ESMO 2026 Lung Cancer Slides

[Slide 2] 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 data captured as a table — no point estimates read from the curve] Endpoint: Overall survival 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) {'timepoint': '6 months', 'arm': 'Tarlatamab + Durvalumab', 'value': '92%', 'ci': None} {'timepoint': '6 months', 'arm': 'Durvalumab', 'value': '75%', 'ci': None} {'timepoint': '12 months', 'arm': 'Tarlatamab + Durvalumab', 'value': '79%', 'ci': None} {'timepoint': '12 months', 'arm': 'Durvalumab', 'value': '53%', 'ci': None} Note: Median follow-up 15.9 vs 16.1 months. Landmark labels printed on curve at dashed 6- and 12-month lines. Event counts in footnote not reliably legible. Durvalumab at-risk row has an extra value at 27 mo (0); tarlatamab row ends at 24 mo (0).

[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 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] 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. 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) CRSa, n (%) Any grade | 140 (49.6) | 4 (1.4) Gr ≥ 3 | 1 (0.4) | 0 ICANSa, 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 4] 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 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 data captured as a table — no point estimates read from the curve] Endpoint: OS 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) {'timepoint': '6 mo', 'arm': 'Tarlatamab + Durvalumab', 'value': '92%', 'ci': None} {'timepoint': '12 mo', 'arm': 'Tarlatamab + Durvalumab', 'value': '79%', 'ci': None} {'timepoint': '6 mo', 'arm': 'Durvalumab', 'value': '75%', 'ci': None} {'timepoint': '12 mo', 'arm': 'Durvalumab', 'value': '53%', 'ci': None} Note: Median FU 15.9 vs 16.1 mo; DCO 7 JUL 2026; landmark timepoints inferred from dashed lines at 6 and 12 months [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 data captured as a table — no point estimates read from the curve] Endpoint: PFS 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) {'timepoint': '6 mo', 'arm': 'Tarlatamab + Durvalumab', 'value': '53%', 'ci': None} {'timepoint': '12 mo', 'arm': 'Tarlatamab + Durvalumab', 'value': '38%', 'ci': None} {'timepoint': '6 mo', 'arm': 'Durvalumab', 'value': '17%', 'ci': None} {'timepoint': '12 mo', 'arm': 'Durvalumab', 'value': '10%', 'ci': None} Note: Median FU 13.9 vs 13.9 mo; DCO 7 JUL 2026; landmark timepoints inferred from dashed lines at 6 and 12 months [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 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

[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 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 data captured as a table — no point estimates read from the curve] Endpoint: Overall survival (ITT) 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) {'timepoint': '6 months', 'arm': 'Tarlatamab + Durvalumab', 'value': '92%', 'ci': None} {'timepoint': '6 months', 'arm': 'Durvalumab', 'value': '75%', 'ci': None} {'timepoint': '12 months', 'arm': 'Tarlatamab + Durvalumab', 'value': '79%', 'ci': None} {'timepoint': '12 months', 'arm': 'Durvalumab', 'value': '53%', 'ci': None} Note: Median FU 15.9 vs 16.1 months. Data cut-off 7 JUL 2026. Event counts from small footnote. [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 data captured as a table — no point estimates read from the curve] Endpoint: Progression-free survival 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) {'timepoint': '6 months', 'arm': 'Tarlatamab + Durvalumab', 'value': '53%', 'ci': None} {'timepoint': '6 months', 'arm': 'Durvalumab', 'value': '17%', 'ci': None} {'timepoint': '12 months', 'arm': 'Tarlatamab + Durvalumab', 'value': '38%', 'ci': None} {'timepoint': '12 months', 'arm': 'Durvalumab', 'value': '10%', 'ci': None} Note: Median FU 13.9 months both arms. Data cut-off 7 JUL 2026. [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

[Slide 3] 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) [Sankey chart of treatment lines 1–3 by therapy group; labels illegible] Miner[?], Lung Cancer 2026 Stephen Liu, MD Stephen Liu, Invited Discussant LBA3 [Slide 4] 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

[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 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 data captured as a table — no point estimates read from the curve] Endpoint: Progression-free survival (PFS) 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) {'timepoint': '6 months', 'arm': 'Tarlatamab + Durvalumab', 'value': '53%', 'ci': None} {'timepoint': '6 months', 'arm': 'Durvalumab', 'value': '17%', 'ci': None} {'timepoint': '12 months', 'arm': 'Tarlatamab + Durvalumab', 'value': '38%', 'ci': None} {'timepoint': '12 months', 'arm': 'Durvalumab', 'value': '10%', 'ci': None} Note: Median FU 13.9 vs 13.9 months. Data cut-off 7 JUL 2026. Key secondary endpoint.

[Slide 2] 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

[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 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

[Slide 3] 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

[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 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

[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 data captured as a table — no point estimates read from the curve] Endpoint: OS 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) {'timepoint': '6 mo', 'arm': 'Tarlatamab + Durvalumab', 'value': '92%', 'ci': None} {'timepoint': '6 mo', 'arm': 'Durvalumab', 'value': '75%', 'ci': None} {'timepoint': '12 mo', 'arm': 'Tarlatamab + Durvalumab', 'value': '79%', 'ci': None} {'timepoint': '12 mo', 'arm': 'Durvalumab', 'value': '53%', 'ci': None} Note: HR 95% CI lower bound 0.32 legible; upper bound not reliably legible. Median follow-up illegible. [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 data captured as a table — no point estimates read from the curve] Endpoint: PFS 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) {'timepoint': '6 mo', 'arm': 'Tarlatamab + Durvalumab', 'value': '53%', 'ci': None} {'timepoint': '6 mo', 'arm': 'Durvalumab', 'value': '17%', 'ci': None} {'timepoint': '12 mo', 'arm': 'Tarlatamab + Durvalumab', 'value': '38%', 'ci': None} {'timepoint': '12 mo', 'arm': 'Durvalumab', 'value': '10%', 'ci': None} Note: Median follow-up 13.9 months both arms.

[Slide 2] 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 data captured as a table — no point estimates read from the curve] Endpoint: Progression-free survival 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) {'timepoint': '6 months', 'arm': 'Tarlatamab + Durvalumab', 'value': '53%', 'ci': None} {'timepoint': '6 months', 'arm': 'Durvalumab', 'value': '17%', 'ci': None} {'timepoint': '12 months', 'arm': 'Tarlatamab + Durvalumab', 'value': '38%', 'ci': None} {'timepoint': '12 months', 'arm': 'Durvalumab', 'value': '10%', 'ci': None} Note: Median FU 13.9 vs 13.9 months. Numbers at risk (0-24 mo, q3): T+D 282,225,147,122,86,32,14,1,0; D 281,106,46,36,18,8,2,0.

[Slide 2] 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] Tarlatamab: DeLLphi-305: Efficacy [Thumbnail of DeLLphi-305 OS KM curve: Median OS 21.1 (20.8–NE) vs 12.7 (11.2–14.8); Median FU 15.9 vs 16.1; HR 0.42 (0.32–0.54); P < 0.001; labels 92%, 79%, 75%, 53%; 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 Sands, ESMO 2026 Stephen Liu, MD

[Slide 1] 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 data captured as a table — no point estimates read from the curve] Endpoint: Overall survival 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) {'timepoint': '6 months (dashed line; inferred)', 'arm': 'Tarlatamab + Durvalumab', 'value': '92%', 'ci': None} {'timepoint': '12 months (dashed line; inferred)', 'arm': 'Tarlatamab + Durvalumab', 'value': '79%', 'ci': None} {'timepoint': '6 months (dashed line; inferred)', 'arm': 'Durvalumab', 'value': '75%', 'ci': None} {'timepoint': '12 months (dashed line; inferred)', 'arm': 'Durvalumab', 'value': '53%', 'ci': None} Note: Median FU 15.9 vs 16.1 months. Landmark timepoints are not printed as text; dashed vertical lines sit at ~6 and ~12 months.

[Slide 4] 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 1] Tarlatamab: DeLLphi-305: Efficacy [KM curve legend] 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% | 75% (6 months); 79% | 53% (12 months) Tarlatamab + Durvalumab Durvalumab 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 [Table] | 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 202[cut off] Stephen Liu, MD [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival 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) {'timepoint': '6 months', 'arm': 'Tarlatamab + Durvalumab', 'value': '92%', 'ci': None} {'timepoint': '6 months', 'arm': 'Durvalumab', 'value': '75%', 'ci': None} {'timepoint': '12 months', 'arm': 'Tarlatamab + Durvalumab', 'value': '79%', 'ci': None} {'timepoint': '12 months', 'arm': 'Durvalumab', 'value': '53%', 'ci': None} Note: Median FU 15.9 vs 16.1 months. Landmark % are labels printed on the curve. Comparison table (discussant-compiled) also shows median PFS 7.4 vs 1.9m, 1y PFS 38% vs 10%, PFS HR 0.34 for DeLLphi-305; IMforte median OS 13.2 vs 10.6m, 1y OS 56% vs 44%, OS HR 0.73, median PFS 5.4 vs 2.1m, 1y PFS 21% vs 12%, PFS HR 0.54. [Slide 2] 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 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] 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)

[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 [MADRID 2026 ESMO congress logo] [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 [MADRID 2026 ESMO congress logo] [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. [MADRID 2026 ESMO congress logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS (single-arm): PFS1 from start of chemotherapy and PFS0 from start of osimertinib 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 {'timepoint': '12 mo', 'arm': 'PFS1', 'value': '58%', 'ci': '43-72'} {'timepoint': '24 mo', 'arm': 'PFS1', 'value': '22%', 'ci': '9-35'} {'timepoint': '12 mo', 'arm': 'PFS0', 'value': '74%', 'ci': '61-87'} {'timepoint': '24 mo', 'arm': 'PFS0', 'value': '31%', 'ci': '16-45'} Note: Single-arm phase 2; two time origins of the same N=50 ITT cohort, not two randomized arms. Median follow-up 30.0 mo (PFS1) / 31.5 mo (PFS0).

[Slide 1] 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 [ESMO Congress Madrid 2026 logo] [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 [ESMO Congress Madrid 2026 logo] [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. [QR code] [ESMO Congress Madrid 2026 logo]

[Slide 3] 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 [MADRID 2026 ESMO congress logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS (single-arm): OS1 from start of chemotherapy and OS0 from start of osimertinib 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 {'timepoint': '12 mo', 'arm': 'OS1', 'value': '87%', 'ci': '78-97'} {'timepoint': '24 mo', 'arm': 'OS1', 'value': '66%', 'ci': '51-81'} {'timepoint': '12 mo', 'arm': 'OS0', 'value': '92%', 'ci': '84-100'} {'timepoint': '24 mo', 'arm': 'OS0', 'value': '72%', 'ci': '59-86'} Note: Single-arm phase 2; two time origins of same N=50 cohort. Median follow-up 30.2 mo (OS1) / 32.4 mo (OS0). Slightly angled photo but numbers legible.

[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 [Inset box] 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 [ESMO Congress logo] 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 [ESMO Congress logo] ALMERIA AUDITORIUM - HALL 9

[Slide 2] 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. Janne et al. AACR 2024, 3. Janne et al. NEJM 2025, 4. Yang et al. NEJM 2025, 5. Li et al. Lancet Oncology 2026 Martin Sebastian [Slide 3] 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 data captured as a table — no point estimates read from the curve] Endpoint: Overall survival (OS1 from start of chemotherapy; OS0 from start of osimertinib), single-arm OS1 (from start of chemotherapy) | 50 | 29.2 | 24.4-42.4 OS0 (from start of osimertinib) | 50 | 31.5 | 27.0-44.4 {'timepoint': '12 months', 'arm': 'OS1', 'value': '87%', 'ci': '78-97'} {'timepoint': '24 months', 'arm': 'OS1', 'value': '66%', 'ci': '51-81'} {'timepoint': '12 months', 'arm': 'OS0', 'value': '92%', 'ci': '84-100'} {'timepoint': '24 months', 'arm': 'OS0', 'value': '72%', 'ci': '59-86'} Note: Single-arm phase 2; two time origins for the same cohort, not comparative arms. Medians in months. Median follow-up 30.2 mo (OS1) and 32.4 mo (OS0).

[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

[Slide 2] 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. [ESMO Congress Madrid 2026 logo] ALMERIA AUDITORIUM - HALL 9 [Survival data captured as a table — no point estimates read from the curve] 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 {'timepoint': '12 mo', 'arm': 'PFS1', 'value': '58%', 'ci': '43-72'} {'timepoint': '24 mo', 'arm': 'PFS1', 'value': '22%', 'ci': '9-35'} {'timepoint': '12 mo', 'arm': 'PFS0', 'value': '74%', 'ci': '61-87'} {'timepoint': '24 mo', 'arm': 'PFS0', 'value': '31%', 'ci': '16-45'} Note: Single-arm phase 2; no comparator HR. Median f/u 30.0 mo (PFS1), 31.5 mo (PFS0). Number-at-risk rows illegible in livestream screenshot.

[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. [ESMO Congress Madrid 2026 logo] [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. [ESMO Congress Madrid 2026 logo] [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. [ESMO Congress Madrid 2026 logo] [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. [ESMO Congress Madrid 2026 logo]

[Slide 1] 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.

[Slide 3] [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 [ESMO Congress Madrid 2026 logo] [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 [QR code] [ESMO Congress Madrid 2026 logo] [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. [QR code] [ESMO Congress Madrid 2026 logo] [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 [QR code] [ESMO Congress Madrid 2026 logo] [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 1] 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. [ESMO Congress Madrid 2026 logo]

[Slide 2] [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] [ESMO Congress Madrid 2026 logo] 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] [ESMO Congress Madrid 2026 logo] [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 [ESMO Congress Madrid 2026 logo] [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 [ESMO Congress Madrid 2026 logo]

[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 data captured as a table — no point estimates read from the curve] 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) {'timepoint': '6 months', 'arm': 'Tarlatamab', 'value': '31%', 'ci': None} {'timepoint': '6 months', 'arm': 'Chemotherapy', 'value': '23%', 'ci': None} {'timepoint': '12 months', 'arm': 'Tarlatamab', 'value': '20%', 'ci': None} {'timepoint': '12 months', 'arm': 'Chemotherapy', 'value': '4%', 'ci': None} Note: Median follow-up 11.0 mo. RMST at 12 mo 5.3 vs 4.3 months. Landmark percentages printed at dashed lines at 6 and 12 months. [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 data captured as a table — no point estimates read from the curve] Endpoint: Duration of response (responders) Tarlatamab | 89 | 6.9 months Chemotherapy | 52 | 5.5 months {'timepoint': '6 months', 'arm': 'Tarlatamab', 'value': '56%', 'ci': None} {'timepoint': '6 months', 'arm': 'Chemotherapy', 'value': '29%', 'ci': None} {'timepoint': '12 months', 'arm': 'Tarlatamab', 'value': '41%', 'ci': None} {'timepoint': '12 months', 'arm': 'Chemotherapy', 'value': '13%', 'ci': None} Note: Responder n taken from at-risk row at time 0 (89 vs 52); DOR table header shows ITT n=254/255. ORR difference not formally tested. [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 data captured as a table — no point estimates read from the curve] 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) {'timepoint': '6 months', 'arm': 'Tarlatamab', 'value': '76%', 'ci': None} {'timepoint': '6 months', 'arm': 'Chemotherapy', 'value': '62%', 'ci': None} {'timepoint': '12 months', 'arm': 'Tarlatamab', 'value': '53%', 'ci': None} {'timepoint': '12 months', 'arm': 'Chemotherapy', 'value': '37%', 'ci': None} Note: Primary analysis; median follow-up 11.2 vs 11.7 mo. Landmarks printed at dashed lines at 6 and 12 months.

[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] 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 data captured as a table — no point estimates read from the curve] Endpoint: PFS by BICR Anbenitamab plus HB1801 THP HR: 0.46 HR CI: 0.37 to 0.56 P: P<0.001 {'timepoint': '24 months', 'arm': 'Anbenitamab plus HB1801', 'value': '58.5', 'ci': None} {'timepoint': '24 months', 'arm': 'THP', 'value': '34.4', 'ci': None} Note: Landmark values printed at dashed 24-month line (24-mo inferred from dashed line position at x=24). No numbers-at-risk or medians printed. [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 data captured as a table — no point estimates read from the curve] Endpoint: Not printed (likely EFS; curve title not shown) Durvalumab-EV Control {'timepoint': '12 months', 'arm': 'Durvalumab-EV', 'value': '73.1', 'ci': None} {'timepoint': '24 months', 'arm': 'Durvalumab-EV', 'value': '63.5', 'ci': None} {'timepoint': '12 months', 'arm': 'Control', 'value': '50.3', 'ci': None} {'timepoint': '24 months', 'arm': 'Control', 'value': '45.1', 'ci': None} Note: Endpoint label not printed on image; 63.5 vs 45.1 at 24 mo matches the EFS figures in the tweet caption, but endpoint is not on the image. Timepoints from dashed lines at 12 and 24 months. No HR/p printed.

[Slide 1] 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.

[Slide 1] 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

[Slide 1] [Photo of presenter standing beside poster; poster board 4404P] Decoding tumor gene expression from circulating chromatin at single-gene resolution [Dana-Farber Cancer Institute logo] [Harvard Medical School logo] 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 [ESMO congress Madrid 2026 logo] [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 [ESMO congress Madrid 2026 logo] [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 [ESMO Congress Madrid 2026 logo] [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 [ESMO Congress Madrid 2026 logo]

[Slide 1] 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 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. [ESMO Congress Madrid 2026 logo] [Slide 3] [ESMO Congress Madrid 2026 logo] 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 data captured as a table — no point estimates read from the curve] Endpoint: Time to deterioration (TTD) in cough 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 Note: PRO endpoint; tolerability panel n=226 vs 220. [Slide 4] [ESMO Congress Madrid 2026 logo] 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 1] 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.

[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 data captured as a table — no point estimates read from the curve] Endpoint: PFS and OS, EGFRwt overall cohort (Temab-A) EGFRwt overall - PFS | 48 | 6.8 months | 5.4-7.3 EGFRwt overall - OS | 48 | 13.8 months | 9.0-25.2 Note: Single-arm cohort; no comparator/HR. DCO 02 June 2026; median f/u 30.1 mo. ORR 50% (24/48), mDOR 5.6 mo. [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 data captured as a table — no point estimates read from the curve] Endpoint: PFS and OS, EGFRmut overall cohort (Temab-A) EGFRmut overall - PFS | 41 | 10.9 months | 8.3-12.3 EGFRmut overall - OS | 41 | 16.2 months | 11.6-19.9 Note: Single-arm cohort; no comparator/HR. DCO 02 June 2026; median f/u 29.5 mo. ORR 63% (26/41), mDOR 9.7 mo.

[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. [ESMO Congress Madrid 2026 logo] [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. [ESMO Congress Madrid 2026 logo] [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. [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS by baseline TP53 mutation status (CROWN ctDNA analysis) 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 {'timepoint': '7 years', 'arm': 'Lorlatinib TP53 mut (-)', 'value': '65.0%', 'ci': '54.5-77.4'} {'timepoint': '7 years', 'arm': 'Lorlatinib TP53 mut (+)', 'value': '40.8%', 'ci': '27.3-61.1'} {'timepoint': '7 years', 'arm': 'Crizotinib TP53 mut (-)', 'value': '2.3%', 'ci': '0.3-15.7'} {'timepoint': '7 years', 'arm': 'Crizotinib TP53 mut (+)', 'value': '6.9%', 'ci': '1.9-24.5'} Note: Within-arm HRs mut(-) vs mut(+): lorlatinib 0.48 (0.268-0.857); crizotinib 0.78 (0.513-1.186). No between-arm HR printed.

[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) [ESMO Congress Madrid 2026 logo] [DESTINY-Lung04 logo] [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 [DESTINY-Lung04 logo] [ESMO Congress Madrid 2026 logo] [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 [DESTINY-Lung04 logo] [ESMO Congress Madrid 2026 logo]

[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 [ESMO Congress Madrid 2026 logo] 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 [ESMO Congress Madrid 2026 logo] 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 [ESMO Congress Madrid 2026 logo] 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 data captured as a table — no point estimates read from the curve] Endpoint: PFS (lorlatinib arm biomarker comparisons) HR: 0.48 HR CI: 0.268–0.857 Note: Four panels; HR/CI shown is TP53 mutation− vs mutation+. Others: EML4::ALK variant 3 vs 1/2 HR 1.09 (0.463–2.579); baseline shedding undetected vs detected HR 0.17 (0.04–0.69); week-4 ctDNA cleared vs persistent HR 0.48 (0.26–0.86), p = 0.013 printed on that panel. Inset medians too small to read reliably — not filled.

[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

[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 [ESMO Congress Madrid 2026 logo] [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 [ESMO Congress Madrid 2026 logo] [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 [ESMO Congress Madrid 2026 logo] [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 [ESMO Congress Madrid 2026 logo]

[Slide 1] 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. [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS (CROWN, 7-year update reprinted from Shaw et al. Ann Oncol 2026) Lorlatinib | 149 | NR | 68.5-NR Crizotinib | 147 | 9.1 | 7.4-10.9 HR: 0.19 HR CI: 0.13-0.26 Note: Curve has printed percentage annotations (lorlatinib 70/65/63/60/57/55%; crizotinib 15/10/10/8/7/3%) but timepoint for each is not legibly labeled in this angled photo, so not entered as landmarks.

[Slide 2] 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] [MADRID 2026 ESMO congress logo] 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] [ESMO Congress Madrid 2026 logo] 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] [ESMO Congress Madrid 2026 logo] 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 data captured as a table — no point estimates read from the curve] Endpoint: PFS by Week 1 and Week 2 plasma EGFRm clearance status (FLAURA2, DCO 03 April 2023) 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 Note: Subgroup HRs (95% CI): Wk1 clearance 0.58 (0.33, 1.02); Wk1 no clearance 0.53 (0.35, 0.80); Wk2 clearance 0.53 (0.35, 0.81); Wk2 no clearance 0.57 (0.34, 0.96). Unstratified Cox. n = number at risk at time 0. Medians in months. [Slide 4] [ESMO Congress Madrid 2026 logo] 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] 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 [ESMO Congress Madrid 2026 logo] [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 [ESMO Congress Madrid 2026 logo] [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 [ESMO Congress Madrid 2026 logo]

[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 [ESMO Congress Madrid 2026 logo] [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 [ESMO Congress Madrid 2026 logo] [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 [ESMO Congress Madrid 2026 logo]

[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 [ESMO Congress Madrid 2026 logo]

[Slide 2] 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

[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.

[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. [ESMO Congress Madrid 2026 logo] [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. [ESMO Congress Madrid 2026 logo] [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. [ESMO Congress Madrid 2026 logo] [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. [ESMO Congress Madrid 2026 logo]

[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 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. [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS by baseline EML4::ALK fusion variant (CROWN ctDNA analysis) 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 {'timepoint': '7 years', 'arm': 'Lorlatinib variants 1/2', 'value': '52.7%', 'ci': '35.7-77.7'} {'timepoint': '7 years', 'arm': 'Lorlatinib variant 3', 'value': '42.6%', 'ci': '24.1-75.4'} {'timepoint': '7 years', 'arm': 'Crizotinib variants 1/2', 'value': '5.1%', 'ci': '0.8-33.7'} {'timepoint': '7 years', 'arm': 'Crizotinib variant 3', 'value': '0%', 'ci': 'NA-NA'} Note: Within-arm HRs variant 3 vs variants 1/2: lorlatinib 1.09 (0.463-2.579); crizotinib 1.27 (0.697-2.318). Variant 2 (n=6): ORR 100%, median PFS NR. [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. [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS by baseline methylation-based tumor fraction (lorlatinib group, CROWN) 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 Note: HR is ND vs D within lorlatinib group (prognostic, not treatment comparison). Assay: Guardant Infinity (755 genes). [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. [ESMO Congress Madrid 2026 logo]

[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] 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 4] 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 [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS by histology (BAT8008 + BAT1308, 1L NSCLC, single arm) All-comer | 98 | 12.42 months | 6.67-NE Squamous | 52 | 7.72 months | 5.26-NE Non-squamous | 46 | NR | 7.89-NE Note: Single-arm study; no HR. Date cutoff Sep 22, 2026. Spider plot N=50/45 differs from KM N=52/46 (as printed).

[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

[Slide 1] 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 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 [ESMO Congress Madrid 2026 logo; presenter on video feed] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS according to ctDNA response (sevabertinib, HER2 TKD-mutant NSCLC; Le et al NEJM 2025) 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 Note: Number at risk at 0/6/12/18/24/30 mo: never detected 12,8,5,1,0,0; clearance 28,20,12,7,3,1; persistent 25,11,5,1,0,0. Ensartinib panel values small/angled - not transcribed. [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 [ESMO Congress Madrid 2026 logo; presenter on video feed] [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. [ESMO Congress Madrid 2026 logo; discussant on video feed]

[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

[Slide 1] 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 [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS by Week 3 / Week 6 plasma EGFRm clearance status (FLAURA2 exploratory ctDNA analysis) 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 Note: Four subgroup panels. n from number-at-risk at time 0 (small text, angled photo). DCO 03 April 2023 (PFS). Unstratified HRs.

[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 [ESMO Congress Madrid 2026 logo] [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 [ESMO Congress Madrid 2026 logo] [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 [ESMO Congress Madrid 2026 logo] [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 [MADRID 2026 ESMO congress logo]

[Slide 2] [Livestream frame] 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic CHAIRS: PASI A. JANNE, XIUNING LE, EGBERT SMIT [DESTINY-Lung04 logo] 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 [ESMO Congress Madrid 2026 logo] [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 data captured as a table — no point estimates read from the curve] Endpoint: Time to sustained deterioration, NSCLC-SAQ cough (also chest pain, dyspnea) 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 Note: Other HRs: chest pain 0.80 (0.54, 1.18); dyspnea 0.77 (0.54, 1.09). Full analysis set. DCO June 9, 2026. [Slide 3] [Livestream frame] 08:30 - 10:00 Rapid oral 1: NSCLC, metastatic CHAIRS: PASI A. JANNE, XIUNING LE, EGBERT SMIT [DESTINY-Lung04 logo] 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 [ESMO Congress Madrid 2026 logo] [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 [DESTINY-Lung04 logo] 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 [QR code] 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). [ESMO Congress Madrid 2026 logo] [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 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 [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS by plasma EGFRm clearance status (Week 1 and Week 2) 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 Note: Data cut-off 03 April 2023 (PFS). Low-resolution capture: small numbers read with moderate confidence - vision-verify against higher-res source before publishing.

[Slide 4] 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).

[Slide 1] Time to deterioration in pulmonary symptoms (NSCLC-SAQ) [DESTINY-Lung04 logo] 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) No. at risk (months 0-36 by 3): 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 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) No. at risk: 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 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) No. at risk: T-DXd: 227 85 62 50 36 29 23 18 11 9 7 5 Pembro + chemo: 227 66 36 21 16 9 5 5 3 3 1 1 Y axis: Probability of no sustained deterioration; X axis: Time from randomization (months) 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 [Survival data captured as a table — no point estimates read from the curve] Endpoint: Time to sustained deterioration in pulmonary symptoms (NSCLC-SAQ) - cough (chest pain and dyspnea in notes) T-DXd | 227 | 26.2 months | 17.1, NC Pembro + chemo | 227 | 11.0 months | 7.4, NC HR: 0.66 HR CI: 0.44, 0.99 Note: Chest pain: T-DXd NC (11.5, NC) vs 14.3 (8.3, NC), HR 0.80 (0.54, 1.18). Dyspnea: T-DXd 10.1 (6.5, NC) vs 5.1 (3.2, 11.0), HR 0.77 (0.54, 1.09). No p-values printed. Full analysis set. [Slide 2] Physical functioning (EORTC QLQ-C30 physical functioning scale) [DESTINY-Lung04 logo] [Line chart: Proportion of patients reporting 'maintained or improved' physical function (%) by Week 2-56, plus EOT] T-DXd (n=226) Pembro + chemo (n=220) 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 E[cut off at frame edge] 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

[Slide 1] A [Waterfall plot] Change from baseline (%) Best response: CR / SD / PR / PD [y-axis 60 to -120; dotted reference lines at +20 and -30] B [Spider plot] Change from baseline (%) Clinical response: CR / SD / PR / PD Baseline | After Cycle 2 | After Cycle 4 [Slide 2] C Event-free survival (%) Events: 8 (32%) Median EFS: Not reached 12-month EFS rate: 72.0 (95%CI 50.1-85.5) 24-month EFS rate: 67.8 (95%CI 45.7-82.4) Time from enrolment (months) Number at risk 0: 25 | 6: 22 | 12: 18 | 18: 12 | 24: 5 D Overall survival (%) Deaths: 3 (12%) Median OS: Not reached 12/24-month OS rate: 88.0 (95%CI 67.3-96.0) Time from enrolment (months) Number at risk 0: 25 | 6: 23 | 12: 22 | 18: 16 | 24: 7 [Survival data captured as a table — no point estimates read from the curve] Endpoint: EFS and OS (single arm) Toripalimab + TIP (single arm) - EFS | 25 | Not reached Toripalimab + TIP (single arm) - OS | 25 | Not reached {'timepoint': '12 months', 'arm': 'EFS', 'value': '72.0%', 'ci': '50.1-85.5'} {'timepoint': '24 months', 'arm': 'EFS', 'value': '67.8%', 'ci': '45.7-82.4'} {'timepoint': '12 months', 'arm': 'OS', 'value': '88.0%', 'ci': '67.3-96.0'} {'timepoint': '24 months', 'arm': 'OS', 'value': '88.0%', 'ci': '67.3-96.0'} Note: Single-arm phase 2; arm label not printed on slide, taken as the single study cohort (n=25 at risk at time 0).

[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 data captured as a table — no point estimates read from the curve] 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 Note: Slide prints 'mOS 19.3 vs 24.8 months' and 'mPFS 10.6 vs 11.0 months' without per-arm labels, so medians left null per arm. Tweet caption attributes OS 19.3 to surgery and 24.8 to no surgery. Source: Lim E, Lancet Resp Med 2024. [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

[Slide 3] 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 data captured as a table — no point estimates read from the curve] Endpoint: PFS and OS (single arm, EGFRmut overall) EGFRmut overall - PFS | 41 | 10.9 | 8.3–12.3 EGFRmut overall - OS | 41 | 16.2 | 11.6–19.9 Note: Single-arm; no HR. ORR 63% (26/41), median DOR 9.7 mo (8.3–13.5). Median f/u 29.5 mo.

[Slide 1] [ESMO Congress Madrid 2026 logo] 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] [ESMO Congress Madrid 2026 logo] 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 data captured as a table — no point estimates read from the curve] 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) Note: Values read from a photographed slide at an angle; several CI digits are low-confidence (esp. Week 1 clearance lower CI 10.3, Week 1 no-clearance lower CI 16.8, Week 2 HR CIs). Medians in months. DCO 03 April 2023 (PFS). Re-verify against source before publishing. [Slide 3] [ESMO Congress Madrid 2026 logo] 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 data captured as a table — no point estimates read from the curve] Endpoint: PFS by baseline plasma EGFRm detection (osi + plat-pem vs osi mono) 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 Note: Medians in months. Some CI digits low-confidence (detected-arm CIs 19.8/27.9 and 13.6/16.6). Reference: Janne PA et al. Cancer Res 2024 (AACR Clinical Trials Plenary). No DCO printed on this slide.

[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 [ESMO Congress Madrid 2026 logo] [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 [ESMO Congress Madrid 2026 logo] [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) [Institution logos] [Slide 4] Conclusion [Institution logos] ➢ 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

[Slide 3] 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. [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS by week-4 ctDNA tumor fraction clearance (lorlatinib group, TF detected at baseline) 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 Note: Week 24 panel: D/ND n=58 median NR (60.0-NR); D/D n=15 median 32.9 (16.4-NR); HR 0.45 (0.21-0.96). Title says 7-year PFS but no 7-year landmark value is printed.

[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

[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 [EZN Europäisches Zentrum für Neuroonkologie logo] [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. [EZN Europäisches Zentrum für Neuroonkologie logo] [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 [EZN Europäisches Zentrum für Neuroonkologie logo]

[Slide 1] [MADRID 2026 ESMO congress logo] 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] [MADRID 2026 ESMO congress logo] 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] [MADRID 2026 ESMO congress logo] 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 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 [Speaker panel: David Planchard] 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 [ESMO Congress Madrid 2026 logo] [Speaker panel: David Planchard] ALMERIA AUDITORIUM - HALL 9 [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS (investigator) by Week 3 plasma EGFRm clearance - clearance subgroup (other subgroups in notes) 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 Note: Week 3 no clearance: 19.6 (10.1, NR) n=21 vs 11.1 (5.6, 16.5) n=40, HR 0.62 (0.33, 1.14). Week 6 clearance: 27.6 (22.0, NR) n=92 vs 16.4 (13.8, 19.4) n=122, HR 0.54 (0.38, 0.77). Week 6 no clearance: 11.8 (5.1, 22.1) n=13 vs 5.4 (1.6, 5.7) n=15, HR 0.39 (0.16, 0.94). n = number at risk at time 0. No p-values printed.

[Slide 1] [ESMO congress Madrid 2026 logo] 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 data captured as a table — no point estimates read from the curve] Endpoint: PFS by baseline plasma EGFRm status 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) Note: Low-resolution screenshot; detected-subgroup HR and lower CI digits are small and slightly blurred (read as 0.60, 0.45 - verify vs source). Unstratified log-rank HR. Data from Jänne AACR 2024 CT017 per footnote. [Slide 2] [ESMO congress Madrid 2026 logo] 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 data captured as a table — no point estimates read from the curve] Endpoint: PFS (pooled across both arms) by baseline plasma EGFRm 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 Note: Pooled prognostic analysis, both treatment arms combined; unstratified log-rank HR. Footnote small/low-res.

[Slide 1] [ESMO Congress Madrid 2026 logo] 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] 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 [ESMO Congress Madrid 2026 logo]

[Slide 1] 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 [ESMO Congress Madrid 2026 logo]

[Slide 1] 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 [ESMO Congress Madrid 2026 logo]

[Slide 1] 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 [ESMO Congress logo]

[Slide 1] 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 [ESMO Congress Madrid 2026 logo]

[Slide 1] 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 [ESMO congress logo]

[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 [ESMO Congress Madrid 2026 logo] [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 [ESMO Congress Madrid 2026 logo]

[Slide 1] Key Takeaways - In addition to plasma ctDNA NGS to ensure timely complete molecular testing for patients with advanced NSCLC, - Emerging data support new approaches: - Pre-diagnostic Plasma ctDNA NGS (if + treat, if -, await tissue) - Screening test for molecular diagnosis in targeted therapy resistance - Proteomic signature prediction including for immunotherapy - clinical utility data pending - Need more trials to demonstrate clinical utility of using ctDNA to adapt treatment, demonstrating benefit to patients Natasha B. Leighl [ESMO Congress Madrid 2026 logo]

[Slide 1] 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) Overall Survival OS | EGFRwt Overall (n=48) Events, n (%) | 33 (69) Median OS, months (95% CI) | 13.8 ([illegible: 9.0 or 8.0]–25.2) Survival Probability; Months; + Censored Overall [number-at-risk rows - illegible at this resolution] Data cutoff date: 02 June 2026. [footnote line illegible] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS and OS, Temab-A monotherapy, overall EGFRwt non-squamous NSCLC cohort (single-arm) EGFRwt Overall - PFS | 48 | 6.8 mo | 5.4–7.3 EGFRwt Overall - OS | 48 | 13.8 mo | [illegible lower bound: 9.0 or 8.0]–25.2 Note: Single-arm phase 1 cohort - no comparator, no HR. OS CI lower bound not reliably legible in low-res image. Data cutoff 02 June 2026; median follow-up 30.1 months.

[Slide 1] Professional Reviewed Abstracts | AMCP Nexus 2026 | October 26-29 S71 [phone overlay: 73 de 128] [Left column — end of preceding, unrelated abstract on BCG-unresponsive HR-NMIBC:] 10/31/2024). Patients with HR-NMIBC with CIS (as of the HR-NMIBC diagnosis date) were assessed for BCG discontinuation (last instillation before a gap of >180 days, following adequate induction [≥5 instillations in 70 days] and maintenance [≥2 instillations in 180 days]). BCG-unresponsive was defined as receiving subsequent guideline-recommended treatment (index date) within 365 days of BCG discontinuation. EFS (i.e., time without disease recurrence or progression) was defined based on codes for transurethral resection of bladder tumor, biopsy, systemic chemotherapy, non-index intravesical therapy, radiotherapy, radical cystectomy, pembrolizumab, diagnosis of muscle-invasive or metastatic bladder cancer, or death. EFS was described using Kaplan-Meier analyses among patients with bladder-sparing therapy after BCG discontinuation. RESULTS: Overall, there were 130 patients with BCG-unresponsive HR-NMIBC with CIS ± PAP (mean age: 70 years; 77% male; 78% White). Among 108 patients initiating bladder-sparing therapies as next treatment, gemcitabine (52%), mitomycin (16%), pembrolizumab (15%), gemcitabine/docetaxel (14%), nadofaragene firadenovec-vncg (2%), docetaxel (1%), and doxorubicin (1%) were used. Among these patients, median EFS was 5.9 months (95% CI: 4.7, 12.1). The proportion of patients with recurrence, progression, or death was 28% at 3 months, 50% at 6 months, 59% at 12 months, 69% at 24 months and 74% at 36 months. CONCLUSIONS: Treatment choice varied among patients with BCG-unresponsive HR-NMIBC with CIS. Despite radical cystectomy being recommended for BCG-unresponsive patients, most received bladder-sparing therapy, with 50% having disease recurrence, progression, or death within 6 months. These findings highlight a significant unmet need for more effective bladder-sparing therapies among BCG-unresponsive patients with HR-NMIBC with CIS ± PAP. SPONSORSHIP: Johnson & Johnson 269 IMforte: Characterization of hospitalization and health care resource utilization with lurbinectedin + atezolizumab maintenance therapy for extensive-stage small cell lung cancer in a phase 3 randomized trial Nguyen A1, Paz-Ares L2, Borghaei H3, Liu S4, Peters S5, Reck M6, Cuchelkar V7, Mohan S7, Lin Y7, Fuller D1, Profant D8, Bhatt K1, Herbst R9; amy.nguyen@jazzpharma.com 1Jazz Pharmaceuticals, Philadelphia, PA, USA; 2Department of Medical Oncology, Hospital Universitario 12 de Octubre, H12O-CNIO Lung Cancer Unit, Universidad Complutense and Ciberonc, Madrid, Spain; 3Department of Hematology and Oncology, Fox Chase Cancer Center, Philadelphia, PA, USA; 4Department of Medical Oncology, Lombardi Comprehensive Cancer Center, Georgetown University, Washington, DC, USA; 5Department of Oncology, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland; 6LungenClinic, Airway Research Center North, German Center for Lung Research, Grosshansdorf, Germany; 7Genentech, Inc., South San Francisco, CA, USA; 8Jazz Pharmaceuticals, Carlsbad, CA, USA; 9Department of Medical Oncology and Hematology, Yale School of Medicine, New Haven, CT, USA BACKGROUND: Extensive-stage small cell lung cancer (ES-SCLC) is an aggressive disease with poor prognosis. Among patients (pts) with ES-SCLC, rates of complications that drive health care resource utilization (HCRU) are high. The phase 3 IMforte (NCT05091567) trial demonstrated statistically significant and clinically meaningful improvements in progression-free and overall survival with lurbinectedin + atezolizumab (L+A) vs atezolizumab alone (A) for first-line maintenance treatment (1Lm) of ES-SCLC. OBJECTIVE: To describe hospitalizations and HCRU from IMforte pts receiving 1Lm with L+A vs A. METHODS: Eligible pts with ES-SCLC who were progression free after induction with atezolizumab, carboplatin, and etoposide were randomized 1:1 to receive 1Lm L+A or A every 3 weeks. Details of all-cause and treatment-related inpatient hospitalizations and supportive care during 1Lm were summarized descriptively. RESULTS: As of February 12, 2025, the median treatment duration was 4.4 (L)/4.8 (A) and 2.1 months in the L+A and A arms. One or more treatment-related hospitalizations occurred in 25/242 (10%) pts in the L+A arm and 6/240 (3%) pts in the A arm. The median (quartile 1 [Q1]–Q3) number and total length of treatment-related hospitalizations were 1 (1–1) admission and 7 (4–14) days in the L+A arm and 1 (1–1) admission and 7 (2–13) days in the A arm, respectively. Regardless of relationship to study drug, 60/242 (25%) and 34/240 (14%) pts had an all-cause hospitalization in the L+A and A arms, respectively. The median (Q1–Q3) number and total length of all-cause hospitalizations were 1 (1–2) admission and 7 (5–14) days in the L+A arm and 1 (1–1) admission and 9 (4–17) days in the A arm, respectively. The most common preferred term used among all-cause hospitalizations was pneumonia (L+A, n=5 [2%]; A, n=6 [3%]). Most pts (215/242 [89%]) in the L+A arm used granulocyte colony-stimulating factor, compared to 28/240 (12%) in the A arm. CONCLUSIONS: Consistent with the extended treatment duration, L+A appeared to be associated with a higher incidence of hospitalization than A. Nonetheless, the frequency remained low: 90% of all pts in the L+A arm did not experience a treatment-related hospitalization. Among pts who were hospitalized, the number of visits and lengths of stay were similar Vol. 33, No. 10-d | October 2026 | JMCP.org S72 Professional Reviewed Abstracts | AMCP Nexus 2026 | October 26-29 between treatment arms, suggesting L+A did not increase inpatient burden on a per-event basis. Overall, L+A showed a manageable and predictable HCRU pattern, supporting its feasibility from clinical and health economic perspectives. SPONSORSHIP: F. Hoffmann-La Roche Ltd and Jazz Pharmaceuticals [right column of S72 — start of unrelated abstract:] perceived benefits of switching from bid to qd varied depending on overall pill burden and daily routines. CONCLUSIONS: Treatment frequency, disease severity, and pill burden influence patient preference and anticipated adherence in MF, PV, and GVHD. Patients demonstrated a clear preference for qd regimens, which may support adherence through simplified treatment routines. Willingness to switch [cut off]

[Slide 1] [ESMO Congress Madrid Spain 23-27 October 2026 logo] 4430P – ETOP Lungscope: Concordance of MTAP Expression Between Primary NSCLC and Matched Metastases by IHC [ETOP·IBCSG Foundation for International Lung & Breast Cancer Research logo] [50 Years IBCSG logo] K. Kerr, S. Finn, U. Dafni, R. Kammler, M. Christoffe, L. Joseph, P. Hofman, T. Radonic, [obscured by presenter] M. Haberecker, S. Berezowska, A.-M. Baird, K. Monkhorst, E. Matsouka, S. Peters, A. Stenzinger, [obscured] L. Bubendorf [affiliations — small, partially legible] Background • ETOP Lungscope is a retrospective translational biobank comprising clinically annotated tissue samples from primary resected non-small cell lung cancer (NSCLC) and matched concomitant or sequential locoregional or distant metastases. • The biobank enables investigation of the frequency and consistency of the expression of relevant biomarkers and their association with clinical parameters in NSCLC. • MTAP (methylthioadenosine phosphorylase) loss/deletion is becoming an important predictive biomarker for drugs targeting the PRMT5–MAT2A pathway. • Here, we present data on MTAP loss determined by immunohistochemistry in primary tumors and matched metastases and its association with clinical parameters. Methods • MTAP protein expression was assessed by IHC in primary and metastatic samples from the ETOP Lungscope biobank using the MSVA-741R antibody (MS Validated Antibodies, Hamburg, Germany) on the Leica BOND-PRIME staining platform. • For each sample, 3 tumor cores (1mm Ø) were independently assessed by 3 scorers and classified as MTAP-deficient, MTAP-proficient, equivocal, or non-evaluable, based on tumor cells versus internal controls. • Final MTAP status was determined separately for each primary and metastatic sample based on the consensus classification of evaluable cores. In cases of discordance among the 3 cores, the majority classification was assigned. • Association between MTAP status of the primary samples, baseline characteristics, selected biomarkers and overall survival (OS) were analysed. Acknowledgment Financial support from Bärbel und Paul Geissbühler Stiftung Results • 349 patients in Lungscope database • 333 patients underwent MTAP expression testing by IHC from 12 centers/7 countries • Among 333 paired samples, 58% had synchronous metastasis, 40% metachronous and 2% both • Inter-observer initial agreement: 89% unanimous (3/3 scorers), 10% partial (2/1), 1% discordance Primary-metastatic MTAP concordance [Sankey diagram; see full transcription in 2113622829031985382_s2] 256/266* (96%) matched pairs Cohen's k=0.89 (proficient vs deficient) *excluding non-evaluable, equivocal samples [Results table and OS curve partially obscured by presenter] This presentation is the intellectual property of the authors and the ETOP IBCSG Partners Fou[ndation] [Adjacent posters partially visible: 4429P (BRCA wild-type / PARP inhibitor) and 'Prediction of circulating tumor ... BRCA testing in metastatic pro[state cancer]' — not transcribed] [Slide 2] [ESMO Congress Madrid Spain 23-27 October 2026 logo] 4430P – ETOP Lungscope: Concordance of MTAP Expression Between Primary NSCLC and Matched Metastases by IHC [ETOP·IBCSG Foundation for International Lung & Breast Cancer Research logo] [50 Years IBCSG logo] K. Kerr1, S. Finn2, U. Dafni3, R. Kammler4, M. Christoffe14, L. Joseph5, P. Hofman6, T. Radonic7, L. Hillen8, L. Horvath9, M. Haberecker10, S. Berezowska11, A.-M. Baird12, K. Monkhorst13, E. Matsouka14, S. Peters15, A. Stenzinger16, R. A. Stahel17, J. Naidoo18, L. Bubendorf19 [affiliations — small print, largely illegible] Background • ETOP Lungscope is a retrospective translational biobank comprising clinically annotated tissue samples from primary resected non-small cell lung cancer (NSCLC) and matched concomitant or sequential locoregional or distant metastases. • The biobank enables investigation of the frequency and consistency of the expression of relevant biomarkers and their association with clinical parameters in NSCLC. • MTAP (methylthioadenosine phosphorylase) loss/deletion is becoming an important predictive biomarker for drugs targeting the PRMT5–MAT2A pathway. • Here, we present data on MTAP loss determined by immunohistochemistry in primary tumors and matched metastases and its association with clinical parameters. Methods • MTAP protein expression was assessed by IHC in primary and metastatic samples from the ETOP Lungscope biobank using the MSVA-741R antibody (MS Validated Antibodies, Hamburg, Germany) on the Leica BOND-PRIME staining platform. • For each sample, 3 tumor cores (1mm Ø) were independently assessed by 3 scorers and classified as MTAP-deficient, MTAP-proficient, equivocal, or non-evaluable, based on tumor cells versus internal controls. • Final MTAP status was determined separately for each primary and metastatic sample based on the consensus classification of evaluable cores. In cases of discordance among the 3 cores, the majority classification was assigned. • Association between MTAP status of the primary samples, baseline characteristics, selected biomarkers and overall survival (OS) were analysed. Acknowledgment Financial support from Bärbel und Paul Geissbühler Stiftung Results • 349 patients in Lungscope database • 333 patients underwent MTAP expression testing by IHC from 12 centers/7 countries • Among 333 paired samples, 58% had synchronous metastasis, 40% metachronous and 2% both • Inter-observer initial agreement: 89% unanimous (3/3 scorers), 10% partial (2/1), 1% discordance Primary-metastatic MTAP concordance Primary sample: proficient (N=242; 73%); deficient (N=68; 20%); non-evaluable (N=23; 7%) Metastatic sample: proficient (N=224; 67%); deficient (N=63; 19%); equivocal (N=2; 1%); non-evaluable (N=44; 13%) [Sankey flow values: 206, 28, 7, 3, 50, 1, 14, 1, 15, 6, 2] 256/266* (96%) matched pairs Cohen's k=0.89 (proficient vs deficient) *excluding non-evaluable, equivocal samples MTAP* association with baseline characteristics and biomarkers Characteristics | MTAP-proficient (n=242) | MTAP deficient (n=68) | p-value Age Median (Range) | 66 (34-85) | 65 (45-83) | 0.63 Sex Male / Female | 131 (82%) / 111 (73%) | 28 (18%) / 40 (27%) | 0.074 Ethnicity Caucasian / Asian/Other / Missing | 173 (80%) / 2 (40%) / 67 (76%) | 44 (20%) / 3 (60%) / 21 (24%) | 0.065 Histology Adenocarcinoma / Squamous cell / Other | 185 (82%) / 33 (62%) / 23 (77%) | 41 (18%) / 20 (38%) / 7 (23%) | 0.011 Smoking History Yes / No | 225 (80%) / 14 (64%) | 58 (20%) / 8 (36%) | 0.10 Stage at initial diagnosis (8th TNM edition) I / II / III / IV / Missing | 23 (82%) / 21 (78%) / 93 (76%) / 25 (68%) / 80 (83%) | 5 (18%) / 6 (22%) / 29 (24%) / 12 (32%) / 16 (17%) | 0.57 EGFR mutation Mutated / Non-mutated | 14 (58%) / 130 (83%) | 10 (42%) / 27 (17%) | 0.012 KRAS mutation Mutated / Non-mutated | 53 (81%) / 77 (78%) | 12 (19%) / 22 (22%) | 0.69 PD-L1 expression Positive (≥1%) / Negative (<1%) | 81 (81%) / 47 (78%) | 19 (19%) / 13 (22%) | 0.69 TP53 mutation Mutated / Non-mutated | 48 (89%) / 22 (69%) | 6 (11%) / 10 (31%) | 0.042 *MTAP status based on primary sample Missing categories not presented in the table Overall Survival by MTAP* MTAP proficient — Deaths (%) 137 (57%) — Median OS (95% CI) 43.2m (32.5-49.4) MTAP deficient — Deaths (%) 46 (68%) — Median OS (95% CI) 25.6m (21.0-34.9) Log-rank p=0.07 Median follow-up: 61.0/65.6 m +Censored Number at risk (Months 0 / 24 / 48 / 72 / 96 / 120 / 144 / 168): MTAP proficient: 242 (0) / 130 (43) / 61 (73) / 33 (86) / 16 (95) / 8 (101) / 1 (104) / 0 (105) MTAP deficient: 68 (0) / 34 (6) / 10 (16) / 5 (19) / 4 (20) / 3 (21) / 2 (21) / 0 (22) *MTAP status based on primary sample MTAP deficiency was significantly associated with worse OS after adjustment for age, EGFR and TP53 status (adjHR=1.55, 95%CI: 1.09–2.19; p = 0.014). Conclusion • In this large clinically annotated cohort of matched primary and metastatic NSCLCs, MTAP-deficient tumors were identified in ~20% of cases and enriched in squamous histology. • MTAP status showed high concordance between primary and matched metastases, supporting biologic stability across disease sites and time. • Adjusted for baseline clinicopathological characteristics, MTAP deficiency was significantly associated with worse OS. • These findings support the use of archival tissue for MTAP IHC to identify patients for biomarker-driven clinical trials. This presentation is the intellectual property of the authors and the ETOP IBCSG Partners Foundation. Please contact Lungscope@etop.ibcsg.org for permission to reprint and/or distribute. [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival by MTAP status (primary sample) MTAP proficient | 242 | 43.2m | 32.5-49.4 MTAP deficient | 68 | 25.6m | 21.0-34.9 HR: adjHR=1.55 (deficient vs proficient, adjusted for age, EGFR, TP53) HR CI: 1.09–2.19 P: Log-rank p=0.07 (unadjusted); adjusted p=0.014 Note: Median follow-up 61.0/65.6 m. Number-at-risk transcribed in ocr_text; landmark rates not printed.

[Slide 1] Financial toxicity of lung cancer: Findings from a pan-European survey of people living with lung cancer and caregivers Lung Cancer Europe Authors: Diego Villalón, Angeliki Paraskevopoulou-Soun, Alexandre Brutt [?], Lavinia Magee [?], Merel Hennink, Nicoline Ehrhardt, Debra Montague MÁS QUE IDEAS Foundation, Barcelona, Spain; Lung Cancer Europe, Bern, Switzerland BACKGROUND Lung cancer is the leading cause of global cancer mortality. Beyond its physical and mental burden, lung cancer imposes a substantial and often overlooked financial impact on people living with lung cancer and caregivers. Understanding this impact is increasingly important as the cost-of-living crisis amplifies economic strain. METHODS An online survey (26 May–05 July 2023) examined the costs associated with lung cancer diagnosis, treatment, and follow-up care, as well as the impacts on employment, income, and access to resources. Descriptive statistics were used to analyse responses from 1,161 participants across 28 European countries (834 people living with lung cancer and 327 caregivers). 1,161 PARTICIPANTS | 834 PARTICIPANTS LIVING WITH LUNG CANCER | 327 CAREGIVERS | 28 EUROPEAN COUNTRIES RESULTS Financial difficulties were reported by 66.7% of participants; 45.7% struggled to pay for lung cancer-related expenses. Of those experiencing financial difficulties, 82.1% viewed finances as a barrier to treatment, recovery, and care, and 67.5% reported mental health impacts. Most reported a mix of medical (73.5%) and non-medical expenses (87.2%), with travel for medical reasons (83.7%), household utilities such as heating (77.3%), and personal care items such as dressings, bandages, and creams (72.3%) most reported. The financial impact of these costs was compounded by a decrease in household income for 62.9% of participants; 49.5% experiencing reductions of >30%. Causes were mainly related to employment changes: 39.7% cited an inability to work, and 27.9% and 26.8% lost income due to work absence and retirement due to illness, respectively. Support for those affected was limited: 39.2% and 46.7% of participants reported that financial benefits for people living with lung cancer and caregivers, respectively, were not available in their countries. Healthcare coverage and social welfare provisions vary notably across countries surveyed; the impacts were highest in Ukraine, Greece, Croatia, and Poland. 88.4% reported an impact on ≥1 area of daily life. THE FINANCIAL BURDEN OF LUNG CANCER Direct and indirect costs | Impact on household income | Support doesn't reach everyone 66.7% of participants reported financial difficulties | 62.9% of participants have faced a decrease in household income | 39.2% reported that financial benefits for people living with lung cancer were not available in their country 82.1% of those who reported financial difficulties viewed finances as a barrier to treatment and recovery | 49.5% reported reductions of more than 30% of their household income | 46.7% of participants reported that caregivers do not receive financial support in their country • Burden of extra medical and non-medical expenses • Employment changes • Additional personal care items related to treatment | • Inability to work • Lost income due to work absence • Premature retirement due to illness | • Healthcare coverage and social welfare vary substantially • Countries most affected: Ukraine, Greece, Croatia, and Poland CONCLUSION Despite widespread financial hardship, targeted support for those impacted by lung cancer across Europe is limited. Routine financial toxicity screening, support for those at risk, and expanded income replacement policies are needed for those affected.