchers Raquel Perez-Lopez, Narmin Ghaffari Laleh, Faisal Mahmood & Jakob Nikolas Kather 2000s: Classical Machine learning methods 2010s: Single-purpose Deep learning models (No cancer / Cancer) 2020: Foundation models trained with self-supervised learning (Foundation model → Downstream models) 2023: Generalist models (e.g. large language models) (Generalist model → Predicted drug response / Cancer type and subtype / Radiology assessment) 2025: AI agents (User query / input → LLM → Contextual understanding) Single task models | Multi task models | Autonomy Perez-Lopez et al., Nat Rev Cancer, 2024, https://t.ly/sTgCX [Slide 2] ESMO Guidance on the Use of Large Language Models in Clinical Practice (ELCAP) Type 1 Patient-facing AI systems: Symptom Queries / Patient Education / Lifestyle Tracking Type 2 HCP-facing AI systems: Clinical Decision Support / Multilingual Communication / Education and Training Type 3 Background AI systems: Data Extraction / Background Alert Systems Which evidence is required? ESMO EVALLM, work in progress Wong et al., Annals of Oncology (2025) Slide 7 www.kather.ai KATHER LAB [Slide 3] AI in a European hospital Data architecture: Access to all clinical data for training & real time deployment AI applications / Clinical Data Repository / Primary clinical systems (EHR, PACS, ...) GPU compute on premise or in the Cloud Substantial demand for inference, and much more demand when models are also trained AI-friendly governance and effective administration AI can be deployed in a safe and compliant way GDPR / EU AI act / MDR are navigable Slide 15 www.kather.ai KATHER LAB

#3842P Ifinatamab deruxtecan (I-DXd) in extensive-stage small cell lung cancer: IDeate-Lung01 updated results Myung-Ju Ahn, Melissa L. Johnson, Luis Paz-Ares, Makoto Nishio, Christine L. Hann, Nicolas Girard, Pedro Rocha, Hidetoshi Hayashi, Tetsuya Sakai, Yu Jung Kim, Haichuan Hu, Meng Qian, Cecilio Roi Garcia, Juliette Godard, Mei Tang, Charles M. Rudin OBJECTIVE • To provide updated efficacy and safety results from patients with previously treated ES-SCLC who received I-DXd 12 mg/kg in the Phase 2 IDeate-Lung01 study (NCT05280470), with an additional 6 months of follow-up from the primary analysis • Results are presented for the total 12-mg/kg group (N=137) and for subgroups of patients who received I-DXd 12 mg/kg as either 2L (n=32) or 3L+ (n=105) therapy CONCLUSIONS • With 6 months of additional follow-up from the primary analysis, I-DXd 12 mg/kg continued to demonstrate remarkable efficacy in patients with previously treated ES-SCLC • In the total 12-mg/kg group, cORR was 48.9% (95% CI, 40.3–57.6), mPFS was 5.1 months (95% CI, 4.2–5.5), and mOS was 10.4 months (95% CI, 9.1–12.4) • Efficacy was particularly encouraging in the 2L subgroup, with numerical improvements observed in cORR, mDOR, mPFS, and mOS compared with the primary analysis • The safety profile was consistent with prior data, with no new safety signals identified • These results further support I-DXd as a promising novel therapy for patients with ES-SCLC in the 2L+ setting • I-DXd is being further investigated in the ongoing global Phase 3 IDeate-Lung02 study (NCT06203210), which is comparing the efficacy and safety of I-DXd 12 mg/kg versus physician's choice of therapy in patients with ES-SCLC with disease progression on or after one prior platinum-based regimen INTRODUCTION • Despite recent advancements in 2L+ treatment options for patients with ES-SCLC, outcomes remain poor, and there is still no globally accepted standard of care • Furthermore, existing chemotherapeutic options are associated with clinically relevant hematologic toxicities, and tarlatamab is associated with CRS and ICANS • I-DXd, a B7-H3–directed ADC designed to enhance selective tumor-cell death and reduce systemic exposure, comprises a humanized anti–B7-H3 IgG1 mAb, a tetrapeptide-based cleavable linker, and a topoisomerase I inhibitor payload (DXd) • In the primary analysis of the Phase 2 IDeate-Lung01 study (data cutoff: March 3, 2025), I-DXd 12 mg/kg demonstrated a cORR of 48.2%, mPFS of 4.9 months, and mOS of 10.3 months, as well as a manageable safety profile, in 137 patients with previously treated ES-SCLC • We present updated results from these patients, with an additional 6 months of follow-up from the primary analysis METHODS • IDeate-Lung01 (NCT05280470) is a Phase 2, multicenter, randomized, open-label study (Figure 1) • Patients with histologically or cytologically documented ES-SCLC who were previously treated with ≥1 line of PBC and ≤3 lines of systemic therapy were eligible • This analysis included patients who received I-DXd 12 mg/kg IV Q3W in either the dose-optimization or extension parts of the study Figure 1. IDeate-Lung01 study design Part 1: Dose optimization: R 1:1 Arm 1: I-DXd 8 mg/kg IV Q3W n=46; Arm 2: I-DXd 12 mg/kg IV Q3W n=42 Part 2: Extension: I-DXd 12 mg/kg IV Q3W n=95 Primary endpoints • ORR by BICR RESULTS • At the September 3, 2025 data cutoff, a total of 137 patients had received I-DXd 12 mg/kg as either 2L (n=32) or 3L+ (n=105) treatment • At data cutoff, 5 (3.6%) patients remained on treatment and 132 (96.4%) had discontinued treatment (PD, n=92 [67.2%]; AE, n=18 [13.1%]; death, n=10 [7.3%]; clinical progression, n=8 [5.8%]; patient withdrawal, n=4 [2.9%]) • In the total 12-mg/kg group, median follow-up was 18.6 months (95% CI, 18.0–19.2), and median treatment duration was 4.8 months (range, 0.7–29.0) – 2L subgroup: median follow-up, 18.7 months (95% CI, 17.1–NE); median treatment duration, 5.6 months (range, 0.7–19.8) – 3L+ subgroup: median follow-up, 18.6 months (95% CI, 17.8–19.2); median treatment duration, 4.5 months (range, 0.7–29.0) Table 1. Baseline demographic and clinical characteristics (Total 12 mg/kg N=137) Age, median (range), years 63 (34–79) Brain metastases at baseline by BICR, n (%) 52 (38.0) Median number of prior lines of systemic therapy (range) 2 (1–3) Prior anti–PD-(L)1 therapy, n (%) 111 (81.0) • In the total 12-mg/kg group, cORR was 48.9% (95% CI, 40.3–57.6), and DCR was 87.6% (95% CI, 80.9–92.6; Figure 2) – 2L subgroup: cORR, 59.4% (95% CI, 40.6–76.3); DCR, 96.9% (95% CI, 83.8–99.9) – 3L+ subgroup: cORR, 45.7% (95% CI, 36.0–55.7); DCR, 84.8% (95% CI, 76.4–91.0) • In 11 patients previously treated with DLL3-targeting T-cell engagers, cORR was 72.7% (95% CI, 39.0–94.0), comprising 1 CR and 7 PRs Figure 2. Best percentage change from baseline in sum of lesion diameters by BICR by LOT (Total 12 mg/kg N=137 | 2L n=32 | 3L+ n=105) cORR, % (95% CI) 48.9 (40.3–57.6) | 59.4 (40.6–76.3) | 45.7 (36.0–55.7) CR, n (%) 3 (2.2) | 0 | 3 (2.9) PR, n (%) 64 (46.7) | 19 (59.4) | 45 (42.9) DCR, % (95% CI) 87.6 (80.9–92.6) | 96.9 (83.8–99.9) | 84.8 (76.4–91.0) • In the total 12-mg/kg group, mTTR was 1.4 months (range, 1.0–8.1), and mDOR was 5.5 months (95% CI, 4.1–6.5; Figure 3) – 2L subgroup: mTTR, 1.4 months (range, 1.2–4.3); mDOR, 9.7 months (95% CI, 3.6–NE) – 3L+ subgroup: mTTR, 1.4 months (range, 1.0–8.1); mDOR, 4.3 months (95% CI, 3.7–5.8) Figure 3. Change in sum of lesion diameters over time by BICR • In the total 12-mg/kg group, mPFS was 5.1 months (95% CI, 4.2–5.5), and mOS was 10.4 months (95% CI, 9.1–12.4; Figure 4) – 2L subgroup: mPFS, 6.0 months (95% CI, 4.6–8.4); mOS, 12.6 months (95% CI, 7.3–16.2) – 3L+ subgroup: mPFS, 4.5 months (95% CI, 4.2–5.4); mOS, 10.3 months (95% CI, 9.1–12.3) Figure 4. (A) PFS by BICR and (B) OS in the total 12-mg/kg group and by LOT (Total 12 mg/kg N=137 | 2L n=32 | 3L+ n=105) mPFS (95% CI), months 5.1 (4.2–5.5) | 6.0 (4.6–8.4) | 4.5 (4.2–5.4) PFS rate, % (95% CI): 3 months 68.8 (60.2–75.9) | 74.3 (55.1–86.2) | 67.1 (57.1–75.2); 6 months 38.3 (28.2–44.4) | 51.7 (33.1–67.4) | 31.6 (22.8–40.7); 9 months 20.7 (14.3–28.1) | 31.5 (16.1–48.1) | 17.4 (10.8–25.5) mOS (95% CI), months 10.4 (9.1–12.4) | 12.6 (7.3–16.2) | 10.3 (9.1–12.3) OS rate, % (95% CI): 6 months 77.4 (69.4–83.5) | 71.9 (52.9–84.3) | 79.0 (69.9–85.7); 12 months 43.1 (34.7–51.2) | 50.0 (31.9–65.7) | 41.0 (31.5–50.1) • In the total 12-mg/kg group, any-grade TRAEs were reported in 124 (90.5%) patients, with Grade ≥3 TRAEs occurring in 50 (36.5%) patients (Table 2) Table 2. Safety summary (Total 12 mg/kg N=137) Any-grade TRAEs 124 (90.5); Grade ≥3 50 (36.5) Serious TRAEs 25 (18.2) TRAEs associated with dose delay/interruption 38 (27.7) TRAEs associated with dose reduction 22 (16.1) TRAEs associated with treatment discontinuation 16 (11.7) TRAEs associated with death 6 (4.4) Figure 5. Most common TRAEs in the total 12-mg/kg group (N=137) (any grade / Grade ≥3, %) Nausea 43.1 / 2.2; Fatigue (grouped term) 40.9 / 5.1; Neutropenia (grouped term) 34.3 / 14.6; Anemia (grouped term) 34.3 / 10.2; Decreased appetite 33.6 / 1.5; Leukopenia (grouped term) 23.4 / 4.4; Lymphopenia (grouped term) 19.7 / 12.4; Thrombocytopenia (grouped term) 19.0 / 5.8; Diarrhea 15.3 / 1.5; Vomiting 12.4 / 0.7; Constipation 12.4 / 0; AST increased 10.2 / 0.7; Pneumonitis 10.2 / 0.7; Stomatitis 10.2 / 0 • In the total 12-mg/kg group, any-grade adjudicated treatment-related ILD/pneumonitis was reported in 18 (13.1%) patients (Grade 1, n=3 [2.2%]; Grade 2, n=9 [6.6%]; Grade 3, n=4 [2.9%]; Grade 5, n=2 [1.5%]; Table 3) Table 3. Adjudicated treatment-related ILD/pneumonitis (Total 12 mg/kg N=137): Any-grade 18 (13.1); Grade 1 3 (2.2); Grade 2 9 (6.6); Grade 3 4 (2.9); Grade 4 0; Grade 5 2 (1.5) This study is sponsored by Daiichi Sankyo Inc. and Merck Sharp & Dohme LLC. Presented at the European Society for Medical Oncology (ESMO) Congress; October 23–27, 2026; Madrid, Spain

[Slide 1] ctDNA dynamics with targeted therapies in AGA + NSCLC On-treatment ctDNA clearance has good prognosis with TKI treatment in AGA+ NSCLC Sevabertinib in HER2 TKD mutation NSCLC E Progression-free Survival According to ctDNA Response Never detected: Median, 12.3 mo (95% CI, 2.5-NE) Clearance: Median, 14.7 mo (95% CI, 6.9-NE) Persistent detection: Median, 7.0 mo (95% CI, 3.9-16.4) No. at Risk (Months 0, 6, 12, 18, 24, 30) Never detected 12 8 5 1 0 0 Clearance 28 20 12 7 3 1 Persistent detection 25 11 5 1 0 0 Le et al NEJM 2025 Ensartinib in METex14 skipping NSCLC B [stacked bar: PR / SD / PD by ctDNA residual, ctDNA negative, ctDNA clearance; Percentage, %] C [KM: Probability of Progression-free Survival (%) - ctDNA clearance, ctDNA negative, ctDNA residual; Time (months)] Xia Yang et al eClinicalMedicine 2025 Xiuning Le MD PhD MD Anderson Cancer Center Survival table Endpoint: PFS according to ctDNA response (sevabertinib, HER2 TKD-mutant NSCLC; Le et al NEJM 2025) Arm | N | Median | Median CI Never detected | 12 | 12.3 mo | 2.5-NE Clearance | 28 | 14.7 mo | 6.9-NE Persistent detection | 25 | 7.0 mo | 3.9-16.4 [Slide 2] Take home message Non-detectable: - For patients with baseline non-detectable ctDNA, the prognosis is good, there might be an option for osimertinib monotherapy Median PFS 30 month Detectable: - For patients started on osimertinib monotherapy, but without ctDNA clearance of EGFRm at 3-4 weeks, the addition of platinum-chemotherapy is indicated Median PFS 17.8 month - For patients who had a clearance of ctDNA EGFRm, the magnitude of benefit of osi-chemo is large, with 12 months absolute improvement of PFS, HR 0.51. - Therefore, all eligible patients should start with osi-chemo because on treatment ctDNA EGFRm clearance was NOT predictive of benefit with osi+chemo vs. osimertinib monotherapy Median PFS 24 month Xiuning Le MD PhD MD Anderson Cancer Center [Slide 4] CROWN: co-alterations beyond the ALK driver [Oncoprints: EPs (n=22) and LTRs (n=30); gene rows incl. ALK, EML4, TP53, ATM, ZFHX3, CHEK2, NOTCH3, NTRK3, ARID1A, NF1, KRAS, MYC, CIC, SPEN, APC, MTOR, PREX2, ADMSC, PCDH15, ROS1, TET1, ARID2, ATRX, PDGFRA, DICER1, BRIP1, IGR, ROBO2, CDC73, BRCA1, PBRM1, DNMT3B, SETD2; legend SNV, Indel, CNA + amplification, Fusion] Early progression vs long-term control TP53: 50% vs 17% Altered genes: median 10 vs 6 "Inflammatory" pathway Baseline: 32% vs 7% New at progression: 7/17 pairs Tumor origin? TF adjustment? CHIP filtering? Co-alterations suggest residual risk. The inflammatory signal remains hypothesis-generating. Solomon et al. ESMO 2026, 2636RO, slides 4-5. Razavi et al. Nat Med 2019;25:1928-1937.

IFCT ESMO 2026 Encorafenib plus binimetinib in patients (pts) with BRAFV600E-mutant advanced non-small cell lung cancer: updated efficacy and safety results from the IFCT-1904 ENCO-BRAF phase 2 trial C. Ricordel1 · J. Mazieres2 · C. Mascaux3 · F. Guisier4 · P. Tomasini5 · M. Wislez6 · D. Moro-Sibilot7 · S. Cousin8 · S. Couraud9 · V. Fallet10 · C. Guguen11 · A. Lagrange12 · A. Madroszyk-Flandin13 · J. Otto14 · M. Pérol15 · B. Roch16 · E. Amour17 · F. Morin17 · A.B. Cortot18 · D. Planchard19 Background Methods and Objectives Open-label, Phase II trial Treatment naive Planned N = 60 Previously treated Planned N = 59 Confirmed best objective re[obscured] and best overall response Treatment naive N=63 [obscured]R (CR + PR): 68.3% [95% CI: 56.8%-79.7%] Previously treated N=31 best ORR (CR + PR): 61.3% [95% CI: 44.1%-78.4%] Quality of Life – VAS score Safety – TRAE any grade ≥10% DOI ESMO 2026 CT.gov identifier: NCT04526782 2651P

[Slide 1] MADRID 2026 ESMO congress OP: 1000O HS-20122, a Novel EGFR/c-Met Bispecific Antibody Drug Conjugate, in Patients with Advanced Solid Tumors Hongyun Zhao, Yuxiang Ma, Fangling Ning, Qingfeng Li, Zhiyong He, Zhiye Zhang, Ruiling Ning, Wei Zheng, Qiming Wang, Shengxiang Ren, Juan Li, Yongchang Zhang, Yongsheng Wang, Haifeng Liu, Minglei Zhuo, Wenzhao Qin, Chenge Song, Lu Chen, Xiaoqing Zhang, Li Zhang [Affiliations in small print, including Sun Yat-sen University Cancer Center ... Shanghai Hansoh Bio Medical Co., Ltd, Shanghai, China] Presented by: Hongyun Zhao, M.D. Department of Clinical Research, Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, Guangzhou, China Madrid Spain, October 24, 2026 [Slide 2] Background Target, Structure and Preclinical • There are signaling cross-talk between EGFR and cMet, so co-targeting EGFR and c-Met with ADC products may limit compensatory pathway activation and improve overall efficacy. • HS-20122 is a bi-specific novel ADC, including a 1:1 EGFR/c-Met antibody, cleavable linker, Topoisomerase1 inhibitor payload(DAR=4). • HS-20122 is highly effective in multiple preclinical models, including lung cancer, gastrointestinal cancer, head and neck cancer. HS-20122 Diagram EGFR binding Fab | c-Met binding Fab | Cleavable linker | Topoisomerase 1 inhibitor (HS-9265) We assessed safety, tolerability, and preliminary efficacy of HS-20122 in patients with solid tumors in a first-in-human (FIH) trial. Abbreviations: ADC, Antibody Drug Conjugate; c-Met, cellular Mesenchymal-Epithelial Transition factor; DAR, drug-to-antibody ratio; EGFR, Epidermal Growth Factor Receptor; Fab, Fragment of Antigen Binding Hongyun Zhao, M.D. [Slide 3] Efficacy in EGFRm NSCLC (All doses) All patients had received prior EGFR-TKIs(90% 3rd generation EGFRTKIs) and 26% had received prior chemotherapy. Best Overall Response (N=106) ORR: 56% DCR: 93% Best % Change in SoD from Baseline in Target Lesion Size PR | SD | PD | NE Treatment Duration (N=106) 6-mo PFS rate: 77% 95% CI (61.7, 86.4) Ongoing | End of Treatment PR | SD | PD | NE Time Since Start of Treatment (Weeks) Abbreviations: NE, Not Evaluable; ORR, Objective Response Rate; PD, Progressive Disease; PFS, Progression-Free Survival; PR, Partial Response; SD, Stable Disease; TKI, Tyrosine Kinase Inhibitor. Hongyun Zhao, M.D. Data Cut-off: 06-Jul-2026 [Slide 4] Efficacy in Squamous NSCLC and HNC All patients had received platinum-based chemotherapy and anti-PD-(L)1. Best Overall Response in Squamous NSCLC (N=28) ORR : 32.1% DCR: 92.9% Best Overall Response in Head and Neck Cancer (N=12) ORR : 50.0% DCR : 91.7% Bar labels: HNSCC, HNSCC, HNSCC, NPC, HNSCC, NPC, HNSCC, NPC, NPC, NPC, NPC, HNSCC PR | SD | PD Best % Change in SoD from Baseline in Target Lesion Size Abbreviations: HNSCC, Head and Neck Squamous Cell Cancer; NPC, Nasopharyngeal Carcinoma; PD, Progressive Disease; PR, Partial Response; SD, Stable Disease. Hongyun Zhao, M.D. Data Cut-off: 06-Jul-2026

VORONOI | 1067P Early Clinical Anti-tumor Activity of VRN101099 Across Diverse HER2-altered Solid Tumors LUMIN-HER2(NCT06806982): Phase 1 Study in HER2-Driven Advanced Solid Tumors Y.H. Park, S.Y. Rha, S-A. Im, J.H. Youk, R. Cosman, C. Steer, J.H. Kim, J.H. Kim, K.H. Jung, J.L. Lee, J.Y.Park, S.Y. Zhang Background - HER2 alterations comprise amplification/overexpression (HER2-positive disease) and activating mutations, arising within the tyrosine kinase domain (TKD; exon 20 insertions, L755) and outside it (extracellular S310F/Y, transmembrane V659/S653). CNS progression is a frequent and poorly addressed failure mode across these HER2-driven tumors. - VRN101099 (VRN10) is an oral, once-daily, brain-penetrant covalent HER2 TKI binding an open-state activation loop conformation, promoting HER2 internalization and degradation. It is selective over wild-type EGFR, is not a BCRP substrate, and retains preclinical activity in T-DXd-resistant and intracranial models. - Here we report Phase 1a safety, exposure-response and early activity across TKD and non-TKD HER2-mutant and HER2-positive solid tumors (LUMIN-HER2, NCT06806982). Trial design - Phase I study design - Standard "3 + 3" dose escalation - 18 to 30 pts in dose escalation, plus up to 72 backfill pts (up to 102 pts in total) - DLT assessment: first cycle of treatment (i.e. Cycle 1, 21 days of IP) - Phase Ib Monotherapy (dose optimization and expansion): HER2 activating mutation NSCLC, Phase Ib combination (dose finding and expansion): HER2 positive solid tumors. [Schema: Phase Ia dose escalation (Monotherapy) 80 mg, 160 mg, 240 mg, 320 mg, 480 mg, 560 mg with Backfill cohorts; Phase Ib dose optimization (Monotherapy & Combi-therapy): Dose A/B (Mono) -> Mono Expansion HER2 activating mutation NSCLC; Dose A/B (Combi) -> Combi Expansion (with HER2-ADC or HP) HER2 positive cancer. First patient dosed in March 2025; Phase Ib initiation scheduled for early 2027] Phase Ia dose escalation key eligible patients - Age >=18 years with HER2-altered solid tumors (confirmed by IHC 1+/2+/3+, FISH/ISH, or NGS of tissue/ctDNA) - HER2 positive mBC or mGC with PD after prior anti-HER2 therapy - Other HER2-altered solid tumors with no approved HER2-targeted SoC, progressed on all available therapies - NSCLC with HER2-activating mutations after prior HER2 TKI or ADC. Primary endpoints: Safety; Tolerability; Determine the MTD or RP2D Secondary endpoints: PK (Pharmacokinetics); ORR, DOR, DCR, PFS, and TTP per RECIST v1.1 Exploratory endpoints: PD (Pharmacodynamics): ctDNA, pHER2; Intracranial ORR per RANO-BM Demographics Characteristics | VRN101099 N=50, (80-480 mg) Median age, years (range) | 63 (35-82) Sex, n (%) Male / Female | 26 (52%) / 24 (48%) Race, n (%) Asian / White | 40 (80%) / 10 (20%) ECOG PS, n (%) 0 / 1 | 30 (60%) / 20 (40%) HER2 aberration, n (%) HER2 positive / mutation | 21 (42%) / 16 (32%) Primary tumor site: Breast | 11 (22%) Gastrointestinal | 13 (26%) Hepatobiliary | 7 (14%) Lung | 4 (8%) others | 15 (30%) CNS metastasis, n (%) Yes / No | 10 (20%) / 40 (80%) Median No. prior systemic therapies, n (range) | 3 (0 - 11) Prior HER2 Abs | 24 (48%) Prior HER2 ADC, n (%) | 26 (52%) Prior HER2 TKIs, n (%) | 7 (14%) Clinical results A. HER2 mutation patients (N=12): TKDm ORR: 40% (2/5); Non-TKDm ORR: 43% (3/7) [waterfall; TKD mutation / Non-TKD mutation; best responses PD, SD, cPR] B. HER2 positive patients (N=18): ORR: 17% (3/18) [waterfall; Breast cancer / Other solid tumor] Figure 1. Best percentage change from baseline in target lesion size with VRN101099 monotherapy. Evaluable patients had at least one post-baseline tumor assessment. (A) HER2m tumors: responses occurred in both TKD (PR, 2/5) and non-TKD (PR, 3/7) mutations. (B) HER2-positive tumors (BC and other solid tumors): most patients had received prior HER2-targeted therapy, yet tumor shrinkage was observed despite heavy pretreatment. Safety - Overall, the safety profile was tolerable and manageable across all dose levels. - Diarrhea was effectively managed with loperamide administered on an as-needed basis upon symptom onset. - Escalation to 560 mg is ongoing, with MTD not yet reached. TRAEs occurring in >=5% of patients are shown. TRAE | VRN101099, 80-480mg (N=50) All | >= Grade 3 Any | 38 (76%) | 3 (6%) Diarrhea | 26 (52%) | 2 (4%) Nausea | 16 (32%) | 0 Vomiting | 6 (12%) | 0 Decreased appetite | 3 (6%) | 0 Rash | 4 (8%) | 0 Fatigue | 3 (6%) | 0 Stomatitis | 3 (6%) | 0 Exposure-Response correlation Figure 2. Exposure-response relationship of VRN101099. (A) VRN101099 exposure (unbound Cavg, AUC0-24/24) versus best change in target lesion size. Higher exposure was associated with greater tumor shrinkage, and shrinkage occurred when exposure exceeded the fu x GI50 range (shaded) of HER2-altered cell lines. (B) In vitro GI50 across HER2-altered cell lines. VRN101099 induced HER2 internalization Figure 3. Enhanced activity of VRN101099 in combination with T-DXd. (A) VRN101099 promoted HER2 internalization, enhancing T-DXd uptake vs. zongertinib (Incucyte). (B) VRN101099 + T-DXd induced greater tumor regression than zongertinib + T-DXd in HER2+ BT-474 xenografts. (C) In vivo tumor HER2 degradation by VRN101099 (PD analysis; 6 h post-dose, Day 5). Together, these data suggest that VRN101099-induced HER2 internalization enhances T-DXd uptake, contributing to the improved combination efficacy. T-DXd resistance model (A) AKR1C has been reported to mediate metabolism of the T-DXd payload DXd. (B) Both DXd and Zongertinib are substrates of BCRP, whereas VRN101099 is not. Figure 4. VRN101099 retains antitumor activity in T-DXd-resistant models. (A) Mice bearing AKR1C overexpressing(OE), T-DXd-resistant NCI-N87 xenografts were treated with vehicle, T-DXd, zongertinib, or VRN101099, and tumor volume was measured over 21 days. (B) Mice bearing BCRP-overexpressing, T-DXd-resistant SK-BR-3 tumors were treated as in (A), and tumor burden was monitored by bioluminescence imaging over 14 days. Data are mean ± SEM. CNS activity in intracranial models Kp,uu,CSF >=1 Figure 5. CNS penetration and intracranial activity of VRN101099. (A) VRN101099 Kp,uu,CSF in mouse. (B) Greater activity versus HER2 TKIs in intracranial Ba/F3 HER2 YVMA model (HER2m NSCLC). (C) Greater activity versus zongertinib and T-DXd in intracranial BT-474 model (HER2+ BC). Intracranial response case study Patient A. HER2 A775_G776insYVMA - NSCLC: -56.14% PR (C1D1, C3D1, C5D1, C7D1); 2 Lung TL PR achieved at First Tumor Assessment (-56.1%); Deep partial response following relapse on sevabertinib, ELVN-002, and T-DXd; Brain NTL also decreased (RANO-BM: Non-CR/Non-PD) Patient B. HER2 positive Breast Cancer: -7.88% SD, -12.50% SD, -14.67% SD; Heavily pretreated: 7 prior systemic regimens, including 5 HER2-directed therapies (prior T-DXd); Brain target lesion reduced; non-target lesions stable (SD per RANO-BM); Duration of treatment: 10.6 months (ongoing) Conclusion - Partial responses occurred in both TKD (2/5) and non-TKD (3/7) HER2-mutant tumors, Non-TKD mutation which lack an approved HER2 TKI. - Activity was observed despite prior T-DXd, received by 17 of 30 evaluable patients (Phase Ib-eligible population), and favorable brain penetration and CNS activity. - VRN10 was tolerable at 80-480 mg (any TRAE 76%, >= G3 6%); MTD was not reached. - Escalation continues to define the RP2D for the planned Phase 1b monotherapy and combination cohorts.

[Slide 2] [Livestream frame] 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic CHAIRS: PASI A. JANNE, XIUNING LE, EGBERT SMIT Time to deterioration in pulmonary symptoms (NSCLC-SAQ) Cough: Median TTD, months (95% CI): T-DXd (n=227) 26.2 (17.1, NC) | Pembro + chemo (n=227) 11.0 (7.4, NC); Hazard ratio (95% CI) 0.66 (0.44, 0.99) Chest pain: Median TTD, months (95% CI): T-DXd NC (11.5, NC) | Pembro + chemo 14.3 (8.3, NC); Hazard ratio (95% CI) 0.80 (0.54, 1.18) Dyspnea: Median TTD, months (95% CI): T-DXd 10.1 (6.5, NC) | Pembro + chemo 5.1 (3.2, 11.0); Hazard ratio (95% CI) 0.77 (0.54, 1.09) Y axis: Probability of no sustained deterioration; X axis: Time from randomization (months) 0-36 No. at risk - Cough: T-DXd 227 106 73 58 41 34 25 17 11 7 4 3 0; Pembro + chemo 227 92 54 38 21 12 10 7 6 4 3 2 0 No. at risk - Chest pain: T-DXd 227 98 70 55 37 30 22 15 10 8 7 4 0; Pembro + chemo 227 91 56 40 25 14 10 7 6 2 2 1 0 No. at risk - Dyspnea: T-DXd 227 85 62 50 36 29 23 18 11 9 7 5 0; Pembro + chemo 227 66 36 21 16 9 5 5 3 3 1 1 0 T-DXd monotherapy delayed time to, and reduced risk of, deterioration in cough versus pembro + chemo, with similar trends observed for chest pain and dyspnea Assessed in the full analysis set. Compliance rates for the NSCLC-SAQ: 67.8% T-DXd vs 71.9% pembro + chemo (baseline); 76.1% T-DXd vs 69.0% pembro + chemo (overall). Sustained deterioration was defined as a meaningful worsening at two or more consecutive post-baseline timepoints relative to baseline (regardless of whether the patient withdrew from treatment or received another anticancer therapy prior to deterioration). Patients with no sustained deterioration were censored at the time of their last PRO assessment with an evaluable score or at Cycle 1 Day 1 NC, not calculable; TTD, time to deterioration [Presenter video inset] Silvia Novello - First-line (1L) trastuzumab derux[tecan] (T-DXd) in patients (pts) with HER[2]-mutant (HER2m) NSCLC: Patient-reported outcomes (PROs) from DESTINY-Lung04 (DL-04) ALMERIA AUDITORIUM - HALL 9 Survival table Endpoint: Time to sustained deterioration, NSCLC-SAQ cough (also chest pain, dyspnea) Arm | N | Median | Median CI T-DXd - cough | 227 | 26.2 | 17.1, NC Pembro + chemo - cough | 227 | 11.0 | 7.4, NC T-DXd - chest pain | 227 | NC | 11.5, NC Pembro + chemo - chest pain | 227 | 14.3 | 8.3, NC T-DXd - dyspnea | 227 | 10.1 | 6.5, NC Pembro + chemo - dyspnea | 227 | 5.1 | 3.2, 11.0 HR: 0.66 (cough) HR CI: 0.44, 0.99 [Slide 3] [Livestream frame] 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic CHAIRS: PASI A. JANNE, XIUNING LE, EGBERT SMIT Overall treatment tolerability (PGI-TT) "In the last 7 days, how bothered were you by the side effects of your cancer treatment?" [Stacked bar chart: Weekly assessments (BL, weeks 1-12) and Q3W assessments (weeks 15-69, EOT); T-DXd n=226 (top), Pembro + chemo n=220 (bottom); Y axis: Proportion of patients bothered by treatment side effects (%)] Legend: Not at all; A little bit; Somewhat; Quite a bit; Very much During the first 12 weeks, the proportion of patient-weeks with high side-effect bother* was 13.4% in both arms (score of 'very much' or 'quite a bit') Proportions of patients reporting overall and high side-effect bother were similar between arms throughout treatment Timepoints after Week 69 through Week 105 have been excluded owing to low sample size (<10% of patients in the safety analysis set, either arm) Assessed in the safety analysis set. Compliance rates: 65.9% T-DXd vs 70.5% pembro + chemo (baseline); 75.6% T-DXd vs 72.9% pembro + chemo (overall). The PGI-TT assessed how a patient perceived the overall bother of cancer treatment-related side effects over the past week *Calculated as the number of weeks with high side-effect bother divided by the number of total weeks on treatment across all patients (restricted to first 12 weeks of treatment and excluding missing visits) BL, baseline [Presenter video inset] Silvia Novello - First-line (1L) trastuzumab derux[tecan] (T-DXd) in patients (pts) with HER[2]-mutant (HER2m) NSCLC: Patient-reported outcomes (PROs) from DESTINY-Lung04 (DL-04) ALMERIA AUDITORIUM - HALL 9 [Slide 4] [Livestream frame] 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic CHAIRS: PASI A. JANNE, XIUNING LE, EGBERT SMIT Conclusions - Median treatment duration was longer with T-DXd vs pembro + chemo (12.3 vs 7.1 months, respectively) - Time to deterioration in patient-reported pulmonary symptoms (cough, chest pain, dyspnea) favored T-DXd vs pembro + chemo - Patient-reported treatment tolerability was generally comparable between arms throughout treatment - Physical functioning was similar across arms through Week 18 and higher with T-DXd vs pembro + chemo at most timepoints from Week 20 onward, including at end of treatment - Overall and high side-effect bother were similar between arms - For the majority of symptoms measured, patient-reported symptom experience was generally better with T-DXd vs pembro + chemo DESTINY-Lung04 PRO data, alongside efficacy and safety results from the primary analysis, support T-DXd as a new 1L treatment option for patients with HER2-mutant NSCLC Scan the Quick Response (QR) code for: - Presentation slides - Plain language summary infographic Copies of materials obtained through the QR code are for personal use only and may not be reproduced without permission from the author This study was sponsored by AstraZeneca. In March 2019, AstraZeneca entered into a global development and commercialization collaboration agreement with Daiichi Sankyo for trastuzumab deruxtecan (T-DXd; DS-8201). Thank you to the patients and their families for their participation, and the study site staff for their contributions. Medical writing support was funded by AstraZeneca and provided by Rosanna Jackson, PhD (Helios Medical Communications, part of the Helios Global Group). [Presenter video inset] Silvia Novello - First-line (1L) trastuzumab derux[tecan] (T-DXd) in patients (pts) with HER[2]-mutant (HER2m) NSCLC: Patient-reported outcomes (PROs) from DESTINY-Lung04 (DL-04) ALMERIA AUDITORIUM - HALL 9

131P Prospective validation of artificial intelligence–assisted RECIST 1.1 in a phase 1 study of fetrastobart vedotin (Fetra-V) in patients with non-small cell lung cancer Elisa Fontana, Chong Duan, Michelle Saul, Paul Galette, Stephen Willingham, Jen-Tang Lu, Allison Wehr, Shivani Gupta, Mike Stecher, Marc Oliva, Lisle Nabell Conclusions - Results from this study show feasibility of using AI-assisted RECIST version 1.1 (RECIST 1.1) in a clinical trial - AI-assisted confirmed ORR and DCR were consistent with investigator-assessed outcomes - The concordance rate between investigator-assessed and AI-assisted read for patient-level confirmed best overall response and time point–level overall response was 84.0% and 72.5%, respectively - These findings support the reliability of AI-assisted response assessment and provide independent confirmation of efficacy observed with Fetra-V, supporting its continued evaluation in the phase 3 study PADL1NK-005 (NCT07144280) - AI-assisted read showed the potential to improve inter-reader reproducibility in tumor size measurements and significantly reduced RECIST 1.1 read time, improving fidelity and accelerating clinical trial conduct - Future work will evaluate volumetric lesion-level response metrics beyond RECIST 1.1 and associations with long-term outcomes of Fetra-V treatment Methods - C5851001 (NCT05208762) is a phase 1 study of Fetra-V as monotherapy in patients with advanced solid tumors and in combination with pembrolizumab in patients with metastatic or unresectable HNSCC or NSCLC (Figure 1) - This analysis included dose-escalation, dose-optimization, and dose-expansion cohorts that included patients with relapsed/refractory NSCLC independent of PD-L1 expression by TPS treated with Fetra-V monotherapy - In this study, AI-assisted RECIST 1.1 assessment was implemented as an exploratory endpoint - A total of 55 patients with NSCLC were treated with Fetra-V 1.5 mg/kg monotherapy - Both investigator read and AI-assisted read per RECIST 1.1 were included in this analysis - Data cutoff was November 22, 2025 Results - AI-assisted confirmed ORR and DCR were consistent with investigator-assessed outcomes in the phase 1 study (Table 1) – As of November 22, 2025, investigator-assessed confirmed ORR was 32.4% in PD-L1–positive and 0% in PD-L1–negative relapsed/refractory NSCLC; AI-assisted confirmed ORR was 29.7% and 5.6%, respectively - The concordance rate between investigator-assessed and AI-assisted read for patient-level confirmed best overall response and time point–level overall response was 84.0% (Table 2) and 72.5%, respectively (Table 3) - Results from this reader study showed that AI-assisted RECIST 1.1 read improved reproducibility between readers and reduced the time required for response assessment by 40% at baseline and 52% at follow-up time points (P<0.001) (Figure 3) Table 1. Investigator-Assessed and AI-Assisted Response by TPS and Histology [small-font cells; not reliably legible] Table 2. Concordance Rate: Patient-Level Confirmed Best Overall Response AI-assisted RECIST 1.1 | Investigator CR | PR | SD | PD | Total CR | 0 | 0 | 0 | 0 | 0 PR | 0 | 11 (22) | 1 (2) | 0 | 12 (24) SD | 0 | 1 (2) | 21 (42) | 2 (4) | 24 (48) PD | 0 | 0 | 4 (8) | 10 (20) | 14 (28) Total | 0 | 12 (24) | 26 (52) | 12 (24) | 50 (100) Overall agreement (95% CI), % | 84.0 (70.9-92.8) Cohen κ (95% CI) | 0.74 (0.58-0.91) Weighted κ (95% CI) | 0.84 (0.73-0.95) Table 3. Concordance Rate: Time Point–Level Overall Response AI-assisted RECIST 1.1 | Investigator CR | PR | SD | PD | Total CR | 0 | 0 | 0 | 0 | 0 PR | 1 (1) | 33 (22) | 8 (5) | 3 (2) | 45 (30) SD | 0 | 7 (5) | 50 (34) | 9 (6) | 66 (44) PD | 0 | 1 (1) | 12 (8) | 25 (17) | 38 (26) Total | 1 (1) | 41 (28) | 70 (47) | 37 (25) | 149 (100) Overall agreement (95% CI), % | 72.5 (64.6-79.5) Cohen κ (95% CI) | 0.57 (0.46-0.69) Weighted κ (95% CI) | 0.68 (0.57-0.79) Figure 3. AI-Assisted Read Improves Reproducibility and Efficiency Preliminary results from this ongoing reader study demonstrated that AI assistance (A) improved inter-reader reproducibility in tumor size measurements (in the illustrated case, all 3 readers accepted the AI-suggested contour in the AI-assisted mode, resulting in identical tumor measurements) and (B) significantly improved reader efficiency, reducing the time required for RECIST assessments by 40% at baseline and 52% at on-treatment follow-up compared with unassisted reading (both P<0.001). Presented at the ESMO Congress 2026 | October 23-27, 2026 | Madrid, Spain

Article Longitudinal ultrasensitive ctDNA monitoring for high-resolution lung cancer risk prediction James R.M. Black, Takahiro Karasaki, Charles W. Abbott, Bailiang Li, Selvaraju Veeriah, Maise Al Bakir, Wing Kin Liu, Ariana Huebner, Carlos Martinez-Ruiz, Piotr Pawlik, David A. Moore, Daniele Marinelli, Oliver Shutkever, Cian Murphy, Lydia Y. Liu, Charlotte Grieco, Karen Grimes, Fabio C.P. Navarro, Rachel Marty Pyke, Gabor Bartha, Kathleen C. Keough, Steven Dea, Neeraja Ravi, John Lyle, Jason Harris, Katherine D. Brown, Fiona H. Blackhall, Fatemah Hassani, Dean A. Fennell, Nicholas McGranahan, Jacqui A. Shaw, Christopher Abbosh, TRACERx Consortium, Allan Hackshaw, Mariam Jamal-Hanjani, Alexander M. Frankell, Sean M. Boyle, Richard O. Chen, and Charles Swanton SUMMARY Biomarkers accurately informing prognostic assessment and therapeutic strategy are critical for improving patient outcome in oncology. Here, we apply a whole-genome, tumor-informed circulating tumor DNA (ctDNA) detection approach to address this challenge, leveraging 1,800 variants across 2,994 plasma samples from 431 patients with non-small cell lung cancer (NSCLC) from the TRACERx study. We show that ultrasensitive ctDNA detection below 80 parts per million both pre- and postoperatively is highly prognostic, and combinatorial analysis of the pre- and postoperative ctDNA status identifies an intermediate risk group, improving disease stratification. ctDNA kinetics demonstrate clinical utility during adjuvant therapy, where patients that "clear" ctDNA during adjuvant therapy experience improved outcomes. Moreover, characterization of patterns in postoperative ctDNA kinetics reveals insights into the timing, risk, and anatomical pattern of relapses. By incorporating longitudinal ultrasensitive ctDNA detection, we propose a refined schema for guiding the stratification and treatment recommendations in early stage NSCLC. Cell 2025 Preoperative ctDNA status | Landmark ctDNA status | ctDNA risk | Potential clinical recommendations Preoperative ctDNA(-) -> Low: • Infrequent (e.g. annual) intrathoracic radiographic surveillance Stage I: de-escalation of surveillance acceptable • Systemic screening could be omitted in this group • Consider clinical trials for de-escalation of adjuvant therapy Preoperative ctDNA(+) -> Landmark ctDNA(-) -> Medium: • Infrequent (e.g 6 monthly) intra and extrathoracic radiographic surveillance Stage I: de-escalation of surveillance acceptable • Give adjuvant therapy where appropriate (clinical trials of de-escalation of adjuvant therapy could be considered) • High-cadence longitudinal ctDNA monitoring to detect impending relapse Preoperative ctDNA(+) -> Landmark weak ctDNA(+) -> High: • More frequent scanning (e.g. intra and extrathoracic radiographic surveillance 3-6 monthly) • Consider adjuvant therapy even in early-stage 1 disease • High-cadence longitudinal ctDNA monitoring through adjuvant therapy for clearance and thereafter Preoperative ctDNA(+) -> Landmark strong ctDNA(+) -> Very high: • More frequent scanning (e.g. intra and extrathoracic radiographic surveillance 3 monthly) • Consider adjuvant therapy even in early-stage 1 disease • Consider clinical trials for escalation of adjuvant therapy • High-cadence longitudinal ctDNA monitoring for treatment response Proposed ctDNA-guided patient care schema

In the EGFRmut Cohort Unselected for c-Met, Median PFS Was 10.9 Months and Median OS Was 16.2 Months in Patients Who Received Temab-A • At the June 02, 2026, data cutoff, median follow-up was 29.5 months (range, 0.4–34.8) • ORR was 63% (26/41) (95% CI, 46.9–77.9) ○ Median DOR was 9.7 months (95% CI, 8.3–13.5) Progression-Free Survival PFS | EGFRmut Overall (n=41) Events, n (%) | 39 (95) Median PFS, months (95% CI) | 10.9 (8.3–12.3) + Censored Overall (at risk) 41 40 40 37 36 34 31 29 29 26 20 16 15 10 9 7 4 4 4 3 3 3 3 3 2 2 2 1 1 0 0 0 0 0 0 Overall Survival OS | EGFRmut Overall (n=41) Events, n (%) | 31 (76) Median OS, months (95% CI) | 16.2 (11.6–19.9) + Censored Overall (at risk) 41 40 40 39 38 38 37 36 35 34 30 27 26 25 25 24 21 18 16 15 13 11 10 10 9 9 9 8 6 6 3 1 1 1 1 0 0 Data cutoff date: 02 June 2026. DOR, duration of response; EGFRmut, epidermal growth factor receptor mutated; OS, overall survival; PFS, progression-free survival; Temab-A, telisotuzumab adizutecan. Survival table Endpoint: PFS and OS (single arm, EGFRmut overall) Arm | N | Median | Median CI EGFRmut overall - PFS | 41 | 10.9 | 8.3–12.3 EGFRmut overall - OS | 41 | 16.2 | 11.6–19.9

[Slide 1] Study rationale • TIL therapy successful in advanced melanoma1, but preconditioning NMA chemotherapy and adjuvant IL2 limit the use in NSCLC • Carboplatin-Paclitaxel reduces tumour-induced elevated levels of circulating total myeloid cells [Diagram: Tumour excised → Tumour cells / Tumour-infiltrating Lymphocytes (TILs) → TILs Isolated → T-cell expansion (REP) → Preconditioning and T-cell infusion] Adapted from Svane, 2016 Use a triple approach to induce optimal anti-tumor effect 1 ChemoTx to reduce immunosuppressive cells in TME 2 TIL infusion 3 Checkpoint Ab to unleash full potential of TIL 1. Rohaan et al. NEJM 2022;387 [Slide 2] TRIAL DESIGN KEY ELIGIBILITY INCLUSION • Age ≥ 18 years, ECOG 0 or 1 • One lesion > 1 cm, primary or metastatic, amendable for TIL retrieval • Stable disease on ≥ 2 courses of first-line single-agents immunotherapy or chemoimmunotherapy, or progressive disease on or after first-line treatment • Asymptomatic CNS ≤ 1 cm before resection may be eligble Biopsy or resection for TIL retrieval → Cohort A 1st chemotherapy: 3 cycles carboplatin AUC 5 + paclitaxel 175 mg/m² i.v. TIL: 2.5–10 × 10^8 cells i.v. at weeks 0 and 3 → Cohort B 2nd chemotherapy: 3 cycles carboplatin AUC 5 + paclitaxel 175 mg/m² i.v. TIL: 2.5–10 × 10^8 cells i.v. at weeks 0 and 3 pembrolizumab: Once every 3 weeks 100 / 150 / 200 mg by body weight i.v. 1 crossover (Cohort B → Cohort A) Primary endpoints: Toxicity according to CTCAE 5.0 criteria. Secondary endpoints: PFS, OS, feasibility, translation immunological parameters [Slide 3] TOXICITY ← Chemotherapy-related (n=28) | ICI-related (n=3) → Chemotherapy-related (no. of patients): Anemia 5; Thrombopenia 4; Nausea 3; Neuropathy 3; Bronchial infection 1; Constipation 1; Dermatitis 1; Fatigue 1; Folliculitis 1; Hypomagnesiemie 1; Lymphopenia 1; Myalgia 1; Neutropenic fever 1; Pneumoniae 1; Pruritis 1; Pulmonary embolus 1; UTI 1 ICI-related (no. of patients): Hepatitis 1; Nephritis 1; Pruritis 1 Toxicity grade: Grade 1 | Grade 2 | Grade 3 | Grade 4 No. of patients • Mostly chemotherapy related toxicity • Grade ≥3 treatment related toxicity 29% Only 1 grade 4 toxicity (Thrombopenia) • 2 Patients with IO-related toxicity • 1 Patients had renal tubular acidosis gr 1 and hepatitis gr 3, occurring before TIL infusion. • No toxicity related to TIL infusion [Slide 4] EFFICACY Overall Survival [swimmer plot, patients 1–13, Cohort A / Cohort B; markers: Continued Response, Disease Progression, Lost to follow-up, Patient Alive, Patient Died, Subsequent systemic treatment; ↑ PDL1 >50%] Months Events | N | 12 Median OS | months [IQR months] | 8.0 [4.0 - 19.0] Median PFS | months [IQR months] | 4.0 [2.0 - 6.0] Best overall response PR | N (%) | 3 (25) SD | N (%) | 5 (42) PD | N (%) | 4 (33) ORR | % | 23 DCR | % | 69 Mean follow-up time for OS at data cut off 1-9-2026 8.1 months (1.2-20months)

[Slide 2] [Session banner] 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic CHAIRS: PASI A. JÄNNE, XIUNING LE, EGBERT SMIT PFS improved with osi + plat–pem versus osi mono irrespective of Week 1 and 2 plasma EGFRm clearance status Panels: Plasma EGFRm clearance at Week 1 | No Plasma EGFRm clearance at Week 1 | Plasma EGFRm clearance at Week 2 | No Plasma EGFRm clearance at Week 2 (y: Probability of PFS; x: Time from randomisation (months) 0-36) No. at risk: Clearance Wk1: osi+plat-pem 39 32 26 21 13 4 0; osi mono 55 46 37 22 14 7 0 No clearance Wk1: osi+plat-pem 65 55 49 30 15 8 0; osi mono 76 57 42 22 11 4 0 Clearance Wk2: osi+plat-pem 72 63 55 42 21 6 0; osi mono 95 74 56 31 10 9 0 No clearance Wk2: osi+plat-pem 41 35 27 22 10 6 0; osi mono 48 30 26 14 7 4 0 Week 1 table: mPFS (95% CI), months | Clearance | No clearance Osi + plat–pem | NR (18.3, NR) | 24.8 (16.8, 30.6) Osi mono | 16.6 (13.6, 21.0) | 13.6 (11.1, 16.4) HR (95% CI) | 0.58 (0.33, 1.02) | 0.53 (0.35, 0.80) Week 2 table: mPFS (95% CI), months | Clearance | No clearance Osi + plat–pem | 27.6 (19.7, NR) | 22.1 (11.8, 30.6) Osi mono | 13.9 (13.6, 19.4) | 13.8 (5.9, 16.8) HR (95% CI) | 0.53 (0.35, 0.81) | 0.57 (0.34, 0.96) Data cut-off: 03 April 2023 (PFS). HRs were calculated using an unstratified Cox proportional hazards model. Clearance was defined as undetected plasma EGFRm (Ex19del, L858R) at Week 1, 2, 3, or 6, where it was detected at baseline. [abbreviations footnote] David Planchard, MD, PhD ALMERIA AUDITORIUM - HALL 9 Survival table Endpoint: PFS by plasma EGFRm clearance status (Week 1 and Week 2) Arm | N | Median | Median CI Osi + plat-pem, clearance Wk1 | 39 | NR | 18.3, NR Osi mono, clearance Wk1 | 55 | 16.6 mo | 13.6, 21.0 Osi + plat-pem, no clearance Wk1 | 65 | 24.8 mo | 16.8, 30.6 Osi mono, no clearance Wk1 | 76 | 13.6 mo | 11.1, 16.4 Osi + plat-pem, clearance Wk2 | 72 | 27.6 mo | 19.7, NR Osi mono, clearance Wk2 | 95 | 13.9 mo | 13.6, 19.4 Osi + plat-pem, no clearance Wk2 | 41 | 22.1 mo | 11.8, 30.6 Osi mono, no clearance Wk2 | 48 | 13.8 mo | 5.9, 16.8 HR: Wk1 clearance 0.58; Wk1 no clearance 0.53; Wk2 clearance 0.53; Wk2 no clearance 0.57 HR CI: 0.33-1.02; 0.35-0.80; 0.35-0.81; 0.34-0.96

ESMO 2026 478P Early postoperative recurrence identifies a biologically aggressive KRAS/MAPK-enriched phenotype in cholangiocarcinoma Natalia S. Tissera 1,2, Andrea Casadei-Gardini 3, Virginia Genovesi 4,5, Anthony Turpin 6, Caterina Vivaldi 4,5, Cristina Dopazo 7, Maria Teresa Salcedo 8, Francesca Ratti 9, Silvia Buonanno 9, Remo Alessandris 9,10, Viviane Gnemmi 11, Emmanuel Boleslawski 12, Sharela Vega 1, Sergio Pérez-Fernandez 1, Ana Vivancos 13, Alejandro García 1, Jorge Hernando 1, Jaume Capdevila 1, Tian Tian 14,15, Teresa Macarulla 14,15 BACKGROUND Early recurrence (ER) after curative-intent resection for cholangiocarcinoma (CCA) is frequent and associated with [illegible] prognosis; however, its molecular determinants remain [illegible] METHODS Multicenter and retrospective study (2001–2025) 4 European centers RECURRENCE GROUPS ANALYTIC COHORT: 289 PATIENTS G3* Late recurrence ≥36 mo / no recurrence with ≥36 mo f-up 48 (16.6%) SURVIVAL OUTCOMES Median follow-up: 61.8 months (95% CI 48.6–85.0) Disease-free survival Median 10.8 months (95% CI 9.5–12.0) Overall survival Median 36.5 months (95% CI 30.9–42.0) Number at risk All 289 127 66 48 25 14 12 9 Number at risk All 289 166 64 39 19 12 OVERALL SURVIVAL BY RECURRENCE GROUP 12-month landmark sensitivity analysis: Recurrence ≤12 months was associated with subsequent mortality: adjusted HR 2.50 (95% CI 1.66–3.78), p<0.001. Median OS: G1 24.2, G2 38.8, G3 119.5 months G1 <12 months; G2 12-<36 months; G3 >=36/no recurrence Log-rank p<0.001 Months from surgery Number at risk G1 162 23 7 3; G2 79 38 10 3; G3 48 48 22 9 FIRST RECURRENCE ANATOMY Evaluable distant-component recurrences (n=161) Liver only | Lung only G1 40/94 (43%) | 4/94 (4%) G2 19/49 (39%) | 4/49 (8%) G3a 5/18 (28%) | 4/18 (22%) PATHOLOGIC PHENOTYPE G1 G2 G3 All BH-FDR: q <0.05 LN positive 71 (52%) | 19 (33%) | 3 (8.3%) pT3-4 63 (41%) | 25 (37%) | 4 (10%) R1 margin 58 (39%) | 18 (25%) | 7 (16%) Vascular invasion 103 (69%) | 43 (63%) | 17 (46%) Moderate/poor differentiation 137 (97%) | 56 (85%) | 29 (74%) WHOLE-COHORT MULTIVARIABLE COX MODEL LN involvement: DFS HR 2.02 (95% CI 1.33–3.08); OS HR 2.27 (95% CI 1.45–3.54) EXPLORATORY EXTREME-PHENOTYPE SUBGROUP: G1a (≤6 months) vs G3b LN positive 56% vs 11% q=0.003 pT3–4 38% vs 5% q=0.009 MOLECULAR FINDINGS 248/289 profiled (85.8%) KRAS: 57/232 (24.6%) KRAS/MAPK BY RECURRENCE GROUP G1 47/142 (33.1%) | G2 10/68 (14.7%) | G3 10/38 (26.3%) Global p=0.019 DFS altered vs WT Median DFS: 7.6 (6.2–11.5) vs 11.4 (10.3–13.8) months. Adjusted HR 1.25 (95% CI 0.85–1.84)* OS adjusted HR 1.01 (95% CI 0.61–1.67)* *Adjusted for LN status, tumour size, CCA subtype, margin status and adjuvant therapy. POST HOC ANALYSIS FOR KRAS/MAPK BY ANATOMICAL SUBTYPE eCCA 47/114 (41.2%) iCCA 20/134 (14.9%) Fisher p<0.001 EXTREME PHENOTYPES: G1a 23/51 (45.1%) vs G3b 2/9 (22.2%) Fisher p=0.28; BH q=1.00 Prespecified pathways: MAPK; tumour suppressor/cell-cycle; iCCA-enriched alterations; PI3K/PTEN. No non-MAPK pathway differed after FDR; no group×anatomy interaction (p=0.715). CONCLUSION ER identifies a clinically and biologically aggressive CCA phenotype with nominal KRAS/MAPK enrichment, warranting future studies evaluating the potential benefit of KRAS-targeted strategies.

Liquid Biopsy in Fusion-Driven NSCLC: Informative Rates, Resistance Detection, and Clinical Impact 2677P GUSTAVE ROUSSY Maxime BORGEAUD, Daniela MILIZIANO, Arianna MARINELLO, Voreak SUYBENG, Olfa TRABELSI-GRATI, Lodovica ZULLO, Marie MORFOUACE, Maud NGOCAMUS, Adrien ROUSSEAU, Marie PORTE, Pernelle LAVAUD, Pamela ABDAYEM, Maxime FRELAUT, Jordi REMON, Anas GAZZAH, Fabrice BARLESI, David PLANCHARD, Etienne ROULEAU, Antoine ITALIANO, Benjamin BESSE, Mihaela ALDEA Background • Liquid biopsy (LBx) enables non-invasive molecular profiling, yet its performance in fusion-driven non-small-cell lung cancer (NSCLC) remains poorly defined. • This study evaluated the performance of LBx in the detection of driver fusions and co-alterations at baseline and on tyrosine kinase inhibitors (TKI) progression. We assessed the yield of LBx to detect resistance mechanisms to TKI, and how often the results impacted patient management. Methods • Retrospective study at Gustave Roussy, January 2021– July 2026, including patients with advanced ALK-, ROS1-, RET- or NTRK-rearranged NSCLC and ≥1 FoundationOne Liquid CDx (324 genes) LBx. Baseline LBx = before 1st line TKI, Progression LBx = after ≥1 fusion-directed TKI. • A LBx was considered informative when ≥1 somatic alteration was detected (excluding germline and CHIP). Resistance alterations were classified as on-target, bypass*, secondary fusion or potential histologic switch** • Resistance detection rates were assessed for each driver, according to tumor fraction (TF). • Treatment proposals and adaptation, and best responses were retrieved from review of molecular tumor board (MTB) records and medical chart review. Results Patients and samples A total of 282 patients were included: 150 with ALK, 61 with ROS1, 66 with RET, and 5 with NTRK rearrangement. Median age was 58 years, 56% were women and 60% never smokers. In total, 109 baseline LBx (107 patients) and 280 progression LBx (200 patients: ALK 165 LBx/121 patients, ROS1 67/47, RET 48/32) were analyzed. TF was reported in 266 (95%) and was ≥1% in 98 (35%). Driver | Fusion detected, baseline | Fusion detected, progression | Informative, baseline | Informative, progression ALK | 22/37 (59%) | 102/165 (62%) | 26/37 (70%) | 120/165 (73%) ROS1 | 11/18 (61%) | 32/67 (48%) | 14/18 (78%) | 41/67 (61%) RET | 38/48 (79%) | 28/48 (58%) | 42/48 (88%) | 31/48 (65%) All* | 74/109 (68%) | 162/280 (58%) | 86/109 (79%) | 192/280 (69%) Table 1. Driver fusion detected and informative LBx (≥1 tumor-derived alteration), n/N (%). *Including NTRK (baseline only) Detection of resistance mechanisms A resistance mechanism was confirmed in 69/280 progression LBx (25%): 51% with TF ≥1% vs 10% with TF <1% or not reported, across drivers and TKIs (Figure 2). 77% of LBx without a mechanism had TF <1% or not reported. TF ≥1% was less frequent when the TKI was ongoing at LBx (30% vs 54%, p=0.001; n=172), but not after adjustment for metastatic site and ECOG (adjusted OR 0.41, 95% CI 0.15–1.06). No patient with brain-only (0/5) or intrathoracic-only disease (0/7) had TF ≥1%. TF ≥1% | TF <1% or not reported By driver | OR (95% CI) All drivers 19/182 | 50/98 | 8.9 (4.8–16.6) ALK 10/103 | 27/62 | 7.2 (3.1–16.3) ROS1 7/50 | 13/17 | 20.0 (5.0–79.1) RET 2/29 | 10/19 | 15.0 (2.8–81.7) By most recent TKI Lorlatinib (ALK) 3/50 | 16/31 | 16.7 (4.3–65.3) 2nd-gen. ALK TKI 6/32 | 7/13 | 5.1 (1.2–20.6) Selective RET TKI 1/27 | 8/16 | 26.0 (2.8–240.5) Progression LBx with a confirmed resistance mechanism Figure 2. Proportion of progression LBx with a confirmed resistance mechanism according to TF (n/N), by driver and most recent TKI; odds ratios with 95% CI Resistance mechanisms by driver On-target mutations predominated in ROS1 and bypass alterations in RET, whereas both were similarly frequent in ALK (Figure 3). The most frequent bypass alterations are shown in Figure 4. MET amplification was found in only 7 of 280 LBx (2.5%) with a median TF of 19.5 (1.2-36) in these samples, consistent with the limitation of LBx to detect CNVs. Figure 4. Bypass alterations detected in progression LBx, by driver (number of LBx) Treatment adaptation and response A mechanism-directed treatment was proposed after 53/69 LBx with a mechanism [obscured] given after 33 (62%), i.e. 11.8% of progression LBx (Figure 5); ORR was 45% (Figure 6). ORR was higher with adaptation in ALK, but low in ROS1 and RET regardless of adaptatio[obscured] (Figure 7). With a bypass alteration, ORR was 40% (4/10) with bypass-directed treatment vs 18% (3/17[obscured] with a fusion-directed TKI alone. ALK Resistance explained 37/165 (22.4%) Treatment proposed 28/165 (17.0%)

[Slide 1] Efficacy: Group 1 (≥2L NSCLC, monotherapy) CS2009 monotherapy demonstrated promising clinical efficacy in heavily pre-treated NSCLC | All Evaluable Patients | 30 mg/kg | Squamous, 30 mg/kg | Non-squamous, 30 mg/kg ORR, % (n) | 16.7% (9/54) | 28.0% (7/25) | 33.3% (4/12) | 23.1% (3/13) DCR, % (n) | 68.5% (37/54) | 60.0% (15/25) | 75.0% (9/12) | 46.2% (6/13) mDoR, month | 10.5 | 10.5 | - | - [Waterfall plot: best change from baseline, % by dose (10/20/30/45 mg/kg), prior anti-angiogenic, prior IO; spider plot: change from baseline, % vs Months] CS2009 is well tolerated with treatment duration exceeded 12 months at 10, 20, and 30 mg/kg. Dr. Anwen Xiong [Slide 2] Efficacy: Group 2 (2/3L NSCLC, in combination with docetaxel) CS2009 + docetaxel demonstrated promising preliminary clinical efficacy in IO-pretreated NSCLC | All Evaluable Patients | 20 mg/kg | 30 mg/kg ORR, % (n) | 55.6% (10/18) | 55.6% (5/9) | 55.6% (5/9) DCR, % (n) | 100% (18/18) | 100% (9/9) | 100% (9/9) [Waterfall plot: best change from baseline, % (20 mg/kg, 30 mg/kg), histology SQ/NSQ and PD-L1 (%) per patient; spider plot: change from baseline, % vs Months; Treatment ongoing] Dr. Anwen Xiong [Slide 3] Efficacy: Group 3 (1L NSCLC, monotherapy) CS2009 monotherapy demonstrated compelling clinical efficacy in first-line, PD-L1 TPS ≥1% NSCLC | All Evaluable Patients | 20 mg/kg | 30 mg/kg | Squamous | Non-squamous TPS ≥1% ORR, % (n) | 66.7% (32/48) | 58.3% (14/24) | 75.0% (18/24) | 69.0% (20/29) | 63.2% (12/19) TPS ≥1% DCR, % (n) | 93.8% (45/48) | 95.8% (23/24) | 91.7% (22/24) | 96.6% (28/29) | 89.5% (17/19) TPS ≥50% ORR, % (n) | 87.5% (21/24) | 76.9% (10/13) | 100% (11/11) | 91.7% (11/12) | 83.3% (10/12) TPS ≥50% DCR, % (n) | 95.8% (23/24) | 92.3% (12/13) | 100% (11/11) | 100% (12/12) | 91.7% (11/12) [Waterfall plot: Best change from baseline, %; TPS ≥1%; 20 mg/kg vs 30 mg/kg; per-patient PD/SD/PR labels, histology and PD-L1 (%) row] [Spider plot: Change from baseline, % vs Months; 20 mg/kg, 30 mg/kg, First New Lesion, Treatment ongoing; TPS ≥1%] Median DoR and median PFS were not reached due to short follow-up. Dr. Anwen Xiong DCR: disease control rate; DoR: duration of response; L: line; NSCLC: non-small cell lung cancer; NSQ: non-squamous; ORR: objective response rate; PD: progressive disease; PFS: progression-free survival; PR: partial response; SD: stable disease; SQ: squamous; TPS: tumor proportion score. [Slide 4] Efficacy: Group 4 (1L SQ NSCLC, in combination with paclitaxel and carboplatin) CS2009 + paclitaxel + carboplatin demonstrated compelling clinical efficacy in first-line squamous NSCLC | All Evaluable Patients | 20 mg/kg | 30 mg/kg | PD-L1 TPS <1% | PD-L1 TPS 1-49%* | PD-L1 TPS ≥50% ORR, % (n) | 80.8% (21/26) | 76.9% (10/13) | 84.6% (11/13) | 100% (8/8) | 66.7% (10/15) | 100% (2/2) DCR, % (n) | 100% (26/26) | 100% (13/13) | 100% (13/13) | 100% (8/8) | 100% (15/15) | 100% (2/2) * 14/15 with PD-L1 TPS ≤5% Note: One additional patient with unknown PD-L1 status achieved a PR. [Waterfall plot: Best change from baseline, %; 20 mg/kg vs 30 mg/kg; all SQ; per-patient SD/PR labels and PD-L1 (%) row] [Spider plot: Change from baseline, % vs Months; 20 mg/kg, 30 mg/kg, First New Lesion, Treatment ongoing] Median DoR and median PFS were not reached due to short follow-up. Dr. Anwen Xiong DCR: disease control rate; DoR: duration of response; L: line; NSCLC: non-small cell lung cancer; NSQ: non-squamous; ORR: objective response rate; PD-L1: programmed death-ligand 1; PFS: progression-free survival; PR: partial response; SD: stable disease; SQ: squamous; UK: unknown.

2622P Clinical and Genomic Landscape of de novo MET Amplified Non-Small Cell Lung Cancer (LC-SCRUM-Asia) Takanori Enomoto, Hiroki Izumi, Shingo Matsumoto, Naoki Furuya, Ichiro Nakachi, Shuji Murakami, Masahiro Kodani, Eiji Iwama, Hironori Ashinuma, Kazumi Nishino, Reiko Matsuzawa, Ryo Toyozawa, Jun Sugisaka, Mariko Okamoto, Shoichi Kuyama, Taku Nakagawa, Jun Sakakibara-Konishi, Kiyotaka Yoh, Koichi Goto Background ● De novo MET Amplification (Amp) is observed in approximately 1–5% of non-small cell lung cancer (NSCLC). ● In recent years, novel therapies for NSCLC with de novo MET high Amp have been developed. ● However, the clinical, pathological, and genomic features and treatment outcomes of NSCLC with de novo MET Amp, and the optimal next-generation sequencing (NGS)-based cutoff for defining MET high Amp remain unclear. Objective ● To investigate the clinical, pathological, and genomic features and treatment outcomes of NSCLC with de novo MET Amp. ● To determine the optimal NGS-based cutoff corresponding to a FISH MET/CEP7 ratio of ≥5 for defining MET high Amp. Materials and Methods ● We retrospectively analyzed patients with NSCLC from the LC-SCRUM clinical-genomic database (2013–2025). ● MET copy number variant (CNV) was analyzed by Oncomine Comprehensive or Precision Assay. MET Amp was defined as CNV ≥4. ● To establish an NGS-based cutoff for MET high Amp, linear regression analysis was performed on paired NGS and fluorescence in situ hybridization (FISH) results from the same tissue specimens. ● We evaluated clinical, pathological, and genomic features and treatment outcomes of NSCLC with de novo MET Amp. ● PD-L1 TPS was assessed by 22C3 assay and categorized as negative (<1%), low (1–49%), and high (≥50%). TMB was assessed by Oncomine Tumor Mutation Load Assay. Results Figure 1. Study flow chart Patients with lung cancer enrolled in LC-SCRUM, February 2013 – November 2025 N = 22,990 SCLC n = 1,220 NSCLC n = 21,770 Patients without NGS data n = 1,545 NSCLC with NGS results n = 20,225 NGS and FISH paired for MET Amp n = 848 (NGS-based cutoff validation cohort) Treatment-resistant NSCLC n = 2,088 Treatment-naive NSCLC n = 18,137 Patients with de novo MET Amp n = 194 (1.1%) (low 153, high 41) (Clinical, pathological, and genomic features evaluation cohort) 35 excluded - 18 had MET exon 14 skipping - 14 had EGFR mutations - 2 had ROS1 fusions - 1 had a HER2 mutation Patients without other driver alterations n = 159 (low 125, high 34) 35 excluded - 15 were BSC alone - 12 were not advanced - 8 received other treatments (5 received non-platinum chemotherapy; 2 received an investigational drug; 1 received a TKI) Patients treated with ICI±Platinum-Based Chemotherapy or Platinum-Based Chemotherapy as first-line treatment n = 124 (low 97, high 27) Patients treated with ICI+Platinum-Based Chemotherapy n = 41 (low 31, high 10); ICI Monotherapy n = 21 (low 17, high 4); Platinum-Based Chemotherapy n = 62 (low 49, high 13) Table 1. Patient Characteristics (All de novo MET Amp (CNV≥4) N = 194 | MET low Amp (4≤CNV<10) n = 153 | MET high Amp (CNV≥10) n = 41 | Non-MET Amp n = 17,922 | P value) Sex Male | 163 (84) | 129 (84) | 34 (83) | 11,716 (65) | < 0.01 Age, years Median (range) | 66 (28–89) | 66 (28–87) | 65 (37–89) | 69 (12–97) | < 0.01 Smoking Ever | 172 (89) | 135 (88) | 37 (90) | 13,047 (73) | < 0.01 Histology Adeno | 158 (81) | 128 (84) | 30 (73) | 13,364 (75) | < 0.01 Figure 2. Relationship Between FISH (Gold Standard) and NGS for Defining MET High Amp N = 95 (NGS CNV≥4 of 848 paired cases) y = 1.53x + 3.05, r = 0.56, P < 0.01 NGS MET high Amp (CNV≥10) vs FISH MET high Amp (MET/CEP7≥5): PPV 81% (17 / 21); NPV 72% (53 / 74); PPA 45% (17 / 38); NPA 93% (53 / 57); OPA 74% (70 / 95) Figure 3. Frequency of targetable gene alterations All de novo MET Amp (CNV≥4) N = 194: MET exon 14 skipping, 9%; EGFR mutation, 7%; ROS1 fusion, 1%; HER2 mutation, 1%; Negative, 82% MET high Amp (CNV≥10) n = 41: MET exon 14 skipping, 10%; EGFR mutation, 7%; Negative, 83% Figure 4. PD-L1 Expression and TMB by MET Amp Level A) PD-L1 status: High (≥50%) All de novo MET Amp 57%, MET low Amp 64%, MET high Amp 29%, Non-MET Amp 27% B) TMB, mutations/Mb median: All de novo MET Amp 7.6 (N = 17); MET low Amp 7.6 (n = 14); MET high Amp 34.8 (n = 3); Non-MET Amp 4.2 (n = 3,492) Figure 5. Progression-Free Survival (PFS) by First-Line Treatment A) ICI+Platinum-Based Chemotherapy: mPFS High 7.7, Low 7.8, All 7.8 months; HR (95% CI) 0.88 (0.40–1.94) vs Low reference; Log-rank P = 0.75 B) ICI Monotherapy: mPFS High 1.4, Low 6.3, All 6.3 months; HR 0.67 (0.19–2.41); Log-rank P = 0.54 C) Platinum-Based Chemotherapy: mPFS High 1.4, Low 4.4, All 3.5 months; HR 3.96 (1.98–7.91); Log-rank P < 0.01 Conclusions ● Compared with non-MET Amp NSCLC, de novo MET Amp was more frequent in males, ever-smokers and adenocarcinoma, and showed higher PD-L1 positivity and TMB. MET high Amp tended to show higher TMB than MET low Amp. ● With FISH as the reference, NGS CNV≥10, corresponding to a FISH MET/CEP7 ratio≥5, showed a high PPV; however, it may underestimate MET high Amp defined by FISH. ● Among patients receiving first-line treatment, MET high Amp was associated with poorer outcomes than MET low Amp with Platinum-Based Chemotherapy, whereas no significant difference in PFS was observed between the two groups following ICI+Platinum-Based Chemotherapy. Takanori Enomoto taenomot@east.ncc.go.jp

[Slide 1] Introduction The 5 Methyl Thio Adenosine Phosphorylase (MTAP) is involved in methionine salvage pathway, and its gene is often co-deleted with CDKN2A/p16INK4a gene, at 9p21.3 locus. MTAP deficiency prevalence in mesothelioma ranges from 40 to 60% depending on testing methods, histological subtypes and series MTAP loss of expression is promising marker: – for diagnosis, as a CDKN2A homozygous deletion (HD) surrogate marker of malignancy, – for prognosis, being previously reported as a poor prognostic factor – for prediction of PRMT5/MT2A inhibitors' response, Chromosome 9 9p21.3 Tel 5' | IFN gene cluster | MTAP | CDKN2A | CDKN2B | 3' Cen 222 kb 1A | 1B MAT2A, ATP+ Methinonine, SAM, Polyamine biosynthesis, Adenine+, MTR, MTA, MTAP MAT2A inhibitors, Cooperative PRMT5 inhibitors, Immune microenvironment, 9p21 loss Bray C et al. Frontiers in Oncology, 2023 , Brune MM et al. Lung cancer 2024, Gorbokon N et al. AJSP 2024, Davidson B et al. Appl Immunohistochem Mol Morphol 2024, Febres-Aldana et al. Modern Pathol 2024, Vrugt B et al. Cancers (Basel) 2023, Brcic L et al. Virchow Archiv. 2023 [Slide 2] Material 1.433 samples were retrieved from the MESOBANK between 2024 and 2026. All diagnoses (Epithelioid Mesothelioma, Biphasic M, Sarcomatoid M, Benign mesothelial lesions) were certified by NETMESO-MESOPATH and RENAPE pathologists. Clinical data (gender, age, asbestose exposure, OS), as well as histomolecular features were retreived for the MESOBANK database. 1433 samples 92 benign lesions 1152 PM (86%) | 189 PeM (14%) 784 EM (68% of PM) | 164 EM (86% of PeM) 182 BM (15,7%) | 18 BM (10%) 179 SM (15,5%) | 7 SM (4%) | Pleural samples n=1.232 | Peritoneal samples n=201 | Total n=1.433 Gender Male | 890 (88.6%) | 114 (11.4%) | 1004 (100.0%) Gender Female | 341 (79.7%) | 87 (20.3%) | 428 (100.0%) Age Mean (SD) | 76.817 (9.828) | 67.930 (15.392) | 75.570 (11.209) Age Median (Range) | 78.000 (29.000, 97.000) | 72.000 (21.000, 92.000) | 77.000 (21.000, 97.000) Age groups [0,65] | 136 (64.8%) | 74 (35.2%) | 210 (100.0%) Age groups [65,100] | 1096 (89.6%) | 127 (10.4%) | 1223 (100.0%) Vital status Dead | 572 (85.9%) | 94 (14.1%) | 666 (100.0%) Vital status Alive | 660 (86.0%) | 107 (14.0%) | 767 (100.0%) Asbestose exposure No | 140 (92.1%) | 12 (7.9%) | 152 (100.0%) Asbestose exposure Yes | 390 (97.0%) | 12 (3.0%) | 402 (100.0%) Asbestose exposure Unknwon | 702 (79.9%) | 177 (20.1%) | 879 (100.0%) [Slide 3] Results (cont.) : Is MTAP loss, as assessed by immunohistochemistry, a reliable surrogate marker for CDKN2A gene HD and malignancy? When comparing MTAP IHC with CDKN2A FISH : - The concordance rate was 88% - No benign lesions showed a loss of MTAP expression - The sensitivity, specificity, positive predictive value and negative predictive value of the MTAP IHC test were of 81%, 91%, 81% and 91% %, respectively N=108 (67 mesotheliomas and 41 benign lesions) | MTAP retained (by IHC) n=68 | MTAP loss (by IHC) n=32 | MTAP partial loss (by IHC) n=8 CDKN2A homozygous deletion (n=35) | 6 (5%) 2 EM 1 BM and 3 SM | 26 (24%) | 3 (3%) 3 EM CDKN2A heterozygous deletion (n=19) | 13 (12%) incl 4 benign lesions | 2 (2%) 1 SM and 1 EM | 4 (4%), 2EM and 2 BM No CDKN2A deletion (n=54) | 49 (45%) incl 37 benign lesions | 4 (4%) 1 EM, 2 BM, 1 SM | 1 (0,9%) 1 BM [Slide 4] Results (cont.): What is the clinical and pathological profile of MTAP loss in mesothelioma? A loss of MTAP expression was observed in 48% of PM; it was rare event in PeM (16%). It was predominantly observed in BM and SM subtypes. No correlation was observed with asbestos exposure, architecture or grade in EM subtypes, necrosis, inflammation, BAP1 expression or mutations. For pleural lesions: | MTAP loss (n=600) | MTAP retained (n=632) | Total (n=1232) | p value Gender Males | 459 (51.6%) | 431 (48.4%) | 890 (100.0%) | 0.001 Gender Females | 141 (41.3%) | 200 (58.7%) | 341 (100.0%) Age Mean (SD) | 78.427 (8.231) | 75.288 (10.922) | 76.817 (9.828) | < 0.001 Age Median (Range) | 79.000 (43.000, 97.000) | 77.000 (29.000, 95.000) | 78.000 (29.000, 97.000) Age groups [0,65] | 38 (27.9%) | 98 (72.1%) | 136 (100.0%) | < 0.001 Age groups (65,100] | 562 (51.3%) | 534 (48.7%) | 1096 (100.0%) Bap1 IHC 0 | 304 (47.2%) | 340 (52.8%) | 644 (100.0%) | 0.223 Bap1 IHC 1_2_3_4 | 296 (50.7%) | 288 (49.3%) | 584 (100.0%) Histology EM | 331 (42.4%) | 449 (57.6%) | 780 (100.0%) | < 0.001 Histology BM | 127 (69.8%) | 55 (30.2%) | 182 (100.0%) Histology SM | 131 (73.2%) | 48 (26.8%) | 179 (100.0%) Mutations | ARID1A/B, ATM, BAP1, LATS1/2, SETD2/DB1, NF2 | ATM, BAP1, LATS1/2, SETD2/DB1,BAP1, MTAP (1), NF2 Fusions | BAP1, CDKN2A (1) MTAP (5), NF2 | BAP1, MTAP (1), NF2

Study design Ph1a Dose Escalation & Backfill (mCRM & BLRM ) in Australia + China + United States DLT Observation Stage (Before MTD), DLT window: 21 days | No DLT Observation Key inclusion criteria: 1. Age ≥18 2. Locally advanced unresectable or metastatic solid tumor 3. Standard treatment does not exist, is no longer effective, or is not acceptable to the subject 4. ≥1 evaluable lesion per RECIST V1.1 5. ECOG PS 0 or 1 IBI3001 1mg/kg IV Q3W N=1 IBI3001 3mg/kg IV Q3W N=3 IBI3001 6mg/kg IV Q3W N=3 IBI3001 10mg/kg IV Q3W N=3 IBI3001 8mg/kg IV Q3W N=3+3 IBI3001 Evaluated dose groups Backfill* *Backfill enrollment will be focused on specific tumor types. Additional cohorts are permitted to add at the current, previously explored, or intermediate dose levels. If dose escalation is ongoing, backfill patients should be observed and recorded for DLTs. If the MTD has been determined or dose escalation has stopped, newly enrolled backfill patients will not undergo DLT observation All patients treated until PD, toxicity intolerance, withdrawal of informed consent or treatment duration reaches 24 months Primary Endpoint: dose escalation-safety, DLTs, RP2D; dose expansion-efficacy Secondary Endpoints: PK, immunogenicity Exploratory Endpoint: relationship between efficacy/safety and B7-H3/EGFR expression level • Dose Escalation: Intermediate dose between non-tolerated dose and preceding lower dose and altered treatment interval is permitted. From 6mg/kg dose, at least one patient in AU or CN is encouraged. • Dose Optimization: Two or more RDEs will be selected based on safety, PK and efficacy data to further determine safety, efficacy and RP2D. Randomization study will be done to obtain an optimal dose level at 1-2 selected indications. • Abbreviations: Modified Continual Reassessment Method (mCRM), Bayesian Logistic Regression Model (BLRM), Maximum Tolerated Dose (MTD), Dose-Limiting Toxicities (DLTs).

Poster 2529P Real-world treatment patterns and outcomes in patients with unresected stage I–II non-small-cell lung cancer: The TERESA study Solange Peters, Stephen V. Liu, Tatiana Shcherbakova, Yao Qiao, Ling Cai, Alice Wang, Jan Sieluk, Joe Y. Chang Objective • To describe treatment patterns and clinical outcomes among patients with unresected stage I–II NSCLC. Conclusions • In this contemporary real-world analysis, >40% of patients with stage I–II NSCLC did not undergo surgery; of these, ~88% received RT, most commonly stereotactic RT. • Use of systemic therapy as part of the initial treatment regimen was limited. For example, only ~6% of patients treated with stereotactic RT were recorded as also receiving systemic treatment as part of their initial treatment regimen. • Patients in the unresected cohort who received stereotactic RT had the most favourable outcomes, likely due to the survival benefit associated with SBRT, in addition to a higher proportion of stage I disease and T1 tumours relative to patients treated with other or no RT. • Even with stereotactic RT, >40% of patients were estimated to have experienced recurrence or death within 2 years. • These findings highlight a clear unmet need in unresected early-stage NSCLC and support evaluation of tolerable systemic therapies to complement SBRT and improve clinical outcomes in this setting. Introduction • Surgery remains the primary curative-intent Tx for early-stage NSCLC; however, ~30% of patients do not undergo definitive surgery due to medical inoperability or patient preference. • SBRT (also called SABR), defined as high-dose, precisely focused radiation typically delivered over 1–5 fractions, is the standard of care for patients with early-stage, lymph node–negative NSCLC who do not undergo surgery. • SBRT has greatly improved the management of early-stage NSCLC: it achieves excellent local control (≥90% at 5 years) with low rates of grade ≥3 toxicity and has enabled significantly more patients to receive curative-intent Tx. • Given technological advances in SBRT delivery over recent decades, existing studies of SBRT in unresected early-stage NSCLC may not fully reflect outcomes achieved with contemporary approaches. Consequently, contemporary data in this patient population remain limited. Methods • Eligible patients were identified from the US-based Flatiron Health EDM for early-stage NSCLC. Figure 1. Study design for the TERESA study Study population • Adults • Primary diagnosis of NSCLC • Stage I–II at initial diagnosis • No known ALK aberration • No other malignancies • With follow-up after NSCLC diagnosis Identification period Jan 2019 – May 2024; Data cutoff Nov 2024 Outcome measures • Tx patterns in unresected patients • TTRD in unresected patients • TTDRD in unresected patients • OS in unresected patients • Patient disposition and characteristics Figure 2. Patient selection in TERESA Adults with primary diagnosis of NSCLC (N=10189) -> Stage I–II at initial diagnosis (N=6958) -> Initial NSCLC diagnosis from Jan 2019–May 2024 (N=6713) -> No known ALK aberration (N=6665) -> With follow-up after NSCLC diagnosis (N=6654) -> No cancer diagnosis prior to NSCLC diagnosis (N=4921; FAS) -> With N0 nodal status (N=3942; N0 subgroup) Results and interpretation Disposition and local Tx patterns • Overall, 4921 patients with stage I–II NSCLC were eligible for inclusion in TERESA (Figure 3) • Of these, 40.4% comprised the unresected cohort – Among patients with unresected disease, 87.9% received RT – Stereotactic RT was the most common type of RT, accounting for 72.4% of RT-treated patients • Findings in the N0 subgroup were consistent with those observed in the FAS Figure 3. Patient disposition in the (A) FAS and (B) N0 subgroup: did not undergo surgery 40.4% | 38.9%; received RT 87.9% | 89.9%; received stereotactic RT 72.4% | 79.0% [arm n values illegible] Systemic Tx patterns • In the unresected cohort, 5.7%, 60.9%, and 24.5% of patients treated with stereotactic RT, other RT, and no RT, respectively, were recorded as receiving systemic Tx as part of their initial Tx regimen (Table 1) • In the N0 subgroup, proportionally fewer patients in the other RT and no RT cohorts were recorded as receiving systemic Tx. Clinical outcomes • In patients with unresected disease in the FAS and N0 subgroup, median follow-up (range) was 18.3 mo (0.2–70.8) and 19.3 mo (0.2–70.8), respectively. • More favourable outcomes were observed among patients who received stereotactic RT compared with those who received other or no RT (Figure 4A–C); however, even with stereotactic RT, 40.4% (calculated as 100% minus the 2-year probability of remaining free from recurrence or death; 59.6% [95% CI: 56.4–62.7]) of patients were estimated to have experienced recurrence or death within 2 years (Figure 4A). • When deaths were censored rather than counted as events, the overall pattern of results across groups was consistent with the main TTRD and TTDRD analyses (Figure 5A–B); 27.2% (calculated as 100% minus the 2-year probability of remaining free from recurrence; 72.8% [95% CI: 69.5–75.7]) of patients who received stereotactic RT were estimated to have experienced recurrence within 2 years (Figure 5A). • OS following recurrence was poor across unresected subgroups; median OS (95% CI) was 14.7 mo (12.9–19.3) with stereotactic RT, 11.7 mo (8.4–15.6) with other RT, and 20.5 mo (10.0–36.3) with no RT (Figure 5C). • In the N0 subgroup, clinical outcomes were consistent with those observed in the FAS (Figure 4D–F; Figure 5D–F). Table 1. Systemic Tx patterns in unresected patients in the FAS and N0 subgroup [values illegible] Table 2. Patient and disease characteristics in the FAS [values illegible] Figure 4. TTRD, TTDRD, and OS in the (A–C) FAS and (D–F) N0 subgroup [legend values illegible] Figure 5. TTR, TTDR, and OS following recurrence in the (A–C) FAS and (D–F) N0 subgroup [legend values illegible] Corresponding author email address: solange.peters@chuv.ch Poster presented at the European Society for Medical Oncology (ESMO), Madrid, Spain, 23–27 Oct 2026

[Slide 1] Study design • Open label, single-arm, three-cohort, non-comparative, Simon's two-stage design, phase II trial. Key inclusion criteria • Age ≥18 years • Histologically documented mBC (cohort 1), aNSCLC (cohort 2), or advanced solid tumors with LMD (cohort 3). • KPS ≥70%. • ECOG PS ≤2. • LVEF ≥50 %. • Newly diagnosed BMs or BMs progressing after local therapy (cohorts 1 and 2). • ≥1 measurable brain lesions (≥10 mm; cohorts 1 and 2). • ≥1 line of systemic treatment in the advanced setting (cohorts 1 and 2). COHORT 1 mBC with active BMs n = 21* COHORT 2 aNSCLC with active BMs n = 20 COHORT 3 LMD from solid tumors n = 22** First Patient In 12 Dec 2023 | Stage I 10 patients/cohort | Interim analysis 28 Oct 2024 | Stage II 10 patients/cohort | Primary analysis 28 Feb 2025 | End of study and final analysis 3 April 2026 HER3-DXd 5.6 mg/kg IV Q3W Until disease progression, unacceptable toxicity, or withdrawal for any reason * The last two patients from cohort 1 were enrolled simultaneously at different sites and both were finally included in the study. ** 20 patients were initially treated with HER3-DXd; however, 2 were subsequently determined not to meet the inclusion criteria. They were excluded from the full analysis set but included in the safety analysis set. 2 additional patients were enrolled. Primary endpoints • Intracranial ORR (cohorts 1 and 2). • 3-month OS rate (cohort 3). Key secondary endpoints • ORR, CBR, DCR, PFS, DoR, TTR for intracranial, extracranial, and bicompartmental lesions as per central review. • OS. • Safety and tolerability (CTCAE v.5.0). • Quality of life and neurocognitive function (EORTC QLQ-C30, and BN20). • Neurologic function (NANO scale). Key exploratory endpoints • Association between HER3 expression with efficacy endpoints. aNSCLC, advanced non-small cell lung cancer; BMs, brain metastases; CBR, clinical benefit rate; CTCAE v.5.0, Common Terminology Criteria for Adverse Events version 5.0; DCR, disease control rate; DoR, duration of response; ECOG PS, Eastern Cooperative Oncology Group Performance Status; EORTC QLQ, European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire; HER3, human epidermal growth factor receptor 3; HER3-DXd: patritumab deruxtecan; IV, intravenously; KPS, Karnofsky Performance Status; LMD, leptomeningeal disease; LVEF, left ventricular ejection fraction; mBC, metastatic breast cancer; NANO, Neurologic Assessment in Neuro-Oncology; ORR, overall response rate; OS, overall survival; PFS, progression-free survival; QW3, once every 3 weeks; TTR, time to response. [Slide 2] Overall survival Cohort 1: mBC with active BMs • Median follow-up: 6.7 months (range, 1.4-26.2). • Events: 81.0% (17/21). Median OS: 7.9 months (95% CI, 5.7-12.8) Number at risk (number censored) (months 0-24, every 3) 21 (0) 17 (1) 12 (1) 7 (2) 6 (2) 5 (2) 3 (2) 3 (2) 1 (4) BMs, brain metastases; CI, confidence interval; mBC, metastatic breast cancer; OS, overall survival. Survival table Endpoint: Overall survival (cohort 1, mBC with active BMs) Arm | N | Median | Median CI HER3-DXd, cohort 1 | 21 | 7.9 months | 5.7-12.8 [Slide 3] Overall survival Cohort 3: LMD from solid tumors • Median follow-up: 5.4 months (range, 0.8-23.1). • Events: 70.0% (14/20). Median OS: 10.5 months (95% CI, 2.1-13.1) Number at risk (number censored) (months 0-21, every 3) 20 (0) 13 (1) 10 (2) 8 (3) 5 (3) 4 (3) 3 (3) 3 (3) CI, confidence interval; OS, overall survival. Survival table Endpoint: Overall survival (cohort 3, LMD from solid tumors) Arm | N | Median | Median CI HER3-DXd, cohort 3 | 20 | 10.5 months | 2.1-13.1 [Slide 4] Conclusions • HER3-DXd demonstrated clinically meaningful intracranial activity in patients with active BMs of breast and lung cancers, and LMD of solid tumors. • With extended follow-up, median overall survival reached 7.9 months in cohort 1, 6.7 months in cohort 2, and 10.5 months in cohort 3, suggesting clinical benefit across all cohorts. • Neurological symptoms, quality of life, and neurocognitive function remained stable or improved during HER3-DXd treatment. • No new safety signals with HER3-DXd were observed. TUXEDO-3 supports the therapeutic value of HER3-DXd and reinforces the role of antibody-drug conjugates as a promising treatment approach in secondary CNS disease. BMs, brain metastases; CNS, central nervous system; HER3-DXd, patritumab deruxtecan; LMD, leptomeningeal disease.

[Slide 1] Study rationale and design • Platinum-based chemotherapy remains standard first-line treatment for advanced thymic carcinoma, with ORR 22–36% and median PFS 5.0–7.5 months.1,2 • Phase II studies support combining anti-angiogenic agents or ICIs with chemo3,4. • iTHYM is a multicenter, single-arm, phase II trial evaluating Ivonescimab (the first anti-PD1–VEGF bispecific antibody) combined with chemo in treatment-naïve advanced TC • Hypothesis: adding Ivo to chemo will result in a PFS6m of 60% Key eligibility criteria • Pathologically confirmed thymic carcinoma • Recurrent or metastatic disease • No prior systemic therapy • Age ≥18 years • ECOG PS 0 or 1 Primary: 6-month PFS rate Secondary: ORR, DCR, DoR, PFS, OS and safety Treatment Ivonescimab + carboplatin + paclitaxel / nab-paclitaxel Q3W, 4–6 cycles → Ivonescimab maintenance Until PD / intolerable toxicity Simon's two-stage design Stage 1 n = 8 At 6 months ≤3 PFS events → Stage 2 +16 patients Total N = 24 At 6 months >3 PFS events → Trial termination At 6 months ≤10 PFS events → Endpoint achieved ≥14/24 alive and progression-free at 6 months At 6 months >10 PFS events → Endpoint not achieved PFS event: progression or death from first dose. p0 = 30%, p1 = 60%, α = 0.05, power = 80% 1. Lemma, JCO 2011. 2. Hirai, Ann Oncol 2015. 3. Shukuya, Lancet Oncol 2025. 4. Proto, Ann Oncol 2024. [Slide 2] Efficacy Best tumor size change of target lesions: Median 43.9% (IQR, 30.8%-52.8%) PD-L1 TPS - <1% + 1–49% ++ ≥50% NA NA PR threshold −30% Best change from baseline (%) Median TTR: 1.7 months (IQR, 1.5-3.7) Change from baseline (%) PR SD PD Time from first dose (months) Time on Treatment PD-L1 PD Dose interrupt /discontinuation due to AE Ongoing Efficacy | N = 24 ORR | 20/24 (83.3%) PR | 20/24 (83.3%) SD | 4/24 (16.7%) DCR | 24/24 (100%) Median follow-up | 10.2 months (IQR:6.8-16.4) 6-month PFS rate* | 19/19 (100%) *Analysis in 19 patients. Maturity: 19/24 (79.2%). Data cutoff: July 31, 2026 ORR: objective response rate; DCR: disease control rate; PR: partial response; SD: stable disease TTR: time to response [Slide 3] Safety Safety summary | n/N (%) Any TRAE | 24/24 (100%) Grade ≥3 TRAE | 18/24 (75.0%) SAE acute coronary syndrome | 1/24 (4.2%) Permanent discontinuation due to TRAE / SAE | 1/24 (4.2%) TRAEs >15% All grades Grade ≥3 Neutrophil count decreased | Grade ≥3 13/24, 54.2% | 17/24, 70.8% Fatigue | 16/24, 66.7% Platelet count decreased | Grade ≥3 6/24, 25.0% | 15/24, 62.5% Anemia | 12/24, 50.0% Anorexia | 11/24, 45.8% Nausea | 9/24, 37.5% Constipation | 7/24, 29.2% Rash maculopapular | 7/24, 29.2% White blood cell decreased | Grade ≥3 3/24, 12.5% | 6/24, 25.0% Gingival pain | 6/24, 25.0% Oral hemorrhage | 6/24, 25.0% Peripheral sensory neuropathy | Grade ≥3 1/24, 4.2% | 5/24, 20.8% Cough | 5/24, 20.8% Diarrhea | 5/24, 20.8% Oral mucositis | 4/24, 16.7% Vomiting | 4/24, 16.7% Patients (%) irAE | Any grade, n (%) | Grade ≥3, n (%) Any irAE | 10 (41.7%) | 0 (0.0%) Rash | 7 (29.2%) | 0 (0.0%) Pruritus | 2 (8.3%) | 0 (0.0%) Adrenal insufficiency | 1 (4.2%) | 0 (0.0%) Hypothyroidism | 1 (4.2%) | 0 (0.0%) Possibly VEGF-related AE | Any grade, n (%) | Grade ≥3, n (%) Any antiangiogenic AE | 17 (70.8%) | 3 (12.5%) Oral mucositis | 4 (16.7%) | 0 (0.0%) Hemorrhagic events* | 9 (37.5%) | 0 (0.0%) Proteinuria | 3 (12.5%) | 2 (8.3%) Hypertension | 1 (4.2%) | 0 (0.0%) Gingival pain | 6 (25.0%) | 0 (0.0%) Creatinine increased | 2 (8.3%) | 0 (0.0%) Acute coronary syndrome | 1 (4.2%) | 1 (4.2%) Safety population N = 24. Highest CTCAE v5.0 grade per patient. DCO: July 31, 2026. TRAE: treatment-related adverse events; SAE: serious adverse events; irAE: immuno-related adverse events *Hemorrhagic events: oral hemorrhage and epistaxis [Slide 4] Conclusions • The iTHYM study met its primary endpoint of 6-month PFS rate in accordance with Simon's two-stage design. • Ivonescimab plus chemo demonstrated encouraging antitumor activity, with an ORR of 20/24 (83.3%) in advanced thymic carcinoma. • Toxicity was manageable, and predominantly chemotherapy-related. All irAEs were grade 1–2. No new safety signals were observed. • Follow-up remains ongoing to determine the final 6-month PFS rate. • These preliminary results support further evaluation of ivonescimab plus chemo as first-line treatment for advanced thymic carcinoma.

[Slide 1] Surgery or not? The MARS-2 earthquake MARS 2 trial schema • Tissue confirmed mesothelioma • Disease in one hemi-thorax • Deemed surgically resectable • PS (0-1) • No end-organ failure -> 2 cycles platinum + pemetrexed -> CT (still resectable) -> R -> (extended) pleurectomy decortication -> Up to 4 cycles platinum + pemetrexed / Up to 4 cycles platinum + pemetrexed [OS curve] Legend: No surgery, Surgery mOS 19.3 vs 24.8 months RMST at 24 months=-1.8 (95% CI -3.3 to -0.3) p=0.019 Number at risk (number censored) No surgery 166 (1) | 129 (4) | 82 (14) | 37 (7) | 15 (7) | 6 (6) Surgery 169 (1) | 115 (6) | 64 (14) | 24 (6) | 15 (5) | 7 (7) [PFS curve] mPFS 10.6 vs 11.0 months HR=0.90 (95% CI 0.72 to 1.11) p=0.33 Time since randomisation (months) Number at risk (number censored) No surgery 166 (1) | 78 (2) | 27 (4) | 9 (1) | 3 (1) | 1 (1) Surgery 169 (0) | 76 (5) | 33 (5) | 11 (1) | 6 (3) | 3 (3) [Subgroup OS curve] Epithelioid and no surgery; Epithelioid and surgery; Non-epithelioid and no surgery; Non-epithelioid and surgery RMST epithelioid=-1.5 (-3.2 to 0.2) RMST non-epithelioid=-4.2 (-8.6 to 0.3) p value for interaction=0.0020 Number at risk (number censored) Epithelioid and no surgery 144 (1) | 116 (4) | 73 (14) | 31 (7) | 12 (6) | 5 (5) Epithelioid and surgery 146 (1) | 105 (6) | 62 (13) | 24 (6) | 15 (5) | 7 (7) Non-epithelioid and no surgery 22 (0) | 12 (0) | 9 (0) | 6 (0) | 3 (1) | 1 (1) Non-epithelioid and surgery 23 (0) | 10 (0) | 2 (1) | 0 (0) | 0 (0) | 0 (0) Overall survival for T1-2 N0 epithelioid only cohort Legend: No surgery, Surgery Number at risk (number censored) No surgery 82 (1) 67 (3) 46 (11) 19 (4) 8 (5) 3 Surgery 78 (0) 58 (6) 39 (9) 15 (5) 7 (2) 3 [QoL panels: Median (IQR) QLQ-C30 score; MD=-5.81 (95% CI -9.73 to -1.89) p=0.0036; Median (IQR) EQ-5D score; MD=-0.14 (95% CI -0.19 to -0.10) p=0.0001 — small text, [?]] Paolo Bironzo, MD, PhD Lim E, Lancet Resp Med 2024 Survival table Endpoint: Overall survival / Progression-free survival (MARS-2) No surgery | 166 Surgery (extended pleurectomy decortication) | 169 HR: PFS HR=0.90 HR CI: 95% CI 0.72 to 1.11 P: PFS p=0.33; OS RMST at 24 months=-1.8 (95% CI -3.3 to -0.3) p=0.019 [Slide 2] MARS-2: some debated issues MARS 2 trial schema • Tissue confirmed mesothelioma • Disease in one hemi-thorax • Deemed surgically resectable • PS (0-1) • No end-organ failure -> 2 cycles platinum + pemetrexed -> CT (still resectable) -> R -> (extended) pleurectomy decortication -> Up to 4 cycles platinum + pemetrexed / Up to 4 cycles platinum + pemetrexed All histologies allowed (13% of non-epithelioid in surgical arm) PET/CT scan (& mediastinal staging) not mandatory Some patients underwent surgery even with PD as best response (9%) High rate of diaphragm and pericardial resections High rate of lung parenchymal invasion 9% 90 day mortality in the surgical arm Table S7 Surgical details | Received surgery (n=158*) Surgical procedure Extended pleurectomy/decortication | 139/157 (89%) Pleurectomy decortication | 13/157 (8%) Partial pleurectomy | 3/157 (2%) Exploration, no pleurodesis | 1/157 (1%) Other | 1/157 (1%) Resection and reconstruction Diaphragm resection | 130/157 (83%) Diaphragm reconstructed | 128/157 (82%) Pericardium resection | 105/157 (67%) Pericardium reconstructed | 84/157 (54%) [Remaining table rows (chest wall resection, completeness of resection, pleura and chest wall, focus, lung parenchymal invasion, mediastinal invasion, etc.) in small print — not transcribed reliably] Study design EORTC 1205 resectable pleural mesothelioma -> R 1:1 -> ARM A – immediate surgery (surgery -> chemotherapy (3 cycles)); ARM B – deferred surgery (chemotherapy (3 cycles) -> surgery) [Flow chart: 64 borderline resectable pleural mesothelioma; Extended pleurectomy/decortication -> Adjuvant chemotherapy vs Neoadjuvant chemotherapy -> Extended pleurectomy/decortication; boxes 'Completed in time', 'Alive, no progressive disease', 'No toxicity'; 70% vs 50%; inset 'mOS 33.8 months [?]', '90-day mortality 1.7% [?]'] Paolo Bironzo, MD, PhD Lim E, Lancet Resp Med 2024; Larkin J, Eur Resp J 2024 [Slide 4] What did we learn from 1st line ICIs trials? #1 ICIs-based treatments have stronger impact in non-epithelioid PM [Four OS Kaplan-Meier panels (nivolumab plus ipilimumab vs chemotherapy; chemotherapy vs chemotherapy plus pembrolizumab exploratory analyses non-epithelioid and epithelioid; subgroup curves) — embedded numbers too small to read reliably] #2 CT+ anti PD(L)1 leads to higher ORR [Bar chart, y-axis 0-100: CheckMate 743 exp | Checkmate 743 ctrl | IND227 exp | IND227 ctrl | DREAM3R exp | DREAM3R ctrl; legend: Global ORR, ORR in epithelioid, ORR in non-epithelioid] #3 Early progressions seems more common with double ICIs [PFS Kaplan-Meier panels: nivolumab plus ipilimumab vs chemotherapy (early crossing circled); chemotherapy vs chemotherapy plus pembrolizumab — embedded numbers too small to read reliably] Paolo Bironzo, MD, PhD Scherpereel A, J Clin Oncol 2026 Chu Q, Lancet 2023; Chu Q, ASCO 2023 Nowak AK, ESMO 2025

2583P Real-World Uptake and Characteristics of Patients Receiving Telisotuzumab Vedotin-tllv (Teliso-V) in Non-Small Cell Lung Cancer From a US Claims Database David Planchard1, Nicolas Girard2, Sumie Kakehi3, Sophia Ng3, Arun Prashar3, Rajesh Kamalakar3, Alexis Cortot4 OBJECTIVE To describe demographics and clinical characteristics of patients prescribed Teliso-V as well as treatment, prescriber, and payer data on Teliso-V utilisation CONCLUSIONS This is the first real-world (RW) study describing early uptake and characteristics of patients prescribed Teliso-V as a second-line or later (2L+) therapy Patients prescribed Teliso-V had high metastatic disease burden and existing comorbidities Patients prescribed Teliso-V had substantial prior treatment exposure; prior docetaxel or paclitaxel was the most common treatment, reported in 45% of patients INTRODUCTION • c-Met (also known as MET) protein is a transmembrane receptor tyrosine kinase involved in cell proliferation and survival; dysregulation of c-Met signalling pathway has been associated with tumourigenesis, resistance to therapy, and poor prognosis in patients with non-small cell lung cancer (NSCLC),1,2 where it is overexpressed in approximately 25% of patients3 • Teliso-V is a first-in-class c-Met protein–directed antibody-drug conjugate comprising a humanised monoclonal antibody telisotuzumab conjugated to the cytotoxic microtubule polymerisation inhibitor monomethyl auristatin E, via a cleavable valine-citrulline linker4 • Based on the LUMINOSITY study (NCT03539536), the US Food and Drug Administration (FDA) granted accelerated approval to Teliso-V for the treatment of adult patients with locally advanced/metastatic non-squamous NSCLC with high c-Met protein overexpression (3+, ≥50%), as determined by an FDA-approved test, who had received a prior systemic therapy5a • Here, we describe RW data on Teliso-V utilisation and characteristics of patients prescribed Teliso-V aThe FDA-recommended Teliso-V dose is 1.9 mg/kg, as an intravenous infusion every 2 weeks, until disease progression or unacceptable toxicity. Teliso-V Mechanism of Action METHODS • This US-based retrospective cohort study utilised Komodo Health (New York, NY) prescription and medical claims data – Komodo Health is an administrative claims database that contains de-identified, patient-level information on medical and pharmacy claims from a geographically diverse, large set of commercial and government (Medicare and Medicaid) electronic claims processors across the United States • Included patients were aged ≥18 years with ≥1 claim of 2L+ Teliso-V monotherapy on or after May 14, 2025,a who met longitudinal data availability requirements – ≥2 unique claims dates within 90 days of index Teliso-V, except when death occurred earlier – ≥2 unique claims dates within first year prior to index Teliso-V • For treatment history, additional criteria to support longer lookback were applied – ≥2 unique claims dates within second year prior to index Teliso-V • The following outcomes are reported using descriptive analyses: – Patient demographics and clinical characteristics – Treatment history prior to Teliso-V – Prescriber and payer data aDate of FDA accelerated approval of Teliso-V for the treatment of adults with locally advanced/metastatic nonsquamous NSCLC with high c-Met protein overexpression (3+, ≥50%). RESULTS Attrition Patient Disposition Index Teliso-V on or after May 14, 2025 N=212 Adults aged 18 years and older at index Teliso-V n=210 ≥2 unique claims dates within 90 days of index Teliso-V, except when death occurred earlier n=196 ≥2 unique claims dates within first year prior to index Teliso-V n=196 Treatment history: ≥2 unique claims dates within second year prior to index Teliso-V n=181 Patient, prescriber, and payer characteristics: 2L+ Teliso-V monotherapy n=178 2L+ Teliso-V monotherapy n=166 • A total of 178 patients with 2L+ Teliso-V monotherapy were included; 166 met the additional criteria for a longer lookback period for treatment history Patient Demographicsa | Overall (N=178) Age at index, years, median (range) 66.7 (38–88) Sex, n (%): Female 109 (61); Male 69 (39) Region, n (%): Midwest 44 (25); Northeast 36 (20); South 62 (35); West 36 (20) aRace and ethnicity data were unknown for all patients. Patient Demographics and Clinical Characteristics • Median age was 66.7 (range 38–88) years, and 61% of patients were female • Comorbidities included mild liver disease (33%), diabetes (28%), congestive heart failure (21%), and peripheral vascular disease (15%) • Among patients with ≥3 available Teliso-V claims (n=106), the median treatment interval for Teliso-V was 14 (interquartile range 14–17) days (data not shown) Patient Clinical Characteristicsa | Overall (N=178) Year of initial diagnosis, n (%): 2021 or earlier 42 (24); 2022 18 (10); 2023 40 (23); 2024 47 (26); 2025 31 (17) Year of first Teliso-V, n (%): 2025 150 (84); 2026 28 (16) Follow-up post Teliso-V, days, median (range) 85 (8–223) Year of 1L start, n (%): 2021 or earlier 31 (17); 2022 15 (8); 2023 37 (21); 2024 47 (26); 2025 48 (27) Metastatic disease at 1L, n (%) 152 (85) Brain metastases, n (%): At 1L 46 (26); At Teliso-V 84 (47) Liver metastases, n (%): At 1L 22 (12); At Teliso-V 46 (26) Mild liver disease, n (%) 58 (33) Diabetes, n (%) 49 (28) Congestive heart failure, n (%) 38 (21) Peripheral vascular disease, n (%) 26 (15) Charlson Comorbidity Score, mean (SD) 1.2 (1.5) aLookback period for comorbidities was Teliso-V claim to 1 year prior to earliest lung cancer diagnosis. 1L, first-line; SD, standard deviation; Teliso-V, telisotuzumab vedotin-tllv. Prior Exposurea | Overall (N=166) Prior platinum exposure, n (%): Yes 138 (83); No 28 (17) Prior taxane (docetaxel or paclitaxel) exposure, n (%): Yes 74 (45); No 92 (55) Prior IO exposure, n (%): Yes 128 (77); No 38 (23) Prior platinum + chemotherapy + IO exposure, n (%): Yes 93 (56); No 73 (44) Prior targeted exposure, n (%): EGFR 9 (5); ALK 3 (2); Multi-targeted kinase inhibitors 7 (4); c-Met 21 (13) aPrior lines of treatment are not mutually exclusive; patients may have received more than 1 prior treatment. ALK, anaplastic lymphoma kinase; EGFR, epidermal growth factor receptor; IO, immuno-oncology. Prescriber Characteristics • In patients with available prescriber data, most common prescriber type was physician (n/N=142/146 [97%]) Prescriber Type | Overall (N=178): Physician, n (%) 142 (80); Nurse practitioner, n (%) 4 (2); Unknown, n (%) 32 (18) Payer Characteristics • Most common payer type was Medicare (58%), followed by commercial (30%) and Medicaid (4%) Healthcare Plan Type | Overall (N=178): Commercial, n (%) 53 (30); Medicaid, n (%) 7 (4); Medicare, n (%) 104 (58); Missing, n (%) 14 (8) LIMITATIONS • Due to the open-claims nature of the dataset, enrolment information was unavailable; therefore, a longitudinal activity requirement was applied

#3825P SHR-7787, a DLL3-targeted bispecific T-cell engager, in relapsed small-cell lung cancer and neuroendocrine carcinoma: results from a phase 1/2 trial Shengxiang Ren, Jia Yu, Qiming Wang, Yan Zhang, Aiqin Gao, Lingxiang Liu, Chen Zhou, Yong Fang, Ruoyu Wang, Xiang Li, Yu Yao, Jie Li, Xin Wang, Yajie Zhu, Longhua Sun, Jinhua Wen, Xiao Gu, Chi Xu, Ke Ma, Wei Shi MADRID SPAIN 23-27 OCTOBER 2026 Background • Delta-like ligand 3 (DLL3), a protein that inhibits Notch signaling, is highly expressed on small-cell lung cancer (SCLC) and other neuroendocrine carcinomas (NECs). • T-cell–engaging bispecific antibodies simultaneously bind tumour-associated antigens and CD3 on T cells, thereby promoting T-cell recruitment and activation to induce tumor-cell killing. • SHR-7787 is a novel, subcutaneously administered, IgG1 bispecific antibody targeting DLL3 and CD3. We conducted a first-in-human, open-label phase 1/2 trial (NCT06605222) of SHR-7787 in relapsed SCLC and NEC and here report preliminary findings from phase 1. Methods • Phase 1 part included dose escalation and expansion (Fig. 1). Patients (pts) with SCLC and NECs who had progressive/recurrent disease after standard therapy were enrolled. • SHR-7787 (full dose [FD], 0.1–100 mg) was given s.c. using a fixed-dose or step-up dose (SUD) schedule. SHR-7787 at 10–60 mg was selected for expansion (all SUD). Key eligibility criteria • Histologically or cytologically confirmed relapsed/recurrent SCLC and other NECs • Had progressive/recurrent disease after standard treatment* • ECOG PS of 0 or 1 Primary objectives • Safety, tolerability, MTD, RP2D; Secondary objectives • PK profile, immunogenicity, preliminary efficacy Dose escalation schema: 0.1 mg, 1 mg (Accelerated titration); 6 mg, 10 mg, 20 mg, 30 mg, 40 mg, 60 mg, 100 mg (BOIN); s.c. Figure 1. Study design (phase 1) * Immunotherapy + platinum-based chemotherapy for SCLC (unless patient refusal or investigator decision); standard of care or platinum-based chemotherapy for other NECs. Results Patients • 78 pts (SCLC, n=74; other NECs, n=4) were enrolled and treated (Table 1). • As of data cutoff (Mar 14, 2026), median follow-up was 4.8 mo. Table 1. Baseline characteristics (10 mg (n=15) | 20 mg (n=14) | 30 mg (n=13) | 40 mg (n=15) | 60 mg (n=14) | All patients* (n=78)) Median age (range), y 57 (51–68) | 62 (46–72) | 60 (53–69) | 62 (48–70) | 62 (49–71) | 60 (46–72) Male 13 (86.7) | 12 (85.7) | 12 (92.3) | 13 (86.7) | 11 (78.6) | 66 (84.6) ECOG PS 1 15 (100.0) | 12 (85.7) | 11 (84.6) | 10 (66.7) | 10 (71.4) | 65 (83.3) Smoking history Never 6 (40.0) | 5 (35.7) | 4 (30.8) | 3 (20.0) | 4 (28.6) | 26 (33.3); Former/current 9 (60.0) | 9 (64.3) | 9 (69.2) | 12 (80.0) | 10 (71.4) | 52 (66.7) Brain metastasis 6 (40.0) | 4 (28.6) | 3 (23.1) | 3 (20.0) | 5 (35.7) | 21 (26.9) Liver metastasis 8 (53.3) | 2 (14.3) | 7 (53.8) | 6 (40.0) | 2 (14.3) | 29 (37.2) Clinical stage IV 12 (80.0) | 12 (85.7) | 9 (69.2) | 13 (86.7) | 11 (78.6) | 62 (79.5) Pathological type SCLC 14 (93.3) | 14 (100.0) | 11 (84.6) | 15 (100.0) | 14 (100.0) | 74 (94.9); LCNEC 0 | 0 | 1 (7.7) | 0 | 0 | 1 (1.3); NEC 1 (6.7) | 0 | 1 (7.7) | 0 | 0 | 3 (3.8) Prior line of Therapy 1: 5 (33.3) | 6 (42.9) | 5 (38.5) | 5 (33.3) | 5 (35.7) | 29 (37.2); 2: 6 (40.0) | 6 (42.9) | 6 (46.2) | 8 (53.3) | 7 (50.0) | 35 (44.9); ≥3: 4 (26.7) | 2 (14.3) | 2 (15.4) | 2 (13.3) | 2 (14.3) | 14 (17.9) Prior immunotherapy 14 (93.3) | 12 (85.7) | 11 (84.6) | 14 (93.3) | 14 (100.0) | 69 (88.5) Data are n (%) or otherwise indicated. LCNEC, large cell neuroendocrine carcinoma. * Pts treated at 0.1 mg (n=1), 1 mg (n=1), 6 mg (n=4), and 100 mg (n=1). Safety • 1 DLT occurred at 10 mg and 20 mg respectively, and MTD was still not reached. • TRAEs occurred in 76 (97.4%) pts and 22 (28.2%) had grade ≥3 TRAEs (Table 2). TRAEs led to dose discontinuation in 2 (2.6%) pts. • 21 (26.9%) pts experienced CRS (grade ≥3, n=2). 1 pt had grade 2 ICANS. All cases of CRS/ICANS occurred during the first cycle (Fig. 2). Figure 2. Incidence of CRS/ICANS over time (ASTCT Grade1–2 | Grade ≥3): Cycle 1: SUD CRS 24.0 | 1.3; ICANS 1.3 | 0; Cycle 1: FD CRS 7.2 | 1.4; ICANS 0 | 0; ≥Cycle 2 CRS 0 | 0; ICANS 0 | 0. All CRS/ICANS cases are included; denominators are the numbers of pts treated during the corresponding dosing period. Table 2. TRAEs (All patients (n=78): Any grade | Grade ≥3) Any TRAE 76 (97.4) | 22 (28.2) Anaemia 44 (56.4) | 2 (2.6) Injection site reaction 36 (46.2) | 0 Lymphocyte count decreased 28 (35.9) | 14 (17.9) Hypoalbuminaemia 28 (35.9) | 0 Hyperglycaemia 22 (28.2) | 0 CRS 21 (26.9) | 2 (2.6) Hyponatraemia 21 (26.9) | 0 Hypokalaemia 20 (25.6) | 3 (3.8) Alanine aminotransferase increased 19 (24.4) | 1 (1.3) Hypocalcaemia 19 (24.4) | 1 (1.3) Aspartate aminotransferase increased 18 (23.1) | 1 (1.3) Platelet count decreased 17 (21.8) | 3 (3.8) Hypomagnesaemia 17 (21.8) | 1 (1.3) Hypoproteinaemia 16 (20.5) | 0 Decreased appetite 15 (19.2) | 0 White blood cell count decreased 14 (17.9) | 1 (1.3) Blood lactate dehydrogenase increased 12 (15.4) | 0 Gamma-glutamyltransferase increased 12 (15.4) | 0 Data are n (%). TRAEs occurring in ≥15% of patients are listed. Efficacy • In 69 evaluable SCLC pts, ORR was 34.8% and DCR was 58.0% (Fig. 3). Higher response rate was observed in higher expansion dose. The ORR and DCR were 46.7%/86.7% and 50.0%/90.0% at 40 mg and 60 mg, respectively (Table 3). Consistent activity was observed in SCLC with brain metastatic lesions (ORR, 42.1% [8/19]; Fig. 4). • 3 (75.0%) of 4 pts with other NECs had a response (Fig. 3). Table 3. Tumor response in efficacy evaluable set (SCLC: 10 mg (n=14) | 20 mg (n=14) | 30 mg (n=11) | 40 mg (n=15) | 60 mg (n=10) | Total (n=69); All patients (n=73)) ORR* (95% CI), % 28.6 (8.4–58.1) | 28.6 (8.4–58.1) | 36.4 (10.9–69.2) | 46.7 (21.3–73.4) | 50.0 (18.7–81.3) | 34.8 (23.7–47.2) | 37.0 (26.0–49.1) DCR (95% CI), % 28.6 (8.4–58.1) | 42.9 (17.7–71.1) | 54.5 (23.4–83.3) | 86.7 (59.5–98.3) | 90.0 (55.5–99.7) | 58.0 (45.5–69.8) | 60.3 (48.1–71.5) TTR (IQR), mo 1.4 (1.4–2.1) | 1.4 (1.4–1.5) | 2.1 (1.4–2.8) | 1.4 (1.4–1.6) | 1.4 (1.2–1.4) | 1.4 (1.4–1.6) | 1.4 (1.4–1.6) * Includes confirmed and unconfirmed responses. DCR, disease control rate; ORR, objective response rate; TTR, time to response. Figure 3. Best change from baseline in target lesions in evaluable patients. Figure 4. Best change from baseline in target lesions in evaluable SCLC patients with brain metastasis. PK • SHR-7787 exposure increased approximately dose-proportionally, with a median Tmax of about 72 h and a t1/2 near 10 d. • PK and cytokine dynamics supported the two-SUD regimen. Conclusions • SHR-7787 showed encouraging activity with a manageable safety profile in relapsed SCLC and NEC. • Dose expansion is ongoing to identify RP2D. Funding: The study was sponsored by Jiangsu Hengrui Pharmaceuticals. Contact: Prof. Shengxiang Ren

Northwestern Medicine Characterizing Hallucinations in Large Language Models (LLMs) Using Real-World Oncology Cases for Clinical Decision Support Jiyon Lyu1, Trie Arni Djunadi2, Akhil Deepak Vatvani3, Shinkyo Yoon1, Sung Mi Yoon4, Daniel Sanghoon Shin5, Joongheum Park6, Jungmin Kim7, Ronald Min1, Jiyoon Unzain-Lim7, Sangmi Yoo8, Young Kwang Chae1 FPN: 224P INTRODUCTION - Despite the increasing use of AI in clinical decision support, hallucinations remain a major safety concern, and existing evaluations often lack real-world patient data. - This study aimed to characterize hallucinations across multiple LLMs in AI-assisted clinical decision-making using real-world clinical data and preprocessed NCCN guidelines. METHODS - An IRB-approved retrospective study using de-identified real-world data from 20 patients with NSCLC in a first-line decision-making scenario. - Accuracy was defined as concordance with NCCN Guidelines (V3.2025) and evaluated at the sentence level by two independent blinded reviewers. 20 patients with NSCLC in a first-line decision-making scenario 5 LLMs General-purpose: ChatGPT, Gemini, Grok Clinical support: AvoMD, ChatGPT for Clinicians GPT-5.5 Thinking ChatGPT for clinicians Gemini 3.1 Pro Grok 4.2 Reasoning AVO All responses were collected on May 12, 2026 Staging, Summarization Expert group Figure 1. A schematic overview of the study CONCLUSION - Beyond Accuracy: High accuracy does not guarantee hallucination-free clinical decision support. LLM outputs should not be used as standalone clinical judgments, and human oversight remains essential. - Defining Potentially Harmful Errors: Objective, quantitative metrics are needed to assess the potential harm of LLM-generated errors, particularly in clinical management recommendations. RESULTS Table 1. Comparative Accuracy and Confidence-Accuracy Relationships Across LLMs | Accuracy of Staging | Accuracy of Summarization* | Proportion of Confidence-High in Summarization** | Correlation with Confidence-High and Accuracy of Summarization Gemini | 92.5 % (95% CI: 78.5-98.0) | 71.5 % (95% CI: 50.2-92.8) | 0.96 (95% CI: 0.92-1.00) | -0.04, N.S. Grok | 95.0 % (95% CI: 81.8-99.1) | 92.8 % (95% CI: 85.5-100) | 0.86 (95% CI: 0.81-0.92) | -0.01, N.S. ChatGPT | 94.9 % (95% CI: 81.4-99.1) | 88.0 % (95% CI: 76.4-99.6) | 0.78 (95% CI: 0.73-0.82) | -0.15, N.S. AvoMD | 94.7 % (95% CI: 80.9-99.1) | 97.7 % (95% CI: 94.2-100) | 0.56 (95% CI: 0.45-0.68) | 0.81, p=0.02 ChatGPT for Clinicians | 97.5 % (95% CI: 85.3-99.9) | 90.3 % (95% CI: 81.3-99.4) | 0.74 (95% CI: 0.71-0.76) | 0.20, N.S. Figure 2. Accuracy and Confidence Levels Across Five LLMs. (A) Staging and patient summarization accuracy across five LLMs. (B) Proportion of high-confidence sentences across five LLMs. Confidence levels (high, moderate, or low) were predefined in the prompt. Table 2. Input Length-Accuracy Relationships Across LLMs | Correlation with Word Count of Input and Accuracy of Staging | Correlation with Word Count of Input and Accuracy of Summarization Gemini | 0.21, N.S. | 0.76, p=0.004 Grok | 0.21, N.S. | 0.68, p=0.015 ChatGPT | -0.36, p=0.02 | -0.51, N.S. AvoMD | -0.07, N.S. | 0.38, N.S. ChatGPT for Clinicians | 0.08, N.S. | -0.62, p=0.03 Input length was negatively correlated with accuracy in ChatGPT (r=-0.36) and ChatGPT for Clinicians (r=-0.62) but positively correlated with accuracy in Gemini (r=0.76) and Grok (r=0.68) (all p <0.05). Table 3. Characterization of Hallucinations Across LLMs | Main Hallucination Type | Representative Cases | Potential Clinical Harm Gemini | Fabrication | Invented demographics, smoking history, pathology, or molecular markers | Severe Grok | Contextual | Recommended NGS based on guideline logic despite not being supported in the provided case context | Moderate ChatGPT | Overinterpretation | Inferred treatment beyond directly cited evidence | Moderate AvoMD | Contextual | Imported information from outside the relevant note context (from wrong date) | Mild Table 3. Continued ChatGPT for Clinicians | Contextual | Added treatment implications not directly supported by the attached guideline | None Omission errors were consistently observed across all five LLMs, particularly in molecular/PD-L1 data extraction, reconciliation of conflicting smoking histories, and assessment of comorbidities. Discussion 1. High accuracy ≠ Clinical safety. Even high-performing LLMs generated clinically significant hallucinations, underscoring the need for human oversight. 2. High confidence ≠ High reliability. Confident but incorrect responses may pose substantial risks to clinical decision-making. 3. Input length-accuracy associations are model-dependent. Longer inputs correlated positively with accuracy in some LLMs and negatively in others. Footnote * Accuracy of Summarization: Proportion of correct responses across seven predefined categories of patient information. ** Proportion of Confidence-High: Proportion of sentences assigned a high confidence level among all evaluated sentences. Disclosure Statement: Jiyon Lyu, No conflicts of interest to declare Funding: No external funding Contact Information Jiyon Lyu, jiyon.lyu.md@gmail.com Young Kwang Chae, ychae@nm.org

[Slide 1] ABBV-706 FIH phase 1 study design and published results Phase 1 R/R ES-SCLCa • Prior platinum-etoposide chemotherapy and/or immuno-oncology regimens • Prior treated or untreated stable brain metastases were eligible • History of ILD/pneumonitis requiring treatment or active disease were excluded Total monotherapy patient population: ABBV-706 1.3–3.5 mg/kg (N=124) 1.8 mg/kg RP2D for monotherapy ABBV-706 1.8 mg/kg as 2L+ (n=41) IV Q3W in 21-day cycles until disease progression or unacceptable toxicity Key Endpoints Safety, ORR, PFS, OS, SEZ6 by IHC Key Methods1 • Archival or fresh tissue biopsy at screening • SEZ6 IHC (AbbVie proprietary assay) • Plasma ctDNA and serum proGRP were collected at baseline, start of each cycle 1–4, odd cycles thereafter, and end of treatment • The Guardant INFINITY™ ctDNA assay (Guardant Health) was used to monitor methylation-based tumor fraction responses • proGRP was used to monitor tumor response using the Roche Cobas® electrochemiluminescence immunoassay (LabCorp) Patient Characteristics 1.8 mg/kg (n=41)1 Median age, years (range) 64 (48–85) Median prior LoT, n (range) 2 (1–6) Brain mets at baseline 39% (n=16) Chemo-free interval 61% (<90 days) 27% (<30 days) Results 1.8 mg/kg (n=41)1 Safety Grade ≥3 All TRAEs 88% (n=36) Anemia 42% (n=17) Neutropenia 20% (n=8) Efficacyb ORR 56% (n=23) mPFS 6.4 months (4.0–8.1) mOS 12.4 months (8.2–17.3) 1. Byers LA, et al. Nat Med. 2026;32(8):2898-2908. aClinicalTrials.gov. NCT05599984. bSimilar efficacy was observed in the ITT and biomarker eligible populations. 2L+, second-line and later; chemo, chemotherapy; ctDNA; circulating tumor DNA; ES-SCLC; extensive stage small cell lung cancer; FIH, first-in-human; IHC, immunohistochemistry; ILD, interstitial lung disease; ITT, intention to treat; IV, intravenous; LoT, lines of treatment; mets, metastases; mOS, median overall survival; mPFS, median progression-free survival; ORR, overall response rate; OS, overall survival; PFS, progression-free survival; proGRP, pro-gastrin releasing peptide; Q3W, every 3 weeks; RP2D, recommended phase 2 dose; R/R, relapsed/refractory; SEZ6, seizure-related homolog protein 6; TRAE, treatment-related adverse event. [Slide 2] SEZ6: A compelling ADC target with high and stable expression in SCLC In the FIH study, tumors expressing SEZ6 are prevalent in patients with SCLCa SEZ6 IHC (FIH) Weak / 1+ Moderate / 2+ Strong / 3+ Normal lung (93% above a 1% [1+ cutoff] and median H-score of 145)1 SEZ6 expression in SCLC is not affected by prior treatment or biopsy from primary cancer or metastases SEZ6 expression at diagnosis and ≥1 prior LoT % Positive SEZ6 IHC Tx-Naïve (n=63) ≥1 prior LoT (n=45) P=0.163 (two-sided Wilcoxon rank-sum test) SEZ6 expression in primary and metastatic tumor tissue Primary (n=32) Metastatic (n=55) P=0.180 (two-sided Wilcoxon rank-sum test) 1. Byers LA, et al. Nat Med. 2026;32(8):2898-2908. aSEZ6 IHC was performed on all SCLC tumors collected at screening n=108 8 ADC, antibody-drug conjugate; FIH, first-in-human; IHC, immunohistochemistry; LoT, line of treatment; SCLC, small cell lung cancer; SEZ6, seizure-related homolog protein 6; Tx, treatment. [Slide 3] SEZ6 expression is not associated with hematological TRAEs or response • Preliminary data from the phase 1 study of ABBV-706 monotherapy showed a manageable safety profile1 • Most common Grade ≥3 TRAEs in patients with SCLC treated with ABBV-706 1.8 mg/kg Q3W were hematological, including anemia and neutropenia • SEZ6 expression did not correlate with development of any grade anemia or neutropeniaa • SEZ6 expression was comparable between respondersb and nonresponders to ABBV-706 treatment, supporting a biomarker-unselected approacha Hematological Toxicity SEZ6 Expression in Patients With Anemia Grade ≤2 (n=19) Grade ≥3 (n=16) SEZ6 Expression in Patients With Neutropenia Grade ≤2 (n=28) Grade ≥3 (n=7) Response SEZ6 Expression in Nonresponders vs Responders Nonresponder (n=14) Responder (n=21) % Positive SEZ6 IHC 1. Byers LA, et al. Nat Med. 2026;32(8):2898-2908. aSEZ6 IHC correlation with ORR and safety in 35 evaluable tumor samples. bResponders defined as confirmed CR + PR. CR, complete response; IHC, immunohistochemistry; ORR, overall response rate; PR, partial response; Q3W, every 3 weeks; SCLC, small cell lung cancer; SEZ6, seizure-related homolog protein 6; TRAE, treatment-related adverse event. [Slide 4] ctDNA clearance is associated with robust improvements in PFS and OS • ABBV-706 induces rapid ctDNA reduction: ≥50% in 77% (n/N=24/31) and ≥95% in 52% (n/N=16/31) of patients by C3D1 • Complete ctDNA clearance observed in 29% of patients shows a significant trend toward improvements in PFS and OS PFS associated with ctDNA OS associated with ctDNA CC (n=9) MR (n=15) Non-MR (n=7) PFS probability OS probability Time (months) | PFS: Non-MR (n=7) | MR (n=15) | CC (n=9) | OS: Non-MR (n=7) | MR (n=15) | CC (n=9) Total patients N=31 Median outcome, months (95% CI)a | 3.6 (2.6–NR) | 6.2 (4.0–9.5) | 13.6 (8.4–NR) | 6.7 (3.6–NR) | 10.1 (8.2–NR) | NR (NR–NR) Rapid ctDNA reduction was sustained and was associated with improved outcomes aSimilar data were observed with 95% cutoff for ctDNA reductions, as a predictor of OS and PFS. Non-MR: ctDNA reduction <50%; MR: ctDNA reduction ≥50% and <100%; CC: 100% ctDNA reduction. Responder defined as PR+CR. C, Cycle; CC, complete clearance; CI, confidence interval; CR, complete response; ctDNA, circulating tumor DNA; D, day; mPFS, median progression-free survival; MR, molecular responder; NR, not reached; OS, overall survival; PFS, progression-free survival; PR, partial response. Survival table Endpoint: PFS and OS by ctDNA response (N=31) Arm | N | Median | Median CI Non-MR (PFS) | 7 | 3.6 | 2.6–NR MR (PFS) | 15 | 6.2 | 4.0–9.5 CC (PFS) | 9 | 13.6 | 8.4–NR Non-MR (OS) | 7 | 6.7 | 3.6–NR MR (OS) | 15 | 10.1 | 8.2–NR CC (OS) | 9 | NR | NR–NR

ESMO Congress, October 23-27, 2026 2656P · MCLA-129 in patients with EGFR-mutant advanced non-small cell lung cancer (NSCLC) with MET overexpression and/or amplification: Results from a phase II study Qing Zhou*, Tongmei Zhang, Huijuan Wang, Xiaorong Dong, Jianya Zhou, Bo Jin, Yan Yu, Haiyong Wang, Ying Liu, Yang Xia, Lin Wu, Bolin Chen, Jiuwei Cui, Junguo Lu, Yanhong shang, Yu Yao, Xiaoming Hou, Guanming Jiang, Li Mao, Lieming Ding Background • MET overexpression and amplification are important mechanisms of acquired resistance to EGFR tyrosine kinase inhibitors (TKIs) in patients with EGFR-mutant advanced non-small cell lung cancer (NSCLC). • MCLA-129 is a human EGFR/c-MET bispecific antibody that demonstrated clinically meaningful antitumor activity in a phase I study (NCT04930432) in patients with advanced NSCLC harboring MET overexpression and/or amplification, with higher MET gene copy number appearing to be associated with greater activity. • Here, we report updated efficacy and safety results from an ongoing phase II study (NCT06885840) of MCLA-129 in patients with EGFR-mutant advanced NSCLC with MET overexpression and/or amplification after standard treatment failure. Methods • This ongoing, multicenter, open-label, phase II study enrolled patients with EGFR-mutant metastatic or unresectable advanced NSCLC who had experienced disease progression following standard therapy. • Patients were assigned to 2 cohorts: Cohort A included patients with MET amplification and/or MET overexpression detected in tumor tissue, and Cohort B included patients who had progressed on or were intolerant to prior MET inhibitor therapy. • All patients received MCLA-129 at 1500 mg intravenously once every 2 weeks (Q2W). • The primary endpoint was investigator-assessed objective response rate (ORR), with secondary endpoints including disease control rate (DCR), progression-free survival (PFS) and safety. Study Design • Data cutoff: June 10, 2026 for baseline and safety analyses; August 27, 2026 for efficacy analyses. Note: *MET amplification was defined as a MET copy number ≥ 5 by next-generation sequencing, a MET gene copy number ≥ 5 by fluorescence in situ hybridization, or a MET/CEP7 ratio ≥ 2. MET overexpression was defined as IHC 2+ or 3+ staining in ≥ 50% of tumor cells. Patient Demographic (Cohort A N = 65 | Cohort B N = 14 | Total N = 79) Age, years, median (range) 62.0 (38-77) | 61.5 (41-80) | 62.0 (38-80) Gender Male 33 (50.8) | 7 (50.0) | 40 (50.6); Female 32 (49.2) | 7 (50.0) | 39 (49.4) ECOG PS 0: 3 (4.6) | 2 (14.3) | 5 (6.3); 1: 62 (95.4) | 12 (85.7) | 74 (93.7) Histology Adenocarcinoma 61 (93.8) | 14 (100.0) | 75 (94.9); Other 4 (6.2) | 0 (0.0) | 4 (5.1) Prior lines of therapy 1: 24 (36.9) | 1 (7.1) | 25 (31.6); 2: 20 (30.8) | 8 (57.1) | 28 (35.4); ≥3: 21 (32.3) | 5 (35.7) | 26 (33.0) MET overexpression or amplification status 2+: 19 (29.2) | 0 (0.0) | 19 (24.1); 3+: 41 (63.1) | 4 (28.6) | 45 (57.0); Amplification*: 14 (21.5) | 10 (71.4) | 24 (30.4) *MET amplification overlapped partially with MET IHC 3+. Efficacy • As of 27 August 2026, 90 patients were evaluable for efficacy: 75 in Cohort A and 15 in Cohort B. • Among response-evaluable patients, MCLA-129 demonstrated an unconfirmed ORR of 45.3% (34/75; 95% CI 33.8-57.3) in Cohort A and 26.7% (4/15; 95% CI 7.8-55.1) in Cohort B. The lower ORR in Cohort B suggests reduced objective tumor response following prior MET inhibitor exposure (Table 1; Figure 1). • Disease control was observed in 86.7% of patients in both cohorts. In Cohort A, 34 patients achieved PR and 31 achieved SD, whereas in Cohort B, 4 patients achieved PR and 9 achieved SD (Table 1; Figure 1). Table 1. Best Overall Response of MCLA-129 in Cohorts A and B (Cohort A N = 75 | Cohort B N = 15) CR 0 | 0; PR 34 (45.3) | 4 (26.7); SD 31 (41.3) | 9 (60.0); PD 8 (10.7) | 1 (6.7); NE 2 (2.7) | 1 (6.7) ORR 34 (45.3) | 4 (26.7); 95% CI 33.8-57.3 | 7.8-55.1 DCR 65 (86.7) | 13 (86.7); 95% CI 76.8-93.4 | 59.5-98.3 Figure 1. Waterfall Plot of Best Response in Cohort A (A) and Cohort B (B) • Among patients with MET IHC3+ tumors in Cohort A, ORR was 73.7% (14/19) in patients previously treated with EGFR-TKI alone and 48.4% (15/31) in those previously treated with EGFR-TKI plus platinum-based chemotherapy (Table 2). • Compared with the corresponding MET IHC2+/3+ subgroups, ORR was consistently higher in patients with MET IHC3+ tumors, both after EGFR-TKI alone (73.7% vs 57.1%) and after EGFR-TKI plus platinum-based chemotherapy (48.4% vs 38.1%). • DCR remained consistently high across subgroups (89.5%-93.6%). Table 2. Best overall response by MET expression and prior treatment in Cohort A MET IHC 2+/3+ (N = 70): EGFR-TKI (N = 28) | EGFR-TKI + chemotherapy (N = 42); MET IHC 3+ (N = 50): EGFR-TKI (N = 19) | EGFR-TKI + chemotherapy (N = 31) CR 0 | 0 | 0 | 0 PR 16 (57.1) | 16 (38.1) | 14 (73.7) | 15 (48.4) SD 9 (32.1) | 21 (50.0) | 3 (15.8) | 14 (45.2) PD 2 (7.1) | 4 (9.5) | 1 (5.3) | 1 (3.2) NE 1 (3.6) | 1 (2.4) | 1 (5.3) | 1 (3.2) ORR 16 (57.1) | 16 (38.1) | 14 (73.7) | 15 (48.4); 95% CI 37.2-75.5 | 23.6-54.4 | 48.8-90.9 | 30.2-66.9 DCR 24 (90.1) | 38 (90.5) | 17 (89.5) | 29 (93.6); 95% CI 71.8-97.7 | 74.4-96.0 | 66.9-98.7 | 78.6-99.2 Note: Best overall responses presented in this table are unconfirmed and investigator-assessed. Efficacy • A total of 32 PFS events had occurred among 75 patients in Cohort A, with a median PFS of 6.9 months (95% CI 4.9-8.3; Figure 2A). In Cohort B, 7 PFS events had occurred among 15 patients, with a median PFS of 4.1 months (95% CI 2.1-NE; Figure 2B). • Among patients with MET IHC 3+ tumors, median PFS was 8.3 months (95% CI 4.1-NE) in those previously treated with EGFR-TKI alone (Figure 3A) and 7.8 months (95% CI 4.2-NE) in those previously treated with EGFR-TKI plus platinum-based chemotherapy (Figure 3B). Figure 2. Progression-Free Survival in Cohort A (A) and Cohort B (B): A Events/total 32/75, Median (95% CI) 6.9 (4.9-8.3); B Events/total 7/15, Median 4.1 (2.1-NE) Figure 3. Progression-Free Survival in Patients with MET IHC 3+ Tumors in Cohort A by Prior Treatment: EGFR-TKI Only (A) vs EGFR-TKI Plus Platinum-Based Chemotherapy (B) Safety • Infusion-related reaction (IRR) was the most common TEAE, occurring in 44/79 (55.7%) patients, and was predominantly grade 1-2. • Grade ≥3 TEAEs occurred in 34 patients (43.0%) and the most common grade ≥3 TEAE was neutropenia (17.7%). • TEAE leading to permanent discontinuation of study treatment were reported in 4 patients (5.1%). No patient died due to study treatment. TEAEs occurring in at least 10% of patients (%) (All Grade / Grade ≥ 3): IRR 55.7 / 3.8; Hypoalbuminaemia 40.5 / 1.3; Neutropenia 34.2 / 17.7; Leukopenia 30.4 / 3.8; Rash 29.1 / 1.3; ALT increased 19 / 0; AST increased 16.5 / 0; Hypocalcaemia 15.2 / 0; Lipase increased 15.2 / 3.8; LDH increased 12.7 / 0; Stomatitis 12.7 / 0; ALP increased 11.4 / 0; Proteinuria 11.4 / 0; Amylase increased 10.1 / 0; Anaemia 10.1 / 0; Thrombocytopenia 10.1 / 0 Conclusions • MCLA-129 demonstrated clinically meaningful antitumor activity in EGFR-mutant NSCLC with MET overexpression and/or amplification after standard treatment. • Greater activity was observed in MET inhibitor-naïve patients with MET IHC 3+ tumors, particularly in those previously treated with EGFR-TKI alone, whereas activity was more limited after prior MET inhibitor exposure. • MCLA-129 showed a manageable safety profile, supporting further investigation in this molecularly defined population. Declarations / Funding: All authors declare no competing interests. The study was funded by Betta Pharmaceuticals Co., Ltd. Contact Info: Qing Zhou. MD, PhD

[Slide 1] Calderasib (MK-1084), alone or in combination, for KRAS G12C-mutant metastatic non-small cell lung cancer Jhanelle E. Gray; Nir Peled; Laiyu Liu; Adrian Sacher; Jair Bar; Mustafa Erman; Annette Kodahl; Maya Gottfried; Enriqueta Felip; Shun Lu; Carlos I. Rojas; Rafal Dziadziuszko; Matteo Simonelli; Grygorii Ursol; Thomas Jemielita; L. Rhoda Molife; Yewon Choi; Razvan Cristescu; Mark Ayers; Mehmet A. N. Sendur 25 October 2026 #2680RO [Slide 2] Summary and Conclusions • In KANDLELIT-001, calderasib alone or in combination with pembrolizumab ± platinum-based chemotherapy shows clinically meaningful activity in participants with KRAS G12C-mutated metastatic NSCLC – Arms 1 + 3 (2L or later calderasib monotherapy): ORR 35%, median DOR 16.8 months, median PFS 8.3 months – Arm 2 (1L calderasib + pembrolizumab): ▪ TPS ≥50%: ORR 84%, median DOR not reached, median PFS 34.4 months ▪ TPS 1%-49%: ORR 55%, median DOR 26.2 months, median PFS 25.1 months • Reductions in KRAS G12C VAF and MSAF at cycle 3 day 1 indicate robust target engagement • AEs were manageable – Grade 3 TRAEs of increased ALT (3%) and increased AST (3%) with calderasib monotherapy were uncommon; rates were <9% with calderasib + pembrolizumab and <11% with the triplet combination – TRAEs leading to calderasib discontinuation occurred in 1% (monotherapy), 4% (calderasib + pembrolizumab), and 7% (triplet) of participants Our findings support enrollment in the phase 3 KANDLELIT-004 (NCT06345729)1 and KANDLELIT-007 (NCT07190248)2 studies of calderasib + pembrolizumab as first-line therapy in KRAS G12C-mutated metastatic NSCLC 1. https://clinicaltrials.gov/study/NCT06345729: KANDLELIT-004: A randomized, double-blind, phase 3 study of calderasib+pembrolizumab vs placebo+pembrolizumab as 1L therapy for participants with KRAS G12C-mutant metastatic NSCLC with PD-L1 TPS ≥50% 2. https://clinicaltrials.gov/study/NCT07190248: KANDLELIT-007: A randomized, open-label, phase 3 study of calderasib+pembrolizumab SC vs pembrolizumab SC+chemotherapy as 1L therapy participants with KRAS G12C-mutant nonsquamous metastatic NSCLC

209P AI-driven quantification of PD-L1 tumor proportion score (TPS) from immunohistochemistry slides in non-small cell lung cancer (NSCLC) A.D. Ramos-Guerra, J.E. Ortuño, I. Carretero-Barrio, N. Torres, A. Benito, S. Granados-Barrio, A. Compañ, R. Noguera, J. Castellví, E. Conde Gallego, C. Aura, N. Valerdiz Menéndez, M.I. Esteban, A. Argueta, A. Estébanez-Gallo, I. Sánchez Guerri, M. Martínez, J. Palacios, L. Paz-Ares, M. Ledesma-Carbayo, on behalf of the INGENIO consortium BACKGROUND • PD-L1 TPS is required for guiding treatment decisions in NSCLC. • TPS assessment requires expertise, is time-consuming and is subject to interobserver variability and technical heterogeneity [1]. • Automated methods often rely on manual annotations [2] or lack three-category TPS validation in NSCLC [3]. • Scarce studies based on real-world NSCLC cohorts. • Automated slide-level TPS classification may help standardize PD-L1 evaluation in routine clinical practice. METHODS DATA SPLIT: INGENIO 11 Spanish hospitals, 1 247 PD-L1 IHC WSIs, Proprietary; Memorial Sloan Kettering 1 US clinical center, 204 PD-L1 IHC WSIs, Public Hospital-dependent split (~80% / ~20%): Train INGENIO 9 hospitals 977 WSIs; Test INGENIO 2 hospitals 270 WSIs; Test MSK 1 hospital 204 WSIs METHODOLOGY WORKFLOW: Digitized WSI → Tissue segmentation → Segmented WSI → Manual control tissue exclusion → Final tissue mask → Selected tissue → Tissue patches → Patch-level encoder → Patch embeddings → Attention-based MIL → TPS classification (<1%, 1–49%, ≥50%) RESULTS Figure 1. Cohort composition. Distribution of PD-L1 TPS categories, PD-L1 antibody clones and slide scanners in the training cohort (Train-INGENIO) and both test cohorts (Test-INGENIO and Test-MSK). Figure 2. Foundation model benchmark. One-vs-rest AUC per TPS category for eight patch encoders in 5-fold cross-validation on Train-INGENIO. (CONCHv1.5, CTransPath, H-Optimus-1, PLIP, Phikon-v2, RetCCL, UNI 2, Virchow2) Figure 3. Model performance on test cohorts. Performance of the Virchow2-based model at clinical cut-offs in Test-INGENIO and Test-MSK test cohorts. | Test-INGENIO AUC | PPV | NPV | Test-MSK AUC | PPV | NPV TPS≥1% | 0.904 (95%CI 0.865-0.937) | 81.5% | 90.7% | 0.921 (95% CI 0.877-0.956) | 87.6% | 84.3% TPS≥50% | 0.923 (95%CI 0.886-0.956) | 86.1% | 95.7% | 0.861 (95% CI 0.814-0.907) | 97.3% | 80.9% Figure 4. Model interpretability. Example IHC slides, expert-guided cell segmentations and attention heatmaps from the Virchow2-based model. CASE A TPS: 0%; CASE B TPS: 5%; CASE C TPS: 80% CONCLUSIONS • AI-driven quantification accurately classifies PD-L1 TPS from IHC WSIs (AUC 0.86–0.92 in two external cohorts). • Attention maps highlight PD-L1+ tumor regions in TPS≥1% cases, consistent with pathologist criteria. • Future work will target the intermediate TPS 1–49% category and its association with progression-free survival.

Primary endpoint outcomes Cohort 1 (mBC with active BMs) 23.8% intracranial ORR as per RANO-BM criteria Bartsch R et al. Lancet Oncol, 2025 | Intracranial, % (n=21) | Extracranial, % (n=19*) | Bicompartmental, % (n=21) ORR | 19.0 | 10.5 | 14.3 DCR | 61.9 | 57.9 | 66.7 Cohort 2 (aNSCLC with active BMs) 30.0% intracranial ORR as per RANO-BM criteria Fuereder T et al. Lancet Oncol, 2025 | Intracranial, % (n=18*) | Extracranial, % (n=18*) | Bicompartmental, % (n=19*) ORR | 27.8 | 27.8 | 26.3 DCR | 77.8 | 77.8 | 73.7 Cohort 3 (LMD from solid tumors) 65.0% 3-month OS rate Preusser M et al. Nat Med, 2025 | Intracranial, % (n=17*) | Extracranial, % (n=15*) | Bicompartmental, % (n=19*) ORR | 23.5 | 26.7 | 31.6 DCR | 58.8 | 60.0 | 57.9 • We now report mature survival data together with final efficacy, safety, quality of life, and neurological outcomes. * Patients without a valid baseline assessment or without baseline lesions in the corresponding compartment were excluded. ADCs, antibody-drug conjugates; aNSCLC, advanced non-small cell lung cancer; BMs, brain metastases; DCR, disease control rate; LMD, leptomeningeal disease; mBC, metastatic breast cancer; ORR, overall response rate; OS, overall survival; RANO-BM: Response Assessment in Neuro-Oncology Brain Metastases.

IFCT ESMO 2026 Encorafenib plus binimetinib in patients (pts) with BRAFV600E-mutant advanced non-small cell lung cancer: updated efficacy and safety results from the IFCT-1904 ENCO-BRAF phase 2 trial C. Ricordel · J. Mazieres · C. Mascaux · F. Guisier · P. Tomasini · M. Wislez · D. Moro-Sibilot · S. Cousin · S. Couraud · V. Fallet · C. Guguen · A. Lagrange · A. Madroszyk-Flandin · J. Otto · M. Pérol · B. Roch · E. Amour · F. Morin · A.B. Cortot · D. Planchard Background • BRAF-V600 activating mutations are oncogenic drivers in non-small cell lung cancer (NSCLC). • The combination of BRAF inhibitor encorafenib plus MEK inhibitor binimetinib has demonstrated clinical efficacy in pts with BRAFV600E-mutant metastatic NSCLC and has received approval from the EMA and FDA. Methods and Objectives Open-label, Phase II trial Key inclusion criteria • Patients with BRAF-V600E-mutant metastatic NSCLC • ≥ 18 years • WHO performance status 0-1 • BRAFV600E mutation (enrolment by local assay) • No prior anti-BRAF cancer therapy • Stable CNS metastases allowed • Treatment naive or previously treated Treatment naive → Planned N = 60 → Binimetinib: 45 mg bid Encorafenib: 450 mg QD 28-Day cycles Previously treated → Planned N = 59 → Binimetinib: 45 mg bid Encorafenib: 450 mg QD 28-Day cycles • Primary endpoint: investigator-assessed confirmed best ORR (complete response + partial response) • Secondary endpoints included: DOR, PFS, OS, safety Baseline characteristics | Treatment naive (N=64) | Previously treated (N=32) Sex Female N (%) | 34 (53.1) | 16 (50.0) Male N (%) | 30 (46.9) | 16 (50.0) Age (years) Mean ± SD | 69.9 ± 10.7 | 69.2 ± 9.6 Smoking status Current N (%) | 7 (10.9) | 5 (15.6) Former N (%) | 34 (53.1) | 18 (56.3) Never N (%) | 23 (35.9) | 9 (28.1) ECOG PS 0 N (%) | 28 (43.8) | 12 (37.5) 1 N (%) | 36 (56.3) | 20 (62.5) Brain metastasis Yes N (%) | 11 (17.2) | 2 (6.3) No N (%) | 53 (82.8) | 30 (93.8) PDL1 TPS ≥ 50% N (%) | 32 (50.0) | 12 (37.5) ≥ 1% and < 50% N (%) | 17 (26.6) | 10 (31.3) < 1% N (%) | 13 (20.3) | 8 (25.0) Not done N (%) | 2 (3.1) | 2 (6.3) CONCLUSIONS ➤ The updated results of the IFCT-1904 ENCO-BRAF phase II trial further support the clinical activity of the encorafenib plus binimetinib combination in patients with BRAFV600E-mutant metastatic NSCLC. ➤ The safety profile was favourable and consistent with previous studies in lung cancer. ➤ These results confirm the combination as a therapeutic option in both first and second-line settings. Confirmed best objective response rate (ORR) and best overall response Treatment naive N=63 Confirmed best ORR (CR + PR): 68.3% [95% CI: 56.8%-79.7%] Median duration of response (DOR, months): 14.8 [95% CI: 9.3-22.1] DOR > 6 months (N): 33 (76.7%) Previously treated N=31 Confirmed best ORR (CR + PR): 61.3% [95% CI: 44.1%-78.4%] Median duration of response (DOR, months): 9.3 [95% CI: 4.8-NR] DOR > 6 months (N): 10 (52.6%) Progression-Free Survival (PFS) Treatment naive N=63: Median PFS (months): 11.1 [95% CI: 7.1-16.5] Previously treated N=31: Median PFS (months): 9.2 [95% CI: 5.6-11.1] Overall Survival (OS) Treatment naive N=63: Median OS (months): 41.1 [95% CI: 27.4-NR]; Median follow-up (months): 36.2 [95% CI: 31.9-37.5] Previously treated N=31: Median OS (months): 35.7 [95% CI: 14.7-NR]; Median follow-up (months): 20.2 [95% CI: 13.7-34.0] Quality of Life – VAS score Safety – TRAE any grade ≥10% [illegible bar chart labels and values] Special thanks to Independent data monitoring committee members: Eric Dansin (Lille), Charlée Nardin (Besançon) and Marie-Anne Vibet (Nantes). Aquilab for blinded independent central review of images DOI ESMO 2026 DOI C. Ricordel : consulting fees from MSD, Takeda, Roche, Amgen; support for attending meetings and/or travel from Takeda, Roche, Sanofi. Corresponding author: Charles.RICORDEL@chu-rennes.fr Study sponsored by IFCT www.ifct.fr Collaboration / funding: Pierre Fabre CT.gov identifier: NCT04526782 2651P

[Slide 1] Background and Study Design - B7-H3 is overexpressed in lung cancer and several other malignancies but has limited expression in normal tissues, where it promotes tumor progression and immune evasion1. - SKB500 is a B7-H3-targeted ADC comprising an antibody linked to a topoisomerase I inhibitor payload via a cleavable pyrimidine-tripeptide linker. As previously reported, SKB500 demonstrated favorable safety and antitumor activity across multiple tumor types2. Dose Escalation mTPI-2 design 18 mg/kg 16 mg/kg 12 mg/kg 9 mg/kg 6 mg/kg 4 mg/kg 2 mg/kg Expansion ES-SCLC 12 mg/kg (n=40)* 16 mg/kg (n=21)* ESCC, HNSCC, CRC, and others Expansion includes dose expansion and indication expansion Key Eligibility Criteria: - Age 18-75 years - ES-SCLC - Progressed on ≥ 1 prior line of systemic therapy - ECOG PS 0-1 Endpoints: - Primary Endpoints: Safety (DLT, AE) - Secondary Endpoints: Investigator-assessed ORR, DCR, DoR, and PFS; OS Here we present updated safety and antitumor activity of SKB500 in ES-SCLC from the multicenter, open-label phase I study. SKB500 was administered intravenously every 3 weeks (Q3W). *Sample sizes include patients enrolled in the dose escalation phase. Clinical trial identification: NCT06736327. Professor Haifeng Liu 1. Yamato M et al. Mol Cancer Ther. 2022; 21(4):635-646. 2. Liu H et al. 2026 ASCO. Abstract 3011. [Slide 2] Objective Response and Duration of Response | 12 mg/kg (n = 40) | 16 mg/kg (n = 21) ORR, % (95% CI) | 75.0 (58.8, 87.3) | 66.7 (43.0, 85.4) cORR, % (95% CI) | 62.5 (45.8, 77.3) | 66.7 (43.0, 85.4) DCR, % (95% CI) | 95.0 (83.1, 99.4) | 85.7 (63.7, 97.0) DoR, median (95% CI), mo | 5.7 (4.3, 7.1) | 5.8 (2.7, NE) Swimmer plot legend: 12 mg/kg Q3W; 16 mg/kg Q3W; Partial Response (first occurred); Ongoing (no PD/alive) Time (Months) Waterfall plot: Best Change from Baseline (%); 12 mg/kg Q3W; 16 mg/kg Q3W * Percentage change from baseline for target lesions was 0%. Data cutoff: September 16, 2026. cORR, confirmed ORR; NE, not estimable; PD, progressive disease; PR, partial response; SD, stable disease. Professor Haifeng Liu [Slide 3] Progression-Free Survival and Overall Survival | 12 mg/kg (n = 40) | 16 mg/kg (n = 21) mPFS (95% CI), mo | 8.1 (5.7, 8.5) | 5.6 (3.9, 9.3) 6-mo PFS rate,% (95% CI) | 65.3 (47.8, 78.2) | 49.0 (25.9, 68.7) | 12 mg/kg (n = 40) | 16 mg/kg (n = 21) mOS (95% CI), mo | 14.0 (7.7, NE) | 10.6 (4.6, NE) 12-mo OS rate,% (95% CI) | 56.8 (37.7, 72.0) | 46.4 (9.7, 77.7) Progression-Free Survival (%); Overall Survival (%); Time (Months) 12 mg/kg Q3W; 16 mg/kg Q3W; Censored No. at Risk Median follow-up: 9.3 months (12 mg/kg) and 8.3 months (16 mg/kg). Data cutoff: September 16, 2026. Professor Haifeng Liu Survival table Endpoint: PFS and OS Arm | N | Median | Median CI 12 mg/kg Q3W | 40 | PFS 8.1 mo; OS 14.0 mo | PFS 5.7-8.5; OS 7.7-NE 16 mg/kg Q3W | 21 | PFS 5.6 mo; OS 10.6 mo | PFS 3.9-9.3; OS 4.6-NE 6-mo PFS rate: 65.3% (47.8-78.2) vs 49.0% (25.9-68.7) 12-mo OS rate: 56.8% (37.7-72.0) vs 46.4% (9.7-77.7) [Slide 4] Safety Summary | 12 mg/kg (n = 40) | 16 mg/kg (n = 21) Median duration of exposure, mo (range) | 5.5 (1.4, 14.8) | 4.3 (0.7, 13.0) TRAEs, n (%) | 40 (100) | 20 (95.2) Grade ≥ 3 TRAEs | 21 (52.5) | 14 (66.7) Serious TRAEs | 15 (37.5) | 9 (42.9) TRAEs leading to dose reduction | 5 (12.5) | 8 (38.1) TRAEs leading to discontinuation | 1 (2.5) | 1 (4.8) TRAEs leading to death | 0 | 0 Summary of any grade TRAEs with incidence ≥ 20% in either group (12 mg/kg any grade, 12 mg/kg grade≥3 | 16 mg/kg grade≥3, 16 mg/kg any grade) Anemia 88 10 | 19 76 WBC decreased 65 15 | 29 76 Hypoalbuminemia 53 0 | 0 48 NEUT decreased 50 8 | 19 67 AST increased 48 0 | 0 19 PLT decreased 45 10 | 14 57 Nausea 38 3 | 0 52 ALT increased 35 0 | 0 14 LYM decreased 33 13 | 29 48 Decreased appetite 33 5 | 10 57 GGT increased 30 0 | 5 29 Asthenia 25 13 | 0 38 Weight decreased 15 0 | 0 33 Hyponatremia 13 0 | 0 43 Vomiting 10 0 | 0 24 Constipation 8 0 | 0 43 TRAEs (%) 12 mg/kg Q3W, Any Grade; 16 mg/kg Q3W, Any Grade; 12 mg/kg Q3W, Grade≥3; 16 mg/kg Q3W, Grade≥3 Data cutoff: September 16, 2026. ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyltransferase; LYM, lymphocyte count; NEUT, neutrophil count; PLT, platelet count; WBC, white blood cell count. Professor Haifeng Liu

7-year PFS by on-treatment TF clearance in the lorlatinib group • Among patients with TF detected at baseline, TF clearance (D/ND) was associated with longer PFS vs persistent TF detected (D/D) as early as week 4 (HR 0.48) and the association held at week 24 (HR 0.45)a [Left panel] Week 4 TF clearance statusb Probability of PFS; Months PFS, months, median (95% CI) | D/ND n=54: NR (51.7-NR) | D/D n=45: 33.3 (16.4-69.7) HR (95% CI), D/ND vs D/D: 0.48 (0.26-0.86) No. at risk (0, 12, 24, 36, 48, 60, 72, 84, 96 months) D/ND: 54 43 39 33 32 30 30 16 2 D/D: 45 30 20 15 13 10 7 6 0 [Right panel] Week 24 TF clearance statusb PFS, months, median (95% CI) | D/ND n=58: NR (60.0-NR) | D/D n=15: 32.9 (16.4-NR) HR (95% CI), D/ND vs D/D: 0.45 (0.21-0.96) No. at risk (0, 12, 24, 36, 48, 60, 72, 84, 96 months) D/ND: 58 50 40 35 32 31 29 17 2 D/D: 15 12 8 4 4 3 3 2 0 D, detected; HR, hazard ratio; ND, not detected; NR, not reached; PFS, progression-free survival; TF, tumor fraction. aTF is defined as the ratio of observed tumor-specific methylation signals to methylation levels across constitutively methylated and unmethylated regions. ND means TF=0. bThe clearance analyses at Week 4 and Week 24 are based on patients with both a detectable baseline sample and an evaluable paired sample at the respective visit. Survival table Endpoint: PFS by week-4 ctDNA tumor fraction clearance (lorlatinib group, TF detected at baseline) Arm | N | Median | Median CI D/ND (TF cleared at week 4) | 54 | NR | 51.7-NR D/D (TF persistent at week 4) | 45 | 33.3 months | 16.4-69.7 HR: 0.48 HR CI: 0.26-0.86

2712P SOHO-01: Updated efficacy and safety of sevabertinib in advanced HER2-mutant NSCLC, including first-line patients and those with active brain metastases Xiuning Le; Tae Min Kim; Lin Li; Yong Fang; Xiaorong Dong; Herbert H. Loong; Yuki Shinno; Gennaro Daniele; Hye Ryun Kim; Silvia Novello; Gerrina Ruiter; Nicolas Girard; Liyun Miao; Rui Li; Paolo Grassi; Shun Lu; Arsela Prelaj INTRODUCTION • Sevabertinib is a potent, reversible, oral tyrosine kinase inhibitor (TKI) that is FDA approved for adult patients with locally advanced or metastatic non-squamous non-small cell lung cancer (NSCLC) harboring HER2 tyrosine kinase domain (TKD) mutations after previous systemic therapy, and has more recently been approved for treatment-naive patients • In the Phase I/II SOHO-01 study (NCT05099172), sevabertinib previously showed antitumor activity and a manageable safety profile in patients with previously treated (Cohort D) and treatment-naive (Cohort F) HER2-mutant NSCLC – In Cohort F, duration of response (DoR) and progression-free survival (PFS) data were immature at a median follow-up of 9.9 months • Here, we report updated results from Cohorts D and F with longer follow-up, together with findings from Cohort G, which enrolled patients with active, clinically stable brain metastases METHODS • SOHO-01 (NCT05099172) is an open-label, single-arm, multicenter, multicohort, Phase I/II study evaluating the efficacy and safety of sevabertinib (20 mg twice daily) in patients aged ≥18 years with locally advanced, or metastatic HER2-mutant NSCLC (Figure 1) Figure 1. SOHO-01 study design (NCT05099172) D Naive to HER2 ex20ins-targeted therapy; F Naive to systemic anticancer therapy for locally advanced or metastatic disease; G Naive to HER2 ex20ins-targeted TKIs with active, clinically stable CNS metastases PRIMARY ENDPOINTS • ORR per RECIST v1.1 by BICR; SECONDARY ENDPOINTS • DoR and PFS (per RECIST v1.1) by BICR and investigator assessed • Safety and tolerability • RANO-BM (Cohort G) Cut-off date: March 16, 2026 RESULTS: BASELINE CHARACTERISTICS • As of the data cut-off of March 16, 2026, 81 patients in Cohort D, 73 patients in Cohort F, and 31 patients in Cohort G received sevabertinib (20 mg, twice daily, Table 1) – Most patients had both non-squamous histology and HER2 TKD mutations Table 1. Baseline demographics and disease characteristics (Cohort D n=81 | Cohort F n=73 | Cohort G n=31) Median age, years (range) 60 (29–82) | 65 (31–82) | 61 (43–78) Female, n (%) 50 (62) | 46 (63) | 21 (68) Never smoked, n (%) 50 (62) | 57 (78) | 22 (71) HER2 TKD mutation and non-squamous histology, n (%) 70 (86) | 69 (95) | 30 (97) RESULTS: EFFICACY ORR • The ORR in patients with non-squamous histology and HER2 TKD mutations at baseline was 73% in Cohort D and 75% in Cohort F (Table 2) • In patients with TKD mutations at baseline, the ORR was 71% (95% CI 59, 81) in Cohort D and 75% (95% CI 63, 84) in Cohort F DoR • Median DoR was 15.1 months in treatment-naive patients (Cohort F) and 9.5 months in previously treated patients (Cohort D; Table 2) Table 2. Tumor response in Cohorts D and F (BICR, per RECIST v1.1) Cohort D (n=81): All treated patients (n=81) | Treated patients with non-squamous histology + HER2 TKD mutation (n=70) | Treated patients with non-squamous histology + HER2 YVMA mutation (n=49); Cohort F (n=73): All treated patients (n=73) | non-squamous + HER2 TKD mutation (n=69) | non-squamous + HER2 YVMA mutation (n=55) CR 2 (2) | 2 (3) | 2 (4) | 4 (5) | 4 (6) | 2 (4) PR 52 (64) | 49 (70) | 36 (73) | 51 (70) | 48 (70) | 39 (71) SD 18 (22) | 12 (17) | 7 (14) | 13 (18) | 13 (19) | 11 (20) PD 6 (7) | 4 (6) | 3 (6) | 2 (3) | 1 (1) | 1 (2) Not evaluable/Not available 3 (4) | 3 (4) | 1 (2) | 3 (4) | 3 (4) | 2 (4) ORR, % (95% CI) 67 (55, 77) | 73 (61, 83) | 78 (63, 88) | 75 (64, 85) | 75 (64, 85) | 75 (61, 85) DCR, % (95% CI) 81 (71, 89) | 84 (74, 92) | 88 (75, 95) | 89 (80, 95) | 90 (80, 96) | 89 (78, 96) Median DoR, months (95% CI) 9.5 (6.3, 13.5) | 9.5 (6.3, 13.5) | 9.5 (6.3, 15.0) | 15.1 (9.1, NE) | 15.1 (9.1, NE) | 15.1 (8.8, NE) Median follow up, months (range) 22.8 (0.9–41.4) | 23.0 (0.9–41.4) | 23.1 (0.9–41.4) | 17.4 (0.2–24.0) | 17.9 (0.2–24.0) | 17.9 (0.2–24.0) PFS • Median PFS in patients with non-squamous histology and HER2 TKD mutations was 9.6 months (95% CI 6.9, 14.5; Figure 2A) in Cohort D and 16.4 months (95% CI 10.0, NE; Figure 2C) in Cohort F • In all patients treated in Cohorts D and F, median PFS was 8.3 months (95% CI 6.9, 12.3; Figure 2B) and 13.5 months (95% CI 10.0, NE; Figure 2D), respectively Figure 2. Kaplan-Meier curves of PFS by BICR in Cohorts D (A, B) and F (C, D) Cohort G: Patients with active, clinically stable brain metastases • Most patients in Cohort G were pretreated, with 45% (n=14) having received ≥3 prior lines of systemic therapy (Table 3) – In Cohort G, 26% of patients had received prior trastuzumab deruxtecan • At a median follow-up of 13.2 months, the ORR was 26% (n=8/31), of which 6 patients had a complete response and 2 patients had a partial response (Table 4) • Of the patients with measurable intracranial lesions (n=16), 81% had intracranial tumor shrinkage by RANO-BM (Figure 3) Table 3. Cohort G treatment history (n=31): Prior systemic anti-cancer therapies 0: 2 (6); 1: 12 (39); 2: 3 (10); ≥3: 14 (45). Previous anticancer therapy: EGFR ex20ins-targeted therapy 31 (100); Chemotherapy 26 (84); HER2 ex20ins-targeted therapy 11 (35); Trastuzumab deruxtecan 8 (26) Table 4. Tumor response in Cohort G (RANO-BM by BICR) (All patients n=31 | Measurable intracranial target lesions n=16 | Non-measurable intracranial target lesions n=15) CR 6 (19) | 0 | 6 (40) PR 2 (6) | 2 (13) | 0 SD 11 (35) | 6 (38) | 5 (33) PD 8 (26) | 6 (38) | 2 (13) Not evaluable 4 (13) | 2 (13) | 2 (13) ORR, % (95% CI) 26 (12, 45) | 13 (2, 38) | 40 (16, 68) DCR, % (95% CI) 45 (27, 64) | 31 (11, 59) | 60 (32, 84) Median DoR, months (95% CI) 6.9 (2.9, 10.4) | 5.1 (2.9, NE) | 6.9 (4.2, NE) Median PFS, months (95% CI) 4.0 (2.7, 5.6) | 2.8 (2.0, 4.0) | 8.1 (2.7, 14.4) RESULTS: SAFETY • The majority of drug related treatment-emergent adverse events (TEAEs) were below grade 3 in Cohorts D, F, and G (Table 5) • The most common drug-related TEAE was diarrhea, occurring in 86% of patients in Cohort D and 85% of patients in Cohort F (Table 5) – Grade 3 drug-related diarrhea occurred in 25% (n=20) of patients in Cohort D, 5% (n=4) of patients in Cohort F, and 26% (n=8) in Cohort G – There were no treatment discontinuations due to diarrhea Table 5. Incidence and severity of drug-related TEAEs (Cohort D n=81 | Cohort F n=73 | Cohort G n=31) Any 78 (96) | 71 (97) | 30 (97) Grade 1–2 45 (56) | 49 (67) | 16 (52) Grade ≥3 33 (41) | 22 (30)* | 14 (45) Serious 12 (15) | 8 (11) | 5 (16) Leading to dose reduction 20 (25) | 23 (32) | 8 (26) Leading to dose interruption or delay 26 (32) | 24 (33) | 8 (26) Leading to permanent treatment discontinuation 4 (5) | 2 (3) | 1 (3) Most common drug-related TEAEs (>20% of patients): Diarrhea 70 (86) | 62 (85) | 27 (87); Rash 42 (52) | 42 (58) | 11 (35); Paronychia 23 (28) | 20 (27) | 6 (19); Anemia 15 (19) | 20 (27) | 2 (6); Stomatitis 15 (19) | 20 (27) | 7 (23); Increased ALT 14 (17) | 17 (23) | 6 (19); Hypokalemia 15 (19) | 16 (22) | 6 (19) CONCLUSIONS • In this updated analysis of the SOHO-01 study, sevabertinib demonstrated durable antitumor activity in previously treated (Cohort D) and treatment-naive (Cohort F) patients with HER2-mutant NSCLC – Durable efficacy and intracranial activity was also demonstrated in heavily pretreated patients with active brain metastases (Cohort G) • The safety profile of sevabertinib in Cohorts D and F was manageable and consistent with previous reports Acknowledgments: This study was sponsored by Bayer. Presented at the European Society for Medical Oncology Congress 2026, Madrid, Spain, 23–27 October 2026.

[Slide 1] Major Oncology Journal Impact Factors Oncology Journal | Impact Factor | Typical focus CA: A Cancer Journal for Clinicians | 685.2 | Cancer statistics, major reviews and guidelines Nature Reviews Clinical Oncology | 94.6 | Clinical reviews and perspectives Annals of Oncology | 80.4 | Clinical trials, guidelines and translational research Cancer Cell | 56.1 | Fundamental and translational cancer biology JCO | 44.7 | Practice-changing clinical oncology Lancet Oncology | 33.7 | Major clinical trials and global oncology Cancer Discovery | 29.5 | Novel therapeutics and translational discoveries Nature Cancer | 28 | Basic, translational and clinical cancer research JAMA Oncology | 23.9 | Clinical research, outcomes and populationscience Journal of Thoracic Oncology | 23.3 | Lung cancer and thoracic malignancies Cancer Research | 22.6 | Experimental and translational cancerresearch Journal for ImmunoTherapy of Cancer | 11.7 | Immunotherapy and immune-oncology Clinical Cancer Research | 10.9 | Translational research and therapeutic development ESMO Open | 10.6 | Clinical and translational oncology European Journal of Oncology | 7.9 | Clinical and translational cancer research British Journal of Cancer | 7.8 | Clinical, laboratory and epidemiologic research JNCI | 7.7 | Clinical, epidemiologic and cancer-controlresearch Cancer | 5.6 | Clinical oncology and outcomes JCO Oncology Practice | 5.5 | Quality of care, implementation and practice Lung Cancer | 5.3 | Lung cancer clinical and translational research High impact general medical journals: NEJM: 84.5 Lancet: 109 Hope S. Rugo [Slide 2] Writing your Paper • Match the manuscript's level of detail to the readership • Avoid overselling clinical implications beyond the design and data • For most original oncology research, use the Introduction, Methods, Results and Discussion (IMRAD) structure. • ICMJE recommends clear reporting of the study objective, methods, findings, limitations and supporting evidence. • Special consideration: Use of AI • Follow the journal's AI policy • If AI assists with writing, analysis or figure generation, disclose its use as required and independently verify the output • AI tools must not be listed as authors • ICMJE recommendations were updated in January 2026, including expanded guidance on AI in publishing and authors' access to data. • https://www.icmje.org/recommendations/browse/artificial-intelligence/ai-use-by-authors.html Hope S. Rugo [Slide 3] Writing your Paper: Recommended Writing Sequence 1. Prepare the tables and figures first. Establish the final analysis, patient flow, baseline characteristics and primary outcomes before drafting the narrative 2. Write the Methods. Specify the study design, eligibility criteria, treatment or exposure definitions, endpoints, statistical methods, ethics approval and data sources 3. Write the Results. Report findings in the same order as the prespecified objectives. Include denominators, effect estimates, confidence intervals and clinically relevant safety outcomes 4. Write the Discussion. Begin with the principal finding, compare it with relevant studies, explain the clinical significance and acknowledge limitations, summarize findings at the very end 5. Write the Introduction. Use a concise sequence: established knowledge, unresolved gap, and the specific question this study addresses – do not write a review article! 6. Draft the abstract and title last. They should accurately reflect the final analysis and avoid claims stronger than the results support Hope S. Rugo, MD [Slide 4] Oncology Specific Topics Survival analyses: State the time origin, event definition, censoring rules, median follow-up, number of events and methods used to estimate survival. Report hazard ratios with confidence intervals when appropriate. Retrospective studies: Explain selection criteria, missing data, confounding, treatment-era effects and potential selection bias. Biomarker studies: Specify assay methods, specimen handling, cutoff selection, validation strategy and whether analyses were prespecified. Clinical trials: Report registration, protocol deviations, endpoint definitions, adverse events and relevant CONSORT items. AI and predictive models: Explain the training and validation cohorts, missing-data handling, model performance and external validation where applicable. Real-world evidence: Explain data provenance, cohort construction, index dates, treatment exposure definitions and sensitivity analyses. Avoid describing an association as proof of causation in an observational study. A statistically significant result is not automatically clinically meaningful. Hope S. Rugo

[Slide 3] Efficacy: Group 3 (1L NSCLC, monotherapy) CS2009 monotherapy demonstrated compelling clinical efficacy in first-line, PD-L1 TPS ≥1% NSCLC | | All Evaluable Patients | 20 mg/kg | 30 mg/kg | Squamous | Non-squamous TPS ≥1% | ORR, % (n) | 66.7% (32/48) | 58.3% (14/24) | 75.0% (18/24) | 69.0% (20/29) | 63.2% (12/19) TPS ≥1% | DCR, % (n) | 93.8% (45/48) | 95.8% (23/24) | 91.7% (22/24) | 96.6% (28/29) | 89.5% (17/19) TPS ≥50% | ORR, % (n) | 87.5% (21/24) | 76.9% (10/13) | 100% (11/11) | 91.7% (11/12) | 83.3% (10/12) TPS ≥50% | DCR, % (n) | 95.8% (23/24) | 92.3% (12/13) | 100% (11/11) | 100% (12/12) | 91.7% (11/12) [Waterfall plot TPS ≥1%: best change from baseline, % (20 mg/kg, 30 mg/kg); spider plot TPS ≥1%: change from baseline, % vs Months] Median DoR and median PFS were not reached due to short follow-up. Dr. Anwen Xiong [Slide 4] Efficacy: Group 4 (1L SQ NSCLC, in combination with paclitaxel and carboplatin) CS2009 + paclitaxel + carboplatin demonstrated compelling clinical efficacy in first-line squamous NSCLC | All Evaluable Patients | 20 mg/kg | 30 mg/kg | PD-L1 TPS <1% | PD-L1 TPS 1-49%* | PD-L1 TPS ≥50% ORR, % (n) | 80.8% (21/26) | 76.9% (10/13) | 84.6% (11/13) | 100% (8/8) | 66.7% (10/15) | 100% (2/2) DCR, % (n) | 100% (26/26) | 100% (13/13) | 100% (13/13) | 100% (8/8) | 100% (15/15) | 100% (2/2) * 14/15 with PD-L1 TPS [illegible] Note: One additional patient with unknown PD-L1 status achieved a PR [Waterfall plot: best change from baseline, %, 20 mg/kg and 30 mg/kg; spider plot: change from baseline, % vs Months; legend: 20 mg/kg, 30 mg/kg, First New Lesion, Treatment ongoing] Median DoR and median PFS were not reached due to short follow-up. Dr. Anwen Xiong

[Slide 2] First-line chemotherapy + novel agents Study | Design | Treatment | Patients | Key results RELEVENT – TYME (Proto) Proto C et al. Ann Oncol 2024;35(9):817-826 | Single-arm phase II | Carboplatin + paclitaxel + ramucirumab → ramucirumab maintenance | 35 TC | ORR 80% (investigator) / 57.6% (central); PFS 18.1 mo; OS 43.8 mo SWOG S1701 (Tsao) Tsao AS et al. JTO Clin Res Rep 2024;5(12):100738 | Randomized phase II (closed early) | Carboplatin + paclitaxel ± ramucirumab | 21 TC | ORR 88% vs 40%; PFS HR 0.51 (p = 0,13) PFS 8 vs7 mo MARBLE (Shukuya) Shukuya T et al. Lancet Oncol 2025;26(3):331-342 | Multicentre phase II | Atezolizumab + carboplatin + paclitaxel | 48 TC | ORR 56% (central review); PFS 9.6 mo China (Zhu) Zhu K et al. ESMO Open 2026;11(10):108366 | Open-label phase II | Toripalimab + platinum doublet | 24 (B3 + TC) | ORR 37.5%; PFS 19.3 mo; G≥3 irAE 4% China (Ma) Ma JT et al. J Immunother Cancer 2026;14(9) | Phase II (preliminary) | Ivonescimab (PD-1/VEGF) + carboplatin-paclitaxel | 7 TC (ongoing) | ORR 85.7%; DCR 100%; G≥3 TRAE 57% Marina Chiara GARASSINO University of Chicago. @marinagarassino [Slide 3] My treatment future algorithms Thymoma First-line: CAP Further lines: CAPE GEM, PEMETREXED, EVEROLIMUS, RIVOCERANIB PATHOLOGICAL REVIEW TO CONFIRM HISTOLOGY Thymic Carcinoma First-line: CARBO-TAXOL RAMUCIRUMAB, CARBO-TAXOL IVONESCIMAB LONGER FOLLOW UP Further lines: CAPE GEM, SUNITINIB, LENVATINIB PEMBRO*, RIVOCERANIB * Only in patients without clinically significant autoimmune disease; screened for myasthenia gravis. Marina Chiara GARASSINO University of Chicago. @marinagarassino

#2678P A Pooled Analysis of Zongertinib Treatment in HER2-Overexpressed/Amplified NSCLC Alison M. Schram,* Mariano Ponz-Sarvise, Alberto Hernando-Calvo, Ingrid M.E. Desar, François Ghiringhelli, Shigehisa Kitano, Christina Gewinner, Lukas Lunger, Patrizia Sini, Lorenz Uhlmann, John V. Heymach, Noboru Yamamoto Introduction • Zongertinib is an irreversible TKI that selectively inhibits HER2 while sparing wild-type EGFR, thereby minimizing associated toxicities • Zongertinib is approved in the United States (accelerated) and in China (conditional) for patients with advanced HER2-mutant NSCLC, and in Japan for patients with previously treated advanced HER2-mutant NSCLC. Zongertinib is not approved by the EMA • Here, we present a pooled analysis of patients with advanced HER2-overexpressed/amplified NSCLC treated with zongertinib monotherapy from Phase Ia of the Beamion LUNG-1 trial (NCT04886804) and the ongoing Phase II Beamion PANTUMOR-1 trial (NCT06581432), along with supporting preclinical data Methods • Phase Ia of Beamion LUNG-1 included patients with previously treated advanced solid tumors and HER2 alterations (mutations, amplification, or overexpression) who had exhausted all standard treatment options • Beamion PANTUMOR-1 includes patients with previously treated HER2-altered solid tumors; patients could have received prior treatment with a HER2-directed ADC but not HER2 TKIs • Patients in Beamion LUNG-1 received escalating doses of zongertinib; patients in Beamion PANTUMOR-1 received a 120 mg fixed dose of zongertinib • Patients were selected for this pooled analysis according to HER2 overexpression and amplification (IHC 3+, IHC 2+ and ISH+, and/or CN ≥6 by NGS); patients with HER2 mutations were excluded • In this pooled analysis, efficacy (investigator-assessed ORR; RECIST v1.1) and safety (AEs; CTCAE v5.0) were assessed • In addition, preclinical antitumor activity of zongertinib as monotherapy was assessed in commercially available HER2-amplified CDX and PDX mouse models of NSCLC Preclinical Activity in HER2-Amplified CDX and PDX Models • In HER2-amplified CDX and PDX models of NSCLC, zongertinib led to delay of tumor growth (TGI = 41–64%; Calu-3) or complete TGI (TGI = 89–100%) Model | Zongertinib TGI (%) 40 mg/kg QD | 20 mg/kg QD | IHC | ERBB2 Expr Log2(TPM+1) | HER2 CN | HER2 mutational status | EGFR status | RAS status NCI-H2170 | | 89 | 3+ | 11.5 | 10.1 | Not mutated | wt | wt Calu-3 | 41 | 64 | 3+ | 10.3 | 14.0 | Not mutated | wt | wt LU9556 | | 96 | 3+ | 12.3 | 79.1 | Not mutated | wt | wt LU9681 | 100 | 91 | 2+ | 9.1 | 7.2 | Not mutated | wt | wt Key Findings and Conclusions • In this pooled analysis, zongertinib showed encouraging antitumor activity in patients with advanced, HER2-overexpressed/amplified NSCLC, none of whom had HER2 mutations • Further investigation of zongertinib treatment in this patient population is warranted Patient Demographics and Clinical Characteristics • As of the respective cut-off dates for Beamion LUNG-1 (May 19, 2026) and Beamion PANTUMOR-1 (Aug 20, 2026), 27 patients with previously treated HER2-overexpressed/amplified NSCLC were included in this pooled analysis – In Beamion LUNG-1, 1 patient received zongertinib 300 mg QD† – 26 patients in Beamion PANTUMOR-1 received zongertinib 120 mg QD †In the abstract, 2 patients from Beamion LUNG-1 are reported; for 1 of these patients, a HER2 mutation was subsequently found in central testing analysis, therefore this patient has been excluded from the presentation All (n=27): Male, n (%) 15 (55.6); Median age, years (range) 68.0 (31–78); Race White 19 (70.4), Asian 6 (22.2), Black 1 (3.7), Missing 1 (3.7); ECOG PS 0 / 1, n (%) 8 (29.6) / 19 (70.4); Lines of previous systemic therapy, median (range) 2 (1–6); Previous HER2-directed therapy, n (%) 1 (3.7); Previous EGFR-directed therapy, n (%) 4 (14.8) Safety Results • TRAEs (any grade) were reported in 19/27 patients • 1 patient had two grade 3 TRAEs (patient from Beamion LUNG-1) • No grade 4/5 TRAEs were reported • 1 patient had an AE leading to dose reduction (patient from Beamion LUNG-1) • No patients discontinued zongertinib due to AEs • Zongertinib had a manageable safety profile, with no new safety signals identified All | Grade 1 | Grade 2 | Grade 3 Patients with any TRAE, n (%) 19 (70.4) | 9 (33.3) | 9 (33.3) | 1 (3.7) Most common (observed in ≥2 patients) TRAEs, n (%): Diarrhea 8 (29.6) | 4 (14.8) | 4 (14.8) | 0 (0); Rash 4 (14.8) | 3 (11.1) | 0 (0) | 1 (3.7); Anemia 3 (11.1) | 1 (3.7) | 2 (7.4) | 0 (0); Fatigue 3 (11.1) | 2 (7.4) | 1 (3.7) | 0 (0); ALT increased 2 (7.4) | 2 (7.4) | 0 (0) | 0 (0); Amylase increased 2 (7.4) | 2 (7.4) | 0 (0) | 0 (0); AST increased 2 (7.4) | 2 (7.4) | 0 (0) | 0 (0); Asthenia 2 (7.4) | 2 (7.4) | 0 (0) | 0 (0); Decreased appetite 2 (7.4) | 1 (3.7) | 1 (3.7) | 0 (0); Dysgeusia 2 (7.4) | 2 (7.4) | 0 (0) | 0 (0); Ejection fraction decreased 2 (7.4) | 1 (3.7) | 1 (3.7) | 0 (0); Proteinuria 2 (7.4) | 0 (0) | 2 (7.4) | 0 (0); Stomatitis 2 (7.4) | 2 (7.4) | 0 (0) | 0 (0) Efficacy in Patients with HER2-Overexpressed/Amplified NSCLC A. Treatment duration and tumor response per RECIST v1.1 by investigator assessment (n=18)** B. Best change from baseline in target lesion size per RECIST v1.1 by investigator assessment (n=18)** C. Confirmed ORR (n=19): 47.4%. 9 (47.4%) patients achieved a confirmed PR. Confirmed ORR: 47.4% (95% CI: 27.3%–68.3%) D. Confirmed DCR (n=19): 78.9%. 6 (31.6%) patients had SD. Confirmed DCR: 78.9% (95% CI: 56.7%–91.5%) • Among all 27 patients, the median duration of treatment was 2.8 months (range: 0–17 months) • At data cut-off, 16/27 (59.3%) patients remained on treatment • Of 9 responding patients: – 3 (33.3%) had a DoR ≥6 months – 5 (55.6%) remained on treatment at data cut-off **1 of 19 patients discontinued at Day 22 due to clinical disease progression without an assessment of overall response; this patient's response was classed as not evaluable This study was funded by Boehringer Ingelheim. Presented at the European Society for Medical Oncology (ESMO) Congress, Madrid, Spain, October 23–27, 2026

[Slide 2] Methods • MTAP immunohistochemistry was initially evaluated in 43 mesothelioma cases with 2 clones : RBT-MTAP mAb (Bio SB) and mAb 1813 (Cell Signalling Technology) on Ventana Ultra Autostainer • MTAP IHC was then performed using the RBT-MTAP mAb only for all 1.433 cases • CDKN2A FISH analysis was carried out on 108 cases (including 67 mesotheliomas and 41 benign mesothelial lesions) • Full-exome RNA sequencing was conducted on 600 cases collected between 2000 and 2026. 1813 clone (CST) rBT-MTAP clone (Bio SB) Sylvie Lantuejoul [Slide 3] Results : MTAP staining MTAP expression was considered absent when no staining was observed in either the cytoplasm or the nucleus. 12% of mesotheliomas showed heterogeneous MTAP expression. EM in situ EM SM BAP1 loss Calretinin Cytokeratins MTAP loss MTAP loss (heterogeneous) MTAP loss [Slide 4] Results (cont.): Is there any correlation with prognosis and OS ? Whilst the overall effect of MTAP on prognosis is not significant (p=0.0597), there is a significant interaction between MTAP loss and histology (p=0.002217) The purportedly poor prognosis associated with the loss of MTAP gene expression is explained by its prevalence in subtypes with a poor prognosis (i.e. BM and SM) S-PM with retained MTAP expression had a worse overall survival (p=0,0001). MTAP 0 / 1_2_3_4 Panels: Begnin | BM | EM | SM Survival probability Time HR in multivariate analysis Cohorte | PM (N=1232) | PeM (N=201) Gender Males | ref | ref Gender Females | 1.04 | 0.92 Age [0,65] | ref | ref Age [65,100] | 1.74 | 1.44 BAP1 IHC 0 | ref | ref BAP1 IHC 1_2_3_4 | 1.34 | 1.20 MTAP IHC 0 | ref | ref MTAP IHC 1_2_3_4 | (blank) | 0.49 Histology EM | ref | Histology Benign | 0.14 | Histology BM | 1.50 | Histology SM (MTAP loss) | 1.65 | Histology SM (MTAP retained) | 3.24 | Grade (WHO) Low grade | ref | ref Grade (WHO) High grade | 1.95 | 3.43 Grade (WHO) NR | 1.39 | 1.11 HR < 1 = HR signif (p < 0.05) HR > 1 = HR signif (p < 0.05) Survival table Endpoint: OS by MTAP IHC status (0 vs 1_2_3_4), faceted by histology PM cohort | 1232 PeM cohort | 201 P: overall MTAP effect p=0.0597; MTAP x histology interaction p=0.002217; S-PM retained MTAP worse OS p=0,0001

[Slide 1] Background and study design • Metastatic thymic epithelial tumours (TETs) have no standard beyond platinum based cytotoxic chemotherapy and immune checkpoint inhibitors are limited by immune-related toxicity, especially in thymoma • Multitargeted VEGFR TKIs are active mainly in thymic carcinoma (lenvatinib ORR 38%, sunitinib 21-26%) but limited efficacy shown in thymoma (ORR 0-6%) • Rivoceranib is a multitargeted oral TKIs but highly selective VEGFR-2 inhibitor. 700 mg once daily, the global development dose, had not been tested prospectively in TETs Eligibility Metastatic thymoma (A-B3) or thymic carcinoma, with prior treatment of at least ≥1 systemic therapy, N = 40 Rivoceranib 700 mg orally once daily, 28-day cycles Dose reduction 500 → 300 → 200 mg Tumour assessment RECIST v1.1 (investigator) every 8 weeks in year 1, then every 12 weeks Until progression or unacceptable toxicity 11 KCSG sites, Korea Primary endpoint: Objective response rate (ORR) by RECIST v1.1, investigator-assessed Secondary endpoints: DCR, PFS, DoR, safety (CTCAE v5.0), OS and WES / WTS / AI-based TME analysis (Lunit SCOPE IO) exploratory Statistics: Simon two-stage; H0 ORR 20% vs H1 40% (1-sided α 0.05, power 80%); ≥11/33 responders required; N = 40 allowing for dropout Sehhoon Park, MD, PhD [Slide 2] Result - Primary endpoint Best response PR SD PD THYMOMA THYMIC CARCINOMA 35.0% Objective response rate 85.0% Disease control rate 11.2 mo Median DOR Best change from baseline (%) Best response (n = 40): PR 14 (35.0%), SD 20 (50.0%), PD 2 (5.0%), not evaluable 4 (10.0%) Thymoma ORR 31.2% (5/16) · DCR 93.8% , Thymic carcinoma ORR 37.5% (9/24) · DCR 79.2% Sehhoon Park, MD, PhD [Slide 3] Result - Survival Analysis Progression-free survival (N = 40) Median 11.8 mo (95% CI 7.4-NR) Overall survival (N = 40) Median not reached (95% CI 12.6-NR) Progression-free survival by histology Thymoma Thymic carcinoma Thymoma NR vs Thymic carcinoma 9.2 mo PFS HR 1.87 (95% CI 0.67-5.20) Number at risk Months Median follow-up 13.6 months (cut-off April 2026) PFS similar across treatment lines (P = 0.79) and in patients with early dose reduction (P = 0.33) Sehhoon Park, MD, PhD Survival table Endpoint: PFS and OS (all patients); PFS by histology All patients - PFS | 40 | 11.8 mo | 7.4-NR All patients - OS | 40 | NR | 12.6-NR Thymoma - PFS | 16 | NR Thymic carcinoma - PFS | 24 | 9.2 mo HR: 1.87 HR CI: 0.67-5.20 [Slide 4] Result - Exploratory Analysis Whole-transcriptome sequencing, PR (n = 6) vs SD+PD (n = 6) Hallmark GSEA: PR vs SD+PD Epithelial Mesenchymal Transition TNFA Signaling via NFKB Inflammatory Response Coagulation Complement KRAS Signaling Up Angiogenesis IL6 JAK STAT3 Signaling IL2 STAT5 Signaling Allograft Rejection DNA Repair Mitotic Spindle MYC Targets V2 MYC Targets V1 E2F Targets G2M Checkpoint Normalized enrichment score (NES) VEGFR, VEGF ligand, and EPAS1 expression by response FLT1 (VEGFR1) KDR (VEGFR2) VEGFA VEGFB VEGFC VEGFD EPAS1 (HIF-2alpha) Gene Expression Lunit SCOPE IO immune phenotype PR (n=6) SD+PD (n=6) NE (n=3) Inflamed (n=10) Excluded (n=3) Desert (n=2) Response group Immune phenotype • Responders: angiogenesis, EMT and inflammatory programmes Non-responders: proliferation (E2F, G2M, MYC) • VEGFR1 (FLT1) and VEGFR2 (KDR) expression was higher in responders (P < 0.05) • 5 of 6 responders had an inflamed immune phenotype vs 3 of 6 non-responders (Lunit SCOPE IO) Sehhoon Park, MD, PhD

132P Abstract 6946 Reasoning LLMs with structured EHR data and for scalable, cross-national clinical trial eligibility screening K. Yoh, H. Iwai, Y. Aoyagi, K. Oda, M. Aki, M. Rioth, A. Petranovich, M. Azuma, Y. Wada, W. Farooqi, J. Hirsch, H. Izumi, Y. Zenke, H. Udagawa, T. Sakai, Y. Tanaka, Y. Murata, M. Takiguchi, K. Mizuguchi, K. Goto BACKGROUND - In the US and Europe, real-world data (RWD) from electronic medical records (EMR) are already integrated into feasibility assessments and rapid identification of eligible patients, accelerating clinical trial enrollment. - In Japan, despite widespread adoption of EMR, structured data expansion and RWD-driven feasibility assessments remain limited, resulting in slower and less efficient patient identification for clinical trials. - The aim of this study, conducted jointly by Fujitsu and Paradigm, is to develop and evaluate RWD-structuring technologies that enable feasibility assessments and rapid patient identification on a U.S. clinical trial platform. METHODS AND COHORT DEFINITIONS Characteristic | Accepted abstract | Expanded cohort Patients | 353 | 3,027 EHR period | Oct 2024–Sep 2025 | Oct 2021–Sep 2025 Genetic-test record | Required | Not required Criteria source | NCT06216301 | Physician-authored mock protocol Prompts / criterion pairs | 10 / 3,530 | 6 / 18,162 Analysis date | 7 May 2026 | 3 Sep 2026 Single-center retrospective study at NCCE. Different periods, entry conditions and prompts preclude a direct cohort performance comparison. Japanese EHR → English → Key/Key Value JSON. A US-developed platform hosted entirely in Japan used gpt-oss-120b with fixed prompts to return decisions, cited evidence and temporal context. Pilot: actionable rulings for most patients on 9 of 10 criteria; histology insufficient evidence in 299 / 350 evaluations. PLATFORM EVALUATION AND EVIDENCE From 115,943 EHR patients to an expanded cohort of 3,027 115,943 EHR patients 22,334 Respiratory care 12,601 Lung cancer-related diagnosis 10,841 Within the data period 3,195 Treatment history 3,027 Expanded cohort RET: CANDIDATE RETENTION AND REVIEW PRIORITY 3,027 patients evaluated 35 FIRST REVIEW · 0–1 missing 60 NEXT · 2 missing criteria 181 LATER · ≥3 missing criteria 2,751 clear exclusions 276 prioritized (9.1%) ONE PATIENT met all six criteria 2 RET-positive retained RET exclusion audit (n = 2,751): no incorrectly excluded candidates identified. 274 / 276 (99.3%) lacked documented RET positivity. Two RET-positive patients remained in the priority group. RESULTS RET: CRITERION DETAIL (n = 276) [table; cell values partially legible] 1. P3 RET fusion | 274 (99.3%) | 2 (0.7%) | 0 (0.0%) | 2 (0.7%) [remaining rows illegible with certainty] EGFR: CRITERION DETAIL (n = 133) [table; values not reliably legible] EGFR review sequence among 540 patients: 57 | 76 | 407 EGFR: EFFICIENCY AND PHYSICIAN-GUIDED REVIEW 540 EGFR-related cases: 57 first reviews, 51 judged appropriate 540 → 57 10.6% first review · 89.4% fewer reviews 57 / 540 selected for first review 47 definitely appropriate (82.5%) | 4 probably appropriate (7.0%) 6 probably unnecessary (10.5%) | 0 definitely unnecessary (0.0%) 57 → 51 89.5% physician-rated appropriate 51 / 57 appropriate for priority review IMPROVEMENT ANALYSIS Physician feedback enables trial-specific refinement. Limitations - Single center, retrospective data, six mock-protocol criteria and one physician reviewer. - Technical portability does not establish cross-country clinical validity. - The cohort retains 377 death records; vital status was outside the six criteria. CONCLUSION: ACTIONABLE REVIEW PRIORITIES RET | priority for verification: 9.1% — 276 / 3,027 prioritized for review; 35 first reviews (1.2% of 3,027) EGFR | focused first review: 89.4% fewer initial reviews 540 → 57; 51 / 57 (89.5%) physician-rated appropriate; 1.1 reviews per appropriate candidate Retrospective review supports two complementary uses of one platform. Trial-specific physician feedback guides refinement; any reduction below 1.1 requires validation. Final eligibility, time savings and enrollment were not evaluated. Funding: This study is funded exclusively by Fujitsu Japan Limited and Paradigm Health, Inc.

[Slide 1] LungMate-014 Study Schema - BTLA is an inhibitory immune checkpoint receptor expressed predominantly on lymphocytes that regulates T-cell activation and function. - Tifcemalimab is a recombinant humanized anti-BTLA monoclonal antibody. Antigen presenting cell HVEM MHC BTLA TCR Grb-2 p85 PI3K SHP-1 SHP-2 T cell TCR signal Key inclusion criteria: - Aged ≥18 years - Histologically or cytologically confirmed potentially resectable stage III (AJCC 8th edition) NSCLC * - ECOG performance status of 0 or 1 * Potentially resectable NSCLC was deemed resectable initially evaluated by multidisciplinary team. Safety run-in phase: Toripalimab (240mg) plus chemotherapy with tifcemalimab (200mg) N = 6 R 1:1 Toripalimab (240mg) plus chemotherapy with tifcemalimab (200mg) (Q3W for 4 cycles) N = 59 Toripalimab (240mg) plus chemotherapy (Q3W for 4 cycles) N = 59 Stratified by: Squamous vs. others; PD-L1 ≥1% vs. <1% or unknown; N stage Receive surgery or radiotherapy decided by MDT 3 weeks after the last treatment Adjuvant treatment by MDT 4-6 weeks after the last treatment Primary endpoint: pathological complete response (pCR) rate Second endpoints: major pathological response (MPR) rate, objective response rate, safety, R0 resection rate, event-free survival, and overall survival [Slide 2] Efficacy Pathological response in ITT population Tori/chemo/tifce arm (n = 59) | Tori/chemo arm (n = 59) pCR: 32.2% (19/59) (95% CI, 20.6-45.6) | 25.4% (15/59) (95% CI, 15.0-38.4) MPR: 45.8% (27/59) (95% CI, 32.7-59.2) | 40.7% (24/59) (95% CI, 28.1-54.3) Pathological response rate (%) pCR MPR Subgroup analyses of pathological response pCR rate (%) PD-L1 TPS Tori/chemo/tifce arm: PD-L1 TPS < 1%: 35.7% (10/28) (95% CI, 18.6-55.9); PD-L1 TPS ≥ 1%: 35.3% (6/17) (95% CI, 14.2-61.7) Tori/chemo arm: PD-L1 TPS < 1%: 25.9% (7/27) (95% CI, 11.1-46.3); PD-L1 TPS ≥ 1%: 18.8% (3/16) (95% CI, 4.0-45.6) Histology Tori/chemo/tifce arm: Non-squamous: 31.6% (6/19) (95% CI, 12.6-56.6); Squamous: 32.5% (13/40) (95% CI, 18.6-49.1) Tori/chemo arm: Non-squamous: 15.0% (3/20) (95% CI, 3.2-37.9); Squamous: 30.8% (12/39) (95% CI, 17.0-47.6) - Surgery was performed in 67.8% (40/59) and 59.3% (35/59) of patients in the Tori/chemo/tifce arm and Tori/chemo arm, with all achieving R0 resection. - Among patients who did not undergo surgery, local therapy was administered to 9 patients in the Tori/chemo/tifce arm (radiotherapy, n = 9) and 10 patients in the Tori/chemo arm (radiotherapy, n = 9; ablation, n = 1). Data cut-off: July 2026 [Slide 3] Survival analysis Event-free survival in ITT population Event-free survival (%) Hazard ratio = 0.78, 95% CI: 0.41-1.48 71.1% (95% CI, 60.4-83.7) 66.0% (95% CI, 54.9-79.3) Overall survival in ITT population Overall survival (%) Hazard ratio = 0.67, 95% CI: 0.30-1.50 83.0% (95% CI, 73.9-93.2) 76.1% (95% CI, 65.9-87.8) Subgroup analyses of OS in non-squamous population Hazard ratio = 0.45, 95% CI: 0.08-2.47 89.5% (95% CI, 76.7-100.0) 79.4% (95% CI, 63.3-99.6) Subgroup analyses of OS in PD-L1 TPS<1% population Hazard ratio = 0.34, 95% CI: 0.12-0.95 88.1% (95% CI, 78.8-98.5) 69.6% (95% CI, 57.1-84.9) toripalimab plus chemotherapy with tifcemalimab toripalimab plus chemotherapy Number at risk Time in months Data cut-off: July 2026; Median follow-up in all patients: 26.6 months Survival table Endpoint: EFS and OS (ITT) plus OS subgroups (non-squamous, PD-L1 TPS<1%); 24-month landmark labels as printed Toripalimab + chemotherapy + tifcemalimab (ITT) | 59 Toripalimab + chemotherapy (ITT) | 59 HR: EFS ITT 0.78; OS ITT 0.67; OS non-squamous 0.45; OS PD-L1 TPS<1% 0.34 HR CI: EFS ITT 0.41-1.48; OS ITT 0.30-1.50; OS non-squamous 0.08-2.47; OS PD-L1 TPS<1% 0.12-0.95 EFS ITT (24 mo): 71.1% (95% CI 60.4-83.7) vs 66.0% (95% CI 54.9-79.3) OS ITT (24 mo): 83.0% (95% CI 73.9-93.2) vs 76.1% (95% CI 65.9-87.8) OS non-squamous (24 mo; n=19 vs 20): 89.5% (95% CI 76.7-100.0) vs 79.4% (95% CI 63.3-99.6) OS PD-L1 TPS<1% (24 mo; n=42 vs 43): 88.1% (95% CI 78.8-98.5) vs 69.6% (95% CI 57.1-84.9) [Slide 4] Adverse events and surgical complications Treatment-related adverse events in patients receiving at least one dose of the treatment Tori / chemo / tifce arm (n = 64)* | Tori / chemo arm (n = 57) AE: Grade ≥ 3; Any grade Event | Tori/chemo/tifce any grade; grade ≥3 | Tori/chemo grade ≥3; any grade Anemia | 87.5; 4.7 | 1.8; 94.7 Neutropenia | 87.5; 68.8 | 71.9; 91.2 Leukopenia | 81.3; 26.6 | 21.1; 91.2 Thrombocytopenia | 78.1; 14.1 | 1.8; 70.1 Alanine aminotransferase increased | 67.2; 10.9 | 3.5; 52.6 Decreased appetite | 57.8 | 36.8 Fatigue / Asthenia | 56.3 | 61.4 Aspartate aminotransferase increased | 51.6; 9.4 | 3.5; 40.4 Alopecia | 50.0; 1.6 | 68.4 Lymphocyte count decreased | 50.0; 6.3 | 8.8; 45.6 Numbness of limbs | 48.4 | 50.9 Pruritus cutaneous | 42.2; 3.1 | 3.5; 45.6 Hypercholesteremia | 39.1 | 1.8; 50.9 Musculoskeletal pain | 39.1 | 31.6 Hypoproteinemia | 35.9 | 29.8 Hyponatraemia | 32.8 | 1.8; 38.6 Blood creatinine increased | 32.8 | 29.8 Hematuria | 31.3 | 28.1 Hyperglycemia | 28.1; 4.7 | 1.8; 33.3 Nausea | 26.6 | 28.1 Hypertriglyceridemia | 26.6 | 26.3 Rash | 25.0; 3.1 | 24.6 Arthralgia | 25.0 | 22.8 Hyperuricemia | 23.4 | 29.8 Proteinuria | 17.2 | 22.8 Troponin increased | 14.1 | 1.8; 15.8 Diarrhea | 14.1; 1.6 | 12.3 Constipation | 14.1 | 10.5 Blood bilirubin increased | 7.8 | 10.5 Infusion reaction | 7.8; 1.6 | 8.8 Fever | 6.3 | 19.3 Pneumonia | 4.7; [illegible] | 3.5 Pancreatitis | 1.6 | 0.0 * The safety population comprised 58 treated randomized patients and 6 patients from the safety run-in phase of the toripalimab / chemotherapy / tifcemalimab regimen (n = 64). | Tori / chemo / tifce arm (n = 64) * | Tori / chemo arm (n = 57) TRAEs, n (%) | 63 (98.4) | 57 (100.0) Grade ≥3 TRAEs, n (%) | 48 (75.0) | 43 (75.4) Serious TRAEs, n (%) | 9 (14.1) | 9 (15.8) TRAEs led to the discontinuation of treatment, n (%) | 6 (9.4) | 2 (3.5) TRAEs led to the delay of treatment, n (%) | 10 (15.6) | 6 (10.5) TRAEs led to the reduction of chemotherapy dose, n (%) | 24 (37.5) | 19 (33.3) | Tori / chemo / tifce arm (n = 40) | Tori / chemo arm (n = 35) Interval from last treatment to surgery (days), median (IQR) | 37.0 (31.8-41.5) | 34.0 (30.5-40.0) Operative time (min), median (IQR) | 170 (122.2-210.0) | 155 (112.5-228.0) Blood loss (mL), median (IQR) | 50 (50.0-100.0) | 100 (50.0-100.0) Blood transfusion, n (%) | 3 (7.5) | 6 (17.1) Duration of chest tube drainage (days), median (IQR) | 9.5 (6.8-14.3) | 12 (6.0-20.5) Length of hospital stay after surgery (days), median (IQR) | 6 (5.0-8.0) | 6 (5.0-8.5) Data cut-off: July 2026

[Slide 1] STUDY DESIGN EPIC-TC is a biomarker-selected, multi-cohort phase II study Single-center, open-label and single-arm study at Fudan University Shanghai Cancer Center 4211RO [Institution logos: Fudan University Shanghai Cancer Center and others] Key Inclusion Criteria • Age ≥18 years • Pathologically confirmed ATC or PDTC, currently in a recurrent or metastatic stage and not eligible for curative resection • EGFR positivity >10% by immunohistochemistry (IHC). • ECOG 0-1 Cohort 1: BRAF V600E wild-type patients (n=30) • MRG003, 2.0 mg/kg, Q3W • Pucotenlimab, 200 mg, Q3W Cohort 2: BRAF V600E mutant patients (n=30) • MRG003, 2.0 mg/kg, Q3W • Pucotenlimab, 200 mg, Q3W Treatment until: Progressive disease (PD) or Intolerable toxicity Primary endpoints: • Objective response rate (ORR) Secondary endpoints: • Disease control rate (DCR) • Time to response (TTR) • safety and tolerability Yu Wang, Fudan University Shanghai Cancer Center 5 [Slide 2] Objective response rate, in 31 evaluable cases Investigator-assessed response by RECIST v1.1 4211RO Best response | Total (n=31) | 95% CI | ATC (n=26) | 95% CI | PDTC (n=5) | 95% CI Pathological complete response (pCR) | 2 | | 2 | | 0 | Partial response (PR) | 13 | | 13 | | 0 | Stable disease (SD) | 5 | | 2 | | 3 | Progressive disease (PD) | 11 | | 9 | | 2 | Objective response rate (ORR) | 48.4% | 30.2%-66.9% | 57.7% | 36.9%-76.6% | 0.0% | Disease control rate (DCR) | 64.5% | 45.4%-80.8% | 65.4% | 44.3%-82.8% | 60.0% | 14.7%-94.7% Waterfall plot: Best change from baseline (%) by Individual Patient Response; ATC (n=26), PDTC (n=5) Legend: Partial response; Stable disease; Progressive disease; * Progressive disease due to new lesions; # Subsequent surgery confirmed pathological complete response after radiographic PR 48.4% Overall ORR (95% CI 30.2–66.9) 64.5% Overall DCR (95% CI 45.4–80.8) 57.7% ATC ORR (95% CI 36.9–76.6) BRAF-mutant vs wild-type: ORR 54.5% vs 45.0%; DCR 72.7% vs 60.0% (exploratory) Median TTR 43 days | PDTC n=5: ORR 0%, DCR 60.0% Yu Wang, Fudan University Shanghai Cancer Center 6 [Slide 3] ORR and DCR between BRAF-mutant and BRAF-wild patients Best response | Total (n=31) | 95% CI | BRAF-mutant (n=11) | 95% CI | BRAF-wild (n=20) | 95% CI Partial response (PR) | 15# | | 6 | | 9# | Stable disease (SD) | 5 | | 2 | | 3 | Progressive disease (PD) | 11 | | 3 | | 8 | Objective response rate (ORR) | 48.4% | 30.2%-66.9% | 54.5% | 23.4%-83.3% | 45.0% | 23.1%-68.5% Disease control rate (DCR) | 64.5% | 45.4%-80.8% | 72.7% | 39.0%-94.0% | 60.0% | 36.1%-80.9% # Two patients underwent subsequent surgery and achieved pathological complete remission (pCR) BRAF status_ORR (Case number; SD+PD / CR+PR): Mutant type: SD+PD 45.5% 95%CI (16.7,76.6); CR+PR 54.5% 95%CI (23.4,83.3) Wild type: SD+PD 55.0% 95%CI (31.5,76.9); CR+PR 45.0% 95%CI (23.1,68.5) BRAF status_DCR (Case number; PD / CR+PR+SD): Mutant type: PD 27.3% 95%CI (16.7,76.6); CR+PR+SD 72.7% 95%CI (39.0,94.0) Wild type: PD 40.0% 95%CI (19.1,63.9); CR+PR+SD 60.0% 95%CI (36.1,80.9) [Slide 4] Conclusion ➢ This is the first study for ADC in treating EGFR-positive ATC/PDTC patients in a recurrent or metastatic stage and not eligible for curative resection. ➢ Becotatug Vedotin in combination with pucotenlimab showed an encouraging antitumor activity and a manageable safety profile in this trial. ➢ ATC is an extremely rare and highly aggressive tumor with limited effective systemic treatments. Conducting a phase III randomized controlled clinical trial is challenging. This study may provide clinical evidence to change current practice. Yu Wang, Fudan University Shanghai Cancer Center

2643P Updated Efficacy and Safety of Taletrectinib in Patients With Advanced ROS1+ NSCLC After Prior Entrectinib Exposure: Results From the Global TRUST-II Study Maurice Pérol, Sara Cresta, Yuki Shinno, Pascale Tomasini, Lyudmila Bazhenova, Ernest Nadal, Wei Wang, Xianyu Zhang, Huan Huang, Santiago Viteri, Geoffrey Liu Background • Taletrectinib is a next-generation, CNS-active, selective ROS1 TKI approved in the United States, Japan, and China for patients with locally advanced or metastatic ROS1+ NSCLC • In the global Phase 2 TRUST-II (NCT04919811) study, taletrectinib demonstrated robust overall and IC efficacy, and a manageable safety profile, in both TKI-naïve and TKI-pretreated patients with ROS1+ NSCLC • Here we report updated efficacy and safety data from the TRUST-II study in patients previously treated with entrectinib, another CNS-active TKI Methods • TRUST-II (NCT04919811) is a global, multicenter, pivotal, Phase 2 study of taletrectinib in patients with ROS1+ NSCLC, including both TKI-naïve and TKI-pretreated patients TRUST-II Study Design Key eligibility criteria • Locally advanced or metastatic NSCLC with ROS1 fusion • Age ≥18 years • ECOG PS 0–1 Cohort 1 TKI-naïve; Cohort 2 TKI-pretreated -> Taletrectinib 600 mg QD Endpoints: Primary • IRC-assessed cORR per RECIST v1.1; Secondary • DOR • IC-ORR • TTR • PFS • OS • Safety • Patients in Cohort 2 received prior treatment with crizotinib or entrectinib; patients may also have received prior chemotherapy with or without immunotherapy • Here, we report updated efficacy and safety in the 10 patients from Cohort 2 with ≥1 measurable baseline lesion per RECIST v1.1 by IRC, who experienced disease progression on entrectinib and subsequently received taletrectinib 600 mg QD Results: Baseline Characteristics Patient Demographics and Baseline Characteristics (Entrectinib-Pretreated n=10 | All TKI-Pretreated N=50) Median age, years (range) 54.0 (27–79) | 55.5 (27–79) Female, n (%) 5 (50) | 27 (54) Never smoker, n (%) 6 (60) | 30 (60) Region: Asia / non-Asia, n (%) 3 (30) / 7 (70) | 22 (44) / 28 (56) Stage IV disease, n (%) 10 (100) | 49 (98) ECOG PS 1, n (%) 8 (80) | 26 (52) Prior chemotherapy, n (%) 4 (40) | 19 (38) Brain metastases at baseline, n (%) 7 (70) | 28 (56); Measurable 4 (40) | 16 (32); Non-measurable 3 (30) | 12 (24) Median time on prior entrectinib, months (range) 15.5 (6.3–84.1) | NA BOR on prior entrectinib, n (%): CR 2 (20) | NA; PR 8 (80) | NA • Among entrectinib-pretreated patients, entrectinib was the most recent anticancer therapy prior to taletrectinib, with all patients having experienced disease progression on entrectinib • Seven patients had baseline brain metastases (measurable and non-measurable) per mRECIST v1.1 and four patients had received prior chemotherapy Results: Efficacy Taletrectinib Demonstrated Robust and Durable Efficacy in Entrectinib-Pretreated Patients Efficacy (Entrectinib-Pretreated n=10) Median follow-up, months (range) 27.1 (18.7–35.9) cORR, % (95% CI) 80.0 (44.4–97.5) Median DOR, months (95% CI) 10.7 (2.6–18.2) Median PFS, months (95% CI) 9.8 (1.6–19.6) Taletrectinib Treatment Beyond PD: Patients treated beyond PD, n (%) 6 (60.0); Median duration of treatment beyond PD, months (range) 6.8 (2.3–18.1); ≥3 months 4 (66.7); ≥6 months 3 (50.0); ≥9 months 3 (50.0) Data cutoff: August 31, 2025. • Among the seven entrectinib-pretreated patients with baseline brain metastases (measurable and non-measurable), IC-ORR was 57.1% (95% CI, 18.4–90.1) and median IC-PFS was 13.7 months (95% CI, 4.3–NR) • Six entrectinib-pretreated patients continued taletrectinib treatment after investigator-assessed PD for a median of 6.8 months (range, 2.3–18.1) Results: Safety • The median duration of taletrectinib exposure in patients who received prior entrectinib treatment was 11.7 months (range, 4.2–31.9), with half of the patients receiving taletrectinib for at least 12 months • AEs of interest commonly associated with entrectinib, such as weight gain, visual disorders, or congestive heart failure, were not observed with taletrectinib in the population reported here • Despite a small sample size, the overall safety profile of taletrectinib in entrectinib-pretreated patients was consistent with that observed in the full TKI-pretreated cohort Most Frequent TEAEs Occurring in ≥20% of Entrectinib-Pretreated Patients (Entrectinib-Pretreated n=10 | All TKI-Pretreated N=50) Diarrhea 8 (80) | 31 (62); Increased AST 7 (70) | 39 (78); Increased ALT 7 (70) | 36 (72); Vomiting 4 (40) | 16 (32); Nausea 3 (30) | 19 (38); Decreased appetite 3 (30) | 6 (12); Lymphopenia 3 (30) | 3 (6); Anemia 2 (20) | 11 (22); Constipation 2 (20) | 8 (16); Dizziness 2 (20) | 6 (12); Headache 2 (20) | 6 (12); Dyspnea 2 (20) | 5 (10); Pyrexia 2 (20) | 5 (10) Conclusions • With longer follow-up, taletrectinib continued to demonstrate high overall and IC response rates in patients with ROS1+ NSCLC who were previously treated with entrectinib, another CNS-active ROS1 TKI – Confirmed ORR was 80% with a median DOR and median PFS of approximately 11 months and 10 months, respectively – Among entrectinib-pretreated patients, 60% continued treatment with taletrectinib beyond first PD, with half of those continuing for ≥9 months – Among patients with baseline brain metastases, median IC-PFS was approximately 14 months • Among patients previously treated with entrectinib, the safety profile of taletrectinib was consistent with that observed in the full TKI-pretreated cohort • These data support the use of taletrectinib as a subsequent treatment option for patients with ROS1+ NSCLC after prior entrectinib treatment and suggest sustained clinical benefit and tolerability of taletrectinib treatment beyond radiographic disease progression This study was sponsored by Nuvation Bio Inc. Presented at the European Society of Medical Oncology Congress | October 23–27, 2026 | Madrid, Spain

[Slide 1] FLAURA2 study design¹ Scan QR code to view slides and supplementary material [QR code] Patients with EGFR-mutated advanced NSCLC (N=557) Key inclusion criteria: • Aged ≥18 years (Japan: ≥20 years) • Pathologically confirmed non-squamous NSCLC • Ex19del / L858R (local / central test) • WHO PS 0 / 1 • No prior systemic therapy for advanced NSCLC • Stable CNS metastases were allowed • Brain scan at baseline (MRI / CT; mandatory) R 1:1* Osimertinib 80 mg (QD) + pemetrexed 500 mg/m² + carboplatin AUC5 or cisplatin 75 mg/m² (Q3W for 4 cycles of plat-based Tx) → Maintenance osimertinib 80 mg (QD) + pemetrexed 500 mg/m² (Q3W)† Osi 80 mg (QD)‡ • Primary endpoint: PFS (by investigator) per RECIST 1.1 • Key secondary endpoint: OS • Data cut-off: 03 April 2023 (PFS) and 12 June 2025 (OS) Exploratory endpoint: Correlation of plasma EGFRm clearance at Weeks 1, 2, 3, and 6 with PFS and OS to understand if monitoring ctDNA dynamics at earlier timepoints can be used to predict outcomes Plasma sample collection: Baseline, Week 1, Week 2, Week 3, Week 6 Baseline and Weeks 1, 2, 3, and 6 plasma samples§ from both treatment arms analysed by ddPCR (Biodesix) [Footnotes small/partly illegible:] FLAURA2 (NCT04035486) is a global, open-label, Phase III study. [illegible] 1. Planchard D, et al. N Engl J Med 2023;389:1935–1948. [remaining footnote text illegible] David Planchard, MD, PhD [Slide 2] PFS improved with osi + plat–pem versus osi mono irrespective of Week 1 and 2 plasma EGFRm clearance status Panels: Plasma EGFRm clearance at Week 1 | No Plasma EGFRm clearance at Week 1 | Plasma EGFRm clearance at Week 2 | No Plasma EGFRm clearance at Week 2 Y axis: Probability of PFS; X axis: Time from randomisation (months) Table (Week 1): Clearance No clearance mPFS (95% CI), months Osi + plat-pem NR (10.3, NR) 24.8 (16.8, 30.6) Osi mono 16.6 (13.6, 21.0) 13.6 (11.1, 16.4) HR (95% CI) 0.58 (0.33, 1.02) 0.53 (0.35, 0.80) Table (Week 2): Clearance No clearance mPFS (95% CI), months Osi + plat-pem 27.0 (19.7, NR) 22.1 (11.8, 30.0) Osi mono 13.9 (13.8, 19.4) 13.8 (5.9, 16.8) HR (95% CI) 0.63 (0.35, 0.81) 0.57 (0.24, 0.98) No. at risk (Week 1 clearance), months 0/6/12/18/24/30/36: row 1: 39 32 26 21 13 4 0 row 2: 65 49 37 22 14 7 0 No. at risk (Week 1 no clearance): row 1: [illegible, ~83] 66 49 38 15 8 0 row 2: 76 57 42 22 11 4 0 No. at risk (Week 2 clearance): row 1: 72 63 55 42 21 0 0 row 2: 85 74 53 31 19 8 0 No. at risk (Week 2 no clearance): row 1: 43 35 27 22 10 6 0 row 2: 48 30 39 14 7 4 0 Data cut-off: 03 April 2023 (PFS). [Footnotes small, partly illegible: HRs calculated using an unstratified Cox proportional hazards model. Clearance was defined as undetectable plasma EGFRm (Ex19del, L858R) at Week 1, 2, 3, or 6, where it was detected at baseline. Abbreviations.] David Planchard, MD, PhD Survival table Endpoint: PFS by plasma EGFRm clearance at Week 1 and Week 2 (osi + plat-pem vs osi mono) Week 1 clearance - Osi + plat-pem | NR | 10.3, NR Week 1 clearance - Osi mono | 16.6 | 13.6, 21.0 Week 1 no clearance - Osi + plat-pem | 24.8 | 16.8, 30.6 Week 1 no clearance - Osi mono | 13.6 | 11.1, 16.4 Week 2 clearance - Osi + plat-pem | 27.0 | 19.7, NR Week 2 clearance - Osi mono | 13.9 | 13.8, 19.4 Week 2 no clearance - Osi + plat-pem | 22.1 | 11.8, 30.0 Week 2 no clearance - Osi mono | 13.8 | 5.9, 16.8 HR: Week 1 clearance 0.58; Week 1 no clearance 0.53; Week 2 clearance 0.63; Week 2 no clearance 0.57 HR CI: 0.33, 1.02; 0.35, 0.80; 0.35, 0.81 (low confidence); 0.24, 0.98 (low confidence) [Slide 3] PFS improved with osi + plat–pem in patients with baseline-detected plasma EGFRm versus osi mono¹ • Baseline-detected plasma EGFRm was prognostic and may select for a higher degree of clinical benefit with osi + plat–pem versus osi mono PFS in patients with baseline-detected plasma EGFRm mPFS (95% CI), months Osi + plat-pem (n=147) 24.8 (19.8, 27.9) Osi mono (n=161) 13.9 (13.6, 16.6) HR (95% CI) 0.60 (0.45, 0.80) No. at risk (0/6/12/18/24/30/36 months): 147 122 101 78 38 15 0 161 128 98 56 31 13 0 PFS in patients with baseline non-detected plasma EGFRm mPFS (95% CI), months Osi + plat-pem (n=65) 33.3 (23.8, NC) Osi mono (n=48) 30.3 (25.0, NC) HR (95% CI) 0.93 (0.51, 1.72) No. at risk: 65 60 56 44 25 5 0 48 41 34 30 21 7 0 Y axis: Probability of PFS; X axis: Time from randomisation (months) HR was calculated by an unstratified log-rank test. 1. Jänne PA, et al. Cancer Res 2024;84(7_Supplement): CT017 / AACR Clinical Trials Plenary Session. [partly illegible] [Abbreviation footnote, partly illegible] David Planchard, MD, PhD Survival table Endpoint: PFS by baseline plasma EGFRm detection (osi + plat-pem vs osi mono) Arm | N | Median | Median CI Baseline-detected plasma EGFRm - Osi + plat-pem | 147 | 24.8 | 19.8, 27.9 Baseline-detected plasma EGFRm - Osi mono | 161 | 13.9 | 13.6, 16.6 Baseline non-detected plasma EGFRm - Osi + plat-pem | 65 | 33.3 | 23.8, NC Baseline non-detected plasma EGFRm - Osi mono | 48 | 30.3 | 25.0, NC HR: Baseline-detected 0.60; baseline non-detected 0.93 HR CI: 0.45, 0.80; 0.51, 1.72

[Slide 1] LBA94 vs 4236O: Two Routes to ppm-Level MRD • tumour-informed (needs tumour tissue) o tumour-naive (plasma only) Approximate LOD95, ppm (log scale) more sensitive → Axis: 10,000 (1%) | 1,000 (0.1%) | 100 (0.01%) | 10 (0.001%) | 1 (0.0001%) Assays A–H plotted; ddPCR · LBA94 comparator; Precise MRD Ultrasensitive threshold; Precise MRD LOD95 <5 ppm; single-digit ppm; ppmSeq lowest detection 2.1 ppm, LOD95 not reported here A–H: commercial assays, analytical LOD95 from company data (Sep 2024); not head-to-head. ppmSeq LOD95 not reported here, so shown as a band | LBA94 (ppmSeq) | 4236O (Precise MRD) Design | Tumour WGS → plasma WGS, no bespoke panel; median 16,417 variants, error-corrected | Tumour WGS → bespoke panel; up to 1,000 variants + depth Plasma data reusable? | Yes: re-analysis, tumour-agnostic signals (copy number, fragments) | Limited to the panel Cohorts | NSCLC (73), CRC (50); retrospective, CRC case-control | CRC (240 evaluable); prospective, unselected Lowest detection | 2.1 ppm; 28% of landmark detections <10 ppm | 53% of 1-month positives <100 ppm; lowest not stated Jeanne Tie [Slide 2] Has Ultrasensitivity Closed the Gap? | First-generation (CRC) | LBA94, CRC (ppmSeq) | 4236O, CRC (PreciseMRD) Pre-op detection | 91% (GALAXY) | 96% | 100% Post-op landmark | 53-58% (GALAXY, DYNAMIC-III) | 72% | 91% Serial sensitivity | 75% (GALAXY) | Not reported | 100% Follow-up | GALAXY 17–23 mo; DYNAMIC-III 47 mo | Not reported | 8.7 mo, 23 events Other ultrasensitive WGS-informed CRC cohorts | Post-op landmark | Follow-up CRC II–III, AccuScan (day 30 landmark) | 63% (27/43) | 36 mo CRC I–III, Pathlight (structural variant-base) | 60% (28/47) | 33 mo CRC I–IV, Next Personal | 82% (22/27) | ~21 mo Landmark 60–82% with about 2–3 years of follow-up The gain is real, but likely smaller than 91% Jeanne Tie GALAXY: Kotani Nat Med 2023, Nakamura Nat Med 2024; Tie Nat Med 2025; Hønholt AACR 2026; Merk GI ASCO 2026; Solar Vasconcelos ASCO 2026 [Slide 3] What Does Ultrasensitivity Buy Clinically? Per 100 patients after surgery 30% recurrence risk, specificity 98% | First-gen ~55% | Ultra-sens. ~80% Recurrences missed | 14 of 30 | 6 of 30 Relapse if ctDNA– | 16% | 8% ctDNA+ (treat / enrol) | 18 | 25 False positives | 1 | 1 DYNAMIC-III: ctDNA clearance with adjuvant chemotherapy, by post-operative ctDNA quartile Q1 (lowest) 83% | Q2 64% | Q3 61% | Q4 (highest) 28% ctDNA Burden So what? Fewer missed recurrences (14 → 6 per 100): ctDNA-negative relapse falls from 16% to 8% Low-level ctDNA clears more often with chemotherapy (83% vs 28%) But the extra positives are mostly low-level, where the best treatment is unknown Tie et al. Nat Med 2025 (DYNAMIC-III end-of-treatment ctDNA analysis) Jeanne Tie

2642P Osimertinib rechallenge after interstitial lung disease during osimertinib treatment in EGFR-mutated advanced non-small cell lung cancer: A retrospective chart review study Ryosuke Tsugitomi, Hisayuki Osoreda, Kyoichi Okishio, Seigo Katakura, Jin-Yuan Shih, Ji-Youn Han, Takehito Shukuya, Young-Chul Kim, Masahiro Seike, Makiko Yomota, Cheng-Yu Chang, Myung-Ju Ahn, Keith L. Davis, Abby Greenblatt, YongJin Kim, Feruza Nasirova, Melinda C. Power, Terufumi Kato Objective • To report the retrospectively collected safety and outcomes of patients with EGFR-mutated advanced NSCLC who experienced ILD during 1L osimertinib treatment, comparing those who were subsequently rechallenged with osimertinib in the 1L setting with those who were not rechallenged Conclusions • In our study, approximately one quarter (33 / 118) of patients with NSCLC who had ILD during 1L osimertinib treatment were rechallenged with osimertinib in the 1L setting • Patients who were selected by their clinicians for rechallenge were younger and had less severe ILD than those not rechallenged • Compared with those who were not rechallenged, those who were rechallenged had: – Longer rwPFS, both overall, and after restriction to those with less severe ILD (grade 1 / 2) and better ECOG PS (0 / 1) to account for differences in ILD severity across the groups – Early evidence of a trend towards rwOS benefit in the overall sample only; longer follow-up and greater patient numbers may be required to generate a clearer understanding of the impact of rechallenge on rwOS – A higher prevalence of ILD recurrence (11 / 33); more than half were grade 1 or 2 • Considering the benefits and risks, rechallenge with osimertinib may be a potential treatment option after ILD during osimertinib treatment in the 1L setting for selected patients with EGFR-mutated advanced NSCLC Introduction • Osimertinib, a 3rd-generation, oral EGFR-TKI, has demonstrated efficacy in EGFR-mutated NSCLC, including in CNS metastases • Treatment with EGFR-TKIs, including osimertinib, may lead to drug-induced ILD, a rare and potentially serious complication – In the FLAURA study, the incidence of ILD was 4% in the osimertinib arm and 2% in the standard EGFR-TKI arm – Prevalence of ILD during EGFR-TKI treatment in Asian countries is higher than in global studies and highest in Japan • Permanent discontinuation of osimertinib is recommended when ILD / pneumonitis is diagnosed during osimertinib treatment • However, emerging evidence suggests rechallenge with osimertinib may be feasible in patients with EGFR-mutated NSCLC who have experienced ILD during osimertinib treatment • In this retrospective study, we describe safety and outcomes in patients with ILD during 1L osimertinib with and without rechallenge Methods • This was a multicentre, retrospective chart review study conducted in 12 oncology centres based in Japan, the Republic of Korea and Taiwan (Figure 1) • Patient data were retrospectively collected from registries and medical record systems • Descriptive statistics were reported • Multivariable logistic regression analyses were conducted on baseline factors that varied between patients who were rechallenged and not rechallenged to identify factors associated with rechallenge • Median rwPFS and rwOS were estimated with Kaplan-Meier methods in the overall sample. We conducted parallel analyses in a sample restricted to those patients with ECOG PS 0 / 1 and ILD grade 1 / 2 given demonstrated differences in outcomes by performance status and observed differences in rechallenge status by severity of patient ILD, allowing consideration in more comparable groups • We summarised real-world prevalence of rechallenge with osimertinib after ILD during 1L treatment, as well as reporting patient characteristics and outcomes of patients who were and were not rechallenged Figure 1. Study design Eligible patients: • ≥18 years (Japan: ≥20 years) • Stage IIIB / C or IV EGFR-mutated (Ex19del / L858R) NSCLC confirmed diagnosis • Initiated 1L osimertinib in January 2018–December 2023 • Radiologically or clinically verified diagnosis of ILD after 1L osimertinib initiation, leading to interruption or discontinuation of osimertinib PRIMARY ENDPOINT • Percentage of patients rechallenged with osimertinib SECONDARY ENDPOINTS • Patient & ILD characteristics by: – Rechallenge status – ILD recurrence status after rechallenge • Subsequent treatments by rechallenge status EXPLORATORY ENDPOINTS • rwPFS & rwOS by rechallenge status • Factors associated with rechallenge *Or death, or end of the study period (30 June 2025 for Japan / Republic of Korea or 15 February 2026 for Taiwan), whichever is earliest. Results and interpretation Patients • A total of 118 patients with EGFR-mutated advanced NSCLC who had 1L osimertinib interrupted or discontinued due to ILD were included • All patients received osimertinib monotherapy, with 98% of patients receiving 1L osimertinib 80 mg QD • Patients had a median (range) age of 75 (47–93) years at advanced NSCLC diagnosis, 36% were male and 43% were current or former smokers • Median (IQR) time to onset of ILD during 1L osimertinib treatment was 2.1 (1.0–4.3) months • Osimertinib was stopped immediately on ILD diagnosis for most patients Prevalence of rechallenge • A total of 33 / 118 patients (28%) were rechallenged with osimertinib after 1L osimertinib was interrupted / discontinued due to ILD – Rechallenge took place at a median (IQR) of 0.9 (0.6–1.2) months after discontinuation / interruption of 1L osimertinib therapy – Likelihood of rechallenge varied by centre; rechallenge rates varied between 12–43% among centres who contributed >5 patients Characteristics by rechallenge status • Patients who were rechallenged versus those who were not rechallenged were more likely to be younger (median 70 vs 77 years), male (45% vs 32%) and current or former smokers (45% vs 42%) • ECOG PS was similar; both groups had 82% of patients with ECOG PS of 0 or 1 at 1L osimertinib initiation • Time to onset of ILD during 1L osimertinib treatment was similar across those who were and were not rechallenged: median (IQR) of 2.0 (0.9, 4.1) vs 2.2 (1.2, 4.3) months, respectively • Patients who were rechallenged had less severe ILD than those who were not rechallenged (Figure 2) • Multivariable logistic regression analysis found that patients who were older, had more severe ILD and had ILD that had been treated were less likely to be rechallenged after ILD (Figure 3) Figure 2. ILD severity by rechallenge status (All patients N=118; Rechallenged n=33; Not rechallenged n=85) Grade 1: 26% | 58% | 14%; Grade 2: 31% | 15% | 36%; Grade 3: 27% | 18% | 31%; Unknown: 11% | 9% | 12% Figure 3. Factors associated with rechallenge (Adjusted OR (95% CI)) ILD recurrence • ILD recurrence after ILD resolution was more common in patients who were rechallenged vs those who were not (34% vs 10%; Figure 4) Figure 4. ILD recurrence ILD recurrence among ILD resolved, rechallenged: No recurrence 66% n=22; Recurrence 34% n=11. In patients with ILD recurrence, 6 / 11 (55%) had grade 1 or 2 ILD. Grade 1-5, U: 3 3 4 0 0 1 ILD recurrence among ILD resolved, not rechallenged: No recurrence 90% n=56; Recurrence 10% n=6. In patients with ILD recurrence, 3 / 6 (50%) had grade 1 or 2 ILD. Grade 1-5, U: 2 1 1 0 1* 1 *Grade 5 ILD recurrence occurred >3 years after osimertinib was discontinued and ~2.5 years after initiating 2L 1st- / 2nd-generation EGFR-TKI therapy. • Patient characteristics were broadly similar among patients with and without ILD recurrence during rechallenge: median age 71 vs 69 years, 45% vs 45% male, 36% vs 50% current / former smokers • However, the original ILD event was often more severe in those with recurrence than those without recurrence during rechallenge (grade 1 / 2 / 3 / unknown: 45 / 9 / 36 / 9% vs 64 / 18 / 9 / 9%) Efficacy outcomes by rechallenge status • Median rwPFS was longer for rechallenged patients than for those who were not rechallenged in the overall sample (Figure 5), and after restricting the sample to patients with ECOG PS 0 / 1 and ILD grade 1 / 2 (Figure 6) to account for the difference in ILD severity across the groups • Rechallenged patients had numerically longer median rwOS, compared with those who were not rechallenged in the overall sample (Figure 7). This pattern did not persist after restricting to patients with ECOG PS 0 / 1 and ILD grade 1 / 2 (Figure 8). Wide overlapping CIs warrant further investigation with larger patient numbers and longer follow-up time for strong conclusions to be drawn Figure 5. rwPFS (overall sample): Patients not rechallenged (n=85) events 69 (81), median rwPFS 4.4 (3.6, 6.1) months; Rechallenged patients (n=33) events 26 (79), median 19.0 (8.9, 26.4) Figure 6. rwPFS (restricted sample: patients with ECOG PS 0 / 1, ILD grade 1 / 2): Not rechallenged (n=38) events 34 (89), median 5.6 (4.0, 7.6); Rechallenged (n=22) events 19 (86), median 16.6 (8.9, 27.4) Figure 7. rwOS (overall sample): Not rechallenged (n=85) events 58 (68), median rwOS 25.5 (19.3, 32.6); Rechallenged (n=33) events 19 (58), median 32.4 (26.7, 41.3) Figure 8. rwOS (restricted sample: patients with ECOG PS 0 / 1, ILD grade 1 / 2): Not rechallenged (n=38) events 22 (58), median 39.6 (21.1, 49.2); Rechallenged (n=22) events 13 (59), median 32.4 (26.7, 51.9) Subsequent treatment patterns • 1st- / 2nd-generation EGFR-TKIs and platinum-based CT were the most common first subsequent therapy for both rechallenged and not rechallenged patients • In rechallenged patients, IO was the most common second subsequent therapy, followed by 1st- / 2nd-generation EGFR-TKIs and non-platinum-based CT • In those who were not rechallenged, IO use was uncommon, with patients receiving either 1st- / 2nd-generation EGFR-TKIs or CT as second subsequent therapy Limitations: Given the observational nature of the study, rechallenge decisions were at the discretion of treating physicians; therefore, differences between groups should not be interpreted as causal Acknowledgements: This study was funded by AstraZeneca. Poster presented at the European Society for Medical Oncology (ESMO) Congress; Madrid, Spain; 23–27 October 2026

Long Term Follow-up of a Phase 1 Dose Escalation and Expansion Study of Alveltamig (ZG006), a Trispecific T Cell Engager Targeting DLL3/DLL3/CD3, as Monotherapy in Patients with Refractory Small Cell Lung Cancer (SCLC) or Neuroendocrine Carcinoma (NEC) Qiming Wang, Jianhua Shi, Xiaoli Chai, Li Zheng, Lihua Wu, Haibo Mou, Rongbo Lin, Xiujuan Qu, Linlin Liu, Qi Xu, Dan Cao, Yu Yao, Hanli Shui, Hewen Yin, Liqing Wu, Jason Wu, Binhua Lyu, Andrew X Zhu Background Alveltamig (ZG006) is a uniquely designed T cell engager, targeting CD3 and delta-like ligand 3 (DLL3) with two distinct DLL3 epitopes. ZG006 bridges tumor cells and T cells together by strongly binding to DLL3 on tumor cells and CD3 on T cells, thereby mediating T cell-specific killing of DLL3-expressing tumor cells such as small cell lung cancer (SCLC) or neuroendocrine carcinoma (NEC). Methods This is a multi-center, open-label, phase 1 dose escalation and expansion study of ZG006 as monotherapy in patients with SCLC or NEC who failed or were intolerant to the standard therapies (NCT05978284). An accelerated titration plus 3+3 dose-escalation design was employed in this trial. Patients received ZG006 intravenously with escalated doses from 0.1 to 100 mg, once every 2 weeks, and dose expansion is conducted in the 10 mg, 30 mg and 60 mg dose cohorts. A step-up dose of 1 mg was given at 10 mg and above. Results The dose-escalation phase has been completed as planned. Here we report the long term follow-up results for the three effective Q2W cohorts of 10, 30, and 60 mg, as of June 2026. A total of 39 patients (34 were SCLC patients) were enrolled, with 5, 19 and 15 participants in the 10 mg, 30 mg and 60 mg cohorts, respectively, with a median follow-up time of approximately 20 months. Baseline characteristics for all three groups are summarized in Table 1. All subjects (100%) in the three cohorts experienced adverse reactions. Grade ≥3 events were reported in 40.0%, 57.9%, and 53.3% of subjects, respectively. The incidence of CRS was 80.0%, 73.7%, and 73.3%, respectively, across the three cohorts, and events were predominantly Grade 1–2, with only one subject in the 10 mg group experienced Grade 3 CRS. Detailed adverse reaction profiles are provided in Table 2. Among patients with SCLC, the confirmed ORR was 75.0% in the 10 mg Q2W cohort, 61.1% in the 30 mg Q2W cohort, and 91.7% in the 60 mg Q2W cohort; the DCR was 75.0%, 72.2%, and 91.7%, respectively. The mPFS was 12.3 months, 5.4 months, and 11.2 months, respectively. The mOS was 19.4 months and 15.6 months in the 10 mg and 30 mg Q2W cohorts, respectively, and was not reached in the 60 mg Q2W cohort. The 12- and 18-month OS rates were 100.0%/50.0%, 55.6%/48.6%, and 75.0%/66.7%, respectively. Additionally, ZG006 demonstrated promising antitumor activity in patients with baseline high, medium, and low levels of DLL3 expression. Detailed efficacy results are presented in Table 3 and Figure 1. Table 1. Baseline Characteristics (10mg Q2W (N=5) | 30mg Q2W (N=19) | 60mg Q2W (N=15)) Median Age 56.0 | 59.0 | 59.0 Gender Male 5 (100.0) | 13 (68.4) | 11 (73.3); Female 0 | 6 (31.6) | 4 (26.7) ECOG 0: 0 | 2 (10.5) | 3 (20.0); 1: 5 (100.0) | 17 (89.5) | 12 (80.0) Cancer Type SCLC 4 (80.0) | 18 (94.7) | 12 (80.0); NEC 1 (20.0) | 1 (5.3) | 3 (20.0) Metastasis Yes 5 (100.0) | 19 (100.0) | 15 (100.0) Prior Lines of Therapy 1: 1 (20.0) | 1 (5.3) | 3 (20.0); 2: 4 (80.0) | 8 (42.1) | 5 (33.3); 3: 0 | 9 (47.4) | 7 (46.7); ≥4: 0 | 1 (5.3) | 0 Prior PD-(L)1 Therapy Yes 3 (60.0) | 15 (78.9) | 13 (86.7); No 2 (40.0) | 4 (21.1) | 2 (13.3) Table 2. Common Treatment-Related AEs for All Grades (10mg Q2W (N=5) | 30mg Q2W (N=19) | 60mg Q2W (N=15)) TRAE 5 (100.0) | 19 (100.0) | 15 (100.0) Anaemia 3 (60.0) | 16 (84.2) | 9 (60.0) Cytokine release syndrome 4 (80.0) | 14 (73.7) | 11 (73.3) White blood cell count decreased 4 (80.0) | 10 (52.6) | 12 (80.0) Pyrexia 4 (80.0) | 7 (36.8) | 9 (60.0) Decreased appetite 1 (20.0) | 13 (68.4) | 8 (53.3) Asthenia 3 (60.0) | 10 (52.6) | 10 (66.7) Pruritus 1 (20.0) | 10 (52.6) | 9 (60.0) Hyponatraemia 0 | 13 (68.4) | 10 (66.7) Hypoalbuminaemia 2 (40.0) | 10 (52.6) | 10 (66.7) Constipation 2 (40.0) | 12 (63.2) | 7 (46.7) Neutrophil count decreased 2 (40.0) | 8 (42.1) | 9 (60.0) Alanine aminotransferase increased 3 (60.0) | 8 (42.1) | 3 (20.0) Hypokalaemia 1 (20.0) | 4 (21.1) | 7 (46.7) Nausea 2 (40.0) | 5 (26.3) | 6 (40.0) Weight decreased 3 (60.0) | 4 (21.1) | 4 (26.7) Table 3. Efficacy in SCLC (Confirmed Response: 10mg Q2W (N=4) | 30mg Q2W (N=18) | 60mg Q2W (N=12) | Total (N=34)) ORR, n (%) 3 (75.0) | 11 (61.1) | 11 (91.7) | 25 (73.5); 95% CI (19.4, 99.4) | (35.8, 82.7) | (61.5, 99.8) | (55.6, 87.1) DCR, n (%) 3 (75.0) | 13 (72.2) | 11 (91.7) | 27 (79.4); 95% CI (19.4, 99.4) | (46.5, 90.3) | (61.5, 99.8) | (62.1, 91.3) mPFS (Months) 12.3 | 5.4 | 11.2 | 9.6; 95% CI (1.4, NE) | (1.6, 11.3) | (4.0, NE) | (4.2, 13.7) mOS (Months) 19.4 | 15.6 | NR | 23.3; 95% CI (12.4, NE) | (6.3, NE) | (7.4, NE) | (11.5, NE) 12-month DoR rate (95% CI) 66.7 (5.4, 94.5) | 33.8 (8.0, 62.7) | 50.0 (18.4, 75.3) | 44.4 (23.1, 63.7) 12-month OS rate (95% CI) 100.0 (100.0, 100.0) | 55.6 (30.5, 74.8) | 75.0 (40.8, 91.2) | 67.6 (49.2, 80.6) 18-month OS rate (95% CI) 50.0 (5.8, 84.5) | 48.6 (24.1, 69.3) | 66.7 (33.7, 86.0) | 55.3 (37.0, 70.2) Figure 1. Best Percentage Changes in Target Lesion Size by Dose Levels in SCLC (DLL3: N:Negative; L:Low expression; M:Middle expression; H:High expression; /:Unknown) Conclusions Based on the updated data from the three effective Q2W dose levels (10 mg, 30 mg, and 60 mg), Alveltamig continues to demonstrate excellent antitumor activity in patients with SCLC who have failed existing standard therapies. With longer-term follow-up, Alveltamig further confirms its pronounced and durable efficacy. Although OS data remain immature, the benefit observed to date is encouraging. The mOS was 19.4 months in the 10 mg cohort, with a pooled mOS of 23.3 months, highlighting the durable survival benefit of Alveltamig.

2627P Enozertinib (ORIC-114), a Brain-Penetrant, Highly Selective EGFR Inhibitor, in Treatment-Naïve Advanced NSCLC with EGFR Atypical Mutations Myung-Ju Ahn, Min Hee Hong, Cheen Leng Lee, Enriqueta Felip Font, Pei Jye Voon, Omar Saavedra, Valentina Boni, Sonam Puri, Thomas John, Adnan Khattak, Byoung-Yong Shim, Anneleen Daemen, Art Weber, Karen Velastegui, Jian Wang, Edna Chow Maneval, Pratik S. Multani, Jiunn-Liang Tan BACKGROUND • Epidermal growth factor receptor (EGFR) mutations occur in approximately 10%–15% of patients with non–small cell lung cancer (NSCLC) in Western populations and up to 50% in Asian populations – Atypical and exon 20 insertion mutations in EGFR comprise 10%–30% of EGFR mutations in NSCLC and are molecularly heterogeneous • Approximately 35% of patients with metastatic EGFR-mutated NSCLC present with de novo central nervous system (CNS) metastases, and approximately 50% develop CNS metastases during their disease course • Enozertinib is an oral, highly brain-penetrant, selective, irreversible inhibitor designed for potent activity against EGFR atypical and exon 20 insertion mutations being evaluated in an ongoing, first-in-human, open-label, multi-center Phase 1b study (NCT05315700) METHODS • A total of 26 patients treated with enozertinib 80 mg orally once daily are presented here with data cutoff date of September 4, 2026 (Figure 1, Table 1) Figure 1. Summary of Key Eligibility Criteria and Endpoints Key Eligibility Criteria: • Treatment-naive, locally advanced or metastatic NSCLC with EGFR atypical mutations • PACC or classical-like • Singleton or compound mutations • Measurable disease per RECIST 1.1 • ECOG PS 0 or 1 • Treated, or untreated and asymptomatic CNS metastases permitted Key Endpoints: • Antitumor activity per RECIST 1.1 – ORR / DOR / PFS • Intracranial response per modified RECIST 1.1 (INV) and RANO-BM (BICR) • ctDNA molecular response • Safety BASELINE CHARACTERISTICS • Seventeen of 26 patients (65%) had brain metastases at baseline, including 12 patients (46%) with untreated brain metastases (Table 1) Table 1. Baseline Characteristics | Patients (N=26) Age, years, median (range) | 66 (26–79) Female, n (%) | 10 (38) Asian / White / Other, n (%) | 20 (77) / 6 (23) / 0 Never smoker, n (%) | 9 (35) ECOG PS 0 / 1, n (%) | 7 (27) / 19 (73) Brain metastases at baseline, n (%) | 17 (65) Previously untreated brain metastases | 12 (46) Atypical mutation type, n (%): PACC 23 (88); L861Q 3 (12) Compound mutations, n (%) | 18 (69) SAFETY • Enozertinib demonstrated a manageable safety profile, consistent with known toxicities associated with wild-type EGFR inhibition (Table 2, Table 3) Table 2. Summary of Treatment-Related Adverse Events in All Patients | N=26 Grade 3 | 6 (23) Serious adverse events | 1 (4) Dose interruptions | 9 (35) Dose reductions | 10 (39) Treatment discontinuations | 0 Data cutoff: September 4, 2026 • TRAEs predominantly Grades 1-2 • All Grade 3 TRAEs resolved with dose modifications • No Grade 4 or Grade 5 TRAEs • No significant off-target toxicities (e.g., cardiac, hematologic, hepatic) • No discontinuations due to TRAEs Table 3. Treatment-Related Adverse Events (TRAEs) Occurring in ≥20% of Patients | Grade 1–2 | Grade 3 Diarrhea | 15 (58) | 2 (8) Paronychia | 15 (58) | 1 (4) Dermatitis acneiform | 12 (46) | 2 (8) Rash | 14 (54) | 0 Stomatitis | 12 (46) | 0 Pruritus | 7 (27) | 0 Decreased appetite | 6 (23) | 0 Dry skin | 6 (23) | 0 Nausea | 6 (23) | 0 SYSTEMIC EFFICACY Table 4. Investigator and BICR-Assessed Response Rates | Response-Evaluable (n=25) Investigator-assessed: Best ORR, % (95% CI) 88 (69, 98); Confirmed ORR, % (95% CI) 84 (63, 96); Partial response, n (%) 21 (84); Stable disease, n (%) 4 (16); DCR, n (%) 25 (100) BICR-assessed: Confirmed ORR, % (95% CI) 80 (59, 93) Data cutoff: September 4, 2026. Figure 2. Tumor Reduction Observed in All Efficacy-Evaluable Patients, Including Patients with CNS Disease at Baseline Figure 3. Responses to Enozertinib Were Observed Early, With Most Patients Remaining on Treatment • At a median follow-up of 10.5 mo (95% CI: 8.4, 13.9), median time on treatment has not been reached. 69% (18/26) of patients remained on study treatment • At a landmark of 6 months, 88% of patients remain progression-free Figure 4. EGFR-Mutant ctDNA Clearance Observed in 82% of Patients by Cycle 3 Day 1 • All 22 ctDNA-evaluable patients had a ≥90% reduction in EGFR-mutant ctDNA clearance by Cycle 3 Day 1, with ctDNA clearance achieved in 18 patients (82%) INTRACRANIAL EFFICACY Table 5: Investigator-Assessed Intracranial Response Rates | Response Rate per modified RECIST 1.1 | Response-Evaluable (n=10) Best ORR, % (95% CI) 80 (44, 98); Confirmed ORR, % (95% CI) 70 (35, 93); Complete response, n (%) 2 (20); Partial response, n (%) 5 (50); Stable disease, n (%) 3 (30); DCR, n (%) 10 (100) Figure 5. In the Context of 90% of Response-Evaluable Patients Having No Prior Treatment for their CNS Metastases, All Patients Achieved CNS Tumor Reductions • No new CNS lesions in patients with or without baseline CNS metastases Table 6. BICR-Assessed Intracranial Response Rates | Response Rate per RANO-BM | Response-Evaluable (n=15) Confirmed ORR, % (95% CI) 67 (38, 88); Complete response, n (%) 6 (40); Partial response, n (%) 4 (27); Stable disease, n (%) 3 (20); Progressive disease, n (%) 2 (13); DCR, n (%) 13 (87) Figure 6. Intracranial Tumor Reduction Observed in Nearly All Efficacy-Evaluable Patients by BICR per RANO-BM • Six complete responses, including in three patients with non-measurable CNS disease Complete Intracranial Response in Patient with Treatment-Naïve EGFR Atypical Mutation and Previously Untreated CNS Metastases • 59yo M with G719C-S768I mutation • Baseline CNS Metastases: 3 target lesions and 2 non-target lesions, all previously untreated • Systemic Response: Confirmed PR (-81%) after Cycle 4 • CNS Response: PR after Cycle 3 and CR after Cycle 14 • TRAEs: G2 dermatitis acneiform (resolved), G1 diarrhea (resolved), G1 dry skin • Duration of Treatment: Cycle 18 (ongoing) CONCLUSIONS • Enozertinib 80 mg once daily demonstrated clinically meaningful antitumor activity in a population of patients with treatment-naive advanced NSCLC harboring EGFR atypical mutations and an exceptionally high rate of CNS metastases – Meaningful clinical responses: confirmed INV-ORR 84%, BICR-ORR 80%; 88% remain progression free at 6 months – Compelling intracranial responses: confirmed INV-ORR 70%, BICR-RANO-BM ORR 67%; multiple complete responses – Molecular responses: 82% of evaluable patients achieved full clearance of EGFR-mutant ctDNA • Enozertinib demonstrated a manageable safety profile, with no significant off-target toxicity and no TRAE-related discontinuations • These data support continued investigation of enozertinib as frontline therapy for patients with NSCLC harboring EGFR atypical mutations This study is sponsored by ORIC Pharmaceuticals, Inc. European Society for Medical Oncology (ESMO) Congress; Madrid, Spain; October 23–27, 2026

[Slide 1] Efficacy: Group 1 (≥2L NSCLC, monotherapy) CS2009 monotherapy demonstrated promising clinical efficacy in heavily pre-treated NSCLC | All Evaluable Patients | 30 mg/kg | Squamous, 30 mg/kg | Non-squamous, 30 mg/kg ORR, % (n) | 16.7% (9/54) | 28.0% (7/25) | 33.3% (4/12) | 23.1% (3/13) DCR, % (n) | 68.5% (37/54) | 60.0% (15/25) | 75.0% (9/12) | 46.2% (6/13) mDoR, month | 10.5 | 10.5 | - | - Waterfall legend: 10 mg/kg, 20 mg/kg, 30 mg/kg, 45 mg/kg, Prior Anti-angiogenic, Prior IO Spider plot legend: 10 mg/kg, 20 mg/kg, 30 mg/kg, 45 mg/kg, First New Lesion, Treatment ongoing CS2009 is well tolerated with treatment duration exceeded 12 months at 10, 20, and 30 mg/kg. Dr. Anwen Xiong DCR: disease control rate; IO: immunotherapy; L: line; mDoR: median duration of response; NSCLC: non-small cell lung cancer; NSQ: non-squamous; ORR: objective response rate; PD: progressive disease; PR: partial response; SD: stable disease; SQ: squamous; UK: unknown. [Slide 2] Efficacy: Group 2 (2/3L NSCLC, in combination with docetaxel) CS2009 + docetaxel demonstrated promising preliminary clinical efficacy in IO-pretreated NSCLC | All Evaluable Patients | 20 mg/kg | 30 mg/kg ORR, % (n) | 55.6% (10/18) | 55.6% (5/9) | 55.6% (5/9) DCR, % (n) | 100% (18/18) | 100% (9/9) | 100% (9/9) Waterfall (20 mg/kg / 30 mg/kg) histology: SQ NSQ SQ NSQ SQ NSQ SQ NSQ SQ NSQ SQ NSQ SQ SQ SQ NSQ SQ SQ PD-L1 (%): 3 5 UK <1 0 0 2 2 UK 0 30 <1 0 <1 UK <1 UK UK Best response: SD SD SD SD SD SD SD SD PR PR PR PR PR PR PR PR PR PR Spider plot legend: 20 mg/kg, 30 mg/kg, Treatment ongoing Dr. Anwen Xiong DCR: disease control rate; DoR: duration of response; IO: immunotherapy; L: line; NSCLC: non-small cell lung cancer; NSQ: non-squamous; ORR: objective response rate; PFS: progression-free survival; PR: partial response; SD: stable disease; SQ: squamous; UK: unknown.

[Slide 1] Phase 1 Dose Escalation and Expansion with Ixo-V Monotherapy in Solid Tumors GCT patient population • Confirmed advanced or metastatic solid tumor malignancy • No available standard of care therapy • Not amenable to surgery • Prior HDCT and ASCT allowed • For dose expansion: CLDN6 positivity^a by IHC was required Key endpoints Primary: • AEs by CTCAE 5.0 Secondary: • Best objective response by investigator per RECIST v1.1 Escalation First-in-human starting dose (0.2 mg/kg Q3W) | Grade ≥2 related AE | Recommended phase 2 dose Accelerated dose titration (1 patient per cohort) | Standard "3 + 3" dose escalation Expansion CLDN6-positive platinum-resistant ovarian cancer expansion CLDN6-positive refractory NSCLC expansion Other CLDN6-positive refractory cancers expansion (including GCT) CLDN6-low/negative expansion CLDN6-low/positive ovarian cancer Ixo-V + combinations (including bevacizumab) CLDN6-positive NSCLC Ixo-V + pembrolizumab [Slide 2] Ixo-V dose level | <2.4 mg/kg (n=3) | 2.4 mg/kg (n=3) | 3 mg/kg^a (n=4) | 3.4 mg/kg^b (n=1) | Total N=11 Median age, years (range) | 30.0 (27.0, 44.0) | 45.0 (31.0, 76.0) | 39.0 (26.0, 46.0) | 23.0 (23.0, 23.0) | 37.0 (23.0, 76.0) Sex, n (%) Male | 3 (100.0) | 3 (100.0) | 3 (75.0) | 1 (100.0) | 10 (90.9) Female | 0 | 0 | 1 (25.0) | 0 | 1 (9.1) ECOG PS at baseline, n (%) 0 | 2 (66.7) | 1 (33.3) | 3 (75.0) | 1 (100.0) | 7 (63.6) 1 | 1 (33.3) | 2 (66.7) | 1 (25.0) | 0 | 4 (36.4) Number of prior lines of therapy, n (%) 1-2 | 0 | 1 (33.3) | 1 (25.0) | 0 | 2 (18.2) ≥3 | 3 (100.0) | 2 (66.7) | 3 (75.0) | 1 (100.0) | 9 (81.8) Median prior lines of therapy, n (range) | 4.0 (3.0, 5.0) | 4.0 (1.0, 4.0) | 3.0 (2.0, 4.0) | 3.0 (3.0, 3.0) | 3.0 (1.0, 5.0) Type of prior therapy, n (%) Platinum compounds | 3 (100.0) | 3 (100.0) | 4 (100.0) | 1 (100.0) | 11 (100.0) Taxanes | 3 (100.0) | 2 (66.7) | 3 (75.0) | 0 | 8 (72.7) HDCT/ASCT | 3 (100.0) | 0 | 2 (50.0) | 1 (100.0) | 6 (54.5) [Slide 3] [Waterfall plot] Legend: 0.4 mg/kg | 1.0 mg/kg | 2.0 mg/kg | 2.4 mg/kg | 3.0 mg/kg | 3.4 mg/kg Ovarian tumor (circle) | Mediastinal tumor (triangle) | Received prior HDCT/ASCT (diamond) | * Tx ongoing Best change from baseline in sum of diameter, % Patient ID: OC04 T01 T06 T04 T03 T07 T08 T02 T05 S07 Parameter^a | All dose levels (N=11) BOR, n (%) PR | 5 (45.5) SD | 3 (27.3) PD | 2 (18.2) NE | 1 (9.1) ORR, % | 45.5 95% CI | (16.7-76.6) DCR, % | 72.7 95% CI | (39.0-94.0) mPFS, months | 5.8 95% CI | (1.4-13.8) ^aDOR was immature at data cutoff date. [Slide 4] TEAE | TRAE Constipation Arthralgia Back pain Nausea Vomiting Fatigue Neutropenia Peripheral neuropathy, NOS Peripheral sensory neuropathy Peripheral motor neuropathy Alopecia Cough Decreased appetite Insomnia Patients, % (axis 60-0-60) TEAE legend: Grade 1 | Grade 2 | Grade 3 | Grade 4 TRAE legend: Grade 1 | Grade 2 | Grade 3 | Grade 4

Baseline Characteristics Characteristic | N = 40, n (%) Age, years Median (range) | 60 (29-82) Sex Male | 25 (62.5) Female | 15 (37.5) ECOG performance status 0 | 13 (32.5) 1 | 27 (67.5) Histology (WHO) Thymic carcinoma | 24 (60.0) Thymoma | 16 (40.0) Type A / B1 | 1 (2.5) / 3 (7.5) Type B2 | 8 (20.0) Type B3 | 4 (10.0) Characteristic Line of rivoceranib 2nd line | 13 (32.5) 3rd line | 11 (27.5) ≥4th line | 16 (40.0) Prior systemic therapy Cisplatin, doxorubicin, cyclophosphamide | 23 (57.5) Carboplatin, paclitaxel | 18 (45.0) Etoposide + cisplatin or carboplatin | 13 (32.5) Immune checkpoint inhibitor | 9 (22.5) Sites of metastasis Pleura | 25 (62.5) Lung | 20 (50.0) Lymph node | 14 (35.0) Liver | 10 (25.0) Bone | 8 (20.0) Patient disposition • 43 screened; 40 received rivoceranib (3 screen failures) • Data cut-off April 2026

[Slide 1] Results: Cohort A Primary Tumor Detection Sensitivity: 23/23 (100%) Specificity: 6/7 (85.7%) Accuracy: 29/30 (96.7%) Median SUVmax: primary ccRCC, 64.65; tumor thrombus, 34.83 Metastatic lesion detection 20 additional metastatic lesions detected 82.6% increase in lesion yield: 23 → 42 Sensitivity: 97.7% (42/43) A Lymph node B Adrenal glands C Lung D Bone Representative [68Ga]Ga-DPI-4452 PET/CT images of distant metastases in ccRCC [Slide 2] Results: Cohort B Per-lesion Detection Performance [68Ga]Ga-DPI-4452 Detected / Not Detected All lesions 95.6% 585/612 Kidney 100.0% 34/34 Lung 96.7% 319/330 Bone 93.5% 115/123 Lymph node 88.4% 61/69 Muscle 100.0% 18/18 Adrenal gland 100.0% 15/15 Pancreas 100.0% 14/14 Tumor thrombus 100.0% 9/9 Detection rate(%) 2-[18F]FDG Detected / Not Detected All lesions 61.4% 376/612 Kidney 88.2% 30/34 Lung 53.3% 176/330 Bone 79.7% 98/123 Lymph node 44.9% 31/69 Muscle 77.8% 14/18 Adrenal gland 73.3% 11/15 Pancreas 64.3% 9/14 Tumor thrombus 77.8% 7/9 Detection rate(%) Lesion SUVmax [68Ga]Ga-DPI-4452 / 2-[18F]FDG All P< 0.001 Kidney 60.05 / 4.35 Lung 6.10 / 1.32 Bone 11.60 / 4.00 Lymph node 14.50 / 1.21 Muscle 49.05 / 3.21 Adrenal gland 28.10 / 3.21 Pancreas 26.20 / 2.19 Tumor thrombus 39.10 / 3.60 Median SUVmax

2671P High-dose firmonertinib combined with bevacizumab and pemetrexed in patients with EGFR-mutated NSCLC and LM: An ambispective real-world study Qi Zhao, Haiyang Chen, Qiming Wang* Background • Leptomeningeal metastasis (LM) is a devastating complication of EGFR-mutated (EGFRm) non-small cell lung cancer (NSCLC), with an increasing incidence of up to 11.1%. • Our previous study demonstrated that high-dose firmonertinib achieved a median overall survival (mOS) of 8.43 months. • In our multicentre retrospective FURMO-LM study of 618 patients, furmonertinib-based triplet therapy was associated with longer OS than monotherapy or doublet regimens. • Building on these findings, we evaluated the efficacy and safety of high-dose furmonertinib combined with bevacizumab and pemetrexed in patients with EGFR-mutant NSCLC and LM. Methods • This study enrolled patients with EGFRm NSCLC and LM diagnosed via EANO-ESMO criteria. Participants received firmonertinib (240 mg QD) combined with bevacizumab (15 mg/kg, Q3W) and pemetrexed (either 50 mg intrathecal [IT] or 500 mg/m² intravenous [IV] Q3W). • The primary endpoint was OS. Secondary endpoints included time to treatment failure (TTF), LM objective response rate (ORR-LM) per RANO-LM criteria, clinical response rate, and safety. Results Patients • From March 2023 to May 2025, 61 patients were enrolled at Henan Cancer Hospital, 57 were evaluable (4 excluded due to inadequate treatment duration). • Baseline characteristics: 18 (31.6%) 19del, 34 (59.6%) L858R, 46(80.7%) had prior third-generation EGFR-TKI therapy, and 7(12.3%) received > 3 prior lines of systemic treatment. Table 1. Baseline characteristics of patients (Patient, n (%)) Age (years) Median 57 (34-82) Sex Male 27 (47.4%); Female 30 (52.6%) Histology Adenocarcinoma 56 (98.2%); Adenocarcinoma with squamous differentiation 1 (1.8%) Smoking history Never 39 (68.4%); Current/Former 18 (31.6%) ECOG performance status 1: 6 (10.5%); 2: 20 (35.1%); 3: 24 (42.1%); 4: 7 (12.3%) EGFR mutation type L858R 34 (59.6%); Ex19del 18 (31.6%); G719X 1 (1.8%); L861Q 1 (1.8%); C797S 1 (1.8%); G719A and S768I 1 (1.8%); Ex18delins 1 (1.8%) Intracranial pressure before treatment ≤180 mmH2O 17 (29.8%); >180 mmH2O 39 (68.4%); NA 1 (1.8%) Type of LM Type I 56 (98.2%); Type II 1 (1.8%) Prior 3-generation EGFR-TKI therapy No 11 (19.3%); Yes 46 (80.7%) Prior brain radiotherapy No 39 (68.4%); WBRT 8 (14.0%); CSI 1 (1.8%); Radiotherapy for brain metastases 9 (15.8%) Route of Pemetrexed administration Intravenous (IV) 8 (14.0%); Intrathecal (IT) 41(71.9%); Switch from IV to IT 8 (14.0%) Efficacy • At the data cutoff of 9 May 2026,The mOS was 15.5 months, mTTF was 12.6 months and median survival from LM diagnosis was 19.7 months (Figure 1).the clinical response rate was 70.2%, Among 40 radiographically evaluable patients, ORR-LM was 52.5% and disease control rate (DCR) was 95% (Figure 2). Figure 1. Kaplan-Meier curves of OS, TTF and OS from LM diagnosis A Median OS: 15.5 (10.9−20.0) B Median TTF:12.6 (9.4−17.3) C Median OS from LM diagnosis: 19.7 (16.3−34.5) Figure 2. Clinical, radiographic and cytological responses A No response 17 (29.8%); Response 40 (70.2%) B Radiographic assessment of LM: CR 3 (5.3%); PR 18 (31.6%); SD 17 (29.8%); PD 2 (3.5%); NE 17 (29.8%) C CSF cytology evaluation: Negative 7 (12.3%); Positive 14 (24.6%); NE 36 (63.2%) • Firmonertinib dose was reduced in 10 patients (17.5%): to 200 mg (n=3), 160 mg (n=6), or 120 mg (n=1); IT pemetrexed was reduced to 30 mg in 3 patients (5.3%). Table 2. Treatment-related adverse events (TRAEs) | Overall, n (%) Any grade | Grade 1-2 | Grade 3-4 Rash | 8 (14.0) | 7 (12.3) | 1 (1.8) Diarrhea | 19 (33.3) | 17 (29.8) | 2 (3.5) Nausea | 7 (12.3) | 6 (10.5) | 1 (1.8) Increased creatine kinase | 9 (15.8) | 9 (15.8) | 0 (0) Increased alanine aminotransferase | 3 (5.3) | 3 (5.3) | 0 (0) Increased aspartate aminotransferase | 5 (8.8) | 5 (8.8) | 0 (0) Increased γ-glutamyltransferase | 4 (7.0) | 3 (5.3) | 1 (1.8) Leucopenia | 25 (43.9) | 13 (22.8) | 12 (21.1) Neutropenia | 22 (38.6) | 12 (21.1) | 10 (17.5) Anemia | 25 (43.9) | 23 (40.4) | 2 (3.5) Decreased platelet count | 14 (24.6) | 10 (17.5) | 4 (7.0) Prolonged electrocardiogram QT interval | 1 (1.8) | 1 (1.8) | 0 (0) Hyperuricemia | 3 (5.3) | 3 (5.3) | 0 (0) Increased creatinine | 6 (10.5) | 4 (7.0) | 2 (3.5) Oral mucositis | 9 (15.8) | 9 (15.8) | 0 (0) Paronychia | 3 (5.3) | 3 (5.3) | 0 (0) Epistaxis | 3 (5.3) | 3 (5.3) | 0 (0) Proteinuria | 2 (3.5) | 2 (3.5) | 0 (0) Safety • Treatment-related adverse events (TRAEs) of any grade occurred in 86% of patients. • Grade 3/4 TRAEs occurred in 28.1% (16/57) of patients and were primarily hematological: leukopenia (21.1%), neutropenia (17.5%), and thrombocytopenia (7.0%). Conclusions • The triple combination of high-dose firmonertinib, bevacizumab, and pemetrexed demonstrates promising clinical efficacy and a manageable safety profile in patients with EGFRm NSCLC and LM. • This regimen may offer a potent therapeutic option, even for patients who have progressed on prior third-generation EGFR-TKIs. *Corresponding to: Qiming Wang, MD. Henan Cancer Hospital, China

3817P ZL-1310 (DLL3 ADC) plus atezolizumab ± carboplatin in Extensive-stage Small Cell Lung Cancer Jun Zhao, Valentina Gambardella, Wenxiu Yao, Zhen Wang, Pedro Rocha, Jie Hu, María Eugenia Olmedo García, Qiming Wang, David Vicente Baz, Luis Paz-Ares, Mariam Alexander, Anne C. Chiang, Manuel Cobo Dols, Yingying Du, Grace K. Dy, Martin Gutierrez, Alexander Spira, Liza Villaruz, Renke Zhou, Yi-Long Wu Background • Despite advances in first-line chemoimmunotherapy, extensive-stage small cell lung cancer (ES-SCLC) remains associated with a poor prognosis; in addition, treatment-related toxicities from chemotherapy may limit the tolerability, highlighting the need for more effective and better tolerated therapies. • Delta-like ligand 3 (DLL3) is highly expressed in SCLC. zocilurtatug pelitecan (ZL-1310) is a novel DLL3-targeted antibody–drug conjugate (ADC) utilizing the TMALIN® platform to deliver a topoisomerase I inhibitor payload to tumor cells (Figure 1). • In the ongoing Ph1 study (NCT06179069), ZL-1310 monotherapy at 1.6 mg/kg Q3W demonstrated encouraging intracranial activity, and was generally well tolerated. Figure 1. Structure and Mechanism of Action of ZL-1310: A: Humanized anti-DLL3 IgG1 mAb; B: Cleavable tripeptide-based linker; C: Camptothecin derivative payload, C24 Method • Phase I, open-label, dose-escalation and expansion study of ZL-1310 as monotherapy and in combination with atezolizumab, with (triplet) or without (doublet) carboplatin in ES-SCLC (NCT06179069) (Figure 2). • For patients that received doublet therapy: • The dose-escalation cohorts enrolled previously treated patients. • The back-fill cohorts enrolled 1L or 1L maintenance patients at 1.2 mg/kg and 1.6 mg/kg after DLT was cleared. • The expansion cohort enrolled 1L ES-SCLC patients at 1.6 mg/kg. • For patients that received triplet therapy: all patients received study drugs as 1L treatment. • Efficacy and safety reported for 1L patients who had the opportunity of at least 12-week follow-up with data cutoff date of Sept 9, 2026. Assessments • Safety: AE assessed per CTCAE v5.0 • Response was assessed by investigator per RECIST v1.1. Tumor evaluation is scheduled every 6 weeks for the first 30 weeks, then every 9 weeks till disease progression or death / withdrawn consent / lost to follow-up. Figure 2. ZL-1310-001 Study Schema ZL-1310 plus atezolizumab ± carboplatin Doublet: ZL-1310 + atezolizumab 1200 mg IV Q3W; dose escalation ZL-1310 0.8 mg/kg + atezolizumab (n=3), 1.2 mg/kg + atezolizumab (n=6), 1.6 mg/kg + atezolizumab (n=7); Backfill; Expansion. This report includes data from 43 1L ES-SCLC patients, including 33 treated with ZL-1310 at 1.6 mg/kg and 10 at 1.2 mg/kg. Triplet: ZL-1310 + atezolizumab 1200 mg + carboplatin as induction, followed by ZL-1310 + atezolizumab as maintenance, IV Q3W; ZL-1310 0.8 mg/kg + atezolizumab + carboplatin 5 mg/ml/min; ZL-1310 0.8 mg/kg + atezolizumab + carboplatin 4 mg/ml/min; ZL-1310 1.2 mg/kg + atezolizumab + carboplatin 4 mg/ml/min. This report includes data from all 13 patients who received the triplet regimen as 1L treatment for ES-SCLC. Baseline Characteristics for 1L patients ZL-1310 + atezolizumab doublet: • At ZL-1310 1.6 mg/kg, 33 patients were enrolled; the first 15 in the back fill cohort had the opportunity for ≥6 months of follow-up, with a median follow-up of 10.0 months. • At ZL-1310 1.2 mg/kg, 10 patients were enrolled; all had ≥6 months follow-up, with a median follow-up of 12.6 months. ZL-1310 + atezolizumab + carboplatin triplet: • In the dose-escalation triplet cohorts, 13 patients were enrolled (Table 1), with median follow-up of 11.8 months. Table 1. Demographics and Baseline Characteristics for 1L Patients (Doublet: 1.2 mg/kg (n=10) | 1.6 mg/kg (n=33) | Total (n=43); Triplet: Carbo AUC5 0.8 mg/kg (n=4) | Carbo AUC4 0.8 mg/kg (n=3) | Carbo AUC4 1.2 mg/kg (n=6) | Total (n=13)) Male, n (%) 8 (80.0) | 26 (78.8) | 34 (79.1) | 3 (75.0) | 3 (100) | 4 (66.7) | 10 (76.9) Age, median (range), years 62 (51, 75) | 63 (46, 75) | 63 (46, 75) | 61 (58, 66) | 69 (66, 69) | 68 (59, 75) | 66 (58, 75) ECOG PS = 1, n (%) 7 (70.0) | 25 (75.8) | 32 (74.4) | 3 (75.0) | 3 (100) | 5 (83.3) | 11 (84.6) Race Asian 10 (100) | 30 (90.9) | 40 (93.0) | 3 (75.0) | 2 (66.7) | 2 (33.3) | 7 (53.8); White 0 | 3 (9.1) | 3 (7.0) | 0 | 1 (33.3) | 4 (66.7) | 5 (38.5) Brain Metastasis-Yes, n (%) 2 (20.0) | 6 (18.2) | 8 (18.6) | 1 (25.0) | 0 | 2 (33.3) | 3 (23.1) Stage IV disease at initial diagnosis, n (%) 8 (80.0) | 25 (75.8) | 33 (76.7) | 3 (75.0) | 2 (66.7) | 5 (83.3) | 10 (76.9) Results Efficacy of doublet as 1L therapy • High overall response rates were observed in patients receiving ZL-1310 + atezolizumab as 1L therapy (ORR for all dose level: 81.4%). Table 2. Best Overall Response per RECIST v1.1 by Investigator Assessment (ZL-1310 1.2 mg/kg + Atezolizumab (N=10) | ZL-1310 1.6 mg/kg + Atezolizumab (N=33)) CR 0 | 0; PR 8 (80.0) | 27 (81.8); SD 2 (20.0) | 2 (6.1); PD 0 | 2 (6.1) Confirmed ORR, % (95% CI) 80.0 (44.4, 97.5) | 81.8 (64.5, 93.0) • Responses occurred rapidly: 34/35 (97.1%) responders achieving their response at the first post-baseline tumor assessment. • The responses were durable (Figure 4). • In the 10 patients receiving ZL-1310 1.2 mg/kg + atezolizumab as 1L, mPFS and mDOR was 8.9 (4.0, NA) and 8.3 (4.0, NA) months, respectively. • In the first 15 patients receiving ZL-1310 1.6 mg/kg + atezolizumab as 1L, mPFS is not reached after a median follow-up of 10.0 months. 66.7% of these patients had not yet experienced disease progression or death at this data cut-off. mDOR is not reached. Efficacy of triplet as 1L therapy • Patients receiving ZL-1310 in combination with atezolizumab and carboplatin as 1L therapy demonstrated a high overall response rate (ORR) of 92.3%. • mPFS is 8.4 months (95% CI 3.7, NA) and mDOR is 7.7 months (2.6, NA). Safety of doublet as 1L therapy • ZL-1310 plus atezolizumab is well tolerated. • No DLTs were observed in the dose escalation phase. • The majority of AEs were Grade 1 or 2, and non-serious. • Grade ≥3 AEs related to ZL-1310 occurred in 7/43 (16.3%) of patients, and SAEs related to ZL-1310 occurred in 5/43 (11.6%) of patients. • AE rate related to ZL-1310 that led to dose reduction (4.7%, 2/43) or discontinuation (1/43, 2.3%) was low. • No treatment-related deaths were reported. Table 3. Summary of Adverse Events - Doublet Safety Analysis Set (1.2 mg/kg (N=10) | 1.6 mg/kg (N=33) | Total (N=43)) Any 10 (100) | 32 (97.0) | 42 (97.7); Related to ZL-1310 10 (100) | 31 (93.9) | 41 (95.3) Serious 4 (40.0) | 6 (18.2) | 10 (23.3); Related to ZL-1310 2 (20.0) | 3 (9.1) | 5 (11.6) Grade ≥ 3 5 (50.0) | 7 (21.2) | 12 (27.9); Related to ZL-1310 4 (40.0) | 3 (9.1) | 7 (16.3) Leading to ZL-1310 Drug Interruption 5 (50.0) | 9 (27.3) | 14 (32.6); Related to ZL-1310 4 (40.0) | 5 (15.2) | 9 (20.9) Leading to ZL-1310 Dose Reduction 0 | 2 (6.1) | 2 (4.7); Related to ZL-1310 0 | 2 (6.1) | 2 (4.7) Leading to ZL-1310 Drug Discontinuation 1 (10.0) | 0 | 1 (2.3); Related to ZL-1310 1 (10.0) | 0 | 1 (2.3) Safety of triplet as 1L therapy • Compared with the doublet, the triplet was associated with increased hematologic toxicity, more frequent Grade ≥3 AEs, and a higher rate of dose interruptions. Although MTD was not met, the ZL-1310 dose was not escalated above 1.2 mg/kg. • A single DLT, Grade 4 thrombocytopenia, was observed in ZL-1310 0.8 mg/kg + Atezolizumab + Carboplatin AUC 5. No DLTs occurred in the ZL-1310 + Atezolizumab + Carboplatin AUC 4 cohorts. • Grade ≥3 AEs related to ZL-1310 occurred in 7/13 (53.8%) of participants, and SAEs related to ZL-1310 occurred in 3/13 (23.1%). • One patient from ZL-1310 1.2 mg/kg + Atezolizumab + Carboplatin AUC 4 cohort had over-dose of Carboplatin and treatment-related death due to a G5 Sepsis. Table 4. Summary of Adverse Events - Triplet Combination Safety Analysis Set (0.8 mg/kg + AUC 5 (N=4) | 0.8 mg/kg + AUC 4 (N=3) | 1.2 mg/kg + AUC 4 (N=6) | Total (N=13)) Any 4 (100) | 3 (100) | 6 (100) | 13 (100); Related to ZL-1310 3 (75.0) | 2 (66.7) | 5 (83.3) | 10 (76.9) Serious 2 (50.0) | 0 | 5 (83.3) | 7 (53.8); Related to ZL-1310 1 (25.0) | 0 | 2 (33.3) | 3 (23.1) Grade ≥ 3 3 (75.0) | 1 (33.3) | 6 (100) | 10 (76.9); Related to ZL-1310 2 (50.0) | 1 (33.3) | 4 (66.7) | 7 (53.8) Leading to ZL-1310 Drug Interruption 4 (100) | 3 (100) | 4 (66.7) | 11 (84.6); Related to ZL-1310 3 (75.0) | 2 (66.7) | 4 (66.7) | 9 (69.2) Leading to ZL-1310 Dose Reduction 0 | 0 | 1 (16.7) | 1 (7.7); Related 0 | 0 | 1 (16.7) | 1 (7.7) Leading to ZL-1310 Drug Discontinuation 0 | 0 | 1 (16.7) | 1 (7.7); Related 0 | 0 | 1 (16.7) | 1 (7.7) Common AEs related to ZL-1310 and ILD / pneumonitis • Doublet: The most common AEs related to ZL-1310 of all grades (≥20%) were anaemia (48.8%), nausea (39.5%), leukopenia (34.9%), decreased appetite (30.2%), ALT increased, neutropenia (27.9% each), hypertriglyceridaemia and lymphopenia (23.3% each) • Triplet: The most common AEs related to ZL-1310 of all grades (≥20%) were anaemia and neutropenia (61.5% each), thrombocytopenia (53.8%), nausea (46.2%), ALT increased, decreased appetite, and leukopenia (38.5% each), AST increased, constipation, malaise, and vomiting (30.8% each), and alopecia, hyperuricaemia, and hypoaesthesia (23.1% each). • ILD / pneumonitis was low in both regimens: 2/43 (4.7%) in doublet, and 2/13 (15.4%) in triplet, all Grade 1. Conclusion • ZL-1310 1.6 mg/kg and atezolizumab doublet combination demonstrated high response rate (81.8%) and durable response as 1L treatment in ES-SCLC (median PFS for the first 15 patients who had 10-month follow-up not yet obtained). • ZL-1310 1.6 mg/kg and atezolizumab doublet combination is well tolerated. The overall AE rate, Grade 3 and above AE rate, SAE rate with this regimen is not higher than that of ZL-1310 1.6 mg/kg monotherapy as previously reported. Only one G1 ILD / pneumonitis was reported so far in patients receiving ZL-1310 1.6 mg/kg and atezolizumab as 1L treatment (1/33, 3.0%). • ZL-1310 1.6 mg/kg and atezolizumab doublet combination has a more favorable risk / Benefit profile compared to ZL-1310 + atezolizumab + carboplatin triplet regimen. • A Phase III randomized 1L study will be started to further evaluate a combination regimen in ES-SCLC (NCT07796100). Acknowledgement: This study was sponsored by Zai Lab (US) LLC. Atezolizumab study drug was provided by Roche.

#3827P DAREON®-9: A Phase Ib Open-Label Dose Escalation/Confirmation Study of Obrixtamig (BI 764532) Plus Topotecan in Patients with Extensive-Stage Small Cell Lung Cancer (ES-SCLC) – Topotecan Use Analysis Martin Wermke*, Jürgen Alt, Farastuk Bozorgmehr, Florian Fuchs, Pauline du Rusquec, Anas Gazzah, Céline Mascaux, Sean Dulloo, Ewa Kalinka, Aman Chauhan, Estelamari Rodriguez, Anna Mosikian, Iryna Tsyhankova, Dian Yang, Cyrus Sayehli Introduction • SCLC is a type of NEC with poor survival outcomes • DLL3 is an inhibitory Notch ligand that is expressed on ~94% of SCLC tumor cells Obrixtamig is a novel DLL3/CD3 IgG-like T-cell engager • DAREON®-9 (NCT05990738) is an ongoing Phase I dose-escalation/confirmation trial assessing the combination of obrixtamig + topotecan in patients with advanced relapsed/refractory ES-SCLC • Data from the dose-escalation phase demonstrated encouraging efficacy and an acceptable safety profile in patients with previously treated ES-SCLC • This analysis assessed whether topotecan treatment patterns affect the safety or efficacy of the combination Methods Part A: dose escalation (BLRM with overdose control; N=44): Obrixtamig with 3 target dose levels (10, 30, 60 mg) + topotecan -> RP2D -> Part B: dose confirmation (N=16): Obrixtamig 60 mg + topotecan Obrixtamig was given as step-up dosing followed by target dose (3 dose levels; IV Q1W for three weeks, then Q3W); topotecan was given per label (QD on Days 1–5 of each cycle, 2.3 mg/m² orally or 1.5 mg/m² IV). G-CSF use was permitted from the start of topotecan treatment, per label and local clinical practice Patients were treated until disease progression, undue toxicity, consent withdrawn, or any other reasons requiring discontinuation. Patients could continue obrixtamig ± topotecan beyond disease progression (according to RECIST v1.1) if, in the opinion of the investigator, they were deriving clinical benefit Key inclusion criteria • Histologically/cytologically confirmed SCLC • ES-SCLC that progressed/recurred following ≥1 line of platinum-based treatment as per SoC (including anti-PD-1/PD-L1) • Eligible for topotecan treatment per label • Adequate organ function Key exclusion criteria • Previous treatment with T-cell engagers or DLL3-targeting therapies • Untreated/symptomatic brain metastases • Prior anticancer therapy ≤14 days prior to first obrixtamig treatment • Topotecan relative dose intensity (RDI) for Cycle 1 to PD was calculated as: RDI (%) = 100 x (cumulative total dose received) / (planned cumulative total dose), where planned cumulative total dose is calculated as the dose that would have been administered per protocol without dose modifications, up to the DLP date, death, or last treatment, whichever occurred first • For the sensitivity analysis, RDI for Cycles 1–X was calculated as the mean of per-cycle RDIs through the cycle of PD; cycles after PD were excluded Patients • As of November 11, 2025, 60 patients had been enrolled and received ≥1 dose of obrixtamig Total (N=60): Median age, years (range) 64.5 (33–81); Male, n (%) 37 (62); Prior lines of treatment 1/2/≥3, n (%) 31 (52) / 18 (30) / 11 (18); Prior use of PD-1/PD-L1 antibodies, n (%) 57 (95); ECOG PS 0/1, n (%) 24 (40) / 36 (60); Median number of cycles of obrixtamig / topotecan, n (range) 8 (1–28) / 8 (1–16); Median treatment exposure of obrixtamig / topotecan, months (range) 5.4 (0.3–19.6) / 6.7 (0.2–10.8) • 15 (25%) patients continued receiving topotecan beyond PD and 20 (33%) patients continued receiving obrixtamig beyond PD Safety Dose modifications due to AEs (Due to topotecan-related AEs (N=60) | Due to obrixtamig-related AEs (N=60)) Dose modification of topotecan, n (%): Dose delay 8 (13) | 4 (7); Dose reduction 20 (33) | 2 (3); Discontinuation 12 (20) | 2 (3) Dose modification of obrixtamig, n (%): Dose delay 9 (15) | 7 (12); Discontinuation 0 | 1 (2) • There were no dose reductions of obrixtamig Dose modifications of both treatments due to AEs were primarily attributed to topotecan-related AEs Most common hematologic TRAEs (≥5% of patients) Total (N=60): Any grade | Grade ≥3 Any TRAE, n (%) 60 (100) | 51 (85) Topotecan-related hematologic AEs: Thrombocytopenia/decreased platelets 52 (87) | 35 (58); Anemia 47 (78) | 19 (32); Neutropenia/decreased neutrophils 37 (62) | 27 (45); Decreased lymphocytes 7 (12) | 7 (12); Febrile neutropenia 5 (8) | 5 (8); Leukopenia 3 (5) | 2 (3) Obrixtamig-related hematologic AEs: Decreased lymphocytes 20 (33) | 15 (25); Anemia 5 (8) | 0 Most hematologic AEs observed in this study were attributed to topotecan. The most common obrixtamig-related hematologic AE was decreased lymphocytes, consistent with the expected effects of obrixtamig's mechanism of action Safety (cont.) • Median duration of grade ≥3 neutropenia: 8 days (range: 1–16) • Median time to onset of grade ≥3 neutropenia: 8 days (range: 5–29) Grade ≥3 neutropenia mostly occurred in Cycles 1–2; Most patients with grade ≥3 neutropenia had 1–2 episodes Infection events ≤14 days since neutropenia onset: Gr 1–2 9 (15%); Gr ≥3 3 (5%); Median duration 7 days (1–184). Hospitalizations due to: Neutropenia 2 (3%); Febrile neutropenia 5 (8%) • No grade 5 neutropenia or treatment-related infection occurred • Of patients with grade ≥3 treatment-emergent thrombocytopenia and anemia, 33% and 85% received platelet and RBC transfusions, respectively • G-CSF was commonly used (primary/secondary prophylaxis: 93%/18%) • Obrixtamig-related CRS: grade 1: 45%; grade 2: 12%; grade 3: 3% • Median time to CRS onset: 2 days (range: 1–29) • Obrixtamig-related ICANS/ICANS-like neurotoxicity: n=3 (all grade 1) • DLTs during MTD evaluation period: n=3 (grade 3 tumor pain; grade 2 increased blood creatinine; grade 3 fatigue) Efficacy • Efficacy was evaluable in 46 patients (≥1 assessment performed) Efficacy (cont.) Efficacy by topotecan RDI • Overall, confirmed ORR was 65%: 72% in patients with RDI <80% and 61% in patients with RDI ≥80% • Overall, confirmed mDoR was 6.9 months: 5.6 months in patients with RDI <80% and 7.3 months in patients with RDI ≥80% • Overall, confirmed mPFS was 5.6 months and was identical in patients with RDI <80% and ≥80% (5.6 months in each group) • Overall, 33 patients (55%) discontinued topotecan before obrixtamig • Of these patients, median time from topotecan discontinuation to PD was 2.6 months (95% CI: 1.4–4.3) Topotecan RDI did not impact ORR or PFS, although mDoR tended to be longer with full-dose topotecan. A sensitivity analysis showed no evidence of poorer outcomes with lower topotecan exposure during early cycles. These findings are limited by the small patient numbers Key Findings and Conclusions • Obrixtamig + topotecan showed a manageable safety profile in relapsed/refractory ES-SCLC • Neutropenia occurred early, was manageable, and was consistent with the known topotecan safety profile • Many patients continued obrixtamig after discontinuing topotecan, including beyond PD, suggesting ongoing clinical benefit • Minor differences between topotecan RDIs indicate topotecan dose modifications did not have a significant negative impact on overall promising efficacy This study was funded by Boehringer Ingelheim. Presented at the European Society for Medical Oncology (ESMO) Congress, Madrid, Spain, October 23–27, 2026 *Corresponding author email address: martin.wermke@ukdd.de

2026 ESMO #2659P Phase 2 study of SHR-1826, a c-MET directed antibody-drug-conjugate, combined with PD-L1 inhibitor and CTLA-4 inhibitor in previously treated advanced/metastatic non-squamous NSCLC Yi Hu, Fan Zhang, Yongzhong Luo, Xingxiang Pu, Zhiye Zhang, Yajie Zhu, Linlin Wang, Li Wang, Tianqing Chu, Yan Yu, Xinghua Han, Yanwei Yin, Yongxu Jia, Yina Wang, Huijuan Wang, Changliang Yang, Yan Li, Huang Xu, Li Song, Zongyan Xie MADRID SPAIN 23-27 OCTOBER 2026 Background • c-MET dysregulation is common in NSCLC and is associated with poor prognosis and resistance to EGFR-TKI. • Although telisotuzumab adizutecan, a novel c-Met–targeting ADC, is being investigated in patients with c-Met–overexpressing tumors, effective therapies remain lacking for the majority of patients with c-Met overexpression who do not harbor MET gene alterations. • SHR-1826, a novel c-MET-targeting ADC comprising a humanized IgG2 antibody conjugated to a topoisomerase I inhibitor, has demonstrated a manageable safety profile in patients with advanced solid tumors in a phase 1 study. Method • This study (NCT06844474) was an open-label, multicenter Phase II clinical trial to evaluate the safety, tolerability and efficacy of SHR-1826 for injection in patients with NSCLC (Figure 1). • This presentation reports results in previously treated non-squamous NSCLC. Screening: Key eligibility criteria • Advanced or metastatic NSCLC • EGFR sensitizing mutation-negative and ALK fusion-negative • ECOG PS of 0 or 1 • ≥1 measurable lesion per RECIST v1.1 Treatment Period: SHR-1826 4 mg/kg IV Q3W + SHR-8068 (a CTLA-4 inhibitor) 280 mg IV once on Day 1, then 70 mg IV Q6W from Cycle 5 + Adebrelimab 1200 mg IV Q3W Safety Lead-in Phase • First enrolled 3–6 NSCLC patients who failed prior first-line standard therapy, with ≤ 2 prior lines of therapy • SMC reviewed safety event incidence, safety, and efficacy data → decided on expansion to ~25–50 patients Efficacy Expansion Phase • Treatment-naïve NSCLC • Squamous NSCLC and Non-squamous NSCLC; ~25–50 patients per sub-cohort Follow-up: Safety Follow-up Period • Continued until 60 (±7) days after the last SHR-1826 dose, 90 (±7) days after the last SHR-8068 dose, and 90 (±7) days after the last adebrelimab dose, whichever was later. Survival Follow-up • After treatment and safety follow-up, continue survival follow-up (OS, etc.) every 60 ± 7 days from 90 days post-last dose, until death, withdrawal of consent, loss to follow-up, or study termination, whichever occurs first. • Primary Endpoints: Safety, ORR • Secondary Endpoints: DCR, DoR, PFS, OS Figure 1. Study design Result Patients • As of Mar. 9, 2026, 37 patients were enrolled and received the treatment (median age, 63.0 years; ECOG PS 1, 86.5%; second line therapy, 32.4%. • The baseline characteristics are summarized in Table 1. With a median follow-up of 4.3 months, 28 (75.7%) patients were still on treatment (Figure 2). Table 1. Baseline characteristics (SHR-1826 + Adebrelimab + SHR-8068 (N=37)) Age, years, median (range) 63.0 (36.0-75.0) Female, n (%) 23 (62.2) ECOG PS of 1, n (%) 32 (86.5) Clinical staging at screening n(%) 37 (100): IIIC 2 (5.4); IVA 9 (24.3); IVB 26 (70.3) Prior lines of therapy, n (%) 37 (100): 1 25 (67.6); 2 12 (32.4) c-MET positive, n (%) 24 (64.9); c-MET 2+/3+ ≥ 50%* 11 (29.7) PD-L1 TPS, n (%) 37 (100): < 1 27 (73.0); ≥ 1 9 (24.3); Unknown 1 (2.7) Prior therapy, n (%) 37 (100): Immunotherapy 35 (94.6); Non-platinum Chemotherapy 34 (91.9); Platinum Chemotherapy 35 (94.6) * ≥50% of tumor cells showing 2+ or 3+ staining intensity. Efficacy • The unconfirmed ORR was 42.4% (95% CI 25.5, 60.8), and the unconfirmed DCR was 87.9% (95% CI 71.8, 96.6); among the 14 responders, the median duration of response was immature at the data cutoff (Table 3). • Most patients had a reduction of tumor size from baseline (Figure 2. 3). • Median PFS was 8.3 months (95% CI: 5.32, NR) (Table 3, Figure 4). Median overall survival was not matured. Table 3. Efficacy summary (SHR-1826 + Adebrelimab + SHR-8068) Unconfirmed Best overall response, n 33 Unconfirmed Complete response, n (%) 0 (0) Unconfirmed Partial response, n (%) 14 (42.4) Stable disease, n (%) 15 (45.5) Progressive disease, n (%) 3 (9.1) Not evaluable, n (%) 1 (3) Unconfirmed Objective response rate, % (95% CI) 42.4 (25.5, 60.8) Unconfirmed Disease control rate, % (95% CI) 87.9 (71.8, 96.6) Progression-free survival, months, median (95% CI) 8.3 (5.3, NR) Rates at 3 months, % (95% CI) 85.2 (64.9, 94.3) Rates at 6 months, % (95% CI) 57.2 (30.2, 77.1) Rates at 9 months, % (95% CI) 42.9 (14.2, 69.3) Figure 2. Target lesion diameter relative to baseline Figure 3. The optimal percentage change in target lesion diameter relative to baseline Figure 4. Kaplan-Meier plot of progression-free survival; Number at risk 37 31 24 15 13 11 6 5 4 3 2 1 0 (0–12 months) Safety • Overall safety was manageable, with hematologic and gastrointestinal toxicities predominating, a slightly higher incidence of rash (Table 2). • One patient (2.7%) experienced a TRAE of grade 3 interstitial lung disease that led to treatment discontinuation. • One had a TRAE leading to death with unknown cause. Table 2. Safety profile and common TRAEs (N=37) TRAEs 35 (94.6); Grade ≥ 3 TRAEs 22 (59.5); Serious TRAEs 19 (51.4); TRAEs leading to treatment interruption 17 (45.9); TRAEs leading to treatment reduction 5 (13.5); TRAEs leading to treatment withdrawn 1 (2.7); TRAEs leading to death 1 (2.7); Immune related AEs 6 (16.2); Grade ≥3 2 (5.4) TRAEs occurring in ≥ 20% of patients (Any grade | Grade ≥3): Anaemia 23 (62.2) | 1 (2.7); White blood cell count decreased 17 (45.9) | 7 (18.9); Rash 16 (43.2) | 4 (10.8); Nausea 14 (37.8) | 1 (2.7); Platelet count decreased 14 (37.8) | 7 (18.9); Neutrophil count decreased 13 (35.1) | 11 (29.7); Decreased appetite 13 (35.1) | 0; Hypoalbuminaemia 10 (27.0) | 0; Asthenia 9 (24.3) | 0; Gamma-glutamyltransferase increased 9 (24.3) | 0; Vomiting 8 (21.6) | 1 (2.7); Alopecia 8 (21.6) | 0 Conclusion SHR-1826 combined with Adebrelimab and SHR-8068 showed encouraging antitumor activity and manageable safety profile in patients with preciously treated non-squamous NSCLC. Follow-up is ongoing for long-term efficacy and safety data. Funding: This study was sponsored by Jiangsu Hengrui Pharmaceuticals Co., Ltd. Corresponding author: Prof. Yi Hu

2660P Phase 2 trial of the PD-1/VEGF bispecific antibody, tivunatamig (PF-08634404 / SSGJ-707), plus chemotherapy in advanced NSCLC: updated results Lin Wu, Haipeng Xu, Yulan Sun, Fangling Ning, Siyuan Huang, Dingzhi Huang, Yan Yu, Feng Ye, Dongqing Lv, Zhidong Pei, Junguo Lu, Tao Xu, Xingxiang Pu, Xiang Li, Ruoyu Wang, Laiyu Liu, Junfang Tang, Anwen Liu, Eeman Shaikh, Jing Lou Conclusions • Tivunatamig (PF'4404) combined with platinum-based chemotherapy (chemo) continues to demonstrate clinically meaningful and durable antitumor activity in first-line (1L) treatment of patients with advanced non-small cell lung cancer (NSCLC) across tumor histology and programmed death ligand 1 (PD-L1) expression – In non-squamous NSCLC, the confirmed objective response rate (ORR) was 77.5% with tivunatamig 10 mg/kg + chemo vs 44.7% with tislelizumab + chemo; median progression-free survival (PFS) was 13.8 vs 9.7 months, respectively – In squamous NSCLC, the confirmed ORR was 80.7% with tivunatamig 10 mg/kg + chemo (paclitaxel) vs 67.7% with tislelizumab + chemo; median PFS was 13.6 vs 8.3 months, respectively • Tivunatamig combined with platinum-based chemo demonstrated a safety profile that was manageable and consistent with the known safety profile of chemo combined with programmed death 1 (PD-1) and vascular endothelial growth factor (VEGF) inhibitors – Grade ≥3 treatment-related adverse events (TRAEs) occurred in 55.8% of patients with tivunatamig + chemo (all doses) vs 49.3% for tislelizumab + chemo – Safety was consistent independent of tumor histology • Exploratory analysis of circulating tumor DNA (ctDNA) clearance via a tumor-agnostic 172-gene panel showed correlation with PFS with better PFS in patients with ctDNA clearance at cycle 3 day 1 (C3D1); 61.9% of patients had ctDNA clearance after 2 cycles of tivunatamig + chemo • These results continue to support 5 ongoing Symbiotic trials in lung cancer, including the ongoing phase 3 pivotal Symbiotic-Lung-01 trial of tivunatamig 10 mg/kg combined with chemo in 1L treatment of NSCLC regardless of histology and PD-L1 expression (see QR code) Background • Tivunatamig is a novel and structurally distinct PD-1×VEGF bispecific engineered with augmented VEGF inhibition and enhanced cooperative binding with PD-1 to enable robust immune activation in the tumor microenvironment (Figure 1) • Tivunatamig monotherapy demonstrated promising efficacy and manageable safety in patients with treatment-naive tumor proportion score (TPS) ≥1% NSCLC – With a median follow-up duration of 18.2 months, confirmed ORR was 67.6%, and median PFS was 12.4 months • Earlier data from the phase 2 study (NCT06412471) of tivunatamig combined with platinum-based chemo in 1L treatment of NSCLC demonstrated encouraging efficacy and manageable safety regardless of tumor histology and PD-L1 expression • Here, we report updated efficacy and safety results from this study after an additional 13 months of follow-up Methods • This is an open-label, multicenter, randomized phase 2 trial of tivunatamig or tislelizumab (anti–PD-1) combined with platinum-based chemo for the treatment of patients with treatment-naïve advanced NSCLC (Figure 2) • The primary objectives are to evaluate the safety, tolerability, and antitumor activity of different dosing regimens of tivunatamig combined with platinum-based chemo • Eligible patients have histologically confirmed treatment-naïve stage IIIB/C or IV NSCLC, regardless of PD-L1 expression – Patients are excluded if they have EGFR or ALK alterations, BRAF V600E mutations, MET exon 14 skipping mutations, RET fusions, or ROS1 fusions • In part 1, patients with non-squamous NSCLC were randomized to receive tivunatamig at 5/10/20 mg/kg or tislelizumab with carboplatin + pemetrexed • In part 2 cohort A, patients with squamous NSCLC were randomized to receive tivunatamig at 5/10/20 mg/kg + carboplatin + paclitaxel – Based on the selected dose from cohort A, part 2 cohort B evaluated tivunatamig 10 mg/kg or tislelizumab with paclitaxel + carboplatin or tivunatamig 10 mg/kg + nab-paclitaxel + carboplatin in squamous NSCLC • Baseline demographics and efficacy results are shown from the recommended phase 3 dose of tivunatamig 10 mg/kg every 3 weeks (Q3W); safety and ctDNA results include all tivunatamig doses evaluated. ctDNA was determined using a tumor-agnostic 172-gene panel (HapOnco® mClean Lung Cancer MRD) Figure 2. Study design: Part 1: 1L non-squamous NSCLC (n=122): Tivunatamig 5 mg/kg Q3W + pemetrexed + carboplatin (n=40); Tivunatamig 10 mg/kg Q3W + pemetrexed + carboplatin (n=41); Tislelizumab + pemetrexed + carboplatin (n=40). Part 2: 1L squamous NSCLC (n=139): Cohort A: Tivunatamig 5 mg/kg Q3W + paclitaxel + carboplatin (n=30); Tivunatamig 10 mg/kg Q3W + paclitaxel + carboplatin (n=30); Cohort B: Tivunatamig 10 mg/kg Q3W + paclitaxel + carboplatin (n=32); Tivunatamig 10 mg/kg Q3W + nab-paclitaxel + carboplatin (n=14); Tislelizumab + paclitaxel + carboplatin (n=31) Results Efficacy results • At the data cutoff (August 6, 2026), 122 patients with non-squamous NSCLC and 139 with squamous NSCLC were enrolled in parts 1 and 2, respectively • Across the treatment groups, the majority of patients were male and had former/current smoking status; >40% of the patients had TPS <1% NSCLC (Table 1) • Median follow-up was between 16.2 and 18.6 months across the tivunatamig 10 mg/kg and tislelizumab cohorts (Table 2) • High confirmed ORR and durable responses were observed irrespective of histology (Table 3 and QR code) • Median PFS was 13.8 months with tivunatamig vs 9.7 months with tislelizumab in patients with non-squamous NSCLC; in patients with squamous NSCLC, median PFS was 13.6 months (paclitaxel) and 13.4 months (nab-paclitaxel) vs 8.3 months (Figure 3) Table 2. Patient disposition, follow-up time, and doses administered (Tivunatamig 10 mg/kg + carboplatin + pemetrexed (n=41) | Tislelizumab + carboplatin + pemetrexed (n=40) | Tivunatamig 10 mg/kg + carboplatin + paclitaxel (n=62) | Tivunatamig 10 mg/kg + carboplatin + nab-paclitaxel (n=14) | Tislelizumab + carboplatin + paclitaxel (n=31)) Patients continuing on treatment, n (%) 13 (31.7) | 12 (30.0) | 21 (33.9) | 6 (42.9) | 10 (32.3) Follow-up time, median (95% CI), months 18.5 (14.3-20.1) | 18.4 (15.0-20.5) | 17.6 (15.5-19.8) | 18.6 (16.9-20.5) | 16.2 (13.9-17.5) Doses administered, median (range), n 14.0 (1-30) | 10.0 (1-30) | 11.5 (1-34) | 14.0 (2-29) | 11.0 (1-29) Table 3. Best overall response (Part 1: Non-squamous NSCLC: Tivunatamig 10 mg/kg + carboplatin + pemetrexed | Tislelizumab + carboplatin + pemetrexed; Part 2: Squamous NSCLC: Tivunatamig 10 mg/kg + carboplatin + paclitaxel | Tivunatamig 10 mg/kg + carboplatin + nab-paclitaxel | Tislelizumab + carboplatin + paclitaxel) Patients evaluable, n 40 | 38 | 57 | 13 | 31 Confirmed ORR, % (95% CI) 77.5 (61.5-89.2) | 44.7 (28.6-61.7) | 80.7 (68.1-90.0) | 69.2 (38.6-90.9) | 67.7 (48.6-83.3) CR 0 | 0 | 1 (1.8) | 0 | 0 PR 31 (77.5) | 17 (44.7) | 45 (78.9) | 9 (69.2) | 21 (67.7) SD 8 (20.0) | 16 (42.1) | 10 (17.5) | 4 (30.8) | 9 (29.0) Unconfirmed PR 0 | 0 | 0 | 1 (7.7) | 0 PD 1 (2.5) | 4 (10.5) | 1 (1.8) | 0 | 1 (3.2) NE 0 | 1 (2.6) | 0 | 0 | 0 Figure 3. Progression-free survival Non-squamous NSCLC: Tivunatamig + chemo (n=41) events 23, median PFS 13.8 (8.0-15.0); Tislelizumab + chemo (n=40) events 22, median 9.7 (6.9-NE); curve annotations: 55% (37-69) [tivunatamig], 45% (28-60) [tislelizumab] Squamous NSCLC: Tivunatamig + chemo (paclitaxel) (n=62) events 31, median 13.6 (7.8-NE); Tivunatamig + chemo (nab-paclitaxel) (n=14) events 7, median 13.4 (4.1-NE); Tislelizumab + chemo (n=31) events 21, median 8.3 (4.4-13.7); curve annotations: Paclitaxel: 50% (36-63); Nab-paclitaxel: 54% (25-76); 34% (17-52) [tislelizumab] Safety results • Grade ≥3 TRAEs occurred in 55.8% of patients across all doses of tivunatamig + chemo vs 49.3% for tislelizumab + chemo (Table 4) • At the tivunatamig 10 mg/kg dose level, grade ≥3 TRAEs occurred in 59.0% of patients; safety was consistent across non-squamous and squamous NSCLC – The most common TRAEs were anemia (55%), decreased white blood cell count (43%), and decreased neutrophil count (40%) (Figure 4) – Any-grade VEGF-related adverse events (AEs) occurred in 56.4% of patients; grade ≥3 AEs occurred in 12.0% (Table 5) • Occurrence of grade ≥3 hemorrhage and hemoptysis was low and similar across non-squamous and squamous histologies – Immune-related AEs (irAEs) occurred in 16.2% of patients; grade ≥3 irAEs occurred in 6.8% (Table 5) Table 4. Safety summary (Tivunatamig (all doses) + chemo (n=190) | Tivunatamig 10 mg/kg + chemo (n=117) | Tislelizumab + chemo (n=71)) Any grade 183 (96.3) | 116 (99.1) | 65 (91.5) Grade ≥3 106 (55.8) | 69 (59.0) | 35 (49.3) Grade 5 3 (1.6) | 3 (2.6) | 3 (4.2) SAE 57 (30.0) | 38 (32.5) | 28 (39.4) Leading to treatment delay 85 (44.7) | 58 (49.6) | 33 (46.5) Leading to discontinuation 8 (4.2) | 7 (6.0) | 7 (9.9) Circulating tumor DNA results • In patients who received any dose of tivunatamig with detectable ctDNA at baseline and samples collected at C3D1 (n=126), 61.9% had ctDNA clearance at C3D1 (Figure 5) • Median PFS was 16.4 months (95% CI, 10.6-not reached) for patients with ctDNA clearance at C3D1 vs 9.6 months (95% CI, 5.7-13.8) in those who did not clear ctDNA at C3D1 This study was originally sponsored by 3SBio; tivunatamig (PF-08634404 / SSGJ-707) was licensed to Pfizer in July 2025. Presented at the ESMO Congress 2026 | October 23-27, 2026 | Madrid, Spain Correspondence: Lin Wu

RC148 (ABBV-1480) monotherapy or in combination with docetaxel in first- or second-line advanced non-small-cell lung cancer (NSCLC) without actionable genomic alterations (AGA): updated results from a Phase 1/2 study Jinming Yu, Yuping Sun, Qing Wen, Meili Sun, Bihui Li, Yun Fan, Xia Song, Zhiyong He, Qichuan Wang, Yan Yu, Wei Zheng, Xiaoyan Li, Siyuan Huang, Runxiang Yang, Juan Li, Yongchang Zhang, Xin Lu, Guoguang Ma, Michael M. Shi, Jianmin Fang Background • RC148 (ABBV-1480) is a novel investigational bispecific antibody with a differentiated molecular design for dual targeting of PD-1 and VEGF. • RC148 has demonstrated encouraging antitumor activity and manageable safety as first-line (1L) monotherapy in patients with PD-L1-positive locally advanced or metastatic NSCLC, and in combination with docetaxel in patients without AGA who are previously treated with a PD-(L)1 inhibitor and platinum-based chemotherapy based on previous analysis of a Phase 1/2 study (NCT06016062). • We report here the updated efficacy and safety results with extended follow-up. The data cutoff date for this analysis was August 22, 2026. Methods Figure 1. Study design Key Eligibility Criteria: • Locally advanced or metastatic NSCLC • Measurable target lesion as per RECIST v1.1 • ECOG performance status of 0-1 1L NSCLC • PD-L1 TPS ≥1%: N=22 RC148 (20 mg/kg) monotherapy Previously treated NSCLC • Without AGA • Prior treatment with PD-(L)1 inhibitor and platinum-based chemotherapy: R 1:1 N=21 RC148 (10 mg/kg) + Docetaxel (75 mg/m²); N=21 RC148 (20 mg/kg) + Docetaxel (75 mg/m²) • All study drugs were administered IV Q3W. Treatment was continued until disease progression, intolerable toxicity, consent withdrawal, or initiation of new anti-cancer treatment, or for up to 24 months (mo), whichever occurred first. • Tumors were assessed as per RECIST v1.1 at baseline and thereafter every 6 weeks until disease progression, consent withdrawal, or initiation of new anti-cancer treatment, whichever occurred first. Primary endpoint • ORR assessed by investigators as per RECIST v1.1, and safety Efficacy Results – 1L NSCLC Patients • 22 patients were enrolled (median age: 65.0 years; sqNSCLC: 54.5%; clinical stage IV: 63.6%). The median follow-up duration was 15.61 mo (IQR: 9.20-17.35); 6 (27.3%) patients remained on treatment. Antitumor activity • The confirmed ORR (cORR) was 57.1% (95% CI: 34.0-78.2) among the 21 patients with at least one post-baseline tumor assessment. Durable tumor response was observed in all patients (median DoR: 12.52 mo [95% CI: 4.14-NE]) and across subgroups. Figure 2. Spider plot of tumor change in 1L NSCLC sqNSCLC (N = 12) | nsqNSCLC (N = 9) | TPS 1%-49% (N = 13) | TPS ≥50% (N = 9) cORR (95% CI) 63.6 (30.8-89.1) | 44.4 (13.7-78.8) | 50.0 (21.1-78.9) | 66.7 (29.9-92.5) Median DoR (95% CI), mo 12.52 (4.14-NE) | NR | 12.52 (4.14-NE) | NR PFS • PFS data was immature, with 10 (45.5%) events reported. The median PFS was 15.21 mo in all patients, and the PFS benefit was consistently observed in the subgroups. Figure 3. PFS in 1L NSCLC Median PFS: 15.21 mo (95% CI: 6.87-NE) 15-mo rate 58.9% (95% CI: 34.3-77.0) No. at risk RC148 20 mg/kg 22 18 17 14 13 11 1 sqNSCLC (N = 12) | nsqNSCLC (N = 9) | TPS 1%-49% (N = 13) | TPS ≥50% (N = 9) Median PFS (95% CI), mo 12.62 (2.76-NE) | NR | 15.21 (6.87-NE) | NR 15-mo PFS rate (95% CI) 45.5 (16.7-70.7) | 87.5 (38.7-98.1) | 64.8 (31.0-85.2) | 50.0 (15.2-77.5) Efficacy Results – Previously Treated NSCLC Patients • 21 patients were randomized to each group (median age: 61.0 vs 66.0 years for the 10 vs 20 mg/kg groups; sqNSCLC: 52.4% vs 66.7%; stage IV: 95.2% vs 85.7%). The median follow-up was 16.33 mo (IQR: 8.61-17.68). Antitumor activity • In a checkpoint inhibitor (CPI)-experienced setting, the 20 mg/kg group exhibited higher cORR versus the 10 mg/kg group (66.7% [95% CI: 43.0-85.4] vs 33.3% [95% CI: 14.6-57.0]), with durable response observed (median DoR: 8.38 mo [95% CI: 5.65-12.68] vs not reached). Figure 4. Spider plot of tumor change in previously treated NSCLC RC148 (20 mg/kg) + Docetaxel: sqNSCLC (N = 14) | nsqNSCLC (N = 7) | TPS <1% (N = 11) | TPS ≥1% (N = 10) cORR (95% CI) 64.3 (35.1-87.2) | 71.4 (29.0-96.3) | 54.5 (23.4-83.3) | 80.0 (44.4-97.5) Median DoR (95% CI), mo 8.38 (4.07-12.68) | NR | 6.98 (4.07-NE) | 12.68 (5.65-NE) PFS • In a CPI-experienced setting, the 20 mg/kg group achieved a longer median PFS than the 10 mg/kg group (8.48 vs 5.82 mo), with promising benefit observed across subgroups. Figure 5. PFS in previously treated NSCLC RC148 (10 mg/kg) +Docetaxel (N = 21) Median PFS (95% CI), mo 5.82 (4.14-11.37); RC148 (20 mg/kg) +Docetaxel (N = 21) 8.48 (5.98-12.45) No. at risk 10 mg/kg group 21 19 10 7 6 6 2; 20 mg/kg group 21 17 14 9 6 3 RC148 (20 mg/kg) + Docetaxel: sqNSCLC (N = 14) | nsqNSCLC (N = 7) | TPS <1% (N = 11) | TPS ≥1% (N = 10) Median PFS (95% CI), mo 7.16 (2.99-12.45) | 10.35 (8.28-NE) | 7.16 (2.86-11.14) | 11.88 (7.13-NE) Safety Results – Overall Safety • The safety profiles for both RC148 monotherapy and RC148 in combination with docetaxel were consistent with the previous report. • RC148 in combination with docetaxel did not increase the overall safety risk as compared with docetaxel monotherapy. • The most common (incidence ≥10.0%) grade ≥3 treatment-related adverse events (TRAEs) for monotherapy were hypertension and proteinuria; for combination therapy were neutrophil count decreased and white blood cell count decreased. • Most VEGF- and immune-related TRAEs were grade 1-2 and were consistent with previous report. Table 1. Safety summary, n (%): 1L NSCLC RC148 (20 mg/kg) Monotherapy (N = 22) | Previously treated NSCLC RC148 (10 mg/kg) + Docetaxel (N = 21) | RC148 (20 mg/kg) + Docetaxel (N = 21) Any TRAE* 22 (100) | 21 (100) | 21 (100) ≥ Grade 3 14 (63.6) | 18 (85.7) | 16 (76.2) Leading to treatment discontinuation 1 (4.5) | 2 (9.5) | 4 (19.0) Leading to death 0 | 0 | 0 Any VEGF-related TRAEs 16 (72.7) | 9 (42.9) | 13 (61.9) ≥ Grade 3 9 (40.9) | 1 (4.8) | 1 (4.8) Any immune-related TRAEs 17 (77.3) | 2 (9.5) | 9 (42.9) ≥ Grade 3 6 (27.3) | 0 | 1 (4.8) *Related to any study drug Conclusion • RC148, alone and with docetaxel, showed durable efficacy and manageable safety in advanced NSCLC, supporting pivotal studies. • In a CPI-experienced setting, NSCLC 20 mg/kg seems to provide superior efficacy as compared to 10 mg/kg and needs further investigation, while 10 mg/kg is RP3D in 1L NSCLC. • RC148 is being evaluated in three Phase 3 studies in NSCLC, both in China and globally (RC148-C301: NCT07416474; RC148-C304: NCT07829549; AMUR-Lung-01: NCT07756645). Contact and Funding • Presenting author: Dr. Jinming Yu • Funding: This study was sponsored by RemeGen Co., Ltd., Yantai, China. Abstract FPN: 2669P. Presented at the European Society For Medical Oncology (ESMO) Congress, October 23–27, 2026, Madrid, Spain.

[Slide 1] ABBV-706: A First-in-Class SEZ6-Targeted Topoisomerase 1 Inhibitor Antibody-Drug Conjugate With Robust Antitumor Activity in Relapsed/Refractory Small Cell Lung Cancer Anne C. Chiang,1 Nadine Jahchan,2 Emily Faivre,2 Irene Lee,2 Lujia Wang,2 Klea Richardson,2 Axel Hernandez,2 Michael Barnes,2 Sheila Bheddah,2 Weilong Zhao,2 Lloyd T. Lam,2 Song Wang,2 Tammy Palenski,2 Daniel Martins,2 Noura Choudhury,3 Alissa J. Cooper,4 Afshin Dowlati,5 Lauren Byers,6 Luis Paz-Ares7 1Yale School of Medicine, New Haven, CT, USA; 2AbbVie Inc., North Chicago, IL, USA; 3Department of Medicine-Hematology and Oncology, University of Chicago, Chicago, IL, USA; 4Department of Thoracic Oncology, Memorial Sloan Kettering Cancer Center, New York, NY, USA; 5University Hospitals Seidman Cancer Center and Case Western Reserve University, Cleveland, OH, USA; 6University of Texas MD Anderson Cancer Center, Houston, TX, USA; 7Hospital Universitario 12 de Octubre, Madrid, Spain Presented at the European Society for Medical Oncology (ESMO), 23-27 October 2026, Madrid, Spain [DV-020855] [Slide 2] ABBV-706 is a first-in-class SEZ6-targeting ADC being evaluated in ES-SCLC ABBV-706 is a first-in-class SEZ6-targeting ADC - Relapsed SCLC is a highly aggressive disease with poor outcomes and limited effective treatment options following first-line platinum-based chemoimmunotherapy1,2 - SEZ6, is a neuroendocrine lineage marker overexpressed in SCLC with limited expression in normal tissues,3-5 and still expressed across molecular subtypes6 - ABBV-706 is a first-in-class ADC consisting of a SEZ6-targeting monoclonal antibody conjugated via a stable linker to a potent Top1i payload (adizutecan)7 EXTRACELLULAR SPACE Potent topoisomerase 1i payload (Adizutecan) Linker (Joins antibody and cytotoxin) ABBV-706 SEZ6 Antibody SEZ6 Protein TUMOR CELL DNA Damage Here, we characterize SEZ6 as a clinical biomarker for SCLC in the ongoing FIH phase 1 study in patients with R/R SCLC treated with ABBV-706 at the RP2D of 1.8 mg/kg Q3W [Slide 3] ABBV-706 FIH phase 1 study design and published results Phase 1 R/R ES-SCLCa - Prior platinum-etoposide chemotherapy and/or immuno-oncology regimens - Prior treated or untreated stable brain metastases were eligible - History of ILD/pneumonitis requiring treatment or active disease were excluded Total monotherapy patient population: ABBV-706 1.3-3.5 mg/kg (N=124) 1.8 mg/kg RP2D for monotherapy ABBV-706 1.8 mg/kg as 2L+ (n=41) IV Q3W in 21-day cycles until disease progression or unacceptable toxicity Key Endpoints Safety, ORR, PFS, OS, SEZ6 by IHC Key Methods1 - Archival or fresh tissue biopsy at screening - SEZ6 IHC (AbbVie proprietary assay) - Plasma ctDNA and serum proGRP were collected at baseline, start of each cycle 1-4, odd cycles thereafter, and end of treatment - The Guardant INFINITY ctDNA assay (Guardant Health) was used to monitor methylation-based tumor fraction responses - proGRP was used to monitor tumor response using the Roche Cobas electrochemiluminescence immunoassay (LabCorp) Patient Characteristics 1.8 mg/kg (n=41)1 Median age, years (range) 64 (48-85) Median prior LoT, n (range) 2 (1-6) Brain mets at baseline 39% (n=16) Chemo-free interval 61% (<90 days) 27% (<30 days) Results 1.8 mg/kg (n=41)1 Safety Grade ≥3 All TRAEs 88% (n=36) Anemia 42% (n=17) Neutropenia 20% (n=8) Efficacyb ORR 56% (n=23) mPFS 6.4 months (4.0-8.1) mOS 12.4 months (8.2-17.3) 1. Byers LA, et al. Nat Med. 2026;32(8):2898-2908. aClinicalTrials.gov. NCT05599984. bSimilar efficacy was observed in the ITT and biomarker eligible populations. [Slide 4] Conclusions SEZ6 is highly expressed in SCLC, stable across LoT, and not correlated with response rates or hematological adverse events following ABBV-706 monotherapy The immunomodulatory effects induced by the adizutecan payload support the mechanistic rationale for combining ABBV-706 with PD-L1 in the ongoing phase 2 1L SCLC SEZanne studya ctDNA clearance is associated with improvements in PFS and OS following ABBV-706 monotherapy ABBV-706 monotherapy has demonstrated promising clinical benefit, with biomarker and preclinical data supporting SEZ6 as a therapeutic target in SCLC; the phase 3 SEZanne-02 studyb in patients with R/R SCLC previously treated with tarlatamab is ongoing AbbVie Inc. and the authors thank the participants, study sites, and investigators who participated in this clinical trial. Medical writing support was provided by Atreju Lackey, PhD, and Andrey Verendeev, PhD, of Avalere Health Global Ltd, which was funded by AbbVie. QR code expiration: September 27, 2027 aClinicalTrials.gov. NCT07155174. bClinicalTrials.gov. NCT07365241.

2635P Telisotuzumab Adizutecan (Temab-A) for EGFR Mutated Non-Squamous Non-Small Cell Lung Cancer: Post Hoc Analysis From a Phase 1 First-in-Human Study in Patients With Brain Metastases Judith Raimbourg, D Ross Camidge, Tae Min Kim, Yun-Gyoo Lee, Jonathan Goldman, Takako Eguchi Nakajima, Byoung Chul Cho, Ki Hyeong Lee, Iván Victoria, Enriqueta Felip, Zoë Hunter, Kristiyana Kaneva, Rui Li, Adam Luo, Tony Navas, Dae Ho Lee OBJECTIVE To assess the safety and efficacy of telisotuzumab adizutecan (ABBV-400, Temab-A) from a Phase 1 first-in-human study in patients with epidermal growth factor receptor mutated (EGFRmut) non-squamous non-small cell lung cancer (NSCLC) and brain metastases (BMs) CONCLUSIONS Temab-A induced deep and durable intracranial responses in patients with EGFRmut non-squamous NSCLC and brain metastases, with an intracranial ORR of 37% and median intracranial PFS of 8.2 months Systemic responses in patients with brain metastases are encouraging, with median PFS and OS of 13.6 and 19.5 months, respectively The safety profile was consistent with the broader Temab-A profile observed in non-squamous NSCLC as well as other tumor types Study findings support further evaluation of Temab-A in patients with metastatic disease and intracranial involvement BACKGROUND • Increased c-Met protein (also known as MET protein) expression is common in solid tumors, including NSCLC, and is linked to poor clinical outcomes • Telisotuzumab adizutecan (ABBV-400; Temab-A) is an antibody-drug conjugate comprising the c-Met–targeting antibody telisotuzumab conjugated to a novel potent topoisomerase 1 inhibitor payload, adizutecan (Figure 1) • In an ongoing Phase 1 study (NCT05029882), Temab-A monotherapy showed manageable safety and promising efficacy with an objective response rate of 63%, a median progression-free survival of 10.9 months, and a median overall survival of 16.2 months in patients with advanced non-squamous NSCLC and EGFRmut tumors • Herein, we present a post hoc, exploratory subgroup analysis of patients with stable BMs and EGFRmut NSCLC from this study METHODS • Three cohorts were pooled to identify 51 patients with EGFRmut NSCLC who had both baseline and on-treatment imaging • Of these, an Independent Review Committee identified 19 patients (37%) with BMs at baseline • Patients with a history of BMs were eligible to enroll provided they showed clinical and radiographic stability for at least 14 days after prior radiotherapy Study Design (NCT05029882) RESULTS • Among the 19 patients with BMs, median follow-up was 28.2 months • Baseline characteristics in the patients with BMs were generally consistent with the study population, with some imbalances in sex, race, prior osimertinib use, and MET amplification Baseline Clinical Characteristics in Patients With and Without Brain Metastases (Baseline BMs n=19 | No Baseline BMs n=32 | Total N=51) Median age (range), years | 60 (45–79) | 63 (43–81) | 63 (43–81) Median number of prior lines of therapy, n (range) | 3.0 (1–6) | 3.0 (1–8) | 3.0 (1–8) Type of prior TKI use, Osimertinib | 10 (53) | 26 (81) | 36 (71) Prior RT to brain, Yes | 13 (68) | 3 (9) | 16 (31) MET amplification, Amplified | 9 (47) | 7 (22) | 16 (31) Efficacy Outcomes in Patients With Brain Metastases Per Investigator (RECIST v1.1) and IRC (CNS-RECIST) Assessment Baseline BMs (n=19): Investigator-assessed systemic outcomes Prior RT (n=13) | Overall (n=19); IRC-assessed intracranial outcomes Prior RT (n=13) | Overall (n=19) ORR, n (%) [95% CI] | 10 (77) [46–95] | 15 (79) [54–94] | 3 (23) [5–54] | 7 (37) [16–62] Confirmed BOR, n (%): CR | 0 | 0 | 2 (15) | 6 (32) PR | 10 (77) | 15 (79) | 1 (8) | 1 (5) SD | 3 (23) | 4 (21) | 4 (31) | 4 (21) PD | 0 | 0 | 0 | 0 Non-CR/Non-PD | N/A | N/A | 6 (46) | 8 (42) CBR12, n (%) [95% CI] | 11 (85) [55–98] | 17 (89) [67–99] | 9 (69) [39–91] | 15 (79) [54–94] Median DOR [95% CI], months | 13.7 [2.2–14.3] | 13.5 [4.2–14.3] | 19.8 [2.2–NE] | 19.8 [2.2–NE] Median PFS, months [95% CI] | 14.1 [3.5–15.4] | 13.6 [5.4–15.4] | 7.7 [3.9–15.6] | 8.2 [4.0–21.0] Data cutoff date: 02 June 2026 Intracranial Progression-Free Survival in Patients With Brain Metastases Per IRC (CNS RECIST) Assessment: Baseline BMs (n=19) Events, n 13; Median PFS, months (95% CI) 8.2 (4.0–21.0) Best Percent Change in Sum of Target Lesion Measurements From Baseline in Patients With Brain Metastases Per IRC CNS-RECIST Assessment • Although small sample size limits interpretation, among the 3 patients with CR in target lesions per IRC CNS-RECIST assessment, 2 had MET amplification and all 3 were 3+, ≥25% c-Met expression by IHC (supplemental information accessible via the QR code) Percent Change in Target Lesion Measurement From Baseline Over Time in Patients With Brain Metastases Per IRC Overall Survival in Patients With Brain Metastases: Baseline BMs (n=19) Death, n 12; Median OS, months (95% CI) 19.5 (7.7–NE) Summary of Safety and Duration of Treatment (Baseline BMs n=19 | No Baseline BMs n=32 | Total N=51) Median duration of treatment (range), months | 13.8 (2.3–26.9) | 10.8 (1.4–38.6) | 11.2 (1.4–38.6) TRAEs, n (%) | 19 (100) | 32 (100) | 51 (100) Grade ≥3 TRAEs, n (%) | 14 (74) | 20 (63) | 34 (67) Serious TRAEs, n (%) | 4 (21) | 6 (19) | 10 (20) TRAEs leading to discontinuation, n (%) | 1 (5) | 5 (16) | 6 (12) TRAEs leading to dose interruption, n (%) | 10 (53) | 21 (66) | 31 (61) TRAEs leading to dose reduction, n (%) | 8 (42) | 18 (56) | 26 (51) TRAEs leading to death, n (%) | 0 | 1 (3) | 1 (2) Data cutoff date: 02 June 2026. • The safety profile in patients with baseline BMs was generally consistent with that of the overall population Presented at the European Society for Medical Oncology (ESMO) Annual Meeting, October 23–27, 2026, Madrid (Spain).

[Slide 1] ISM6331 Demonstrates Antitumor Activity Across Dose Cohorts 130 mg COHORT ORR: 40.0% — 2/5 Patients (Deepest Shrinkage) 30 & 50 mg COHORT ORR: 25.0% — 2/8 Patients (Early Signal) OVERALL MESOTHELIOMA ORR: 18.2% — 4/22 Patients (Dose Range 30 –130 mg) RAPID RESPONSE ONSET: 75.0% — 3/4 Responders at Week 8 Heavily Pretreated Mesothelioma (n=22) • Responses across dose cohorts: PR observed at 30 mg (1/2), 50 mg (1/1), and 130 mg (2/5) • Durable disease control: 68.2% DCR (at 1st tumor scan), 3/4 PRs ongoing, including 1 PR lasting ~ 1 year • Rapid Response in heavily pretreated patients: 75% of PRs observed at 1st scan (Week 8); PRs seen in patients with 5L and 6L prior lines [Waterfall and spider plots by dose: 15 mg, 30 mg, 50 mg, 75 mg, 100 mg, 130 mg; y-axis %, x-axis Days 0–330] # 28.3% target lesion reduction, NTL1 (abdominal peritoneal nodule) disappearance *33% target lesion reduction, NTL1 (mid right omentum) disappearance, new lesions at Week 16 Other Refractory Solid Tumors (n=8) • No objective responses by RECIST • Target lesion shrinkage: Observed in NSCLC & HNSCC (7L & 8L prior lines respectively) • Extended Disease Control: SD (≥ 4 months) observed in EHE (15 mg) and CRC (30 mg) [Waterfall: CRC PD, HNSCC PD, NSCLC PD, HNSCC PD, EHE SD, CRC SD, NSCLC PD, HNSCC SD; spider plot 15/30/50 mg] CRC: colorectal cancer, NSCLC: non-small cell lung cancer, EHE: Epithelioid hemangioendothelioma HNSCC: Head & Neck squamous cell carcinoma, PR: partial response, SD: stable disease, LOTs: lines of treatment [Slide 2] Case Series 63 y/o Asian Female with Pleural Mesothelioma | 50mg Patient Demographics & Baseline Status Age / Sex: 63y / Female Primary Disease: Pleural Mesothelioma Histology / ECOG: Epithelioid / 1 Metastases Sites: Lymph Node, Chest wall Cohort 3: 50 mg Single Dose: 06Aug2025 Multiple Dose: 09Aug2025 -46.8% Best % Change (Sum of Diameters) Baseline: 125.3 mm Confirmed PR (Week 48, Ongoing) Path to Current Trial (Clinical History) 1L: PEM + Bev + CDDP (PD) -> 2L: RC88 (PD) -> 3L: NIV + IPI (PD) -> 4L: PEM + Bev + CDDP (PD) -> 5L: KY0301 (UNK) -> Current Trial (Cohort 3) PEM: pemetrexed; Bev: bevacizumab; CDDP: cisplatin; NIV: nivolumab; IPI: ipilimumab; PD: progressive disease; RC88: MSLN targeted ADC; KY0301: MET-EGFR bispecific ADC No interventions included photodynamic surgery and particle implantation. Safety & Biomarker Profile - Safety: Gr. 1 lymphocyte count decreased (No Grade 3+ TRAE or dose modification) - PD biomarker: -32% max inhibition of Blood CCN2 Baseline: 51.9mm | Week 24: 28.0 mm 65y/o White Male with Peritoneal Mesothelioma | 130mg Patient Demographics & Baseline Status Age / Sex: 65 y / Male Primary Disease: Peritoneal Mesothelioma Histology / ECOG: Epithelioid / 0 Metastases Sites: Omentum Cohort 6: 130 mg Single Dose: 16Jun2026 Multiple Dose: 18Jun2026 -57.6% Best % Change (Sum of Diameters) Baseline: 33 mm Initial PR (Week 8, Ongoing) Path to Current Trial (Clinical History) 1L: PEM + CDDP (PD) -> 2L: Atezo + Bev (PD) -> Current Trial (Cohort 6) PEM: pemetrexed; Bev: bevacizumab; CDDP: cisplatin; Atezo: Atezolizumab; PD: progressive disease; The patient previously underwent laparoscopy of the peritoneum. Safety & Biomarker Profile - Safety: No Grade 3+ TRAE or dose modification - PD biomarker: Blood CCN2: Not available - Molecular: BAP1 loss-of-function, NF2 alteration was not detected Baseline: 33.0 mm | Week 8: 14.0 mm

[Slide 1] Leptomeningeal Metastasis (LM) - update BLOOM study with Osimertinib, 41 patients with cytologically confirmed LM received 160 mg daily: • LM objective response rate by blinded central review: 62%. • Median PFS: 8.6 months. • Median OS: 11.0 months. • Confirmed CSF tumour-cell clearance: 28%. TBCRC049 — Phase II: Tucatinib + trastuzumab + capecitabine; 17 women with newly diagnosed HER2-positive LM. • Median OS 10.0 months • Time to CNS progression 6.9 months • Neurological improvement (7/12 evaluable) 58% • Composite LM response rate was 38% among 13 evaluable patients. • Therapeutic tucatinib concentrations were demonstrated in CSF. PMID: 31809241 PMID: 41851506 [Slide 3] Solid Metastasis – NSCLC: molecularly directed CNS therapy • EGFR-mutant: osimertinib-based therapy; combination strategies have relevant CNS efficacy. • ALK-positive: lorlatinib and other CNS-active ALK inhibitors can achieve durable intracranial control. • RET, ROS1, METex14, NTRK: evaluate CNS-specific evidence rather than systemic response alone. • For asymptomatic, non-threatening lesions, systemic-first therapy with close MRI monitoring may be appropriate. [Slide 4] Solid Metastasis – melanoma immunotherapy 51% Intracranial response — ipilimumab/nivolumab (ABC trial) 42% 7-year intracranial PFS 48% 7-year overall survival • Combination immunotherapy can produce durable intracranial disease control. • Evidence is strongest for asymptomatic, immunotherapy-naïve patients. • Cave: Corticosteroid requirement and neurological symptoms • -> Balance immunotherapy benefit against the need for urgent local therapy and steroid exposure. PMID: 39978875

[Slide 2] 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic CHAIRS: PASI A. JÄNNE, XIUNING LE, EGBERT SMIT FLAURA2 study design1 Scan QR code to view slides and supplementary material [QR code] Patients with EGFR-mutated advanced NSCLC (N=557) Key inclusion criteria: • Aged ≥18 years (Japan: ≥20 years) • Pathologically confirmed non-squamous NSCLC • Ex19del / L858R (local / central test) • WHO PS 0 / 1 • No prior systemic therapy for advanced NSCLC • Stable CNS metastases were allowed • Brain scan at baseline (MRI / CT; mandatory) R 1:1* Osimertinib 80 mg (QD) + pemetrexed 500 mg/m2 + carboplatin AUC5 or cisplatin 75 mg/m2 (Q3W for 4 cycles of plat-based Tx) -> Maintenance osimertinib 80 mg (QD) + pemetrexed 500 mg/m2 (Q3W)† Osi 80 mg (QD)† • Primary endpoint: PFS (by investigator) per RECIST 1.1 • Key secondary endpoint: OS • Data cut-off: 03 April 2023 (PFS) and 12 June 2025 (OS) Exploratory endpoint: Correlation of plasma EGFRm clearance at Weeks 1, 2, 3, and 6 with PFS and OS to understand if monitoring ctDNA dynamics at earlier timepoints can be used to predict outcomes Plasma sample collection: Baseline, Week 1, Week 2, Week 3, Week 6 -> Baseline and Weeks 1, 2, 3, and 6 plasma samples‡ from both treatment arms analysed by ddPCR (Biodesix) FLAURA2 (NCT04035486) is a global, open-label, Phase III study. We acknowledge the potential presence of immortal time bias of analysing PFS and OS (measured from randomisation) by post-baseline subgroup as patients would need to remain progression-free or alive until Week 1, 2, 3 and 6 to be assessed in that subgroup. 1. Planchard D, et al. N Engl J Med 2023;389:1935-1948. [Remaining footnotes and abbreviations in small print - partially legible: stratification by race (Asian Chinese / Asian non-Chinese / non-Asian), EGFR test (local / central) and WHO PS (0 / 1); plasma sample schedule details] David Planchard, MD, PhD ALMERIA AUDITORIUM - HALL 9 [Slide 4] 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic CHAIRS: PASI A. JÄNNE, XIUNING LE, EGBERT SMIT PFS improved with osi + plat-pem versus osi mono irrespective of Week 3 and 6 plasma EGFRm clearance status [4 KM panels; Y: Probability of PFS; X: Time from randomisation (months) 0-36] Plasma EGFRm clearance at Week 3 - No. at risk (0,6,12,18,24,30,36): osi + plat-pem 88 79 69 53 26 12 0; osi mono 100 90 65 37 19 10 0 No Plasma EGFRm clearance at Week 3 - No. at risk: osi + plat-pem 21 16 12 9 4 2 0; osi mono 40 22 18 10 6 3 0 Week 3 table: mPFS (95% CI), months | Clearance | No clearance Osi + plat-pem | 27.6 (21.0, NR) | 19.6 (10.1, NR) Osi mono | 15.2 (13.6, 19.3) | 11.1 (5.6, 16.5) HR (95% CI) | 0.51 (0.35, 0.75) | 0.62 (0.33, 1.14) Plasma EGFRm clearance at Week 6 - No. at risk: osi + plat-pem 92 83 70 53 28 14 0; osi mono 122 110 83 48 26 12 0 No Plasma EGFRm clearance at Week 6 - No. at risk: osi + plat-pem 13 10 6 5 1 0 0; osi mono 15 3 3 1 0 0 0 Week 6 table: mPFS (95% CI), months | Clearance | No clearance Osi + plat-pem | 27.6 (22.0, NR) | 11.8 (5.1, 22.1) Osi mono | 16.4 (13.8, 19.4) | 5.4 (1.6, 5.7) HR (95% CI) | 0.54 (0.38, 0.77) | 0.39 (0.16, 0.94) Data cut-off: 03 April 2023 (PFS). HRs were calculated using an unstratified Cox proportional hazards model. Clearance was defined as undetected plasma EGFRm (Ex19del, L858R) at Week 1, 2, 3, or 6, where it was detected at baseline. CI, confidence interval; EGFR, epidermal growth factor receptor; EGFRm, EGFR mutation; Ex19del, Exon 19 deletion; HR, hazard ratio; m, median; mono, monotherapy; NR, not reached; osi, osimertinib; pem, pemetrexed; PFS, progression-free survival; plat, platinum. David Planchard, MD, PhD ALMERIA AUDITORIUM - HALL 9 Survival table Endpoint: PFS (investigator) by Week 3 plasma EGFRm clearance - clearance subgroup (other subgroups in notes) Arm | N | Median | Median CI Osi + plat-pem (Week 3 clearance) | 88 | 27.6 | 21.0, NR Osi mono (Week 3 clearance) | 100 | 15.2 | 13.6, 19.3 HR: 0.51 HR CI: 0.35, 0.75

[Slide 1] Tobacco, women and lung cancer: a delayed epidemic and a screening gap Tobacco remains the leading preventable risk factor for lung cancer.1 Globally, rates of smoking continue to increase among women. The pattern is shifting, in 2020, female incidence surpassed males in the USA Screening gap3. NLST enrolled 59% men and NELSON about 16% women4,5 Screening criteria may miss women at risk based on age and pack-year thresholds while screening effectiveness appears less certain for SCLC. Women may be more susceptible to tobacco. Differences in carcinogen-metabolizing pathways, including CYP1A1-mediated activation and GSTM1-mediated detoxification, higher levels of tobacco-related DNA adducts, and possibly less efficient DNA repair3. [Right panel] Trends in tobacco consumption and lung cancer mortality by sex in the United States, [line chart; small labels illegible] Life Magazine 1939, Sponsor Philip Morris & Co [historical photograph] Sex trends and outcomes in SCLC in Spain2 [small chart; text illegible] 1 American Cancer Society, The Cancer Atlas, 2025 · 2 Garrido P, ASCO 2026 · 3MolenaD, Garrido P, JTOCR 2026; 4de Koning HJ, NEJM 2020 · 5 NLST Research Team, NEJM 2011 Prof. Pilar Garrido [Slide 2] Risk factors in lung cancer II Lung cancer in never smokers: an important and distinct clinical entity Never smokers • Two-thirds of cases occur in women, most commonly in East Asian populations • Never smokers: >55% of Asian and >30% of Hispanic women with NSCLC in the USA • Younger patients are more likely to be never smokers • 8th leading cause of cancer death in the USA, 5th worldwide Genetic risk • Familial lung cancer: germline EGFR T790M, HER2, BRCA2, CHEK2, MET and YAP1 mutations • More common in women Environmental risk • Indoor air pollution (solid fuels, high-temperature frying): higher risk in women • Second-hand smoke: higher risk after living ≥30 years with a smoker • Air pollution: women potentially more susceptible1 Estimated number of global deaths attributed to 37 cancer types in 2020 [horizontal bar chart; labels illegible] LoPiccolo J, Nat Rev Clin Oncol 2024 [Diagram: RISK FACTORS - ENVIRONMENTAL EXPOSURE, SECOND-HAND TOBACCO EXPOSURE, FAMILY HISTORY, AIR POLLUTION] 1Alhattab R, BMJ 2025 Prof. Pilar Garrido [Slide 3] Sex differences in lung cancer survival Women consistently show better survival than men Early stages • Surgery: lower risk of death in women (HR 0.73)1 • Absolute survival difference at 1, 5 and 10 years: 3.0%, 10% and 12%1 • Large registry (n = 33,790): advantage persists after adjustment (stage I-II adenocarcinoma, HR 0.74)2 Advanced stages • Male sex represents an independent, negative prognostic factor in NSCLC2 Is the magnitude of benefit with IO sex dependent? • Some data from meta-analyses of RCTs suggest a different benefit by sex3-4 • Men: large benefit with ICIs alone versus control, even in NSCLC with PD-L1 >50% • Women: large benefit with ICIs + chemotherapy • Another meta-analysis of 23 RCTs in advanced solid tumors does not find a sex difference in OS from IO5 OS KM curves stratified by sex from surgery1 [four small KM panels: Adenocarcinoma, Squamous cell cancer, Stage IA and IB, Stage IIA and IIB; numbers illegible] HR death according to patients' sex: 1L NSCLC4 [forest plot A/B; text illegible] Death HR according to patients' sex, pooled analysis3 [forest plot; text illegible] Differences in efficacy of immunotherapy in men and women by subgroups5 [Table 2. Differences in Efficacy of Immunotherapy in Men and Women by Subgroups; Overall row partially legible: 23 studies, Men 9322, Women 4399, 0.75 (0.69-0.81), 0.77 (0.67-0.88), I2 38, P .60; remaining rows too small to transcribe reliably] 1Sachs F, Chest 2021 · 2Radkiewicz C, PLoS One 2019 · 3Conforti F, Lancet Oncol 2018 · 4Conforti F, ESMO Open 2021 · 5Wallis CJD, JAMA Oncol 2019 Prof. Pilar Garrido

[Slide 1] PFS improved with osi + plat–pem in patients with baseline-detected plasma EGFRm versus osi mono¹ • Baseline-detected plasma EGFRm was prognostic and may select for a higher degree of clinical benefit with osi + plat–pem versus osi mono PFS in patients with baseline-detected plasma EGFRm mPFS (95% CI), months Osi + plat–pem (n=147): 24.8 (19.6, 27.9) Osi mono (n=161): 13.9 (13.6, 16.6) HR (95% CI): 0.60 (0.45, 0.80) No. at risk (0, 6, 12, 18, 24, 30, 36 months): Osi + plat–pem: 147 122 101 78 38 15 0 Osi mono: 161 128 96 56 31 13 0 PFS in patients with baseline non-detected plasma EGFRm mPFS (95% CI), months Osi + plat–pem (n=65): 33.3 (23.8, NC) Osi mono (n=48): 30.3 (25.0, NC) HR (95% CI): 0.93 (0.51, 1.72) No. at risk (0, 6, 12, 18, 24, 30, 36 months): Osi + plat–pem: 65 60 56 44 25 5 0 Osi mono: 48 41 34 30 21 7 0 Time from randomisation (months) HR was calculated by an unstratified log-rank test 1. Jänne PA, et al. Cancer Res 2024;84(7_Supplement): CT017 / AACR Clinical Trials Plenary Session. CI, confidence interval; EGFRm, epidermal growth factor receptor mutation; HR, hazard ratio; mono, monotherapy; NC, not calculable; osi, osimertinib; pem, pemetrexed; PFS, progression-free survival; plat, platinum David Planchard, MD, PhD Survival table Endpoint: PFS by baseline plasma EGFRm status Arm | N | Median | Median CI Baseline-detected: Osi + plat-pem | 147 | 24.8 mo | 19.6, 27.9 Baseline-detected: Osi mono | 161 | 13.9 mo | 13.6, 16.6 Baseline non-detected: Osi + plat-pem | 65 | 33.3 mo | 23.8, NC Baseline non-detected: Osi mono | 48 | 30.3 mo | 25.0, NC HR: 0.60 (detected); 0.93 (non-detected) HR CI: 0.45-0.80 (detected); 0.51-1.72 (non-detected) [Slide 2] Baseline-detected plasma EGFRm correlated with shorter PFS in the pooled ctDNA analysis set¹ • Pooled analysis across both arms showed patients with baseline-detected plasma EGFRm had shorter median PFS (18.3 months) compared with those with baseline non-detected plasma EGFRm (30.5 months) • This correlation is consistent with prior ctDNA analyses reported from the FLAURA and ELIOS studies²,³ mPFS (95% CI), months Detected plasma EGFRm (n=308*): 18.3 (15.4, 21.0) Non-detected plasma EGFRm (n=113): 30.5 (25.0, NC) HR (95% CI): 0.55 (0.42, 0.73) No. at risk (0, 6, 12, 18, 24, 30, 36 months): Detected: 308 250 197 134 69 28 0 Non-detected: 113 101 90 74 46 12 0 Time from randomisation (months) 1. Jänne PA, et al. Cancer Res 2024;84(7_Supplement): CT017 / AACR Clinical Trials Plenary Session; 2. Gray JE, et al. Clin Cancer Res 2023;29:3340–3351; 3. Ahn M-J, et al. Ann Oncol 2022;33:S1563–S1564 *147 and 161 patients in the osimertinib plus platinum-pemetrexed and osimertinib mono arms, respectively, had baseline-detected plasma EGFRm. HR was calculated by an unstratified log-rank test. CI, confidence interval; ctDNA, circulating tumour DNA; EGFRm, epidermal growth factor receptor mutation; HR, hazard ratio; NC, not calculable; mono, monotherapy; PFS, progression-free survival David Planchard, MD, PhD Survival table Endpoint: PFS (pooled across both arms) by baseline plasma EGFRm Arm | N | Median | Median CI Detected plasma EGFRm | 308 | 18.3 mo | 15.4, 21.0 Non-detected plasma EGFRm | 113 | 30.5 mo | 25.0, NC HR: 0.55 HR CI: 0.42-0.73

Conclusions • On-treatment plasma EGFRm clearance was prognostic, but not predictive of PFS or OS benefit with osimertinib + platinum–pemetrexed at all timepoints using a single-gene assay • Plasma EGFRm clearance rates were comparable across treatment arms at each timepoint • Studies using ultra-sensitive multi-marker ctDNA assays are planned to investigate if early ctDNA dynamics beyond plasma EGFRm can serve as a predictive biomarker [Banner] Efficacy of osimertinib + platinum–pemetrexed was superior to osimertinib monotherapy regardless of plasma EGFRm clearance status, reinforcing osimertinib + platinum–pemetrexed as a preferred first-line treatment option for EGFR-mutated advanced NSCLC

[Slide 1] MADRID 2026 ESMO congress Updated Results of Randomized Phase 3 Study of Selpercatinib vs Cabozantinib or Vandetanib (C/V) in Advanced, Kinase Inhibitor-Naïve, RET-mutant Medullary Thyroid Cancer (MTC) LIBRETTO-531 (NCT04211337) Jaume Capdevila, Marcia S. Brose, Bruce Robinson, Rossella Elisei, Eric Sherman, Ana O. Hoff, Mimi I. Hu, Makoto Tahara, Ming Gao, Barbara Jarzab, Dagmar Fuehrer-Sakel, Bhumsuk Keam, Fernanda Vaisman, Yu Wang, Jonathan Wadsley, Marcus Ngantcha, Patricia Maeda, Rob O'Shea, Lori Wirth, Julien Hadoux Jaume Capdevila, MD, PhD Vall d'Hebron University Hospital, Vall d'Hebron Institute of Oncology (VHIO), IOB Quiron-Teknon, Barcelona, Spain 26 October 2026 [Slide 2] LIBRETTO-531 Study Design Data cut off: 12 February 2026 Key Eligibility Criteria • Unresectable locally advanced or metastatic RET-mutant MTC • No prior treatment with kinase inhibitors • Documented RECIST 1.1 PD within 14 months (centrally reviewed) • RET mutation identified via NGS or PCR Stratification Factors • Mutation status (M918T vs other) • Physician's choice of treatment if randomized to control arm, with declared intent before randomization 2:1 Selpercatinib (160 mg BID) N=193 Physician's choice of Cabozantinib (140 mg QD) or Vandetanib (300 mg QD) N=98 Optional Crossover Selpercatinib (160 mg BID) N=43 Primary endpoint: PFS per RECIST 1.1 by BICR Secondary endpoints: • Efficacy (TFFS¹ by BICR and investigator, PFS by investigator, ORR by BICR and investigator, and OS) • Safety LIBRETTO-531 final update analysis Overall survival analysis was to be performed six years after first patient visit. Endpoints included OS, PFS by investigator, PFS2, and safety (including TEAEs, dose modifications, treatment discontinuations). LIBRETTO-531 trial (NCT04211337). ¹TFFS was tested conditionally on achieving statistical significance for PFS, adjusted for multiplicity and tested against a two-sided significance level of 0.05. BICR, blinded Independent Central Review; NGS, next-generation sequencing; PCR, polymerase chain reaction; PFS, progression-free survival; TFFS, treatment failure-free survival; ORR, overall response rate; OS, overall survival, PD, progressive disease. Jaume Capdevila, MD, PhD [Slide 3] LIBRETTO-531 Overall Survival Overall Survival Probability Time from randomization (months) Selpercatinib Cab or Van No. Patients No. Events Median Follow-up Months (95% CI) Selpercatinib 193 23 45.4 (43.3, 47.4) Cab or Van 98 28 47.0 (43.7, 50.7) HR (95% CI) 0.338 (0.194, 0.587) P-value <0.0001 87.1% 66.8% No. at Risk With a median follow-up of 45 months, overall survival favored selpercatinib over the cabozantinib or vandetanib group Hazard ratios and 95% confidence intervals were calculated with a Cox proportional-hazards model, with stratification according to trial group (selpercatinib vs. control). Two-sided P values were calculated with a log-rank test, with stratification according to trial group. Tick marks indicate censored data. Patients in the control group were permitted to cross over to selpercatinib following BICR-confirmed disease progression. Jaume Capdevila, MD, PhD Survival table Endpoint: OS Selpercatinib | 193 Cabozantinib or Vandetanib | 98 HR: 0.338 HR CI: 0.194, 0.587 P: <0.0001 48 months (dashed line): Selpercatinib 87.1% vs Cab or Van 66.8% (printed labels)

When should we order plasma genotyping for patients with advanced NSCLC? Diagnostic approach in newly diagnosed advanced NSCLC Sequential approach → Tissue NGS → If unsuccessful → Plasma NGS; If successful → Start treatment Complementary approach → Concurrent tissue and plasma NGS testing (tissue scant or time limited) → Start treatment Plasma-first approach → Plasma NGS (tissue not available) → If unsuccessful → Reflex tissue testing; If successful → Start treatment cfDNA: cell free DNA; NCCN Clinical Practice Guidelines in Oncology for Non-Small Cell Lung Cancer V.5.2026; *ESMO, ASCO recommend RNA where possible/appropriate; Figure from Ghazali & Leighl Canadian Oncology Today. 2024; Pascual J et al. Ann Oncol 2022; 33:750-768 ; 2. V.9.2025. Rolfo C... Leighl... Gandara J Thorac Oncol 2021; ASCO Living Guidelines for NSCLC with Driver Alterations V 2026 3.0 Puri, Leighl et al. J Clin Oncol 2026 Natasha B. Leighl

Plasma ctRNA For Gene Expression Profiling In NSCLC Detecting fusions ctRNA for detecting ALK fusions (BRIGHTSTAR)1 [oncoprint figure; row labels ctDNA / ctRNA; values 15/38 [illegible], 2/38 [illegible]] • ctRNA detected 19 ALK rearrangements vs 14 by ctDNA • ctDNA detection associated with worse PFS (HR=5.0, p=0.047), but not ctRNA (HR=3.1, p=0.179) ctRNA for detecting ALK, ROS1, RET fusions (Japanese study)2 [pie charts ctRNA vs ctDNA; Detected / Not detected] • Sensitivity at chemo-naive/PD: • ctRNA: 77.8% (7/9) • ctDNA: 33.3% (3/9) • Longitudinal ctRNA profiling detects progression and treatment response Detecting mechanisms of resistance to EGFR TKI3 [sigmoid plot: Spikes detected (%) vs H1975 RNA (%); Spike type: In silico / In vitro] RARE-seq (targeted cfRNA sequencing) has ~50-fold better analytical sensitivity than whole-coding-transcriptome sequencing (LOD95: 0.05% vs 2.8%) LOD95 = 0.05% Example: SCLC transformation [box plot Pre-treatment vs Post-resistance; timeline figure Time from diagnosis (days); ctRNA signature: LUAD ES, SCLC ctRNA ES; ctDNA variant: EGFR exon19del, EGFR(T790M)] 1. Heeke S, et al. JTO Clin Res Rep. 2025;6:100780; 2. Hasegawa N et al. Cancer Science. 2021;112:4393-4403; 3. Nesselbush MC et al. Nature. 2025;641(8063):759-768 Natasha B. Leighl

Detecting Molecular Tumour Type (TT) and Lung Cancer Subtype (ST) Through Cell-Free Methylomics Patient case [Guardant report excerpt: BIOMARKERS ASSOCIATED WITH FDA-APPROVED TREATMENT OPTIONS; MOLECULAR TUMOR TYPE - text largely illegible] [radial plot] Molecular Tumor Typing Primary Prediction: Lung, >99% Molecular Lung Subtype Lung adenocarcinoma: >99% Lung squamous cell carcinoma: <1% Small cell lung carcinoma: <1% [partly illegible] Zhang et al. 2026 study [radial plot; legend High Confidence / Low Confidence / [illegible]] • N=145 NSCLC patients underwent cfDNA methylation profiling • 47 (32%) had reported TT or ST prediction • Overall accuracy TT prediction: 100% • Accuracy of ST prediction: 94% (33/35) SIBYL study (NCT05935384) Lung-Subtype Caller vs G360 Test Requisition Forms [heatmap: Lung-subtype caller output (SCLC/LUSC/LUAD) vs As described in G360 TRF (LUAD/LUSC/SCLC)] SCLC row: 31.5% | 8.5% | 70.3% LUSC row: 21.4% | 85.2% | 2.7% LUAD row: 73.7% | 6.3% | 1.8% • N=7910 adenocarcinoma (LUAD) • N=1603 squamous (LUSC) • N=467 small cell (SCLC) • High concordance between histological diagnoses and MLSP was supported by RWE analyses in 9,980 cases Case courtesy of Dr. N. Leighl. https://www.guardantcomplete.com/assets/pdf/ONC-US-00552_G360_Liquid_CDx_Spec_Sheet.pdf (accessed 13 Oct 2026) Zhang J et al. ASCO 2026 (Abstract #8538); de Nanteyamo MGP et al. ESMO 2025 (Abstract #1924P) Natasha B. Leighl

Pre-diagnostic Plasma NGS In Patients With Suspected Advanced Lung Cancer Can ↓ Time To Treatment By ~50% Liquid and Tissue Biopsies taken at same time [icon: blood tube + microscope] Liquid Biopsy at presentation, before tissue biopsy (required before treatment--> delay) [icon] Time to Treatment Initiation (Days) [horizontal bar chart; legend: Plasma Actionable / Plasma / Tissue] Western Uni.: Plasma 12; Tissue 22 U. Penn: Plasma Actionable 10; Plasma 12; Tissue 20 Royal Marsden: Plasma Actionable 9; Tissue 25 Princess Margaret: Plasma 33; Tissue 62 Princess Margaret: Plasma Actionable 33; Plasma 39; Tissue 61 LIBELULE: Plasma Actionable 21; Plasma 29; Tissue 33 Holjiak E et al. J Liq Biopsy. 2025;10:100443; Thompson JC et al. JTO Clin Res Rep. 2022; Cui W et al. Lung Cancer. 2022; Garcia-Pardo M...Leighl. Ther Adv Med Oncol. 2022; Garcia-Pardo M...Leighl. JAMA Netw Open. 2023; Swalduz A et al. JTO 2025 Natasha B. Leighl

Pre-diagnostic cfDNA Testing Now Funded Standard Of Care In Patients In The UK With Suspected Advanced Lung Cancer At Presentation • N=1,296; 50% male; 18% never smokers • 24% emergency hospitalizations • 10% could not undergo tissue biopsy • 20% underwent ≥1 biopsy • 6% biopsy complication rate • 62% confirmed NSCLC; 11% SCLC; 12% other • TAT: 39d vs 22d from sample to genomics report (44%↓) • Cost effective: £11M savings to NHS • Now funded as standard of care in the UK • Wider implementation planned [Report cover image: A Health Economics Evaluation of ctDNA testing in NSCLC; QR code] Slide modified, original courtesy of Dr. Sanjay Popat https://norththamesgenomics.nhs.uk/wp-content/uploads/2026/01/Edge-Health-Health-economics-evaluation-of-ctDNA-testing-in-NSCLC-for-pub4.pdf Natasha B. Leighl

Plasma Epigenomics For Detecting SCLC Transformation cfDNA nucleosome and methylation signatures capture SCLC transformation in EGFRm NSCLC [Schematic: Noninvasive Detection of Transformed SCLC via Tissue-Informed Epigenomic cfDNA Analysis — Plasma Collection from Patients with Metastatic Lung Cancer → cfDNA Epigenomic Profiling (Histone Post-Translational Modifications; DNA Methylation; Chromatin Accessibility) → Apply Tumor-Derived Classifiers to Epigenomic cfDNA Data → ROC Curves; small labels [illegible]] Integrated Epigenomic cfDNA Classifier [Box plot: SCLC Risk Score; EGFRm LUAD (n = 12) vs EGFRm tSCLC (n = 9); P = 0.00036] [ROC curve: Sensitivity (%) vs 100 - Specificity (%)] Optimal cut-off = -0.029; Sensitivity = 89%; Specificity = 91%; Likelihood Ratio = 9.8; AUROC = 0.94, P = [illegible] Correlation of SCLC Risk Score with cfDNA Tumor Fraction [scatter: SCLC Risk Score vs cfDNA Tumor Fraction] EGFRm tSCLC R2 = 0.79, P = [illegible]; EGFRm LUAD R2 = 0.01, P = [illegible] Case 1: 73-year-old woman diagnosed with EGFR mutant (exon 19 del) LUAD metastatic to brain. [Timeline plot: SCLC Risk Score vs Time Since Diagnosis; labels: Lung biopsy at diagnosis: LUAD; Ongoing PR; Clinical progression; cfDNA: EGFR T790M; cfDNA TP#1; cfDNA TP#2; New liver mets; Liver biopsy: tSCLC; cfDNA TP#3; treatments Erlotinib, Osimertinib] Case 2: 62-year-old woman diagnosed with EGFR mutant (exon 19 del) LUAD metastatic to liver, bone, and brain. [Timeline plot: SCLC Risk Score vs Time Since Diagnosis; labels: Liver biopsy at diagnosis: LUAD; PD with small increase in LAD; cfDNA: no genomic resistance mechanism; cfDNA TP#1; Further PD in adrenal glands; cfDNA TP#2; New liver mets; Liver biopsy: tSCLC; cfDNA TP#3; treatment Osimertinib] El Zarif et al. Clin Cancer Res. 2024;30(17):3798-3811 Natasha B. Leighl

[Slide 1] Incorporating Plasma ctDNA Testing Into Treatment Decisions in Early-Stage NSCLC At Diagnosis | After Surgery | During Follow Up ctDNA testing ctDNA+ -> Neoadjuvant Therapy ctDNA- -> Surgery Post-surgery ctDNA testing ctDNA+ -> Adjuvant Therapy ctDNA- -> Observation Post-treatment ctDNA surveillance ctDNA+ -> Adjuvant Therapy 2 ctDNA- -> Observation Longitudinal ctDNA monitoring ctDNA- -> Observation ctDNA+ -> Drug Therapy Adapted from Soh J et al. Cancers (Basel). 2021;13(16):4035; Made with Biorender.com Natasha B. Leighl [Slide 2] Plasma monitoring integrated into routine surveillance (FFPE tumor tissue collected) Baseline q3m Month 0 | Month 3 | Month 6 | Month 9 | Month 12 EQ-5D-5L [questionnaire icon at each timepoint] Liquid biopsies [blood tube at each timepoint] CT CAP* [CT scanner at each timepoint] * +/- brain MRI if CNS involvement; **or until progression, whichever occurs earlier; CT CAP: CT chest abdomen pelvis *CONFIDENTIAL; NOT FOR DISTRIBUTION* Natasha B. Leighl