ESMO 2026 GI Cancer Slides

[Slide 1] Abstract Proffered Paper session LBA78 - Active surveillance versus standard surgery after neoadjuvant chemoradiotherapy for oesophageal cancer: Long-term results of a cluster-randomised controlled trial (SANO) Speakers Sanjiv S. Gangaram-Panday (Rotterdam, Netherlands) Lecture Time 13:40 - 13:50 Abstract Background The Surgery As Needed for Oesophageal cancer (SANO)-trial demonstrated that active surveillance is non-inferior to standard surgery in terms of two-year overall survival in patients with a complete clinical response after neoadjuvant chemoradiotherapy (nCRT). Active surveillance resulted in improved quality of life. Here, we report 5-year outcomes. [Slide 2] Methods From 2017 to 2021, a multicentre stepped-wedge cluster-randomised non-inferiority trial was conducted. Patients with locally advanced oesophageal cancer who achieved complete clinical response (defined as no tumour detected on endoscopic biopsies, -ultrasound, or PET-CT) underwent active surveillance or standard surgery. Cox regression was used to compare 5-year overall and disease-free survival, and the cumulative incidence of distant metastases was estimated using competing risk analysis. [Slide 3] Results A total of 309 patients were included, 198 underwent active surveillance and 111 underwent standard surgery. Median follow-up was 76 months (IQR 67–86). Five-year overall survival was not statistically significantly different between active surveillance and standard surgery (50%, 95%CI 34-57 vs. 54%, 95%CI 46-64; HR 1.17, 95%CI 0.77-1.77) with similar results across histologic subtypes. Disease-free survival was not statistically significantly different (HR 1.39, 95% CI 0.94-2.06). The cumulative incidence of distant metastases at five years was significantly higher in the active surveillance group compared to the standard surgery group (46%, 95%CI 39-53 vs. 29%, 95%CI 21-37; P<.01). Overall, 85 of 198 patients (43%) were spared non-beneficial oesophagectomy, including those with sustained complete clinical response at last follow-up or distant metastases detected within six months after nCRT. Conclusions Overall survival at 5 years is not statistically significantly different between active surveillance and standard surgery in patients with oesophageal cancer and a complete clinical response after nCRT, despite a higher incidence of distant metastases in the active surveillance group. These findings support active surveillance as a feasible treatment strategy in these patients.

[Slide 1] Background The Surgery As Needed for Oesophageal cancer (SANO)-trial demonstrated that active surveillance is non-inferior to standard surgery in terms of two-year overall survival in patients with a complete clinical response after neoadjuvant chemoradiotherapy (nCRT). Active surveillance resulted in improved quality of life. Here, we report 5-year outcomes. Methods From 2017 to 2021, a multicentre stepped-wedge cluster-randomised non-inferiority trial was conducted. Patients with locally advanced oesophageal cancer who achieved complete clinical response (defined as no tumour detected on endoscopic biopsies, -ultrasound, or PET-CT) underwent active surveillance or standard surgery. Cox regression was used to compare 5-year overall and disease-free survival, and the cumulative incidence of distant metastases was estimated using competing risk analysis. Results A total of 309 patients were included, 198 underwent active surveillance and 111 underwent standard surgery. Median follow-up was 76 months (IQR 67–86). Five-year overall survival was not statistically significantly different between active surveillance and standard surgery (50%, 95%CI 34-57 vs. 54%, 95%CI 46-64; HR 1.17, 95%CI 0.77-1.77) with similar results across histologic subtypes. Disease-free survival was not statistically significantly different (HR 1.39, 95% CI 0.94-2.06). The cumulative incidence of distant metastases at five years was significantly higher in the active surveillance group compared to the standard surgery group (46%, 95%CI 39-53 vs. 29%, 95%CI 21-37; P<.01). Overall, 85 of 198 patients (43%) were spared non-beneficial oesophagectomy, including those with sustained complete clinical response at last follow-up or distant metastases detected within six months after nCRT. Conclusions Overall survival at 5 years is not statistically significantly different between active surveillance and standard surgery in patients with oesophageal cancer and a complete clinical response after nCRT, despite a higher incidence of distant metastases in the active surveillance group. These findings support active surveillance as a feasible treatment strategy in these patients.

[Slide 1] Methods Design • Phase III, stepped-wedge, cluster-randomised trial • Non-inferiority design Patients • Locally advanced oesophageal cancer • nCRT according to the CROSS regimen • cCR at 12 weeks after nCRT (CRE-2) [MADRID 2026 ESMO congress logo] [Slide 2] Cause-specific mortality Legend: Active surveillance (red); Surgery (blue); death by oesophageal cancer (solid); death by other cause (dashed) Y axis: Cumulative incidence (%); X axis: Time (months) 0-60 End-of-curve labels at 60 months: 40%, 30%, 15%, 10% death by oesophageal cancer: sHR 1.67, 95%CI 1.02-2.73, p=0.04 death by other cause: sHR 0.49, 95%CI 0.22-1.09, p=0.08 Number at risk Active surveillance: 198 178 150 120 106 98 Surgery: 111 97 78 70 63 60 [MADRID 2026 ESMO congress logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: Cause-specific mortality (cumulative incidence; competing risks) Active surveillance | 198 Surgery | 111 HR: sHR 1.67 (death by oesophageal cancer); sHR 0.49 (death by other cause) HR CI: 1.02-2.73; 0.22-1.09 P: 0.04; 0.08 {'timepoint': '60 mo', 'arm': 'Active surveillance - death by oesophageal cancer', 'value': '40%', 'ci': None} {'timepoint': '60 mo', 'arm': 'Surgery - death by oesophageal cancer', 'value': '30%', 'ci': None} {'timepoint': '60 mo', 'arm': 'Surgery - death by other cause', 'value': '15%', 'ci': None} {'timepoint': '60 mo', 'arm': 'Active surveillance - death by other cause', 'value': '10%', 'ci': None} Note: End-of-curve percentages printed at 60 months; arm/line attribution read from curve colour and line style (solid red 40%, solid blue 30%, dashed blue 15%, dashed red 10%). n from number at risk at time 0. [Slide 3] Distant dissemination rate Legend: Active surveillance; Surgery Y axis: Cumulative incidence (%); X axis: Time (months) 0-60 End-of-curve labels at 60 months: 46% (active surveillance), 29% (surgery) sHR 2.04, 95%CI 1.25-3.34, p < 0.01 Number at risk Active surveillance: 198 156 123 98 94 85 Surgery: 111 88 69 63 60 56 [MADRID 2026 ESMO congress logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: Distant dissemination (cumulative incidence) Active surveillance | 198 Surgery | 111 HR: sHR 2.04 HR CI: 1.25-3.34 P: < 0.01 {'timepoint': '60 mo', 'arm': 'Active surveillance', 'value': '46%', 'ci': None} {'timepoint': '60 mo', 'arm': 'Surgery', 'value': '29%', 'ci': None} Note: End-of-curve percentages printed at 60 months. n from number at risk at time 0. [Slide 4] Conclusions • 5-year overall survival was not statistically significantly different • Distant metastases were significantly higher with active surveillance • 43% of patients avoided oesophagectomy • These findings support the concept of active surveillance after nCRT [MADRID 2026 ESMO congress logo]

[Slide 1] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH Conclusions • 5-year overall survival was not statistically significantly different • Distant metastases were significantly higher with active surveillance • 43% of patients avoided oesophagectomy • These findings support the concept of active surveillance after nCRT [MADRID 2026 ESMO congress logo] Sanjiv Gangaram-Panday Active surveillance versus standard surgery after neoadjuvant chemoradiotherapy for oesophageal cancer: Long-term results of a cluster-randomised controlled trial (SANO) ALMERIA AUDITORIUM - HALL 9 [Livestream screen capture with presenter video inset]

[Slide 1] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH Disease-free survival Legend: Active surveillance / Surgery HR 1.39, 95% CI 0.94-2.06, p = 0.10 Y-axis: Disease-free survival (%); X-axis: Time (months) 0-60 Number at risk (number censored) (months 0, 12, 24, 36, 48, 60) Active surveillance: 198 (0) 156 (0) 118 (0) 97 (1) 91 (1) 82 (1) Surgery: 111 (0) 87 (0) 67 (0) 62 (0) 59 (0) 55 (0) [ESMO Congress Madrid 2026 logo] Sanjiv Gangaram-Panday Active surveillance versus standard surgery after neoadjuvant chemoradiotherapy for oesophageal cancer: Long-term results of a cluster-randomised controlled trial (SANO) ALMERIA AUDITORIUM - HALL 9 [Survival data captured as a table — no point estimates read from the curve] Endpoint: Disease-free survival Active surveillance | 198 Surgery | 111 HR: 1.39 HR CI: 95% CI 0.94-2.06 P: 0.10 Note: No medians or landmark rates printed. [Slide 2] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH Distant dissemination rate Legend: Active surveillance / Surgery sHR 2.04, 95%CI 1.25-3.34, p < 0.01 Y-axis: Cumulative incidence (%); X-axis: Time (months) 0-60 Curve end labels: 46% (active surveillance), 29% (surgery) Number at risk (months 0, 12, 24, 36, 48, 60) Active surveillance: 198 156 123 98 94 85 Surgery: 111 88 69 63 60 56 [ESMO Congress Madrid 2026 logo] Sanjiv Gangaram-Panday Active surveillance versus standard surgery after neoadjuvant chemoradiotherapy for oesophageal cancer: Long-term results of a cluster-randomised controlled trial (SANO) ALMERIA AUDITORIUM - HALL 9 [Survival data captured as a table — no point estimates read from the curve] Endpoint: Distant dissemination (cumulative incidence) Active surveillance | 198 Surgery | 111 HR: sHR 2.04 HR CI: 95% CI 1.25-3.34 P: <0.01 {'timepoint': '60 mo (curve end label)', 'arm': 'Active surveillance', 'value': '46%', 'ci': None} {'timepoint': '60 mo (curve end label)', 'arm': 'Surgery', 'value': '29%', 'ci': None} Note: Cumulative incidence (competing-risk subdistribution HR). [Slide 3] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH Cause-specific mortality Legend: Active surveillance / Surgery / — death by oesophageal cancer / - - death by other cause death by oesophageal cancer: sHR 1.67, 95%CI 1.02-2.73, p=0.04 death by other cause: sHR 0.49, 95%CI 0.22-1.09, p=0.08 Y-axis: Cumulative incidence (%); X-axis: Time (months) 0-60 Curve end labels: 40%, 30% (solid lines, death by oesophageal cancer); 15%, 10% (dashed lines, death by other cause) Number at risk (months 0, 12, 24, 36, 48, 60) Active surveillance: 198 178 150 120 106 98 Surgery: 111 97 78 70 63 60 [ESMO Congress Madrid 2026 logo] Sanjiv Gangaram-Panday Active surveillance versus standard surgery after neoadjuvant chemoradiotherapy for oesophageal cancer: Long-term results of a cluster-randomised controlled trial (SANO) ALMERIA AUDITORIUM - HALL 9 [Survival data captured as a table — no point estimates read from the curve] Endpoint: Cause-specific mortality (cumulative incidence) Active surveillance | 198 Surgery | 111 HR: death by oesophageal cancer sHR 1.67; death by other cause sHR 0.49 HR CI: 1.02-2.73; 0.22-1.09 P: 0.04; 0.08 {'timepoint': '60 mo (curve end label), death by oesophageal cancer', 'arm': 'Active surveillance', 'value': '40%', 'ci': None} {'timepoint': '60 mo (curve end label), death by oesophageal cancer', 'arm': 'Surgery', 'value': '30%', 'ci': None} {'timepoint': '60 mo (curve end label), death by other cause', 'arm': 'Surgery', 'value': '15%', 'ci': None} {'timepoint': '60 mo (curve end label), death by other cause', 'arm': 'Active surveillance', 'value': '10%', 'ci': None} Note: Arm assignment of end labels inferred from line color (red=active surveillance, blue=surgery); 15% dashed line appears blue, 10% dashed appears red. [Slide 4] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH Complete clinical responders Active surveillance (n=198): • 16% (n=32) developed distant metastases (11% (n=22) within 6 months post-nCRT) • 51% (n=101) developed locoregional regrowth • 1% (n=1) developed second primary oesophageal tumour • 32% (n=64) maintained CCR Hypothetical 43% avoided oesophagectomy [ESMO Congress Madrid 2026 logo] Sanjiv Gangaram-Panday Active surveillance versus standard surgery after neoadjuvant chemoradiotherapy for oesophageal cancer: Long-term results of a cluster-randomised controlled trial (SANO) ALMERIA AUDITORIUM - HALL 9

[Slide 1] Overall survival Legend: Active surveillance; Surgery HR 1.17, 95% CI 0.77-1.77, p = 0.46 5-year RMST: -0.32 months 95% CI -6.29 to 5.65, p = 0.92 Overall survival (%); Time (months) Number at risk (number censored) (0, 12, 24, 36, 48, 60): Active surveillance: 198 (0), 178 (0), 150 (0), 120 (1), 106 (1), 98 (1) Surgery: 111 (0), 97 (0), 78 (0), 70 (0), 63 (0), 60 (0) [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival Active surveillance | 198 Surgery | 111 HR: 1.17 HR CI: 0.77-1.77 P: 0.46 Note: 5-year RMST difference -0.32 months (95% CI -6.29 to 5.65), p = 0.92. No medians or landmark rates printed. [Slide 2] Disease-free survival Legend: Active surveillance; Surgery HR 1.39, 95% CI 0.94-2.06, p = 0.10 Disease-free survival (%); Time (months) Number at risk (number censored) (0, 12, 24, 36, 48, 60): Active surveillance: 198 (0), 156 (0), 118 (0), 97 (1), 91 (1), 82 (1) Surgery: 111 (0), 87 (0), 67 (0), 62 (0), 59 (0), 55 (0) [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: Disease-free survival Active surveillance | 198 Surgery | 111 HR: 1.39 HR CI: 0.94-2.06 P: 0.10 Note: No medians or landmark rates printed. [Slide 3] Complete clinical responders Active surveillance (n=198): • 16% (n=32) developed distant metastases (11% (n=22) within 6 months post-nCRT) • 51% (n=101) developed locoregional regrowth • 1% (n=1) developed second primary oesophageal tumour • 32% (n=64) maintained CCR Hypothetical 43% avoided oesophagectomy [ESMO Congress Madrid 2026 logo]

[Slide 1] Disease-free survival Legend: Active surveillance — Surgery HR 1.39, 95% CI 0.94-2.06, p = 0.10 Y-axis: Disease-free survival (%) X-axis: Time (months) Number at risk (number censored) Active surveillance (red): 198 (0) | 156 (0) | 118 (0) | 97 (1) | 91 (1) | 82 (1) Surgery (blue): 111 (0) | 87 (0) | 87 (0) | 62 (0) | 59 (0) | 55 (0) (time points 0, 12, 24, 36, 48, 60 months) [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: Disease-free survival Active surveillance | 198 Surgery | 111 HR: 1.39 HR CI: 0.94-2.06 P: 0.10 Note: No medians or landmark rates printed. Number-at-risk values partly blurry (e.g. surgery 24-mo '87' could be '67'); transcribed as seen. [Slide 2] Distant dissemination rate Legend: Active surveillance — Surgery sHR 2.04, 95%CI 1.25-3.34, p < 0.01 Y-axis: Cumulative incidence (%) X-axis: Time (months) End-of-curve labels: Active surveillance 4[illegible]% ; Surgery 2[illegible]% Number at risk Active surveillance (red): 198 | 156 | 123 | 98 | 94 | 85 Surgery (blue): 111 | 88 | 69 | 63 | 60 | 56 (time points 0, 12, 24, 36, 48, 60 months) [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: Distant dissemination rate (cumulative incidence) Active surveillance | 198 Surgery | 111 HR: sHR 2.04 HR CI: 1.25-3.34 P: <0.01 Note: 60-month labels printed at curve ends but second digit illegible (approx '4?%' and '2?%'); not filled. At-risk row digits blurry (best reading). [Slide 3] Cause-specific mortality Legend: Active surveillance — Surgery — death by oesophageal cancer --- death by other cause death by oesophageal cancer: sHR 1.67, 95%CI 1.02-2.73, p=0.04 death by other cause: sHR 0.49, 95%CI 0.22-1.09, p=0.08 Y-axis: Cumulative incidence (%) X-axis: Time (months) End-of-curve labels: 40% ; 30% ; 15% ; 10% Number at risk Active surveillance (red): 198 | 178 | 150 | 120 | 108 | 98 Surgery (blue): 111 | 97 | 78 | 70 | 63 | 60 (time points 0, 12, 24, 36, 48, 60 months) [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: Cause-specific mortality (cumulative incidence; death by oesophageal cancer and by other cause) Active surveillance | 198 Surgery | 111 HR: death by oesophageal cancer: sHR 1.67; death by other cause: sHR 0.49 HR CI: 1.02-2.73; 0.22-1.09 P: 0.04; 0.08 Note: End-of-curve labels 40%/30%/15%/10% printed but arm-to-label mapping is not explicit (and labels sit above the drawn curve levels), so not entered as landmarks.

[Slide 1] Proffered Paper session LBA78 - Active surveillance versus standard surgery after neoadjuvant chemoradiotherapy for oesophageal cancer: Long-term results of a cluster-randomised controlled trial (SANO) Speakers Sanjiv S. Gangaram-Panday (Rotterdam, Netherlands) Lecture Time 13:40 - 13:50 Abstract Background The Surgery As Needed for Oesophageal cancer (SANO)-trial demonstrated that active surveillance is non-inferior to standard surgery in terms of two-year overall survival in patients with a complete clinical response after neoadjuvant chemoradiotherapy (nCRT). Active surveillance resulted in improved quality of life. Here, we report 5-year outcomes. Methods From 2017 to 2021, a multicentre stepped-wedge cluster-randomised non-inferiority trial was conducted. Patients with locally advanced oesophageal cancer who achieved complete clinical response (defined as no tumour detected on endoscopic biopsies, -ultrasound, or PET-CT) underwent active surveillance or standard surgery. Cox regression was used to compare 5-year overall and disease-free survival, and the cumulative incidence of distant metastases was estimated using competing risk analysis.

[Slide 1] [KM figure with shaded confidence bands] Legend: Active surveillance | Surgery Y-axis: Overall survival (%) ; X-axis: Time (months) 0, 12, 24, 36, 48, 60 Number at risk (number censored): Active surveillance: 198 (0), 178 (0), 150 (0), 120 (1), 106 (1), 98 (1) Surgery: 111 (0), 97 (0), 78 (0), 70 (0), 63 (0), 60 (0) [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival Active surveillance | 198 Surgery | 111 Note: No HR, median, or landmark values printed on figure; n taken from number-at-risk at time 0. Number censored shown in parentheses.

[Slide 1] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH [RASolute 302 logo] RASolute 302 Study Design Key eligibility criteria: - Adults with mPDAC - One prior fluoropyrimidine- or gemcitabine-based regimen in the metastatic setting - ECOG PS 0-1 - Documented tumor RAS mutational status^a by local testing R 1:1 -> Daraxonrasib 300 mg, PO QD Investigator's choice of chemotherapy (1 of 4): GnP, mFOLFIRINOX, NAL-IRI + 5-FU/LV, FOLFOX Dual primary endpoints - OS and PFS (per BICR)^b (RAS G12 population) Key secondary endpoints - OS and PFS (per BICR)^b (overall population^c) - ORR^b (per BICR) (RAS G12 and overall populations) - TTD in PROs^d (RAS G12 and overall populations) Stratification factors: - ECOG PS (0 vs 1) - Metastatic disease at diagnosis (yes vs no) - Liver metastases at baseline (yes vs no) - Tumor RAS mutational status (RAS G12D/V vs other RAS G12 vs RAS G13 or Q61 mutation or no RAS mutation identified) Global enrollment: 59 sites across 6 countries NCT06625320. ^aEither mutant, defined as nonsynonymous mutations in KRAS, NRAS, or HRAS at codons 12, 13, or 61 (G12, G13, or Q61), or no RAS mutation identified. ^bPer RECIST 1.1. ^cIncluding patients whose tumors harbor RAS G12, G13, or Q61 mutations, or had no RAS mutation identified. ^dTTD in symptom of pain and in global health status/quality of life. Presented by Zev A. Wainberg, MD, MSc [Speaker panel] Zev Wainberg - Daraxonrasib, a RAS(ON) multi-selective inhibitor vs chemotherapy in previously treated metastatic pancreatic adenocarcinoma (mPDAC): Additional subgroup analyses from the ph3 RASolute 302 study ALMERIA AUDITORIUM - HALL 9 [ESMO Congress Madrid 2026 logo] [Slide 2] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH [RASolute 302 logo] RASolute 302 Primary Analysis Results - In the Phase 3 RASolute 302 study in patients with previously treated mPDAC, daraxonrasib demonstrated statistically significant and clinically meaningful improvements in OS, PFS, ORR, and quality of life vs SOC chemotherapy in both the RAS G12 and overall populations^1 - Daraxonrasib is approved in the US for patients with mPDAC who have received at least one prior systemic therapy or who are not candidates for multiagent systemic therapy^2 - Daraxonrasib is recommended in the NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines®)^3 - Daraxonrasib also received a positive CHMP opinion in the EU for the treatment of patients with mPDAC who have received at least one prior systemic therapy^4 OS in the Overall Population | | Daraxonrasib (n=248) | Chemo (n=252) | No. of events, n (%) | 79 (32) | 141 (56) mOS, months (95% CI) | 13.2 (10.0-NE) | 6.7 (5.8-8.0) HR (95% CI)^a | 0.40 (0.30-0.53) P-value^b | p=4.6×10^-11 Curve labels at 12 months: 53.2% (daraxonrasib), 17.3% (chemo) No. at risk (Months 0-16) Daraxonrasib: 248 240 237 225 208 188 165 125 90 62 37 24 11 7 4 1 0 Chemo: 252 235 217 193 161 126 108 78 53 35 21 10 5 2 0 ^aHRs and 95% CIs were based on the stratified Cox model with Efron's method of tie handling. ^bP-values were calculated using the stratified log-rank test. 1. O'Reilly EM, et al. N Engl J Med. 2026;395(4):325-337; 2. FDA Press Release. August 26, 2026. https://www.fda.gov/drugs/resources-information-approved-drugs/fda-approves-daraxonrasib-metastatic-pancreatic-adenocarcinoma. Accessed August 27, 2026; 3. Referenced with permission from the NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines®) for Pancreatic Adenocarcinoma V.4.2026. © National Comprehensive Cancer Network, Inc. 2026. All rights reserved. Accessed October 5, 2026. To view the most recent and complete version of the guideline, go online to NCCN.org. NCCN makes no warranties of any kind whatsoever regarding their content, use or application and disclaims any responsibility for their application or use in any way. 4. European Medicines Agency. October 15, 2026. CHMP Summary of Positive Opinion for Rasonque (Daraxonrasib). CHMP Ref No. EMADOC-1829012207-72049. Presented by Zev A. Wainberg, MD, MSc [Speaker panel] Zev Wainberg - Daraxonrasib, a RAS(ON) multi-selective inhibitor vs chemotherapy in previously treated metastatic pancreatic adenocarcinoma (mPDAC): Additional subgroup analyses from the ph3 RASolute 302 study ALMERIA AUDITORIUM - HALL 9 [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS, overall population Daraxonrasib | 248 | 13.2 months | 10.0-NE Chemo (SOC) | 252 | 6.7 months | 5.8-8.0 HR: 0.40 HR CI: 0.30-0.53 P: 4.6×10^-11 {'timepoint': '12 months', 'arm': 'Daraxonrasib', 'value': '53.2%', 'ci': None} {'timepoint': '12 months', 'arm': 'Chemo', 'value': '17.3%', 'ci': None} Note: Landmark labels printed on curve at the 12-month dashed line. Stratified Cox HR; stratified log-rank p. Primary analysis per O'Reilly NEJM 2026. [Slide 3] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH [RASolute 302 logo] PFS by RAS Genotype Subgroups PFS by BICR | PFS by investigator Tumor RAS mutational status^a | Events/N Daraxonrasib | Chemo | PFS HR (95% CI)^c || Events/N Daraxonrasib | Chemo | PFS HR (95% CI)^c Overall | 127/248 | 130/252 | 0.55 (0.43-0.70) || 135/248 | 149/252 | 0.48 (0.38-0.61) RAS G12D | 53/95 | 74/121 | 0.48 (0.33-0.69) || 53/95 | 79/121 | 0.43 (0.30-0.62) RAS G12V | 39/102 | 29/76 | 0.54 (0.33-0.87) || 46/102 | 36/76 | 0.51 (0.33-0.80) RAS G12R | 19/30 | 16/31 | 0.80 (0.41-1.56) || 22/30 | 17/31 | 0.73 (0.38-1.39) Other RAS mutation^b | 9/11 | 7/14 | 1.29 (0.47-3.51) || 7/11 | 9/14 | 0.57 (0.21-1.56) RAS Q61 | 9/11 | 5/11 | 1.55 (0.51-4.69) || 7/11 | 7/11 | 0.63 (0.22-1.83) No RAS mutation identified | 6/9 | 2/7 | 1.92 (0.39-9.52) || 6/9 | 6/7 | 0.62 (0.20-1.94) [axis 0.05 0.5 1.0 5 10; Favors Daraxonrasib | Favors Chemo] PFS results generally favored daraxonrasib versus SOC chemotherapy across RAS genotype subgroups, although treatment-effect estimates differed between BICR and investigator assessments, particularly in smaller subgroups Data cutoff: 10 Feb 2026. ^aLocally determined. Only subgroups with ≥5 patients in at least one treatment arm were included. ^bOther RAS mutations include RAS Q61 and RAS G13. ^cHRs and 95% CIs overall and for subgroups were based on the unstratified Cox model with Efron's method of tie handling. Presented by Zev A. Wainberg, MD, MSc [Speaker panel] Zev Wainberg - Daraxonrasib, a RAS(ON) multi-selective inhibitor vs chemotherapy in previously treated metastatic pancreatic adenocarcinoma (mPDAC): Additional subgroup analyses from the ph3 RASolute 302 study ALMERIA AUDITORIUM - HALL 9 [Slide 4] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH [RASolute 302 logo] OS by Preselected SOC Chemotherapy Regimen^a GnP: Daraxonrasib (n=140) | Chemo (n=141); HR (95% CI)^c 0.45 (0.32-0.65) mFOLFIRINOX/FOLFOX^b: Daraxonrasib (n=30) | Chemo (n=30); HR (95% CI)^c 0.46 (0.23-0.95) NAL-IRI + 5-FU/LV: Daraxonrasib (n=78) | Chemo (n=81); HR (95% CI)^c 0.32 (0.18-0.58) No. at risk (Months 0-16) GnP Daraxonrasib: 140 136 136 127 116 110 94 72 50 32 17 12 3 2 0 GnP Chemo: 141 131 121 110 88 69 59 43 27 19 9 5 3 2 0 mFOLFIRINOX/FOLFOX Daraxonrasib: 30 27 26 25 22 21 20 15 13 8 5 4 3 2 2 0 mFOLFIRINOX/FOLFOX Chemo: 30 28 25 22 18 14 13 8 6 5 4 3 2 0 NAL-IRI + 5-FU/LV Daraxonrasib: 78 77 75 73 70 57 51 38 27 22 15 8 5 3 2 1 0 NAL-IRI + 5-FU/LV Chemo: 81 76 71 61 55 43 36 27 20 11 8 2 0 Data cutoff: 10 Feb 2026. Median (range) follow-up time was 8.5 (3.2-15.9) months. ^aInvestigator's choice of chemotherapy was determined prior to randomization. ^bmFOLFIRINOX and FOLFOX regimens were pooled due to small individual sample sizes. ^cHRs and 95% CIs were based on the unstratified Cox model with Efron's method of tie handling. Presented by Zev A. Wainberg, MD, MSc [Speaker panel] Zev Wainberg - Daraxonrasib, a RAS(ON) multi-selective inhibitor vs chemotherapy in previously treated metastatic pancreatic adenocarcinoma (mPDAC): Additional subgroup analyses from the ph3 RASolute 302 study ALMERIA AUDITORIUM - HALL 9 [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS by preselected SOC chemotherapy regimen GnP cohort - Daraxonrasib | 140 GnP cohort - Chemo | 141 mFOLFIRINOX/FOLFOX cohort - Daraxonrasib | 30 mFOLFIRINOX/FOLFOX cohort - Chemo | 30 NAL-IRI + 5-FU/LV cohort - Daraxonrasib | 78 NAL-IRI + 5-FU/LV cohort - Chemo | 81 HR: GnP 0.45; mFOLFIRINOX/FOLFOX 0.46; NAL-IRI + 5-FU/LV 0.32 HR CI: GnP 0.32-0.65; mFOLFIRINOX/FOLFOX 0.23-0.95; NAL-IRI + 5-FU/LV 0.18-0.58 Note: Three subgroup curves; no medians or p-values printed. Unstratified Cox HRs. Number-at-risk rows small; minor digit uncertainty possible. DCO 10 Feb 2026; median follow-up 8.5 (3.2-15.9) mo.

[Slide 1] Proffered Paper session LBA82 - Daraxonrasib, a RAS(ON) multi-selective inhibitor vs chemotherapy in previously treated metastatic pancreatic adenocarcinoma (mPDAC): Additional subgroup analyses from the ph3 RASolute 302 study Speakers Zev A. Wainberg (Los Angeles, United States of America) Lecture Time 14:22 - 14:32 Abstract Background Multiagent systemic chemo offers limited benefit in mPDAC. Aberrant RAS(ON) signaling drives PDAC; >90% of cases harbor mutant (mut) RAS, and RAS WT PDAC remains dependent on RAS pathway signaling. Daraxonrasib is an oral, RAS(ON) multi-selective, tri-complex inhibitor of active, GTP-bound mut and WT RAS. Daraxonrasib significantly improved OS and PFS vs SOC chemo in pts with previously treated mPDAC overall (with or without an identified tumor RAS mut; mOS 13.2 vs 6.7 mo; HR, 0.40; mPFS 7.2 vs 3.6 mo; HR, 0.49) and in pts with RAS G12 mut mPDAC (OS HR, 0.40; PFS HR, 0.45) with manageable safety in the RASolute 302 primary analysis (NCT06625320). We report additional subgroup analyses from RASolute 302. [Slide 2] OS in subgroups | | n (%) Daraxonrasib n=248 | Chemo n=252 | OS HR (95% CI) | Tumor RAS mut status | G12D | 95 (38) | 121 (48) | 0.49 (0.33-0.74) | | G12V | 102 (41) | 76 (30) | 0.32 (0.19-0.54) | | G12R | 30 (12) | 31 (12) | 0.61 (0.29-1.30) | | RAS Q61 | 11 (4) | 11 (4) | 0.37 (0.09-1.44) | | No RAS mut identified | 9 (4) | 7 (3) | 0.43 (0.09-1.98) | Conclusions Daraxonrasib demonstrated OS benefit vs chemo across clinically relevant subgroups including pts with or without an identified tumor RAS mut. Directionally consistent OS benefit across these subgroups further reinforces the potential for daraxonrasib to be a new SOC for eligible pts with mPDAC. [Slide 3] Methods Pts with previously treated mPDAC were randomized 1:1 to daraxonrasib 300 mg PO QD or investigator's choice of SOC chemo (GnP, mFOLFIRINOX, Nal-IRI+5-FU/LV, or FOLFOX). Dual primary endpoints (EPs) were OS and PFS in the RAS G12 mut population; key secondary EPs included OS and PFS in the overall population. Additional analyses evaluated OS in clinically relevant subgroups by locally determined tumor RAS mut status, preselected chemo regimen, and disease setting of prior treatment. Results At data cutoff (10 Feb 2026), median follow-up was 8.5 mo (range, 3.2-15.9). OS benefit with daraxonrasib vs chemo was observed across all RAS mut subgroups (Table). Daraxonrasib showed a consistent OS treatment effect across preselected chemo subgroups (HRs, 0.32-0.46). OS data by disease setting of prior treatment will also be presented. Table: LBA82

[Slide 2] ORR by RAS Genotype Subgroups [RASolute 302 logo] Tumor RAS mutational status(b) | ORR by BICR (PR+CR/n(a)), % Daraxonrasib (n=248) | Chemo (n=252) | ORR by investigator (PR+CR/n(a)), % Daraxonrasib (n=248) | Chemo (n=252) Overall | 32% (75/237) | 11% (27/241) | 24% (60/248) | 8% (21/252) RAS G12D | 24% (23/95) | 9% (11/116) | 18% (17/95) | 6% (7/121) RAS G12V | 50% (47/94) | 15% (11/73) | 37% (38/102) | 14% (11/76) RAS G12R | 7% (2/27) | 14% (4/29) | 10% (3/30) | 3% (1/31) Other RAS mutation(c) | 18% (2/11) | 0% (0/14) | 18% (2/11) | 7% (1/14) RAS Q61 | 18% (2/11) | 0% (0/11) | 18% (2/11) | 9% (1/11) No RAS mutation identified | 11% (1/9) | 17% (1/6) | 0% (0/9) | 14% (1/7) Daraxonrasib generally showed ORR improvements compared with SOC chemotherapy across RAS genotype subgroups by both BICR and investigator assessments, with differences seen between assessments Data cutoff: 10 Feb 2026. (a)Patients with measurable disease at baseline were included. (b)Locally determined. Only subgroups with ≥5 patients in at least one treatment arm were included. (c)Other RAS mutations include RAS Q61 and RAS G13. Presented by Zev A. Wainberg, MD, MSc [ESMO Congress Madrid 2026 logo] [Slide 3] OS by Preselected SOC Chemotherapy Regimen(a) [RASolute 302 logo] GnP: Daraxonrasib (n=140) | Chemo (n=141) — HR (95% CI)(c) 0.45 (0.32–0.65) mFOLFIRINOX/FOLFOX(b): Daraxonrasib (n=30) | Chemo (n=30) — HR (95% CI)(c) 0.46 (0.23–0.95) NAL-IRI + 5-FU/LV: Daraxonrasib (n=78) | Chemo (n=81) — HR (95% CI)(c) 0.32 (0.18–0.58) [Three KM panels; y-axis: OS Probability (%); x-axis: Months 0–16] No. at risk (low resolution, monthly 0 onward): GnP — Daraxonrasib: 140 136 136 127 116 110 94 72 50 32 17 12 3 2 0; Chemo: 141 131 121 110 88 69 59 43 27 19 9 5 3 2 0 mFOLFIRINOX/FOLFOX — Daraxonrasib: 30 27 26 25 22 21 20 15 13 8 5 4 3 2 2 0; Chemo: 30 28 25 22 18 14 13 8 6 5 4 3 2 0 NAL-IRI + 5-FU/LV — Daraxonrasib: 78 77 75 73 70 67 51 38 27 22 15 8 5 3 2 1 0; Chemo: 81 76 71 61 55 43 36 27 20 11 8 2 0 Data cutoff: 10 Feb 2026. Median (range) follow-up time was 8.5 (3.2–15.9) months. (a)Investigator's choice of chemotherapy was determined prior to randomization. (b)mFOLFIRINOX and FOLFOX regimens were pooled due to small individual sample sizes. (c)HRs and 95% CIs were based on the unstratified Cox model with Efron's method of tie handling. Presented by Zev A. Wainberg, MD, MSc [ESMO logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival by preselected SOC chemotherapy regimen (subgroups) GnP subgroup: Daraxonrasib | 140 GnP subgroup: Chemo | 141 mFOLFIRINOX/FOLFOX subgroup: Daraxonrasib | 30 mFOLFIRINOX/FOLFOX subgroup: Chemo | 30 NAL-IRI + 5-FU/LV subgroup: Daraxonrasib | 78 NAL-IRI + 5-FU/LV subgroup: Chemo | 81 HR: GnP 0.45; mFOLFIRINOX/FOLFOX 0.46; NAL-IRI + 5-FU/LV 0.32 HR CI: 0.32–0.65; 0.23–0.95; 0.18–0.58 Note: No medians, p-values or landmarks printed. Unstratified Cox HRs. Number-at-risk rows are small/low-resolution — verify before citing. [Slide 4] OS by Disease Setting of Prior Treatment [RASolute 302 logo] Metastatic disease <6 months after neoadjuvant/adjuvant therapy(a): Daraxonrasib (n=42) | Chemo (n=34) — HR (95% CI)(b) 0.36 (0.17–0.75) One prior line of therapy for metastatic disease: Daraxonrasib (n=206) | Chemo (n=218) — HR (95% CI)(b) 0.43 (0.32–0.58) [Two KM panels; y-axis: OS Probability (%); x-axis: Months 0–16] No. at risk (monthly from 0): Metastatic <6 mo — Daraxonrasib: 42 41 41 37 37 36 32 27 22 16 9 6 2 2 2 0; Chemo: 34 31 27 24 22 18 17 15 9 4 1 1 1 0 One prior line — Daraxonrasib: 206 199 196 188 171 152 133 98 68 46 28 18 9 5 2 1 0; Chemo: 218 204 190 169 139 108 91 63 44 31 20 9 4 2 0 Data cutoff: 10 Feb 2026. Median (range) follow-up time was 8.5 (3.2–15.9) months. (a)Patients received prior neoadjuvant or adjuvant treatment but were diagnosed with metastatic disease <6 months from the last dose. Under the study criteria, this was considered first-line treatment in the metastatic setting. (b)HRs and 95% CIs were based on the unstratified Cox model with Efron's method of tie handling. Presented by Zev A. Wainberg, MD, MSc [ESMO logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival by disease setting of prior treatment (subgroups) Metastatic <6 mo after (neo)adjuvant: Daraxonrasib | 42 Metastatic <6 mo after (neo)adjuvant: Chemo | 34 One prior metastatic line: Daraxonrasib | 206 One prior metastatic line: Chemo | 218 HR: Metastatic <6 mo 0.36; One prior line 0.43 HR CI: 0.17–0.75; 0.32–0.58 Note: No medians, p-values or landmarks printed. Unstratified Cox HRs.

[Slide 1] [RASolute 302 logo] RAS Genotype Distribution • Consistent with the expected distribution of RAS mutations in PDAC, RAS G12 variants predominated; G13, Q61, and no RAS mutation identified were uncommon RAS genotypes in PDAC1 Reported real-world prevalence KRAS G12D (39.9%) KRAS G12V (28.6%) KRAS G12R (14.6%) KRAS G12C (1.7%) KRAS Q61 (6.3%) KRAS G13 (0.8%) NRAS (0.2%) RAS WT (8%) RAS genotypes in RASolute 302a RAS G12D (43.2%) RAS G12V (35.6%) RAS G12R (12.2%) Other RAS G12c (0.8%) RAS Q61 (4.4%) RAS G13 (0.6%) No RAS mutation identifiedb (3.2%) aLocally determined at a data cutoff of 10 Feb 2026. bPatients with no identified tumor RAS mutation but with other known driver mutations (eg, BRAF V600E, MSI-H/dMMR, NTRK gene fusion) for which there is local regulatory approval must have received appropriate targeted therapy. cOther RAS G12 includes KRAS G12N, KRAS G12A, KRAS G12C, and KRAS G12L mutations. 1. Pant S, et al. J Clin Oncol. 2025;43:777. Presented by Zev A. Wainberg, MD, MSc [ESMO Congress logo] [Slide 3] [RASolute 302 logo] PFS by RAS Genotype Subgroups PFS by BICR | PFS by investigator Tumor RAS mutational statusa | Events/N Daraxonrasib | Chemo | PFS HR (95% CI)c | Events/N Daraxonrasib | Chemo | PFS HR (95% CI)c Overall | 127/248 | 130/252 | 0.55 (0.43–0.70) | 135/248 | 149/252 | 0.48 (0.38–0.61) RAS G12D | 53/95 | 74/121 | 0.48 (0.33–0.69) | 53/95 | 79/121 | 0.43 (0.30–0.62) RAS G12V | 39/102 | 29/76 | 0.54 (0.33–0.87) | 46/102 | 36/76 | 0.51 (0.33–0.80) RAS G12R | 19/30 | 16/31 | 0.80 (0.41–1.56) | 22/30 | 17/31 | 0.73 (0.38–1.39) Other RAS mutationb | 9/11 | 7/14 | 1.29 (0.47–3.51) | 7/11 | 9/14 | 0.57 (0.21–1.56) RAS Q61 | 9/11 | 5/11 | 1.55 (0.51–4.69) | 7/11 | 7/11 | 0.63 (0.22–1.83) No RAS mutation identified | 6/9 | 2/7 | 1.92 (0.39–9.52) | 6/9 | 6/7 | 0.62 (0.20–1.94) [forest plot axis 0.05 0.5 1.0 5 10; Favors Daraxonrasib | Favors Chemo] PFS results generally favored daraxonrasib versus SOC chemotherapy across RAS genotype subgroups, although treatment-effect estimates differed between BICR and investigator assessments, particularly in smaller subgroups Data cutoff: 10 Feb 2026. aLocally determined. Only subgroups with ≥5 patients in at least one treatment arm were included. bOther RAS mutations include RAS Q61 and RAS G13. cHRs and 95% CIs overall and for subgroups were based on the unstratified Cox model with Efron's method of tie handling. Presented by Zev A. Wainberg, MD, MSc [ESMO Congress logo] [Slide 4] [RASolute 302 logo] Conclusions • In the Phase 3 RASolute 302 study, daraxonrasib demonstrated significant improvement in OS in the overall patient population (mOS: 13.2 months vs 6.7 months; HR, 0.40 [95% CI: 0.30–0.53]) • OS improvement was observed across all RAS genotype subgroups, regardless of identified tumor RAS mutational status, as expected based on daraxonrasib's mechanism of action • ORR and PFS (by BICR and investigator) generally favored daraxonrasib across subgroups • OS improvement with daraxonrasib was also maintained across preselected standard of care chemotherapy regimens or prior treatment settings, including early metastatic recurrence following perioperative therapy Consistent OS improvement across key patient subgroups reinforces daraxonrasib as a new standard of care for previously treated mPDAC Presented by Zev A. Wainberg, MD, MSc [ESMO Congress logo]

[Slide 3] [RASolute 302 logo] OS by RAS Genotype Subgroups Events/N Tumor RAS mutational status(a) | Daraxonrasib | Chemo | OS HR (95% CI)(c) Overall | 79/248 | 141/252 | 0.42 (0.32-0.55) RAS G12D | 37/95 | 71/121 | 0.49 (0.33-0.74) RAS G12V | 22/102 | 39/76 | 0.32 (0.19-0.54) RAS G12R | 12/30 | 16/31 | 0.61 (0.29-1.30) Other RAS mutation(b) | 3/11 | 10/14 | 0.29 (0.08-1.06) RAS Q61 | 3/11 | 7/11 | 0.37 (0.09-1.44) No RAS mutation identified | 4/9 | 4/7 | 0.43 (0.09-1.98) [Forest plot axis: 0.05, 0.5, 1.0, 2.0; Favors Daraxonrasib <- | -> Favors Chemo] Daraxonrasib showed OS improvement compared with SOC chemotherapy across all RAS genotype subgroups Data cutoff: 10 Feb 2026. (a)Locally determined. (b)Other RAS mutations include RAS Q61 and RAS G13. (c)The size of diamonds is relative to the size of the population in each subgroup. HRs and 95% CIs overall and for subgroups were based on the unstratified Cox model with Efron's method of tie handling. Presented by Zev A. Wainberg, MD, MSc [ESMO Congress Madrid 2026 logo]

[Slide 2] [RASolute 302 logo] [phone UI overlay: "1 de 1"] OS by RAS Genotype Subgroups Tumor RAS mutational status^a | Events/N Daraxonrasib | Chemo | OS HR (95% CI)^c Overall | 79/248 | 141/252 | 0.42 (0.32–0.55) RAS G12D | 37/95 | 71/121 | 0.49 (0.33–0.74) RAS G12V | 22/102 | 39/76 | 0.32 (0.19–0.54) RAS G12R | 12/30 | 16/31 | 0.61 (0.29–1.30) Other RAS mutation^b | 3/11 | 10/14 | 0.29 (0.08–1.06) RAS Q61 | 3/11 | 7/11 | 0.37 (0.09–1.44) No RAS mutation identified | 4/9 | 4/7 | 0.43 (0.09–1.98) Axis: 0.05 0.5 1.0 2.0 — Favors Daraxonrasib / Favors Chemo Daraxonrasib showed OS improvement compared with SOC chemotherapy across all RAS genotype subgroups Data cutoff: 10 Feb 2026. ^aLocally determined. ^bOther RAS mutations include RAS Q61 and RAS G13. ^cThe size of diamonds is relative to the size of the population in each subgroup. HRs and 95% CIs overall and for subgroups were based on the unstratified Cox model with Efron's method of tie handling. Presented by Zev A. Wainberg, MD, MSc [ESMO congress logo]

[Slide 3] [RASolute 302 logo] Baseline Demographics and Disease Characteristics Columns: Overall^1: Daraxonrasib (n=248) | Chemo (n=252); Preselected SOC chemo regimen^b: GnP: Daraxonrasib (n=140) | Chemo (n=141); mFOLFIRINOX/FOLFOX^c: Daraxonrasib (n=30) | Chemo (n=30); NAL-IRI + 5-FU/LV: Daraxonrasib (n=78) | Chemo (n=81) Age, median (range), years: 66 (30-88) | 65 (36-86) | 64 (30-85) | 63 (36-86) | 70 (47-88) | 66 (51-82) | 67 (43-87) | 68 (46-86) ECOG PS, n (%) 0: 129 (52) | 118 (47) | 80 (57) | 70 (50) | 13 (43) | 5 (17) | 36 (46) | 43 (53) 1^a: 119 (48) | 134 (53) | 60 (43) | 71 (50) | 17 (57) | 25 (83) | 42 (54) | 38 (47) Metastatic disease at diagnosis, n (%): 154 (62) | 154 (61) | 93 (66) | 102 (72) | 21 (70) | 15 (50) | 40 (51) | 37 (46) Prior 5-FU-based regimen, n (%): 181 (73) | 186 (74) | 139 (99) | 141 (100) | 6 (20) | 9 (30) | 36 (46) | 36 (44) Prior gemcitabine-based regimen, n (%): 110 (44) | 114 (45) | 5 (4) | 4 (3) | 29 (97) | 30 (100) | 76 (97) | 80 (99) Liver metastases at baseline, n (%): 174 (70) | 176 (70) | 105 (75) | 109 (77) | 22 (73) | 21 (70) | 47 (60) | 46 (57) Tumor RAS mutational status, n (%) RAS G12D/V: 197 (79) | 197 (78) | 111 (79) | 109 (77) | 22 (73) | 22 (73) | 64 (82) | 66 (81) Other RAS G12: 31 (13) | 34 (13) | 19 (14) | 24 (17) | 4 (13) | 5 (17) | 8 (10) | 5 (6) RAS G13 or Q61, or no RAS mutation identified: 20 (8) | 21 (8) | 10 (7) | 8 (6) | 4 (13) | 3 (10) | 6 (8) | 10 (12) - Baseline characteristics for patients diagnosed with metastatic disease <6 months from the last dose of neoadjuvant/adjuvant therapy (n=76) and those who received one prior line of therapy for metastatic disease (n=424) were balanced between arms Key baseline characteristics and prognostic factors for subgroups were generally consistent with the overall population and balanced between arms Data cutoff: 10 Feb 2026. ^aThree patients (including 2 patients randomized to the daraxonrasib arm and 1 patient randomized to the chemotherapy arm) had ECOG PS >1 at baseline. These patients were not dosed due to no longer meeting study eligibility criteria. ^bInvestigator's choice of chemotherapy was determined prior to randomization. ^cmFOLFIRINOX and FOLFOX regimens were pooled due to small individual sample sizes. 1. O'Reilly EM, et al. N Engl J Med. 2026;395:325-337. Presented by Zev A. Wainberg, MD, MSc [ESMO Congress Madrid 2026 logo]

[Slide 1] PFS by RAS Genotype Subgroups [RASolute 302 logo] Tumor RAS mutational status(a) | PFS by BICR: Events/N Daraxonrasib | Chemo | PFS HR (95% CI)(c) || PFS by investigator: Events/N Daraxonrasib | Chemo | PFS HR (95% CI)(c) Overall: 127/248 | 130/252 | 0.55 (0.43-0.70) || 135/248 | 149/252 | 0.48 (0.38-0.61) RAS G12D: 53/95 | 74/121 | 0.48 (0.33-0.69) || 53/95 | 79/121 | 0.43 (0.30-0.62) RAS G12V: 39/102 | 29/76 | 0.54 (0.33-0.87) || 46/102 | 36/76 | 0.51 (0.33-0.80) RAS G12R: 19/30 | 16/31 | 0.80 (0.41-1.56) || 22/30 | 17/31 | 0.73 (0.38-1.39) Other RAS mutation(b): 9/11 | 7/14 | 1.29 (0.47-3.51) || 7/11 | 9/14 | 0.57 (0.21-1.56) RAS Q61: 9/11 | 5/11 | 1.55 (0.51-4.69) || 7/11 | 7/11 | 0.63 (0.22-1.8[cut off]) No RAS mutation identified: 6/9 | 2/7 | 1.92 (0.39-9.52) || 6/9 | 6/7 | 0.62 (0.20-1.9[cut off]) [Forest plots, axis 0.05-10; Favors Daraxonrasib / Favors Chemo] PFS results generally favored daraxonrasib versus SOC chemotherapy across RAS genotype subgroups, although treatment-effect estimates differed between BICR and investigator assessments, particularly in smaller subgroups Data cutoff: 10 Feb 2026. aLocally determined. Only subgroups with >=5 patients in at least one treatment arm were included. bOther RAS mutations include RAS Q61 and RAS G13. cHRs and 95% CIs overall and for subgroups were based on the unstratified Cox model with Efron's method of tie handling. Presented by Zev A. Wainberg, MD, MSc [Orange hand-drawn annotation circling RAS G12R row]

[Slide 1] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH G-HOPE-001 study design An open-label, phase 3 study conducted in China and Japan (NCT06238843) Key inclusion criteria: - Adults ≥18 years old - ECOG PS 0-1 - Locally advanced unresectable or metastatic G/GEJA - CLDN18.2-positive by central or local immunohistochemistry* - Received and progressed on 2-4 prior therapies which must include fluoropyrimidine, platinum, and a taxane or irinotecan Key exclusion criteria: - Positive HER2(†) - History of treatment with topoisomerase inhibitor-based ADCs R(‡) 1:1 Arcotatug tavatecan (AT) 6 mg/kg Q3W Treatment of physician's choice (TPC): Irinotecan, paclitaxel, trifluridine/tipiracil Treatment continuation until: Completion of 24 months of study treatment; Disease progression and/or lack of clinical benefit(||); Loss of follow-up; Death; Intolerable toxicity; Withdrawal of informed consent Primary endpoints: PFS by IRRC per RECIST v1.1; OS(§) Secondary endpoints included: ORR per RECIST v1.1; Safety *Defined as CLDN18.2 IHC membrane staining intensity ≥2+ in ≥75% of tumour cells. For participants who have previously received any anti-CLDN18.2 therapy, tumour samples were obtained after discontinuation of prior anti-CLDN18.2 therapy. †Defined as IHC 3+, or IHC 2+ and positive by in situ hybridization. ‡Stratification factors include region (China vs. Japan), primary tumour site (stomach vs. gastroesophageal junction) and history of prior gastrectomy (yes vs. no). ||Patients could continue to receive study treatment after disease progression as long as they continued to show clinical benefit. §In Japan the single primary endpoint is OS. ADC, antibody-drug conjugate; CLDN18.2, claudin18.2; ECOG PS, Eastern Cooperative Oncology Group performance status; G/GEJA, gastric or gastroesophageal junction adenocarcinoma; HER2, human epidermal growth factor receptor 2; IHC, immunohistochemistry; IRRC, independent radiological review committee; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; Q3W, every 3 weeks; R, randomisation; RECIST v1.1, Response Evaluation Criteria in Solid Tumors version 1.1 [Speaker panel] Kohei Shitara - Arcotatug tavatecan (IBI343/TAK-921) versus treatment of physician's choice (TPC) in previously treated, advanced gastric or gastroesophageal junction adenocarcinoma (G/GEJA): First interim analysis of the randomized, open-label, phase 3, G-HOPE-001 ... ALMERIA AUDITORIUM - HALL 9 [ESMO Congress Madrid 2026 logo] [Slide 2] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH Final PFS by IRRC at IA1: primary endpoint | Arcotatug tavatecan (N=203) | TPC (N=205) Event, n (%) | 74 (36.5) | 90 (43.9) Median (95% CI) | 5.5 (4.3, 8.2) | 2.9 (2.6, 3.4) HR (95% CI) | 0.47 (0.34, 0.65) P-value (2-sided) | <0.0001 Curve labels at 6 months: 48.5% (AT); 24.0% (TPC) Y-axis: Progression free survival (%); X-axis: Months (0-15) At risk (censored) at 0/3/6/9/12/15 months: Arcotatug tavatecan: 203 (0), 78 (82), 29 (110), 10 (124), 1 (129), 0 (129) TPC: 205 (0), 43 (92), 7 (111), 2 (113), 1 (114), 0 (115) - AT demonstrated a statistically significant and clinically meaningful improvement in PFS demonstrating a ~53% reduction in risk of progression or death. Consistent benefit was observed by investigator assessment (HR 0.39; 95% CI, 0.30-0.53). ITT population. Median and 95% CI are estimated using Kaplan-Meier method. P-value is from the stratified log-rank test; HR and 95% CI are from stratified Cox proportional hazards model. Median follow-up duration for PFS by IRRC at IA1: AT, 4.2 months (95% CI, 2.9-4.4); TPC, 2.9 months (95% CI, 2.7-3.4). 3L+, third-line or later; CI, confidence interval; CLDN18.2, claudin18.2; G/GEJA, gastric or gastroesophageal junction adenocarcinoma; HR, hazard ratio; IA1, interim analysis 1; IRRC, independent radiological review committee; ITT, intent-to-treat; IA1, interim analysis 1; PFS, progression-free survival; TPC, treatment of physician's choice. [Speaker panel] Kohei Shitara - Arcotatug tavatecan (IBI343/TAK-921) versus treatment of physician's choice (TPC) in previously treated, advanced gastric or gastroesophageal junction adenocarcinoma (G/GEJA): First interim analysis of the randomized, open-label, phase 3, G-HOPE-001 ... 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 IRRC at IA1 (primary endpoint, ITT) Arcotatug tavatecan | 203 | 5.5 mo | 4.3, 8.2 TPC | 205 | 2.9 mo | 2.6, 3.4 HR: 0.47 HR CI: 0.34, 0.65 P: <0.0001 (2-sided) {'timepoint': '6 mo', 'arm': 'Arcotatug tavatecan', 'value': '48.5%', 'ci': None} {'timepoint': '6 mo', 'arm': 'TPC', 'value': '24.0%', 'ci': None} Note: Investigator-assessed PFS HR 0.39 (95% CI 0.30-0.53). Median follow-up AT 4.2 mo, TPC 2.9 mo. [Slide 3] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH OS at IA2: primary endpoint | Arcotatug tavatecan (N=230) | TPC (N=230) Event, n (%) | 122 (53.0) | 159 (69.1) Median (95% CI) | 10.2 (9.1, 12.3) | 6.9 (5.7, 7.8) HR (95% CI) | 0.57 (0.45, 0.73) P-value (2-sided) | <0.0001 Curve labels: 6 months 71.6% (AT), 54.5% (TPC); 12 months 43.2% (AT), 24.7% (TPC) Y-axis: Overall survival (%); X-axis: Months (0-24) At risk (censored) at 0/3/6/9/12/15/18/21/24 months: Arcotatug tavatecan: 230 (0), 218 (0), 149 (16), 88 (52), 43 (78), 14 (98), 7 (103), 1 (107), 0 (108) TPC: 230 (0), 181 (3), 113 (14), 52 (45), 26 (56), 10 (65), 3 (70), 2 (71), 0 (71) - AT demonstrated a statistically significant and clinically meaningful improvement in OS vs. TPC for CLDN18.2+ G/GEJA in the 3L+ setting, demonstrating a ~43% reduction in risk of death ITT population. Median and 95% CI are estimated using Kaplan-Meier method. Median follow-up duration for OS at IA2: AT, 11.1 months (95% CI, 9.8, 12.1); TPC, 11.2 months (95% CI, 9.9, 12.7). 3L+, third-line or later; AT, arcotatug tavatecan; CI, confidence interval; CLDN18.2, claudin18.2; G/GEJA, gastric or gastroesophageal junction adenocarcinoma; HR, hazard ratio; IA2, interim analysis 2; ITT, intent-to-treat; OS, overall survival; TPC, treatment of physician's choice. [Speaker panel] Kohei Shitara - Arcotatug tavatecan (IBI343/TAK-921) versus treatment of physician's choice (TPC) in previously treated, advanced gastric or gastroesophageal junction adenocarcinoma (G/GEJA): First interim analysis of the randomized, open-label, phase 3, G-HOPE-001 ... ALMERIA AUDITORIUM - HALL 9 [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS at IA2 (primary endpoint, ITT) Arcotatug tavatecan | 230 | 10.2 mo | 9.1, 12.3 TPC | 230 | 6.9 mo | 5.7, 7.8 HR: 0.57 HR CI: 0.45, 0.73 P: <0.0001 (2-sided) {'timepoint': '6 mo', 'arm': 'Arcotatug tavatecan', 'value': '71.6%', 'ci': None} {'timepoint': '6 mo', 'arm': 'TPC', 'value': '54.5%', 'ci': None} {'timepoint': '12 mo', 'arm': 'Arcotatug tavatecan', 'value': '43.2%', 'ci': None} {'timepoint': '12 mo', 'arm': 'TPC', 'value': '24.7%', 'ci': None} Note: Median follow-up AT 11.1 mo, TPC 11.2 mo. Caption OS values match slide. [Slide 4] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH Conclusions - The phase 3 G-HOPE-001 study evaluating AT vs. TPC in patients with CLDN18.2+ G/GEJA in the 3L+ setting reached prespecified primary endpoints with statistically significant and clinically meaningful improvements in both PFS and OS: - PFS: At IA1, median 5.5 vs. 2.9 months (HR 0.47; 95% CI 0.34, 0.65; p<0.0001); at IA2, median 7.1 vs. 3.1 months (descriptive) - OS: At IA2, median 10.2 vs. 6.9 months (HR 0.57; 95% CI 0.45, 0.73; p<0.0001) - AT demonstrated a favourable safety profile, with manageable GI and haematological toxicities that were consistent with findings from prior studies - These results support arcotatug tavatecan (AT), a novel Fc-silenced CLDN18.2-targeted ADC, as an effective treatment option in the 3L+ setting for patients with CLDN18.2-positive, HER2-negative G/GEJA 3L+, third-line or later; ADC, antibody-drug conjugate; AT, arcotatug tavatecan; CI, confidence interval; CLDN18.2, claudin18.2; Fc, fragment crystallizable; G/GEJA, gastric or gastroesophageal junction adenocarcinoma; GI, gastrointestinal; HER2, human epidermal growth factor receptor 2; HR, hazard ratio; IA1, interim analysis 1; IA2, interim analysis 2; OS, overall survival; PFS, progression-free survival; TPC, treatment of physician's choice. [Speaker panel] Kohei Shitara - Arcotatug tavatecan (IBI343/TAK-921) versus treatment of physician's choice (TPC) in previously treated, advanced gastric or gastroesophageal junction adenocarcinoma (G/GEJA): First interim analysis of the randomized, open-label, phase 3, G-HOPE-001 ... ALMERIA AUDITORIUM - HALL 9 [ESMO Congress Madrid 2026 logo]

[Slide 1] Background Arcotatug tavatecan is a novel CLDN18.2-targeted ADC comprising an Fc-silenced antibody designed to reduce on target, off-tumor toxicity and conjugated to an exatecan-derived TOPO I inhibitor payload for selective delivery to tumor cells. Here, we report the prespecified first interim analysis (IA1) of the G-HOPE-001 study. Methods Patients (pts) with unresectable or metastatic G/GEJA, CLDN18.2 IHC ≥2+ in ≥75% of tumor cells who had progressed on 2-4 prior regimens were randomized 1:1 to arcotatug tavatecan 6mg/kg Q3W or TPC (irinotecan, paclitaxel or trifluridine/tipiracil) until progression or unacceptable toxicity. Stratification factors were regions, primary tumor site and prior gastrectomy. Primary endpoints were IRRC-assessed PFS per RECIST v1.1 and OS. The prespecified IA1 included the PFS final analysis and an OS futility analysis, with a data cutoff of February 13, 2026. Results In IA1, 408 pts from China and Japan (male: 62.0%, median age: 58.0, ECOG PS 1: 71.6%, primary tumor at stomach: 81.6%, primary tumor resection: 43.6%) were randomized to arcotatug tavatecan (N=203) and TPC (N=205) arms. IRRC-assessed median PFS was 5.5 months vs 2.9 months (HR: 0.47 [95% CI: 0.34-0.65], p-value <0.0001). With 34.8% of events, median OS was 9.9 months vs 7.4 months (HR: 0.63 [95% CI: 0.45-0.89]). IRRC-assessed ORR was 23.6% in arcotatug tavatecan and 5.3% in TPC. Grade ≥3 (G3+) TEAEs occurred in 45.8% pts with arcotatug tavatecan vs 57.9% pts with TPC. Common gastrointestinal TEAEs (≥10%) with arcotatug tavatecan vs TPC were nausea (38.9% vs 40.6%), vomiting (22.2% vs 28.4%) and diarrhea (12.8% vs 34.5%). Interstitial lung disease was not observed. TRAEs leading to treatment discontinuation occurred in 1.0% of pts in each arm. No TRAE led to death. Conclusions Arcotatug tavatecan demonstrated a statistically significant and clinically meaningful improvement in PFS with a manageable safety profile versus TPC for CLDN18.2+ G/GEJA in the 3L+ setting. OS results will be evaluated at the prespecified IA2.

[Slide 2] OS at IA2: primary endpoint [KM curve: Overall survival (%) vs Months] Landmarks printed: 6 mo: AT 71.6%, TPC 54.5%; 12 mo: AT 43.2%, TPC 24.7% Table: Arcotatug tavatecan (N=230) | TPC (N=230) Event, n (%): 122 (53.0) | 159 (69.1) Median (95% CI): 10.2 (9.1, 12.3) | 6.9 (5.7, 7.8) HR (95% CI): 0.57 (0.45, 0.73) P-value (2-sided): <0.0001 At risk (censored), months 0/3/6/9/12/15/18/21/24: Arcotatug tavatecan: 230 (0), 218 (0), 149 (16), 88 (52), 43 (78), 14 (98), 7 (103), 1 (107), 0 (108) TPC: 230 (0), 181 (3), 113 (14), 52 (45), 26 (56), 10 (65), 3 (70), 2 (71), 0 (71) • AT demonstrated a statistically significant and clinically meaningful improvement in OS vs. TPC for CLDN18.2+ G/GEJA in the 3L+ setting, demonstrating a ~43% reduction in risk of death ITT population. Median and 95% CI are estimated using Kaplan–Meier method. Median follow-up duration for OS at IA2: AT, 11.1 months (95% CI, 9.8, 12.1); TPC, 11.2 months (95% CI, 9.9, 12.7). 3L+, third-line or later; AT, arcotatug tavatecan; CI, confidence interval; CLDN18.2, claudin18.2; G/GEJA, gastric or gastroesophageal junction adenocarcinoma; HR, hazard ratio; IA2, interim analysis 2; ITT, intent-to-treat; OS, overall survival; TPC, treatment of physician's choice. [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS (primary endpoint), IA2, ITT Arcotatug tavatecan | 230 | 10.2 | 9.1, 12.3 TPC | 230 | 6.9 | 5.7, 7.8 HR: 0.57 HR CI: 0.45, 0.73 P: <0.0001 (2-sided) {'timepoint': '6 mo', 'arm': 'Arcotatug tavatecan', 'value': '71.6%', 'ci': None} {'timepoint': '6 mo', 'arm': 'TPC', 'value': '54.5%', 'ci': None} {'timepoint': '12 mo', 'arm': 'Arcotatug tavatecan', 'value': '43.2%', 'ci': None} {'timepoint': '12 mo', 'arm': 'TPC', 'value': '24.7%', 'ci': None} Note: Median follow-up OS at IA2: AT 11.1 mo, TPC 11.2 mo. Landmark values printed on curve. [Slide 3] PFS by IRRC updates at IA2 [KM curve: Progression-free survival (%) vs Months] Landmarks printed: 6 mo: AT 55.2%, TPC 23.1% Table: Arcotatug tavatecan (N=230) | TPC (N=230) Event, n (%): 117 (50.9) | 137 (59.6) Median (95% CI): 7.1 (5.3, 9.7) | 3.1 (2.8, 3.8) HR (95% CI): 0.45 (0.35, 0.58) At risk (censored), months 0/3/6/9/12/15/18: Arcotatug tavatecan: 230 (0), 142 (32), 81 (60), 38 (89), 10 (108), 1 (112), 0 (113) TPC: 230 (0), 82 (54), 16 (85), 6 (90), 3 (93), 1 (93), 0 (93) • The descriptive PFS update at IA2 continued to demonstrate a substantial benefit with AT versus TPC ITT population. Median and 95% CI are estimated using Kaplan–Meier method. Median follow-up duration for PFS by IRRC at IA2: AT, 8.2 months (95% CI, 7.0, 8.5); TPC, 4.4 months (95% CI, 4.2, 5.5). AT, arcotatug tavatecan; CI, confidence interval; HR, hazard ratio; IA2, interim analysis 2; IRRC, independent radiological review committee; ITT, intent-to-treat; PFS, progression-free survival; TPC, treatment of physician's choice. [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS by IRRC (descriptive update), IA2, ITT Arcotatug tavatecan | 230 | 7.1 | 5.3, 9.7 TPC | 230 | 3.1 | 2.8, 3.8 HR: 0.45 HR CI: 0.35, 0.58 {'timepoint': '6 mo', 'arm': 'Arcotatug tavatecan', 'value': '55.2%', 'ci': None} {'timepoint': '6 mo', 'arm': 'TPC', 'value': '23.1%', 'ci': None} Note: Descriptive update; no p-value printed. Median follow-up PFS: AT 8.2 mo, TPC 4.4 mo. [Slide 4] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH ORR by IRRC at IA2 Confirmed responses in patients with a measurable lesion Arcotatug tavatecan (N=182) | TPC (N=182) Best overall response, n (%) Complete response: 3 (1.6) | 1 (0.5) Partial response: 65 (35.7) | 13 (7.1) Stable disease: 84 (46.2) | 85 (46.7) Progressive disease: 22 (12.1) | 49 (26.9) Not evaluable: 2 (1.1) | 16 (8.8) No post-baseline assessment yet: 6 (3.3) | 18 (9.9) Objective response rate,* n (%): 68 (37.4) | 14 (7.7) 95% CI: (30.3, 44.4) | (3.8, 11.6) Median duration of response,† months: 9.9 | NC‡ 95% CI: 8.1, 12.1 | 3.0, NC Disease control rate,§ n (%): 152 (83.5) | 99 (54.4) 95% CI: (78.1, 88.9) | (47.2, 61.6) • AT demonstrated a higher response rate with durable responses Includes all randomised participants with at least one target/measurable lesion at baseline. *Complete response + partial response. †Duration of response calculated in patients with confirmed objective response: AT, n=68; TPC, n=14; ‡Among AT and TPC responders, duration of response events occurred in 21/68 (30.9%) and 5/14 (35.7%), respectively; 10/68 (14.7%) and 4/14 (28.6%) were censored for subsequent therapy, with remaining responses ongoing at data cutoff; §Complete response + partial response + stable disease. AT, arcotatug tavatecan; CI, confidence interval; IA2, interim analysis 2; IRRC, independent radiological review committee; NC, not calculated; ORR, objective response rate; TPC, treatment of physician's choice. [Speaker video] Kohei Shitara Arcotatug tavatecan (IBI343/TAK-921) versus treatment of physician's choice (TPC) in previously treated, advanced gastric or gastroesophageal junction adenocarcinoma (G/GEJA): First interim analysis of the randomized, open-label, phase 3, G-HOPE-001 ... ALMERIA AUDITORIUM - HALL 9 [ESMO Congress Madrid 2026 logo]

[Slide 1] ESMO GASTROINTESTINAL CANCERS Annual Congress | MUNICH GERMANY 26-29 JUNE 2024 BACKGROUND Nonmalignant cells: Polarity; Apical membrane; Claudin 18.2; Basolateral membrane; Basement membrane Cancer cells CLDN18.2 expressed aberrantly on tumor cell surface - promising therapeutic target Nakayama et al, Nat Rev Clin Oncol 2024 [ADC structure diagram: Asn297; spacer1; spacer2] Fully human anti-CLDN18.2 Ab with IgG1 Fc silenced | Site specific conjugation and cleavable linker | Exatecan (topoisomerase inhibitor) payload DAR 4 IBI343 ADC Profile targeting CLDN 18.2 • Promising in vivo safety: Fc-silent with no ADCC-mediated GI toxicity • Promising in vivo efficacy: Strong on-target killing effect & "bystander killing effect" ESMO GASTROINTESTINAL CANCERS 2024 Dr Jia (Jenny) Liu [Slide 2] nature medicine | Search | Log in Content | About | Publish articles > article Download PDF Article | Open access | Published: 16 July 2025 CLDN18.2-targeting antibody–drug conjugate IBI343 in advanced gastric or gastroesophageal junction adenocarcinoma: a phase 1 trial Jia Liu, Jianwei Yang, Yuping Sun, Jifang Gong, Jinbo Yue, Yueyin Pan, Meili Sun, Rongfeng Song, Xiuying Xiao, Andrea Tazbirkova, Jian Ruan, Zhenyang Liu, Zimin Liu, Zhihua Li, Lili Sheng, Yanru Qin, Jieer Ying, Xianjun Yu, Jian Zhang, Yiping Mou, Chuangxin Lu, Ping Chen, Suyi Li, Jie Li, Xiujuan Qu, Ting Deng, Juan Du, Aiping Zhou, [author list continues] [Slide 4] Overall safety summary at IA2 | | Arcotatug tavatecan (N=230) | TPC (N=222) | Median (range) treatment duration, months | 5.0 (0.7–18.8) | 2.0 (0.5–14.3) Any TEAEs, n (%) | 226 (98.3) | 216 (97.3) Treatment-related | 222 (96.5) | 208 (93.7) Any Grade ≥3 TEAEs, n (%) | 134 (58.3) | 139 (62.6) Treatment-related | 109 (47.4) | 124 (55.9) Any TESAEs, n (%) | 90 (39.1) | 84 (37.8) Treatment-related | 51 (22.2) | 53 (23.9) Any TEAEs leading to dose interruption, n (%) | 109 (47.4) | 114 (51.4) Any TEAEs leading to dose reduction, n (%) | 41 (17.8) | 78 (35.1) Any TEAEs leading to drug discontinuation, n (%) | 8 (3.5) | 5 (2.3) Treatment-related | 4 (1.7) | 3 (1.4) Any TEAEs leading to death, n (%) | 8 (3.5) | 5 (2.3) Treatment-related | 1 (0.4)* | 0 • Treatment-related Grade ≥3 TEAEs and TEAEs leading to dose reduction or interruption were less frequent with AT vs. TPC • Incidence of TESAEs was similar between arms, and treatment-related drug discontinuations were uncommon Includes participants who received ≥1 dose of study drug. Events were graded according to Common Terminology Criteria for Adverse Events version 5.0. *Sudden death of unknown cause AT, arcotatug tavatecan; IA2, interim analysis 2; TEAE, treatment emergent adverse ev[obscured by black bar]choice. [ESMO Congress Madrid 2026 logo]

[Slide 4] TEAEs in ≥10% of patients at IA2 [Butterfly bar chart; columns: AE | Arcotatug tavatecan any grade | AT grade ≥3 | TPC grade ≥3 | TPC any grade] Anaemia: 66.1 | 20.4 | 13.1 | 54.1 WBC count decreased: 58.3 | 11.7 | 22.5 | 54.5 Neutrophil count decreased: 55.2 | 17.4 | 32.0 | 58.1 Decreased appetite: 44.3 | 3.9 | 2.3 | 35.1 Nausea: 43.9 | 0.9 | 1.8 | 43.7 Hypoalbuminaemia: 40.4 | 0 | 0.5 | 28.4 Weight decreased: 30.9 | 3.0 | 2.3 | 22.5 Platelet count decreased: 28.7 | 6.5 | 5.4 | 18.9 ALT increased: 25.7 | 0.4 | 0 | 13.5 Vomiting: 25.2 | 2.6 | 3.2 | 32.9 AST increased: 22.6 | 1.7 | 1.4 | 14.0 Hypokalaemia: 22.2 | 4.8 | 12.6 | 24.8 GGT increased: 20.9 | 6.1 | 3.6 | 12.2 Constipation: 20.0 | 0 | 0 | 18.0 Hypocalcaemia: 19.1 | 0.4 | 0.9 | 14.4 Lymphocyte count decreased: 18.7 | 7.4 | 10.8 | 21.6 Hyponatraemia: 16.1 | 1.3 | 0.5 | 11.7 Blood ALP increased: 15.7 | 1.7 | 1.8 | 9.5 Diarrhoea: 15.2 | 1.3 | 4.1 | 40.5 Fatigue: 15.2 | 0.9 | 0 | 11.3 Asthenia: 14.3 | 2.2 | 3.6 | 17.6 Hypoproteinaemia: 13.0 | 0 | 0 | 5.9 Malaise: 12.2 | 0 | 0.9 | 10.4 Protein urine present: 10.9 | 0 | 0 | 6.8 Abdominal pain: 10.0 | 2.2 | 1.4 | 11.7 Alopecia: 1.7 | 0 | 0 | 12.6 Patients experiencing TEAEs (%) • GI toxicities such as nausea, vomiting and diarrhoea were comparable or less frequent with arcotatug tavatecan than TPC; haematological events were manageable Includes participants who received ≥1 dose of study drug. Events were graded according to Common Terminology Criteria for Adverse Events version 5.0. Presented in order of overall frequency with arcotatug tavatecan. ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyl transferase; GI, gastrointestinal; IA2, interim analysis 2; TEAE, treatment emergent adverse event; TPC, treatment of physician's choice; WBC, white blood cell. [ESMO Congress Madrid 2026 logo]

[Slide 1] Proffered Paper session LBA77 - Arcotatug tavatecan (IBI343/TAK-921) versus treatment of physician's choice (TPC) in previously treated, advanced gastric or gastroesophageal junction adenocarcinoma (G/GEJA): First interim analysis of the randomized, open-label, phase 3, G-HOPE-001 study Speakers Kohei Shitara (Kashiwa, Japan, Chiba) Lecture Time 13:30 - 13:40 [Abstract] Background Arcotatug tavatecan is a novel CLDN18.2-targeted ADC comprising an Fc-silenced antibody designed to reduce on target, off-tumor toxicity and conjugated to an exatecan-derived TOPO I inhibitor payload for selective delivery to tumor cells. Here, we report the prespecified first interim analysis (IA1) of the G-HOPE-001 study. [Slide 2] Results In IA1, 408 pts from China and Japan (male: 62.0%, median age: 58.0, ECOG PS 1: 71.6%, primary tumor at stomach: 81.6%, primary tumor resection: 43.6%) were randomized to arcotatug tavatecan (N=203) and TPC (N=205) arms. IRRC-assessed median PFS was 5.5 months vs 2.9 months (HR: 0.47 [95% CI: 0.34-0.65], p-value <0.0001). With 34.8% of events, median OS was 9.9 months vs 7.4 months (HR: 0.63 [95% CI: 0.45-0.89]). IRRC-assessed ORR was 23.6% in arcotatug tavatecan and 5.3% in TPC. Grade ≥3 (G3+) TEAEs occurred in 45.8% pts with arcotatug tavatecan vs 57.9% pts with TPC. Common gastrointestinal TEAEs (≥10%) with arcotatug tavatecan vs TPC were nausea (38.9% vs 40.6%), vomiting (22.2% vs 28.4%) and diarrhea (12.8% vs 34.5%). Interstitial lung disease was not observed. TRAEs leading to treatment discontinuation occurred in 1.0% of pts in each arm. No TRAE led to death. Conclusions Arcotatug tavatecan demonstrated a statistically significant and clinically meaningful improvement in PFS with a manageable safety profile versus TPC for CLDN18.2+ G/GEJA in the 3L+ setting. OS results will be evaluated at the prespecified IA2. [Slide 3] Methods Patients (pts) with unresectable or metastatic G/GEJA, CLDN18.2 IHC ≥2+ in ≥75% of tumor cells who had progressed on 2-4 prior regimens were randomized 1:1 to arcotatug tavatecan 6mg/kg Q3W or TPC (irinotecan, paclitaxel or trifluridine/tipiracil) until progression or unacceptable toxicity. Stratification factors were regions, primary tumor site and prior gastrectomy. Primary endpoints were IRRC-assessed PFS per RECIST v1.1 and OS. The prespecified IA1 included the PFS final analysis and an OS futility analysis, with a data cutoff of February 13, 2026.

[Slide 1] G-HOPE-001 trial: design and IA1 result An open-label, phase 3 study conducted in China and Japan (NCT06238843) Key inclusion criteria: • ECOG PS 0–1 • Locally advanced unresectable or metastatic G/GEJA • CLDN18.2 IHC 2+/3+ in ≥75% TC • 3L+ • Prior zolbetuximab and irinotecan allowed Key exclusion criteria: • Positive HER2ᵃ • History of treatment with topoisomerase inhibitor-based ADCs R‡ 1:1 Arcotatug tavatecan (AT) 6 mg/kg Q3W Treatment of physician's choice (TPC): irinotecan, paclitaxel, trifluridine/tipiracil ‡Stratification factors include region (China vs. Japan), primary tumour site (stomach vs. gastroesophageal junction) and history of prior gastrectomy (yes vs. no) Cross-over not allowed Dual primary endpoint PFS met! Events maturity: 40% [Embedded KM curve: Progression-free survival (%) vs Months; inset table (Arcotatug tavatecan vs TPC: Event n (%), Median (95% CI), HR (95% CI), P-value) - values too small/blurry to read reliably: illegible] HR 0.47, absolute gain +2.6 months (median)/+25% (@6mos) Filippo Pietrantonio, MD [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS (dual primary endpoint), IA1 Arcotatug tavatecan TPC HR: 0.47 Note: Only HR 0.47, events maturity 40%, and 'absolute gain +2.6 months (median)/+25% (@6mos)' are legibly printed; embedded inset table values illegible - do not infer medians. [Slide 2] Overall survival: the other dual primary endpoint OS at IA2, median follow-up ≈ 11 mos Events maturity: 61% [KM curve: Overall survival vs months 0-24; landmark labels on curve at 6 and 12 months, partly blurry: AT ~71.6% / TPC ~56.6% at 6 mo; AT 43.7% / TPC 24.7% at 12 mo] Arcotatug tavatecan (N=230) TPC (N=230) Event, n (%) 122 (53.0) 159 (69.1) Median (95% CI) 10.2 (9.1, 12.3) 6.9 (5.7, 7.8) HR (95% CI) 0.57 (0.45, 0.73) P-value (2-sided) <0.0001 HR 0.57, absolute gain +3.3 months (median) PFS may be sufficient for regulatory purposes in some settings, but OS and quality of life remain the only endpoints that matter to patients, especially in the 3L+ setting Filippo Pietrantonio, MD [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS (dual primary endpoint), IA2 Arcotatug tavatecan | 230 | 10.2 mo | 9.1, 12.3 TPC | 230 | 6.9 mo | 5.7, 7.8 HR: 0.57 HR CI: 0.45, 0.73 P: <0.0001 (2-sided) {'timepoint': '12 months', 'arm': 'Arcotatug tavatecan', 'value': '43.7%', 'ci': None} {'timepoint': '12 months', 'arm': 'TPC', 'value': '24.7%', 'ci': None} Note: Inset table text is small and somewhat blurry: N=230 per arm, event counts and CI bounds are best reads - verify against primary presentation. 6-month landmark labels (~71.6% / ~56.6%) too blurry to record. Median follow-up ~11 mo; events maturity 61%.

[Slide 1] Camupepimut (ELI-002 7P) Mechanism of Action Targeting the Lymph Nodes with Amphiphile (AMP) to Orchestrate Anti-tumor T cell Response Tissue Injection Site: AMP; Endogenous Albumin; Albumin-bound AMP Lymph Node: Albumin-bound AMP; Antigen Presenting Cell; T Cell APC Activation T Cell Response ↑ Expansion ↑ Persistence ↑ Effector function ↑ Phenotype 1 Subcutaneous injection -> 2 Albumin binding -> 3 Lymph node targeting -> 4 Delivery to immune cells -> 5 T Cell Response Wainberg, O'Reilly, et al. Nature Medicine. 2025; Pant, O'Reilly, et al. Nature Medicine. 2024; Liu, Irvine, et al. Nature. 2014; Moynihan, Irvine, et al. Nature Medicine. 2016; Ma, Irvine, et al. Science. 2019 Eileen M. O'Reilly, MD, FASCO [ESMO Congress Madrid 2026 logo] [Slide 2] AMPLIFY-7P Phase 2 Primary Endpoint: DFS endpoint in ITT not met n=144 | Intention-to-Treat: Camupepimut (n=96) Median 23.8 months Observation (n=48) Median 21.2 months HR: 0.85 (0.52-1.41) P=0.5329 Dosing Period (shaded, ~0-4.5 months) Y axis: Disease-free Survival (%); X axis: Months At risk (0,3,...,27 mo): Camupepimut: 96 82 64 50 42 40 24 15 6 0 Observation: 48 36 29 28 24 22 16 10 4 0 Eileen M. O'Reilly, MD, FASCO Primary DFS analyses used stratified Cox model and specified stratified log-rank test; DCO: April 20, 2026 [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: DFS (primary, ITT) Camupepimut | 96 | 23.8 months Observation | 48 | 21.2 months HR: 0.85 HR CI: 0.52-1.41 P: 0.5329 Note: ITT n=144. Primary endpoint not met. DCO April 20, 2026. [Slide 3] Exploratory Analysis: DFS Supervised by mKRAS T Cell Response Strength of mKRAS-specific T cell response significantly correlates to improved RFS / DFS in Phase 1, Phase 2 n=25 | Phase 1: RFS Supervised by T cell Response >=9.17x (n=17) HR: 0.12 (0.02-0.62) Median: Not Reached <9.17x (n=8) P=0.0002 Median: 3.02 months Y axis: Relapse-free Survival (%); X axis: Months At risk: >=9.17x 17 13 10 8 6 6 3 1 0; <9.17x 8 3 1 1 1 1 0 n=90 | Phase 2: DFS Supervised by T cell Response >9.17x (n=74) HR: 0.22 (0.08-0.63) Median: Not Reached <9.17x (n=16) P<0.0001 Median: 6.08 months Y axis: Disease-free Survival (%); X axis: Months At risk: >9.17x 74 68 53 46 40 38 24 15 6 0; <9.17x 16 9 7 3 1 1 0 Eileen M. O'Reilly, MD, FASCO Phase 1 ELI-002-2P: Wainberg, O'Reilly, et al. Nature Medicine. 2025; mKRAS T cell fold-change assessed among n=90 evaluable patients; DCO: April 20, 2026 [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: Phase 1 RFS / Phase 2 DFS by mKRAS T-cell response fold-change (exploratory) Phase 1 >=9.17x | 17 | Not Reached Phase 1 <9.17x | 8 | 3.02 months Phase 2 >9.17x | 74 | Not Reached Phase 2 <9.17x | 16 | 6.08 months HR: Phase 1 0.12; Phase 2 0.22 HR CI: Phase 1 0.02-0.62; Phase 2 0.08-0.63 P: Phase 1 P=0.0002; Phase 2 P<0.0001 Note: Exploratory, non-randomized comparison by T-cell response strength. Phase 1 = ELI-002-2P. DCO April 20, 2026. [Slide 4] Pre-specified DFS Analysis: ctDNA Positive vs ctDNA Negative Camupepimut treatment effect was consistent in ctDNA+ and ctDNA- n=119 | Prespecified: ctDNA Positive vs Negative ctDNA: Negative - Camupepimut / Observation: HR: 0.51 (0.25-1.02) P=0.0280 ctDNA: Positive - Camupepimut / Observation: HR: 0.57 (0.22-1.48) P=0.0489 Camupepimut, ctDNA- Median: Not Reached Observation, ctDNA- Median: 21.2 months Camupepimut, ctDNA+ Median: 4.3 months Observation, ctDNA+, Median: 1.9 months Y axis: Disease-free Survival (%); X axis: Months At risk (0,3,...,27 mo): Camupepimut ctDNA-: 62 57 50 42 37 35 21 14 5 0 Observation ctDNA-: 31 24 20 19 17 16 12 8 3 0 Camupepimut ctDNA+: 20 13 7 1 1 1 1 0 0 0 Observation ctDNA+: 6 2 1 1 0 0 0 0 0 0 Eileen M. O'Reilly, MD, FASCO Unstratified analyses, non-adjusted P-values were calculated using two-sided Wilcoxon test; Primary DFS analyses used specified stratified log-rank test; ITT HR 0.85; DCO: April 20, 2026 [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: DFS by baseline ctDNA status (pre-specified) Camupepimut, ctDNA- | Not Reached Observation, ctDNA- | 21.2 months Camupepimut, ctDNA+ | 4.3 months Observation, ctDNA+ | 1.9 months HR: ctDNA- 0.51; ctDNA+ 0.57 HR CI: ctDNA- 0.25-1.02; ctDNA+ 0.22-1.48 P: ctDNA- P=0.0280; ctDNA+ P=0.0489 (unstratified, non-adjusted, two-sided Wilcoxon) Note: n=119 analysed. Arm n not printed (at-risk at t=0: 62/31/20/6). DCO April 20, 2026.

[Slide 1] AMPLIFY-7P: Randomized Phase 2 Trial in Adjuvant mKRAS+ PDAC NCT05726864 (2:1 Randomized) Testing whether camupepimut (ELI-002 7P) improves DFS following standard locoregional therapy Camupepimut Monotherapy • 7 mKRAS peptide antigens Phase 2: Key Criteria • Includes: mKRAS G12D/R/V/C/A/S/G13D • Resected Stage I-III mKRAS PDAC • Complete R0/R1 resection • Radiographic NED status < 6 months • MRD agnostic (biomarker +/-) • Enrollment Completed 12/2024 R 2:1 — Stratification: Nodal Status N=96 Camupepimut Monotherapy N=48 Standard of Care: Observation Optional crossover @ confirmed PD N=144 pts, 24 U.S. Sites Endpoints Primary Endpoint: Disease Free Survival (DFS) Tumor Biomarker Response Overall Survival, Safety Exploratory: Immunogenicity DFS assessment by iRECIST (new lesions confirmed by biopsy/imaging); R0/R1: Absence or presence of microscopic residual disease; NED: No Evidence of Disease. MRD: Minimal Residual Disease. PD: Progressive Disease Eileen M. O'Reilly, MD, FASCO [ESMO Congress Madrid 2026 logo] [Slide 4] Key Takeaways Published Today in Nature Medicine [QR code] George, O'Reilly, et al. Nature Medicine. 2026 Primary Endpoint Not Met • Primary DFS Endpoint: Not met in ITT population (HR: 0.85 P=0.5329) • However, the biologically, clinically relevant signal informs future Phase 3 Clear Clinical Signal in Prespecified Subgroups Consistent treatment effect observed in ctDNA negative (HR: 0.51 P=0.0280), ctDNA positive (HR: 0.57 P=0.0489), no prior radiation (HR: 0.59 P=0.0221) Biological Validation Strong mKRAS-specific T cell response (>9.17x) correlated with DFS (HR: 0.22 P<0.0001) Favorable Safety Camupepimut (ELI-002 7P) maintained a favorable safety profile DFS: Disease-free Survival; ITT: Intention-to-treat Eileen M. O'Reilly, MD, FASCO [ESMO Congress Madrid 2026 logo]

[Slide 1] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH Key Takeaways Published Today in Nature Medicine | [QR code] George, O'Reilly, et al. Nature Medicine. 2026 Primary Endpoint Not Met | • Primary DFS Endpoint: Not met in ITT population (HR: 0.85 P=0.5329) • However, the biologically, clinically relevant signal informs future Phase 3 Clear Clinical Signal in Prespecified Subgroups | Consistent treatment effect observed in ctDNA negative (HR: 0.51 P=0.0280), ctDNA positive (HR: 0.57 P=0.0489), no prior radiation (HR: 0.59 P=0.0221) Biological Validation | Strong mKRAS-specific T cell response (>9.17x) correlated with DFS (HR: 0.22 P<0.0001) Favorable Safety | Camupepimut (ELI-002 7P) maintained a favorable safety profile Eileen M. O'Reilly, MD, FASCO DFS, Disease-free Survival; ITT, Intention-to-treat [ESMO Congress Madrid 2026 logo] [Speaker card] Eileen O'Reilly - Seven-valent mKRAS-specific lymph node targeted amphiphile immunotherapy versus observation in resected pancreatic ductal adenocarcinoma (PDAC): The AMPLIFY-7P randomized phase 2 study ALMERIA AUDITORIUM - HALL 9 [Slide 2] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive [partially cropped] AMPLIFY-7P Phase 2 Primary Endpoint: DFS endpoint in ITT not met n=144 Intention-to-Treat: Camupepimut (n=96) Median 23.8 months Observation (n=48) Median 21.2 months HR: 0.85 (0.52-1.41) P=0.5329 [KM curve; y-axis: Disease-free Survival (%); x-axis: Months 0-30; shaded Dosing Period] At risk (months 0, 3, 6, 9, 12, 15, 18, 21, 24, 27): Camupepimut: 96, 82, 64, 50, 42, 40, 24, 15, 6, 0 Observation: 48, 36, 29, 28, 24, 22, 16, 10, 4, 0 Primary DFS analyses used stratified Cox model and specified stratified log-rank test; DCO: April 20, 2026 Eileen M. O'Reilly, MD, FASCO [ESMO Congress Madrid 2026 logo] ALMERIA AUDITORIUM - HALL 9 [Survival data captured as a table — no point estimates read from the curve] Endpoint: DFS (primary endpoint, ITT) Camupepimut | 96 | 23.8 months Observation | 48 | 21.2 months HR: 0.85 HR CI: 0.52-1.41 P: 0.5329 Note: Primary endpoint not met. Stratified Cox / stratified log-rank. DCO April 20, 2026. Randomization 2:1 (96 vs 48), total n=144. [Slide 3] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH AMPLIFY-7P Phase 2 Exploratory Analysis: mKRAS T Cell Response Potent mKRAS-specific T cell responses observed in patients randomized to camupepimut T Cell Response: 99% (89 / 90) Responders; 1% Non-responders T cell Fold Change Over Baseline: 146x Mean; 44x Median; 9.17x [y-axis: Fold-change from baseline 0-1300; Baseline vs Max Response] mKRAS T cell Response >9.17x: 82% (74 / 90) CD4 / CD8: 85% (68 / 80) CD4 + CD8; 9% CD4 Only; 6% CD8 Only mKRAS Variant Specificity (% with Response): G12D 89%; G12R 87%; G12V 82%; G12C 83%; G12A 88%; G12S 89%; G13D 88% [y-axis: Average fold change from baseline; Baseline vs Max Response] Among evaluable patients with assessable baseline and on-treatment samples; DCO: April 20, 2026 Eileen M. O'Reilly, MD, FASCO [ESMO Congress Madrid 2026 logo] [Slide 4] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH AMPLIFY-7P Phase 2 Pre-specified DFS Analysis: No Prior Radiation Prior radiation observed to influence response to camupepimut; Treatment effect observed in no prior radiation n=106 Prespecified: No Prior Radiation Camupepimut (n=73) HR: 0.59 (0.33-1.03) Observation (n=33) P=0.0221 Median: 23.8 months [camupepimut] Median: 10.4 months [observation] [KM curve; y-axis: Disease-free Survival (%); x-axis: Months 0-30] At risk (months 0, 3, 6, 9, 12, 15, 18, 21, 24, 27): Camupepimut: 73, 64, 51, 40, 34, 33, 20, 11, 3, 0 Observation: 33, 21, 17, 17, 14, 13, 8, 4, 2, 0 Baseline Lymphocytes: Absolute Lymphocyte Count (x10^9 / L); P < 0.0001; Prior Radiation No vs Yes; Median 1.7 vs 1.2 mKRAS T Cell Response: Fold-change from baseline; P = 0.0011; Prior Radiation No vs Yes; Median 58x vs 15x mKRAS T cell Response >9.17x: OR: 0.21 (0.07-0.64) P=0.005 Unstratified analyses, non-adjusted P-values were calculated using two-sided Wilcoxon test; Primary DFS analyses used specified stratified log-rank test; ITT HR 0.85; DCO: April 20, 2026 Eileen M. O'Reilly, MD, FASCO [ESMO Congress Madrid 2026 logo] ALMERIA AUDITORIUM - HALL 9 [Survival data captured as a table — no point estimates read from the curve] Endpoint: DFS, pre-specified subgroup: no prior radiation Camupepimut | 73 | 23.8 months Observation | 33 | 10.4 months HR: 0.59 HR CI: 0.33-1.03 P: 0.0221 Note: Subgroup (n=106). Note HR 95% CI crosses 1 (0.33-1.03) while printed P=0.0221 - both as printed. ITT primary endpoint not met (HR 0.85). DCO April 20, 2026.

[Slide 1] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH Primary Endpoint: DFS endpoint in ITT not met [AMPLIFY-7P Phase 2] n=144 | Intention-to-Treat: [KM curve; y-axis: Disease-free Survival (%); x-axis: Months 0-30; shaded Dosing Period] Camupepimut (n=96) Median 23.8 months Observation (n=48) Median 21.2 months HR: 0.85 (0.52-1.41) P=0.5329 At risk (0, 3, 6, 9, 12, 15, 18, 21, 24, 27, 30) Camupepimut 96 82 64 50 42 40 24 15 6 0 Observation 48 36 29 28 24 22 16 10 4 0 Eileen M. O'Reilly, MD, FASCO Primary DFS analyses used stratified Cox model and specified stratified log-rank test; DCO: April 20, 2026 [Speaker card] Eileen O'Reilly — Seven-valent mKRAS-specific lymph node targeted amphiphile immunotherapy versus observation in resected pancreatic ductal adenocarcinoma (PDAC): The AMPLIFY-7P randomized phase 2 study ALMERIA AUDITORIUM - HALL 9 [Survival data captured as a table — no point estimates read from the curve] Endpoint: DFS (ITT, primary) Camupepimut | 96 | 23.8 months Observation | 48 | 21.2 months HR: 0.85 HR CI: 0.52-1.41 P: 0.5329 Note: Stratified Cox / stratified log-rank. At-risk rows: Camupepimut 96,82,64,50,42,40,24,15,6,0; Observation 48,36,29,28,24,22,16,10,4,0 (0-27 mo by 3).

[Slide 3] Pre-specified DFS Analysis: 6 Month DFS Rate AMPLIFY-7P Phase 2 n=144 Intention-to-Treat: 6 months Camupepimut: 75.7% Observation: 61.7% [faded legend: Camupepimut (n=96) Median 23.8 months; Observation (n=48) Median 21.2 months; HR: 0.85 (0.52-1.41) P=0.5329] Pre-specified DFS Rate | 6 months Camupepimut 75.7% Observation 61.7% DFS Rate Difference 14.0% P-value 0.0557 Exploratory HR from t0 to 6 months 0.55 Dosing Period At risk (months 0/3/6/9/12/15/18/21/24/27) Camupepimut: 96 82 64 50 42 40 24 15 6 0 Observation: 48 36 29 28 24 22 16 10 4 0 Eileen M. O'Reilly, MD, FASCO Primary DFS analyses used stratified Cox model and specified stratified log-rank test; DCO: April 20, 2026 [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: DFS (ITT) - pre-specified 6-month DFS rate Camupepimut (ELI-002 7P) | 96 | 23.8 months Observation | 48 | 21.2 months HR: 0.85 HR CI: 0.52-1.41 P: 0.5329 {'timepoint': '6 months', 'arm': 'Camupepimut (ELI-002 7P)', 'value': '75.7%', 'ci': None} {'timepoint': '6 months', 'arm': 'Observation', 'value': '61.7%', 'ci': None} Note: 6-month DFS rate difference 14.0%, P=0.0557 (not significant at 0.05). Exploratory HR t0 to 6 months 0.55. DCO April 20, 2026. [Slide 4] Methods AMPLIFY-7P is a randomized, open label phase 2 trial in patients (pts) with stages I-III, KRAS-mutated PDAC who received curative-intent therapy and no recurrence at ≤ 6 months. Pts randomized 2:1, stratified by lymph node status (N0 vs N1/2), to 10 doses camupepimut (prime, boost) or observation. Primary endpoint: investigator DFS; secondary: safety, 6-month DFS; success required Hazard Ratio (HR) ≤ 0.70, 2-sided alpha ~0.10. Results From 01-12/2024, 144 pts with mKRAS PDAC; 44.4% G12D, 35.4% G12V, 16.7% G12R: randomized to camupepimut (n=96) or observation (n=48) at 24 centers. Stratification balanced nodal status (N1/N2: 57[59.4%] vs 29[60.4%]) but ctDNA(+) 20[20.8%] vs 6[12.5%]. Primary DFS endpoint not met (HR 0.85, p=0.539), however, pre-specified 6-month DFS advantage of 14% (75.7% vs 61.7%; HR=0.55; p=0.0557) and mKRAS-specific T cells correlated with efficacy (mDFS not reached vs 6.08 mo; HR=0.22; p<0.0001); subgroups promising DFS:. Table: LBA83 Subgroup | n(%) | DFS (mo) | HR | Nominal p ctDNA(-) | 93(65%) | NR vs 21.2 | 0.51 | 0.0280 No radiation | 106(74%) | 23.8 vs 10.4 | 0.59 | 0.0221 Post-hoc multivariate Cox regression analysis DFS: camupepimut adjusted HR=0.45; 95%CI 00.25-0.80; p=0.0064; ctDNA adjusted HR=6.72; 95%CI 3.49-12.94; p<0.0001. Grade ≥ 3 AEs (10.4 vs 14.6%); serious AEs (6.3 vs 8.3%); no new safety signals. Immunogenicity: mKRAS-specific T cell responses 99% (89/90); 85% (68/80) CD8+ and CD4+ T cells and antigen spreading 87% (13/15). Conclusions Prespecified 6-month subgroup DFS increased and DFS associated with mKRAS-specific T cell response, although primary DFS endpoint not met. Baseline ctDNA(+) adversely prognostic/imbalanced. Phase III planned with refined stratification and pt selection factors.

[Slide 3] AMPLIFY-7P | Phase 2 Pre-specified DFS Analysis: No Prior Radiation Prior radiation observed to influence response to camupepimut; Treatment effect observed in no prior radiation n=106 Prespecified: No Prior Radiation Disease-free Survival (%) Camupepimut (n=73) HR: 0.59 (0.33-1.03) Observation (n=33) P=0.0221 Median: 23.8 months [Camupepimut] Median: 10.4 months [Observation] Months At risk (0, 3, 6, 9, 12, 15, 18, 21, 24, 27) Camupepimut: 73 64 51 40 34 33 20 11 3 0 Observation: 33 21 17 17 14 13 8 4 2 0 Baseline Lymphocytes Absolute Lymphocyte Count (x10⁹ / L) P < 0.0001 Prior Radiation: No | Yes Median ▸ 1.7 | 1.2 mKRAS T Cell Response Fold-change from baseline P = 0.0011 Prior Radiation: No | Yes Median ▸ 58x | 15x 9.17x [threshold line] mKRAS T cell Response >9.17x OR: 0.21 (0.07-0.64) P=0.005 Unstratified analyses, non-adjusted P-values were calculated using two-sided Wilcoxon test; Primary DFS analyses used specified stratified log-rank test; ITT HR 0.85; DCO: April 20, 2026 Eileen M. O'Reilly, MD, FASCO [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: DFS - prespecified subgroup: no prior radiation (n=106) Camupepimut | 73 | 23.8 months Observation | 33 | 10.4 months HR: 0.59 HR CI: 0.33-1.03 P: 0.0221 (unstratified, non-adjusted) Note: Subgroup analysis; ITT HR 0.85. Side panels: baseline lymphocytes median 1.7 vs 1.2 (P<0.0001), mKRAS T-cell fold-change median 58x vs 15x (P=0.0011); T-cell response >9.17x OR 0.21 (0.07-0.64), P=0.005. DCO April 20, 2026.

[Slide 2] Summary and Conclusions [HERIZON GEA-301 logo] Zanidatamab Demonstrated Superiority Over Trastuzumab • At this FFA, zanidatamab + CT demonstrated a statistically significant and clinically meaningful improvement in OS (HR, 0.76) vs trastuzumab + CT • All primary endpoints have been met, showing PFS and OS benefits for both zanidatamab-containing regimens Sustained Long-Term Survival With Zanidatamab + Tislelizumab + CT • With longer follow-up, zanidatamab + tislelizumab + CT sustained a robust OS benefit over trastuzumab + CT, with an improved HR vs IA1 (HR, 0.65 vs 0.72)¹ and a 40% 3-year OS rate vs 23% with trastuzumab + CT • Adding tislelizumab to zanidatamab + CT provided a meaningful trend toward further OS benefit irrespective of PD-L1 status Long-Term Safety and Tolerability • The safety profiles remained manageable with no new safety signals, supporting the tolerability of prolonged treatment These results solidify zanidatamab with chemotherapy ± tislelizumab as the new standard of care in 1L HER2-positive mGEA, replacing trastuzumab-based regimens [QR code] A copy of the presentation can be accessed via the QR code. Copies of this slide deck obtained through QR code are for personal use only and may not be reproduced without written permission from the author of the slides 1L, first-line; CT, chemotherapy; FFA, final formal analysis; HER2, human epidermal growth factor receptor 2; HR, hazard ratio; IA1, first interim analysis of OS; mGEA, advanced or metastatic gastro-oesophageal adenocarcinoma; OS, overall survival; PD-L1, programmed death-ligand 1; PFS, progression-free survival. 1. Shitara K, et al. N Engl J Med. 2026;394:2002-14. Kohei Shitara, MD [ESMO Congress Madrid 2026 logo]

[Slide 1] LBA79 - Zanidatamab (zani) + chemotherapy (CT) ± tislelizumab (TIS) for first-line (1L) HER2-positive (HER2+) advanced/metastatic gastroesophageal adenocarcinoma (mGEA): Second interim analysis (IA) from HERIZON-GEA-01 Speakers Kohei Shitara (Kashiwa, Japan, Chiba) Lecture Time 16:15 - 16:20 [Abstract button] Background In the primary analysis of progression-free survival (PFS) and first IA (IA1) of overall survival (OS) of the phase 3 HERIZON-GEA-01 trial (NCT05152147), zani + CT ± TIS significantly prolonged PFS and, with TIS, provided a significant OS benefit (HR 0.72) vs trastuzumab (tras) + CT (Shitara, et al. NEJM. 2026), supporting the US approval of these regimens in 1L HER2+ mGEA. Here, we present the planned second IA of OS (IA2). Methods Patients (pts) with untreated HER2+ mGEA were randomized 1:1:1 to zani (1800 mg [<70 kg]/2400 mg [≥70 kg] IV every 3 weeks [Q3W]) + CT (capecitabine/oxaliplatin or 5-FU/cisplatin) ± TIS (200 mg IV Q3W) or tras + CT. PFS (by blinded independent central review) and OS were primary endpoints. At IA2, formal OS comparisons were tested in sequence: zani + CT vs tras + CT, then zani + CT + TIS vs zani + CT. Results 914 pts were randomized. With a median follow-up of 35.3 mo (data cutoff: 15 July 2026; 9.4 mo longer than IA1), zani + CT significantly improved OS vs tras + CT (HR 0.76 [95% CI: 0.62–0.93]; P = 0.0088; median OS, 22.9 mo [95% CI: 19.9–27.6] vs 19.6 mo [95% CI: 16.9–22.2]). A 4-mo longer median OS was observed with zani + TIS + CT (27.0 mo [95% CI: 23.7–32.2]) vs zani + CT (HR 0.81 [95% CI: 0.65–1.00]), although the difference was not statistically significant (P = 0.0502). The OS benefit of zani + TIS + CT vs tras + CT observed at IA1 (HR 0.72) was sustained at IA2 with an improved HR (0.65 [95% CI: 0.52–0.80]). Estimated 36-mo OS was 39.8% for zani + CT + TIS, 31.5% for zani + CT, and 22.9% for tras + CT. Median duration of treatment was 46.6 (40.5 for zani) wks with zani + CT + TIS, 31.1 (30.1 for zani) wks with zani + CT, and 30.0 (28.2 for tras) wks with tras + CT. The safety profiles were consistent with IA1. Conclusions At IA2, zani + CT led to a statistically significant and clinically meaningful OS improvement vs tras + CT. With longer follow-up, zani + CT + TIS showed improved OS benefit over tras + CT and there was a meaningful trend for OS improvement vs zani + CT. Unchanged safety profiles support the tolerability of prolonged treatment. These results continue to support zani combined with CT ± TIS as a new HER2-directed 1L standard of care in HER2+ mGEA.

[Slide 1] Zanidatamab + Tislelizumab + CT: OS in the ITT Population Zanidatamab + tislelizumab + CT continued to demonstrate an OS benefit vs trastuzumab + CT, with an improved HR from IA1 [HERIZON GEA-01 logo] Median OS (95% CI), mo: Zanidatamab + tislelizumab + CT 27.0 (23.7-32.2) | Trastuzumab + CT 19.6 (16.9-22.2) HR (95% CI): 0.65 (0.52-0.80)^a Curve annotations: 55.8% (95% CI, 49.8-61.4) [24 mo, zanidatamab + tislelizumab + CT]; 40.7% (95% CI, 34.9-46.5) [24 mo, trastuzumab + CT]; 39.8% (95% CI, 33.2-46.3) [36 mo, zanidatamab + tislelizumab + CT]; 22.9% (95% CI, 17.3-28.5) [36 mo, trastuzumab + CT] Y-axis: Overall survival (%); X-axis: Months from randomisation (0-57); + Censored Patients at risk (0,3,6,...,57 months): Zani + TIS + CT: 302 267 246 228 207 185 168 148 129 101 72 64 53 42 35 28 11 6 2 0 Tras + CT: 308 284 261 227 199 171 147 121 95 70 48 39 33 21 15 12 5 1 1 0 Data cutoff: 15 July 2026. ^aAnalysis was descriptive, as prespecified criteria for significance were previously met. CT, chemotherapy; HR, hazard ratio; IA1, first interim analysis of OS; ITT, intent-to-treat; OS, overall survival; TIS, tislelizumab; Tras, trastuzumab; Zani, zanidatamab. Kohei Shitara, MD [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS (ITT) Zanidatamab + tislelizumab + CT | 302 | 27.0 mo | 23.7-32.2 Trastuzumab + CT | 308 | 19.6 mo | 16.9-22.2 HR: 0.65 HR CI: 0.52-0.80 {'timepoint': '24 mo', 'arm': 'Zanidatamab + tislelizumab + CT', 'value': '55.8%', 'ci': '49.8-61.4'} {'timepoint': '24 mo', 'arm': 'Trastuzumab + CT', 'value': '40.7%', 'ci': '34.9-46.5'} {'timepoint': '36 mo', 'arm': 'Zanidatamab + tislelizumab + CT', 'value': '39.8%', 'ci': '33.2-46.3'} {'timepoint': '36 mo', 'arm': 'Trastuzumab + CT', 'value': '22.9%', 'ci': '17.3-28.5'} Note: Descriptive analysis (prespecified significance criteria previously met at IA1). DCO 15 July 2026. Landmark-to-arm assignment by label position near each curve. Trastuzumab 36-mo upper CI read as 28.5 here but 28.9 on a clearer photo (2113638372698685867_s3) - verify. [Slide 2] Zanidatamab + Tislelizumab + CT vs Trastuzumab + CT: OS in Key Prespecified Subgroups The OS benefit with zanidatamab + tislelizumab + CT vs trastuzumab + CT was generally consistent across prespecified subgroups, including in patients with PD-L1-negative disease [HERIZON GEA-01 logo] Subgroup | Category | Events/Patients Zanidatamab + tislelizumab + CT | Events/Patients Trastuzumab + CT | OS HR^a (95% CI) All patients | | 157/302 | 212/308 | 0.65 (0.52-0.80) Age | <65 years | 78/163 | 116/162 | 0.56 (0.42-0.74) Age | >=65 years | 79/139 | 96/146 | 0.79 (0.59-1.07) Geographic region | Asia | 70/159 | 111/165 | 0.55 (0.41-0.74) Geographic region | EU/NA | 60/95 | 65/93 | 0.84 (0.59-1.20) Geographic region | ROW | 27/48 | 36/50 | 0.69 (0.42-1.14) ECOG PS | 0 | 53/122 | 67/120 | 0.70 (0.49-1.00) ECOG PS | 1 | 103/179 | 145/188 | 0.64 (0.49-0.82) Prior gastrectomy | Yes | 21/51 | 34/56 | 0.61 (0.36-1.06) Prior gastrectomy | No | 136/251 | 178/252 | 0.65 (0.52-0.82) Anatomical subtype | Gastric | 96/208 | 155/226 | 0.55 (0.43-0.71) Anatomical subtype | GEJ | 51/74 | 45/60 | 0.98 (0.66-1.47) Anatomical subtype | Oesophageal | 10/20 | 12/22 | 0.73 (0.32-1.71) HER2 status | IHC 3+ | 127/251 | 172/255 | 0.65 (0.51-0.82) HER2 status | IHC 2+/ISH+ | 30/51 | 39/52 | 0.70 (0.43-1.13) PD-L1 status | TAP <1% | 47/90 | 75/98 | 0.50 (0.35-0.73) PD-L1 status | TAP >=1% | 90/187 | 121/188 | 0.68 (0.52-0.89) Forest plot axis: 0.1 - 1.0 - 10.0; Favours Zani + TIS + CT | Favours Tras + CT Data cutoff: 15 July 2026. ^aThe widths of the CIs were not adjusted for multiplicity and cannot be used to infer treatment effects. CT, chemotherapy; ECOG PS, Eastern Cooperative Oncology Group Performance Status; EU, European Union; GEJ, gastro-oesophageal junction; HER2, human epidermal growth factor receptor 2; HR, hazard ratio; IHC, immunohistochemistry; ISH, in situ hybridisation; NA, North America; OS, overall survival; PD-L1, programmed death-ligand 1; ROW, rest of the world; TAP, tumour area positivity; TIS, tislelizumab; Tras, trastuzumab; Zani, zanidatamab Kohei Shitara, MD [ESMO Congress Madrid 2026 logo] [Slide 3] Zanidatamab + Tislelizumab + CT vs Zanidatamab + CT: OS in the ITT Population A consistent trend toward improved OS was observed irrespective of PD-L1 status [HERIZON GEA-01 logo] Median OS (95% CI), mo: Zanidatamab + tislelizumab + CT 27.0 (23.7-32.2) | Zanidatamab + chemotherapy 22.9 (19.9-27.6) ITT HR (95% CI): 0.81 (0.65-1.00); P = 0.0502 PD-L1 TAP <1% HR (95% CI): 0.76 (0.52-1.10) PD-L1 TAP >=1% HR (95% CI): 0.77 (0.58-1.02) Y-axis: Overall survival (%); X-axis: Months from randomisation (0-57); + Censored Patients at risk (0,3,6,...,57 months): Zani + TIS + CT: 302 267 246 228 207 185 168 148 129 101 72 64 53 42 35 28 11 6 2 0 Zani + CT: 304 278 256 228 202 182 162 142 122 94 71 51 42 28 23 18 5 3 1 0 Data cutoff: 15 July 2026 CT, chemotherapy; HR, hazard ratio; ITT, intent-to-treat; OS, overall survival; PD-L1, programmed death-ligand 1; TAP, tumour area positivity; TIS, tislelizumab; Zani, zanidatamab Kohei Shitara, MD [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS (ITT) Zanidatamab + tislelizumab + CT | 302 | 27.0 mo | 23.7-32.2 Zanidatamab + chemotherapy | 304 | 22.9 mo | 19.9-27.6 HR: 0.81 HR CI: 0.65-1.00 P: 0.0502 Note: PD-L1 TAP <1% HR 0.76 (0.52-1.10); TAP >=1% HR 0.77 (0.58-1.02). DCO 15 July 2026. Comparison between the two zanidatamab arms (adding tislelizumab).

[Slide 1] Background • Zanidatamab is a bispecific HER2-directed antibody that binds to HER2 ECD 2 and 4 in a trans configuration, facilitating crosslinking associated with the formation of distinct HER2 clusters on the cell surface. It has several key mechanisms of action1: – Increasing HER2 internalisation; reducing phosphorylation of EGFR, HER2, and HER3 and blocking downstream signalling; and inducing immune-mediated effects (CDC, ADCC, and ADCP) • Tislelizumab is a high-affinity ICI targeting PD-1 and is specifically engineered to minimise Fcγ receptor binding on macrophages Adults with untreated, unresectable, locally advanced, recurrent or metastatic GEA • HER2 IHC 3+ or IHC 2+/ISH+ per central testing • ECOG PS 0 or 1 • No prior HER2-targeted agents or immunotherapy in any setting Stratification factors • Geographic region • HER2 status • ECOG PS R 1:1:1 N = 914 Trastuzumab + chemotherapy(a) Zanidatamab(b) + chemotherapy(a) Zanidatamab(b) + tislelizumab(c) + chemotherapy(a) Prophylaxis to prevent IRRs and diarrhoea was mandatory in the zanidatamab-containing arms Treatment until disease progression/death/unacceptable toxicity. Chemotherapy could be discontinued after 6 cycles Tumour imaging(d) Q6W Dual primary endpoints • PFS (per BICR) • OS Data cutoff for IA2: 15 July 2026 Here we report results from the planned second interim analysis of OS (IA2) from HERIZON-GEA-01 (a)Physician's choice of CAPOX (capecitabine plus oxaliplatin) or FP (5-fluorouracil plus cisplatin). Chemotherapy was administered for at least 6 cycles or until disease progression, unacceptable toxicity, or another criterion for treatment discontinuation was met. (b)Zanidatamab 1800 mg (<70 kg) or 2400 mg (≥70 kg) was administered IV Q3W. (c)Tislelizumab 200 mg was administered IV Q3W. (d)Computed tomography/magnetic resonance imaging scans were performed Q6W for the first 54 weeks, then every 9 weeks. ADCC, antibody-dependent cellular cytotoxicity; ADCP, antibody-dependent cellular phagocytosis; BICR, blinded independent central review; CDC, complement-dependent cytotoxicity; ECD, extracellular domain; ECOG PS, Eastern Cooperative Oncology Group Performance Status; EGFR, epidermal growth factor receptor; Fcγ, fragment crystallisable-gamma; GEA, gastro-oesophageal adenocarcinoma; HER2, human epidermal growth factor receptor 2; HER3, human epidermal growth factor receptor 3; IA2, planned second interim analysis of OS; ICI, immune checkpoint inhibitor; IHC, immunohistochemistry; IRR, infusion-related reaction; ISH, in situ hybridisation; IV, intravenously; OS, overall survival; PD-1, programmed cell death protein 1; PFS, progression-free survival; Q3W, every 3 weeks; Q6W, every 6 weeks; R, randomisation. 1. Weisser NE, et al. Nat Commun. 2023;14(1):1394. Kohei Shitara, MD [HERIZON GEA-301 logo] [ESMO Congress Madrid 2026 logo] [Slide 2] Zanidatamab + Tislelizumab + CT vs Zanidatamab + CT: OS in the ITT Population A consistent trend toward improved OS was observed irrespective of PD-L1 status Zanidatamab + tislelizumab + CT | Zanidatamab + chemotherapy Median OS (95% CI), mo: 27.0 (23.7–32.2) | 22.9 (19.9–27.6) ITT HR (95% CI): 0.81 (0.65–1.00); P = 0.0502 PD-L1 TAP <1% HR (95% CI): 0.76 (0.52–1.10) PD-L1 TAP ≥1% HR (95% CI): 0.77 (0.58–1.02) Y-axis: Overall survival (%); X-axis: Months from randomisation; + Censored Patients at risk (0,3,6,...,57 mo): Zani + TIS + CT: 302 267 246 228 207 185 168 148 129 101 72 64 53 42 35 28 11 6 2 0 Zani + CT: 304 278 256 228 202 182 162 142 122 94 71 51 42 28 22 18 5 3 1 0 Data cutoff: 15 July 2026. CT, chemotherapy; HR, hazard ratio; ITT, intent-to-treat; OS, overall survival; PD-L1, programmed death-ligand 1; TAP, tumour area positivity; TIS, tislelizumab; Zani, zanidatamab. Kohei Shitara, MD [HERIZON GEA-301 logo] [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS (ITT) Zanidatamab + tislelizumab + CT | 302 | 27.0 mo | 23.7–32.2 Zanidatamab + chemotherapy | 304 | 22.9 mo | 19.9–27.6 HR: 0.81 HR CI: 0.65–1.00 P: 0.0502 Note: PD-L1 TAP <1% HR 0.76 (0.52–1.10); TAP ≥1% HR 0.77 (0.58–1.02). Data cutoff 15 July 2026 (IA2). n from at-risk row. [Slide 3] Safety Summary The AE profile remained consistent with IA1, and no unexpected safety signals were observed Columns: Zanidatamab + tislelizumab + CT (n = 294)(a) | Zanidatamab + CT (n = 305)(a) | Trastuzumab + CT (n = 302) Duration of treatment, weeks Mean (SD): 64.1 (58.0) | 52.6 (52.6) | 42.3 (39.2) Median (IQR): 46.6 (17.6–101.4) | 31.1 (14.3–74.7) | 30.0 (18.1–54.1) Duration of zanidatamab or trastuzumab treatment, weeks Mean (SD): 60.2 (57.4) | 51.0 (52.9) | 40.8 (39.4) Median (IQR): 40.5 (14.4–97.1) | 30.1 (12.1–72.7) | 28.2 (16.3–52.1) Any-grade TEAE, n (%): 293 (99.7) | 301 (98.7) | 297 (98.3) TRAE, n (%): 289 (98.3) | 296 (97.0) | 291 (96.4) Grade ≥3: 213 (72.4) | 180 (59.0) | 181 (59.9) Serious TEAEs, n (%): 176 (59.9) | 155 (50.8) | 131 (43.4) Treatment-related: 124 (42.2) | 87 (28.5) | 62 (20.5) TEAEs leading to death, n (%): 30 (10.2) | 26 (8.5) | 22 (7.3) Treatment-related: 7 (2.4) | 1 (0.3) | 4 (1.3) Discontinuation due to TRAEs, n (%) Any component: 132 (44.9) | 107 (35.1) | 88 (29.1) Zanidatamab or trastuzumab(b): 35 (11.9) | 28 (9.2) | 7 (2.3) Tislelizumab(c): 49 (16.7) | — | — Any-grade diarrhoea, n (%): 244 (83.0) | 241 (79.0) | 161 (53.3) Grade ≥3: 74 (25.2) | 61 (20.0) | 39 (12.9) AESIs(d), n (%): 103 (35.0) | 94 (30.8) | 56 (18.5) IRRs: 74 (25.2) | 77 (25.2) | 40 (13.2) Noninfectious pulmonary toxicities: 22 (7.5) | 5 (1.6) | 4 (1.3) Left ventricular dysfunction: 27 (9.2) | 20 (6.6) | 13 (4.3) Immune-mediated AEs(d), n (%): 118 (40.1) | 44 (14.4) | 32 (10.6) • Diarrhoea occurred primarily in cycles 1 and 2, and the incidence was consistent with IA1 Data cutoff: 15 July 2026. (a)Five patients who were assigned to the zanidatamab + tislelizumab + CT arm did not receive tislelizumab. Data from these patients are summarized in the zanidatamab + CT arm. (b)Zanidatamab- or trastuzumab-related TEAEs that led to discontinuation of zanidatamab or trastuzumab. (c)Tislelizumab-related TEAEs that led to discontinuation of tislelizumab. (d)AESIs for zanidatamab were IRRs, non-infectious pulmonary toxicities, and left ventricular dysfunction; AESIs for tislelizumab were IRRs and immune-mediated AEs. Immune-mediated AEs are excluded from AESIs and reported separately. AESIs for zanidatamab and tislelizumab are reported in all treatment groups, even if the study agent was not administered in that group. AE, adverse event; AESI, AE of special interest; CT, chemotherapy; IA1, first interim analysis of OS; IQR, interquartile range; IRR, infusion-related reaction; TEAE, treatment-emergent AE; TRAE, treatment-related AE. Kohei Shitara, MD [HERIZON GEA-301 logo] [ESMO Congress Madrid 2026 logo] [Slide 4] Zanidatamab + CT: OS in the ITT Population Zanidatamab + CT demonstrated a statistically significant and clinically meaningful improvement in OS vs trastuzumab + CT (FFA criteria met) Zanidatamab + CT | Trastuzumab + CT Median OS (95% CI), mo: 22.9 (19.9–27.6) | 19.6 (16.9–22.2) HR (95% CI): 0.76 (0.62–0.93); P = 0.0088(a) Landmark labels: 24 mo: 49.6% (95% CI, 43.6–55.3) [Zani + CT]; 40.7% (95% CI, 34.9–46.5) [Tras + CT] 36 mo: 31.5% (95% CI, 25.4–37.8) [Zani + CT]; 22.9% (95% CI, 17.3–28.9) [Tras + CT] Y-axis: Overall survival (%); X-axis: Months from randomisation; + Censored Patients at risk (0,3,6,...,57 mo): Zani + CT: 304 278 256 228 202 182 162 142 122 94 71 51 42 28 22 18 5 3 1 0 Tras + CT: 308 284 261 227 199 171 147 121 95 70 48 39 33 21 15 12 5 1 1 0 Data cutoff: 15 July 2026. (a)The critical 2-sided alpha was 0.0279. CT, chemotherapy; FFA, final formal analysis; HR, hazard ratio; ITT, intent-to-treat; OS, overall survival; Tras, trastuzumab; Zani, zanidatamab. Kohei Shitara, MD [HERIZON GEA-301 logo] [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS (ITT) Zanidatamab + CT | 304 | 22.9 mo | 19.9–27.6 Trastuzumab + CT | 308 | 19.6 mo | 16.9–22.2 HR: 0.76 HR CI: 0.62–0.93 P: 0.0088 {'timepoint': '24 mo', 'arm': 'Zanidatamab + CT', 'value': '49.6%', 'ci': '43.6–55.3'} {'timepoint': '24 mo', 'arm': 'Trastuzumab + CT', 'value': '40.7%', 'ci': '34.9–46.5'} {'timepoint': '36 mo', 'arm': 'Zanidatamab + CT', 'value': '31.5%', 'ci': '25.4–37.8'} {'timepoint': '36 mo', 'arm': 'Trastuzumab + CT', 'value': '22.9%', 'ci': '17.3–28.9'} Note: Critical 2-sided alpha 0.0279; FFA criteria met. Data cutoff 15 July 2026. n from at-risk row. Footnote text small/blurry.

[Slide 2] Progression-Free Survival in the ITT Population PFS was consistent with the primary analysis (IA1) HERIZON GEA-301 Zanidatamab + Tislelizumab + CT | Zanidatamab + CT | Trastuzumab + CT Median PFS (95% CI), mo | 12.7 (9.8–16.8) | 12.4 (9.8–14.5) | 8.3 (7.0–8.9) HR (95% CI) vs Tras + CT | 0.61 (0.50–0.76) | 0.65 (0.53–0.80) Progression-free survival (%) + Censored Months from randomisation Patients at risk (0, 3, 6, ... 54 months) [low resolution; digits marked [?] are uncertain] Zani + TIS + CT: 302 239 183 151 124 109 90 71 66[?] 49 38 36[?] 28[?] 28[?] 17 14 3 2 0 Zani + CT: 304 231 175 140 115 86[?] 77 63 55 38 29 27 21 15 10 7 1 1 0 Tras + CT: 308 247 168[?] 101 68[?] 49 33 25 19 12 9 8 6[?] 5 4 4 0 Data cutoff: 15 July 2026 CT, chemotherapy; HR, hazard ratio; ITT, intent-to-treat; PFS, progression-free survival; TIS, tislelizumab; Tras, trastuzumab; Zani, zanidatamab. Kohei Shitara, MD [Survival data captured as a table — no point estimates read from the curve] Endpoint: Progression-free survival, ITT (final formal analysis) Zanidatamab + Tislelizumab + CT | 302 | 12.7 mo | 9.8–16.8 Zanidatamab + CT | 304 | 12.4 mo | 9.8–14.5 Trastuzumab + CT | 308 | 8.3 mo | 7.0–8.9 HR: 0.61 (zani+tisle+CT vs tras+CT); 0.65 (zani+CT vs tras+CT) HR CI: 0.50–0.76; 0.53–0.80 Note: N from at-risk at time 0. Two HRs, each vs trastuzumab + CT. No p-values printed. Data cutoff 15 July 2026 (small print, legible). Some later at-risk digits uncertain due to resolution.

[Slide 1] Patient Disposition Median follow-up was 35.3 months at IA2, 9.4 months longer than at IA1 [HERIZON GEA-01 logo] Randomised N = 914 Zanidatamab + tislelizumab + CT: n = 302; 299 patients treated^a Ongoing in study, n = 122 (40%): Ongoing study treatment, n = 67; Survival follow-up, n = 55 Discontinued study, n = 180 (60%): Death, n = 157; Withdrawal of consent, n = 20; Lost to follow-up, n = 3 Zanidatamab + CT: n = 304; 300 patients treated^a Ongoing in study, n = 103 (34%): Ongoing study treatment, n = 48; Survival follow-up, n = 55 Discontinued study, n = 201 (66%): Death, n = 178; Withdrawal of consent, n = 19; Lost to follow-up, n = 3; Investigator's decision, n = 1 Trastuzumab + CT: n = 308; 302 patients treated Ongoing in study, n = 75 (24%): Ongoing study treatment, n = 19; Survival follow-up, n = 56 Discontinued study, n = 233 (76%): Death, n = 211; Withdrawal of consent, n = 21; Lost to follow-up, n = 1 Data cutoff: 15 July 2026. ^aTreated includes all randomised patients who received any amount of any study treatment and does not necessarily reflect the safety analysis set. Five patients assigned to the zanidatamab-tislelizumab-CT arm did not receive tislelizumab and are included in the safety analysis set for the zanidatamab-CT arm. CT, chemotherapy; IA1, first interim analysis of overall survival; IA2, planned second interim analysis of overall survival Kohei Shitara, MD [ESMO Congress Madrid 2026 logo; presenter video panel] [Slide 2] Zanidatamab + CT: OS in the ITT Population Zanidatamab + CT demonstrated a statistically significant and clinically meaningful improvement in OS vs trastuzumab + CT (FFA criteria met) [HERIZON GEA-01 logo] Median OS (95% CI), mo: Zanidatamab + CT 22.9 (19.9-27.6) | Trastuzumab + CT 19.6 (16.9-22.2) HR (95% CI): 0.76 (0.62-0.93); P = 0.0088^a Curve annotations: 49.6% (95% CI, 43.6-55.3) [24 mo, zanidatamab + CT]; 40.7% (95% CI, 34.9-46.5) [24 mo, trastuzumab + CT]; 31.5% (95% CI, 25.4-37.8) [36 mo, zanidatamab + CT]; 22.9% (95% CI, 17.3-28.9) [36 mo, trastuzumab + CT] Y-axis: Overall survival (%); X-axis: Months from randomisation (0-57); + Censored Patients at risk (partly obscured by camera): Zani + CT: [304] 278 256 228 202 182 162 142 122 94 71 51 42 28 22 18 5 3 1 0; Tras + CT: [308] 284 261 227 199 171 147 121 95 70 48 39 33 21 15 12 5 1 1 0 Data cutoff: [15] July 2026. ^aThe critical 2-sided alpha was 0.0279. CT, chemotherapy; FFA, final formal analysis; HR, hazard ratio; ITT, intent-to-treat; OS, overall survival; Tras, trastuzumab; Zani, zanidatamab Kohei Shitara, MD [partly obscured] [ESMO Congress Madrid 2026 logo; presenter video panel; edge of previous slide visible at left] [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS (ITT) Zanidatamab + CT | 304 | 22.9 mo | 19.9-27.6 Trastuzumab + CT | 308 | 19.6 mo | 16.9-22.2 HR: 0.76 HR CI: 0.62-0.93 P: 0.0088 (critical 2-sided alpha 0.0279) {'timepoint': '24 mo', 'arm': 'Zanidatamab + CT', 'value': '49.6%', 'ci': '43.6-55.3'} {'timepoint': '24 mo', 'arm': 'Trastuzumab + CT', 'value': '40.7%', 'ci': '34.9-46.5'} {'timepoint': '36 mo', 'arm': 'Zanidatamab + CT', 'value': '31.5%', 'ci': '25.4-37.8'} {'timepoint': '36 mo', 'arm': 'Trastuzumab + CT', 'value': '22.9%', 'ci': '17.3-28.9'} Note: Final formal analysis criteria met. n from disposition slide / first at-risk value (partly obscured). Note: trastuzumab 36-mo CI printed 17.3-28.9 here vs 17.3-28.5 read on Kotani photo of companion slide - small print, verify. [Slide 4] Zanidatamab + Tislelizumab + CT vs Zanidatamab + CT: OS in the ITT Population Median OS was 4 months longer with zanidatamab + tislelizumab + CT compared with zanidatamab + CT; however, this OS comparison had limited power for superiority [HERIZON GEA-01 logo] Median OS (95% CI), mo: Zanidatamab + tislelizumab + CT 27.0 (23.7-32.2) | Zanidatamab + chemotherapy 22.9 (19.9-27.6) HR (95% CI): 0.81 (0.65-1.00); P = 0.0502 Curve annotations: 55.8% (95% CI, 49.8-61.4) [24 mo, zani + tis + CT]; 49.6% (95% CI, 43.6-55.3) [24 mo, zani + CT]; 39.8% (95% CI, 33.2-46.3) [36 mo, zani + tis + CT]; 31.5% (95% CI, 25.4-37.8) [36 mo, zani + CT] Y-axis: Overall survival (%); X-axis: Months from randomisation (0-57); + Censored Patients at risk: Zani + TIS + CT: 302 267 246 228 207 185 168 148 129 101 72 64 53 42 35 28 11 6 2 0 Zani + CT: 304 278 256 228 202 182 162 142 122 94 71 51 42 28 22 18 5 3 1 0 Data cutoff: 15 July 2026 CT, chemotherapy; HR, hazard ratio; ITT, intent-to-treat; OS, overall survival; TIS, tislelizumab; Zani, zanidatamab Kohei Shitara, MD [ESMO Congress Madrid 2026 logo; presenter video panel] [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS (ITT) Zanidatamab + tislelizumab + CT | 302 | 27.0 mo | 23.7-32.2 Zanidatamab + CT | 304 | 22.9 mo | 19.9-27.6 HR: 0.81 HR CI: 0.65-1.00 P: 0.0502 {'timepoint': '24 mo', 'arm': 'Zanidatamab + tislelizumab + CT', 'value': '55.8%', 'ci': '49.8-61.4'} {'timepoint': '24 mo', 'arm': 'Zanidatamab + CT', 'value': '49.6%', 'ci': '43.6-55.3'} {'timepoint': '36 mo', 'arm': 'Zanidatamab + tislelizumab + CT', 'value': '39.8%', 'ci': '33.2-46.3'} {'timepoint': '36 mo', 'arm': 'Zanidatamab + CT', 'value': '31.5%', 'ci': '25.4-37.8'} Note: Limited power for superiority per slide. DCO 15 July 2026. Different subtitle than the version in 2113637094992052251_s3 (which shows PD-L1 subgroup HRs) - likely two slide versions/builds.

[Slide 1] [ACTICCA-1 logo] [UKE Universitatsklinikum Hamburg-Eppendorf logo] Study Design ACTICCA-1 is an investigator-initiated, randomised, controlled, phase 3 study. Key inclusion criteria - Age 18 years or older - Histologically confirmed CCA or muscle-invasive gallbladder cancer - Mixed tumour types and ampullary cancers excluded - Macroscopically complete resection with curative intent (R0 and R1) - ECOG 0 or 1 - Adequate renal, haematological, and liver function - Start of adjuvant treatment within 6 to 16 weeks after surgery R 1:1 Stratification - Tumour location - N0 vs. N1 - R0 vs. R1 Group A CisGem (ABC 02): 8 cycles cisplatin 25 mg/m2 and gemcitabine 1000 mg/m2 on day 1 and 8 of a 21 days cycle CT or MRI q 3 months for 2 years, imaging q 6 months for 3 additional years Group B Capecitabine (BILCAP): 8 cycles capecitabine 1.250 mg/m2 BID on days 1 through 14 of a 21 days cycle Primary endpoint - Disease-free survival (DFS), i.e. recurrence or death due to any cause Secondary endpoints - Overall survival (OS) - Cumulative Recurrence - Safety (CTCAE version 4.03) - Quality of life (QoL) monitored with EORTC QLQ-C30 and BIL 21 ACTICCA-1 was conducted in 67 hepatobiliary sites in Australia, Austria, Denmark, Germany, Italy, New Zealand, The Netherlands, and the United Kingdom. Henning Wege, LBA21, ACTICCA-1 [ESMO Congress Madrid 2026 logo] [Slide 2] Primary Endpoint (DFS) [KM: Disease-free survival probability vs Time since randomization (years)] Group A CisGem / Group B Capecitabine DFS rate at 24 months 52% vs. 44% - Median DFS 25.7 vs. 17.9 months - Log rank p=0.097 - Cox HR 0.82 (95% CI 0.67-1.01), p=0.058 - Non-constant treatment effect (Schoenfeld residuals) Number at risk: Group A 297 191 140 110 75 35 Group B 294 169 122 101 72 30 Henning Wege, LBA21, ACTICCA-1 [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: Disease-free survival (primary) Group A CisGem | 297 | 25.7 months Group B Capecitabine | 294 | 17.9 months HR: 0.82 (Cox) HR CI: 0.67-1.01 (95% CI) P: Cox p=0.058; log rank p=0.097 {'timepoint': '24 months', 'arm': 'Group A CisGem', 'value': '52%', 'ci': None} {'timepoint': '24 months', 'arm': 'Group B Capecitabine', 'value': '44%', 'ci': None} Note: Non-constant treatment effect (Schoenfeld residuals). Primary endpoint not statistically significant. [Slide 3] Secondary Endpoints (OS & Recurrence) RMST OS (3 years) Δ 1.44 months, p=0.076 [KM: Overall survival probability vs Time since randomization (years)] Number at risk: Group A 297 193 101 2; Group B 294 177 98 1 Median OS 58.1 vs. 48.2 months, Cox HR 0.82, p=0.103 Cumulative incidence of recurrence or disease-related death (Aalen-Johansen estimates) [Curve: Cumulative incidence vs Time since randomization (years)] Group A CisGem / Group B Capecitabine Henning Wege, LBA21, ACTICCA-1 [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival (secondary) Group A CisGem | 297 | 58.1 months Group B Capecitabine | 294 | 48.2 months HR: 0.82 (Cox) P: 0.103 Note: RMST OS (3 years) difference 1.44 months, p=0.076. Number-at-risk row small (Group A 297/193/101/2; Group B 294/177/98/1). OS not statistically significant. [Slide 4] Safety (Selected Events) Graded via CTCAE version 4.03 Group A CisGem, N=297 | Group B Capecitabine, N=294 | Total, N=591 All events (%), G3/4 (%) 1967 (74.7), 348 (23.2) | 1662 (69.5), 158 (14.3) | 3629 (72.1), 506 Severe 130 | 113 | 243 Related 51 | 47 | 98 Haematology (%), G3/4 577 (25.9), 234 | 180 (6.5), 18 | 757 (16.2), 252 Anaemia 114 | 21 | 135 Neutropenia 233 | 29 | 262 Gastrointestinal (%), G3/4 361 (11.8), 11 | 452 (15.0), 40 | 813 (13.4), 51 Diarrhoea 65 | 139 | 204 Nausea 91 | 85 | 176 Stomatitis 18 | 23 | 41 Skin (%), G3/4 105 (3.0), 1 | 353 (16.0), 44 | 458 (9.5), 45 Alopecia 29 | 9 | 38 PPES 4 | 262 | 266 General (%), G3/4 294 (9.1), 20 | 214 (12.2), 6 | 508 (10.7), 26 Fatigue 167 | 112 | 279 PPE = Palmar-plantar erythrodysestesia syndome Henning Wege, LBA21, ACTICCA-1 [ESMO Congress Madrid 2026 logo]

[Slide 3] Secondary Endpoints (OS & Recurrence) RMST OS (3 years) Δ 1.44 months, p=0.076 Y-axis: Overall survival probability; X-axis: Time since randomization (years) 0–6 Group A 297 193 101 2 Group B 294 177 99 1 Median OS 58.1 vs. 48.2 months, Cox HR 0.82, p=0.103 Cumulative incidence of recurrence or disease-related death (Aalen-Johansen estimates) Y-axis: Cumulative incidence; X-axis: Time since randomization (years) 0–5 Group A CisGem; Group B Capecitabine Henning Wege, LBA21, ACTICCA-1 [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival (secondary endpoint) Group A CisGem | 297 | 58.1 months Group B Capecitabine | 294 | 48.2 months HR: 0.82 (Cox) P: 0.103 (Cox); RMST OS (3 years) Δ 1.44 months, p=0.076 Note: Right panel: cumulative incidence of recurrence or disease-related death (Aalen-Johansen estimates) — no numeric values printed. Tweet says 'p=0.1' for OS; slide prints p=0.103.

[Slide 1] Presentation Adjuvant chemotherapy with gemcitabine and cisplatin compared to standard of care after curative intent resection of cholangiocarcinoma and gallbladder carcinoma (ACTICCA-1 trial) Saturday, 24 Oct 2026 10:25 - 10:35 Alicante Auditorium - Hall 6 Save | Note Info Background Despite surgery and adjuvant capecitabine, about 50% of biliary tract cancer (BTC) patients recur or die within two years. Gemcitabine and cisplatin (GC) has been the established chemotherapy regimen in advanced BTC, providing a strong rationale for its evaluation in the adjuvant setting. [Slide 2] Background Despite surgery and adjuvant capecitabine, about 50% of biliary tract cancer (BTC) patients recur or die within two years. Gemcitabine and cisplatin (GC) has been the established chemotherapy regimen in advanced BTC, providing a strong rationale for its evaluation in the adjuvant setting. Methods ACTICCA-1 is a randomized, phase III investigator-initiated trial. Patients were enrolled at 60 sites across 7 countries between September 2017 and January 2023 and randomized to adjuvant GC (arm A) or capecitabine (arm B), both for 24 weeks. The primary endpoint is disease-free survival (DFS) in the full analysis set (FAS). Secondary endpoints include overall survival, safety and tolerability of chemotherapy, quality of life, and patterns of disease recurrence. Results At the data cut-off date of Dec 31, 2025, 591 patients had been randomized. Median follow-up was 4.8 years. Median age was 63, 46.9% were female, and median interval between surgery and enrolment was 67 days. The location was intrahepatic in 28.3%, perihilar in 19%, distal in 36.5%, and gallbladder in 16.2%. N+ was 47.9%. DFS favoured arm A compared to arm B (median 2.14 vs. 1.49 years, log rank p=0.097). As pre-defined in the statistical analysis plan (SAP), due to violation of the proportional hazard assumption Cox model (HR 0.82, 95% CI 0.67-1.01) was not feasible and restricted mean survival time (RMST) was analysed. Patients in arm A experienced on average 2.6 additional DFS months compared to arm B (RMST in the first 3 years p=0.01) and a strong trend in overall survival (p=0.076). Toxicity showed a slightly higher rate of severe adverse events (130 vs. 113 with 51 vs. 47 treatment related) and adverse events rate (74.4% vs. 69.5% of patients) in arm A compared to B. Further endpoints will be displayed at the meeting. Conclusions GC is an adjuvant treatment option for BTC. Clinical trial identification NCT02104882; 2012-005078-70. Legal entity responsible for the study University Medical Center Hamburg Eppendorf.

[Slide 2] Patient Allocation Randomisation 1:1 N=601 Allocated to CisGem N=301 • 297 received allocated intervention • 4 recurrence at inclusion Analysed N=297 Discontinuation reason at cut-off • 96 death • 15 lost to follow-up • 14 withdrawal of consent • 6 patient decision • 3 choice of investigator • 8 other reason • 72 no reason provided Allocated to Capecitabine N=300 • 294 received allocated intervention • 6 recurrence at inclusion Analysed N=294 Discontinuation reason at cut-off • 102 death • 13 lost to follow up • 1 SAE reason • 4 withdrawal of consent • 3 patient decision • 6 choice of investigator • 13 other reason • 81 no reason provided • Recruitment Sep. 2017 until Jan. 2023 • Data cut-off Dec. 31st in 2025 • Median follow-up 4.8 years (95% CI 4.6-4.9) • 591 patients evaluated as modified ITT population (i.e. all randomized patients regardless of treatment received) Henning Wege, LBA21, ACTICCA-1 [ACTICCA-1 logo] [UKE Universitätsklinikum Hamburg-Eppendorf logo] [ESMO Congress Madrid 2026 logo] [Slide 3] Patient Characteristics | Stratification parameters Columns: Group A CisGem, N=297 | Group B Capecitabine, N=294 | Total, N=591 Age in years [range] | 63.0 [27-82] | 63.0 [26-83] | 63.0 [26-83] Male (%) Male | 154 (51.9) | 160 (54.4) | 314 (53.1) ECOG 0 | 172 (58.3) | 179 (58.0) | 342 (58.2) ECOG 1 | 120 (40.7) | 119 (40.6) | 239 (40.6) Localisation (%) Intrahepatic | 85 (28.6) | 82 (27.9) | 167 (28.3) Perihilar | 55 (18.5) | 57 (19.4) | 112 (19.0) Distal | 109 (36.7) | 107 (36.4) | 216 (36.5) Gallbladder | 48 (16.2) | 48 (16.3) | 96 (16.2) T-Status (%) T1 | 49 (16.9) | 52 (17.9) | 101 (17.4) T2 | 134 (46.2) | 141 (48.5) | 275 (47.3) T3 | 95 (32.8) | 90 (30.9) | 185 (31.8) T4 | 11 (3.8) | 6 (2.1) | 17 (2.9) R-Status (%) R0 | 201 (67.7) | 199 (67.7) | 400 (67.7) Lymph node (%) Negative | 155 (52.2) | 153 (52.0) | 308 (52.1) Time since surgery (%) <2 months | 121 (40.7) | 110 (37.4) | 231 (39.1) Henning Wege, LBA21 [ACTICCA-1 logo] [UKE logo] [ESMO Congress Madrid 2026 logo]

[Slide 2] Subgroup Analysis • RMST DFS (3 years) of subgroups showed differences with CIs excluding 0* for male, ECOG 1, T3-4, distal CCA, lymph node positive, R0. • In exploratory interaction testing, there were no significant signals of potential effect heterogeneity. Legend: Group A CisGem | Group B Capecitabine Columns: Subgroup | Group A events/n(%) | Group A RMST | Group B events/n(%) | Group B RMST | Difference RMST | Favours CisGem | interaction p-value Sex (interaction p 0.614) f | 83/143 (58.0%) | 23.8 (22.0-25.8) | 76/134 (56.7%) | 21.9 (19.8-24.1) | 2.0 (-0.9-4.8) m | 93/154 (60.4%) | 22.7 (20.9-24.7) | 113/160 (70.6%) | 19.6 (17.8-21.6) | 3.1 (0.5-5.8) * Age (0.825) ≤63 | 100/151 (66.2%) | 22.3 (20.5-24.2) | 102/150 (68.0%) | 19.7 (17.9-21.8) | 2.6 (-0.1-5.3) >63 | 76/146 (52.1%) | 24.4 (22.5-26.4) | 87/144 (60.4%) | 21.5 (19.6-23.7) | 2.8 (0.0-5.6) Type of surgery (0.535) Open | 146/231 (63.2%) | 22.5 (21.0-24.1) | 143/219 (65.3%) | 20.2 (18.6-22.0) | 2.3 (0.0-4.6) Laparoscopic | 23/52 (44.2%) | 26.7 (23.8-29.9) | 33/58 (56.9%) | 23.4 (20.5-26.7) | 3.3 (-1.0-7.6) Robotic | 7/13 (53.8%) | 22.9 (17.0-30.7) | 13/16 (81.2%) | 15.7 (11.0-22.4) | 7.1 (-1.6-15.9) ECOG status (0.085) 0 | 101/172 (58.7%) | 23.1 (21.4-25.0) | 103/170 (60.6%) | 22.0 (20.2-24.0) | 1.1 (-1.5-3.7) 1 | 73/120 (60.8%) | 23.6 (21.6-25.7) | 83/119 (69.7%) | 19.0 (16.9-21.3) | 4.6 (1.6-7.6) * T classification (0.511) T1-2 | 93/183 (50.8%) | 25.5 (23.8-27.2) | 106/193 (54.9%) | 23.5 (21.8-25.4) | 2.0 (-0.5-4.5) T3-4 | 78/106 (73.6%) | 19.2 (17.0-21.6) | 79/96 (82.3%) | 15.2 (13.1-17.7) | 4.0 (0.7-7.2) * L classification (0.709) L0 | 52/117 (44.4%) | 27.1 (25.1-29.2) | 58/116 (50.0%) | 24.3 (22.2-26.7) | 2.7 (-0.3-5.8) L1 | 87/121 (71.9%) | 19.6 (17.6-21.8) | 97/127 (76.4%) | 16.8 (14.9-19.1) | 2.7 (-0.2-5.7) Grading (0.298) G1-G2 | 105/177 (59.3%) | 23.4 (21.6-25.2) | 107/170 (62.9%) | 21.5 (19.7-23.5) | 1.9 (-0.7-4.4) G3-G4 | 45/76 (59.2%) | 22.4 (19.8-25.4) | 61/86 (70.9%) | 18.0 (15.5-20.8) | 4.5 (0.7-8.3) Type of cancer (0.710) intrahepatic (CCA) | 41/85 (48.2%) | 25.5 (23.1-28.1) | 49/82 (59.8%) | 22.2 (19.6-25.1) | 3.3 (-0.3-7.0) perihilar / Klatskin (CCA) | 33/55 (60.0%) | 23.3 (20.4-26.6) | 34/57 (59.6%) | 22.5 (19.5-25.9) | 0.8 (-3.6-5.2) distal extrahepatic (CCA) | 75/109 (68.8%) | 20.9 (18.8-23.2) | 81/107 (75.7%) | 17.2 (15.1-19.6) | 3.7 (0.5-6.8) * muscle invasive gallbladder carcinoma | 27/48 (56.2%) | 24.5 (21.5-28.0) | 25/48 (52.1%) | 23.4 (20.1-27.2) | 1.1 (-3.6-5.9) Lymph node status (0.899) Negative | 74/155 (47.7%) | 26.4 (24.7-28.3) | 80/153 (52.3%) | 24.3 (22.4-26.4) | 2.2 (-0.5-4.9) Positive | 102/142 (71.8%) | 19.8 (17.9-21.9) | 109/141 (77.3%) | 16.6 (14.7-18.8) | 3.2 (0.3-6.0) * Tumor resection (0.109) R0 | 100/201 (49.8%) | 25.7 (24.1-27.4) | 118/199 (59.3%) | 22.0 (20.3-23.9) | 3.7 (1.3-6.2) * R1 | 76/96 (79.2%) | 18.4 (16.2-20.8) | 71/95 (74.7%) | 17.4 (15.1-20.0) | 1.0 (-2.3-4.3) Forest axis 0 5 10 15 Henning Wege, LBA21, ACTICCA-1 [ACTICCA-1 logo] [UKE Universitätsklinikum Hamburg-Eppendorf logo] [ESMO Congress Madrid 2026 logo] [Slide 4] Summary & Conclusions • ACTICCA-1 is the largest RCT in adjuvant BTC comparing two active regimens (CisGem vs. capecitabine). • CisGem did not significantly improve DFS at 24 months (52% vs. 44%, Δ 8%, Cox HR 0.82, p=0.058) and no significant interaction was noted in the subgroup analyses. • Despite a difference in median OS of nearly 10 months (58.1 vs. 48.2 months), the secondary endpoint OS did not show a significant improvement for CisGem compared to capecitabine (Cox HR 0.82, p=0.103). • In the first 3 years, CisGem significantly increased DFS by RMST analysis with Δ 2.6 months (p=0.011). • As expected, CisGem demonstrated an increased (and mainly haematological) toxicity. CisGem should not be considered as a standard of care, however is an adjuvant treatment option for patients with BTC. Henning Wege, LBA21, ACTICCA-1 [ACTICCA-1 logo] [UKE Universitätsklinikum Hamburg-Eppendorf logo] [ESMO Congress Madrid 2026 logo]

[Slide 1] Kaplan-Meier curve for the primary endpoint Disease-free survival Legend: CisGem; Capecitabine Disease-free survival probability / Time since randomization [years] 0-5 CisGem: Number at risk 297 191 140 110 75 35; Cumulative number censored 0 20 24 29 52 87 Capecitabine: Number at risk 294 169 122 101 72 30; Cumulative number censored 0 14 15 19 41 75 [Survival data captured as a table — no point estimates read from the curve] Endpoint: DFS (primary endpoint) CisGem | 297 Capecitabine | 294 Note: Figure prints no median, HR or p-value; only curve + number at risk (years 0-5).

[Slide 2] [Left: video of presenter at podium, ESMO congress Madrid 2026 logos] [ESMO Congress Madrid 2026 logo] MASTERPLAN: randomised phase II trial KEY ELIGIBILITY CRITERIA • Age ≥18 • ECOG PS 0-1 • One of: ➢ Operable disease with ≥1 high-risk feature ➢ Borderline resectable ➢ Locally advanced Open-label multicentre trial 14 hospitals R (2:1) Australia & New Zealand CHEMOTHERAPY ALONE: 3 months chemotherapy CHEMOTHERAPY + SBRT: 3 months chemotherapy + SBRT 40 Gy in 5 fractions → +/- SURGERY → ADJUVANT/ONGOING CHEMOTHERAPY PRIMARY ENDPOINT: 12-month locoregional recurrence (LRR) SECONDARY ENDPOINTS: PFS, OS, toxicity, (R0) resection rate, post operative complications NCT04089150 • ACTRN12619000409178 • Recruitment: Feb 2020–Jan 2025 4 [Slide 3] B Median overall survival, months (95% CI) Chemotherapy plus SBRT 22·0 (17·5–34·7) Chemotherapy alone 14·0 (8·9–19·2) HR 0·45 (0·25–0·81); logrank p=0·0061 Y-axis: Overall survival (%); X-axis: Time from randomisation (months) Number at risk (censored) at 0, 3, 6, 9, 12, 15, 18, 21, 24, 27, 30 months: Chemotherapy plus SBRT: 49 (0), 47 (2), 47 (2), 45 (2), 37 (2), 35 (3), 26 (6), 20 (8), 17 (9), 16 (9), 14 (10) Chemotherapy alone: 25 (0), 23 (0), 21 (0), 17 (0), 16 (0), 12 (0), 7 (2), 5 (3), 4 (4), 2 (4), 1 (4) Figure 3: Kaplan–Meier curve for progression-free survival (A) and overall survival (B) for the overall population receiving chemotherapy plus SBRT and those receiving chemotherapy alone The p value for overall survival is descriptive. HR=hazard ratio. SBRT=stereotactic body radiotherapy. [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival Chemotherapy plus SBRT | 49 | 22.0 mo | 17.5-34.7 Chemotherapy alone | 25 | 14.0 mo | 8.9-19.2 HR: 0.45 HR CI: 0.25-0.81 P: 0.0061 (logrank, descriptive) Note: Journal figure (Lancet Oncology, in press per caption). OS p value is descriptive per figure legend; OS is a secondary endpoint (primary = 12-month LRR). [Slide 4] Perioperative chemotherapy and stereotactic body radiotherapy versus perioperative chemotherapy alone in patients with high-risk, localised pancreatic cancer (MASTERPLAN): an open-label, phase 2, multicentre, randomised trial of the Australasian Gastro-Intestinal Trials Group Andrew Oar, Andrew Kneebone, Katrin M Sjoquist, David Espinoza, Mark Lee, Hien Le, Leily Gholamrezaei, Julie Chu, James Lynam, Girish Mallesara, Dominique Lee, Dinesh Vignarajah, Iain Ward, Jaswinder Samra, Andrew Barbour, Angela Whitehead, Susan Brooks, Anna Nicholson, Tim Wei Wang, Weng Ng, Fathalla Ali, Derrick Siu, Juanita Lopez Gaitan, Olivia Cook, Alisha Moore, Martinus B Oostendorp, Sonia Yip, John Shakeshaft, David Pryor, Nam Nguyen, David Goldstein, on behalf of the Australasian Gastro-Intestinal Trials Group (AGITG) Summary

[Slide 1] 3136RO - Randomised phase II study of mFOLFIRINOX ± SBRT in patients with high risk, borderline resectable or locally advanced pancreatic cancer: Results of the AGITG MASTERPLAN trial Speakers Andrew Oar (Southport, Australia) Lecture Time 17:14 - 17:19 Abstract Background Stereotactic body radiotherapy (SBRT) delivers high-dose, precision radiotherapy (RT). Despite advances in chemotherapy (CT) for pancreatic cancer (PC), locoregional recurrences (LRR) are frequent and contribute to morbidity/mortality. Methods Eligible pts had operable disease with ≥1 high risk feature (>4 cm, extrapancreatic extension, or node positive), or BRPC, or LAPC. High-risk operable and BRPC pts were classified as potentially operable PC (POPC), while LAPC/medically inoperable pts as inoperable PC (IPC). Staging followed AGITG guidelines with central radiology review. CT was clinician choice: mFOLFIRINOX or gem/nab-paclitaxel. Pts were randomised 2:1 to 3mths of CT+SBRT (I) or to 3mths CT alone (C). SBRT was 40Gy in 5 fractions with fiducial marker guidance and real time RT quality assurance. After CT±SBRT, pts were evaluated for surgery. Following resection pts received 3mths adjuvant CT. Pts unsuitable for resection, continued CT for further 3mths. The primary endpoint was 12mth LRR. Secondary endpoints included resection rate, margin status, toxicity per CTCAE (v5.0), progression-free survival (PFS), and overall survival (OS). Results Across 14 sites 75 pts were enrolled with 1 ineligible, including 48 POPC and 26 IPC. 12mth LRR rates were 20.2 % (95% CI: 9.9-33.2%) v 52.0% (95% CI: 30.3-69.9%) in I and C respectively (p=0.022). In POPC, 12mth LRR rates were 19.7% (95% CI: 7.8–35.6) (I) v 52.9% (95% CI: 25.9–74.1) (C) (p=0.031), and in IPC 21.4% (95% CI: 4.8–45.8) (I) v 50.0% (95% CI: 11.4-80.2) (C) (p=0.16). Overall Grade 3+ toxicity occurred in 70.2% (I) and 64.0% (C) (p=0.27). Median PFS was 11.9 v 6.3 mths (p=0.002) and median OS 22.0 v 14.0 mths (p=0.006) in I and C, respectively. Table: 3136RO Clinicodemographics and outcomes | | C=Control (n=25) | I=Investigational (n=49) | P value | | Demographics | | | | | Median age (IQR) | 66 (55-70) | 65 (58-70) | 0.89 | | % Male | 72% | 51% | 0.08 | | Baseline Ca19.9 (median, IQR) | 340 (60-977) | 180 (34-1152) | 0.60 | | Staging | - | - | | | Operable with ≥1 high-risk feature | 3 (12%) | 1 (2.0%) | - | | BRPC | 14 (56%) | 30 (61%) | - | | IPC | 8 (32%) | 18 (37%) | 0.69 | | cN0 | 60% | 71% | 0.32 | | Treatment | | | | | CT regimen | - | - | - | | mFOLFIRINOX | 23 (92%) | 44 (90%) | - | | Gemcitabine/nab-paclitaxel | 2 (8%) | 3 (6%) | 1.0 | | Withdrawal prior to chemotherapy | 0 (0%) | 2 (4%) | | | SBRT commenced | NA | 39 (80%) | NA | | Surgical Outcomes (POPC only; n=48) | C=Control (n=17) | I=Investigational (n=31) | | | Resection performed n (%) | 10 (59%) | 21 (68%) | 0.54 | | R0 resection (>1mm) | 4/10 (40%) | 13/21 (62%) | 0.44 | Conclusions SBRT improved LRR and PFS without additional toxicity. A significant OS signal was observed in this cohort. Further research to optimise the role of SBRT is required.

[Slide 2] [Madrid 2026 ESMO Congress logo] Primary outcome: lower 12-month locoregional recurrence (LRR) [Cumulative incidence plot: LRR Cumulative Incidence vs Time from randomisation - months; curves Chemotherapy, Chemotherapy + SBRT] HR: 0.44 (95% CI: 0.22-0.88, p(Gray's)=0.02) Number at risk (0, 3, 6, 9, 12, 15, 18, 21, 24 months) Chemotherapy 25 20 15 7 4 3 2 1 1 SBRT 49 44 38 31 22 17 13 11 10 12-month LRR 20.2% SBRT + chemotherapy 95% CI 9.9–33.1% 12-month LRR 52.0% Chemotherapy alone 95% CI 30.3–69.9% Treatment effect HR 0.44 95% CI 0.22–0.88; p=0.022 Absolute difference ~32% at 12 months Primary endpoint: ITT analysis. LRR includes synchronous LRR + distant progression. Sensitivity analysis for initial/isolated LRR: HR 0.35; p=0.044 [Survival data captured as a table — no point estimates read from the curve] Endpoint: 12-month locoregional recurrence (cumulative incidence) Chemotherapy + SBRT | 49 Chemotherapy alone | 25 HR: 0.44 HR CI: 0.22–0.88 P: 0.022 (Gray's) {'timepoint': '12 months', 'arm': 'Chemotherapy + SBRT', 'value': '20.2% LRR', 'ci': '9.9–33.1%'} {'timepoint': '12 months', 'arm': 'Chemotherapy alone', 'value': '52.0% LRR', 'ci': '30.3–69.9%'} Note: Cumulative incidence (not survival). n from at-risk at time 0. Sensitivity (initial/isolated LRR) HR 0.35, p=0.044. [Slide 3] [Madrid 2026 ESMO Congress logo] PFS and OS: both favoured the addition of SBRT * PFS: 11.9 vs 6.3 mths HR 0.44 (95% CI 0.26–0.75); (p=0.002) [in-plot] HR: 0.44 (95% CI: 0.26-0.75, p(logrank)=0.002) Number at risk (0, 3, 6, 9, 12, 15, 18, 21, 24 months) Chemotherapy+SBRT 49 44 38 31 23 19 15 12 10 Chemotherapy 25 20 15 7 5 4 2 1 1 OS: 22.0 vs 14.0 mths HR 0.45 (95% CI 0.25–0.81); (p=0.006) [in-plot] HR: 0.45 (95% CI: 0.25-0.81, p(logrank)=0.006) Number at risk (0, 3, 6, 9, 12, 15, 18, 21, 24, 27, 30 months) Chemotherapy+SBRT 49 47 47 45 37 35 26 20 17 16 14 Chemotherapy 25 23 21 17 16 12 7 5 4 2 1 * Secondary endpoint 8 [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS and OS (secondary endpoints) Chemotherapy + SBRT (PFS) | 49 | 11.9 mo Chemotherapy (PFS) | 25 | 6.3 mo Chemotherapy + SBRT (OS) | 49 | 22.0 mo Chemotherapy (OS) | 25 | 14.0 mo HR: PFS 0.44; OS 0.45 HR CI: PFS 0.26–0.75; OS 0.25–0.81 P: PFS 0.002; OS 0.006 (log-rank) Note: Secondary endpoints; conclusions slide says OS improvement requires further evaluation in adequately powered trials. [Slide 4] CHAIRS: MARIA IGNEZ BRAGHIROLI, FENG WANG, LUCJAN WYRWICZ [Madrid 2026 ESMO Congress logo] Conclusions • MASTERPLAN first prospective randomised trial showing significant benefit of SBRT in high risk localised PDAC • SBRT significantly reduced LRR and improved PFS • SBRT was safe with a low incidence of Grade ≥3 toxicity • Improved OS was observed in MASTERPLAN, however requires further evaluation in adequately powered randomised trials. • Future studies should explore biomarkers, novel systemic treatment agents, treatment intensification including SBRT in patients with high risk localised PDAC 10

[Slide 3] [ESMO Congress Madrid 2026 logo] PFS and OS: both favoured the addition of SBRT * PFS: 11.9 vs 6.3 mths HR 0.44 (95% CI 0.26–0.75); (p=0.002) [on plot] HR: 0.44 (95% CI: 0.26-0.75, p(logrank)=0.002) Number at risk (0,3,6,9,12,15,18,21,24 months): Chemotherapy+SBRT 49 44 38 31 23 19 15 12 10 Chemotherapy 25 20 15 7 5 4 2 1 1 OS: 22.0 vs 14.0 mths HR 0.45 (95% CI 0.25–0.81); (p=0.006) [on plot] HR: 0.45 (95% CI: 0.25-0.81, p(logrank)=0.006) Number at risk (0,3,6,9,12,15,18,21,24,27,30 months): Chemotherapy+SBRT 49 47 47 45 37 35 26 20 17 16 14 Chemotherapy 25 23 21 17 16 12 7 5 4 2 1 * Secondary endpoint 8 [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS (secondary) Chemotherapy + SBRT | 49 | 11.9 mo Chemotherapy alone | 25 | 6.3 mo HR: 0.44 HR CI: 0.26-0.75 P: 0.002 Note: OS (secondary): 22.0 vs 14.0 mo, HR 0.45 (95% CI 0.25-0.81), p=0.006 (logrank). Both PFS and OS are secondary endpoints.

[Slide 3] [Madrid 2026 ESMO Congress logo] Safety and surgical outcomes: no clear increase in clinically important toxicity Grade ≥3 adverse events 70.2% vs 64.0% SBRT + chemotherapy vs chemotherapy alone; p=0.59 SBRT-attributed serious toxicity 1 patient One grade ≥3 SAE attributed to SBRT (pancreatitis) Grade ≥3 GI events None reported No grade ≥3 perforation or GI haemorrhage with SBRT Surgical outcomes Outcome | Chemotherapy + SBRT | Chemotherapy alone Surgical exploration | 29/47 (62%) | 11/25 (44%) Resection - POPC + IPC | 25/47 (53%) | 11/25 (44%) R0 resection – POPC | 13/21 (62%) | 4/10 (40%) 30-day postoperative complications | 4/25 (16.7%) | 4/11 (36.4%) Safety population: all treated participants. Resection figures exclude open-and-close procedures.

[Slide 1] [Madrid 2026 ESMO congress logo] COMPANION-002: Phase 3 U.S. Biliary Tract Cancer (BTC) Study KEY ELIGIBILITY Advanced, metastatic or unresectable BTC One prior gemcitabine/platinum containing regimen for advanced disease ≥1 measurable lesion by RECIST v1.1 ECOG PS 0-1 Excluded known actionable mutations if available targeted therapy for second-line R (2:1) N=168 28-day cycles Tovecimig (10 mg/kg) Days 1 & 8 + Paclitaxel (80 mg/m2) Days 1, 8 & 15 N= 111 → Follow-up Q 3 months 28-day cycles Paclitaxel (80 mg/m2) Days 1, 8 & 15 N=57 → Crossover permitted upon BICR confirmed PD → 28-day cycles Tovecimig (10 mg/kg) Days 1 & 8 + Paclitaxel (80 mg/m2) Days 1, 8 & 15 N=31; Follow-up Q 3 months STRATIFICATION FACTORS • Tumor Location - Intrahepatic vs others (extrahepatic, gallbladder or ampullary) • ECOG PS (0 vs 1) • Disease Status - Locally advanced vs metastatic PRIMARY OBJECTIVE • Objective Response Rate (ORR) by blinded independent central review (BICR) SECONDARY OBJECTIVES • Progression free survival (PFS) per BICR • Disease control rate (DCR) • Overall survival (OS) • Duration of response (DoR) • Progression free survival 2 (PFS2) • Safety profile • Quality of Life (EORTC QLQ-C30 and BIL21) Nilofer Azad, MD [Slide 2] [Madrid 2026 ESMO congress logo] COMPANION-002: Tovecimig + Paclitaxel Significantly Improves ORR and DCR Tovecimig + paclitaxel responses were durable: median DoR 4.9 months n (%) | Tovecimig + Paclitaxel (n=111) | Paclitaxel (n=57) ORR (CR + PR) Yes | 20 (18.0) | 3 (5.3) p-value | 0.0228 BOR CR | 1 (0.9) | 0 PR | 19 (17.1) | 3 (5.3) SD | 49 (44.1) | 18 (31.6) Non-CR/non-PD* | 8 (7.2) | 2 (3.5) PD | 18 (16.2) | 25 (43.9) NE** | 16 (14.4) | 9 (15.8) DCR (CR + PR + SD) | 69 (62.2) | 21 (36.8) p-value | 0.0018 [Waterfall plot: Change from Baseline (%) by Patient; legend Tovecimig + Paclitaxel, Paclitaxel] *Non-CR / Non-PD: patients enrolled based on local radiology scan results but displayed no clearly definable target lesions as determined by independent central radiology. ** Not Evaluable: patients who did not receive a Week-8 scan; these patients are not evaluable for response only but will be evaluable for PFS/OS analyses. Nilofer Azad, MD [Slide 3] [Madrid 2026 ESMO congress logo] COMPANION-002: Tovecimig + Paclitaxel Significantly Improves PFS (BICR) Legend: + Censored; Tovecimig + Paclitaxel; Paclitaxel Median PFS, months (95% CI) Tovecimig + Paclitaxel: 4.7 (3.9-5.6) Paclitaxel: 2.6 (1.9-3.6) HR, 0.44 (95% CI, 0.30-0.66) p < .0001 Time (months) Number at Risk (Censored) Tovecimig + Paclitaxel: 111(0) 73 (10) 30 (11) 11 (10) 8 (2) 5 (1) 3 (1) 2 (1) 1 (1) 1 (0) Paclitaxel: 57 (0) 25 (4) 9 (1) 4 (1) 2 (0) 2 (1) Nilofer Azad, MD [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS (BICR) Tovecimig + Paclitaxel | 111 | 4.7 months | 3.9-5.6 Paclitaxel | 57 | 2.6 months | 1.9-3.6 HR: 0.44 HR CI: 0.30-0.66 P: <.0001 Note: No landmark rates printed. [Slide 4] [Madrid 2026 ESMO congress logo] COMPANION-002: OS Analysis Confounded by Crossover Paclitaxel alone survival curve includes 31 patients (54%) who crossed over and received tovecimig Legend: + Censored; Tovecimig + Paclitaxel; Paclitaxel Median OS, months (95% CI) Tovecimig + Paclitaxel: 8.9 (7.0-11.2) Paclitaxel: 9.4 (6.5-13.5) HR, 1.05 (95% CI, 0.7-1.5) p = .779 Time (months) Number at Risk (Censored) Tovecimig + Paclitaxel: 111 (0) 95 (2) 72 (1) 53 (0) 34 (1) 26 (0) 22 (1) 13 (6) 9 (3) 5 (4) 2 (4) Paclitaxel: 57 (0) 53 (1) 39 (2) 27 (0) 22 (0) 19 (0) 9 (2) 6 (3) 2 (2) 1 (1) Nilofer Azad, MD [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS Tovecimig + Paclitaxel | 111 | 8.9 months | 7.0-11.2 Paclitaxel | 57 | 9.4 months | 6.5-13.5 HR: 1.05 HR CI: 0.7-1.5 P: .779 Note: Paclitaxel arm includes 31 (54%) crossover patients who received tovecimig. Paclitaxel CI partly overlapped by highlight box ('6.5-13.5' read).

[Slide 4] [ESMO Congress Madrid 2026 logo] Phase 2/3 COMPANION-002 Conclusions • Tovecimig + paclitaxel met the primary endpoint: ORR 18.0% vs 5.3% (p=0.0228), with durable responses (median DOR 4.9 months) • Disease control rate was also significantly improved: 62.2% vs 36.8% (p=0.0018) • PFS was significantly prolonged: 4.7 vs 2.6 months (HR=0.44, p<0.0001) • Overall survival may have been confounded by crossover, with 54% of paclitaxel patients receiving tovecimig • Addition of tovecimig to paclitaxel in cross-over setting doubled first line paclitaxel only PFS (3.9 vs 1.5 months, HR=0.40, p=0.0035) and those crossover patients survived for a median of 12.8 months. • Safety was manageable and consistent with prior studies and the anti-VEGF class, with no new safety signals • Overall, 28.8% of patients treated with tovecimig + paclitaxel discontinued the treatment regimen due to an adverse event; most TEAEs leading to discontinuation were consistent with known anti-angiogenic and chemotherapeutic toxicities. • Tovecimig + paclitaxel is a promising option for patients with BTC who have been previously treated Nilofer Azad, MD

[Slide 1] In summary, this might be what we were looking for Question | My take | My Comments Is this study promising? | YES | • Promising MoA and targets • Low bar in the second line setting Did it meet its primary endpoint? | YES | • ORR 18% vs 5.3% (p=0.0228) • Clear signs of clinical activity PFS HR 0.44 • We need to identify patients who benefit the most Will it change practice? | ? | • We would want a phase III trial with FOLFOX as control arm demonstrating a clear OS benefit • Given the high unmet need of these patients, a formal review by a relevant guideline committee is warranted Sebastián Mondaca, MD, MBA

[Slide 1] [ESMO Congress Madrid 2026 logo] Tovecimig: Next-generation Angiogenesis Disrupting Bispecific Antibody [Antibody schematic: VEGF-A and DLL4 binding arms; 2x2 target binding valency] Anti-VEGF-A MAb — VEGF/VEGF-R signaling — Disrupts tumor vessel formation (proven anti-angiogenic mechanism) Anti-DLL4 MAb — DLL4-Notch-1 signaling — Targets tumor DLL4 expression and alters perfusion in tumor vessels (novel angiogenesis target) • Dual blockade: VEGF-A – validated target for blockbuster oncology therapeutics (e.g.: Avastin®) DLL4 (Notch-1 ligand) – mediates resistance to anti-VEGF therapies • Bispecific anchors in tumor microenvironment (DLL4) to disrupt angiogenesis • Only DLL4 × VEGF bispecific to demonstrate monotherapy activity in patients with CRC and GC¹ • Preclinical combination activity and an encouraging early BTC response signal supported further development of tovecimig + paclitaxel in Phase 2; the same dose and schedule were subsequently carried forward into COMPANION-002. Nilofer Azad, MD ¹ Lee, J et al. 2021, October 7-10. Plenary Presentation AACR-NCI-EORTC Virtual International Conference on Molecular Targets and Cancer Therapeutics 2 Oh DY, et al. ASCO Gastrointestinal Cancers Symposium, January 19-21, 2023. Abstract 54[illegible]

[Slide 2] COMPANION-002: A Randomized Phase 2/3 Trial of Tovecimig Plus Paclitaxel versus Paclitaxel Alone as a Second-line Therapy in Patients with Advanced Biliary Tract Cancer COMPANION-002: randomized, open-label phase 2/3 trial (34 US sites) Advanced BTC progressed on 1L gemcitabine + platinum N = 168 → R 2:1 Tovecimig 10 mg/kg IV D1, D15 + paclitaxel 80 mg/m² D1, D8, D15, q28d (n = 111) Crossover after BICR-confirmed PD: 31/57 (54%) Paclitaxel alone 80 mg/m² D1, D8, D15, q28d (n = 57) Primary endpoint: ORR by BICR (RECIST v1.1) • Key secondary endpoints: PFS, OS, DCR Objective response rate (primary) P = .0228: 18.0% | 5.3% Disease control rate P = .0018: 62.2% | 36.8% Progression-free survival (BICR): Median 4.7 vs 2.6 mo; HR 0.44 (95% CI 0.30–0.66); P < .0001 No. at risk: 111 30 8 3 1 | 57 9 2 Months since randomization Target lesion response [waterfall; 204%; +20% (PD); −30% (PR)]; Individual patients (T+P, n = 88; P, n = 47) Tovecimig + paclitaxel (T+P) | Paclitaxel alone (P) Relevance: The combination of tovecimig, a novel bispecific antibody targeting DLL4/VEGFA, and taxane chemotherapy demonstrated interesting activity in previously treated biliary cancers. Development continues in treatment-naiive biliary cancer.* Relevance statement written by JCO Associate Editor Eileen M. O'Reilly, MD, FASCO. Fig. Design and key efficacy outcomes of COMPANION-002 (NCT05506943), a randomized, open-label phase 2/3 trial. Patients with advanced BTC that progressed on one prior gemcitabine/platinum-based regimen were randomized 2:1 to tovecimig plus paclitaxel (n = 111) or paclitaxel alone (n = 57); crossover was permitted after BICR-confirmed progression. ORR, DCR, and PFS were assessed by BICR per RECIST v1.1 in the ITT population; P values are two-sided. Waterfall plot shows maximum percentage change from baseline in target lesion size (evaluable patients); dashed lines indicate RECIST v1.1 thresholds for PR (−30%) and PD (+20%). 1L, first-line; BICR, blinded independent central review; BTC, biliary tract cancer; D, day; DCR, disease control rate; HR, hazard ratio; ITT, intent-to-treat; ORR, objective response rate; OS, overall survival; PD, progressive disease; PR, partial response; PFS, progression-free survival; q28d, every 28 days; R, randomization. Azad et al. (2026) | Journal of Clinical Oncology | DOI: 10.1200/JCO-26-02188 Journal of Clinical Oncology An American Society of Clinical Oncology Journal

[Slide 1] Interim Analysis (15 Jan 2026): 25 Pts with 1 year follow-up and completed 12-Month evaluation (12M) [Pie chart:] cCR at 1 year n=14 (61%); Surgery n=9 (36%); Not treated n=1; Not evaluated at 1 Y n=1 TRG Becker (n=9) TRG 1a: no residual tumor 0 TGR 2: 10% to 50% residual tumor cells 3 (33%) TRG 3: >50% residual tumor cells 6 (67%) •25 Pts enrolled; 1 Pt excluded (Biermer disease, no dostarlimab treatment). •23 Pts reached 1 year evaluation: 14/23 (61%) achieved cCR. •1 Pt after the interim analysis achieved cCR at 1 year • Overall cCR rate: 15/24 (62,5%). •9/24 Pts underwent surgery. •Median cumulative dostarlimab dose during the first 6 months: 4,000 mg. Interim endpoint met: ≥8 Pts (33.3%) achieved cCR at 1 year; study to continue. Thierry André [ESMO Congress Madrid 2026 logo] [Slide 2] DEWI phase II: Safety and Outcomes Planned Interim Analysis (15 Jan 2026): 25 Pts with at least 1 year follow-up Median follow-up = 14.06 months in Feb 2026 SAFETY (N=24) •Grade 3-4 dostarlimab-related AEs: 21%. •Treatment interruption for SAE (n=5, 4 Pts had surgery and one had cCR after 8 dostarlimab doses as administered intravenously) •Surgery performed in 9/24 Pts: • 4 fistulas including 2 esophagogastric anastomotic fistulas • 1 postoperative death related to a fistula complications. DISEASE CONTROL AND OVERALL SURVIVAL •1 Pts with metastatic disease progression (sus clav nodes after surgery) •1 death: 1 postoperative death related to a fistula complications. [ESMO Congress Madrid 2026 logo] [Slide 3] At Analysis with data cutoff on 1 Sep 2026: 43 Pts treated and with 1 year follow-up and completed 1 year evaluation Median follow-up: 14.9 months (Q1=14 months, Q3=15.7 months) [Pie chart:] cCR at 1 year with organ preservation n=29; Surgery n=14 (32.5%) TRG Becker (n=14) TRG 1a: no residual tumor 1 (7%) TGR 2: 10% to 50% residual tumor cells 6 (43%) TRG 3: >50% residual tumor cells 7 (50%) Among the 29 Pts in cCR under a Watch-and-Wait approach with organ preservation - Median treatment duration: 10 months (range, 3-11). - All Pts in cCR at M6 remained in cCR at 1 year. - No relapse and no death. - One patient refuse Watch and Wait in CCR. Thierry André [ESMO Congress Madrid 2026 logo] [Slide 4] Conclusion and Future Directions • Interim analysis met primary endpoint. • About 2/3 of patients in this preliminary results with 43 patients are in CCR at 1 year with organ preservation without relapse or death. • Additional cohort of 20 surgery-unfit Pts (elderly and/or with comorbidities) currently enrolled, 1-year cCR analysis planned in 202. • Target enrollment: 89 surgery-fit and 39 surgery-unfit Pts. • Translational studies (ctDNA and transcriptomic profiling) is ongoing to identify predictive biomarkers to support early selection for surgery versus watch-and-wait management. Thierry André [ESMO Congress Madrid 2026 logo]

[Slide 1] 16:15 - 17:45 Rapid oral 1: Gastrointestinal tumour, upper digestive CHAIRS: MARIA IGNEZ BRAGHIROLI, FENG WANG, LUCJAN WYRWICZ DEWI Phase II: Study design KEY ELIGIBILITY CRITERIA • ECOG PS of 0-1, • Age ≥18 years, • Pts aged >70 years: G8 score of ≥14 and/or fit for surgery, • Non-metastatic dMMR/MSI-H gastric or GEJ adenocarcinoma, cT2-T4, Nx, M0 after: - CT of the thorax-abdomen-pelvis (TAP-CT), - Echo-endoscopy (EUS), - Optional coelioscopy. PRIMARY OBJECTIVE • To evaluate clinical complete response (cCR) at 1 year, defined as the proportion of Pts who, at 1 year after the start of treatment with dostarlimab, are: - alive, - without tumor resection, - free of disease progression (on endoscopy & Thoraco-Abdo-Pelv CT scan) - with negative biopsies - with complete or near complete at endoscopy (Grade 3 or 4) STATISTICAL ANALYSES: Interim Safety/Futility Analysis • Current design: Simon two-stage, • Planned design: Fleming three-stage, • One-sided type I error: 0.54%, power: 95%, • A total of 80 evaluable Pts with cCR status available at 1 year are required. • A cCR rate at 1 year of 50% (P1) is expected. • Interim analysis: planned after 25 Pts first pts included with 1 year of follow-up. The study will continue if ≥8 Pts (33.3%) achieve cCR at 1 year. ClinicalTrials.gov identifier: NCT06059495 EU CT Number: 2023-506102-39-00 Thierry André [GERCOR logo] [Fondazione GONO GI logo] [ESMO Congress Madrid 2026 logo] Thierry André — Watch-and-wait (W&W) strategy for dostarlimab in localized deficient dMMR/MSI-H gastroesophageal adenocarcinoma (GEA): Results from the first interim analysis of the GERCOR-GONO DEWI phase II study ALMERIA AUDITORIUM - HALL 9 [Slide 2] 16:15 - 17:45 Rapid oral 1: Gastrointestinal tumour, upper digestive CHAIRS: MARIA IGNEZ BRAGHIROLI, FENG WANG, LUCJAN WYRWICZ DEWI Phase II: Study Schedule cCR=clinical complete response; W&W=watch-and-wait; PD=disease progression Inclusion -> DOSTARLIMAB 500 mg IV q3w, 4 cycles: C1-C4, W1-W4-W7-W10 Week 12 Evaluation: (1) Biopsy(es) with or without tumoral cells and tumoral regression at endoscopy (stage 2-3) and without PD, W&W -> DOSTARLIMAB 500 mg IV q3w*, 4 cycles: C5-C8, W13-W15-W18-W21 -> Week 24 Evaluation: cCR -> DOSTARLIMAB 1000 mg IV q6w*, 4 cycles: C9-C12 -> cCR: W&W / If PD STOP DOSTARLIMAB** No cCR -> SURGERY with TGR 1a, 1b, 2 (Becker) / SURGERY with TGR 3 (Becker) -> If PD STOP DOSTARLIMAB** (2) cCR, W&W -> DOSTARLIMAB 1000 mg IV q6w*, 2 cycles: C5-C6, W13-W19 -> Week 24 Evaluation: cCR -> DOSTARLIMAB 1000 mg IV q6w*, 4 cycles: C7-C10, W25-W31-W37-W43 -> cCR: W&W / If PD STOP DOSTARLIMAB** No cCR -> SURGERY proposed -> If surgery** (3) Local/loco regional PD and biopsy(es) with tumoral cells -> SURGERY with TGR 1a, 1b, 2 (Becker) -> DOSTARLIMAB 1000 mg IV q6w*, 6 cycles: C5-C10 -> If PD STOP DOSTARLIMAB** SURGERY with TGR 3 (Becker) -> STOP DOSTARLIMAB** First Pt included: Jan 22, 2024 Thierry André [ESMO Congress Madrid 2026 logo] Thierry André — Watch-and-wait (W&W) strategy for dostarlimab in localized deficient dMMR/MSI-H gastroesophageal adenocarcinoma (GEA): Results from the first interim analysis of the GERCOR-GONO DEWI phase II study ALMERIA AUDITORIUM - HALL 9 [Slide 3] 16:15 - 17:45 Rapid oral 1: Gastrointestinal tumour, upper digestive CHAIRS: MARIA IGNEZ BRAGHIROLI, FENG WANG, LUCJAN WYRWICZ Interim Analysis (15 Jan 2026): 25 Pts with 1 year follow-up and completed 12-Month evaluation (12M) [Pie chart:] cCR at 1 year n=14 (61%); Surgery n=9 (36%); Not treated n=1; Not evaluated at 1 Y n=1 TRG Becker (n=9) TRG 1a: no residual tumor 0 TGR 2: 10% to 50% residual tumor cells 3 (33%) TRG 3: >50% residual tumor cells 6 (67%) •25 Pts enrolled; 1 Pt excluded (Biermer disease, no dostarlimab treatment). •23 Pts reached 1 year evaluation: 14/23 (61%) achieved cCR. •1 Pt after the interim analysis achieved cCR at 1 year • Overall cCR rate: 15/24 (62,5%). •9/24 Pts underwent surgery. •Median cumulative dostarlimab dose during the first 6 months: 4,000 mg. Interim endpoint met: ≥8 Pts (33.3%) achieved cCR at 1 year; study to continue. Thierry André [ESMO Congress Madrid 2026 logo] Thierry André — Watch-and-wait (W&W) strategy for dostarlimab in localized deficient dMMR/MSI-H gastroesophageal adenocarcinoma (GEA): Results from the first interim analysis of the GERCOR-GONO DEWI phase II study ALMERIA AUDITORIUM - HALL 9 [Slide 4] 16:15 - 17:45 Rapid oral 1: Gastrointestinal tumour, upper digestive CHAIRS: MARIA IGNEZ BRAGHIROLI, FENG WANG, LUCJAN WYRWICZ At Analysis with data cutoff on 1 Sep 2026: 43 Pts treated and with 1 year follow-up and completed 1 year evaluation Median follow-up: 14.9 months (Q1=14 months, Q3=15.7 months) [Pie chart:] cCR at 1 year with organ preservation n=29; Surgery n=14 (32.5%) TRG Becker (n=14) TRG 1a: no residual tumor 1 (7%) TGR 2: 10% to 50% residual tumor cells 6 (43%) TRG 3: >50% residual tumor cells 7 (50%) Among the 29 Pts in cCR under a Watch-and-Wait approach with organ preservation - Median treatment duration: 10 months (range, 3-11). - All Pts in cCR at M6 remained in cCR at 1 year. - No relapse and no death. - One patient refuse Watch and Wait in CCR. Thierry André [ESMO Congress Madrid 2026 logo] Thierry André — Watch-and-wait (W&W) strategy for dostarlimab in localized deficient dMMR/MSI-H gastroesophageal adenocarcinoma (GEA): Results from the first interim analysis of the GERCOR-GONO DEWI phase II study ALMERIA AUDITORIUM - HALL 9

[Slide 2] DEWI Phase II: Study Schedule cCR=clinical complete response W&W=watch-and-wait PD=disease progression Inclusion -> DOSTARLIMAB 500 mg IV q3w 4 cycles: C1-C4 W1-W4-W7-W10 Week 12 Evaluation: - Biopsy(es) with or without tumoral cells and tumoral regression at endoscopy (stage 2-3) and without PD W&W -> DOSTARLIMAB 500 mg IV q3w* 4 cycles: C5-C8 W13-W15-W18-W21 -> Week 24 Evaluation: cCR -> DOSTARLIMAB 1000 mg IV q6w* 4 cycles: C9-C12 -> cCR: W&W / If PD STOP DOSTARLIMAB**; No cCR -> SURGERY with TGR 1a, 1b, 2 (Becker) / SURGERY with TGR 3 (Becker) -> If PD STOP DOSTARLIMAB** - cCR W&W -> DOSTARLIMAB 1000 mg IV q6w* 2 cycles: C5-C6 W13-W19 -> cCR -> DOSTARLIMAB 1000 mg IV q6w* 4 cycles: C7-C10 W25-W31-W37-W43 -> cCR: W&W / If PD STOP DOSTARLIMAB**; No cCR -> SURGERY proposed -> If surgery** - Local/loco regional PD and biopsy(es) with tumoral cells -> SURGERY with TGR 1a, 1b, 2 (Becker) -> DOSTARLIMAB 1000 mg IV q6w* 6 cycles: C5-C10 -> If PD STOP DOSTARLIMAB**; SURGERY with TGR 3 (Becker) -> STOP DOSTARLIMAB** First Pt included: Jan 22, 2024 Thierry André [ESMO Congress Madrid 2026 logo]

[Slide 2] DEWI phase II Interim Analysis (15 Jan 2026): Pts treated with 1 year follow-up Overall Population (n=24) | n | % Male/Female | 18/6 | 75/25 Age, median (min-max) Yrs | 65 (59-71) | - ECOG PS at baseline 0/1 | 18/6 | 75/25 Tumor localisation Gastric/ Gastro-esophageal Junction | 14/9 | 62.5/37.5 Echo-Endoscopy/scanner stage uT2 | 5 | 21 uT3 | 18 | 75 uT4 | 1 | 4 uN0/uN+/uNx | 9/10/5 | 37/42/21 dMMR (Immunochemistry) | 24/24 | 100 loss of MLH1 and/or PMS2 | 22 | 92 loss of MSH2 and/or MSH6 | 2 | 8 MSI (PCR) | 12/12 | - Lynch syndrome* (Yes Germline) Yes/No (germline)/sporadic or unknown | 1/21/2 | 4/88/8 Thierry André [MADRID 2026 ESMO congress logo]

[Slide 1] OS in the ITT population Compared with chemotherapy alone, anbenitamab plus chemotherapy significantly reduced the risk of death by 61% | | Anbenitamab group | Control group | Median OS, mo, 95%CI | 19.6 (16.8,23.4) | 9.6 (7.7,13.7) HR (95% CI) 0.39 (0.28, 0.54) P value 3.97*10^[illegible exponent] [KM curves with landmark labels: Anbenitamab group 67.3% (15 mo), 56.2% (18 mo), 37.0% (24 mo); Control group 56.9% (15 mo), 25.8% (18 mo), 14.0% (24 mo)] Number at risk (events) at 0 / 3 / 6 / 9 / 12 / 15 / 18 / 21 / 24 / 27 / 30 / 33 / 36 / 39 / 42 months Anbenitamab group 125 (0) / 122 (1) / 114 (3) / 101 (3) / 82 (13) / 63 (24) / 48 (29) / 32 (38) / 19 (45) / 15 (47) / 7 (52) / 4 (55) / 2 (57) / 2 (57) / 0 (59) Control group 121 (0) / 108 (0) / 83 (1) / 63 (2) / 45 (10) / 32 (14) / 21 (17) / 13 (20) / 6 (24) / 4 (26) / 3 (27) / 3 (27) / 1 (29) / 0 (30) Ruixing Zhang [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival, ITT population Anbenitamab + chemotherapy | 125 | 19.6 mo | 16.8-23.4 Placebo + chemotherapy (control) | 121 | 9.6 mo | 7.7-13.7 HR: 0.39 HR CI: 0.28-0.54 P: 3.97*10^[illegible exponent] {'timepoint': '15 months', 'arm': 'Anbenitamab', 'value': '67.3%', 'ci': None} {'timepoint': '18 months', 'arm': 'Anbenitamab', 'value': '56.2%', 'ci': None} {'timepoint': '24 months', 'arm': 'Anbenitamab', 'value': '37.0%', 'ci': None} {'timepoint': '15 months', 'arm': 'Control', 'value': '56.9%', 'ci': None} {'timepoint': '18 months', 'arm': 'Control', 'value': '25.8%', 'ci': None} {'timepoint': '24 months', 'arm': 'Control', 'value': '14.0%', 'ci': None} Note: Printed control 15-mo label 56.9% sits near a curve point visually ~35%; transcribed as printed - verify. Number-at-risk parenthetical values appear to be cumulative events/censored. P-value exponent illegible. [Slide 2] Study design KC-WISE is a multicenter, randomized, double-blind, phase 3 study conducted at 51 sites in China (NCT05427383). Study Population • Centrally confirmed HER2 positive GC/GEJC • Progressed on previous therapy containing trastuzumab • ECOG PS of 0 or 1 R 1:1 → Anbenitamab 30 mg/kg Q3W and chemotherapy* → Placebo and chemotherapy* Primary endpoints: ➢ IRC assessed PFS ➢ OS [boxed] Secondary endpoints: ➢ IRC and investigator assessed ORR, DoR, DCR, investigator assessed PFS, and safety Stratification factors: − Chemotherapy type (taxanes [paclitaxel and docetaxel] or irinotecan) − HER2 expression (IHC 3+ or IHC2+/FISH+) − Previous lines of therapy (1 or ≥2) The second interim analysis was planned when approximate 150 OS events were observed. DCR, disease control rate; DoR, duration of response; ECOG PS, Eastern Cooperative Oncology Group performance status; FISH, fluorescence in situ hybridization; GC/GEJC, gastric cancer or gastroesophageal junction carcinoma; HER2, human epidermal growth factor receptor 2; IHC, immunohistochemistry; IRC, independent review committee; ORR, objective response, rate; OS, overall survival; PFS, progression-free survival; Q3W, every 3 weeks. Ruixing Zhang [ESMO Congress Madrid 2026 logo] [Slide 3] IRC and investigator-assessed ORR, DoR and PFS | | Anbenitamab (n=125) IRC-assessed | Anbenitamab Investigator-assessed | Control (n=121) IRC-assessed | Control Investigator-assessed | Best overall response, n (%) Complete response | 0 | 1 (0.8) | 1 (0.8) | 0 Partial response | 70 (56.0) | 65 (52.0) | 16 (13.2) | 22 (18.2) Stable disease | 19 (15.2) | 27 (21.6) | 29 (24.0) | 26 (21.5) Non-CR/Non-PD | 14 (11.2) | 14 (11.2) | 8 (6.6) | 8 (6.6) Progressive disease | 19 (15.2) | 14 (11.2) | 56 (46.3) | 53 (43.8) Not evaluable | 3 (2.4) | 4 (3.2) | 11 (9.1) | 12 (9.9) Objective response rate, %, 95% CI | 56.0 (46.8-64.9) | 52.8 (43.7-61.8) | 14.0 (8.4-21.5) | 18.2 (11.8-26.2) Disease control rate, %, 95% CI | 82.4 (74.6-88.6) | 85.6 (78.2-91.2) | 44.6 (35.6-53.9) | 46.3 (37.2-55.6) Median DoR, mo, 95% CI | 7.9 (5.6-9.7) | 8.6 (6.2-10.6) | 3.6 (2.7-7.8) | 3.4 (2.8-7.7) Median PFS, mo, 95% CI | 7.1 (5.4-9.0) | 8.8 (6.9-10.3) | 2.7 (1.5-3.0) | 2.8 (2.5-3.0) At cutoff date of July 15, 2026, the median follow-up duration was 22.8 months (95% CI, 18.9 to 24.4) in the anbenitamab group and 22.3 months (95% CI, 18.6 to 24.6) in the control group. BOR, best overall response; CI, confidence interval; CR, complete response; DoR, duration of response; IRC, independent review committee; PD, progressive disease. Ruixing Zhang [ESMO Congress Madrid 2026 logo]

[Slide 4] Subsequent anti-tumor therapy | Anbenitamab(n=125) | Control(n=121) Subsequent therapy | 75 (60.0) | 78 (64.5) Anti-tumor agents | 71 (56.8) | 76 (62.8) Immunotherapy | 30 (24.0) | 28 (23.1) HER2 targeted therapy | 50 (40.0) | 56 (46.3) Radiotherapy | 7 (5.6) | 3 (2.5) Surgery | 3 (2.4) | 3 (2.5) Other therapies | 4 (3.2) | 4 (3.3) Ruixing Zhang [ESMO Congress Madrid 2026 logo]

[Slide 2] Anbenitamab (KN026) is a HER2-targeted bispecific antibody with superior binding affinity, inhibitory potency, and HER2 internalization relative to trastuzumab or pertuzumab17; it has recently been approved by the National Medical Products Administration (China) for the second- and later-line treatment of HER2-positive advanced gastric cancer on the basis of the KC-WISE trial.18 HB1801 is an albumin-bound docetaxel. Previous clinical studies showed that HB1801 had improved antitumor activity in advanced gastric cancer.19 [Slide 3] FIGURE 2 A Progression-free Survival as Assessed by Blinded Independent Central Review Anbenitamab plus HB1801 (58.5 at 24 months); THP (34.4 at 24 months) Hazard ratio for disease progression or death, 0.46 (95% CI, 0.37 to 0.56) P<0.001 Months since Randomization No. at Risk Anbenitamab plus HB1801: 445 430 407 345 324 217 173 113 86 38 7 0 THP: 440 410 368 274 232 144 110 75 50 28 5 0 B Subgroup Analysis of Progression-free Survival Subgroup | Anbenitamab plus HB1801 | THP (no. of patients with event/total no. of patients) | Hazard Ratio for Disease Progression or Death (95% CI) All patients | 149/445 | 252/440 | 0.46 (0.38 to 0.57) Age <65 yr | 135/389 | 218/362 | 0.45 (0.37 to 0.56) Age ≥65 yr | 14/56 | 34/78 | 0.47 (0.25 to 0.88) ECOG 0 | 71/251 | 130/236 | 0.38 (0.28 to 0.51) ECOG 1 | 78/194 | 120/202 | 0.59 (0.44 to 0.79) Previous neoadjuvant or adjuvant therapy Yes | 60/165 | 95/161 | 0.49 (0.36 to 0.68) No | 89/280 | 157/279 | 0.45 (0.35 to 0.58) Previous treatment with trastuzumab Yes | 30/81 | 56/91 | 0.47 (0.30 to 0.74) No | 119/364 | 196/349 | 0.46 (0.37 to 0.58) Previous treatment with pertuzumab Yes | 14/29 | 23/32 | 0.46 (0.23 to 0.90) No | 135/416 | 229/408 | 0.46 (0.38 to 0.58) Hormone-receptor status Positive | 95/244 | 150/243 | 0.48 (0.37 to 0.62) Negative | 54/201 | 102/197 | 0.44 (0.31 to 0.61) Visceral metastasis Yes | 114/316 | 190/316 | 0.47 (0.37 to 0.59) No | 35/129 | 62/124 | 0.46 (0.30 to 0.69) Liver metastasis Yes | 57/156 | 111/181 | 0.46 (0.33 to 0.63) No | 92/289 | 141/259 | 0.47 (0.36 to 0.62) Bone metastasis Yes | 85/206 | 119/193 | 0.55 (0.42 to 0.73) No | 64/239 | 133/247 | 0.38 (0.29 to 0.52) No. of distant metastatic sites 1 or 2 | 92/310 | 148/288 | 0.46 (0.36 to 0.60) ≥3 | 56/125 | 98/141 | 0.50 (0.36 to 0.70) HER2 status IHC score of 3+ | 108/374 | 207/385 | 0.43 (0.34 to 0.54) IHC score of 2+ and FISH-positive | 41/71 | 45/55 | 0.44 (0.29 to 0.69) Stage IV disease at initial diagnosis Yes | 85/253 | 139/246 | 0.48 (0.37 to 0.63) No | 64/192 | 113/194 | 0.45 (0.33 to 0.61) Anbenitamab plus HB1801 Better | THP Better Progression-free Survival, Overall and According to Subgroup Analyses (Intention-to-Treat Population). [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS by BICR (ITT) Anbenitamab plus HB1801 | 445 THP | 440 HR: 0.46 HR CI: 0.37 to 0.56 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: Medians not printed. Panel A HR CI 0.37-0.56; forest-plot 'All patients' row shows 0.46 (0.38 to 0.57) - two slightly different CIs printed (likely stratified vs unstratified); do not mix. Event counts from subgroup table. [Slide 4] TABLE 3 Table 3. Summary of Adverse Events.* Anbenitamab plus HB1801 (N=444) All Grades | Grade ≥3 || THP (N=439) All Grades | Grade ≥3 number of patients (percent) Any adverse event | 444 (100) | 204 (45.9) | 434 (98.9) | 151 (34.4) Common adverse events† Anemia | 330 (74.3) | 21 (4.7) | 250 (56.9) | 15 (3.4) Diarrhea‡ | 236 (53.2) | 10 (2.3) | 176 (40.1) | 5 (1.1) Alopecia | 229 (51.6) | 0 | 172 (39.2) | 0 ALT level increased | 192 (43.2) | 6 (1.4) | 163 (37.1) | 6 (1.4) Leukopenia | 190 (42.8) | 43 (9.7) | 144 (32.8) | 24 (5.5) Neutropenia | 181 (40.8) | 58 (13.1) | 146 (33.3) | 37 (8.4) AST level increased | 175 (39.4) | 6 (1.4) | 158 (36.0) | 6 (1.4) Hypoalbuminemia | 167 (37.6) | 0 | 100 (22.8) | 1 (0.2) Weight loss | 155 (34.9) | 3 (0.7) | 106 (24.1) | 2 (0.5) Hypertriglyceridemia | 152 (34.2) | 14 (3.2) | 154 (35.1) | 12 (2.7) Peripheral edema | 128 (28.8) | 3 (0.7) | 64 (14.6) | 1 (0.2) Asthenia | 119 (26.8) | 4 (0.9) | 85 (19.4) | 2 (0.5) Weight gain | 114 (25.7) | 1 (0.2) | 83 (18.9) | 0 Hypokalemia | 113 (25.5) | 15 (3.4) | 61 (13.9) | 7 (1.6) Urinary tract infection | 103 (23.2) | 2 (0.5) | 85 (19.4) | 2 (0.5) Hyperchloesterolemia | 101 (22.7) | 0 | 122 (27.8) | 0 Nausea | 95 (21.4) | 1 (0.2) | 70 (15.9) | 0 * Listed are adverse events that occurred up to 30 days after the last administration of trial drugs or initiation of new antitumor treatment (whichever occurred first). The events are listed in descending order of frequency in the anbenitamab group. Safety analyses were performed in the safety population, which included all the patients who had undergone randomization and received at least one dose of a trial drug. The adverse events were coded according to the Medical Dictionary for Regulatory Activities, version 28.1, and graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events, version 5.0. ALT denotes alanine aminotransferase, and AST aspartate aminotransferase. † Adverse events that occurred in at least 20% of the patients in either group are reported. ‡ Diarrhea led to dose reduction in five patients (1.1%) in the anbenitamab group and two patients (0.5%) in the THP group and to treatment interruption in two patients (0.5%) and one patient (0.2%), respectively. No patients discontinued treatment owing to diarrhea. Primary prophylaxis for diarrhea was administered to 7.2% of the patients in the anbenitamab group and to 8.0% of those in THP group. Summary of Adverse Events.

[Slide 1] 16:15 - 17:45 Rapid oral 1: Gastrointestinal tumour, upper digestive CHAIRS: MARIA IGNEZ BRAGHIROLI, FENG WANG, LUCJAN WYRWICZ OS in the ITT population Compared with chemotherapy alone, anbenitamab plus chemotherapy significantly reduced the risk of death by 61% Anbenitamab group | Control group Median OS, mo, 95%CI: 19.6 (16.8, 23.4) | 9.6 (7.7, 13.7) HR (95% CI): 0.39 (0.28, 0.54) P value: 3.97*10^[illegible] Curve labels - Anbenitamab group: 67.3%, 56.2%, 37.0%; Control group: 56.9%, 25.8%, 14.0% Y-axis: Overall survival (%); X-axis: Time (months) 0-42 Number at risk Anbenitamab group: 125 (0), 122 (1), 114 (3), 101 (3), 82 (13), 63 (24), 48 (29), 32 (38), 19 (45), 15 (47), 7 (52), 4 (55), 2 (57), 2 (57), 0 (59) Control group: 121 (0), 108 (0), 83 (1), 63 (2), 45 (10), 32 (14), 21 (17), 13 (20), 6 (24), 4 (26), 3 (27), 3 (27), 1 (29), 0 (30) Ruixing Zhang [ESMO Congress Madrid 2026 logo] Ruixing Zhang Anbenitamab plus chemotherapy for previously treated HER2-positive gastric or gastroesophageal carcinomas (GC/GEJC): Updated overall survival (OS) analysis of KC-WISE ALMERIA AUDITORIUM - HALL 9 [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival (ITT) Anbenitamab group (anbenitamab + chemotherapy) | 125 | 19.6 mo | 16.8, 23.4 Control group (chemotherapy) | 121 | 9.6 mo | 7.7, 13.7 HR: 0.39 HR CI: 0.28, 0.54 P: 3.97*10^[exponent illegible] {'timepoint': '15 mo (vertical line)', 'arm': 'Anbenitamab group', 'value': '67.3%', 'ci': None} {'timepoint': '18 mo (vertical line)', 'arm': 'Anbenitamab group', 'value': '56.2%', 'ci': None} {'timepoint': '24 mo (vertical line)', 'arm': 'Anbenitamab group', 'value': '37.0%', 'ci': None} Note: Control-arm labels 56.9% / 25.8% / 14.0% are printed near the 15/18/24-mo lines but 56.9% sits above where the control curve is at 15 mo, so their timepoint mapping is uncertain - not tabulated as landmarks. Anbenitamab landmark timepoints inferred from vertical reference lines at 15, 18, 24 months. N-at-risk shown with censored counts in parentheses.

[Slide 1] Confidential Phase 3 KEYNOTE-975 Study Design (NCT04210115) Key Eligibility Criteria • Histologically confirmed ESCC, GEJC, EAC • Tumor staging cTX N+ M0, cT2-T4a NX M0, or cTX N+M1, cervical or upper thoracic esophageal carcinoma with supraclavicular metastases only • Eligible for dCRT • Ineligible for curative surgery • Radiographically evaluable disease • ECOG PS 0-1 • Available tumor tissue R (1:1) N = 703 Pembrolizumab 200 mg Q3W (8 cycles) then 400 mg Q6W (5 cycles) + dCRT (RT + investigator choice of FOLFOX or FP) Placebo Q3W (8 cycles) then Q6W (5 cycles) + dCRT (RT + investigator choice of FOLFOX or FP) Treated until progression, unacceptable toxicity, or withdrawal Stratification • PD-L1 CPS (≥10 vs <10) • Radiation dose (50 Gy vs 60 Gy) • Geographic region and histology (SCC East Asia vs SCC ROW and adenocarcinoma regardless of origin) Endpoints • Primary: EFS and OS in PD-L1 CPS≥10, CPS≥1, all participants • Secondary: Safety and tolerability [Slide 2] Confidential EFS in Participants With PD-L1 CPS ≥10 Tumors | | Events n (%) | Median EFS (95% CI), mo | HR (95% CI) P-value | | Pembrolizumab + dCRT | 137 (66%) | 15.2 (11.2 - 19.4) | 0.97 (0.77 - 1.23) | | Placebo + dCRT | 147 (69%) | 14.7 (11.4 - 19.9) | P = 0.399a | Y-axis: Event-free survival (%); X-axis: Time, months 24-mo rate: 39% / 37% 48-mo rate: 30% / 27% No. at Risk Pembrolizumab + dCRT: 209 169 116 101 81 68 59 41 34 22 13 5 3 0 0 Placebo + dCRT: 214 176 127 101 86 63 52 39 33 24 15 11 3 0 0 Data cutoff: Jan 21, 2026. aDid not meet the pre-specified boundary for significance (P = 0.0193) [Survival data captured as a table — no point estimates read from the curve] Endpoint: EFS, PD-L1 CPS ≥10 Pembrolizumab + dCRT | 209 | 15.2 | 11.2-19.4 Placebo + dCRT | 214 | 14.7 | 11.4-19.9 HR: 0.97 HR CI: 0.77-1.23 P: 0.399 {'timepoint': '24 mo', 'arm': 'Pembrolizumab + dCRT', 'value': '39%', 'ci': None} {'timepoint': '24 mo', 'arm': 'Placebo + dCRT', 'value': '37%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Pembrolizumab + dCRT', 'value': '30%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Placebo + dCRT', 'value': '27%', 'ci': None} Note: Boundary footnote read as P = 0.0193 (vs 0.0183 read on livestream capture 2113601481865937254_s1; small text — verify). [Slide 3] Confidential OS in Participants With PD-L1 CPS ≥10 Tumors | | Events n (%) | Median OS (95% CI), mo | HR (95% CI) P-value | | Pembrolizumab + dCRT | 111 (53%) | 32.2 (24.0 - 48.0) | 1.00 (0.77 - 1.30) | | Placebo + dCRT | 118 (55%) | 35.0 (25.2 - 50.1) | P = 0.500a | Y-axis: Overall survival (%); X-axis: Time, months 24-mo rate: 58% / 59% 48-mo rate: 43% / 44% No. at Risk Pembrolizumab + dCRT: 209 182 163 143 126 109 87 70 64 50 33 17 7 1 0 Placebo + dCRT: 214 196 171 149 135 109 95 75 64 55 41 23 11 2 0 Data cutoff: Jan 21, 2026. aNominal p-value as OS was not tested for statistical significance as did not meet the pre-specified boundary for significance in EFS in PD-L1 CPS ≥10. [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS, PD-L1 CPS ≥10 Pembrolizumab + dCRT | 209 | 32.2 | 24.0-48.0 Placebo + dCRT | 214 | 35.0 | 25.2-50.1 HR: 1.00 HR CI: 0.77-1.30 P: 0.500 (nominal) {'timepoint': '24 mo', 'arm': 'Pembrolizumab + dCRT', 'value': '58%', 'ci': None} {'timepoint': '24 mo', 'arm': 'Placebo + dCRT', 'value': '59%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Pembrolizumab + dCRT', 'value': '43%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Placebo + dCRT', 'value': '44%', 'ci': None} Note: Nominal p-value. [Slide 4] Confidential EFS and OS in Key Subgroups in all Participants EFS — Events/Patients, N | HR (95% CI) Overall 468/703 0.94 (0.78-1.12) Age Category: <65 years 210/300 0.95 (0.72-1.25); ≥65 years 258/403 0.94 (0.73-1.20) Sex: Male 349/508 1.00 (0.81-1.23); Female 119/195 0.81 (0.57-1.17) Race: Asian 172/289 1.04 (0.77-1.41); Non-Asian 289/402 0.88 (0.70-1.11) ECOG PS: 0 156/241 0.78 (0.57-1.07); 1 312/462 1.06 (0.85-1.32) Geographic Region and Histology: Adenocarcimona 73/96 0.67 (0.42-1.06); ESCC in East Asia 164/281 1.04 (0.77-1.42); ESCC in Rest of the World 231/326 0.93 (0.72-1.21) Chemotherapy Backbone: FP 216/350 1.11 (0.85-1.46); FOLFOX 252/353 0.81 (0.63-1.04) Radiation Dose: 50 Gy 397/583 0.97 (0.79-1.18); 60 Gy 68/116 0.80 (0.50-1.30) Histology: Adenocarcinoma 73/96 0.67 (0.42-1.06); Squamous Cell Carcinoma 395/607 0.98 (0.80-1.19) Current Disease Stage: II 122/213 0.98 (0.69-1.40); III 278/391 0.96 (0.76-1.22); IV 64/92 0.97 (0.59-1.58) Axis: 0.1 – 1 – 10; Favors pembrolizumab + dCRT ← → Favors placebo + dCRT OS — Events/Patients, N | HR (95% CI) Overall 385/703 0.99 (0.81-1.21) Age Category: <65 years 169/300 1.04 (0.77-1.41); ≥65 years 216/403 0.95 (0.73-1.24) Sex: Male 291/508 1.09 (0.86-1.37); Female 94/195 0.77 (0.51-1.16) Race: Asian 135/289 1.08 (0.77-1.51); Non-Asian 246/402 0.93 (0.72-1.19) ECOG PS: 0 126/241 0.74 (0.52-1.05); 1 259/462 1.15 (0.90-1.47) Geographic Region and Histology: Adenocarcimona 61/96 0.76 (0.46-1.25); ESCC in East Asia 129/281 1.11 (0.79-1.57); ESCC in Rest of the World 195/326 1.00 (0.75-1.33) Chemotherapy Backbone: FP 176/350 1.24 (0.92-1.66); FOLFOX 209/353 0.82 (0.63-1.08) Radiation Dose: 50 Gy 330/583 0.96 (0.78-1.20); 60 Gy 52/116 1.23 (0.71-2.13) Histology: Adenocarcinoma 61/96 0.76 (0.46-1.25); Squamous Cell Carcinoma 324/607 1.04 (0.83-1.29) Current Disease Stage: II 102/213 1.05 (0.71-1.55); III 230/391 1.01 (0.78-1.31); IV 50/92 1.00 (0.57-1.74) Axis: 0.1 – 1 – 10; Favors pembrolizumab + dCRT ← → Favors placebo + dCRT Data cutoff: Jan 21, 2026. Subgroups with fewer than 20 participants were not included in the forest plot.

[Slide 4] Confidential EFS and OS in Participants With PD-L1 CPS ≥1 Tumors EFS Table: Events n (%) | Median EFS (95% CI), mo | HR (95% CI) P-value Pembrolizumab + dCRT: 206 (64%) | 16.9 (12.8 - 19.5) | 0.92 (0.76 - 1.11) Placebo + dCRT: 229 (69%) | 14.7 (11.6 - 18.6) | P = 0.187a 24-mo rate: 40% / 39%; 48-mo rate: 32% / 28% No. at Risk (months 0,5,...,70): Pembro: 324 269 195 165 133 107 92 63 51 30 18 8 5 0 0 Placebo: 331 275 200 158 136 108 85 66 58 43 28 20 9 1 0 OS Table: Events n (%) | Median OS (95% CI), mo | HR (95% CI) P-value Pembrolizumab + dCRT: 172 (53%) | 33.2 (25.6 - 46.7) | 0.95 (0.77 - 1.17) Placebo + dCRT: 186 (56%) | 35.0 (25.3 - 43.6) | P = 0.324a 24-mo rate: 59% / 57%; 48-mo rate: 43% / 44% No. at Risk (months 0,5,...,70): Pembro: 324 287 259 232 204 174 137 110 94 72 47 27 12 2 0 Placebo: 331 301 267 226 205 172 145 119 103 88 70 43 18 3 0 Data cutoff: Jan 21, 2026. aNominal p-value as EFS and OS in PD-L1 CPS ≥1 was not tested for statistical significance. [Survival data captured as a table — no point estimates read from the curve] Endpoint: EFS and OS, PD-L1 CPS ≥1 (not formally tested) EFS: Pembrolizumab + dCRT | 324 | 16.9 mo | 12.8-19.5 EFS: Placebo + dCRT | 331 | 14.7 mo | 11.6-18.6 OS: Pembrolizumab + dCRT | 324 | 33.2 mo | 25.6-46.7 OS: Placebo + dCRT | 331 | 35.0 mo | 25.3-43.6 HR: EFS 0.92; OS 0.95 HR CI: EFS 0.76-1.11; OS 0.77-1.17 P: EFS P = 0.187; OS P = 0.324 (nominal) {'timepoint': '24 mo', 'arm': 'EFS Pembro', 'value': '40%', 'ci': None} {'timepoint': '24 mo', 'arm': 'EFS Placebo', 'value': '39%', 'ci': None} {'timepoint': '48 mo', 'arm': 'EFS Pembro', 'value': '32%', 'ci': None} {'timepoint': '48 mo', 'arm': 'EFS Placebo', 'value': '28%', 'ci': None} {'timepoint': '24 mo', 'arm': 'OS Pembro', 'value': '59%', 'ci': None} {'timepoint': '24 mo', 'arm': 'OS Placebo', 'value': '57%', 'ci': None} {'timepoint': '48 mo', 'arm': 'OS Pembro', 'value': '43%', 'ci': None} {'timepoint': '48 mo', 'arm': 'OS Placebo', 'value': '44%', 'ci': None} Note: DCO Jan 21, 2026. Small text; number-at-risk rows lower confidence.

[Slide 1] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH [Speaker video] Manish Shah Pembrolizumab or placebo plus concurrent definitive chemoradiotherapy in esophageal and gastroesophageal junction (GEJ) cancer: The phase 3 randomized KEYNOTE-975 study ALMERIA AUDITORIUM - HALL 9 [ESMO Congress Madrid 2026 logo] Confidential EFS in Participants With PD-L1 CPS ≥10 Tumors | | Events n (%) | Median EFS (95% CI), mo | HR (95% CI) P-value | | Pembrolizumab + dCRT | 137 (66%) | 15.2 (11.2 - 19.4) | 0.97 (0.77 - 1.23) | | Placebo + dCRT | 147 (69%) | 14.7 (11.4 - 19.9) | P = 0.399a | Y-axis: Event-free survival (%); X-axis: Time, months 24-mo rate: 39% (pembrolizumab) / 37% (placebo) 48-mo rate: 30% (pembrolizumab) / 27% (placebo) No. at Risk (0,5,...,70 mo) Pembrolizumab + dCRT: 209 169 116 101 81 68 59 41 34 22 13 5 3 0 0 Placebo + dCRT: 214 176 127 101 86 63 52 39 33 24 15 11 3 0 0 Data cutoff: Jan 21, 2026. aDid not meet the pre-specified boundary for significance (P = 0.0183) [Survival data captured as a table — no point estimates read from the curve] Endpoint: EFS, PD-L1 CPS ≥10 Pembrolizumab + dCRT | 209 | 15.2 | 11.2-19.4 Placebo + dCRT | 214 | 14.7 | 11.4-19.9 HR: 0.97 HR CI: 0.77-1.23 P: 0.399 {'timepoint': '24 mo', 'arm': 'Pembrolizumab + dCRT', 'value': '39%', 'ci': None} {'timepoint': '24 mo', 'arm': 'Placebo + dCRT', 'value': '37%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Pembrolizumab + dCRT', 'value': '30%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Placebo + dCRT', 'value': '27%', 'ci': None} Note: Did not meet pre-specified significance boundary (footnote boundary read as P = 0.0183 here; the same slide in 2113601888923029958_s2 reads 0.0193 — small text, verify). [Slide 2] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH [Speaker video] Manish Shah Pembrolizumab or placebo plus concurrent definitive chemoradiotherapy in esophageal and gastroesophageal junction (GEJ) cancer: The phase 3 randomized KEYNOTE-975 study ALMERIA AUDITORIUM - HALL 9 [ESMO Congress Madrid 2026 logo] Confidential OS in Participants With PD-L1 CPS ≥10 Tumors | | Events n (%) | Median OS (95% CI), mo | HR (95% CI) P-value | | Pembrolizumab + dCRT | 111 (53%) | 32.2 (24.0 - 48.0) | 1.00 (0.77 - 1.30) | | Placebo + dCRT | 118 (55%) | 35.0 (25.2 - 50.1) | P = 0.500a | Y-axis: Overall survival (%); X-axis: Time, months 24-mo rate: 58% (pembrolizumab) / 59% (placebo) 48-mo rate: 43% (pembrolizumab) / 44% (placebo) No. at Risk (0,5,...,70 mo) Pembrolizumab + dCRT: 209 182 163 143 126 109 87 70 64 50 33 17 7 1 0 Placebo + dCRT: 214 196 171 149 135 109 95 75 64 55 41 23 11 2 0 Data cutoff: Jan 21, 2026. aNominal p-value as OS was not tested for statistical significance as did not meet the pre-specified boundary for significance in EFS in PD-L1 CPS ≥10. [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS, PD-L1 CPS ≥10 Pembrolizumab + dCRT | 209 | 32.2 | 24.0-48.0 Placebo + dCRT | 214 | 35.0 | 25.2-50.1 HR: 1.00 HR CI: 0.77-1.30 P: 0.500 (nominal) {'timepoint': '24 mo', 'arm': 'Pembrolizumab + dCRT', 'value': '58%', 'ci': None} {'timepoint': '24 mo', 'arm': 'Placebo + dCRT', 'value': '59%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Pembrolizumab + dCRT', 'value': '43%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Placebo + dCRT', 'value': '44%', 'ci': None} Note: Nominal p-value; OS not formally tested. [Slide 3] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH [Speaker video] Manish Shah Pembrolizumab or placebo plus concurrent definitive chemoradiotherapy in esophageal and gastroesophageal junction (GEJ) cancer: The phase 3 randomized KEYNOTE-975 study ALMERIA AUDITORIUM - HALL 9 [ESMO Congress Madrid 2026 logo] Confidential EFS and OS in Participants With PD-L1 CPS ≥1 Tumors EFS | | Events n (%) | Median EFS (95% CI), mo | HR (95% CI) P-value | | Pembrolizumab + dCRT | 206 (64%) | 16.9 (12.8 - 19.5) | 0.92 (0.76 - 1.11) | | Placebo + dCRT | 229 (69%) | 14.7 (11.6 - 18.6) | P = 0.187a | 24-mo rate: 40% / 39%; 48-mo rate: 32% / 26% No. at Risk: Pembrolizumab + dCRT 324 269 195 165 133 107 92 63 51 30 18 8 5 0 0 Placebo + dCRT 331 275 200 158 136 108 85 66 58 43 28 26 9 1 0 OS | | Events n (%) | Median OS (95% CI), mo | HR (95% CI) P-value | | Pembrolizumab + dCRT | 172 (53%) | 33.2 (25.6 - 46.7) | 0.95 (0.77 - 1.17) | | Placebo + dCRT | 186 (56%) | 35.0 (25.3 - 43.6) | P = 0.324a | 24-mo rate: 59% / 57%; 48-mo rate: 43% / 44% No. at Risk: Pembrolizumab + dCRT 324 287 259 232 204 174 137 110 94 72 47 27 12 2 0 Placebo + dCRT 331 301 267 226 205 172 145 119 103 88 70 43 18 3 0 Data cutoff: Jan 21, 2026. aNominal p-value as EFS and OS in PD-L1 CPS ≥1 was not tested for statistical significance. [Survival data captured as a table — no point estimates read from the curve] Endpoint: EFS and OS, PD-L1 CPS ≥1 Pembrolizumab + dCRT (EFS) | 324 | 16.9 | 12.8-19.5 Placebo + dCRT (EFS) | 331 | 14.7 | 11.6-18.6 Pembrolizumab + dCRT (OS) | 324 | 33.2 | 25.6-46.7 Placebo + dCRT (OS) | 331 | 35.0 | 25.3-43.6 HR: EFS 0.92; OS 0.95 HR CI: EFS 0.76-1.11; OS 0.77-1.17 P: EFS 0.187; OS 0.324 (nominal) {'timepoint': '24 mo', 'arm': 'Pembrolizumab + dCRT (EFS)', 'value': '40%', 'ci': None} {'timepoint': '24 mo', 'arm': 'Placebo + dCRT (EFS)', 'value': '39%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Pembrolizumab + dCRT (EFS)', 'value': '32%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Placebo + dCRT (EFS)', 'value': '26%', 'ci': None} {'timepoint': '24 mo', 'arm': 'Pembrolizumab + dCRT (OS)', 'value': '59%', 'ci': None} {'timepoint': '24 mo', 'arm': 'Placebo + dCRT (OS)', 'value': '57%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Pembrolizumab + dCRT (OS)', 'value': '43%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Placebo + dCRT (OS)', 'value': '44%', 'ci': None} Note: Small text from livestream capture; numbers read from 2x zoom, some at-risk digits and the placebo 48-mo EFS rate (26% vs 28%) are low confidence. Nominal p-values. [Slide 4] 13:30 - 15:00 Proffered paper 1: Gastrointestinal tumour, upper digestive CHAIRS: FILIPPO PIETRANTONIO, ELIZABETH SMYTH [Speaker video] Manish Shah Pembrolizumab or placebo plus concurrent definitive chemoradiotherapy in esophageal and gastroesophageal junction (GEJ) cancer: The phase 3 randomized KEYNOTE-975 study ALMERIA AUDITORIUM - HALL 9 [ESMO Congress Madrid 2026 logo] Confidential EFS and OS in all Participants EFS | | Events n (%) | Median EFS (95% CI), mo | HR (95% CI) P-value | | Pembrolizumab + dCRT | 224 (64%) | 16.8 (13.0 - 19.4) | 0.94 (0.78 - 1.12) | | Placebo + dCRT | 244 (69%) | 15.8 (12.1 - 19.3) | P = 0.239a | 24-mo rate: 40% / 40%; 48-mo rate: 31% / 28% No. at Risk: Pembrolizumab + dCRT 349 291 212 178 144 117 101 70 54 33 20 9 6 0 0 Placebo + dCRT 354 296 216 173 149 117 90 71 62 47 32 22 10 0 0 OS | | Events n (%) | Median OS (95% CI), mo | HR (95% CI) P-value | | Pembrolizumab + dCRT | 188 (54%) | 32.5 (25.6 - 41.5) | 0.99 (0.81 - 1.21) | | Placebo + dCRT | 197 (56%) | 35.3 (26.3 - 43.6) | P = 0.445a | 24-mo rate: 58% / 58%; 48-mo rate: 42% / 44% No. at Risk: Pembrolizumab + dCRT 349 311 280 249 217 187 148 118 100 76 51 29 13 2 0 Placebo + dCRT 354 324 287 245 221 186 155 127 109 94 75 47 19 3 0 Data cutoff: Jan 21, 2026. aNominal p-value as EFS and OS in all participants was not tested for statistical significance. [Survival data captured as a table — no point estimates read from the curve] Endpoint: EFS and OS, all participants Pembrolizumab + dCRT (EFS) | 349 | 16.8 | 13.0-19.4 Placebo + dCRT (EFS) | 354 | 15.8 | 12.1-19.3 Pembrolizumab + dCRT (OS) | 349 | 32.5 | 25.6-41.5 Placebo + dCRT (OS) | 354 | 35.3 | 26.3-43.6 HR: EFS 0.94; OS 0.99 HR CI: EFS 0.78-1.12; OS 0.81-1.21 P: EFS 0.239; OS 0.445 (nominal) {'timepoint': '24 mo', 'arm': 'Pembrolizumab + dCRT (EFS)', 'value': '40%', 'ci': None} {'timepoint': '24 mo', 'arm': 'Placebo + dCRT (EFS)', 'value': '40%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Pembrolizumab + dCRT (EFS)', 'value': '31%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Placebo + dCRT (EFS)', 'value': '28%', 'ci': None} {'timepoint': '24 mo', 'arm': 'Pembrolizumab + dCRT (OS)', 'value': '58%', 'ci': None} {'timepoint': '24 mo', 'arm': 'Placebo + dCRT (OS)', 'value': '58%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Pembrolizumab + dCRT (OS)', 'value': '42%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Placebo + dCRT (OS)', 'value': '44%', 'ci': None} Note: Small text from livestream capture; read at 2x zoom; some digits low confidence. Overall EFS/OS HRs (0.94/0.99) match forest plot in 2113601888923029958_s4.

[Slide 2] Confidential EFS and OS in all Participants EFS | | Events n (%) | Median EFS (95% CI), mo | HR (95% CI) P-value | | Pembrolizumab + dCRT | 224 (64%) | 16.8 (13.0 - 19.4) | 0.94 (0.78 - 1.12) | | Placebo + dCRT | 244 (69%) | 15.8 (12.1 - 19.3) | P = 0.239a | [KM curve: Event-free survival (%) vs Time, months 0-70] 24-mo rate: 40% / 40%; 48-mo rate: 31% / 28% (pembrolizumab / placebo) No. at Risk (months 0/5/10/.../70): Pembrolizumab + dCRT: 349 291 212 178 144 117 101 70 54 33 20 9 6 0 0 Placebo + dCRT: 354 296 216 173 149 117 90 71 62 47 32 22 10 0 0 OS | | Events n (%) | Median OS (95% CI), mo | HR (95% CI) P-value | | Pembrolizumab + dCRT | 188 (54%) | 32.5 (25.6 - 41.5) | 0.99 (0.81 - 1.21) | | Placebo + dCRT | 197 (56%) | 35.3 (26.3 - 43.6) | P = 0.445a | [KM curve: Overall survival (%) vs Time, months 0-70] 24-mo rate: 58% / 58%; 48-mo rate: 42% / 44% (pembrolizumab / placebo) No. at Risk (months 0/5/10/.../70): Pembrolizumab + dCRT: 349 311 280 249 217 187 148 118 100 76 51 29 13 2 0 Placebo + dCRT: 354 324 287 245 221 186 155 127 109 94 75 47 19 3 0 Data cutoff: Jan 21, 2026. aNominal p-value as EFS and OS in all participants was not tested for statistical significance. [Survival data captured as a table — no point estimates read from the curve] Endpoint: EFS and OS, all participants Pembrolizumab + dCRT (EFS) | 349 | 16.8 mo | 13.0-19.4 Placebo + dCRT (EFS) | 354 | 15.8 mo | 12.1-19.3 Pembrolizumab + dCRT (OS) | 349 | 32.5 mo | 25.6-41.5 Placebo + dCRT (OS) | 354 | 35.3 mo | 26.3-43.6 HR: EFS 0.94; OS 0.99 HR CI: EFS 0.78-1.12; OS 0.81-1.21 P: EFS P=0.239 (nominal); OS P=0.445 (nominal) {'timepoint': '24 months EFS', 'arm': 'Pembrolizumab + dCRT', 'value': '40%', 'ci': None} {'timepoint': '24 months EFS', 'arm': 'Placebo + dCRT', 'value': '40%', 'ci': None} {'timepoint': '48 months EFS', 'arm': 'Pembrolizumab + dCRT', 'value': '31%', 'ci': None} {'timepoint': '48 months EFS', 'arm': 'Placebo + dCRT', 'value': '28%', 'ci': None} {'timepoint': '24 months OS', 'arm': 'Pembrolizumab + dCRT', 'value': '58%', 'ci': None} {'timepoint': '24 months OS', 'arm': 'Placebo + dCRT', 'value': '58%', 'ci': None} {'timepoint': '48 months OS', 'arm': 'Pembrolizumab + dCRT', 'value': '42%', 'ci': None} {'timepoint': '48 months OS', 'arm': 'Placebo + dCRT', 'value': '44%', 'ci': None} Note: Two panels (EFS, OS) on one slide. Data cutoff Jan 21, 2026. Nominal p-values; not tested for significance in all participants.

[Slide 1] Study design Academic proof-of-concept single arm phase II trial, NCT06578559 Key Clinical Eligibility Criteria • ≥18 years of age • ECOG PS ≤ 1 • BRAF-V600E mutant mCRC treated with previous EC (± chemo), with at least PR or SD and PFS ≥ 6 months • One intervening EC-free line of therapy • At least 4 months between the end of EC and retreatment ctDNA analysis* Molecular Eligibility Criteria • BRAF-V600E mutant • RAS wild-type • MAP2K1 wild-type • MET not amplified → Encorafenib + cetuximab re-treatment Primary endpoint: ORR Key Secondary Endpoints: DCR, PFS, OS, safety Fleming single stage design: p0 for ORR = 0.05; p1 for ORR= 0.25; alpha error = 0.05 (two-sided); beta error = 0.20 Target accrual: 16 patients; p0 rejected if ≥ 3 objective responses observed *GUARDANT360 CDx, Guardant Health ECOG PS, Eastern Cooperative Oncology Group Performance Status, mCRC, metastatic colorectal cancer; p0, null hypothesis; p1: alternative hypothesis; PR, partial response; SD, stable disease; ctDNA, circulating tumor DNA; PFS, progression-free survival; ORR, objective response rate; DCR, disease control rate; OS, overall survival Marco Maria Germani, MD [Slide 2] Results, study screening Genomic profile of the 13 molecularly ineligible patients [Oncoprint: MSI status; screening genes BRAF 93%, KRAS 31%, NRAS 7%, MET 7%, MAP2K1 0%] Alteration type: Missense / other SNV; Frameshift indel; Nonsense; Splice-site; Amplification; Wild-type / VUS Microsatellite status: MSI-H; MSS VAF (%) - SNV/indel, log10 (0.05–50); Est. copy number - AMP (2.0–2.8) 2 BRAF-V600E wild-type 7 KRAS mut + 1 KRAS ampl 1 NRAS mut 1 NRAS mut + KRAS mut 1 MET ampl 1 MET ampl + NRAS mut Amp, amplification; mut: mutation; Est, estimated; MSI-H: microsatellite high; MSS, microsatellite stable; mut, mutation; VAF: variant allele fraction; VUS: variance of unknown significance Marco Maria Germani, MD [Slide 3] Results, objective response rate ORR: 0/16 (0%) 95% CI: 0 – 20.6% DCR: 6/16 (37.5%) 95% CI: 15.2 – 64.6% [Waterfall plot of best change in target lesions; legend: Disease Progression / Stable Disease; reference lines +20% and -30%] Two patients were not radiologically assessed for objective response because clinically deteriorated and died before scheduled radiological assessment. Both patients were computed as non-responders. * PD due to non-target lesion progression. ** PD due to occurrence of new lesions CI, confidence interval; DCR, disease control rate; ORR, objective response rate; PD, disease progression Marco Maria Germani, MD

[Slide 1] How do we understand the differences in trial outcomes? 1. Patient selection – in Tridente, only patients with prior CR/PR were included 2. Regimen – could the addition of binimetinib push patients to respond? But actually the clinically important survival interval is very similar between the two studies: mPFS: 2.8 months in TRIDENTE, 2.9 months in BRICKET Rona Yaeger, MD [Slide 2] Rechallenge Trials: TRIDENTE and BRICKET Design TRIDENTE: study design RAS wild-type/BRAF V600E-mutant mCRC -> EC-based therapy -> PD -> ≥1 of SoC -> PD -> ENCO + BINI + CET (N =21) • Encorafenib 300 mg QD • Binimetinib 45 mg BID • Cetuximab initially 400 mg/m2 and subsequently 250 mg/m2 QW Primary endpoint: ORR • CR or PR by RECIST v1.1 • PD within 4 weeks after the last ENCO administration ≥ 4 months Key eligibility criteria • ≥ 20 years of age • ECOG PS 0-1 • Refractory or intolerant to fluoropyrimidine-based chemotherapy including oxaliplatin or irinotecan • Measurable disease per RECIST v1.1 BRICKET: study design RAS wild-type/BRAF V600E-mutant mCRC -> EC-based therapy -> PD -> 1 of SoC (EC free) -> PD -> Molecular Eligibility Criteria -> ENCO + CET (N =16) Molecular Eligibility Criteria • BRAF-V600E mutant • RAS wild-type • MAP2K1 wild-type • MET not amplified • Encorafenib 300 mg QD • Cetuximab initially 400 mg/m2 and subsequently 250 mg/m2 QW Primary endpoint: ORR • PR or SD by RECIST v1.1 • PFS ≥ 6 months ≥ 4 months Key eligibility criteria • ≥ 18 years of age • ECOG PS 0-1 • Measurable disease per RECIST v1.1 Rona Yaeger, MD

[Slide 2] 14:45 - 16:15 Rapid oral: Gastrointestinal tumours, lower digestive CHAIRS: LISA SALVATORE, KAI-KEEN SHIU, RONA YAEGER Results, Survival outcomes Progression-free survival n | Events | Median PFS (mos) | 95% CI 16 | 16 | 2.94 | 2.01 – 3.39 PFS probability ; Months 0-8 Number at risk (number censored): 16 (0) | 13 (0) | 3 (0) | 2 (0) | 0 (0) Overall survival n | Events | Median OS (mos) | 95% CI 16 | 11 | 9.11 | 4.01 – 13.0 OS probability ; Months 0-14 Number at risk (number censored): 16 (0) | 14 (0) | 12 (1) | 8 (2) | 6 (3) | 4 (3) | 3 (4) | 1 (4) CI, confidence interval; mos, months; n, number; PFS, progression free survival; OS, overall survival Marco Maria Germani, MD [Speaker panel] Marco Maria Germani — Circulating tumor DNA (ctDNA)-guided re-treatment (re-tx) with encorafenib and cetuximab (EC) in patients (pts) with chemorefractory BRAFV600E mutated metastatic colorectal cancer (mCRC): The proof-of-concept single arm phase II BRICKET trial by GONO ALICANTE AUDITORIUM - HALL 6 [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS and OS (single arm) Encorafenib + cetuximab re-treatment — PFS | 16 | 2.94 mo | 2.01 – 3.39 Encorafenib + cetuximab re-treatment — OS | 16 | 9.11 mo | 4.01 – 13.0 Note: Single-arm phase II; two endpoints on one slide, listed as two rows. PFS at-risk (censored) at 0/2/4/6/8 mo: 16(0), 13(0), 3(0), 2(0), 0(0). OS at-risk at 0/2/4/6/8/10/12/14 mo: 16(0), 14(0), 12(1), 8(2), 6(3), 4(3), 3(4), 1(4). Small print on livestream capture; read carefully but low resolution. [Slide 3] 14:45 - 16:15 Rapid oral: Gastrointestinal tumours, lower digestive CHAIRS: LISA SALVATORE, KAI-KEEN SHIU, RONA YAEGER Results, objective response rate ORR: 0/16 (0%) 95% CI: 0 – 20.6% DCR: 6/16 (37.5%) 95% CI: 15.2 – 64.6% Legend: Disease Progression; Stable Disease Waterfall reference lines: +20%, -30% Two patients were not radiologically assessed for objective response because clinically deteriorated and died before scheduled radiological assessment. Both patients were computed as non-responders. * PD due to non-target lesion progression ** PD due to occurrence of new lesions CI, confidence interval; DCR, disease control rate; ORR, objective response rate; PD, disease progression Marco Maria Germani, MD [Speaker panel] Marco Maria Germani — Circulating tumor DNA (ctDNA)-guided re-treatment (re-tx) with encorafenib and cetuximab (EC) in patients (pts) with chemorefractory BRAFV600E mutated metastatic colorectal cancer (mCRC): The proof-of-concept single arm phase II BRICKET trial by GONO ALICANTE AUDITORIUM - HALL 6

[Slide 3] Results, Survival outcomes Progression-free survival n | Events | Median PFS (mos) | 95% CI 16 | 16 | 2.94 | 2.01 – 3.39 Number at risk (number censored): 16 (0) [0 mo], 13 (0) [2], 3 (0) [4], 2 (0) [6], 0 (0) [8] Overall survival n | Events | Median OS (mos) | 95% CI 16 | 11 | 9.11 | 4.01 – 13.0 Number at risk (number censored): 16 (0) [0], 14 (0) [2], 12 (1) [4], 8 (2) [6], 6 (3) [8], 4 (3) [10], 3 (4) [12], 1 (4) [14] CI, confidence interval; mos; months; n, number; PFS, progression-free survival; OS, overall survival Marco Maria Germani, MD [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS and OS (single arm, encorafenib + cetuximab re-treatment) EC re-treatment (PFS) | 16 | 2.94 mos | 2.01 – 3.39 EC re-treatment (OS) | 16 | 9.11 mos | 4.01 – 13.0 Note: Single-arm; no landmark rates printed. [Slide 4] Results, treatments and survival since LB in eligible and not eligibile patients Eligible, n=16 — Lines post-LB — [swimmer plot] — 6/16 treated after PD to EC (37.5%) Not eligible, n=13 — Lines post-LB — [swimmer plot] — 10/13 treated after LB (76.9%) Time since LB (months) No OS difference since LB detected between eligible and not eligible patients Cohort | Events/Pop | Survival after LB (mos, 95% CI) | 95% CI | HR | 95% CI | p value Not eligible | 9/13 | 5.1 | 2.9 – 12.7 | Ref. | – | Eligible | 11/16 | 9.7 | 4.7 – 13.6 | 0.90 | 0.36 – 2.29 | 0.832 OS probability after liquid biopsy / Months 0-14 Number at risk (number censored) Not eligible: 13 (0), 9 (3), 7 (3), 4 (3), 4 (3), 4 (3), 3 (3), 2 (3) Eligible: 16 (0), 14 (0), 13 (0), 10 (1), 7 (2), 5 (3), 3 (3), 1 (4) CI, confidence interval; FTD/TPI: trifluridine-tipiracil; LB: liquid biopsy; mos; months; n, number; NE: not evaluable; OS, overall survival; PD: progressive disease; SD: stable disease Marco Maria Germani, MD [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS since liquid biopsy (eligible vs not eligible) Not eligible | 13 | 5.1 mos | 2.9 – 12.7 Eligible | 16 | 9.7 mos | 4.7 – 13.6 HR: 0.90 HR CI: 0.36 – 2.29 P: 0.832 Note: Not eligible = reference. No OS difference detected.

[Slide 1] 14:45 - 16:15 Rapid oral: Gastrointestinal tumours, lower digestive CHAIRS: LISA SALVATORE, KAI-KEEN SHIU, RONA YAEGER PRABITAS: Study Design A pragmatic, investigator-initiated, open-label, randomized phase 3 trial Japan Registry of Clinical Trials jRCTs041230120 Patients - Histologically confirmed mCRC - Prior regimens including - Fluoropyrimidine, IRI, OX - Anti-VEGF agents - Anti-EGFR antibody (if indicated) - ECOG PS 0-2 Stratification factors ECOG PS (0/1/2), RAS status (wild-type/mutant), Facility Planned N=890 R 1:1 Conventional FTD/TPI+BEV: FTD/TPI 35 mg/m2 BID, days 1-5 and 8-12; BEV 5 mg/kg IV, days 1 and 15; 28-day cycle Biweekly FTD/TPI+BEV: FTD/TPI 35 mg/m2 BID, days 1-5; BEV 5 mg/kg IV, day 1; 14-day cycle Primary endpoint: OS (non-inferiority) Non-inferiority margin: HR 1.25 Secondary endpoints: PFS, TTF, ORR, DCR, safety Pragmatic elements - Broad eligibility criteria - Starting dose, dose modification and CT timing at physician's discretion - Only 10 prespecified AEs, collected twice (first 2 months and whole treatment period) Hiroya Taniguchi Hiroya Taniguchi - Biweekly versus conventional trifluridine/tipiracil plus bevacizumab for refractory metastatic colorectal cancer: The pragmatic phase 3 PRABITAS trial [Slide 2] 14:45 - 16:15 Rapid oral: Gastrointestinal tumours, lower digestive CHAIRS: LISA SALVATORE, KAI-KEEN SHIU, RONA YAEGER Overall Survival (ITT population) | N | Events | Median OS, mo (95% CI) Conventional | 486 | 355 | 10.3 (9.5-11.6) Biweekly | 484 | 336 | 11.3 (10.2-12.5) HR 0.93 (95% CI 0.80-1.08) p<0.0001 for non-inferiority Non-inferiority margin: HR 1.25 Data cutoff: May 20, 2026 Median follow-up: 17.9 mo Time (months): 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 | 24 | 27 Number at risk Conventional | 486 | 457 | 359 | 265 | 186 | 111 | 61 | 33 | 13 | 0 Biweekly | 484 | 451 | 366 | 288 | 203 | 129 | 63 | 33 | 10 | 0 Hiroya Taniguchi Hiroya Taniguchi - Biweekly versus conventional trifluridine/tipiracil plus bevacizumab for refractory metastatic colorectal cancer: The pragmatic phase 3 PRABITAS trial [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival (ITT) Conventional FTD/TPI+BEV | 486 | 10.3 mo | 9.5-11.6 Biweekly FTD/TPI+BEV | 484 | 11.3 mo | 10.2-12.5 HR: 0.93 HR CI: 0.80-1.08 P: <0.0001 for non-inferiority Note: Non-inferiority margin HR 1.25. Data cutoff May 20, 2026; median follow-up 17.9 mo. [Slide 3] 14:45 - 16:15 Rapid oral: Gastrointestinal tumours, lower digestive CHAIRS: LISA SALVATORE, KAI-KEEN SHIU, RONA YAEGER Progression-Free Survival and Tumor Response Progression-Free Survival | N | Events | Median PFS, mo (95%CI) Conventional | 486 | 466 | 3.9 (3.6-4.4) Biweekly | 484 | 449 | 3.9 (3.6-4.4) HR 0.93 (95%CI 0.81-1.06) Time (months): 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 | 24 | 27 Number at risk Conventional | 486 | 307 | 148 | 73 | 33 | 16 | 9 | 3 | 0 | 0 Biweekly | 484 | 284 | 161 | 97 | 45 | 26 | 7 | 2 | 2 | 0 Tumor Response | Conventional n=457 | Biweekly n=449 ORR | 2% | 2% DCR | 47% | 52% Patients with measurable disease (RECIST v1.1) Hiroya Taniguchi Hiroya Taniguchi - Biweekly versus conventional trifluridine/tipiracil plus bevacizumab for refractory metastatic colorectal cancer: The pragmatic phase 3 PRABITAS trial [Survival data captured as a table — no point estimates read from the curve] Endpoint: Progression-free survival Conventional FTD/TPI+BEV | 486 | 3.9 mo | 3.6-4.4 Biweekly FTD/TPI+BEV | 484 | 3.9 mo | 3.6-4.4 HR: 0.93 HR CI: 0.81-1.06 Note: Slide also shows tumor response: ORR 2% vs 2%, DCR 47% vs 52% (conventional n=457, biweekly n=449; RECIST v1.1). [Slide 4] 14:45 - 16:15 Rapid oral: Gastrointestinal tumours, lower digestive CHAIRS: LISA SALVATORE, KAI-KEEN SHIU, RONA YAEGER Treatment-related Adverse Events Conventional | Biweekly Hematological AEs (Grade >=3) / G-CSF use Neutropenia** | 39% | 13% Leukopenia** | 18% | 7% Febrile neutropenia** | 5% | 1% G-CSF use** | 13% | 3% Non-hematological AEs (Grade >=2) Anorexia** | 26% | 15% Nausea* | 18% | 12% Vomiting | 5% | 3% Fatigue | 17% | 15% Diarrhea** | 10% | 4% Oral mucositis | 5% | 6% Infection | 9% | 9% *p<0.05, **p<0.001 CTCAE v5.0; worst grade: first 2 months for neutropenia/leukopenia, entire treatment period for other AEs. Hiroya Taniguchi Hiroya Taniguchi - Biweekly versus conventional trifluridine/tipiracil plus bevacizumab for refractory metastatic colorectal cancer: The pragmatic phase 3 PRABITAS trial

[Slide 2] Overall Survival (ITT population) | N | Events | Median OS, mo (95% CI) Conventional | 486 | 355 | 10.3 (9.5-11.6) Biweekly | 484 | 336 | 11.3 (10.2-12.5) HR 0.93 (95% CI 0.80-1.08) p<0.0001 for non-inferiority Non-inferiority margin: HR 1.25 Data cutoff: May 20, 2026 Median follow-up: 17.9 mo Time (months): 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 | 24 | 27 Number at risk Conventional | 486 | 457 | 359 | 265 | 186 | 111 | 61 | 33 | 13 | 0 Biweekly | 484 | 451 | 366 | 288 | 203 | 129 | 63 | 33 | 10 | 0 Hiroya Taniguchi [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival (ITT) Conventional FTD/TPI+BEV | 486 | 10.3 mo | 9.5-11.6 Biweekly FTD/TPI+BEV | 484 | 11.3 mo | 10.2-12.5 HR: 0.93 HR CI: 0.80-1.08 P: <0.0001 for non-inferiority Note: Non-inferiority margin HR 1.25. Data cutoff May 20, 2026; median follow-up 17.9 mo. [Slide 3] Progression-Free Survival and Tumor Response Progression-Free Survival | N | Events | Median PFS, mo (95%CI) Conventional | 486 | 466 | 3.9 (3.6-4.4) Biweekly | 484 | 449 | 3.9 (3.6-4.4) HR 0.93 (95%CI 0.81-1.06) Time (months): 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 | 24 | 27 Number at risk Conventional | 486 | 307 | 148 | 73 | 33 | 16 | 9 | 3 | 0 | 0 Biweekly | 484 | 284 | 161 | 97 | 45 | 26 | 7 | 2 | 2 | 0 Tumor Response | Conventional n=457 | Biweekly n=449 ORR | 2% | 2% DCR | 47% | 52% Patients with measurable disease (RECIST v1.1) Hiroya Taniguchi [Survival data captured as a table — no point estimates read from the curve] Endpoint: Progression-free survival Conventional FTD/TPI+BEV | 486 | 3.9 mo | 3.6-4.4 Biweekly FTD/TPI+BEV | 484 | 3.9 mo | 3.6-4.4 HR: 0.93 HR CI: 0.81-1.06 Note: Slide also shows tumor response: ORR 2% vs 2%, DCR 47% vs 52% (conventional n=457, biweekly n=449; RECIST v1.1). [Slide 4] Treatment-related Adverse Events Conventional | Biweekly Hematological AEs (Grade >=3) / G-CSF use Neutropenia** | 39% | 13% Leukopenia** | 18% | 7% Febrile neutropenia** | 5% | 1% G-CSF use** | 13% | 3% Non-hematological AEs (Grade >=2) Anorexia** | 26% | 15% Nausea* | 18% | 12% Vomiting | 5% | 3% Fatigue | 17% | 15% Diarrhea** | 10% | 4% Oral mucositis | 5% | 6% Infection | 9% | 9% *p<0.05, **p<0.001 CTCAE v5.0; worst grade: first 2 months for neutropenia/leukopenia, entire treatment period for other AEs. Hiroya Taniguchi

[Slide 1] 14-day versus 28-day trifluridine–tipiracil plus bevacizumab for refractory metastatic colorectal cancer (PRABITAS): a pragmatic, multicentre, open-label, randomised, phase 3, non-inferiority trial in Japan Tomoki Sakakida*, Hiroya Taniguchi*, Masako Asayama*, Yusuke Suwa, Takaya Shimura, Seiichiro Mitani, Akitaka Makiyama, Hiroyuki Arai, Hiroshi Matsuoka, Mitsuru Yokota, Yoshinao Takano, Kyoko Kato, Hisanobu Oda, Takeshi Kajiwara, Kohei Ogawa, Nobuhiro Shibata, Kentaro Sawada, Takato Yomoda, Toshihiro Kudo, Hiroyuki Okuda, Yuta Okumura, Chihiro Tanaka, Masato Nakamura, Takashi Tsuda, Takahito Kitajima, Hiroshi Tsukuda, Eiji Shinozaki, Mitsuhiro Furuta, Tomomi Kashiwada, Koichi Okuya, Hiroyuki Mushiake, Junichi Hasegawa, Koei Nihei, Takashi Mizushima, Takashi Mine, Takuro Mizukami, Chiho Kudo, Ayako Mishima, Kenta Murotani, Toshiki Masuishi, on behalf of the PRABITAS investigators† Summary Background Trifluridine–tipiracil (FTD–TPI) plus bevacizumab is the standard treatment for refractory metastatic colorectal cancer. However, the conventional schedule is frequently associated with haematological and gastrointestinal toxicity, which is clinically significant and relevant to patients. A simplified 14-day schedule has been proposed to reduce toxicity, but whether this treatment preserves survival has not been established in a randomised trial. We therefore aimed to assess whether the 14-day schedule is non-inferior to the conventional schedule for overall survival. Methods PRABITAS was a pragmatic, multicentre, open-label, randomised, phase 3, non-inferiority trial conducted in 178 hospitals in Japan. Patients with refractory metastatic colorectal cancer were randomly assigned (1:1) using a minimisation algorithm, stratified by Eastern Cooperative Oncology Group performance status, RAS mutational status, and institution, to FTD–TPI (35 mg/m² orally twice daily) plus bevacizumab (5 mg/kg intravenously) on either a conventional 28-day (FTD–TPI on days 1–5 and 8–12; bevacizumab on days 1 and 15) or 14-day schedule (FTD–TPI on days 1–5; bevacizumab on day 1). Eligibility criteria were broadened compared with those of conventional explanatory trials, and trial procedures were aligned with routine clinical practice. The primary endpoint was overall survival, analysed in the intention-to-treat population, with a prespecified non-inferiority margin of 1·25. Safety was assessed in all patients who received at least one dose. Missing data were not imputed. This trial is registered with the Japan Registry of Clinical Trials, jRCTs041230120, and is closed to enrolment. Findings Between Dec 26, 2023, and May 20, 2025, we randomly assigned 972 patients, of whom 970 (99%) constituted the intention-to-treat population after two were excluded because they were randomly assigned before site approval (486 [50%] to the 28-day treatment group, 292 [60%] men and 194 [40%] women]; and 484 [50%] to the 14-day treatment group, 264 [55%] men and 220 [45%] women). Median follow-up was 17·9 months (IQR 14·3–22·6). Median overall survival was 10·3 months (95% CI 9·5–11·6) in the 28-day group and 11·3 months (10·2–12·5) in the 14-day group (stratified hazard ratio 0·93 [95% CI 0·80–1·08]). The 12-month overall survival rates were 44·0% (95% CI 39·5–48·5) in the 28-day group and 47·5% (42·9–52·0) in the 14-day group. Neutropenia grade 3 or worse was reduced in the 14-day group (13% vs 39%; absolute difference 25 percentage points [95% CI 20–31]), as was anorexia grade 2 or worse (15% vs 26%; absolute difference 11 percentage points [6–16]). Two treatment-related deaths occurred in the 14-day group and none in the 28-day group. Interpretation A 14-day schedule of FTD–TPI plus bevacizumab was non-inferior to the conventional 28-day schedule for overall survival, while reducing haematological and gastrointestinal toxicity. These findings support this simplified schedule as the new standard for FTD–TPI plus bevacizumab in refractory metastatic colorectal cancer. Funding Japan Society for the Promotion of Science KAKENHI Grant-in-Aid for Challenging Research (Pioneering). Copyright © 2026 Elsevier Ltd. All rights reserved, including those for text and data mining, AI training, and similar technologies. Introduction Colorectal cancer is a leading cause of cancer-related mortality worldwide, with approximately 1·9 million new cases and 900 000 deaths annually.1 For patients with metastatic colorectal cancer who experience disease progression after fluoropyrimidine-based, oxaliplatin-based, and irinotecan-based chemotherapy, international guidelines recommend sequential use of later-line Sidebar: Published Online October 24, 2026 https://doi.org/10.1016/S0140-6736(26)01782-4 | See Online/Comment https://doi.org/10.1016/S0140-6736(26)02034-9 | *Contributed equally to this work as joint first authors | †Investigators listed in the appendix (p 2) | Department of Clinical Oncology, Aichi Cancer Center Hospital, Nagoya, Japan (T Sakakida MD, H Taniguchi MD, A Mishima, T Masuishi MD); Department of Molecular Gastroenterology and Hepatology, Kyoto Prefectural University of Medicine, Kyoto, Japan (T Sakakida); Department of Gastroenterology, Saitama Cancer Center, Kitaadachigun, Japan (M Asayama MD); Department of Surgery, Gastroenterological Center, Yokohama City University Medical Center, Yokohama, Japan (Y Suwa MD); Department of Gastroenterology and Metabolism, Nagoya City University Graduate School of Medical Sciences, Nagoya, Japan (T Shimura MD); Department of Medical Oncology, Kindai University Faculty of Medicine, Sakai, Japan (S Mitani MD); Cancer Center, Gifu University Hospital, Gifu, Japan (A Makiyama MD); Department of Clinical Oncology, St Marianna University School of Medicine, Kawasaki, Japan (H Arai MD, T Mizukami MD); Department of Surgery, Fujita Health University Hospital, Toyoake, Japan (Prof H Matsuoka MD); www.thelancet.com Published online October 24, 2026 https://doi.org/10.1016/S0140-6736(26)01782-4

[Slide 1] Articles 14-day versus 28-day trifluridine–tipiracil plus bevacizumab for refractory metastatic colorectal cancer (PRABITAS): a pragmatic, multicentre, open-label, randomised, phase 3, non-inferiority trial in Japan Tomoki Sakakida*, Hiroya Taniguchi*, Masako Asayama*, Yusuke Suwa, Takaya Shimura, Seiichiro Mitani, Akitaka Makiyama, Hiroyuki Arai, Hiroshi Matsuoka, Mitsuru Yokota, Yoshinao Takano, Kyoko Kato, Hisanobu Oda, Takeshi Kajiwara, Kohei Ogawa, Nobuhiro Shibata, Kentaro Sawada, Takato Yomoda, Toshihiro Kudo, Hiroyuki Okuda, Yuta Okumura, Chihiro Tanaka, Masato Nakamura, Takashi Tsuda, Takahito Kitajima, Hiroshi Tsukuda, Eiji Shinozaki, Mitsuhiro Furuta, Tomomi Kashiwada, Koichi Okuya, Hiroyuki Mushiake, Junichi Hasegawa, Koei Nihei, Takashi Mizushima, Takashi Mine, Takuro Mizukami, Chiho Kudo, Ayako Mishima, Kenta Murotani, Toshiki Masuishi, on behalf of the PRABITAS investigators†

[Slide 1] CheckMate 8HW Study design: CheckMate 8HW • CheckMate 8HW is a randomized, multicenter, open-label phase 3 trial^a Key eligibility criteria: • Histologically confirmed unresectable or metastatic CRC • MSI-H/dMMR status by local testing • Immunotherapy-naïve • ECOG PS 0 or 1 Stratification factors: • Prior lines of treatment (0 vs 1 vs ≥ 2) • Primary tumor location (right vs left) R 2:2:1 n = 353: NIVO 240 mg Q2W for 6 doses, followed by NIVO 480 mg Q4W^b n = 354: NIVO 240 mg + IPI 1 mg/kg Q3W for 4 doses,^c followed by NIVO 480 mg Q4W^b n = 132: Investigator's choice of chemo^d (mFOLFOX6 and FOLFIRI ± bevacizumab or cetuximab) Treatment until disease progression, unacceptable toxicity, withdrawal of consent (all arms), or a maximum treatment duration of 2 years (NIVO and NIVO + IPI arms only) Dual primary endpoints in patients with centrally confirmed MSI-H/dMMR status^e: • PFS by BICR^f (NIVO + IPI vs chemo in 1L) • PFS by BICR^f (NIVO + IPI vs NIVO across all lines) Key secondary endpoints in patients with centrally confirmed MSI-H/dMMR status^e: • PFS by BICR^f (NIVO + IPI vs NIVO in 1L) • ORR • OS Other key endpoints in all treated patients • Safety • At data cutoff (May 28, 2026), median follow-up^g for all randomized patients was 68.1 (range, 37.6–81.4) months across all lines and 63.0 (range, 37.6–80.2) months in the 1L population • The final OS analysis was initiated based on achieving 3 years of minimum follow-up ^aClinicalTrials.gov. NCT04008030. ^bPatients with ≥ 2 prior lines are randomized only to the NIVO or NIVO + IPI arms. ^cPatients can continue NIVO treatment upon early IPI discontinuation. ^dPatients receiving investigator's choice of chemo are eligible to receive NIVO + IPI upon progression (crossover treatment). ^eConfirmed using either immunohistochemistry and/or polymerase chain reaction-based tests. ^fEvaluated using RECIST v1.1. ^gTime between randomization and data cutoff across all 3 treatment arms. 4 [Slide 2] CheckMate 8HW Overall survival: NIVO+IPI vs NIVO across all lines Landmark labels: NIVO+IPI 90% (12 mo), 85% (24), 81% (36), 77% (48), 74% (60); NIVO 83% (12), 72% (24), 69% (36), 65% (48), 62% (60) Centrally confirmed MSI-H/dMMR | NIVO + IPI (n = 296) | NIVO (n = 286) Events (%) | 78 (26) | 111 (39) Median OS, mo 95% CI | NR (NE-NE) | NR (NE-NE) HR (95% CI) | 0.61 (0.46-0.82) P value | 0.0009^a Months: 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 No. at risk NIVO+IPI 296 279 273 267 264 260 257 255 251 247 245 239 238 226 216 209 199 186 176 166 151 141 114 78 62 31 3 0 NIVO 286 275 263 252 238 228 220 213 207 202 201 199 195 185 176 170 163 151 141 134 122 115 94 57 49 17 3 0 • In patients with centrally confirmed MSI-H/dMMR mCRC, NIVO + IPI demonstrated statistically significant OS benefit vs NIVO across all lines (HR, 0.61, 95% CI, 0.46-0.82, P = 0.0009) — These data are consistent with those observed in the all-randomized population by local testing (HR, 0.64, [95% CI, 0.51-0.81]) — OS results were consistent across most prespecified subgroups Median follow-up for all randomized patients with centrally confirmed MSI-H/dMMR mCRC was 68.1 (range, 37.6-81.4) months. ^aStatistical boundary for significance, 0.0058. 7 [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS (centrally confirmed MSI-H/dMMR, all lines) NIVO + IPI | 296 | NR | NE-NE NIVO | 286 | NR | NE-NE HR: 0.61 HR CI: 0.46-0.82 P: 0.0009 {'timepoint': '12 mo', 'arm': 'NIVO + IPI', 'value': '90%', 'ci': None} {'timepoint': '12 mo', 'arm': 'NIVO', 'value': '83%', 'ci': None} {'timepoint': '24 mo', 'arm': 'NIVO + IPI', 'value': '85%', 'ci': None} {'timepoint': '24 mo', 'arm': 'NIVO', 'value': '72%', 'ci': None} {'timepoint': '36 mo', 'arm': 'NIVO + IPI', 'value': '81%', 'ci': None} {'timepoint': '36 mo', 'arm': 'NIVO', 'value': '69%', 'ci': None} {'timepoint': '48 mo', 'arm': 'NIVO + IPI', 'value': '77%', 'ci': None} {'timepoint': '48 mo', 'arm': 'NIVO', 'value': '65%', 'ci': None} {'timepoint': '60 mo', 'arm': 'NIVO + IPI', 'value': '74%', 'ci': None} {'timepoint': '60 mo', 'arm': 'NIVO', 'value': '62%', 'ci': None} Note: Statistical boundary for significance 0.0058. All-randomized by local testing HR 0.64 (0.51-0.81). Median f/u 68.1 mo. [Slide 3] CheckMate 8HW Efficacy summary: PFS and ORR with extended follow-up Patients with centrally confirmed MSI-H/dMMR | All lines: NIVO + IPI (n = 296) | All lines: NIVO (n = 286) | 1L: NIVO + IPI (n = 171) | 1L: NIVO (n = 170) Median PFS, mo (95% CI) | NR (71.7-NE) | 39.3 (23.0-58.7) | NR (64.3-NE) | 60.8 (32.8-NE) HR (95% CI) | 0.60 (0.47-0.77) [all lines] | 0.68 (0.48-0.96) [1L] ORR, n (%) | 213 (72) | 165 (58) | 127 (74) | 104 (61) CR rate, n (%) | 117 (40) | 89 (31) | 70 (41) | 55 (32) DCR, n (%) | 249 (84) | 218 (76) | 145 (85) | 137 (81) • NIVO + IPI continued to show clinical benefit in PFS, and higher ORR rates compared with NIVO across all treatment lines at extended follow-up (descriptive analyses) • Based on descriptive analysis, 1L NIVO + IPI continued to demonstrated clinically meaningful improvements in PFS (HR, 0.21, 95% CI, 0.14-0.32) and higher ORR (74% vs 27%) and CR (41% vs 7%) rates vs chemo, respectively Median follow-up for all randomized patients was 68.1 (range, 37.6-81.4) months across all lines and 62.8 (range, 37.6-80.2) months in the 1L setting. 9 [Slide 4] CheckMate 8HW Summary • CheckMate 8HW delivers 2 firsts in MSI-H/dMMR mCRC — The first significant OS benefit for an IO regimen over chemo in 1L (HR, 0.37, P < 0.0001) — The first significant OS benefit for dual IO over monotherapy across all lines (HR, 0.61, P = 0.0009) • The 1L NIVO + IPI vs NIVO comparison shows a durable, clinically meaningful, and consistent OS benefit (HR, 0.66, P=0.0432), with 5-year OS rates (77% vs 66%), despite not satisfying its prespecified survival statistical threshold (P = 0.0058) • With longer follow-up, descriptive analyses showed consistent PFS and ORR benefit with NIVO + IPI across all comparisons and treatment lines • The safety profile of NIVO + IPI remained stable with longer follow-up, with low frequency of grade 3/4 toxicity despite numerically higher incidence of any-grade TRAEs with the combination, confirming tolerability alongside efficacy • These results reinforce the durable, long-term clinical benefit of NIVO + IPI across all efficacy endpoints tested, supporting it as a global 1L standard of care in MSI-H/dMMR mCRC 11

[Slide 1] CheckMate 8HW Overall survival: NIVO+IPI vs chemo in 1L Centrally confirmed MSI-H/dMMR | NIVO + IPI (n = 171) | Chemo (n = 84) Events (%) | 39 (23) | 41 (49) Median OS, mo (95% CI) | NR (NE-NE) | 50.1 (31.7-NE) HR (95% CI) | 0.37 (0.24-0.57) P value | < 0.0001 Landmark labels (12/24/36/48/60 mo): NIVO+IPI 90% | 85% | 84% | 80% | 77%; Chemo 78% | 66% | 57% | 51% | 46% Months: 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 No. at risk NIVO+IPI | 171 161 158 155 154 152 149 148 146 145 145 143 142 132 123 116 108 96 89 80 66 57 43 27 21 10 1 0 Chemo | 84 78 69 66 62 59 55 53 52 50 49 47 45 41 38 35 33 30 26 25 20 18 15 12 [?] 2 0 0 - In patients with centrally confirmed MSI-H/dMMR mCRC, NIVO + IPI demonstrated statistically significant OS benefit vs chemo in the 1L (HR, 0.37, 95% CI, 0.24-0.57, P < 0.0001) - Benefit from 1L NIVO + IPI observed despite 79.8% (n=67/84) of patients treated with chemo receiving subsequent IO therapy Median follow-up for all randomized patients with centrally confirmed MSI-H/dMMR mCRC in the 1L setting was 62.8 (range, 37.6-80.2) months. aStatistical boundary for significance, 0.0059. [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival, 1L, centrally confirmed MSI-H/dMMR NIVO + IPI | 171 | NR | NE-NE Chemo | 84 | 50.1 mo | 31.7-NE HR: 0.37 HR CI: 0.24-0.57 P: < 0.0001 {'timepoint': '12 mo', 'arm': 'NIVO + IPI', 'value': '90%', 'ci': None} {'timepoint': '24 mo', 'arm': 'NIVO + IPI', 'value': '85%', 'ci': None} {'timepoint': '36 mo', 'arm': 'NIVO + IPI', 'value': '84%', 'ci': None} {'timepoint': '48 mo', 'arm': 'NIVO + IPI', 'value': '80%', 'ci': None} {'timepoint': '60 mo', 'arm': 'NIVO + IPI', 'value': '77%', 'ci': None} {'timepoint': '12 mo', 'arm': 'Chemo', 'value': '78%', 'ci': None} {'timepoint': '24 mo', 'arm': 'Chemo', 'value': '66%', 'ci': None} {'timepoint': '36 mo', 'arm': 'Chemo', 'value': '57%', 'ci': None} {'timepoint': '48 mo', 'arm': 'Chemo', 'value': '51%', 'ci': None} {'timepoint': '60 mo', 'arm': 'Chemo', 'value': '46%', 'ci': None} Note: Median follow-up 62.8 (37.6-80.2) mo; 79.8% (67/84) chemo pts received subsequent IO. Significance boundary 0.0059. [Slide 3] CheckMate 8HW Overall survival: NIVO+IPI vs NIVO in 1L Centrally confirmed MSI-H/dMMR | NIVO + IPI (n = 171) | NIVO (n = 170) Events (%) | 39 (23) | 41 (49)[?] Median OS, mo (95% CI) | NR (NE-NE) | NR (NE-NE) HR (95% CI) | 0.66 (0.44-0.99) P value | 0.0432a Landmark labels (12/24/36/48/60 mo): NIVO+IPI 90% | 85% | 84% | 80% | 77%; NIVO 88% | 78% | 73% | 70% | 66% Months: 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 No. at risk NIVO+IPI | 171 161 158 155 154 152 149 148 146 145 145 143 142 132 123 116 108 96 89 80 66 57 43 27 21 10 1 0 NIVO | 170 168 164 160 150 145 139 135 133 129 128 126 123 116 108 102 96 85 75 69 58 52 39 22 17 5 2 0 - OS favored NIVO+IPI vs NIVO in 1L (HR, 0.66, 95% CI, 0.44-0.99, P = 0.0432), however, the prespecified threshold for statistical significance (P < 0.0058) was not met - Lack of statistical significance likely reflects the smaller sample size and fewer events relative to the all-lines comparison - The 5-year OS rates (77% vs 66%) were consistent with the results across all lines (74% vs 62%) Median follow-up for all randomized patients with centrally confirmed MSI-H/dMMR mCRC in the 1L setting was 62.8 (range, 37.6-80.2) months. aStatistical boundary for significance, 0.0058. [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival, 1L, centrally confirmed MSI-H/dMMR NIVO + IPI | 171 | NR | NE-NE NIVO | 170 | NR | NE-NE HR: 0.66 HR CI: 0.44-0.99 P: 0.0432 (threshold P < 0.0058 not met) {'timepoint': '12 mo', 'arm': 'NIVO + IPI', 'value': '90%', 'ci': None} {'timepoint': '24 mo', 'arm': 'NIVO + IPI', 'value': '85%', 'ci': None} {'timepoint': '36 mo', 'arm': 'NIVO + IPI', 'value': '84%', 'ci': None} {'timepoint': '48 mo', 'arm': 'NIVO + IPI', 'value': '80%', 'ci': None} {'timepoint': '60 mo', 'arm': 'NIVO + IPI', 'value': '77%', 'ci': None} {'timepoint': '12 mo', 'arm': 'NIVO', 'value': '88%', 'ci': None} {'timepoint': '24 mo', 'arm': 'NIVO', 'value': '78%', 'ci': None} {'timepoint': '36 mo', 'arm': 'NIVO', 'value': '73%', 'ci': None} {'timepoint': '48 mo', 'arm': 'NIVO', 'value': '70%', 'ci': None} {'timepoint': '60 mo', 'arm': 'NIVO', 'value': '66%', 'ci': None} Note: NIVO events not reliably legible. Not statistically significant. Median follow-up 62.8 (37.6-80.2) mo.

[Slide 1] A Centrally confirmed MSI-H/dMMR | NIVO+IPI (n = 296) | NIVO (n = 286) Total censored patients, n | 218 | 175 Median OS, months 95% CI | NR NE-NE | NR NE-NE HR (95% CI) | 0.61 (0.46-0.82) P = 0.0009 (Threshold for significance P < 0.006) Landmark labels: NIVO+IPI 90% (12 mo), 85% (24), 81% (36), 77% (48), 74% (60); NIVO 83%, 72%, 69%, 65%, 62% Overall survival probability (%) / Months 0-81 No. at risk NIVO+IPI 296 279 273 267 264 260 257 255 251 247 245 239 238 226 216 209 199 186 176 166 151 141 114 78 62 31 3 0 NIVO 286 275 263 252 238 228 220 213 207 202 201 199 195 185 176 170 163 151 141 134 122 115 94 57 49 17 3 0 B 1L centrally confirmed MSI-H/dMMR | NIVO+IPI (n = 171) | NIVO (n = 170) Total censored patients, n | 132 | 114 Median OS, months 95% CI | NR NE-NE | NR NE-NE HR (95% CI) | 0.66 (0.44-0.99) P = 0.0432 (Threshold for significance P < 0.006) Landmark labels: NIVO+IPI 90%, 85%, 84%, 80%, 77%; NIVO 88%, 78%, 73%, 70%, 66% No. at risk NIVO+IPI 171 161 158 155 154 152 149 148 146 145 145 143 142 132 123 116 108 96 89 80 66 57 43 27 21 10 1 0 NIVO 170 168 164 160 150 145 139 135 133 129 128 126 123 116 108 102 96 85 75 69 58 52 39 22 17 5 2 0 [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival (centrally confirmed MSI-H/dMMR, all lines) - panel A NIVO+IPI | 296 | NR | NE-NE NIVO | 286 | NR | NE-NE HR: 0.61 HR CI: 0.46-0.82 P: 0.0009 {'timepoint': '12 mo', 'arm': 'NIVO+IPI', 'value': '90%', 'ci': None} {'timepoint': '24 mo', 'arm': 'NIVO+IPI', 'value': '85%', 'ci': None} {'timepoint': '36 mo', 'arm': 'NIVO+IPI', 'value': '81%', 'ci': None} {'timepoint': '48 mo', 'arm': 'NIVO+IPI', 'value': '77%', 'ci': None} {'timepoint': '60 mo', 'arm': 'NIVO+IPI', 'value': '74%', 'ci': None} {'timepoint': '12 mo', 'arm': 'NIVO', 'value': '83%', 'ci': None} {'timepoint': '24 mo', 'arm': 'NIVO', 'value': '72%', 'ci': None} {'timepoint': '36 mo', 'arm': 'NIVO', 'value': '69%', 'ci': None} {'timepoint': '48 mo', 'arm': 'NIVO', 'value': '65%', 'ci': None} {'timepoint': '60 mo', 'arm': 'NIVO', 'value': '62%', 'ci': None} Note: Total censored 218 vs 175. Threshold for significance P < 0.006. Panel B (1L, n=171 vs 170): HR 0.66 (0.44-0.99), P=0.0432; landmarks NIVO+IPI 90/85/84/80/77%, NIVO 88/78/73/70/66%.

[Slide 1] 14:45 - 16:15 Rapid oral: Gastrointestinal tumours, lower digestive CHAIRS: LISA SALVATORE, KAI-KEEN SHIU, RONA YAEGER CheckMate 8HW Overall survival: NIVO+IPI vs NIVO across all lines Landmark labels: NIVO+IPI 90% (12 mo), 85% (24), 81% (36), 77% (48), 74% (60); NIVO 83% (12), 72% (24), 69% (36), 65% (48), 62% (60) Centrally confirmed MSI-H/dMMR | NIVO + IPI (n = 296) | NIVO (n = 286) Events (%) | 78 (26) | 111 (39) Median OS, mo 95% CI | NR (NE-NE) | NR (NE-NE) HR (95% CI) | 0.61 (0.46-0.82) P value | 0.0009^a Months: 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 No. at risk NIVO+IPI 296 279 273 267 264 260 257 255 251 247 245 239 238 226 216 209 199 186 176 166 151 141 114 78 62 31 3 0 NIVO 286 275 263 252 238 228 220 213 207 202 201 199 195 185 176 170 163 151 141 134 122 115 94 57 49 17 3 0 • In patients with centrally confirmed MSI-H/dMMR mCRC, NIVO + IPI demonstrated statistically significant OS benefit vs NIVO across all lines (HR, 0.61, 95% CI, 0.46-0.82, P = 0.0009) — These data are consistent with those observed in the all-randomized population by local testing (HR, 0.64, [95% CI, 0.51-0.81]) — OS results were consistent across most prespecified subgroups Median follow-up for all randomized patients with centrally confirmed MSI-H/dMMR mCRC was 68.1 (range, 37.6-81.4) months. ^aStatistical boundary for significance, 0.0058. 7 Sara Lonardi Final overall survival results from CheckMate 8HW: Nivolumab plus ipilimumab versus nivolumab or chemotherapy in MSI-H/dMMR metastatic colorectal cancer ALICANTE AUDITORIUM - HALL 6 [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS (centrally confirmed MSI-H/dMMR, all lines) NIVO + IPI | 296 | NR | NE-NE NIVO | 286 | NR | NE-NE HR: 0.61 HR CI: 0.46-0.82 P: 0.0009 {'timepoint': '12 mo', 'arm': 'NIVO + IPI', 'value': '90%', 'ci': None} {'timepoint': '12 mo', 'arm': 'NIVO', 'value': '83%', 'ci': None} {'timepoint': '24 mo', 'arm': 'NIVO + IPI', 'value': '85%', 'ci': None} {'timepoint': '24 mo', 'arm': 'NIVO', 'value': '72%', 'ci': None} {'timepoint': '36 mo', 'arm': 'NIVO + IPI', 'value': '81%', 'ci': None} {'timepoint': '36 mo', 'arm': 'NIVO', 'value': '69%', 'ci': None} {'timepoint': '48 mo', 'arm': 'NIVO + IPI', 'value': '77%', 'ci': None} {'timepoint': '48 mo', 'arm': 'NIVO', 'value': '65%', 'ci': None} {'timepoint': '60 mo', 'arm': 'NIVO + IPI', 'value': '74%', 'ci': None} {'timepoint': '60 mo', 'arm': 'NIVO', 'value': '62%', 'ci': None} Note: Statistical boundary for significance 0.0058. All-randomized by local testing HR 0.64 (0.51-0.81). Median f/u 68.1 mo.

[Slide 1] Study Design and Methods Multicenter, single-arm, phase II trial (NCT06640166) Inclusion Criteria • mCRC • BRAF V600E • PD on EC in 2nd line • PS 0-1 • Best response to EC: CR, PR or SD lasting ≥3 months. Encorafenib plus cetuximab in combination with FOLFIRI (EC beyond progression) Experimental treatment • Encorafenib 300 mg (75 mgx4 hard capsules) po qd; • Cetuximab 500 mg/m2 iv Q2W; • FOLFIRI iv Q2W (Irinotecan 180 mg/m2, Folinic Acid 400 mg/m2, 5Fluorouracil 400 mg/m2 iv bolus and 2400 mg/m2 iv continuous infusion over 46-48 hours) Primary endpoint Investigator-assessed 6-month PFS rate in the ITT population. Secondary endpoint • mPFS; mOS; • ORR; DCR (per RECIST V1.1); • Safety. Statistical hypothesis Binomial test (single-stage, one-sided α = 0.05, 80% power) H0=10% H1=30% Required to enroll 25 patients and to observe ≥7 pts alive without progression at 6 months to reject the null hypothesis, supporting further investigation. Maria Alessandra Calegari [Slide 2] Results – Primary Endpoint Investigator-assessed 6-month PFS rate in the ITT population Patients evaluable for 6-month PFS rate (n = 18) PFS > 6 months: 12* | PFS < 6 months: 6 6-month PFS rate 66.7% (12/18) * Positivity threshold ≥7 pts alive without progression at 6 months [Swimmer plot: Individual patients; Progression Free Survival (months) 0-20; dashed line at 6 months; legend: Progressive disease, On treatment ongoing, Evaluable for primary endpoint, Not yet evaluable for primary endpoint] Data cutoff: May 8, 2026 Abbreviations: ITT, intention to treat; PFS, progression free survival; pt, patient Maria Alessandra Calegari [Slide 3] Results – Survivals Progression Free Survival Endpoint | Estimate (95%CI) Median PFS, months | 7.1 (6.2 – NA) 6-month PFS rate | 69.5% (51.7 – 93.6) 12-month PFS rate | 22.8% (8.8 – 59.0) Median FU 16.1 months No. at risk: 25 15 12 3 3 3 2 0 0 Events: 0 4 6 13 13 13 13 14 14 Months 0 3 6 9 12 15 18 21 24 Overall Survival Endpoint | Estimate (95%CI) Median OS, months | NR 6-month OS rate | 94.1% (83.6 – 100.0) 12-month OS rate | 83.7% (64.5 – 100.0) Median FU 7.6 months No. at risk: 25 18 14 10 8 7 4 2 0 Events: 0 0 1 1 2 4 5 5 5 Data cutoff: May 8, 2026 Abbreviations: CI, confidence interval; FU, follow up; NA, not assessable; NR, not reached; PFS, progression free survival; OS, overall survival Maria Alessandra Calegari [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS and OS (single arm) Encorafenib + cetuximab + FOLFIRI (PFS) | 25 | 7.1 months | 6.2 – NA Encorafenib + cetuximab + FOLFIRI (OS) | 25 | NR {'timepoint': '6-month PFS', 'arm': 'EC + FOLFIRI', 'value': '69.5%', 'ci': '51.7 – 93.6'} {'timepoint': '12-month PFS', 'arm': 'EC + FOLFIRI', 'value': '22.8%', 'ci': '8.8 – 59.0'} {'timepoint': '6-month OS', 'arm': 'EC + FOLFIRI', 'value': '94.1%', 'ci': '83.6 – 100.0'} {'timepoint': '12-month OS', 'arm': 'EC + FOLFIRI', 'value': '83.7%', 'ci': '64.5 – 100.0'} Note: Events = final value in printed events row. Median FU 16.1 mo (PFS panel), 7.6 mo (OS panel). Data cutoff May 8, 2026. [Slide 4] Results – ORR, DCR Waterfall plot Endpoint | Estimate (95%CI) Objective response rate | 15.0% (3.2-37.9) Disease control rate | 80% (56.3-94.3) Best change from baseline, % Pt 5, Pt 21, Pt 11, Pt 17, Pt 18, Pt 9, Pt 2, Pt 7, Pt 3, Pt 1, Pt 4, Pt 12, Pt 14, Pt 13, Pt 6, Pt 20, Pt 16, Pt 15, Pt 10, Pt 8 Best response: PR, SD, PD Spider plot Change from baseline, % / Time from baseline, months (0-21) Best response: PR, SD, PD; ▲ Disease progression; * Surgery Data cutoff: May 8, 2026 Abbreviations: CI, confidence interval; DCR, disease control rate; ORR, overall response rate; PD, progressive disease; PR, partial response; SD, stable disease Maria Alessandra Calegari

[Slide 1] Conclusions ECLYPse met its primary endpoint, with a 6-month PFS rate of 67% (12/18 pts), a mPFS of 7.1 months, and no new safety signal. Continuing EC beyond progression in combination with ChT appears to extend disease control in this poor-prognosis population. This strategy (EC beyond progression in combination with ChT) should be furtherly assessed in the post-BREAKWATER era (2nd line setting). Maria Alessandra Calegari

[Slide 3] MAJOR STUDIES TO WATCH CANCER VACCINES INTerpath-001 Can individualised mRNA neoantigens help prevent melanoma relapse? KRAS DIRECTED KRASCENDO-1 Will a head-to-head trial raise the standard for KRAS G12C NSCLC? DRUG RESISTANCE SAFFRON Can EGFR/MET biomarker selection overcome acquired resistance? BILIARY TRACT CANCER HARMONi-GI1 Can a PD-1/VEGF bispecific improve first-line BTC care? ANTIBODY–DRUG CONJUGATES TroFuse-005 What role will novel ADCs have in recurrent endometrial cancer? PERIOPERATIVE THERAPY VOLGA Can an ADC–immunotherapy strategy improve bladder cancer care? GASTROESOPHAGEAL CANCER CLARITY-Gastric 01 How far can CLDN18.2-directed treatment extend? [MADRID 2026 ESMO congress logo]

[Slide 1] Pre-specified DFS Analysis: ctDNA Positive vs ctDNA Negative Camupepimut treatment effect was consistent in ctDNA+ and ctDNA- n=119 | Prespecified: ctDNA Positive vs Negative ctDNA: Negative — Camupepimut (–) vs Observation (–): HR: 0.51 (0.25-1.02); P=0.0280 ctDNA: Positive — Camupepimut (+) vs Observation (+): HR: 0.57 (0.22-1.48); P=0.0489 Curve labels: Camupepimut, ctDNA- Median: Not Reached Observation, ctDNA- Median: 21.2 months Camupepimut, ctDNA+ Median: 4.3 months Observation, ctDNA+, Median: 1.9 months Y-axis: Disease-free Survival (%); X-axis: Months (0–30) At risk (months 0, 3, 6, 9, 12, 15, 18, 21, 24, 27) Camupepimut ctDNA –: 62, 57, 50, 42, 37, 35, 21, 14, 5, 0 Observation ctDNA –: 31, 24, 20, 19, 17, 16, 12, 8, 3, 0 Camupepimut ctDNA +: 20, 13, 7, 1, 1, 1, 1, 0, 0, 0 Observation ctDNA +: 6, 2, 1, 1, 0, 0, 0, 0, 0, 0 Eileen M. O'Reilly, MD, FASCO [ESMO Congress Madrid 2026 logo, partially cropped] [Survival data captured as a table — no point estimates read from the curve] Endpoint: Disease-free survival by ctDNA status (pre-specified) Camupepimut, ctDNA- | 62 | Not Reached Observation, ctDNA- | 31 | 21.2 months Camupepimut, ctDNA+ | 20 | 4.3 months Observation, ctDNA+ | 6 | 1.9 months HR: ctDNA-: 0.51; ctDNA+: 0.57 HR CI: ctDNA-: 0.25-1.02; ctDNA+: 0.22-1.48 P: ctDNA-: P=0.0280; ctDNA+: P=0.0489 Note: n per arm taken from printed at-risk row at month 0; total n=119 printed.

[Slide 1] Proportion of patients (%) ; Months 0-108 Single curve flat at 100% with censoring marks from ~40 to ~103 months Number at risk (censored) 0: 111 (0) | 12: 111 (0) | 24: 111 (0) | 36: 111 (0) | 48: 96 (16) | 60: 51 (60) | 72: 29 (82) | 84: 16 (95) | 96: 4 (107) | 108: 0 (111) [Survival data captured as a table — no point estimates read from the curve] Endpoint: Not printed on image (caption describes zero events; endpoint label absent) Neoadjuvant ipilimumab + nivolumab (single arm) | 111 Note: Image shows only axes, a single flat curve at 100%, and number at risk (censored): 111(0), 111(0), 111(0), 111(0), 96(16), 51(60), 29(82), 16(95), 4(107), 0(111) at 0/12/24/36/48/60/72/84/96/108 months. No endpoint name, events, median, or HR printed.

[Slide 1] 3530O - Neoadjuvant short-course radiotherapy followed by camrelizumab and chemotherapy in locally advanced rectal cancer (LARC): The phase 3 UNION trial Speakers Zhen Yu Lin (Wuhan, China) Lecture Time 11:10 - 11:20 Abstract Background Despite the use of neoadjuvant chemoradiotherapy, recurrence and distant metastasis remain common in LARC. Effective immunotherapy-based strategies for the predominantly proficient mismatch repair/microsatellite-stable population remain limited. UNION is the first phase 3 trial to evaluate a neoadjuvant strategy incorporating radiotherapy, immunotherapy, and chemotherapy in LARC. Here we present the prespecified event-free survival (EFS) analysis. Methods Patients with cT3-4 or N+ rectal adenocarcinoma (tumor located ≤10 cm from the anal verge) were randomly assigned (1:1) to receive short-course radiotherapy (SCRT; 25 Gy/5f; experimental group) or long-course chemoradiotherapy (50.4 Gy/28f with concurrent capecitabine 825 mg/m2 twice daily; standard of care group) followed by 2 neoadjuvant cycles and 6 adjuvant cycles of capecitabine and oxaliplatin (CAPOX). The experimental group additionally received neoadjuvant and adjuvant camrelizumab (200 mg, day 1) since CAPOX treatment for up to 17 cycles. The primary endpoint was blinded independent review committee (BIRC)-assessed pathological complete response rate. Key secondary endpoints with prespecified α allocation were investigator-assessed EFS and overall survival. Results Between July 2021 and March 2023, 231 patients were randomized to the experimental group (n=113) and standard of care group (n=118). As of October 1, 2025, the median follow-up duration was 36.8 months (95% CI, 34.8-37.8) and 35.6 months (95% CI, 34.3-36.9), respectively. The 3-year EFS rate was significantly higher in the experimental group compared with standard of care (91.3% [95% CI, 83.9-95.4] vs 77.4% [95% CI, 67.9-84.4], hazard ratio=0.33 [95% CI, 0.15-0.71], P=0.0029). BIRC-assessed EFS consistently favored the experimental group. During the whole treatment period, the incidence of grade ≥3 treatment-related adverse events was 51.3% (59/115) and 48.2% (54/112), respectively. Conclusions UNION is the first phase 3 trial to demonstrate a significant improvement in EFS with an immunotherapy-based regimen in LARC. This neoadjuvant strategy represents a new standard of care.

[Slide 1] Background • BGB-B2033 is a novel, rationally designed IgG-based GPC3 x 4-1BB bispecific antibody that binds a unique epitope of 4-1BB – GPC3: tumor-associated antigen highly expressed in multiple tumors, including in ~90% of HCC with any staining(1) – Activates 4-1BB on CD8+ T-cells only in the presence of GPC3-expressing tumor cells, enabling targeted tumor activity – Has an engineered Fc region designed to extend PK exposure while maintaining localized immune activation(2) – Demonstrated ~linear PK and prominent PD effect of soluble 4-1BB increase at dose levels ≥300 mg(3) [Diagram] Activated T cell | Signal 1 | Signal 2 | Cytokines (e.g., IFN-γ, IL2) | TCR/CD3 complex | MHC I | 4-1BB clustering | GPC3 | Tumor cell | GPC3x41BB bsAb Unique 4-1BB binder: specifically designed epitope for efficient receptor clustering and T cell co-stimulation Silenced Fc: mitigate FcγRs-mediated liver toxicity; YTE mutation extends half-life and prolongs exposure Potent GPC3 anchor: strong affinity provides tumor specific 4-1BB activation 1. Ishiguro et al, Sci. Transl. Med. 2017,9, eaal4291. 2. Li et al. Poster #6009 AACR 2025. 3. Chon et al. J Clin Oncol 2026,44:3016. bsAb, bispecific antibody; GPC3, glypican-3; HCC, hepatocellular carcinoma; PD, pharmacodynamics; PK, pharmacokinetics; YTE, M252Y/S254T/T256E. Presenter: Hong Jae Chon, MD, PhD [Slide 2] Study Design (NCT06427941) Phase 1/2 study of BGB-B2033 alone or in combination with tislelizumab with or without bevacizumab in patients with advanced solid tumors Key Eligibility Criteria • Patients with aHCC who received at least 1 line of systemic therapy • Child Pugh A • BCLC C or BCLC B that is not amenable to or progressed after locoregional therapy and not amenable to curative treatment • ECOG PS 0-1 • At least 1 measurable lesion by RECIST 1.1 • Archival tissue or fresh biopsy for GPC3 testing Monotherapy Dose Escalation and Dose Optimization BGB-B2033 Q3W in Patients with aHCC This pooled analysis Dose Escalation: 1000 mg Q3W n=10 | 600 mg Q3W n=14 | 300 mg Q3W n=15 Dose Optimization (R): 900 mg Q3W n=28 | 450 mg Q3W n=27 Total Enrollment = 94 1-100 mg Q3W n=21 Last patient enrolled: 25 August 2026 | DCO: 22 September 2026 | Median study follow-up was 5.5 (range, 0.3-16.7) months IV infusion Q3W | Infusion duration Cycle 1/2: 60 ± 5 min; Cycle ≥3: 30 ± 5 min Endpoints Primary: • Safety • Recommended dose for expansion • MTD Secondary: • Objective response rate by investigator per RECISTv1.1 • Pharmacokinetics/pharmacodynamics • Immunogenicity aHCC, advanced hepatocellular carcinoma; BCLC, Barcelona Clinic Liver Cancer; DCO, data cut-off; ECOG PS, Eastern Cooperative Oncology Group Performance Status; GPC3, Glypican-3; MTD, maximum tolerate dose; Q3W, every 3 weeks; R, randomized; RECIST, Response Evaluation Criteria in Solid Tumors Presenter: Hong Jae Chon, MD, PhD [Slide 3] Consistent Antitumor Activity with BGB-B2033 was Observed Across Dose Escalation and Optimization Cohorts 300 mg Q3W n=14a Escalation cORR: 28.6% 450 mg Q3W n=25a Optimization cORR: 36.0% 600 mg Q3W n=14a Escalation cORR: 35.7% 900 mg Q3W n=27a Optimization cORR: 22.2% 1000 mg Q3W n=10a Escalation cORR: 30.0% Y-axis: Best Change From Baseline (%); legend: Progressive Disease | Stable Disease | Partial Response | Complete Response • 450 mg RP2D was selected based on linear PK, exposure-efficacy, PD plateau, and a flat exposure-safety relationship Data cut-off: 22 September 2026 a Investigator assessed and efficacy-evaluable patients cORR, confirmed objective response rate; PD, pharmacodynamic; PK, pharmacokinetic; RP2D, recommended phase 2 dose. ☆ The patient has an unconfirmed partial response, remains on treatment and is awaiting confirmation Presenter: Hong Jae Chon, MD, PhD [Slide 4] BGB-B2033 Was Well Tolerated, With Limited Grade ≥3 TRAEs Parameter, n (%) | Dose escalation 300 mg n=15 | 600 mg n=14 | 1000 mg n=10 | Dose optimization 450 mg n=27 | 900 mg n=28 | Total (N=94) Any TEAEs | 11 (73.3) | 9 (64.3) | 10 (100) | 26 (96.3) | 22 (78.6) | 78 (83.0) Leading to Discontinuation | 0 | 0 | 2 (20.0) | 1 (3.7) | 0 | 3 (3.2) Grade ≥3 TEAEs | 5 (33.3) | 1 (7.1) | 6 (60.0) | 5 (18.5) | 4 (14.3) | 21 (22.3) Any Serious TEAEs | 0 | 2 (14.3) | 4 (40.0) | 5 (18.5) | 4 (14.3) | 15 (16.0) Any TRAEs | 9 (60.0) | 7 (50.0) | 5 (50.0) | 18 (66.7) | 15 (53.6) | 54 (57.4) Leading to Discontinuation | 0 | 0 | 1 (10.0)a | 1 (3.7)b | 0 | 2 (2.1) Grade ≥3 TRAEs | 2 (13.3) | 0 | 1 (10.0) | 1 (3.7) | 1 (3.6) | 5 (5.3) IMAEs | 2 (13.3)c | 1 (7.1)d | 0 | 5 (18.5)e | 4 (14.3)f | 12 (12.8) TRAEs Occurring in >5% Patients (Grade 1-2 | Grade ≥3): AST increased 18.1% | 2.1% ALT increased 11.7% | 1.1% IRR 8.5% Platelet decreased 7.4% Bilirubin increased 6.4% Pyrexia 6.4% Hypothyroidism 5.3% Data cut-off: 22 September 2026 a Drug eruption (grade 3). b IRR (grade 2). c Hypothyroidism (n=2). d Rash (n=1). e Hypothyroidism (n=2), maculo-papular rash (n=1), rash (n=2). f Hypothyroidism (n=1), immune-mediated hepatitis (n=1), pancreatitis (n=1; grade 3), rash (n=1). AE, adverse event; ALT, alanine aminotransferase; AST, aspartate aminotransferase; IMAE, immune-mediated adverse events; IRR, infusion-related reaction; TEAE, treatment emergent adverse event; TRAE, treatment-emergent treatment-related adverse event. Presenter: Hong Jae Chon, MD, PhD

[Slide 1] [Stage photo of presented slide] Concomitant somatostatin analog use in second-line therapy of gastroenteropancreatic neuroendocrine tumors: Survival results from the retrospective retroSAUNA study Dr Siddharth Chhajlani et al. Study Design • Retrospective study (471 pts) • In 22 NET expert centers • Between 2010 and 2025 (15 years) • Endpoints: PFS and OS Patients with non-functional GEP NET grade 1-2 with progression on 1L SSA → PRRT-substudy: 2L PRRT (n=292) → with 2L SSA (n=167) / w/o 2L SSA (n=125) → TT-substudy: 2L TT (*) (n=179) → with 2L SSA (n=126) / w/o 2L SSA (n=53) (*) with everolimus or sunitinib [Highlighted box] • Reason to continue the SSA ? • Change during the study period? Study Population | PRRT substudy (n=292): PRRT w/o SSA (n=125) | PRRT with SSA (n=167) | TT substudy (n=179): TT w/o SSA (n=53) | TT with SSA (n=126) Median age at start 2L (years, IQR) | 67 (59-73) | 66 (59-73) | 66 (58-72) | 65 (57-73) Sex (n, %) Female | 56 (45%) | 80 (48%) | 23 (43%) | 57 (45%) Sex Male | 69 (55%) | 87 (52%) | 30 (57%) | 69 (55%) NET grade (n, %) NET G1 | 34 (27%) | 58 (35%) | 14 (26%) | 34 (27%) NET G2 | 91 (73%) | 109 (65%) | 39 (74%) | 92 (73%) Location primary (n, %) pancreas | 51 (41%) | 45 (27%) | 42 (79%) | 66 (52%) small intestine | 50 (40%) | 88 (53%) | 3 (6%) | 36 (29%) other GI | 24 (19%) | 34 (20%) | 8 (15%) | 24 (19%) [Slide 4] [Wide stage photo of presented slide] High antitumor activity was seen across DLL3 subgroups, with a trend toward deeper and more durable benefit in the DLL3high subgroup [Waterfall: Best % change from baseline in sum of diameters; DLL3high / DLL3low / ◆ Ongoing] | All (N=44) | DLL3high (n=24) | DLL3low (n=20) ORR, % (95% CI) | 77.3 (63.0–87.2) | 87.5 (69.0–95.7) | 65.0 (43.3–81.9) PR, n (%) | 34 (77.3) | 21 (87.5) | 13 (65.0) DCR, % (95% CI) | 90.9 (78.8–96.4) | 95.8 (79.8–99.3) | 85.0 (64.0–94.8) SD, n (%) | 6 (13.6) | 2 (8.3) | 4 (20.0) PD, n (%) | 4 (9.1) | 1 (4.2) | 3 (15.0) Median DoR, months (95% CI)* | 12.5 (5.5–NC) | NC (5.1–NC) | 7.0 (5.5–NC) 12-month DoR rate, % (95% CI)† | 51 (29–73) | 54 (24–85) | 46 (13–79) Median PFS, months (95% CI)* | 7.7 (6.0–14.5) | 9.5 (6.0–NC) | 7.6 (4.6–9.0) 12-month PFS rate, % (95% CI)† | 35 (17–54) | 47 (18–75) | 27 (5–49) At data cut-off (February 3, 2026), 22 patients remained on obrixtamig treatment (DLL3high: n=15; DLL3low: n=7) *Median duration of follow up in the overall population: DoR, 7.1 months (95% CI: 4.9–11.0); PFS, 11.6 months (95% CI: 5.8–14.3); †Kaplan–Meier estimates CI, confidence interval; DLL3, delta-like ligand 3; DCR, disease control rate; DoR, duration of response; NC, not calculable; ORR, objective response rate; PD, progressive disease; PFS, progression-free survival; PR, partial response; SD, stable disease Jaume Capdevila [ESMO Congress Madrid 2026 logo]

[Slide 1] 16:15 - 17:45 Rapid oral 1: Gastrointestinal tumour, upper digestive CHAIRS: MARIA IGNEZ BRAGHIROLI, FENG WANG, LUCJAN WYRWICZ IKF-AIO RAMIRIS Phase III Study design Key eligibility criteria - Advanced/ metastatic esophagogastric adenocarcinoma - Progression during/ after first-line platinum + fluoropyrimidine therapy - Prior taxane treatment R 1:1 Stratification: - Progression ≤3 vs >3 months after end of first-line therapy - ECOG performance status 0 vs. 1 Arm A: Ramucirumab 8 mg/ kg i.v., d1+15, q28 + FOLFIRI* i.v., d1+15, q28 Arm B: Ramucirumab 8 mg/kg i.v., d1+15, q28 + Paclitaxel 80 mg/m² i.v., d1, 8, 15, q28 Study endpoints Co-primary: Overall survival (OS); Objective overall response rate (ORR) Secondary: Progression-free survival (PFS); Disease control rate (DCR); Safety; Quality of life Study assumptions OS: Aim: 6.0 -> 8.6 months; target HR = 0.70; one-sided α = 0.020; 80% power; n= 318 required ORR: Aim: 10% -> 25%; one-sided α = 0.005; 80% power, n = 298 required -> Final sample size: n = 318 Study cohort n = 319 randomized n = 317 in ITT population (2 patients excluded) n = 304 in PP population Randomized multicenter phase III trial in 47 sites in Germany, Austria and Italy from Oct 2019 until Oct 2024 * FOLFIRI: irinotecan 180 mg/m²; 5-FU bolus 400 mg/m²; leucovorin 400 mg/m²; 5-FU 2400 mg/m² over 46 h Magdalena Scheck Magdalena Scheck FOLFIRI plus ramucirumab versus paclitaxel plus ramucirumab as second-line therapy for taxane-pretreated patients with advanced or metastatic gastroesophageal adenocarcinoma-Final results of the phase III IKFs602/AIO RAMIRIS Study ALMERIA AUDITORIUM - HALL 9 [ESMO Congress Madrid 2026 logo] [Slide 2] 16:15 - 17:45 Rapid oral 1: Gastrointestinal tumour, upper digestive CHAIRS: MARIA IGNEZ BRAGHIROLI, FENG WANG, LUCJAN WYRWICZ Co-Primary endpoints OS and ORR (ITT population) [KM curve: Survival Probability vs Survival time (months)] | Arm A Ram + FOLFIRI n = 158 | Arm B Ram + Pacli n = 159 Median OS (95% CI) | 8.7 [7.6, 10.6] | 7.5 [6.2, 9.3] HR (95.8% CI) P-value | 0.95 [0.74, 1.22] 0.33 Events n (%) | 139 (88%) | 137 (86%) Median follow-up (mo) | 8.2 | 6.6 No. at risk Arm A: 158 101 46 30 12 7 5 3 2 1 1 Arm B: 159 91 45 26 16 8 4 3 2 1 0 | Arm A Ram + FOLFIRI n = 158 | Arm B Ram + Pacli n = 159 | P-value ORR | 40 (25%) | 36 (23%) | 0.29 CR | 4 (3%) | 0 (0%) PR | 36 (23%) | 36 (23%) SD | 61 (39%) | 61 (38%) PD | 33 (21%) | 46 (29%) Missing | 24 (15%) | 16 (10%) OS: HR from unstratified Cox proportional hazards model; p-value from unstratified log-rank test (one-sided). + Censored. The results for OSS and OR for ITT population are consistent with those for PP population. Magdalena Scheck Magdalena Scheck FOLFIRI plus ramucirumab versus paclitaxel plus ramucirumab as second-line therapy for taxane-pretreated patients with advanced or metastatic gastroesophageal adenocarcinoma-Final results of the phase III IKFs602/AIO RAMIRIS Study ALMERIA AUDITORIUM - HALL 9 [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS (co-primary), ITT Arm A Ram + FOLFIRI | 158 | 8.7 | 7.6, 10.6 Arm B Ram + Pacli | 159 | 7.5 | 6.2, 9.3 HR: 0.95 HR CI: 95.8% CI 0.74, 1.22 P: 0.33 (one-sided log-rank) Note: Median follow-up 8.2 vs 6.6 mo. ORR 25% vs 23%, p=0.29. [Slide 3] 16:15 - 17:45 Rapid oral 1: Gastrointestinal tumour, upper digestive CHAIRS: MARIA IGNEZ BRAGHIROLI, FENG WANG, LUCJAN WYRWICZ Progression-free survival ITT Population | Arm A Ram + FOLFIRI n = 158 | Arm B Ram + Pacli n = 159 Median PFS (95% CI) | 5.5 [4.0, 6.3] | 3.7 [2.9, 4.2] HR (95.8% CI) P-value | 0.71 [0.56, 0.90] 0.002 Events n (%) | 142 (90%) | 149 (94%) No. at risk (partly illegible): Arm A 158 63 20 5 4 1 1 1 0; Arm B 159 43 7 3 1 0 PP Population | Arm A Ram + FOLFIRI n = 150 | Arm B Ram + Pacli n = 154 Median PFS (95% CI) | 5.6 [4.1, 6.4] | 3.7 [2.9, 4.3] HR (95.8% CI) P-value | 0.69 [0.55, 0.88] 0.001 Events n (%) | 138 (92%) | 148 (96%) No. at risk (partly illegible): Arm A 150 63 20 5 4 1 1 1 0; Arm B 154 43 7 3 1 0 Ram plus FOLFIRI resulted in a numerical improvement in PFS vs. SOC HR from unstratified Cox proportional hazards model; p-value from unstratified log-rank test (one-sided). + Censored. SOC standard of care Magdalena Scheck Magdalena Scheck FOLFIRI plus ramucirumab versus paclitaxel plus ramucirumab as second-line therapy for taxane-pretreated patients with advanced or metastatic gastroesophageal adenocarcinoma-Final results of the phase III IKFs602/AIO RAMIRIS Study ALMERIA AUDITORIUM - HALL 9 [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS (secondary), ITT and PP ITT Arm A Ram + FOLFIRI | 158 | 5.5 | 4.0, 6.3 ITT Arm B Ram + Pacli | 159 | 3.7 | 2.9, 4.2 PP Arm A Ram + FOLFIRI | 150 | 5.6 | 4.1, 6.4 PP Arm B Ram + Pacli | 154 | 3.7 | 2.9, 4.3 HR: ITT 0.71; PP 0.69 HR CI: 95.8% CI ITT 0.56, 0.90; PP 0.55, 0.88 P: ITT 0.002; PP 0.001 (one-sided) Note: Number-at-risk rows small and only partly legible. [Slide 4] 16:15 - 17:45 Rapid oral 1: Gastrointestinal tumour, upper digestive CHAIRS: MARIA IGNEZ BRAGHIROLI, FENG WANG, LUCJAN WYRWICZ Safety TRAEs occurring in ≥10% of patients in either arm Ram + FOLFIRI (Grade ≥3 | Grade ≤2) vs Ram + Paclitaxel (Grade ≤2 | Grade ≥3), incidence in % Fatigue: 4 | 34 vs 21 | 3 Nausea: 4 | 31 vs 15 | 1 Diarrhea: 3 | 31 vs 11 | 1 Neutrophil count decreased: 15 | 9 vs 7 | 12 Mucositis: 3 | 19 vs 5 Vomiting: 2 | 15 vs 5 Hypertension: 9 | 4 vs 6 | 7 Platelet count decreased: 2 | 11 vs 2 | 1 Anorexia: 11 vs 4 | 1 Peripheral sensory neuropathy: 1 | 8 vs 20 | 1 Proteinuria: 6 vs 12 | 1 Legend: TRAE Grade ≤ 2; TRAE Grade ≥3 Safety population consists of all randomized patients who received at least one dose of study medication. Magdalena Scheck Magdalena Scheck FOLFIRI plus ramucirumab versus paclitaxel plus ramucirumab as second-line therapy for taxane-pretreated patients with advanced or metastatic gastroesophageal adenocarcinoma-Final results of the phase III IKFs602/AIO RAMIRIS Study ALMERIA AUDITORIUM - HALL 9 [ESMO Congress Madrid 2026 logo]

[Slide 1] DFS and OS in the mITT and PP Population Median follow-up: 52.8 months Disease-free survival — mITT population Modified intention-to-treat: 4-month vs 6-month groups · Center-stratified Cox model · N=1023 HR, 1.01 (95% CI, 0.82–1.25) One-sided P = 0.006 · NI met · margin HR 1.33 Legend: 6 months / 4 months No. at risk (months 0/12/24/36/48/60/72/84/96) 6 months: 512 / 450 / 394 / 358 / 209 / 95 / 40 / 13 / 5 4 months: 511 / 462 / 397 / 352 / 210 / 106 / 44 / 15 / 6 Disease-free survival — PP population Per protocol: 4-month group completing 5–6 cycles vs 6-month group completing 8 cycles · Center-stratified Cox model · N=736 HR, 0.96 (95% CI, 0.75–1.25) One-sided P = 0.007 · NI met · margin HR 1.33 No. at risk 6 months: 309 / 271 / 238 / 219 / 126 / 57 / 24 / 7 / 2 4 months: 427 / 388 / 332 / 296 / 175 / 86 / 33 / 12 / 5 Overall survival — mITT population Modified intention-to-treat: 4-month vs 6-month groups · Center-stratified Cox model · N=1023 HR, 1.06 (95% CI, 0.84–1.34) No. at risk 6 months: 512 / 489 / 439 / 401 / 229 / 107 / 48 / 14 / 5 4 months: 511 / 496 / 450 / 392 / 230 / 113 / 44 / 15 / 6 Overall survival — PP population Per protocol: 4-month group completing 5–6 cycles vs 6-month group completing 8 cycles · Center-stratified Cox model · N=736 HR, 0.95 (95% CI, 0.71–1.27) No. at risk 6 months: 309 / 295 / 264 / 240 / 137 / 66 / 26 / 8 / 2 4 months: 427 / 418 / 380 / 333 / 194 / 92 / 33 / 12 / 5 Dazhi Xu [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: DFS (mITT, primary NI analysis); also DFS PP, OS mITT, OS PP 6 months XELOX (mITT) | 512 4 months XELOX (mITT) | 511 HR: 1.01 HR CI: 0.82-1.25 P: One-sided P = 0.006 (NI met, margin HR 1.33) Note: Other panels: DFS PP N=736 HR 0.96 (0.75-1.25) one-sided P=0.007 NI met; OS mITT HR 1.06 (0.84-1.34); OS PP HR 0.95 (0.71-1.27). PP arms 6 mo n=309, 4 mo n=427. Medians not printed. Median follow-up 52.8 months. [Slide 2] Conclusion • Four-month adjuvant XELOX: noninferior to the standard 6-month regimen for 3-year DFS with fewer high-grade adverse events and higher treatment completion rates • These data support 4-month XELOX as an evidence-based option for selected patients with stage II-IIIB gastric cancer after R0 D2 gastrectomy [ESMO Congress Madrid 2026 logos; presenter video inset] [Slide 3] What does LOMAC tell us? Adjuvant XELOX in GEA: 4 months is non-inferior to 6 months in terms of 3-year DFS and is associated with less neuropathy. DFS: ITT population Disease-free survival — mITT population Modified intention-to-treat: 4-month vs 6-month groups · Center-stratified Cox model · N=1023 Primary endpoint: 3-year DFS rate 6 months: 70.8% 4 months: 69.2% (-1.6%) HR, 1.01 (95% CI, 0.82–1.25) One-sided P = 0.006 · NI met · margin HR 1.33 No. at risk (months 0/12/24/36/48/60/72/84/96) 6 months: 512 / 450 / 394 / 358 / 209 / 95 / 40 / 13 / 5 4 months: 511 / 462 / 397 / 352 / 210 / 106 / 44 / 15 / 6 Toxicities: fewer TRAEs in 4 mos group Endpoint | 4 months | 6 months Grade ≥ 3 AEs | 8.4% | 15.0% Grade ≥ 3 sensory neuropathy | 0.4% | 5.9% Grade ≥ 3 fatigue | 0% | 1.4% Feng Wang 15 [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: DFS (mITT); primary endpoint 3-year DFS rate 6 months XELOX | 512 4 months XELOX | 511 HR: 1.01 HR CI: 0.82-1.25 P: One-sided P = 0.006 (NI met, margin HR 1.33) {'timepoint': '3 years', 'arm': '6 months XELOX', 'value': '70.8%', 'ci': None} {'timepoint': '3 years', 'arm': '4 months XELOX', 'value': '69.2%', 'ci': None} Note: Difference -1.6%. Toxicity table: G>=3 AEs 8.4 vs 15.0%; G>=3 sensory neuropathy 0.4 vs 5.9%; G>=3 fatigue 0 vs 1.4% (4 vs 6 months).

[Slide 1] 16:15 - 17:45 Rapid oral 1: Gastrointestinal tumour, upper digestive CHAIRS: MARIA IGNEZ BRAGHIROLI, FENG WANG, LUCJAN WYRWICZ FORTITUDE-102: Overall Survival by FGFR2 Amplification Status FGFR2 Amplified (Placebo vs Bema) Median OS, months (95% CI): Bema 16.6 (11.8–NE); Placebo 8.0 (5.4–18.8) OS HR (95% CI): 0.67 (0.34–1.32) P = 0.24 Number at Risk (0, 6, 12, 18, 24, 30 months) Placebo 22 14 8 5 0 0 Bema 25 21 14 6 1 0 FGFR2 Non-amplified (Placebo vs Bema) Median OS, months (95% CI): Bema 17.3 (13.7–19.8); Placebo 14.5 (13.0–16.0) OS HR (95% CI): 0.84 (0.61–1.15) P = 0.27 Number at Risk (0, 6, 12, 18, 24, 30 months) Placebo 128 113 69 26 8 0 Bema 121 99 74 33 8 1 Key findings: • In FORTITUDE-102, adding bemarituzumab to mFOLFOX6/CAPOX + nivolumab yielded a numerically greater treatment effect in FGFR2-amplified vs non-amplified patients (OS HR 0.67 [95% CI 0.34–1.32] vs 0.84 [0.61–1.15]; PFS HR 0.77 [95% CI 0.40–1.49] vs 0.87 [0.65–1.17]) • Results from the placebo arm suggest FGFR2 amplification may confer poor prognosis FGFR2 amplification identified by tissue and/or ctDNA NGS. Presented by: Zev Wainberg, MD [Speaker panel: Zev Wainberg — Efficacy of bemarituzumab in patients with advanced gastric and gastroesophageal junction cancer (GC/GEJ) by FGFR2 amplification status: Exploratory biomarker analysis from FORTITUDE-101 (F101) and FORTITUDE-102 (F102)] ALMERIA AUDITORIUM - HALL 9 [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS by FGFR2 amplification status (exploratory) FGFR2 amplified: Bema + mFOLFOX6/CAPOX + nivolumab | 25 | 16.6 mo | 11.8–NE FGFR2 amplified: Placebo + mFOLFOX6/CAPOX + nivolumab | 22 | 8.0 mo | 5.4–18.8 FGFR2 non-amplified: Bema + mFOLFOX6/CAPOX + nivolumab | 121 | 17.3 mo | 13.7–19.8 FGFR2 non-amplified: Placebo + mFOLFOX6/CAPOX + nivolumab | 128 | 14.5 mo | 13.0–16.0 HR: 0.67 (amplified); 0.84 (non-amplified) HR CI: 0.34–1.32 (amplified); 0.61–1.15 (non-amplified) P: 0.24 (amplified); 0.27 (non-amplified) Note: n from number-at-risk at time 0. PFS HRs (text only): amplified 0.77 (0.40–1.49), non-amplified 0.87 (0.65–1.17). Exploratory, not significant.

[Slide 1] 4303 P Cell-free chromatin epigenomic profiling enables non-invasive subtyping of pancreatic cancer Marc Eid*, Karl Semaan*, Damien Vasseur, Gunsagar Gulati, Ji-Heui Seo, Alexander Jordan, Leigh Culnane, Antoine Hollebecque, Antoine Italiano, Toni Choueiri, Kimberly Perez, Jonathan Nowak, Andrew Aguirre#, Brian Wolpin#, Sylvan Baca#, Matthew Freedman#, Harshabad Singh# *co-first authors #co-senior authors Background • Classical (C) and basal [obscured] subtypes are key prognostic biomarkers of pancrea[obscured] (PDAC) and may predict response to treatmen[obscured] • Current subtyping m[obscured] • Here, we report a n[obscured] PDAC subtyping based on profiling circulati[obscured] Methods • PDAC PDX model sub[obscured] bulk RNA-seq and the PurIST algorithm. • ChIP-seq for the pos[obscured] H3K27ac and H3K4me3, MeDIP-se[obscured] PDAC PDX. [remaining methods obscured by presenter] Widespread epigenomic differences between classical and basal-like PDAC A PDX (epi-)genomic datasets: RNA-seq (N=28), ATAC (N=27), H3K27ac (N=28), H3K4me3 (N=27), MeDIP (N=28); Subtype: Basal-like, Classical; Source: DFCI, MDACC B [genome tracks: CLDN18, GATA6, HNF1A, KRT6A, PTGES, GNAI2] C [volcano plots: H3K27ac, H3K4me3, ATAC-seq, MeDIP-seq; up in basal-like / up in classical] Validation of the epigenomic signature in an independent PDAC PDX cohort A [heatmaps H3K27ac, ATAC, H3K4me3; PDAC subtype Basal-like / Classical] B C:B ratio, Classical vs Basal-like: H3K27ac p= 0.0036; H3K4me3 p= 0.0036; MeDIP-seq p= 0.039 PIES enables accurate subtyping of PDAC from patient plasma A 1 mL of plasma; Cohorts: DFCI (N=26); Clinical trials: NCT03520790 (N=29), MOSCATO-02 (N=27) B ctDNA fraction, Classical vs Basal-like, p= 0.32 C AUPRC Classical vs. basal-like: All samples (AUPRC = 0.88); ≥3% ctDNA (AUPRC = 0.91); <3% ctDNA (AUPRC = 0.86) Patients with low PIES (basal-like) have worse outcomes A Progression-free survival multivariable Cox model (Adjusted Hazard Ratio, log scale) PIES: Low vs High 2.99 (1.30-6.90); p=0.01 ctDNA (%) 7.68 (0.73-81.00); p=0.09 Liver met: Yes vs. No 1.54 (0.69-3.42); p=0.29 ECOG: 1-2 vs. 0 1.49 (0.75-2.97); p=0.26 CA 19-9 (U/mL) 1.00 (1.00-1.00); p=0.44 Arm: GnP vs FFX 0.57 (0.26-1.28); p=0.17 B Circulating epigenome – All patients: High PIES vs Low PIES; Log-rank p= 0.046 Number at risk: High PIES 34 24 18 10 5 2; Low PIES 16 10 5 1 1 0 C Circulating epigenome – FFX arm: High PIES vs Low PIES; Log-rank p= 0.00014 Number at risk: High PIES 13 12 8 3 1 0; Low PIES 4 1 0 0 0 0 D Low PIES per circulating epigenome: FFX vs GnP; Log-rank p= 0.0097 Number at risk: FFX 4 1 0 0 0; GnP 12 9 5 1 1 Conclusion • In a multi-institutional cohort, PIES distinguished C and B PDAC subtypes with an AUPRC of 0.88 across all plasma samples, irrespective of ctDNA fraction, increasing to 0.91 in samples with ≥3% ctDNA. • PIES provides a non-invasive approach for determining PDAC transcriptional subtypes, with the potential to facilitate broader implementation of molecular subtyping and inform treatment selection. @marc_eid @KarlSemaan @SylvanBacaLab Presented at ESMO 2026 | Madrid, Spain [Slide 2] 4303 P Cell-free chromatin epigenomic profiling enables non-invasive subtyping of pancreatic cancer Marc Eid*, Karl Semaan*, Damien Vasseur, Gunsagar Gulati, Ji-Heui Seo, Alexander Jordan, Leigh Culnane, Antoine Hollebecque, Antoine Italiano, Toni Choueiri, Kimberly Perez, Jonathan Nowak, Andrew Aguirre#, Brian Wolpin#, Sylvan Baca#, Matthew Freedman#, Harshabad Singh# Background • Classical (C) and basal (B) transcriptional subtypes are key prognostic biomarkers of pancreatic ductal adenocarcinoma (PDAC) and may predict response to treatment1-3. • Current subtyping methods rely on tissue biopsies. • Here, we report a novel non-invasive approach for PDAC subtyping based on profiling circulating tumor chromatin. Methods • PDAC PDX model su[obscured] identified using bulk RNA-seq and the PurIST algorithm. • ChIP-seq for the post-translational modifications H3K27ac and H3K4me3, MeDIP-seq, and ATAC-seq, were performed on PDAC PDX. • Differentially marked regions between C and B subtypes were identified using DESeq2 and were validated in an independent cohort of PDAC PDX models4. • cfChIP-seq for H3K27ac and H3K4me3, cfMeDIP-seq, and low-pass whole-genome sequencing (LPWGS) were performed on 1 mL of plasma from patients with metastatic pancreatic cancer. • ctDNA fraction was estimated from LPWGS data using ichorCNA. • A C/B score was derived for each epigenetic feature in individual patients' plasma samples by calculating the ratio of aggregate signal at classical:basal regions. • For each plasma sample, we combined the different epigenetic features to develop the Pancreatic Integrated Epigenomic Score (PIES). • Performance was evaluated using area under the precision-recall curve (AUPRC). • Progression-free survival (PFS) was compared between patients with Low and High PIES using multivariable Cox regression adjusted for ctDNA fraction, liver metastasis, ECOG, CA19-9, and treatment regimen. References 1. Singh H, et al. Clin Cancer Res. 2024;30:4932–4942. 2. Wenric S, et al. JCO Precis Oncol. 2025;9. 3. Dilly J, et al. Cancer Discov. 2024;14:2135–2161. 4. Lomberk G, et al. Nat Commun. 2018;9:1978. Dana-Farber Cancer Institute | Harvard Medical School | Gustave Roussy | MD Anderson Cancer Center [Right-hand results panels obscured by viewers; same poster as s1]

[Slide 1] 2879RO - Durvalumab plus definitive chemoradiotherapy versus definitive chemoradiotherapy in patients with locally advanced esophagus carcinoma: ARION PRODIGE 67 randomized phase 2 trial Speakers Marie-Anouchka Modesto (Toulouse, France) Lecture Time 17:09 - 17:14 Abstract Background In locally advanced or unresectable esophageal cancer, definitive chemoradiotherapy (CRT) is the standard of care (SoC) combining radiotherapy with platinum-based chemotherapy, resulting in a 2-year disease-free survival of 25%, which remains an unmet medical need. As immune checkpoint inhibitors increase survival in the metastatic setting, a multicentric, this open-label, comparative phase II ARION trial was designed to assess efficacy and safety of PD1/PDL1 blockade by durvalumab (D) added to SoC. Methods Patients (pts) ≥18 years old, WHO performance status ≤1, non-metastatic, locally advanced squamous cell carcinoma (SCC) or unresectable adenocarcinoma (ADK), (stratified by tumor stage, histology, centers) were randomized between arm A with 12 monthly 1500 mg (D) infusions added to SoC (IMRT 50 Gy/25 daily fractions over 5 weeks) and 6 cycles of FOLFOX regimen: oxaliplatin 85 mg/m2, Leucovorin 200 mg/m2, 5-FU 400 mg/m2 bolus then continuous 2400 mg/m2 over 46 h) or arm B (SoC). Primary endpoint was centrally reviewed progression-free survival (cPFS). Overall survival (OS), PFS and safety were secondary endpoints. 74 cPFS events were required to detect a hazard ratio (HR) of 0.55 with 90% power (10% one-sided alpha). Results Between May 2019 and May 2024, 112 pts were randomized (Arm A: SoC + D, n=58 / Arm B: SoC, n=54) in 13 French sites. Analysis was performed after a 24-month follow-up. Median age was 67 years, 68% were male, 77% were SCC, 70% had nodal extension. There was no significant difference between arms for cPFS, PFS and OS (table). Treatment-related adverse events (AE) were higher in arm A. Grade ≥ 3 AE rates were similar in both arms. Table: 2879RO | | Arm A (exp) 12-month rate [95%CI] | Arm B (SoC) 12-month rate [95%CI] | Adjusted HR* [95%CI] [80%CI] | 1-sided p-value | | cPFS | 48.6% [34.6; 61.3] | 48.9% [34.7; 61.7] | 0.91 [0.55; 1.49] [0.66; 1.25] | 0.351 | | PFS | 50.1% [36.2; 62.5] | 48.1% [33.7; 61.2] | 0.92 [0.56; 1.50] [0.67; 1.27] | 0.369 | | OS | 75.1% [61.6; 84.4] | 67.9% [53.6; 78.7] | 0.99 [0.59; 1.68] [0.70; 1.40] | 0.489 | *for Histological subtype, Stage Conclusions In this indication, Durvalumab with CRT compared to SoC did not improve PFS. There was no new safety signal. Longer survival, follow-up and predictive factors for treatment efficacy will be presented.

[Slide 1] 16:15 - 17:45 Rapid oral 1: Gastrointestinal tumour, upper digestive CHAIRS: MARIA IGNEZ BRAGHIROLI, FENG WANG, LUCJAN WYRWICZ DFS and OS in the mITT and PP Population Median follow-up: 52.8 months Disease-free survival - mITT population Modified intention-to-treat: 4-month vs 6-month groups · Center-stratified Cox model · N=1023 HR, 1.01 (95% CI, 0.82-1.25) One-sided P = 0.006 · NI met · margin HR 1.33 No. at risk: 6 months 512 450 394 358 209 95 40 13 5; 4 months 511 462 397 352 210 106 44 15 6 Disease-free survival - PP population Per protocol: 4-month group completing 5-6 cycles vs 6-month group completing 8 cycles · Center-stratified Cox model · N=736 HR, 0.96 (95% CI, 0.75-1.25) One-sided P = 0.007 · NI met · margin HR 1.33 No. at risk: 6 months 309 271 238 219 126 57 24 7 2; 4 months 427 388 332 296 175 86 33 12 5 Overall survival - mITT population HR, 1.06 (95% CI, 0.84-1.34) No. at risk: 6 months 512 489 439 401 229 107 46 14 5; 4 months 511 496 450 392 230 113 44 15 6 Overall survival - PP population HR, 0.95 (95% CI, 0.71-1.27) No. at risk: 6 months 309 295 264 240 137 66 26 8 2; 4 months 427 418 380 333 194 92 33 12 5 Months since randomization (0-96) Dazhi Xu Dazhi Xu Four months of adjuvant XELOX versus six months in stage II-IIIB gastric cancer: The LOMAC randomized trial ALMERIA AUDITORIUM - HALL 9 [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: DFS (primary, non-inferiority) and OS, mITT and PP 6 months XELOX (mITT) | 512 4 months XELOX (mITT) | 511 6 months XELOX (PP) | 309 4 months XELOX (PP) | 427 HR: DFS mITT 1.01; DFS PP 0.96; OS mITT 1.06; OS PP 0.95 HR CI: DFS mITT 0.82-1.25; DFS PP 0.75-1.25; OS mITT 0.84-1.34; OS PP 0.71-1.27 P: DFS mITT one-sided P=0.006 (NI met, margin HR 1.33); DFS PP one-sided P=0.007 Note: Median follow-up 52.8 months. No medians or landmark rates printed; arm Ns from first number-at-risk value.

[Slide 1] TRIDENTE: study design Multicenter, single-arm phase II study RAS wild-type/BRAF V600E-mutant mCRC -> EC-based therapy -> PD -> ≥1 of SoC -> PD -> ENCO + BINI + CET (N = 21) • Encorafenib 300 mg QD • Binimetinib 45 mg BID • Cetuximab initially 400 mg/m2 and subsequently 250 mg/m2 QW • CR or PR by RECIST v1.1 • PD within 4 weeks after the last ENCO administration ≥ 4 months Key eligibility criteria • ≥ 20 years of age • ECOG PS 0-1 • Refractory or intolerant to fluoropyrimidine-based chemotherapy including oxaliplatin or irinotecan • Measurable disease per RECIST v1.1 • Primary endpoint: Investigator-assessed objective response rate (ORR)* • Secondary endpoints: Progression-free survival (PFS), overall survival (OS), disease control rate (DCR), and safety • Exploratory biomarker analysis: circulating tumor DNA at baseline and disease progression * Statistics: Threshold 3%, expected 20%, α = 0.05 (one-sided), power 80% ctDNA analysis using Guardant360® Daisuke Kotani, MD, PhD Clinical trial information: jRCTs031210511 [Slide 2] Objective response rate | N = 21 Complete response | 0 Partial response | 4 (19.0%) Stable disease | 12 (57.1%) Progressive disease | 4 (19.0%) Not evaluable | 1 (4.8%) Objective response rate (90% CI) | 19.0% (6.8-38.4%)* Disease control rate (95% CI) | 76.2% (52.8-91.8%) * Lower bound of the 90% CI (6.8%) exceeded the pre-specified threshold of 3.0%. [Waterfall plot: Best percentage change in target lesion from baseline (%); legend PR, SD, PD, NE] The ORR includes one confirmed and three unconfirmed partial responses. ORR did not differ significantly by prior regimen (EC vs EC + BINI: 7.7% vs 37.5%, p = 0.121) or ENCO-free interval (<6 months vs ≥6 months: 37.5% vs 7.7%, p = 0.121). Daisuke Kotani, MD, PhD [Slide 3] Progression-free survival and overall survival Progression-free survival Median PFS 2.8 months (95% CI, 2.4 – 3.9 months) PFS probability; Months after registration 0 3 6 9 12 Number at risk 21 8 1 1 0 Overall survival Median OS 9.6 months (95% CI, 4.9 – 17.0 months) OS probability; Months after registration 0 3 6 9 12 15 18 21 24 Number at risk 21 19 15 11 7 6 4 2 0 Median follow-up: 9.5 months (range, 1.6 – 23.5 months) Daisuke Kotani, MD, PhD [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS and OS (single arm, ENCO + BINI + CET) ENCO + BINI + CET - PFS | 21 | 2.8 months | 95% CI, 2.4 – 3.9 months ENCO + BINI + CET - OS | 21 | 9.6 months | 95% CI, 4.9 – 17.0 months Note: Single-arm; no HR. Median follow-up 9.5 months (range 1.6-23.5). [Slide 4] ORR by baseline ctDNA status ORR (%) Detectable baseline plasma BRAF V600E and no resistance alterations* | 30.0% (3/10) | 95% CI, 6.7 - 65.2% Undetectable baseline plasma BRAF V600E or any resistance alterations* | 10.0% (1/10) | 95% CI, 0.3 - 44.5% Baseline ctDNA analysis was available in 20 patients (95.2%). There was no statistically significant difference between the groups (p = 0.284). When baseline BRAF V600E detection status was not considered, the ORR was 25.0% in patients without resistance alterations and 12.5% in those with resistance alterations. * Resistance alterations were defined as mutations in KRAS, NRAS, MAP2K1, or EGFR, or amplifications in MET, BRAF, or KRAS. Daisuke Kotani, MD, PhD

[Slide 3] Among Patients With a Confirmed Responsea to BGB-B2033, Tumor Reductions Were Sustained Over Time Across Cohorts DOR (95% CI), % | 450 mg Q3W (n=9)a | All Patients (N=27)a 3-month | 88.9 (43.3-98.4) | 92.4 (73.0-98.1) 6-month | 74.1 (28.9-93.0) | 79.2 (56.9-90.8) Y-axis: Target Lesion Percent Change from Baseline (%); X-axis: Time from First Treatment (Months) Status: Ongoing (solid) | Discontinued (dashed) Data cut-off: 22 September 2026 aInvestigator-assessed confirmed responders. Presenter: Hong Jae Chon, MD, PhD [Slide 4] Conclusion 1 With longer follow up, BGB-B2033, a first-in-class GPC3 x 4-1BB bispecific antibody, continues to demonstrate compelling antitumor activity in patients with heavily pretreated aHCC 2 At the RP2D (450 mg Q3W), cORR was 36% and the 6-month DOR was 74% 3 The favorable tolerability profile is notable as patients with heavily pretreated aHCC often have underlying liver disease that limits therapeutic options 4 Given the compelling antitumor activity and favorable safety profile, global registrational studies are planned to be initiated by the end of the year aHCC, advanced hepatocellular carcinoma; cORR, confirmed objective response rate; DOR, duration of response; GPC3, glypican-3; ORR, objective response rate; RP2D, recommended phase 2 dose; Q3W, every 3 weeks Presenter: Hong Jae Chon, MD, PhD

[Slide 1] OS BY BASELINE ALBI GRADE: STRIDE + TACE vs TACE* ALBI Grade 1 | STRIDE + TACE (n=105) | TACE* (n=111) Total OS events, n (%) | 30 (28.6) | 51 (45.9) Median OS, months (95% CI) | NC (37.8–NC) | 39.2 (28.0–NC) HR (95% CI) | 0.57 (0.36–0.90) OS maturity: 37.5% ALBI Grade 2 | STRIDE + TACE (n=70) | TACE* (n=64) Total OS events, n (%) | 39 (55.7) | 39 (60.9) Median OS, months (95% CI) | 27.9 (17.8–NC) | 20.4 (16.6–30.1) HR (95% CI) | 0.81 (0.52–1.26) OS maturity: 58.2% Proportion of participants alive (%) vs Time from randomisation (months) 0-45 Number of participants at risk (ALBI Grade 1) | 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 | 24 | 27 | 30 | 33 | 36 | 39 | 42 | 45 STRIDE + TACE | 105 | 100 | 98 | 97 | 94 | 87 | 85 | 80 | 77 | 73 | 60 | 37 | 18 | 10 | 4 | 0 TACE | 111 | 109 | 106 | 98 | 92 | 88 | 79 | 76 | 70 | 65 | 53 | 31 | 16 | 10 | 5 | 0 Number of participants at risk (ALBI Grade 2) | 0 | 3 | 6 | 9 | 12 | 15 | 18 | 21 | 24 | 27 | 30 | 33 | 36 | 39 | 42 | 45 STRIDE + TACE | 70 | 65 | 63 | 58 | 53 | 49 | 43 | 39 | 37 | 35 | 31 | 23 | 6 | 1 | 0 | 0 TACE | 64 | 57 | 50 | 50 | 46 | 43 | 35 | 30 | 27 | 27 | 22 | 11 | 4 | 2 | 1 | 0 DCO2: 23 Feb 2026. *Out of the first 175 participants randomised to this treatment arm. Circles represent censored observations. For participants with baseline ALBI grade 1, median (range) follow-up duration for OS was 31.0 (0.7–43.2) months for STRIDE+TACE and 29.4 (2.8–43.9) months for TACE; for participants with baseline ALBI grade 2, median (range) follow-up duration was 27.4 (0.5–39.6) months and 19.7 (0.5–44.0) months, respectively. Masatoshi Kudo, MD, PhD [Survival data captured as a table — no point estimates read from the curve] Endpoint: Overall survival by baseline ALBI grade (STRIDE + TACE vs TACE) ALBI Grade 1 - STRIDE + TACE | 105 | NC | 37.8–NC ALBI Grade 1 - TACE | 111 | 39.2 | 28.0–NC ALBI Grade 2 - STRIDE + TACE | 70 | 27.9 | 17.8–NC ALBI Grade 2 - TACE | 64 | 20.4 | 16.6–30.1 HR: ALBI 1: 0.57; ALBI 2: 0.81 HR CI: ALBI 1: 0.36–0.90; ALBI 2: 0.52–1.26 Note: Two subgroup panels. OS maturity 37.5% (ALBI 1), 58.2% (ALBI 2). DCO2 23 Feb 2026. Medians in months. No p-values printed.

[Slide 1] Study design retrosauna Study design - International retrospective study across 22 NET expert centers to evaluate the impact of continued SSA use alongside 2L therapy - Endpoints: progression-free survival (PFS) and overall survival (OS) after start of respectively PRRT or TT [Schema: Eligible patients -> PRRT-substudy = 2L PRRT -> PRRT with 2L SSA / PRRT w/o 2L SSA; TT-substudy = 2L TT (*) -> TT with 2L SSA / TT w/o 2L SSA; (*) with everolimus or sunitinib] Eligibility criteria Inclusion criteria - Age >=18 years at time of diagnosis - Histologically-proven diagnosis of advanced or metastatic, non-functional, well-differentiated GEP NET grade 1-2 - Radiological progression during 1L SSA - PRRT-substudy: started 2L treatment with PRRT (177Lu-DOTATATE or 177Lu-DOTATOC) with/without 2L SSA between 2010-2024 - TT-substudy: started either 2L treatment with TT (sunitinib or everolimus) with/without 2L SSA between 2010-2014 Exclusion criteria - Presence of high-grade NEN (NET grade 3 or neuroendocrine carcinoma) - Prior treatment with everolimus, sunitinib or PRRT - Presence of functional syndrome - Prior or concomitant palliative, systemic oncological treatment for other malignancy than GEP NET - 2L treatment of GEP NET with chemotherapy Siddharth Chhajlani, MD [ESMO Congress Madrid 2026 logo] [Slide 2] Demographics retrosauna Flowchart: Patients with non-functional GEP NET grade 1-2 with progression on 1L SSA PRRT-substudy: 2L PRRT (n=292) -> with 2L SSA (n=167); w/o 2L SSA (n=125) TT-substudy: 2L TT (*) (n=179) -> with 2L SSA (n=126); w/o 2L SSA (n=53) (*) with everolimus or sunitinib Demographics table: Columns: TT substudy (n=179): TT w/o SSA (n=53) | TT with SSA (n=126) || PRRT substudy (n=292): PRRT w/o SSA (n=125) | PRRT with SSA (n=167) Median age at start 2L (years, IQR): 66 (58-72) | 65 (57-73) || 67 (59-73) | 66 (59-73) Sex (n, %) Female: 23 (43%) | 57 (45%) || 56 (45%) | 80 (48%) Sex (n, %) Male: 30 (57%) | 69 (55%) || 69 (55%) | 87 (52%) NET grade (n, %) NET G1: 14 (26%) | 34 (27%) || 34 (27%) | 58 (35%) NET grade (n, %) NET G2: 39 (74%) | 92 (73%) || 91 (73%) | 109 (65%) Location primary (n, %) pancreas: 42 (79%) | 66 (52%) || 51 (41%) | 45 (27%) Location primary small intestine: 3 (6%) | 36 (29%) || 50 (40%) | 88 (53%) Location primary other GI: 8 (15%) | 24 (19%) || 24 (19%) | 34 (20%) Siddharth Chhajlani, MD [ESMO Congress Madrid 2026 logo] [Slide 3] TT substudy: Progression-free Survival Results retrosauna [KM curve legend: TT with SSA; TT without SSA; y-axis Probability of progression-free survival 0-1.0; x-axis Time (in months) 0-90] log-rank p=0.022 unadjusted HR (0.67, 95% CI 0.47-0.95; p=0.023) Table: TT w/o SSA | TT with SSA n group (events): n=53 (47 events) | n=126 (109 events) mPFS (95% CI): 9.7m (7.1-14.0) | 13.5m (10.4-20.9) log rank p=0.022 - In univariable Cox regression analysis, SSA therapy was associated with improved PFS (HR 0.67, 95% CI 0.47-0.95; p=0.023). - In multivariable Cox regression analysis (*), SSA therapy remained significantly associated with improved PFS (aHR 0.63, 95% CI 0.45-0.90; p=0.010). (*) after adjustment for sex, NET subtype, primary tumor location, and age Number at risk (censored) TT with SSA: 126 (0), 55 (5), 34 (2), 14 (4), 11 (1), 7 (2), 4 (1) TT without SSA: 53 (0), 15 (3), 5 (2), 3 (1), 2 (1), 0 (0), 0 (0) Siddharth Chhajlani, MD [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS (TT substudy) TT with SSA | 126 | 13.5 mo | 10.4-20.9 TT w/o SSA | 53 | 9.7 mo | 7.1-14.0 HR: 0.67 (unadjusted); aHR 0.63 HR CI: 0.47-0.95 (unadjusted); 0.45-0.90 (adjusted) P: log-rank p=0.022; Cox p=0.023 (unadjusted), p=0.010 (adjusted) Note: Retrospective, non-randomized comparison. Adjusted for sex, NET subtype, primary tumor location, age. Number-at-risk at 0,15,30,45,60,75,90 mo. [Slide 4] PRRT substudy: Progression-free Survival Results retrosauna [KM curve legend: PRRT with SSA; PRRT without SSA; y-axis Probability of progression-free survival 0-1.0; x-axis Time (in months) 0-90] log-rank p=0.0013 unadjusted HR 0.63 (95% CI 0.48-0.84; p=0.001) Table: PRRT w/o SSA | PRRT with SSA n group (events): n=125 (94 events) | n=167 (112 events) mPFS (95% CI): 27.0m (23.2-31.5) | 33.3m (28.6-42.8) log rank p=0.0013 - In univariable Cox regression analysis, SSA therapy was associated with improved PFS (HR 0.63, 95% CI 0.48-0.84; p=0.001). - In multivariable Cox regression analysis (*), SSA therapy remained significantly associated with improved PFS (aHR 0.63, 95% CI 0.48-0.83; p=0.001). (*) after adjustment for sex, NET subtype, primary tumor location, and age Number at risk (censored) PRRT with SSA: 167 (0), 128 (8), 73 (19), 35 (19), 18 (3), 10 (2), 6 (1) PRRT without SSA: 125 (0), 86 (8), 38 (15), 14 (6), 4 (2), 2 (0), 1 (0) Siddharth Chhajlani, MD [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS (PRRT substudy) PRRT with SSA | 167 | 33.3 mo | 28.6-42.8 PRRT w/o SSA | 125 | 27.0 mo | 23.2-31.5 HR: 0.63 (unadjusted); aHR 0.63 HR CI: 0.48-0.84 (unadjusted); 0.48-0.83 (adjusted) P: log-rank p=0.0013; Cox p=0.001 (unadjusted and adjusted) Note: Retrospective, non-randomized. Number-at-risk at 0,15,30,45,60,75,90 mo.

[Slide 1] Efficacy Objective Responses in Cohort 2 (33%) and Cohort 3 (40%*), with disease control in ≥89% of patients in each cohort Best overall response by cohort | Cohort 1 N=2 (%) | Cohort 2 N=46 (%) | Cohort 3 N=20 (%) Disease control | 2 (100) | 41 (89.1) | 19 (95.0) ORR | 0 | 15 (32.6) | 8 (40.0) Confirmed PR | 0 | 14 (30.4) | 5 (25.0) Unconfirmed PR | 0 | 1 (2.2) | 3 (15.0)* Stable disease | 2 (100) | 26 (56.5) | 11 (55.0) Progressive disease | 0 | 5 (10.9) | 1 (5.0) ORR GEP-NET SSTR2+ | 0/2 (0.0) | 14/38 (36.8) | 6/17 (35.3) Disease control = CR/PR/SD. *Two additional uPRs reported as of 10 Sep 2026, for a total of 5 uPRs and an ORR of 50% (25% confirmed) for Cohort 3 [Waterfall plot: Percent Change from Baseline, y-axis 50 to -75; legend Dose Level: Cohort 1 / Cohort 2 / Cohort 3; ■ Additional unconfirmed PR as of 10 Sep 2026] All treated patients; Investigator-assessed RECIST v1.1; ORR = confirmed or unconfirmed CR/PR. Waterfall: 67 patients with a post-baseline measurement. CR, complete response; DCO, data cutoff; GEP-NET, gastroenteropancreatic neuroendocrine tumor; ORR, objective response rate; PR, partial response; RECIST v1.1, Response Evaluation Criteria in Solid Tumors version 1.1; SD, stable disease; SSTR2+, somatostatin receptor subtype 2–positive; uPR, unconfirmed partial response for Cohort 3. [ESMO Congress Madrid 2026 logo] [Slide 2] PFS Across ITT population, 72% of patients remained progression-free at 18 months Kaplan-Meier estimate of PFS Legend: Cohort 1 (N=2); Cohort 2 (N=46); Cohort 3 (N=20) Number at risk (months 0 / 5 / 10 / 15 / 20): Cohort 1: 2 / 2 / 2 / 2 / 2 Cohort 2: 46 / 40 / 33 / 23 / 6 Cohort 3: 20 / 16 / 7 / 0 X-axis: Time from Treatment Start (Months); Y-axis: Survival Probability Curves are truncated at 20 months, where few patients remain at risk (Cohort 1, n=2; Cohort 2, n=6). One Cohort 2 PFS event occurred after 20 months. Landmark PFS rate, % (95% CI) | Cohort 1 (N=2) | Cohort 2 (N=46) | Cohort 3 (N=20) | Cohorts 1-3 (N=68) 6 months | 100 | 82.6 (68.2–90.9) | 89.1 (62.7–97.2) | 85.7 (74.9–92.0) 12 months | 100 | 71.2 (55.7–82.2) | 76.3 (38.2–92.7)‡ | 74.2 (61.1–83.5) 18 months | 100 | 68.9 (53.1–80.3) | NE | 71.9 (58.3–81.8) At risk, 6 / 12 / 18 mo | 2 / 2 / 2 | 38 / 30 / 13 | 12 / 1 / 0 | 52 / 33 / 15 PFS events, n (%) | 0 | 15 (32.6) | 3 (15.0) | 18 (26.5) ‡1 patient at risk at 12 months. NE, not estimable (no patient at risk). Investigator-assessed RECIST v1.1; Kaplan-Meier estimates with log-log 95% CI. Cohort 3 is preliminary dose-exploration data and is shown separately in the landmark table. CI, confidence interval; ITT, intention-to-treat; mo, months; NE, not estimable; PFS, progression-free survival; RECIST v1.1, Response Evaluation Criteria in Solid Tumors version 1.1. [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS (investigator-assessed RECIST v1.1) Cohort 1 | 2 Cohort 2 | 46 Cohort 3 | 20 Cohorts 1-3 | 68 {'timepoint': '6 months', 'arm': 'Cohort 1', 'value': '100', 'ci': None} {'timepoint': '6 months', 'arm': 'Cohort 2', 'value': '82.6', 'ci': '68.2–90.9'} {'timepoint': '6 months', 'arm': 'Cohort 3', 'value': '89.1', 'ci': '62.7–97.2'} {'timepoint': '6 months', 'arm': 'Cohorts 1-3', 'value': '85.7', 'ci': '74.9–92.0'} {'timepoint': '12 months', 'arm': 'Cohort 1', 'value': '100', 'ci': None} {'timepoint': '12 months', 'arm': 'Cohort 2', 'value': '71.2', 'ci': '55.7–82.2'} {'timepoint': '12 months', 'arm': 'Cohort 3', 'value': '76.3', 'ci': '38.2–92.7'} {'timepoint': '12 months', 'arm': 'Cohorts 1-3', 'value': '74.2', 'ci': '61.1–83.5'} {'timepoint': '18 months', 'arm': 'Cohort 1', 'value': '100', 'ci': None} {'timepoint': '18 months', 'arm': 'Cohort 2', 'value': '68.9', 'ci': '53.1–80.3'} {'timepoint': '18 months', 'arm': 'Cohort 3', 'value': 'NE', 'ci': None} {'timepoint': '18 months', 'arm': 'Cohorts 1-3', 'value': '71.9', 'ci': '58.3–81.8'} Note: Single-arm cohorts, no HR. Medians not printed (not reached/immature). Cohort 3 12-mo estimate based on 1 patient at risk.

[Slide 1] [ESMO Congress Madrid 2026 logo] Concomitant somatostatin analog use in second-line therapy of gastroenteropancreatic neuroendocrine tumors: Survival results from the retrospective retroSAUNA study Dr Siddharth Chhajlani et al, Study Interpretation • Real-world study showing that SSA continuation is done in current practice. • The main biais is the reason to continue SSA beyond progression: driven per "centers' habits versus driven by prognostic factors? • Prospective randomized studies are ongoing: • STOPNET (NCT06345079) : stop SSA or not under PRRT (PEP = PFS rate at 20 months) • SAUNA (NCT05701241) : stop SSA or not, under PRRT or targeted therapy (PEP = PFS) • What is the quality of life of pursuing injections of SSA vs no ? • What is the cost-effectiveness of pursuing SSA ? • Outside of functioning NET requiring SSA or in selected cases, I will wait the results of prospective studies before pursuing SSA [struck through:] (SSA one day, SSA for the life in metastatic non-functioning NETs !)

[Slide 1] LBA48 – EMERALD-3 PHASE 3 STUDY Study Design and Endpoints [Embedded study schema, small text:] Study population: Confirmed HCC not amenable to curative treatment, but eligible for embolisation; Child-Pugh class A; ECOG PS 0 or 1; No extrahepatic disease; Excludes Vp3 and Vp4; No prior systemic therapy Stratification factors: Geographical region (Japan vs Asia [excluding Japan] vs rest of the world); Prior palliative locoregional therapy (1 >6 months vs 1 ≤6 months vs none); Baseline tumour burden (>Up-to-7 vs ≤Up-to-7) R, N=760 Arm A: STRIDE + L + TACE (n=293) Arm B: STRIDE + TACE (n=175) Arm C: TACE (n=292) *Participants were randomised 1:1:1 to Arms A, B and C until each arm reached 175 participants; following this, randomisation continued 1:1 to Arms A and C until reaching ~275 participants per arm Primary endpoint: PFS for Arm A vs C per RECIST v1.1 by BICR Key secondary endpoints: OS for Arm A vs C; PFS for Arm B vs C per RECIST v1.1 by BICR; OS for Arm B vs C Other secondary endpoints included: PFS2, PROs, safety Post-hoc exploratory assessments: TTNT; Mean ALBI and Child-Pugh scores over the treatment period ➢ Positive phase 3 study ➢ Primary endpoint PFS ➢ Post-hoc exploratory analyses ➢ Liver function trajectory, efficacy, post-progression outcomes and safety by baseline liver function ➢ Liver function is a key factor for HCC management ➢ Liver function may be worsened by repeated TACE ➢ Liver decompensation is a competing risk of death across stages ➢ Post-progression treatment options and outcomes are challenging Pinato DJ. ESMO 2025; Celsa C et al. Hepatology 2024; Cabibbo G et al. Clin Cancer Res 2024 Lorenza Rimassa, MD [Slide 2] LBA48 – EMERALD-3 PHASE 3 STUDY Conclusions > Liver function remained stable over the treatment period with STRIDE ± L + TACE, with consistent patterns for subsequent anti-cancer therapy use, irrespective of baseline liver function > I agree > STRIDE ± L + TACE showed clinically meaningful improvements in post-progression outcomes (TTNT and PFS2), with favorable OS vs TACE, regardless of baseline ALBI grade > I agree, but OS data are not mature > Safety outcomes were broadly consistent with the overall study population, regardless of baseline ALBI grade > I agree, but shorter duration of systemic therapy and lower number of TACE in ALBI grade 2 group > These results support a STRIDE-based regimen in combination with TACE as a potential new treatment option for eeHCC > I agree, but we have to wait for OS for STRIDE+L+TACE; or how can/should we consider the available OS for STRIDE+TACE? Lorenza Rimassa, MD [Slide 3] 1679O – PHASE 1 STUDY OF BGB-B2033 BISPECIFIC AB Study Design and Baseline Characteristics [Embedded study design schematic: Monotherapy Dose Escalation and Dose Optimization, BGB-B2033 Q3W in Patients with aHCC; Key Eligibility Criteria; Endpoints (Primary: Safety, MTD...; Secondary: Objective response rate by investigator per RECIST v1.1, Pharmacokinetics/pharmacodynamics, Immunogenicity); Total Enrollment = 94; dose optimization 300 mg Q3W N=28 [?], 450 mg Q3W N=27 [?]; embedded baseline characteristics table - small text largely illegible] > 94 heavily pretreated patients > 39 dose escalation, 55 dose optimization > Median age 56.5, male 84%, Asian 87.2% > HBV-related HCC 87.2% > Median n. of prior treatment lines 2 (1-8) > ≥3 prior treatment lines 24.5% > Previous systemic therapies > Anti-PD-(L)1 93.6% > TKI 68.1% > Anti-PD-(L)1 and TKI 61.7% > Median follow-up 5.5 (0.3-16.7) months Lorenza Rimassa, MD [Slide 4] 1679O – PHASE 1 STUDY OF BGB-B2033 BISPECIFIC AB Conclusions > BGB-B2033, a first-in-class GPC3 x 4-1BB bispecific antibody, continues to demonstrate compelling antitumor activity in patients with heavily pretreated aHCC > I agree, although DCR, PFS, and OS data are not yet available and a longer follow up is needed > At the RP2D (450 mg Q3W), cORR was 36% and the 6-month DOR was 74% > The favorable tolerability profile is notable as patients with heavily pretreated aHCC… > I agree > Given the compelling antitumor activity and favorable safety profile, global registrational studies are planned to be initiated by the end of the year > Phase 3 study of BGB-B2033 vs investigator's choice of sorafenib or lenvatinib in patients with HCC who progressed on prior systemic treatment > I agree; high unmet medical need: immunotherapy combinations are available as first-line systemic treatment options, but no level 1 evidence therapies are available for previously treated patients Lorenza Rimassa, MD

[Slide 4] GroundBBreaker 301 Study (NCT07836530) Phase 3 study of BGB-B2033 vs investigator's choice of sorafenib or lenvatinib in patients with HCC who progressed on prior systemic treatment Key Eligibility Criteria • Adults (≥18 years) with histologically confirmed HCC not amenable to curative surgical or locoregional therapies • GPC3 expression level for stratification • Previously treated with up to 2 systemic regimens, including PD-(L)1 inhibitors • Measurable disease per RECIST v1.1 • ECOG PS 0 or 1 • Adequate organ function N=492 R 1:1 BGB-B2033 • 450mg Q3W intravenous administration • Until disease progression, unacceptable toxicity, or other discontinuation criteria Investigator's choice • Sorafenib BID or lenvatinib QD • Until disease progression, unacceptable toxicity, or other discontinuation criteria Primary endpoint: OS Secondary endpoints: ORR, PFS, and DOR per BICR and investigator assessment | Safety | PROs BICR, blinded independent central review; BID, twice daily; DOR, duration of response; ECOG PS, Eastern Cooperative Oncology Group performance status; GPC3, glypican-3; HCC, hepatocellular carcinoma; ORR, objective response rate; OS, overall survival; PD-(L)1, programmed cell death protein (ligand) 1; PFS, progression-free survival; PRO, patient-reported outcome; Q3W, every 3 weeks; QD, daily; RECIST, Response Evaluation Criteria in Solid Tumors. Presenter: Hong Jae Chon, MD, PhD

[Slide 1] 906eP Clinical outcomes and prognostic factors with trifluridine/tipiracil plus bevacizumab in RAS/BRAF wild-type metastatic colorectal cancer: real-world evidence from the BeTAS study [2026 ESMO congress logo] Nieves Martínez-Lago1, Iria Parajo Vazquez2, Belen de Frutos Gonzalez3, María Carmen Riesco Martinez4, Sandra López Peraita5, Borja González Gomez6, Carla Antonilli Perezz7, Ana López Alfonso8, Ana Fernández Montes9, Ismael Ghanem10, Martín Pérez Martelo1, Sara Agraso Busto2, Marcos Melian Sosa11, Elena Gallardo Martin12, Ana María Jiménez Gordo13, Luis Cabezón-Gutierrez14, Antía Cousillas Castiñeiras15, Marta Llanos Muñoz16, Maria Pilar Ochoa Rivas17, Margarita Reboredo Lopez2. 1 University Hospital & Health Research Institute (IDIS), Translational Medical Oncology Group (ONCOMET), Santiago de Compostela, Spain; 2 A Coruña University Hospital, A Coruña, Spain; 3 Ramón y Cajal University Hospital, Madrid, Spain; 4 Hospital Universitario 12 de Octubre, Madrid, Spain; 5 Hospital Universitario de Burgos, Burgos, Spain; 6 Hospital Universitario Lucus Augusti, Lugo, Spain; 7 Hospital Universitario de Gran Canaria Dr. Negrin, Las Palmas de Gran Canaria, Spain; 8 Hospital Universitario Infanta Leonor, Madrid, Spain; 9 Complejo Hospitalario Universitario de Ourense, Orense, Spain; 10 La Paz University Hospital, Madrid, Spain; 11 Instituto Valenciano de Oncología (IVO), Valencia, Spain; 12 Hospital Universitario Alvaro Cunqueiro, Vigo, Spain; 13 Hospital Universitario Infanta Sofía, Madrid, Spain; 14 Hospital Universitario De Torrejón, Madrid, Spain; Facultad de Medicina, Universidad Francisco Vitoria, Pozuelo de Alarcón, Madrid, Spain; 15 Complejo Hospitalario Universitario de Pontevedra, Pontevedra, Spain; 16 Hospital Universitario de Canarias, San Cristóbal de La Laguna, Spain; 17 Hospital Central de la Defensa Gómez Ulla, Madrid, Spain. KEY FINDINGS RAS/BRAF wt N = 222 Study population | 3L treatment 50.0% FTD/TPI + BEV | Prior anti-EGFR 98.3% Rechallenge 17.2% | mPFS 4.3 mo 95% CI 3.4–5.2 | mOS 10.7 mo 95% CI 8.8–12.6 | Response ORR 8.5% DCR 51.2% RECIST-evaluable n=201 BACKGROUND ▸ FTD/TPI + BEV is an established later-line treatment option in mCRC following the phase III SUNLIGHT trial. ▸ In RAS/BRAF wild-type mCRC, multiple later-line strategies may be considered, while real-world evidence specifically characterizing outcomes with FTD/TPI + BEV in this molecular subgroup remains limited. OBJECTIVE To evaluate real-world effectiveness and prognostic factors in patients with RAS/BRAF wild-type mCRC treated with FTD/TPI + BEV. METHODS Multicenter real-world BeTAS study evaluating FTD/TPI + BEV in previously treated mCRC. Eligible patients had failure or intolerance to ≥2 prior systemic regimens. Among 593 patients with known RAS/BRAF status, 222 RAS/BRAF wild-type patients, defined according to baseline tumor tissue testing, were included in this analysis. Efficacy endpoints were OS, PFS, ORR and DCR. Prognostic factors were explored using Kaplan–Meier estimates and Cox proportional hazards models. PATIENT PROFILE – RAS/BRAF wt (N=222) 66.5 y Median age | 90.1% ECOG PS 0–1 | 80.6% Time from metastatic diagnosis ≥18 months | 70.7% <3 metastatic sites 69.8% Liver metastases | 50.0% Treated in 3L | 98.3% Anti-EGFR | 17.2% Rechallenge Tabernero classification: BPC 18.0% | GPC 37.8% | PPC 44.2% BPC: best prognostic category - GPC: good prognostic category - PPC: poor prognostic category RESPONSE RATE – RECIST-EVALUABLE (n=201) ORR 8.5% DCR 51.2% CR 0.5% | PR 8.0% | SD 42.8% | PD 48.8% EFFECTIVENESS Median follow-up: 30.6 months Progression-free survival [KM curve] mPFS 4.3 mo 95% CI 3.4–5.2 Overall survival [KM curve] mOS 10.7 mo 95% CI 8.8–12.6 EXPLORATORY MULTIVARIABLE PROGNOSTIC ANALYSIS Independent associations with OS and PFS were explored using multivariable Cox proportional hazards models. Variable | HR (95% CI) OS | P value | HR (95% CI) PFS | P value ECOG PS 0–1 vs. 2 | 0.61 (0.28–1.34) | 0.218 | 0.59 (0.29–1.21) | 0.158 Tabernero classification BPC vs. GPC | 0.64 (0.43–0.96) | 0.032 | 0.68 (0.47–0.99) | 0.050 Tabernero classification BPC vs. GPC | 0.41 (0.22–0.77) | 0.006 | 0.59 (0.35–0.99) | 0.044 Prior anti-VEGF No vs. yes | 0.60 (0.36–1.02) | 0.058 | 0.69 (0.44–1.09) | 0.115 Primary prophylactic G-CSF No vs. yes | 1.44 (0.92–2.25) | 0.115 | 1.45 (0.97–2.18) | 0.070 Grade 3–4 neutropenia No vs. yes | 1.83 (1.16–2.88) | 0.009 | 1.86 (1.24–2.89) | 0.003 HRs are adjusted estimates from multivariable Cox models. CONCLUSIONS 1 Clinically meaningful effectiveness in RAS/BRAF wt mCRC — FTD/TPI + BEV showed clinically meaningful real-world effectiveness in this molecular subgroup. 2 Supported by randomized phase III evidence — The role of FTD/TPI + BEV in later-line mCRC is supported by prospective randomized phase III evidence. 3 A robust evidence-based later-line option — These real-world findings reinforce FTD/TPI + BEV as an evidence-based option for RAS/BRAF wild-type mCRC. In RAS/BRAF wild-type mCRC, FTD/TPI + BEV shows clinically meaningful real-world effectiveness and remains a later-line strategy supported by randomized phase III evidence. CONTACT Nieves Martinez-Lago Nieves.Purificacion.Martinez.Lago@sergas.es X: @DraMartinezLago ACKNOWLEDGEMENTS To the patients and their families and the BeTAS investigators.

[Slide 1] regarding the use of daraxonrasib, when available, in patients with RAS-wt, RAS^G13-mutated or RAS^G61-mutated tumours. The RASolute 302 trial was not powered or stratified to answer this question.^8 Unexplained discrepancies appeared in this small subgroup between a clearly significant OS benefit and a potentially detrimental effect on PFS.^8 Despite being eligible, efficacy and toxicity results in these patients were not analysed in the previous studies.^7,8 Further data and analyses are needed to fully inform on the tolerability and magnitude of benefit across rare subsets as well as tolerability in patients with ECOG PS 2. Currently, the greatest strength of evidence for the use of daraxonrasib is in patients with RAS^G12-mutated tumours with ECOG PS 0-1. An algorithm for the management of metastatic PDAC is shown in Figure 1. [Slide 2] T. Conroy & M. Ducreux Annals of Oncology [Treatment algorithm flowchart] Metastatic PC Previous (neo)adjuvant ChT and metastases <6 months after last ChT dose? No: - ECOG PS 0-1, bilirubin <1.5x ULN and no major comorbidities -> FOLFIRINOX [I, A; MCBS 5]^b / NALIRIFOX [I, A; MCBS 2]^b ; or GN [I, B; MCBS 3]^b - ECOG PS 2 with KPS ≥70 and bilirubin ≤1.5x ULN -> GN [II, A; MCBS 3]^b - ECOG PS 2 with KPS <70 and/or bilirubin >1.5x ULN -> Gemcitabine [I, A] - ECOG PS 3-4 -> Symptom-directed care [IV, A] Yes: - ECOG PS 0-1 and RAS^G12 mutation [ESCAT I-A]^a -> Daraxonrasib [I, A; MCBS 5]^b,c - ECOG PS 2 or no RAS^G12 mutation -> Switch to another ChT regimen Second line after FOLFIRINOX/NALIRIFOX: - ECOG PS 0-1 and RAS^G12 mutation [ESCAT I-A]^a -> Daraxonrasib [I, A; MCBS 5]^b,c - ECOG PS 2 or no RAS^G12 mutation -> If FOLFIRINOX in first line:^d Preferred: GP [I, B]; Alternatives: Gemcitabine [III, C], GN^e [III, C]. If NALIRIFOX in first line:^f Gemcitabine-based regimen [V, C] Second line after GN/Gemcitabine: - ECOG PS 0-1 and RAS^G12 mutation [ESCAT I-A]^a -> Daraxonrasib [I, A; MCBS 5]^b,c - ECOG PS 2 or no RAS^G12 mutation -> Preferred: Nanoliposomal irinotecan-5-FU-LV [I, B; MCBS 3]^b,g; Alternatives: mFOLFOX6 [II, C]^h, OFF [II, C]^h Figure 1. Systemic treatment of metastatic pancreatic cancer. Purple: algorithm title; blue: systemic anticancer therapy or their combination; white: other aspects of management and nontreatment aspects. 5-FU, 5-fluorouracil; ChT, chemotherapy; ECOG PS, Eastern Cooperative Oncology Group performance status; EMA, European Medicines Agency; ESCAT, ESMO Scale for Clinical Actionability of molecular Targets; FDA, Food and Drug Administration; FOLFIRINOX, LV-5-FU-irinotecan-oxaliplatin; GN, gemcitabine-nab-paclitaxel; GP, gemcitabine-paclitaxel; KPS, Karnofsky performance status; LV, leucovorin; MCBS, ESMO-Magnitude of Clinical Benefit Scale; mFOLFOX6, modified LV-5-FU-oxaliplatin; NALIRIFOX, liposomal irinotecan-5-FU-LV-oxaliplatin; OFF, oxaliplatin-5-FU-LV; PC, pancreatic cancer; ULN, upper limit of normal. ^a ESCAT scores apply to alterations from genomic-driven analyses only. These scores have been defined by the ESMO Precision Oncology and Taxonomy Committee.^9 ^b ESMO-MCBS v2.0^10 was used to calculate scores for new therapies/indications approved by the EMA or FDA. The scores have been calculated and validated by the ESMO-MCBS Working Group and reviewed by the authors (https://www.esmo.org/guidelines/esmo-mcbs/esmo-mcbs-evaluation-forms). ^c FDA approved, not EMA approved. ^d After FOLFIRINOX treatment, GP can be offered to patients without a RAS^G12-mut or with ECOG PS 2 and a favourable comorbidity profile [I, B]. Other options are gemcitabine alone [III, C] or GN [III, C; EMA and FDA approved in metastatic PC only (not advanced PC) and not approved as second-line therapy]. ^e EMA and FDA approved in metastatic PC only (not advanced PC). ^f After NALIRIFOX treatment, a gemcitabine-based regimen may be offered to patients without a RAS^G12-mut or with ECOG PS 2 and a favourable comorbidity profile [V, C]. ^g Only in patients with, or who have recovered to, ECOG PS 0-1. ^h If not given previously.

[Slide 1] FOG-001-101 Study Cohorts for Patients with Desmoid Tumors Dose Escalation (N=31)* Dose Level 2 150 mg QW N=1 Dose Level 3 300 mg QW N=12 Dose Level 4 450 mg QW N=16 Dose Level 6 900 mg QW N=2 Dose Optimization (N=45) Dose Level 4 (450 mg) QW x 2 cycles -> Dose Level 4 Q2W | N=15 Dose Level 4 (450 mg) QW x 4 cycles -> Dose Level 2 QW | N=15 Dose Level 2 (150 mg) QW x 4 cycles -> Dose Level 2 Q2W | N=15 Induction Phase | Maintenance Phase * IV weekly in 28-day cycles Dose optimization arms designed to evaluate Phase 3 Induction and Maintenance dose and schedule, and to satisfy Project Optimus requirement • Induction: Achieve rapid tumor reduction • Maintenance: Sustain antitumor benefit and optimize tolerability for chronic administration Mrinal Gounder, MD [Screen: 14:45 - 16:15 Proffered paper: Sarcoma; Chairs: Sebastian Bauer, Joanna Szkandera; Mrinal Gounder - A phase 1/2 trial of zolucatetide, a first-in-class direct β-catenin:TCF inhibitor: Updated safety and efficacy in patients with desmoid tumors; MALAGA AUDITORIUM - HALL 3] [Slide 2] Safety Summary TRAEs in ≥20% of Study Patients TRAEs, n (%) | 150 mg (n=16) All Grade | Grade ≥3 | 300 mg (n=12) All Grade | Grade ≥3 | 450 mg (n=46) All Grade | Grade ≥3 | 900 mg (n=2) All Grade | Grade ≥3 | All patients (N=76) All Grade | Grade ≥3 Any TRAE | 15 (93.8) | 1 (6.3) | 8 (66.7) | 0 | 44 (95.7) | 6 (13.0) | 2 (100) | 2 (100) | 69 (90.8) | 9 (11.8) TRAEs (any grade) in ≥20% of pts Fatigue | 7 (43.8) | 0 | 4 (33.3) | 0 | 27 (58.7) | 1 (2.2) | 2 (100) | 0 | 40 (52.6) | 1 (1.3) Alopecia | 5 (31.3) | 0 | 1 (8.3) | 0 | 21 (45.7) | 0 | 2 (100) | 0 | 29 (38.2) | 0 Nausea | 4 (25.0) | 0 | 3 (25.0) | 0 | 21 (45.7) | 0 | 1 (50.0) | 0 | 29 (38.2) | 0 Epistaxis | 4 (25.0) | 0 | 1 (8.3) | 0 | 16 (34.8) | 0 | 0 | 0 | 21 (27.6) | 0 AST increased | 1 (6.3) | 0 | 0 | 0 | 17 (37.0) | 1 (2.2) | 2 (100) | 1 (50.0) | 20 (26.3) | 2 (2.6) Blood bilirubin increased | 1 (6.3) | 0 | 0 | 0 | 15 (32.6) | 1 (2.2) | 2 (100) | 1 (50.0) | 18 (23.7) | 2 (2.6) Myalgia | 3 (18.8) | 0 | 0 | 0 | 15 (32.6) | 0 | 0 | 0 | 18 (23.7) | 0 Any serious TRAE | 2 (2.6)^1 TRAE leading to discontinuation of treatment | 0 TRAEs leading to death (Grade 5) | 0 ^1Events were one each of grade 1 fever and grade 2 peripheral sensory neuropathy • Most commonly reported TRAEs are low-grade, reversible, and manageable with monitoring and supportive care • No TRAEs resulting in treatment discontinuation or death • No high-grade gastrointestinal or skin toxicities • Not associated with ovarian toxicity • Bilirubin elevations were attributable to OATP transporter inhibition and not to hepatocellular injury or cholestasis Mrinal Gounder, MD Data cutoff: 22-Jun-2026 AST – Aspartate aminotransferase; TRAE – Treatment-related adverse event [Slide 3] Zolucatetide Depth of Response and Duration of Exposure at ≥450 mg Starting Dose Best Percent Change by RECIST 1.1 in Patients (N=46) Receiving ≥ 450 mg Starting 15 PRs | CR SD | PR | CR + Prior GSI Treatment - GSI-Naïve Duration of Exposure in Patients (N=48*) Receiving ≥ 450 mg Starting Dose On Treatment Time on treatment, median (range): 4.7 (0.5-22.3) months* Weeks on Treatment Arrows indicate timing of landmark analyses 46 patients with a starting dose ≥ 450 mg are RECIST-evaluable (≥1 post-baseline scan) • Disease control rate (SD+PR+CR) 100% • No patients have discontinued treatment due to PD • Responses seen irrespective of prior GSI exposure or CTNNB1/APC mutation type * One patient is not efficacy evaluable due to early resection of their desmoid tumor prior to completing a post-baseline scan, without subsequent dosing. A second patient has not had a first, 8-week scan. Mrinal Gounder, MD Data cutoff: 22-Jun-2026 CR – complete response, DL - dose level, PD – progressive disease, PR – partial response, SD – stable disease; RECIST – Response Evaluation Criteria in Solid Tumors [Slide 4] Conclusions – Zolucatetide (FOG-001) in Desmoid Tumors • Zolucatetide is a Helicon™ peptide that directly and selectively inhibits the β-catenin:TCF interaction As of the data cut-off of June 22, 2026: • Encouraging anti-tumor activity that deepens over time – Responses seen in 72% and 62% of patients at 32 and 24 week landmarks, respectively, among patients treated at ≥450 mg starting dose – Disease control rate of 100% – Irrespective of prior exposure to gamma secretase inhibitors or mutations in CTNNB1 or APC – No patients have discontinued treatment due to progressive disease • Favorable safety profile, consisting mainly of low-grade and reversible AEs – No treatment discontinuations due to AEs • A Phase 3 registrational study of zolucatetide in desmoid tumors is planned – 450 mg QW for 4 cycles (induction phase) followed by 300 mg Q2W (maintenance phase) selected as the recommended Phase 3 regimen Enrollment in desmoid tumor-specific cohorts is ongoing (NCT05919264) Mrinal Gounder, MD

[Slide 1] Temab-A + Bev vs FTD-TPI + Bev for Patients With Recurrent Metastatic CRC • Increased expression of c-Met protein is common in CRC and carries a poor prognosis1,2 • Prior results from this Phase 1 studya (median follow-up: 9.8 months) showed that telisotuzumab adizutecan (ABBV-400, Temab-A) + bevacizumab (Bev) resulted in manageable safety and higher antitumor activity vs SOC (trifluridine/tipiracil [FTD-TPI] + Bev) in 3L+ CRC3 ○ The 2.4 mg/kg Temab-A dose exhibited the most favorable benefit-risk profile versus 2.0 mg/kg ○ cORR with Temab-A 2.4 mg/kg vs FTD-TPI + Bev: 30% vs 0% Mechanism of action of Temab-A 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 With a median follow-up of 18.2 months, we report results from patients randomized to receive either Temab-A 2.4 mg/kg + Bev or FTD-TPI + Bev aNCT05029882. 3L+, third-line and beyond; ADC, antibody-drug conjugate; Bev, bevacizumab; cORR, confirmed objective response rate; CRC, colorectal cancer; FTD-TPI, trifluridine/tipiracil; mAbs, monoclonal antibodies; SOC, standard of care; Temab-A, telisotuzumab adizutecan; TKI, tyrosine kinase inhibitor; Top1i, topoisomerase 1 inhibitor. 1. Lee SJ, et al. Clin Colorectal Cancer. 2018;17(3):165–169. 2. Liu Y, et al. World J Gastroenterol. 2015;21(12):3706–3710. 3. Cecchini M, et al. Oral presentation at ESMO 2025 (Abstract 731MO). [Slide 2] Temab-A 2.4 mg/kg + Bev Shows Promising Efficacy Over FTD-TPI + Bev With a Median PFS of 6.8 Months Median follow-up: 18.2 months (range 0.4–22.8) PFS | Temab-A 2.4 mg/kg + Bev (n=30) | FTD-TPI + Bev (n=20) Events, n (%) | 27 (90.0) | 18 (90.0) Median PFS, months (95% CI) | 6.8 (5.4–9.5) | 4.2 (1.8–6.3) HR (95% CI) | 0.48 (0.26–0.89) Legend: Temab-A 2.4 mg/kg + Bev | FTD-TPI + Bev Y-axis: Probability; X-axis: Time (Months) 0–20 Patients at Risk Temab-A 2.4 mg/kg | 30 | 30 | 28 | 26 | 25 | 21 | 19 | 12 | 12 | 11 | 4 | 4 | 4 | 4 | 3 | 2 | 1 | 1 | 1 | 1 | 0 FTD-TPI | 20 | 19 | 15 | 12 | 11 | 9 | 7 | 4 | 4 | 2 | 1 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 Data cutoff date: 02March2026. Bev, bevacizumab; FTD-TPI, trifluridine/tipiracil; HR, hazard ratio; PFS, progression-free survival; Temab-A, telisotuzumab adizutecan. [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS Temab-A 2.4 mg/kg + Bev | 30 | 6.8 months | 5.4–9.5 FTD-TPI + Bev | 20 | 4.2 months | 1.8–6.3 HR: 0.48 HR CI: 0.26–0.89 Note: Phase 1 (NCT05029882) Part 7 randomized dose optimization; median follow-up 18.2 mo (range 0.4–22.8); data cutoff 02 March 2026. No p-value printed. [Slide 3] Median OS for Temab-A 2.4 mg/kg + Bev Was 14.7 Months and 10.2 Months for FTD-TPI + Bev Median follow-up: 18.2 months (range 0.4–22.8) OS | Temab-A 2.4 mg/kg + Bev (n=30) | FTD-TPI + Bev (n=20) Events, n (%) | 22 (73.3) | 17 (85.0) Median OS, months (95% CI) | 14.7 (9.8–17.8) | 10.2 (7.7–14.5) HR (95% CI) | 0.62 (0.33–1.18) Legend: Temab-A 2.4 mg/kg + Bev | FTD-TPI + Bev Y-axis: Probability; X-axis: Time (Months) 0–23 Patients at Risk Temab-A 2.4 mg/kg | 30 | 30 | 29 | 27 | 26 | 25 | 25 | 24 | 24 | 22 | 19 | 19 | 19 | 16 | 15 | 12 | 9 | 9 | 8 | 4 | 1 | 1 | 1 | 0 FTD-TPI | 20 | 19 | 19 | 19 | 19 | 19 | 18 | 16 | 15 | 13 | 10 | 8 | 6 | 6 | 6 | 5 | 3 | 3 | 2 | 1 | 0 | 0 | 0 | 0 Data cutoff date: 02March2026. Bev, bevacizumab; FTD-TPI, trifluridine/tipiracil; HR, hazard ratio; OS, overall survival; Temab-A, telisotuzumab adizutecan. [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS Temab-A 2.4 mg/kg + Bev | 30 | 14.7 months | 9.8–17.8 FTD-TPI + Bev | 20 | 10.2 months | 7.7–14.5 HR: 0.62 HR CI: 0.33–1.18 Note: Phase 1 (NCT05029882) Part 7; median follow-up 18.2 mo; data cutoff 02 March 2026. HR CI crosses 1; no p-value printed.

[Slide 2] Co-Primary endpoints OS and ORR (ITT population) Arm A Ram + FOLFIRI n = 158 | Arm B Ram + Pacli n = 159 Median OS (95% CI): 8.7 [7.6, 10.6] | 7.5 [6.2, 9.3] HR (95.8% CI): 0.95 [0.74, 1.22]; P-value: 0.33 Events n (%): 139 (88%) | 137 (86%) Median follow-up (mo): 8.2 | 6.6 Y-axis: Survival Probability; X-axis: Survival time (months) No. at risk (0,6,12,...,60 mo): Arm A: 158 101 46 30 12 7 5 3 2 1 1 Arm B: 159 91 45 28 16 8 4 3 2 1 0 ORR table: Arm A Ram + FOLFIRI n = 158 | Arm B Ram + Pacli n = 159 | P-value ORR: 40 (25%) | 36 (23%) | 0.29 CR: 4 (3%) | 0 (0%) PR: 36 (23%) | 36 (23%) SD: 61 (39%) | 61 (38%) PD: 33 (21%) | 46 (29%) Missing: 24 (15%) | 16 (10%) OS: HR from unstratified Cox proportional hazards model; p-value from unstratified log-rank test (one-sided). + Censored. The results for OSS and OR for ITT population are consistent with those for PP population. Magdalena Scheck [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: OS (ITT) Arm A Ram + FOLFIRI | 158 | 8.7 mo | 7.6–10.6 Arm B Ram + Pacli | 159 | 7.5 mo | 6.2–9.3 HR: 0.95 HR CI: 0.74–1.22 (95.8% CI) P: 0.33 (one-sided log-rank) Note: Median follow-up 8.2 vs 6.6 mo. ORR 25% vs 23%, p=0.29 (co-primary). Number-at-risk text small. [Slide 3] Progression-free survival ITT Population: Arm A Ram + FOLFIRI n = 158 | Arm B Ram + Pacli n = 159 Median PFS (95% CI): 5.5 [4.0, 6.3] | 3.7 [2.9, 4.2] HR (95.8% CI): 0.71 [0.56, 0.90]; P-value: 0.002 Events n (%): 142 (90%) | 149 (94%) No. at risk (0,6,12,...,48 mo): Arm A: 158 63 20 5 4 1 1 1 0; Arm B: 159 43 7 3 1 0 PP Population: Arm A Ram + FOLFIRI n = 150 | Arm B Ram + Pacli n = 154 Median PFS (95% CI): 5.6 [4.1, 6.4] | 3.7 [2.9, 4.3] HR (95.8% CI): 0.69 [0.55, 0.88]; P-value: 0.001 Events n (%): 138 (92%) | 148 (96%) No. at risk: Arm A: 150 63 20 5 4 1 1 1 0; Arm B: 154 43 7 3 1 0 Y-axis: Probability of Progression-free Survival; X-axis: Time to progression or death (months) Ram plus FOLFIRI resulted in a numerical improvement in PFS vs. SOC HR from unstratified Cox proportional hazards model; p-value from unstratified log-rank test (one-sided). + Censored. SOC standard of care Magdalena Scheck [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS (secondary; ITT) Arm A Ram + FOLFIRI | 158 | 5.5 mo | 4.0–6.3 Arm B Ram + Pacli | 159 | 3.7 mo | 2.9–4.2 HR: 0.71 HR CI: 0.56–0.90 (95.8% CI) P: 0.002 (one-sided log-rank) Note: PP population: 5.6 [4.1-6.4] vs 3.7 [2.9-4.3], HR 0.69 [0.55-0.88], p=0.001, n=150 vs 154. PFS is a secondary endpoint. Number-at-risk text small. [Slide 4] Conclusion ➢ Ram plus FOLFIRI did not achieve the primary endpoints of OS or ORR in taxane-pretreated pts. ➢ OS and ORR were numerically higher in Ram plus FOLFIRI compared to Ram plus paclitaxel. ➢ PFS was clearly improved with Ram plus FOLFIRI. ➢ No new safety signals were observed. Ram plus FOLFIRI showed increased rates of all-grade gastrointestinal toxicity, but reduced rates of peripheral neuropathy. ➢ Ram plus FOLFIRI might offer a potential treatment option for patients who had received one line of prior therapy and progressed upon taxane therapy and/ or with pre-existing neuropathy. Magdalena Scheck [ESMO Congress Madrid 2026 logo]

[Slide 1] [Madrid 2026 ESMO Congress logo] 4212RO Biomarker Phase II Study of Cabozantinib in Advanced Radioiodine-Refractory Differentiated Thyroid Cancer: The GETNE-T2216 / CABOTHYROID Trial Jorge Hernando*, Francheska Cadacio, Guillermo Villacampa, José María Herranz, Teresa Alonso-Gordoa, Javier Molina-Cerrillo, Paula Jiménez-Fonseca, German Iglesias Álvarez, Miguel Navarro Martín, Neus Baste, Alberto Carmona-Bayonas, Lara Iglesias, María Quindós-Varela, Gloria Marquina, Tian Tian, Jaume Capdevila *Department of Medical Oncology, Gastrointestinal and Endocrine Tumor Unit & Neuroendocrine and Endocrine Tumor Translational Research Program (NET-VHIO), Vall d'Hebron Hospital Universitari, Vall d'Hebron Barcelona Hospital Campus, Barcelona, Spain [GETNE — Grupo Español de Tumores Neuroendocrinos y Endocrinos logo] [Podium: ESMO Congress, Madrid Spain 23-27 October 2026] [Slide 2] GETNE-T2216 / CABOTHYROID Trial BIOMARKER ANALYSIS Comparison of methylation patterns between responders and non-responders [Volcano plot: -log10(q-value) vs Methylation Difference (%); legend: Hyper (Resp > Non), Hypo (Resp < Non)] [Pie chart] Promoter (≤1kb) (12.7%); Promoter (2-3kb) (1.4%); 1st Exon (15.5%); Other Exon (1.4%); 1st Intron (21.1%); Other Intron (35.2%); Distal Intergenic (12.7%) Genes identified with higher importance in response to cabozantinib prediction Symbol | Gene name | Chromosome | Annotation ANKRD26P1 | ankyrin repeat domain 26 pseudogene 1 | chr16 | Distal Intergenic CDH20 | cadherin 20 | chr18 | Intron LINC00273 | long intergenic non-protein coding RNA 273 | chr16 | Distal Intergenic FRG1 | FSHD region gene 1 | chr4 | Intron ZNF733P | zinc finger protein 733, pseudogene | chr7 | Distal Intergenic → EM-seq analysis showed differences in methylation patterns between responders vs. non responders. → ML analysis identified ANKRD26P1, LINC00273, and ZNF733P as the genes with higher importance in response prediction. [Slide 3] GETNE-T2216 / CABOTHYROID Trial EFFICACY - FULL ANALYSIS SET [Waterfall plot: Maximum reduction from baseline (%)] Confirmed BoR n=39 CR 0 (0.0%) PR 10 (24.4%) SD 27 (65.9%) PD 2 (4.9%) DCR = 89.7% (95% CI: 70.8-94.4) ORR = 24% Median DoR = 18.4 months (95% CI: 14.0-NR) Median TTR = 2.7 months (95% CI: 2.5-2.8) [PFS curve] Median PFS = 16.1 months (95% CI: 13.8-NR) Number at risk (censored): 41 (0), 31 (1), 22 (5), 9 (11), 1 (16), 0 (17) [0, 6, 12, 18, 24, 30 months] [OS curve] Number at risk (censored): 41 (0), 38 (0), 34 (1), 15 (17), 3 (29), 0 (32) Cabozantinib provides a fast response in in patients with DTC, even after progression to up to two MKIs. [Slide 4] GETNE-T2216 / CABOTHYROID Trial EFFICACY - UNIVARIATE ANALYSIS Symbol | Gene name ANKRD26P1 | ankyrin repeat domain 26 pseudogene 1 LINC00273 | long intergenic non-protein coding RNA 273 ZNF733P | zinc finger protein 733, pseudogene Forest plot of PFS Cox model (Events | Events Ref | HR (95% CI) p-val) Age (y) at inclusion | 24/41 | | 0.98 (0.94 - 1.02, p=0.368) Sex: Male (Ref: female) | 13/23 | 11/18 | 1.02 (0.46 - 2.29, p=0.953) Histology: PTC (Ref: FTC) | 16/25 | 8/16 | 1.38 (0.59 - 3.23, p=0.462) Previous treatments: Lenv+Sor (Ref: Lenv) | 12/17 | 8/13 | 1.07 (0.43 - 2.70, p=0.879) Sor (Ref: Lenv) | 4/11 | 8/13 | 0.29 (0.09 - 1.00, p=0.049) [highlighted] Presence of bone metastasis: Yes (Ref: No) | 12/18 | 12/23 | 1.75 (0.77 - 3.98, p=0.184) ECOG 1-2 (Ref: 0) | 10/20 | 14/21 | 0.67 (0.30 - 1.52, p=0.342) ANKRD26P1 Methylation High (Ref: Low) | 8/15 | 8/15 | 0.73 (0.27 - 1.96, p=0.527) LINC00273 Methylation High (Ref: Low) | 8/15 | 8/15 | 0.87 (0.32 - 2.33, p=0.783) ZNF733P Methylation High (Ref: Low) | 8/15 | 8/15 | 0.78 (0.29 - 2.09, p=0.617) Forest plot of OS Cox model (Events | Events Ref | HR (95% CI) p-val) Age (y) at inclusion | 9/41 | | 0.97 (0.91 - 1.03, p=0.289) Sex: Male (Ref: female) | 3/23 | 6/18 | 0.42 (0.10 - 1.67, p=0.217) Histology: PTC (Ref: FTC) | 4/25 | 5/16 | 0.52 (0.14 - 1.93, p=0.326) Previous treatments: Lenv+Sor (Ref: Lenv) | 4/17 | 5/13 | 0.53 (0.14 - 1.97, p=0.341) Sor (Ref: Lenv) | 0/11 | 5/13 | NE (NE, NE) Presence of bone metastasis: Yes (Ref: No) | 4/18 | 5/23 | 1.06 (0.28 - 3.95, p=0.931) ECOG 1-2 (Ref: 0) | 7/20 | 2/21 | 4.01 (0.83 - 19.3, p=0.083) ANKRD26P1 Methylation High (Ref: Low) | 3/15 | 3/15 | 0.92 (0.18 - 4.58, p=0.918) LINC00273 Methylation High (Ref: Low) | 2/15 | 4/15 | 0.46 (0.08 - 2.51, p=0.368) ZNF733P Methylation High (Ref: Low) | 3/15 | 3/15 | 0.92 (0.18 - 4.58, p=0.918) • Cabozantinib offer a greater disease control benefit in patients who previously received only sorafenib. • Each gene of importance was not prognostic of longer survival individually.

[Slide 1] [Photo of two people standing in front of a poster at the ESMO 2026 poster hall; poster board numbers 4351P (left) and 4350P] [ESMO Congress Madrid 2026 logo] 4350P: Genetic Determinants of Cachexia-Related Phenotypes in Metastatic Pancreatic Cancer: PANTHEIA-SEOM Translational Study [Sección SEOM Evaluación de Resultados y Práctica Clínica logo] [IMDEA nutrition logo] Vilma Pacheco-Barcia1, Andrea Ochaita Aguado2, Andrés J. Muñoz Martín3, Cristina Fernández Díaz2, Axel Mariño-Méndez4, María Ángeles Vicente Conesa5, Paula Cerdà Serdà6, María de Toro Carmena7, Jorge Hernando8, Jorge Adeva9, Silvia Cruz Gil2, Marta Gómez de Cedrón2, Alberto Carmona-Bayonas5, Ricardo Ramos Ruiz2, Paula Jiménez-Fonseca10, Lara Fernández2, Enrique Grande11, Ana Ramírez de Molina2. 1Hospital Universitario La Luz, Madrid, Spain. 2IMDEA Nutrition Institute, Madrid, Spain. 3Hospital General Universitario Gregorio Marañón, Madrid, Spain. 4Hospital Clínic de Barcelona, Barcelona, Spain. 5Hospital Universitario Morales Meseguer, Murcia, Spain. 6Hospital Universitario de la Santa Creu i Sant Pau. 7Hospital Universitario La Princesa, Madrid, Spain. 8Hospital Universitario de Vall d'Hebron, Barcelona, Spain. 9Hospital Universitario 12 de Octubre, Madrid, Spain. 10Hospital U. Central de Asturias, Oviedo, Spain. 11Hospital U. Quirónsalud Pozuelo. BACKGROUND [partially obscured by people] ...ancer cachexia affects most patients with metastatic pancreatic ...ancer and drives functional decline and mortality, yet its inherited determ[illegible] unknown. We explored whether germli[illegible] related to cachexia, ...arcopenia, [illegible] ...diture are associated with weigh[illegible] MET[HODS] [obscured] RESULTS [Figure: A. Weight loss by genotype (GG n=9, GT n=22, TT n=8; genotype, variant near KLF5); B. Association of the T allele across phenotypes - cachexia/weight-loss phenotypes vs body weight/mass; small text illegible] Figure 2. (A) Percentage weight loss from usual weight to the start of chemotherapy by genotype of the variant near KLF5. (B) Standardized effect of each T allele (95% CI) across phenotypes, ordered from cachexia/weight-loss phenotypes (purple) to body [w]eight/mass (blue). Regression adjusted for sex and age; * significant after Bonferroni correction. [Figure 3: Manhattan-style plot of associations across cachexia/weight-loss related phenotypes and body weight/mass; labels KLF5 (near), TRIB2, ACVR1B, MTHFR, BMP4; small text illegible] [...]ciations of the 26 variants (one colour per gene) with 16 phenotypes, ordered from cachexia/weight-loss related [p]urple background) to body weight/mass (blue background). Each dot is the −log10 P of an additive model adjusted [...]; dashed line, Bonferroni threshold (P = 0.05/26); dotted line, P = 0.05. Clinician-reported cachexia (yes/no; 4 [...]ed with Firth logistic regression. Weight loss • Each T allele of the variant near KLF5 was associated with 5.0 percentage points less weight loss (95% CI −7.6 to −2.4; P = 0.0004; Bonferroni P = 0.010), the only association surviving correction for multiple testing. • Median weight loss was 9.8% in GG, 7.9% in GT and 4.1% in TT carriers; cachexia† was present in 8/9 (89%), 16/22 (73%) and 1/8 (13%), respectively (OR per T allele 0.09; 95% CI 0.02–0.47; P = 0.004). Clinician-reported cachexia (4 cases) showed the same direction (OR 0.12; 95% CI 0.01–0.73; P = 0.019; Firth regression). • The effect was robust to clinical adjustment, centre, exclusion of non-European patients, permutation (family-wise P = 0.007) and bootstrap resampling (variant selected in 89% of resamples; effect range −4.7 to −6.3 points). • The association was specific to weight loss: there was no association with usual (pre-illness) body weight or BMI (P = 0.40 and P = 0.74; Figure 2B). • A variant in MTHFR was associated with higher pre-chemotherapy BMI (+3.8 kg/m² per G allele; Bonferroni P = 0.028), largely reflecting higher usual BMI rather than less weight loss. • Across all 416 tests (26 variants × 16 phenotypes), 25 were nominally significant (20.8 expected by chance); only the KLF5-near variant association with weight loss and the MTHFR associations with pre-chemotherapy weight and BMI passed Bonferroni correction (Figure 3). CONCLUSIONS • The T allele of a variant near KLF5, previously linked to lean mass, is associated with less pre-treatment weight loss and less cachexia in metastatic pancreatic cancer, independently of prior body size. • These preliminary findings require replication in a larger, independent cohort. Acknowledgments: Gratitude to the PANTHEIA study researchers, SEOM's Clinical Practice and Outcome Evaluation Section and PANDORA-SEOM. Author COIs: [small text, partially legible: speaker honoraria from Pierre Fabre, Nutricia, Servier, Pfizer, AstraZeneca, Leo Pharma; consultancy or advisory board from Advanced Accelerator Applications, a Novartis company, Pfizer, Nutricia, Servier; travel support from Amgen, Servier, Pfizer, Nutricia, Merck.] Funding: [small text, partially legible: Regional Government of the Community of Madrid, Government of Spain, European Council of Innovation; SEOM Grant 2024. Pfizer grant 2022. Nutricia and Servier research funding without restrictions.] Author contact: [email address, small text] [Badge on presenter: Vilma Pacheco-Barcia, ESMO Member, ESMO Officer, Invited Speaker]

[Slide 1] Temab-A 2.4 mg/kg + Bev Shows Promising Efficacy Over FTD-TPI + Bev With a Median PFS of 6.8 Months Median follow-up: 18.2 months (range 0.4–22.8) PFS | Temab-A 2.4 mg/kg + Bev (n=30) | FTD-TPI + Bev (n=20) Events, n (%) | 27 (90.0) | 18 (90.0) Median PFS, months (95% CI) | 6.8 (5.4–9.5) | 4.2 (1.8–6.3) HR (95% CI) | 0.48 (0.26–0.89) Legend: Temab-A 2.4 mg/kg + Bev; FTD-TPI + Bev Probability ; Time (Months) 0-20 Patients at Risk Temab-A 2.4 mg/kg: 30 | 30 | 28 | 26 | 25 | 21 | 19 | 12 | 12 | 11 | 4 | 4 | 4 | 4 | 3 | 2 | 1 | 1 | 1 | 1 | 0 FTD-TPI: 20 | 19 | 15 | 12 | 11 | 9 | 7 | 4 | 4 | 2 | 1 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 [Survival data captured as a table — no point estimates read from the curve] Endpoint: PFS Temab-A 2.4 mg/kg + Bev | 30 | 6.8 mo | 5.4–9.5 FTD-TPI + Bev | 20 | 4.2 mo | 1.8–6.3 HR: 0.48 HR CI: 0.26–0.89 Note: Median follow-up 18.2 months (range 0.4–22.8). No p-value or landmark rates printed.

[Slide 1] Design of ARION UCGI 33/PRODIGE 67 Randomized phase 2 trial (NCT03777813) ✓ Eligible patients : ✓ LA non-metastatic biopsy-proven squamous cell carcinoma or adenocarcinoma of the esophagus ✓ ≥18 years old, ✓ WHO performance status ≤1, ✓ Treatment Arms : ✓ Standard Arm : SoC = Chemoradiotherapy (CRT) : RT 50 Gy + 6 FOLFOX q2w ✓ Experimental Arm: SoC + PD-L1 blocker Durvalumab 1500 mg q4w up to 12 months ✓ Objectives : ✓ The primary end point : efficacy of durvalumab added to SoC, in terms of centrally reviewed Progression Free Survival (cPFS). ✓ Secondary end points: PFS OS Safety QoL EXPERIMENTAL ARM CRT+ Durvalumab (N=60) Durvalumab 1500 mg q4w during Year 1 RT 50 Gy + FOLFOX q2w Screening Endoscopy Biopsy FDG-PET/CT QoL R 1:1 M6 CT scan QoL | M9 CT scan QoL | Year 1 CT scan QoL | M18 CT scan QoL | Year 2 CT scan QoL STANDARD of CARE ARM (N=60) RT 50 Gy + FOLFOX q2w Stratification factors: • Center • Histological Type: Adenocarcinoma vs Squamous cell carcinoma • Stage: I/II vs III/IV vs unknown Primary endpoint: Centrally Reviewed PFS Secondary endpoints: Safety, OS, QoL Ancillary studies: - Biomarkers CPS score + TIL (CD3 CD8 CD 68) - Gut microbiota - Plasmatic biomarkers of response to IO and / or CRT Assessed for eligibility N= 188 Randomized patients N =112 Arm A: CRT + DURVA N = 58 Received allocated intervention N =57 Did not receive allocated treatment N = 1 Arm B: CRT N = 54 Received allocated intervention N =53 Did not receive allocated treatment N =1 Intent To Treat population N = 58 Safety population N = 57 Intent To Treat population N =54 Safety population N = 53 From May 2019 to May 2024 Modesto A, Tougeron D, Tremolières P, Ronchin P, Jouve AD, Leignel DA, et al. Association of Radiochemotherapy to Immunotherapy in unresectable locally advanced Oesophageal carciNoma-randomized phase 2 trial ARION UCGI 33/PRODIGE 67: the study protocol. BMC Cancer. 2023 Oct 12;23(1):966 Anouchka MODESTO M.D Ph.D. [ESMO Congress Madrid 2026 logo] [Slide 2] Survival Primary endpoint : Centrally reviewed PF Survival One year 50,5% vs 48.9 % adj HR* = 0.86, 95%CI [0.52; 1.41] ; 80%CI [0.62; 1.19] 1-sided p = 0.273 *Adjusted for Stratification factors Y-axis: centralized Progression-free survival; X-axis: Months (0-27) Number at risk (months 0/3/6/9/12/15/18/21/24/27) Arm A: 58 53 42 33 25 20 18 18 9 0 Arm B: 54 46 33 28 24 21 20 18 5 0 Legend: Arm A / Arm B No impact of CPS score One year OS < 65 years 90.5% vs 55.6% (p=0.072) Secondary end point : Overall Survival One year 75.1% vs 67,9 % adj HR* = 0.97, 95%CI [0.57; 1.64] ; 80%CI [0.69; 1.37], 1-sided p = 0.451 *Adjusted for Stratification factors Y-axis: Overall survival; X-axis: Months (0-27) Number at risk Arm A: 58 56 54 50 42 40 35 30 20 0 Arm B: 54 48 43 39 36 35 30 29 13 0 Legend: Arm A / Arm B Anouchka MODESTO M.D Ph.D. [ESMO Congress Madrid 2026 logo] [Survival data captured as a table — no point estimates read from the curve] Endpoint: Centrally reviewed PFS (primary); OS (secondary) Arm A: CRT + durvalumab | 58 Arm B: CRT | 54 HR: 0.86 (adjusted, cPFS) HR CI: 95% CI 0.52-1.41; 80% CI 0.62-1.19 P: 1-sided p = 0.273 {'timepoint': '1 year', 'arm': 'Arm A: CRT + durvalumab', 'value': '50.5% (cPFS)', 'ci': None} {'timepoint': '1 year', 'arm': 'Arm B: CRT', 'value': '48.9% (cPFS)', 'ci': None} {'timepoint': '1 year', 'arm': 'Arm A: CRT + durvalumab', 'value': '75.1% (OS)', 'ci': None} {'timepoint': '1 year', 'arm': 'Arm B: CRT', 'value': '67.9% (OS)', 'ci': None} Note: OS adj HR 0.97, 95% CI 0.57-1.64, 80% CI 0.69-1.37, 1-sided p=0.451. HRs adjusted for stratification factors. Subgroup <65 y: 1-year OS 90.5% vs 55.6% (p=0.072). Medians not printed. [Slide 3] Tolerance and Quality of life ➢ Radiation Therapy completion (50 Gy) : 97% ➢ FOLFOX completion: 70% ➢ Median delivered Durva (Exp arm) : 9 courses ➢ Grade 3-5 TRAE 75,4% vs 75,5% ➢ No safety signal QLQ-C30 : Global health status [line chart, Global health status (mean) by Baseline, M3, M6, M9, M12, M18, M24; Arm A / Arm B] QLQ-OES18 : Dysphagia [line chart, Dysphagia (mean) by Baseline, M3, M6, M9, M12, M18, M24; Arm A / Arm B] Anouchka MODESTO M.D Ph.D. [ESMO Congress Madrid 2026 logo] [Slide 4] Conclusion ✓ Anti PDL-1 Durvalumab + CRT was not associated with improved centrally reviewed PFS ✓ Trend to improve OS if < 65 years : one year OS 90% vs 55,6% (p=0,07) ✓ No safety signal ✓ Pending questions: – Regional lymph nodes at risk of containing microscopic disease received large fields of prophylactic irradiation – Optimal timing of RT and IO : sequential > concurrent? – Patients selection and Biomarkers beyond CPS. [photo of Rodin's The Thinker sculpture] Anouchka MODESTO M.D Ph.D. [ESMO Congress Madrid 2026 logo]

[Slide 1] CRS • Non-specific immune response characterized by excessive and rapid activation of T-cells and other effector cells, leading to hypersecretion of proinflammatory cytokines such as IL-6, IL-10, IL-18, IFN-γ, and TNF-α • Often begins with fever, and can escalate to hypotension, disseminated intravascular coagulation, and acute respiratory distress syndrome with hypoxemia • Rarely severe (Grade 3 & 4: 1 – 7%) • Usually occurs within 48 hours after the first dose • Becomes more attenuated with subsequent treatment → It is recommended to use steroids, antihistamines and paracetamol in the first 2 cycles and discontinue if no CRS is experienced Christophe Le Tourneau, MD, PhD [Slide 2] ICANS • Immune effector Cell-Associated Neurotoxicity Syndrome • Often co-occurs with CRS • Disruption of the blood-brain barrier and transcient central cytokine leak due to cytokine-mediated endothelial dysfunction • Linked to IL-1-mediated neuroinflammation, blood-brain barrier breakdown, and disturbance of neuronal excitability • Early symptoms include tremor, dysgraphia, dysphasia (especially in naming objects), apraxia, and lethargy, with potential progression to seizures, encephalopathy, or coma • Remains rare (1 – 8%) Christophe Le Tourneau, MD, PhD [Slide 3] On-target off-tumor toxicity • Due to antigens expressed on normal tissues • Toxicities depend on the targets: - CEA → GI toxicity - EGFR → skin toxicity - EpCAM → liver toxicity, diarrhea - gp100 → skin toxicity - GPRC5D → skin & nail disorders, loss of sense of taste - BCMA → hypogammaglobulinemia, risk of infection Christophe Le Tourneau, MD, PhD [Slide 4] Infusion-related reactions (IRR) Neurologic and psychiatric: Dizziness, Headache, Weakness, Syncope, Seizure, Anxiety, Sense of impeding doom 40% Respiratory: Dyspnea, Nasal Congestion, Rhinitis, Sneezing, Oropharyngeal or laryngeal edema, Bronchospasm, Tachypnea, Cyanosis, Respiratory arrest 30% Cardiovascular: Tachycardia, Hypotension, Arrhythmia, Chest pain, Ischemia or infarction, Cardiac arrest 90% Cutaneous: Flushing, Erythema, Pruritus, Urticaria, Angioedema, Rash Gastrointestinal: Nausea, Vomiting, Cramping, Diarrhoea Other: Chills, Fever Figure 1. Signs and symptoms of infusion reactions. Christophe Le Tourneau, MD, PhD Barroso et al., ESMO Open 2024;9:1-18

[Slide 2] Study Design A Phase 1/2 study (NCT06741644) is ongoing to evaluate CS2009 as monotherapy or in combination with chemotherapy (XELOX or FOLFOX) in patients with metastatic colorectal cancer (mCRC) across later-line and first-line settings. Key Eligibility Criteria • Patients with mCRC who are previously untreated and ineligible for radical surgery or local therapy, without known MSI-H/dMMR, or those in the later-line setting with progression/intolerance to standard of care • ECOG PS 0-1 • At least one measurable lesion Endpoints: Safety, ORR per RECIST v1.1 Safety was assessed in all treated patients. Efficacy was assessed in patients with at least one post-baseline tumor assessment per RECIST v1.1. Group 1 (≥2L mCRC monotherapy) Pre-treated mCRC → CS2009 monotherapy 30 mg/kg IV Q3W Group 2 (1L mCRC combination) Treatment-naïve mCRC → CS2009 30 mg/kg IV Q3W + XELOX up to 8 cycles, followed by CS2009 + capecitabine maintenance Group 3 (1L mCRC combination) Treatment-naïve mCRC → CS2009 20 mg/kg IV Q2W + FOLFOX up to 12 cycles, followed by CS2009 + leucovorin and 5-FU maintenance Treatment continued until progressive disease, unacceptable toxicity, consent withdrawal, or death Note: Data from Group 1 and Group 2 are presented. Group 3 data remain immature due to short follow-up and are not presented. ECOG PS: Eastern Cooperative Oncology Group performance status; IV: intravenous; L: line; FOLFOX: oxaliplatin + leucovorin + 5-FU; mCRC: metastatic colorectal cancer; MSI-H/dMMR: microsatellite instability-high/deficient mismatch repair; ORR: objective response rate; Q2W: once every 2 weeks; Q3W: once every 3 weeks; RECIST: Response Evaluation Criteria in Solid Tumors; XELOX: capecitabine + oxaliplatin. Dr. Feng Wang [ESMO Congress Madrid 2026 logo] [Slide 3] Safety CS2009 monotherapy and in combination with chemotherapy demonstrated a favorable safety profile in mCRC n (%) | Group 1: Later-line mCRC CS2009 Monotherapy (N=23) Any-grade* | Grade ≥3 | Group 2: First-line mCRC CS2009 + XELOX (N=23) Any-grade* | Grade ≥3 Immune-related adverse event (irAE) Total | 6 (26.1) | 3 (13.0) | 9 (39.1) | 5 (21.7) Adrenal insufficiency | 2 (8.7) | 0 | 0 | 0 Alanine aminotransferase increased | 1 (4.3) | 1 (4.3) | 3 (13.0) | 1 (4.3) Amylase increased | 1 (4.3) | 1 (4.3) | 3 (13.0) | 0 Aspartate aminotransferase increased | 1 (4.3) | 1 (4.3) | 3 (13.0) | 1 (4.3) Hyperthyroidism | 1 (4.3) | 1 (4.3) | 2 (8.7) | 0 Hypothyroidism | 1 (4.3) | 0 | 2 (8.7) | 0 Lipase increased | 0 | 0 | 4 (17.4) | 1 (4.3) Possible anti-VEGF-related TRAE Total | 7 (30.4) | 2 (8.7) | 4 (17.4) | 1 (4.3) Proteinuria | 4 (17.4) | 0 | 2 (8.7) | 1 (4.3) Epistaxis | 2 (8.7) | 0 | 0 | 0 Urinary occult blood positive | 2 (8.7) | 0 | 1 (4.3) | 0 *Any-grade irAEs and possible anti-VEGF-related TRAEs reported in ≥2 patients. irAE: immune-related adverse event; mCRC: metastatic colorectal cancer; TRAE: treatment-related adverse event; VEGF: vascular endothelial growth factor; XELOX: capecitabine + oxaliplatin. Dr. Feng Wang [ESMO Congress Madrid 2026 logo] [Slide 4] Efficacy: Group 1 (Later-line mCRC, monotherapy) Preliminary data demonstrated promising clinical efficacy for CS2009 monotherapy in later-line mCRC ORR 20.0% (3/15), DCR 93.3% (14/15) [Waterfall plot: Best change from baseline, %; legend ■ 30 mg/kg ▲ Prior Anti-angiogenic; bars labelled PD, SD, SD, SD, SD, SD, SD, SD, SD, SD, SD, SD, PR, PR, PR] BRAF Mut.: N UK UK N UK N UK UK N N N UK UK Y UK KRAS Mut.: Y UK Y Y UK N UK UK N N Y UK UK N UK Liver Met.: Y Y Y Y Y N Y N Y Y N Y N N N Location: R Rect. Rect. Rect. Rect. Rect. Rect. Rect. L Rect. L Rect. L R Rect. Prior lines of therapy for patients achieving PR: Patient 1: XELOX → bevacizumab + irinotecan + capecitabine Patient 2: FOLFOX → bevacizumab + FOLFIRI → encorafenib + cetuximab Patient 3: trilaciclib/placebo + bevacizumab + FOLFOXIRI → capecitabine + bevacizumab → trifluridine + bevacizumab [Spider plot: Change from baseline, % vs Months; legend — 30 mg/kg; ★ First New Lesion; ▸ Treatment ongoing] DCR: disease control rate; L: left colon; mCRC: metastatic colorectal cancer; Met.: metastasis; Mut.: mutation; N: no; ORR: objective response rate; PD: progressive disease; PR: partial response; R: right colon; Rect.: rectum; SD: stable disease; UK: unknown; Y: yes. Dr. Feng Wang [ESMO Congress Madrid 2026 logo]

[Slide 1] Age does NOT impact checkpoint inhibitor efficacy Overall survival benefit of immunotherapy by age group Re-expression of Table 1 (OS subgroup analysis) Subgroup | No. of studies | Age <65 years HR (95% CI) | Age ≥65 years HR (95% CI) | P value for subgroup difference All | 26 | 0.77 (0.70-0.85) | 0.77 (0.70-0.85) | 0.99 Primary disease site NSCLC | 10 | 0.71 (0.61-0.84) | 0.78 (0.68-0.90) | 0.39 Melanoma | 6 | 0.76 (0.63-0.92) | 0.72 (0.60-0.87) | 0.68 Other tumors | 10 | 0.83 (0.72-0.97) | 0.78 (0.65-0.94) | 0.61 Type of immunotherapy Anti-PD-1/PD-L1 antibody | 18 | 0.77 (0.68-0.87) | 0.71 (0.64-0.79) | 0.36 Anti-CTLA-4 antibody | 7 | 0.82 (0.71-0.94) | 0.93 (0.80-1.07) | 0.23 Line of treatment 1st line | 11 | 0.74 (0.62-0.89) | 0.81 (0.69-0.95) | 0.49 ≥2nd line | 15 | 0.79 (0.71-0.87) | 0.74 (0.66-0.83) | 0.45 Hazard ratio for overall survival (0.5-1.2); Favors immunotherapy / No benefit; legend: Age <65 years, Age ≥65 years Kim et al, ESMO Open 2022 [Madrid 2026 ESMO congress logo; inset camera feed of speaker at podium: MADRID SPAIN 23-27 OCTOBER 2026] [Slide 2] Tolerability of peri-operative treatment in older patients: CRITICS CRITICS cohort composition ≥70 years (n=172) | <70 years (n=616) Age group: ≥70 years 172 (22%); <70 years 616 (78%); Median age: 59 vs 73 years Sex: Male 121 (70%) [≥70] vs 408 (66%) [<70]; Female 51 (30%) vs 208 (34%) WHO performance status: WHO PS 0 104 (60%) vs 430 (70%); WHO PS 1 54 (31%) vs 155 (25%); Not performed 14 (8%) vs 28 (5%) Patients (%) Based on Table 1 from Slagter et al. Older versus younger adults with gastric cancer receiving perioperative treatment: Results from the CRITICS trial. WHO performance status had 3 missing values overall. Based on Slagter et al, EJC 2020 [Madrid 2026 ESMO logo, cropped] [Slide 3] Tolerability of peri-operative FLOT in older patients RealFLOT cohort composition (n=206) Selected real-world cohort of resectable gastric or gastro-oesophageal junction adenocarcinoma Sex: Male 141 (68%); Female 65 (32%) Age group: <65 years 117 (57%); ≥65 years 89 (43%) ECOG performance status: 0 189 (92%); 1 17 (8%) Patients (%) [0-100] Based on Table 1. Median age 63 years (range 36-77); 43% were aged ≥65 years; all patients had ECOG PS 0-1. Based on Giommoni et al, BMC Cancer 2021 [Madrid 2026 ESMO congress logo] [Slide 4] Age and microbiome The age-related microbiome shift is strongest in healthier older adults MrOS cohort Discovery Cohort (N=599): Walk Speed High n=200 ***; Walk Speed Low n=399; Excellent n=205 *; <Excellent n=394; Low Med n=292 *; High Med n=307; LSC High n=205 *; LSC Low n=394; Composite Healthy n=133 **; Rest of cohort n=466 Validation Cohort (N=308): Walk Speed High n=106; Walk Speed Low n=202; Excellent n=106; <Excellent n=202; Low Med n=141 *; High Med n=167; LSC High n=111; LSC Low n=197; Composite Healthy n=61 *; Rest of cohort n=247 [Forest plots: Spearman Rho (-0.10 to 0.3) and Adj. β-Coef. (-0.01 to 0.01); significance markers ***, **, *, # on adj. β-coef. panel] Osteoporotic Fractures in Men study Based on Wilmanski et al, Nature Metabolism [Madrid 2026 ESMO congress logo]

[Slide 1] STUDY DESIGN EMERALD-3 is a global, randomised, open-label, sponsor-blinded, multicentre, Phase 3 study(1) Study population(1): • Confirmed HCC not amenable to curative treatment, but eligible for embolisation • Child-Pugh class A • ECOG PS 0 or 1 • No extrahepatic disease • Excludes Vp3 and Vp4 • No prior systemic therapy Stratification factors(1): • Geographical region (Japan vs Asia [excluding Japan] vs rest of the world) • Prior palliative locoregional therapy (1 >6 months vs 1 ≤6 months vs none) • Baseline tumour burden (>Up-to-7 vs ≤Up-to-7) R* N=760 Arm A: STRIDE + L + TACE† (n=293) Arm B: STRIDE + TACE† (n=175) Arm C: TACE† (n=292) *Participants were randomised 1:1:1 to Arms A, B and C until each arm reached 175 participants; following this, randomisation continued 1:1 to Arms A and C until reaching ~275 participants per arm Primary endpoint‡: • PFS for Arm A vs C per RECIST v1.1 by BICR Key secondary endpoints‡: • OS for Arm A vs C • PFS for Arm B vs C per RECIST v1.1 by BICR • OS for Arm B vs C Other secondary endpoints included‡: • PFS2, PROs, safety Post-hoc exploratory assessments: • TTNT • Mean ALBI and Child-Pugh scores over the treatment period EMERALD-3: NCT05301842. In this post-hoc exploratory analysis, underlined variables are reported by baseline liver function. †DEB-TACE or conventional TACE. TACE starts ≥7 days after tremelimumab + durvalumab ± lenvatinib for Arms A and B or within 7 days of randomisation for Arm C, followed by additional procedures at the investigator's discretion. Treatment will continue for up to a maximum of 36 months. ‡Under a multiple testing procedure, the primary endpoint and key secondary endpoints are assessed sequentially in hierarchical order. BICR, blinded independent central review; DEB-TACE, drug-eluting bead transarterial chemoembolisation; ECOG PS, Eastern Cooperative Oncology Group performance status; PFS2, time from randomisation to earliest progression following first objective progression after the first subsequent therapy or death; PRO, patient-reported outcome; R, randomised; RECIST, Response Evaluation Criteria in Solid Tumors; TTNT, time from randomisation to first subsequent therapy or death; Vp3, thrombosis in the right or left portal vein; Vp4, thrombosis in the main truck of the portal vein. 1. Kudo M, et al. Lancet Oncol 2026;27:1077–1093. Masatoshi Kudo, MD, PhD [Slide 2] BASELINE CHARACTERISTICS BY ALBI GRADE n (%) unless otherwise indicated | ALBI Grade 1: STRIDE + L + TACE (n=188) | STRIDE + TACE (n=105) | TACE (n=183) | ALBI Grade 2*: STRIDE + L + TACE (n=105) | STRIDE + TACE (n=70) | TACE (n=109) Age, years, median (range) | 66 (25–86) | 66 (33–83) | 64 (34–83) | 67 (33–86) | 63 (32–85) | 67 (43–85) Geographical region: Japan | 15 (8.0) | 7 (6.7) | 10 (5.5) | 7 (6.7) | 4 (5.7) | 12 (11.0) Asia (excluding Japan) | 119 (63.3) | 82 (78.1) | 119 (65.0) | 53 (50.5) | 55 (78.6) | 54 (49.5) Rest of world | 54 (28.7) | 16 (15.2) | 54 (29.5) | 45 (42.9) | 11 (15.7) | 43 (39.4) ECOG PS 0 | 154 (81.9) | 94 (89.5) | 156 (85.2) | 82 (78.1) | 51 (72.9) | 83 (76.1) Aetiology: Hepatitis B virus | 94 (50.0) | 61 (58.1) | 98 (53.6) | 37 (35.2) | 38 (54.3) | 40 (36.7) Hepatitis C virus | 23 (12.2) | 11 (10.5) | 24 (13.1) | 16 (15.2) | 8 (11.4) | 13 (11.9) None / neither | 68 (36.2) | 33 (31.4) | 57 (31.1) | 52 (49.5) | 23 (32.9) | 52 (47.7) Child-Pugh: Class A – score of 5 | 176 (93.6) | 100 (95.2) | 172 (94.0) | 70 (66.7) | 44 (62.9) | 73 (67.0) Class A – score of 6 | 12 (6.4) | 5 (4.8) | 11 (6.0) | 32 (30.5) | 26 (37.1) | 34 (31.2) BCLC stage A | 33 (17.6) | 20 (19.0) | 42 (23.0) | 18 (17.1) | 13 (18.6) | 27 (24.8) B | 123 (65.4) | 69 (65.7) | 112 (61.2) | 64 (61.0) | 45 (64.3) | 59 (54.1) C | 32 (17.0) | 16 (15.2) | 28 (15.3) | 23 (21.9) | 12 (17.1) | 22 (20.2) Tumour burden (Up-to-7 criteria†) ≤7 | 82 (43.6) | 38 (36.2) | 72 (39.3) | 33 (31.4) | 22 (31.4) | 43 (39.4) >7 | 106 (56.4) | 67 (63.8) | 111 (60.7) | 72 (68.6) | 48 (68.6) | 66 (60.6) Portal vein invasion Vp1 or Vp2 | 17 (9.0) | 12 (11.4) | 19 (10.4) | 12 (11.4) | 7 (10.0) | 9 (8.3) Prior palliative embolisation before randomisation‡ | 29 (15.4) | 22 (21.0) | 37 (20.2) | 22 (21.0) | 7 (10.0) | 13 (11.9) DCO1: 2 Sep 2025. Tumour burden and prior palliative embolisation before randomisation are from Interactive Response Technology data. *Includes 1 participant with ALBI grade 3 in the STRIDE + lenvatinib + TACE arm; †Within Up-to-7 criteria: sum of size of largest tumour (in cm) and number of tumours of ≤7; beyond Up-to-7 criteria: sum of size (in cm) and number of tumours >7. ‡Includes participants who received one embolisation procedure at >6 months and those who received one embolisation procedure at ≤6 months. BCLC, Barcelona Clinic Liver Cancer; DCO, data cut-off; Vp1, thrombus beyond second order branches; Vp2, thrombus in second order branches. Masatoshi Kudo, MD, PhD [Slide 3] MEAN ALBI SCORE OVER TREATMENT PERIOD BY BASELINE ALBI GRADE Left panel: ALBI Grade 1 | Right panel: ALBI Grade 2* Legend: STRIDE + L + TACE | STRIDE + TACE | TACE Y-axis: Mean ALBI score (0 to -4); X-axis: Analysis visit number (1, 5, 6, 8 ... 42) Reference bands: ALBI Grade 3: > -1.39 | ALBI Grade 2: >-2.60 to ≤ -1.39 | ALBI Grade 1: ≤ -2.60 Number of participants (ALBI Grade 1): STRIDE + L + TACE 185 158 155 135 141 139 137 127 128 123 117 110 104 98 90 90 84 77 74 73 70 69 65 63 58 55 52 47 44 40 38 33 28 26 26 21 20 19 8 STRIDE + TACE 105 36 86 28 83 84 75 74 68 65 63 59 55 57 55 52 48 50 46 44 44 43 39 37 35 35 35 34 33 34 32 28 29 23 20 17 12 TACE 181 53 109 33 86 88 67 50 34 30 24 19 15 12 11 10 10 9 8 7 7 5 5 5 5 Number of participants (ALBI Grade 2*): STRIDE + L + TACE 102 90 87 78 81 75 71 68 64 59 57 54 51 44 47 44 41 36 36 34 30 31 25 20 19 18 16 14 14 12 10 7 8 7 7 7 6 7 STRIDE + TACE 70 23 62 20 53 52 47 43 38 32 32 32 32 31 29 29 28 23 22 23 21 20 19 18 17 17 16 13 15 16 16 13 15 15 11 12 12 7 TACE 109 34 51 20 38 38 28 19 12 9 8 8 5 5 5 DCO2: 23 Feb 2026. Visits with <5 participants were excluded to avoid overinterpretation of changes in mean values due to small sample sizes. *Includes 1 participant with ALBI grade 3 in the STRIDE + lenvatinib + TACE arm. ALBI scores were derived from albumin and bilirubin separately assessed every 14 days from baseline up to Cycle 3, then every 28 days thereafter until treatment discontinuation or up to Cycle 37 Day 1 (analysis visit 42). For participants with baseline ALBI grade 1, median (range) follow-up duration for ALBI assessment was 11.9 (0–35.2) months for STRIDE+L+TACE, 11.8 (0–34.9) months for STRIDE+TACE, and 2.0 (0–24.1) months for TACE; for participants with baseline ALBI grade 2*, median (range) follow-up duration was 9.3 (0–35.1) months, 5.8 (0–34.3) months, and 0.9 (0–28.4) months, respectively. Duration of follow-up was calculated between final visit and start of treatment. Where the final visit was baseline and occurred prior to start of treatment, duration was set to 0. All visits, including those with fewer than 5 participants, were included in the duration of follow-up calculation. Masatoshi Kudo, MD, PhD [Slide 4] MEAN CHILD-PUGH SCORE OVER TREATMENT PERIOD BY BASELINE SCORE Left panel: Child-Pugh Class A – score of 5 | Right panel: Child-Pugh Class A – score of 6 Legend: STRIDE + L + TACE | STRIDE + TACE | TACE Y-axis: Mean Child-Pugh score; X-axis: Analysis visit number; band label: Child-Pugh Class A Number of participants (score of 5): STRIDE + L + TACE 242 181 171 164 165 157 158 150 143 135 127 122 116 116 105 99 99 93 88 83 80 72 71 66 63 58 54 50 46 37 31 30 28 23 23 21 11 STRIDE + TACE 144 111 107 113 99 94 88 85 82 79 73 74 71 68 65 63 57 56 56 53 50 47 46 44 42 41 40 39 34 35 28 27 24 15 6 TACE 242 158 130 106 83 62 38 31 26 23 17 15 14 13 12 11 10 8 9 6 6 6 5 5 Number of participants (score of 6): STRIDE + L + TACE 43 32 27 30 28 27 22 21 22 20 19 14 14 13 11 8 8 7 7 7 7 5 5 STRIDE + TACE 31 23 18 18 19 18 14 12 9 11 12 13 10 11 12 9 11 12 9 8 9 8 8 9 6 9 8 5 8 8 8 6 6 6 TACE 46 26 19 15 9 7 5 DCO2: 23 Feb 2026. Visits with <5 participants were excluded to avoid overinterpretation of changes in mean values due to small sample sizes. Child-Pugh score for class A-C ranges from 5–15, derived from clinical laboratory assessments collected at baseline then every 28 days until treatment discontinuation or up to Cycle 37 Day 1 (analysis visit 42). For participants with baseline Child-Pugh Class A (score of 5), median (range) follow-up duration for Child-Pugh assessment was 11.9 (0–35.1) months for STRIDE+L+TACE, 11.2 (0–34.9) months for STRIDE+TACE, and 1.8 (0–28.4) months for TACE; for participants [partly obscured] Class A (score of 6), median (range) follow-up duration was 7.8 (0–35.2) months, 5.6 (0–33.2) months and 0.9 (0–17.3) months, respectively. Duration of follow-up was calculated between final visit and start of treatment. Where the final visit was baseline and occurred [obscured] duration was set to 0. All visits, including those with fewer than 5 participants, were included in the duration of follow-up calculation. Masatoshi Kudo, MD, PhD

[Slide 1] What makes arcotatug tavatecan different? A CLDN18.2-targeted "delivery vehicle", rather than an immune-effector antibody Arcotatug tavatecan DAR 4:1 [Antibody-drug conjugate schematic with labels:] Anti-CLDN18.2-antibody; Exatecan payload TOPO I inhibitor; Glycan-based conjugated cleavable linker; Fc-silenced to mitigate side effects Selectively delivers cytotoxin into CLDN18.2+ tumours ➢ CLDN18.2 ADC: internalization releases exatecan, a potent TOP1 inhibitor with bystander killing; myelosuppression is the main toxicity. ➢ Fc-silenced design minimizes ADCC/CDC, distinguishing it from zolbetuximab and potentially reducing on-target/off-tumor gastric mucosal toxicity. ➢ CLDN18.2 expression levels and heterogeneity may be particularly relevant for effective payload delivery. Filippo Pietrantonio, MD Liu J et al, Nat Med 2025; Shitara et al, ESMO 2026 [ESMO Congress Madrid 2026 logo]

[Slide 1] BTC: Delayed Diagnosis an Ongoing Issue1 - 35 patients: iCCA (42.9%), dCCA (17.1%), gallbladder (14.3%), ampullary (14.3%) and pCCA (11.4%) - The majority of patients (60%) had metastases at diagnosis. Only 37.1% underwent surgery and 17.1% received no treatment 1 Treatment options [pie]: Surgery 37.1% | No treatment 17.1% | First-line treatment 45.8% Conclusion - Biliary tumors are often diagnosed at advanced stages due to significant diagnostic delays caused by non-specific symptoms. - These delays limit curative treatment options and worsen prognosis. - Improved early detection strategies and awareness are crucial to enhancing outcomes and survival rates. BTC, biliary tract cancer; dCCA, distal cholangiocarcinoma; iCCA, intrahepatic cholangiocarcinoma; pCCA, perihilar cholangiocarcinoma. 1. Martirena A et al. Presented at CCF 2025. [Slide 2] BTC: Precision Medicine is Real1 Advanced BTC [pie]: GBC (35%) | eCCA (30%) | iCCA (25%) | AMP (5%) Distribution of targetable alterations in advanced BTC [stacked bars for iCCA, eCCA, GBC] Categories: FGFR2 fusions; FGFR2 mutations; IDH1 mutation; IDH2 mutation; HER2 overexpression; HER2 mutation; BRAF V600E; MSI-H/dMMR; TMB-high; RET fusion; NTRK fusion; NRG1 fusion; RAS mutations; MTAP deletion; MDM2 amplification; MET amplification; Others Approved therapies [legend] Overview of the main potentially targetable alterations identified in advanced BTC (aBTC). It is expected that half of patients diagnosed with aBTC will have a diagnosis of eCCA, followed by iCCA (one quarter) and GBC (one quarter). Within the targetable alterations, FGFR2 fusions and IDH1 mutations are most frequent in iCCA, while HER-2 amplifications are mainly seen in eCCA and GBC. Other alterations seem to have a similar distribution within BTC subgroups. Green bars highlight the targets with approved therapies; many others are currently in development. aBTC, advanced biliary tract cancer; AMP, ampullary tumor; BRAF, B-Raf proto-oncogene; BTC, biliary tract cancer; dMMR, deficient mismatch repair; eCCA, extrahepatic cholangiocarcinoma; FGFR2, fibroblast growth factor receptor 2; GBC, gallbladder cancer; HER2, human epidermal growth factor receptor 2; iCCA, intrahepatic cholangiocarcinoma; IDH1/2, isocitrate dehydrogenase 1/2; MDM2, mouse double minute 2 homolog; MET, mesenchymal-epithelial transition factor; MSI-H, microsatellite instability-high; MTAP, methylthioadenosine phosphorylase; NRG1, neuregulin 1; NTRK, neurotrophic tyrosine receptor kinase; RAS, rat sarcoma virus; RET, rearranged during transfection; TMB, tumor mutational burden. 1. Lamarca A et al. J Hepatol. 2026; 85:794-807.

[Slide 1] Neoadjuvant immunotherapy in dMMR/MSI-H GEA Around 60% cCR rate of ICI (3-6 months) in peri-operative dMMR/MSI-H gastric cancer Study | Regimen/Intent | Duration | Result NEONIPIGA | Nivo*6 + Ipi*2 → surgery → nivo*9 cycles | Pre-O 3months Post-O 9 months | 29 operated pCR 59% MPR 79% INFINITY | Durva*3 + Treme*1 → surgery or W &W | Pre-O 3 months | pCR 60% in 15 resected pts cCR 76% in 17 W&W pts Pan-cancer (GEA) | Dostarlimab*9 | Pre-O 6 months | pCR 52.9% in 17 pts IMHOTEP (GEA) | Pembro* 1–2 → surgery → Pembro*9 | Pre-O 1.5-3 months Post-O 9 months | pCR 19.4% in 29 pts Can response-guided immunotherapy modification enable organ preservation in GEA? 1. André T et al., J Clin Oncol. 2023 2. Raimondi A et al., Ann Oncol. 2025 3. Fouchardiere C et al., Ann Oncol. 2024 4. Cercek A et al., N Engl J Med. 2025 Feng Wang 5 [ESMO Congress Madrid 2026 logo] [Slide 2] How to define cCR in GEA? Clinical complete response (cCR) at 1 year, defined as at 1 year after the start of treatment with dostarlimab, are: - alive - without tumor resection - free of disease progression with a CT of the thorax/abdomen/pelvis normal - with complete or near complete endoscopy (Grade 3 or 4) - with negative biopsies cCR under endoscopy [endoscopy image] Residual disease under EUS [EUS image with arrow] 1 year evaluation was not enough. Submucosal residual disease may be missed under endoscopy. Feng Wang 7 [ESMO Congress Madrid 2026 logo] [Livestream side panel: video of speaker at podium] Feng Wang — Invited Discussant 2877RO, 2878RO and LBA81 [Slide 3] Operation related complications vs serious irAE High incidences of surgical complications [Pie chart] Non-Fistulas (n = 5) | Fistulas (n = 4) 1 postoperative death due to a fistula complication The operation-related complication rate was higher than the 2-6% reported for D2 surgery in East Asia. dMMR/MSI-H GEA cohort receiving radical resection without ICI in SYSUCC [KM curve] 5-year RFS rate: 64% [KM curve] 5-year OS rate: 77% Grade 3-4 dostarlimab-related AEs: 21% Treatment interruption for SAE (n=5) What is the incidence of long term irAE? Feng Wang 8 Unpublished Data [ESMO Congress Madrid 2026 logo] [Slide 4] Mono PD-1 Blockade in GEA: is it sufficient, and for how long? Short-course dual ICI significantly increased pCR rates in dMMR colon cancer Inclusion criteria • Previously untreated colon cancer • MSI-H or dMMR • Radiologically assessed stage IIB-III (cT4 or N+) • ECOG PS 0 or 1 R 1:1 → Sintilimab 200 mg Q3W*2 + IBI310 1 mg/kg Q6W*1 | Sintilimab 200 mg Q3W*2 → Surgery Primary endpoint: pCR rate Secondary endpoint: EFS, OS, RVT, R0 resection rate [Waterfall: Pathologic tumor regression (%); legend anti-PD1+anti-CTLA4, anti-PD1] Inter-group difference: 28.3% (95% CI: 10.8 - 45.7) p = 0.0015 pCR rate: anti-PD1 46.7% | anti-PD1+ anti-CTLA4 78.4% A longer course of neoadjuvant ICI was associated with higher CR rates in locally advanced dMMR CRC [Bubble plot: Complete Response (%) vs Median Duration of Treatment (mo); labelled studies include Cercek et al., 2024 (anti-PD-1); Kothari et al., 2022 (anti-PD-1); Pei et al., 2023 (anti-PD-1); Xie et al., 2023 (anti-PD-1); Ludford et al., 2023 (anti-PD-1); J. Li et al., 2024 (PICC); Zhang et al., 2022 (anti-PD-1); Chen et al., 2023 (anti-PD-1); de la Fouchardiere et al., 2024 (IMHOTEP) (anti-PD-1); Y. Li et al., 2024 (anti-PD-L1); Shiu et al., 2024 (NEOPRISM-CRC) (anti-PD-1); Xu et al., 2024 (NEOSHOT) (anti-PD-1); Avallone et al., 2020 (NICOLE) (anti-PD-1); Type of Cancer: Colorectal, Colon, Rectal; No. of Patients 5/10/25/50/75] Probit treatment-duration coefficient, 0.36 (95% CI, 0.26–0.47) P<0.001 for treatment-duration coefficient P=0.08 for testing for goodness of fit 1. Wang F et al., Cancer Cell 2025 2. Rousseau B et al., NEJM 2025 Feng Wang 9 [ESMO Congress Madrid 2026 logo]

[Slide 1] Adjuvant therapy in resected BTC • Capecitabine • S-1 Panel C: RFS (proportion) vs Time (months) 0-144; Treatment: Observation / Capecitabine; HR 0.81 (0.65 to 1.01) No. at risk: Observation At risk | 224 | 124 | 93 | 76 | 71 | 66 | 55 | 49 | 35 | 24 | 16 | 8 | 2 Censored | 0 | 1 | 2 | 5 | 5 | 5 | 14 | 20 | 33 | 43 | 49 | 57 | 63 Events | 0 | 99 | 129 | 143 | 148 | 153 | 155 | 155 | 156 | 157 | 159 | 159 | 159 Capecitabine At risk | 223 | 149 | 109 | 88 | 84 | 71 | 59 | 51 | 39 | 25 | 15 | 7 | 4 Censored | 0 | 6 | 6 | 9 | 9 | 9 | 19 | 27 | 38 | 51 | 61 | 69 | 72 Events | 0 | 68 | 108 | 126 | 130 | 143 | 145 | 145 | 146 | 147 | 147 | 147 | 147 Panel D: RFS (proportion) vs Time (months) 0-144; Treatment: Observation / Capecitabine; HR 0.77 (0.61 to 0.97) No. at risk: Observation At risk | 220 | 121 | 90 | 74 | 69 | 64 | 53 | 47 | 33 | 23 | 16 | 8 | 2 Censored | 0 | 1 | 2 | 5 | 5 | 5 | 14 | 20 | 33 | 42 | 48 | 56 | 62 Events | 0 | 98 | 128 | 141 | 146 | 151 | 153 | 153 | 154 | 155 | 156 | 156 | 156 Capecitabine At risk | 210 | 146 | 108 | 88 | 84 | 71 | 59 | 51 | 39 | 25 | 15 | 7 | 4 Censored | 0 | 2 | 2 | 5 | 5 | 5 | 15 | 23 | 34 | 47 | 57 | 65 | 68 Events | 0 | 62 | 100 | 117 | 121 | 134 | 136 | 136 | 137 | 138 | 138 | 138 | 138 5-year RFS benefit 3% ITT OS HR 0.79 (0.63 - 1.00) Multiple studies with negative results - GEMOX (PRODIGE 12) - BCAT Meta-analysis of 4 phase III RCTs (n=1,308): benefit with 5-FU–based therapy (OS HR 0.78), no benefit with gemcitabine-based therapy (OS HR 1.03) • Bridgewater J et al. Long-Term Outcomes and Exploratory Analyses of the Randomized Phase III BILCAP Study. J Clin Oncol. 2022 Jun 20;40(18):2048-2057 • Akkus E et al. Adjuvant chemotherapy compared to observation in resected biliary tract cancers: Survival meta-analysis of phase-III randomized controlled trials. Eur J Cancer. 2025 May 2:220:115342 Sebastián Mondaca, MD, MBA [Survival data captured as a table — no point estimates read from the curve] Endpoint: RFS (BILCAP long-term; panels C and D, populations not labeled on slide) Observation (panel C) | 224 Capecitabine (panel C) | 223 Observation (panel D) | 220 Capecitabine (panel D) | 210 HR: C: 0.81; D: 0.77 HR CI: C: 0.65 to 1.01; D: 0.61 to 0.97 {'timepoint': '5 years', 'arm': 'Capecitabine vs observation (ITT)', 'value': 'RFS benefit 3%', 'ci': None} Note: Reproduced figure from Bridgewater et al. JCO 2022. Panel C/D population labels not shown on slide. Events = final cumulative events row. OS HR 0.79 (0.63-1.00) printed. Risk-table digits are small; read with moderate confidence. [Slide 2] Study design Key inclusion criteria • Age 18 years or older • Histologically confirmed CCA or muscle-invasive gallbladder cancer • Macroscopically complete resection with curative intent (R0 and R1) • ECOG 0 or 1 • Adequate renal, haematological, and liver function • Start of adjuvant treatment within 6 to 16 weeks after definitive surgery R 1:1 Group A CisGem (ABC 02): 8 cycles cisplatin 25 mg/m2 and gemcitabine 1000 mg/m2 on day 1 and 8 of a 21 day cycle CT or MRI q 3 months for 2 years, imaging q 6 months for 3 additional years Group B Capecitabine (BILCAP): 8 cycles capecitabine 1.250 mg/m2 BID on days 1 through 14 of a 21 day cycle Primary endpoint • Disease-free survival (DFS), i.e. recurrence or death due to any cause Secondary endpoints • Overall survival (OS) • Cumulative Recurrence • Safety (CTCAE version 4.03) • Quality of life (QoL) monitored with EORTC QLQ-C30 and BIL 21 We expect and improvement in DFS from 50.9% with capecitabine to 60.2% with CisGem, resulting in a hazard ratio (HR) of 0.75. Our statistical analysis plan (SAP) pre-defined the difference in restricted mean survival time (RMST) over 3 years to quantify the treatment effect in case of violation of the proportional hazard assumption. Sebastián Mondaca, MD, MBA [Slide 3] Primary outcome: DFS Group A CisGem | Group B Capecitabine RMST (3 years) Δ 2.6 months, p=0.011 Disease-free survival probability ; Time since randomization (years) 0-5 Group A | 297 | 191 | 140 | 110 | 75 | 35 Group B | 294 | 169 | 122 | 101 | 72 | 30 RMST has several advantages: - No PH assumption - Clinically intuitive - Absolute measure ...but it is blind to what happens after the truncation time Tail of the curve Truncation time was prespecified Sebastián Mondaca, MD, MBA [Survival data captured as a table — no point estimates read from the curve] Endpoint: DFS Group A CisGem | 297 Group B Capecitabine | 294 P: 0.011 (RMST difference at 3 years) Note: RMST (3 years) Δ 2.6 months, p=0.011. No medians or HR printed on this slide (DFS HR 0.82 (0.67-1.01) appears on companion slide s4). n = number at risk at time 0. [Slide 4] The non-inferiority mindset At least, Gem-Cis does not seem to be worse than capecitabine (DFS HR 0.82; 95% CI 0.67 - 1.01) 1- An alternative should be preferred if there is a benefit in QoL or in another area 2-When the benefit of the SOC is limited, you have to be careful when adopting an alternative without proven benefit [Chart] QLQ-C30 global health status (+/- SD) vs Time since randomization (months) 3 6 9 12 15 18 21 24 Group A CisGem | Group B Capecitabine Sebastián Mondaca, MD, MBA

[Slide 1] Lesion size and total burden both associate with response T-VEC Response probability by lesion size, n = 27 [Curve: Probability of response, % vs Maximum lesion diameter, mm (0-60)] 50% at 1.5 cm, under 10% above 3 cm RP1 + nivolumab ORR by total target-lesion size, n = 140 20 mm or less | 63% 21 to 50 | 39% 51 to 100 | 35% 101 to 150 | 15% more than 150 | 12% 63% at 20 mm or less, 12% above 150 mm Masoud SJ et al. Ann Surg Oncol 2019;26:4633. Yamada MM et al. JITC 2025;13:e013564. FDA CTGTAC briefing document, 30 July 2026. [Slide 2] Surgical principles of TIL procurement Choose the site for the product and the patient Lymph node and subcutaneous metastases first; manufacture failure 2.8% from nodes, 19.4% from subcutaneous tissue Minimize morbidity, preserve fitness for lymphodepletion Complications in 17% of 93 procurements, almost all Clavien-Dindo I to II; recovery about 21 days The hybrid patient Can harvest and palliation be one operation? Site selection Tumor Size | 1.5 cm or greater Tumor Location | Risk of contamination: avoid intraluminal sites Extent of Surgery | Complete resection often easier but not necessary; Low threshold for drain to avoid infected seroma during subsequent therapy Pathologic features | Reduce adiposity and necrotic tissue; Tumor periphery may be enriched for TIL Prior Therapy | Assessment of radiographic response to prior therapies, areas of progression; Caution in sites in prior radiation fields or previous infections. In general, harvest from prior radiated area only when there is progression after radiation Tas L et al. Br J Surg 2025;112:znaf090. Kirane and Ariyan, J Cell Ther 2024. Table from prior lecture. [Slide 3] Case 2. Procurement and palliation in one operation Progression on anti-PD-1 → Referred for TIL: post-RT axillary tumor, deemed unresectable → Open wound into tumor: Superinfection and sepsis → One operation: TIL procured from a separate site; axillary resection with latissimus advancement flap → Lymphodepletion and infusion → Complete response to TIL Axillary tumor and wound: Bulky axillary tumor; Open wound into tumor; Recurrent infection and sepsis Resection and flap: Palliative resection + latissimus flap; Posterior donor site; TIL procurement from a separate site After TIL: Systemic response — Before TIL / After TIL; H&N nodes; Pancreatic mass; Peritoneal nodules; PET schematic; Recovered for lymphodepletion and TIL; Resolution of previously avid disease Schematic depiction of the clinical course. PET findings illustrated from the case description, not actual patient scans.

[Slide 1] 13:30 - 15:00 Transforming oncology with generative AI and autonomous tools CHAIRS: EVANGELIA CHRISTODOULOU, JAKOB KATHER Context is the bridge across the AI chasm [image of two bridge spans labeled 'Publication & Pilot' and 'Implementation & Impact'] "...the so-called 'AI chasm', that is the gulf between developing a scientifically sound algorithm and its use in any meaningful real-world applications." P. Keane & E. Topol, npj Digital Medicine 2018 The chasm occurs where the development context and the deployment context diverge: 1 The question the model answers vs. the decision in front of us 2 The quality of training data vs. the data at the site of implementation 3 The people (patients, clinicians, staff), process and technology behind each Caroline Chung, MD, MSc [ESMO Congress Madrid 2026 logo] [Speaker panel] Caroline Chung — Validating AI for oncology: Navigating the road to clinical implementation CADIZ AUDITORIUM - NCC [Slide 2] 13:30 - 15:00 Transforming oncology with generative AI and autonomous tools CHAIRS: EVANGELIA CHRISTODOULOU, JAKOB KATHER AI reaches patients as a system Model validation is only evaluating one part of the system System-level validation: the whole path, from patient data to outcome Model-level validation Patient data — Imaging, pathology, molecular, EHR → AI system — Models, retrieval, tools → Interface & workflow — What, when, to whom → Clinical decision quality: Clinician decides → Accept / Modify or override / Abstain or escalate → Action → patient outcome Monitor performance and context over time [feedback loop to Patient data] Each component of the system will influence the overall ability to achieve the set goal (i.e. purpose). Caroline Chung, MD, MSc [ESMO Congress Madrid 2026 logo] [Speaker panel] Caroline Chung — Validating AI for oncology: Navigating the road to clinical implementation CADIZ AUDITORIUM - NCC [Slide 3] 13:30 - 15:00 Transforming oncology with generative AI and autonomous tools CHAIRS: EVANGELIA CHRISTODOULOU, JAKOB KATHER AI headlines often break down in the real world CLAIM | COUNTER-EVIDENCE | SOURCE AI takes a better history | Outside text-chat OSCEs, accuracy drops when vignettes become multi-turn patient conversations | Johri et al., Nat Med 2025 AI out-diagnoses physicians | Real ED notes: physicians 0.83 vs GPT-4-turbo ≤0.58 on admission; only 5% of 519 LLM studies used real data | Williams, Nat Commun 2024 · Bedi, JAMA 2025 Removing the human is free | Unassisted adenoma detection fell 28.4% → 22.4% after AI; wrong AI cut expert mammography accuracy to 45.5% | Budzyń, Lancet Gastro Hep 2025 · Dratsch, Radiology 2023 These are failures to define the purpose, not failures of algorithm design. Caroline Chung, MD, MSc [ESMO Congress Madrid 2026 logo] [Speaker panel] Caroline Chung — Validating AI for oncology: Navigating the road to clinical implementation CADIZ AUDITORIUM - NCC [Slide 4] 13:30 - 15:00 Transforming oncology with generative AI and autonomous tools CHAIRS: EVANGELIA CHRISTODOULOU, JAKOB KATHER Three levels of validation Purpose sets the metric at every level · context is measured throughout Real-world implementation — Does it support the right decision here, for these patients? — Metric: clinical decision quality in the workflow Pressure testing — Raises confidence it will be robust across real-world data quality — Metric: breakdown threshold and how it fails Controlled environment — The foundation: can it do the intended task under known conditions? — Metric: discrimination and calibration of intended performance [arrow] Increasing complexity / Increasing robustness / Baseline level of model performance Caroline Chung, MD, MSc [ESMO Congress Madrid 2026 logo] [Speaker panel] Caroline Chung — Validating AI for oncology: Navigating the road to clinical implementation CADIZ AUDITORIUM - NCC

[Slide 1] SBRT is an effective local therapy for large HCCs • 1 year LC 87% 102 HCC med 7.2cm (1.4 – 23.1cm) (Bujold), 30-54 Gy in 6 • 1 year LC 92% 16 HCC med 7.5 cm (5.1 – 9.7 cm) (Beaton) • 2 year LC 75% 156 HCC med 12.7 cm (5 - 25.7 cm) (Wong), 4Gy x 6-10 +/- TACE • 3 year LC 72% 34 HCC 4-10 cm (Kuo) • 3 year LC 82% 55 HCC >10 cm (Kuo) [Paper excerpts:] Sequential Phase I and II Trials of Stereotactic Body Radiotherapy for Locally Advanced Hepatocellular Carcinoma — Alexis Bujold, Christine A. Massey, John J. Kim, James Brierley, Charles Cho, Rebecca K.S. Wong, Rob E. Dinniwell, Zahra Kassam, Jolie Ringash, Bernard Cummings, Jenna Sykes, Morris Sherman, Jennifer J. Knox, and Laura A. Dawson Radiation dose is associated with improved local control for large, but not small, hepatocellular c[arcinoma — obscured] — Uri Amit, Jahan J Mohiuddin, Shwetha Manjunath, Nikhil Grand[i — obscured] Long-Term Outcome of Stereotactically Designed Hypofractionated Image Guided Radiation Therapy in Large, Unresectable Hepatocellular Carcinoma — Natalie Sean Man Wong, MBBS, MRes(Med), FRCR; Francis Ann Shing Lee, MBChB, FRCR; Venus Wan Yan Lee, PhD; ... Multimodality Treatment with [obscured] Stereotactic Body Radiotherapy for Large Unresectable Hepatocellular Carcinomas – A Single Institution Phase II Study — L. Beaton, E.M. Dunne, R. Yeung, T. Rackley, B. Weber, C. Mar, C.J. Yong-Hing, E.M. Yoshida, K. DeVries, R. Lee, C. Duzenli, S.K. Loewen, M. Liu, D. Schellenberg, R. Ma Impact of tumor size on outcome after stereotactic body radiation therapy for inoperable hepatocellular carcinoma — Hsing-Tao Kuo, Jenny Que, Li-Ching Lin, Ching-Chieh Yang, Lok-Beng Koay, Chia-Hui Lin [Callout box] • Local control high for large HCC • Lower survival and PFS as volume of HCC increases

[Slide 1] [ESMO Congress Madrid 2026 logo] Anti PD(L)-1 Positive phase 3 trials Immunotherapy PD(L)1 • CT + nivolumab • CT + pembrolizumab • CT + tislelizumab • CT + sintilimab • CT + sugemalimab Study | n | Arms | CT | mOS (m); HR CM649 | 1.581 | CT + Nivolumab / CT | Capox or Folfox | 14.4 (CPS ≥5) / 11.1; HR 0.71; p 0.0001 [box: CPS ≥1: 14.0 / 11.3] KN 859 | 1.579 | CT + Pembrolizumab / CT | Capox (86%) or Cisplatin + 5FU | 12.9 (ITT) / 11.5 / HR 0.78 [box: CPS ≥1: 13.0 / 11.4] RATIONALE 305 | 997 | CT + Tislelizumab / CT | Capox (93%) or Cisplatin + 5FU | 15.0 (ITT) / 12.9 / HR: 0.80; p 0.001 [box: TAP ≥5: 17.2 / 12.6 / HR: 0.74; P 0.006] ORIENT16 | 650 | CT + Sintilimab / CT | Capox | 15.2 (ITT) / 12.3 / HR: 0.77, p 0.009 [box: CPS ≥5: 18.4 / 12.9 / HR 0.66; 0.002] GEMSTONE 303 (antiPDL1) | 479 | CT + Sugemalimab / CT | Capox | 15.6 (CPS ≥5) / 12.6 / HR: 0.75; p 0.006 CM649: primary endpoint amendment: OS or PFS by blinded independent central review, in PD-L1 (CPS) ≥5 patients # CM649: Primary endpoint in PDL1 CPS ≥5; KN859: CPS ≥1; Rationale 305: All and TAP ≥5% Orient16: All. CPS ≥5: OS: 18.4 Gemstone 303: CPS ≥5 required for inclusion Anelisa K. Coutinho, MD Janjigian et al. Lancet ;398:27–40, 2021 | Rha et al. Lancet Oncol. 24:1181–1195, 2023 | Xu et al. JAMA. 330:2064–2074, 2023 | Qiu et al. BMJ. 385:e078876, 2024 | Zhang et al. JAMA. 333:1305–1314, 2025 | Janjigian et al. Ann Oncol 37:787–797, 2026 [Photo also shows stage with ESMO Congress Madrid 2026 backdrop and a seated panelist]

[Slide 3] 4355P Biomarker signals in MSS/pMMR metastatic colorectal cancer (mCRC) are predominantly prognostic, not predictive – lessons from the randomized phase-2 METIMMOX trial Benjamin P. Geisler1,2@, Megan Othus3, Sebastian Meltzer2, Paula A. Bousquet2, Emily A. Burger1,4, Eline Aas1,5, Anne H. Ree2,6 1Department of Health Management and Health Economics, University of Oslo, Oslo, Norway; 2Department of Oncology, Akershus University Hospital, Lørenskog, Norway; 3Fred Hutchinson Cancer Center, Seattle WA, United States; 4Center for Health Decision Science, Harvard T. H. Chan School of Public Health, Boston MA, United States; 5Division for Health Services, Norwegian Institute of Public Health, Oslo, Norway; 6Institute of Clinical Medicine, University of Oslo, Oslo, Norway Background [I]mmune checkpoint blockade is ineffective in unselected MSS/pMMR mCRC. [M]ETIMMOX (randomized phase 2; NCT03388190): alternating oxaliplatin-based chemotherapy (FLOX) + nivolumab vs FLOX alone; identical PFS. [P]ost hoc subgroup signals (TMB/BRAF, CRP, TLR) conflated prognostic and predictive effects; we separated them in one pre-specified interaction model per endpoint. Methods ○ 68 complete cases (59 deaths; 63 PFS events). ○ Biomarkers before nivolumab: TMB ≥9 mut/Mb or BRAF V600E; CRP <5 mg/L after 2 FLOX cycles common to both arms (by chance 47% vs 19% positive). ○ TLR ≥10% at the first on-treatment CT: analyzed as intermediate endpoint with a week-9 landmark. ○ Per endpoint (PFS, OS): one Firth-corrected Cox model with age, sex, treatment, biomarker main effects and treatment × biomarker interactions (Figure 1); profile-likelihood 95% CIs, PLRT. ○ Ridge-penalized Cox as co-primary stability estimator. [Figure 1: DAG - Age/Sex, TMB/BRAF, CRP → PFS/OS; partially obscured by presenter] Results Figure 2. Kaplan-Meier survival by biomarker status and treatment arm a) CRP <5 mg/L at week 4 (n = 68) b) TMB ≥9 mut/Mb or BRAF V600E (n = 68) c) TLR ≥10%, week-9 landmark (n = 65) [Overall survival panels a-c; Progression-free survival panels d-f; x-axis Weeks since randomization / Weeks since week-9 landmark] Legend: FLOX/nivolumab, biomarker-positive; FLOX/nivolumab, biomarker-negative; FLOX alone, biomarker-positive; FLOX alone, biomarker-negative ○ No marginal treatment effect: OS HR 0.96 (95% CI 0.58–1.61); PFS HR 0.77 (0.44–1.38). ○ Prognostic: CRP <5 mg/L and TMB/BRAF positivity were associated with longer PFS (HR 0.39 each) and, for CRP, longer OS (HR 0.49), regardless of treatment (Table 1, Figure 2). ○ Not predictive: no treatment × biomarker interaction reached significance (Figure 3). Largest signal: CRP × treatment for PFS, HR 0.26 (0.06–1.34), p = 0.10, but it shrinks toward 1 under ridge (HR 0.66) and is weaker for OS (HR 0.65). ○ TLR: strongly prognostic (landmark HR 0.35 OS, 0.21 PFS), but more frequent with FLOX alone (76% vs 53%), with inconsistent interactions (HR 2.47 OS, 0.88 PFS): a marker of chemotherapy-responsive disease, not of nivolumab benefit. Table 1. Hazard ratios (95% profile-likelihood CI) from Firth-corrected Cox regression Term | Model | OS HR (95% CI) | p | Ridge HR | PFS HR (95% CI) | p | Ridge HR Treatment (FLOX/nivolumab vs FLOX) | Univariable | 0.96 (0.58–1.61) | 0.87 | – | 0.77 (0.44–1.38) | 0.38 | – CRP <5 mg/L at week 4 | Univariable (prognostic) | 0.49 (0.27–0.84) | 0.009 | – | 0.39 (0.20–0.73) | 0.003 | – TMB ≥9 mut/Mb or BRAF V600E | Univariable (prognostic) | 0.68 (0.40–1.13) | 0.14 | – | 0.39 (0.21–0.73) | 0.003 | – TLR ≥10% (week-9 landmark)† | Univariable (prognostic) | 0.35 (0.21–0.61) | <0.001 | – | 0.21 (0.11–0.42) | <0.001 | – CRP × treatment | Unified interaction model | 0.65 (0.19–2.60) | 0.52 | 0.80 | 0.26 (0.06–1.34) | 0.10 | 0.66 TMB/BRAF × treatment | Unified interaction model | 0.95 (0.32–2.82) | 0.92 | 0.91 | 0.65 (0.18–2.41) | 0.52 | 0.74 TLR × treatment | Landmark responder model† | 2.47 (0.75–7.75) | 0.14 | 1.13 | 0.88 (0.21–3.38) | 0.85 | 0.63 [Table footnote: small print, partially illegible] Figure 3. Treatment × biomarker interactions [forest plot, small print illegible] Conclusion In this exploratory analysis, biomarker signals in MSS/pMMR mCRC were predominantly prognostic rather than predictive. The CRP × treatment interaction for PFS is hypothesis-generating, and evaluation in adequately powered prospective studies could be considered. @ b.p.geisler@medisin.uio.no Disclosure: B.P. Geisler has no potential conflicts of interest to declare. Funding: [small print, partially illegible] ...

[Slide 1] a Pre-ICB combination ~ 6 months pre-invika treatment b Baseline (for invika) disease progression post ICB combination c Invika monotherap 9+ months post-treatment [CT chest images, red arrows marking lesions in a and b] [Slide 2] b OBD HPV+ | OBD including TMB-H or MSI-H or immunogenic Best percentage change from baseline in target lesions Tumor types (left to right): Anal, HNSCC, Cervical, Anal, HNSCC | MSS CRC, MSS CRC, NSCLC, Uveal melanoma, TNBC, MSS CRC, DDA, NSCLC, MSS CRC, Esophageal, MSS CRC, MSS CRC, Ovarian, Oropharyngeal, MSS CRC, Pancreatic, Cervical, MSI CRC, Melanoma*, MSS CRC, MSS CRC, NSCLC

[Slide 1] Gastric Cancer -> Biomarker Testing (• HER2 • Claudin 18.2 • PD-L1 • MSI • Biomarker X) -> 1st-Line Targeted Therapy -> 2nd-Line Targeted Therapy -> Repeat Biopsy (Tissue +/- ctDNA) / ctDNA / Disease Progression (Liver Metastases) -> Biomarker Testing (• HER2 • Claudin18.2 • PD-L1 • MSI • Biomarker X) -> 2nd-Line Targeted Therapy -> Progression (Peritoneal Metastases) / Repeat Biopsy (Tissue +/- ctDNA) -> Biomarker Testing (• HER2 • Claudin18.2 • PD-L1 • MSI • Biomarker X) -> 3rd-Line Targeted Therapy Raghav Sundar MD PhD Made on Biorender Leo AI [ESMO Congress Madrid 2026 logo] [Inset: speaker at podium; MADRID SPAIN 23-27 OCTOBER 2026]

[Slide 1] 8. Organoids for Running Virtual Trials KEY ISSUE Limitations of conventional clinical trials. Patient-derived organoids enable ex vivo testing of multiple treatment options, with potential integration into cancer digital twins and virtual trials. KEY INSIGHT Organoid-based virtual trials could improve treatment selection and accelerate drug development. ESMO COMMITMENT ESMO supports translational research and the evaluation of patient-derived organoids as predictive biomarkers for precision oncology. Patient-Derived Organoids Predict Treatment Response in Metastatic Colorectal Cancer Smabers LP et al., Clinical Cancer Research (2025) [ESMO Congress Madrid 2026 logo]

[Slide 1] A Randomized Phase 1 Study of Telisotuzumab Adizutecan (ABBV-400, Temab-A) in Combination With Bevacizumab Versus Trifluridine-Tipiracil With Bevacizumab in Patients With Metastatic Colorectal Cancer Michael Cecchini, Marcia Cruz-Correa, Sae-Won Han, Kanwal Raghav, Ana Landa-Magdalena, John Strickler, Aitana Calvo, Antoine Hollebecque, Takako Eguchi Nakajima, Gali Perl, Vikram Jain, Zoë Hunter, Edward M. Chan, Martha Blaney, Rui Li, Athan Vasilopoulos, Apurvasena Parikh, Manish R. Sharma BACKGROUND • Temab-A is an ADC composed of the c-Met–directed antibody, telisotuzumab, conjugated to a novel topoisomerase 1 inhibitor • Increased expression of c-Met protein is common in CRC and carries a poor prognosis Validated c-Met–targeted antibody (telisotuzumab); Potent Top1i payload (adizutecan); Stable linker - Minimizes free-payload toxicity Antitumor effect of Temab-A does not rely on c-Met signaling, unlike c-Met protein–targeting TKIs and mAbs STUDY DESIGN First in human, phase 1 study of combination Temab-A and bevacizumab in c-Met unselected patients with 3L+ CRC (NCT05029882) Safety lead-in and randomized dose optimization • Median 3 prior lines of therapies • ~90% prior bevacizumab in Temab-A arms Safety Lead-In (Part 7a): DL1 (1.6 mg/kg; n=7); DL2 (2.4 mg/kg; n=8); Backfill DL3 (2.0 mg/kg; n=5); Temab-A + Bev 7.5 mg/kg IV Q3W Dose Optimization (Part 7b): R (1:1:1) N=66 Arm 1: Temab-A 2.0 mg/kg Q3W + Bev (n=22) Arm 2: Temab-A 2.4 mg/kg Q3W + Bev (n=22) Arm 3: Comparator (SOC): TFD-TPI + Bev (n=22) KEY RESULTS Selected dose: Temab-A 2.4 mg/kg + Bev Median follow-up: 18.2 months Safety and Tolerability [Butterfly bar chart of AE rates, Patients (%): Temab-A 2.4 mg/kg + Bev (n=30) vs FTD-TPI + Bev (n=20), Grade ≥3 / Grade 1-2; rows: Anemia, Neutropenia, Thrombocytopenia, Nausea, Fatigue, Vomiting, Decreased appetite, Diarrhea; bar-segment numbers too small to transcribe reliably] | Temab-A 2.4 mg/kg + Bev (n=30) | FTD-TPI + Bev (n=20) Median treatment duration, months (range) | 6.1 (1.4–19.8) | 4.9 (0.9–14.0) Treatment discontinuation, n (%) | 1 (3) | 1 (5) Dose reduction, n (%) | 15 (50) | 6 (30) Dose interruption, n (%) | 22 (73) | 9 (45) • No TRAEs leading to death • Adjudicated ILD/pneumonitis rate was 3% (Grade 3: n=1) in patients treated with 2.4 mg/kg Temab-A + Bev Efficacy ORR 30% vs 0% with FTD-TPI + Bev MEDIAN PFS 6.8 months vs 4.2 months with FTD-TPI + Bev (HR: 0.48) MEDIAN OS 14.7 months vs 10.2 months with FTD-TPI + Bev (HR: 0.62) • c-Met protein expression was detected in the tumors of all patients • Longer survival observed with Temab-A + Bev over FTD-TPI + Bev across c-Met protein expression levels Temab-A 2.4 mg/kg + Bev showed sustained clinically meaningful improvements in efficacy outcomes, and the safety profile was manageable with low discontinuation rates; hematologic adverse events were managed with dose interruptions and reductions. Temab-A 2.4mg/kg is being evaluated in the global randomized Phase 3 trial, AndroMETa-560 (NCT07525206)

[Slide 1] [PeerView Institute for Medical Education (PVI) logo] Progress Towards Personalised Care in Advanced CCA: Impact of an Online Micro-Learning Educational Initiative on Clinician Knowledge, Competence, and Practice Readiness Ben Johnson1, Michael Beyer1, Rianna Hanif1, Jessica Massey1, Angela Lamarca2, Nicola Normanno3, Rachna Shroff4† 1PVI, PeerView Institute for Medical Education, London, UK; 2Fundacion Jimenez Diaz University Hospital, Madrid, Spain; 3IRCCS Istituto Romagnolo per lo Studio dei Tumori "Dino Amadori" – IRST S.r.l., Meldola, Italy; 4University of Arizona Cancer Center (UACC), University of Arizona College of Medicine – Tucson, Tucson, Arizona, USA. † In memoriam: Dr. Shroff was instrumental in the development of this activity. We are grateful for her contributions to this activity and the field of medicine. [QR code] Background & Methods The goal of precision oncology is to deliver the right treatment to the right patient, at the right dose and at the right time. In cholangiocarcinoma (CCA), key actionable alterations, including FGFR2 alterations, IDH1/2 mutations, BRAF V600E mutations, HER2 overexpression and alterations, NTRK fusions and MSI-H/dMMR status, now inform personalised treatment selection. As the therapeutic landscape evolves, clinicians require practical support to interpret molecular findings and integrate evidence-based targeted approaches into patient-centred care. Design: PeerView deployed a CME-certified, online video-based micro-learning activity that blended the clinical journey of a patient with advanced CCA with expert guidance and relevant clinical evidence on implementing targeted treatment approaches. Five assessment questions were administered to learners after the education and to a demographically matched group of non-participants to evaluate differences in knowledge and competence. Here, we report the findings 7 months after launching this initiative. Programme Reach 2,491 learners across 72 countries(a) 34,386 patients with advanced CCA potentially impacted per month(b) 1,309 supplementary material downloads, demonstrating active engagement beyond passive viewing Interprofessional audience comprising oncologists (42%), gastroenterologists (13%), pathologists (11%), nurses/NPs/PAs (9%) and other HCPs (25%) a Data cut-off: 7 months from launch (November 2025-June 2026). b Estimated from participant-reported monthly patient load. Impact on Knowledge & Competence The objective of this initiative was to improve the understanding and practical application of precision approaches in the management of advanced CCA. 40 percentage point (pp) mean difference across all assessed knowledge and competence domains [legend] Non-participants / Learners 57 pp difference — 35% → 92% — in understanding of the clinical landscape for the management of advanced CCA and the differences in available targeted therapies 38 pp difference — 45% → 83% — in the ability to formulate team-based, personalised CCA clinical care plans that align therapy with a patient's molecular profiling results and support treatment across the care continuum 8 pp difference — 75% → 83% — in the ability to integrate evidence-based molecular profiling tools and interpret results to inform therapeutic decision-making in advanced CCA ✓ Following participation, no learners selected 'I'm not sure' on any of the five assessed knowledge/competence questions. These findings suggest meaningfully increased confidence in applying the educational content. ★ Key finding: Understanding of evidence on emerging combination targeted therapy represented the largest knowledge gap between non-participants and learners. On a question assessing this area, 16% of non-participants and 92% of learners responded correctly—a 76-pp difference and the largest of any assessed question. This suggests that clinicians in the target audience may benefit from accessible updates as the therapeutic landscape evolves. Ongoing Unmet Needs Self-reported practice considerations among non-participants (n = 50) 84% reported opportunities for further support in applying updated treatment guidance in routine practice 68% described practical challenges in ensuring timely, comprehensive molecular profiling 72% identified a need for further support in recognising appropriate clinical trial referral opportunities Key Themes Identified for Future Education Open-ended requests for more information/future education among non-participants (n = 50) 1 Therapeutic sequencing after frontline chemoimmunotherapy — Uncertainty about optimal treatment decisions following progression on frontline regimens was the most frequently cited clinical challenge. 2 Targeted therapy safety in community settings — Targeted therapy toxicities were a particular area of uncertainty for non-specialist and community-based clinicians. Recognition and grading of toxicities and dose-modification protocols represent a distinct, actionable educational gap. 3 Keeping pace with the emerging evidence base — Rapidly evolving early-phase combination trial data represented the largest driver of knowledge gaps in this programme. Clinicians require accessible, evidence-based updates to translate emerging findings confidently into clinical decision-making. The corresponding author has no conflicts of interest to share. Corresponding author contact: ben.johnson@peerview.com Jointly accredited CME activity | PVI, PeerView Institute for Medical Education This activity is supported by an educational grant from Servier. [Back to search button]

[Slide 2] RESULTS · PHASE 2 · PATIENT SUMMARY Patient summaries rated 9.0 / 10 by 15 Clinicians & AHPs 145 reviews · 15 MDT members, 6 roles · every case, ready or deferred All Clinicians & AHPs | 9.0 / 10 | 92% | 134 of 145 rated 8-10 Medical oncologists | 9.2 / 10 | 93% | 28 of 30 rated 8-10 Colorectal surgeons | 9.0 / 10 | 94% | 67 of 71 rated 8-10 Mean summary rating by role (1–10) Colorectal surgeon (7) | 8.98 Nurse & MDT coordinator (2) | 9.2 Medical oncologist (3) | 9.2 Pathologist (2) | 10 Shazad Ashraf [Slide 3] Conclusions THE PROBLEM In 14,180 MDT cases, 8.2% were deferred and trials were discussed in only 3.9%. PHASE 1 · ABSTRACT Across 15 synthetic cases, BEAGLE®-MDT scored 9.2-10.0 / 10 for summaries, management and trial matching. PHASE 2 · NEW SINCE ABSTRACT Across 15 MDT members, the new deferral gate matched clinicians in 83% of reviews; open trials identified for 92% of cases. WHAT COMES NEXT Improve trial identification and patient/referrer letters. Shadow-mode and prospective evaluation to demonstrate fewer deferrals, more trial access, and safety. BEAGLE®-MDT is designed to support, not replace, multidisciplinary clinical expertise.