KEYNOTE-361 Trial

[Slide 1] Conclusions In this retrospective, exploratory analysis of KEYNOTE-361, most patients were positive for ctDNA at baseline Baseline ctDNA levels were associated with clinical outcomes for pembrolizumab but not for chemotherapy Chemotherapy induced larger ctDNA decreases from baseline to treatment cycle 2 than pembrolizumab (including ctDNA clearance) Changes with pembrolizumab were more associated with long-term outcomes and were enriched with higher tTMB and PD-L1 levels Short-term ctDNA reductions were not treatment-independent surrogates for clinical outcomes Patients with larger ctDNA reductions in the pembrolizumab arm had a much more favorable os relative to patients with larger - ctDNA reductions in the chemotherapy arm for Data presented herein have demonstrated that ctDNA dynamics can be influenced by the mechanisms of treatment, which is relevant consideration of this biomarker in future combination therapies --- [Slide 2] 20 A --- [Slide 3] KEYNOTE-361 (NCT02853331) Study Design and ctDNA Substudy Design KEYNOTE-361 study design ctDNA substudy Key Eligibility Criteria Evaluable CIDNA at OTHER Histologically or Representative subset baseline 19.2% Pembrolizumab selected for ctDNA cytologically n 8 130 n = 302 analysis by WES confirmed diagnosis n = 132 of locally advanced, CtDNA+ Evaluable CIDNA at unresectable or baseline and C2 80.8% metastatic UC of the n = 115 renal, pelvic, ureter, Pembrolizumab® N = 1010 Baseline bladder or urethra + Chemotherapyb n = 349 ctDNA status No prior systemic therapy for advanced Evaluable ctDNA at disease CIDNA- baseline 12.8% Measurable disease Representative subset n = 130 Chemotherapyᵇ selected for ctDNA per RECIST v1.1 n = 342 analysis by WES ctDNA+ ECOG PS 0-2 n= 131 Tumor tissue sample Evaluable ctDNA at 87.2% baseline and C2 for PD-L1 assessment n = 123 ctDNA substudy objectives: Determine whether baseline or on-treatment changes in ctDNA levels defined by maxVAF were associated with clinical outcomes (BOR, PFS, and baseline OS) Secondary: Primary: Determine whether maxVAF changes from baseline to cycle 2 were associated with outcome in models adjusting for (TMB/PD-L1 subgroups, prognostic cvcle 2: factors, ctDNA. and circulating radiographic tumor DNA; response maxVAF, by RECIST maximum v1.1 variant allele frequency; ITMB, tissue tumor mutational burden. dNA was collected at pre-dose prior to first therapy cycle and at cycle 2 (3-week I --- [Slide 4] Between Arm Comparisons of OS by Baseline ctDNA Levels Baseline maxVAF <median 100 Baseline maxVAF ≥median + Chemotherapy (n = 62) 100 + Chemotherapy (n = 63) + Pembrolizumab (n = 63) + Pembrolizumab (n = 62) 75 75 Overall Survival, % 50 Overall Survival, % 50 25 25 HR = 0.65 (95% CI, 0.42-1.02) HR = 1.06 (95% CI, 0.71-1.58) 0 0 0 10 20 30 40 0 10 20 30 40 Time, months Time, months Median 3.84 Clinical data cutoff: April 29, 2020.

[Slide 1] Between Arm Comparisons of os by Baseline ctDNA Levels Baseline maxVAF >median Baseline maxVAF <median 100 100 Chemotherapy (n = 63) Chemotherapy (n = 62) Pembrolizumab (n = 62) Pembrolizumab (n = 63) 75 75 50 50 ++++++ III 25 25 TH HR = 0.65 (95% CI, 0.42-1.02) HR = 1.06 (95% CI, 0.71-1.58) 0 0 0 10 20 30 40 0 10 20 30 40 Time, months Time, months Median, 3.84. Clinical data cutoff: April 29, 2020. --- [Slide 2] Quantitative Circulating Tumor DNA Assessment in Patients With Advanced Urothelial Carcinoma Treated With Pembrolizumab or Platinum-Based Chemotherapy From the Phase 3 KEYNOTE-361 Trial Thomas Powles¹, Yen-Hwa Chang², Yoshiaki Yamamoto³, Jose Munoz⁴, Felipe Reyes⁵, Avivit Peer⁶, Graham Cohen⁷, Evan Y. Yu8, Anja Lorch⁹, Abhishek Bavle¹⁰, Blanca Homet Moreno¹⁰, Julia Markensohn¹⁰, Mackenzie Edmondson¹⁰, Cai Chen¹⁰, Razvan Cristescu¹⁰, Carol Pena¹⁰, Jared Lunceford¹⁰, Seyda Gunduz¹¹ 'Barts Cancer Institute, Queen Mary University of London, London, United Kingdom; 2Taipei Veterans General Hospital, Taipel, Taiwan; 3Yamaguchi University Hospital, Yamaguchi, Japan; Hospital Universitari i Politècnic La Fe, Valencia, Spain; Fundación Arturo López Pérez, Santiago, Chile; ⁵Rambam Health Care Campus, Haifa, Israel; Mary Potter Oncology Centre, Gauteng, South Africa; *Fred Hutchinson Cancer Center and University of Washington, Seattle, Washington, USA; Universitätsspital Zürich, Zürich, Switzerland; 10Merck & Co., Inc., Rahway, NJ, USA; "Istinye University Liv Hospital, Istanbul, Turkey --- [Slide 3] I ASCO ANNUAL MEETING 2024ASCO --- [Slide 4] Between Arm Comparisons of OS by Cycle 2/Cycle 1 Reductions by Treatment Arm Large Reduction Not Large Reduction Large Reduction Not Large Reduction 100 Chemotherapy (n = 78) 100 Chemotherapy (n = 24) Pembrolizumab (n = 19) Pembrolizumab (n = 72) 3.00 ctDNA Change with Treatment 75 75 2.00 Overall Survival, % 50 1.00 Overall Survival, % 50 0.50 0.25 25 25 HR = 0.25 (95% CI, 0.10-0.62) HR = 0.82 (95% CI, 0.48-1.38) 0.00 0 0 Chemotherapy Pembrolizumab Chemotherapy Pembrolizumab 0 10 20 30 40 0 10 20 30 40 n 78 n 19 n 24 n = 72 Time, months Time, months Treatment ctDNA change with treatment = ratio of C2/C1 maxVAF. Median, 0.18. Large reduction was defined as the ratio of C2/C1 reductions <median; Not large reduction was defined as the ratio of C2/C1 reductions 2 median. Clinical data cutoff: April 29, 2020. --- [Slide 5] Between Arm Comparisons of OS by Baseline ctDNA Levels Baseline maxVAF <median Baseline maxVAF ≥median 100 Chemotherapy (n = 62) 100 + Chemotherapy (n = 63) Pembrolizumab (n = 63) Pembrolizumab (n = 62) 75 75 Overall Survival, % 50 Overall Survival, % 50 25 25 HR = 0.65 (95% CI, 0.42-1.02) HR = 1.06 (95% CI, 0.71-1.58) 0 0 0 10 20 30 40 0 10 20 30 40 Time, months Time, months Median, 3.84. Clinical data cutoff: April 29, 2020.

[Slide 1] Association Between Baseline ctDNA Levels and Clinical Outcomes by Treatment Arm Chemotherapy Pembrolizumab P value reflecting the association between baseline ctDNA levels and 75 clinical outcomes* 50 Outcomes Chemotherapy Pembrolizumab Baseline Tumor-Informed maxVAF n = 125 n = 125 25 BOR per RECIST v1.1 by 0.172 0.009 BICR PFS per RECIST v1.1 by 0.060 2.03 X 10-5 investigator 0 OS 0.088 6.14 X 10-⁵ Nonresponders Responders Nonresponders Responders n = 62 n 63 n 79 n 46 *Analysis showed similar results after adjustment for baseline tumor size and TMB/PD-L1 status Response Association was evaluated using logistic regression (BOR) and Cox proportional hazards regression (PFS and OS), with adjustment for ECOG PS. Multiplicity-adjusted P values were calculated. Significance was prespecified at a # 0.05 Bold font indicates significance. Hypothesis: negative association Clinical data cutoff: April 29, 2020. --- [Slide 2] Best 9-Week Percentage Radiological Change From Baseline in Tumor Size Chemotherapy Pembrolizumab 100 ctDNA Reduction Category ctDNA clearance ctDNA+, reduction ≥median 50 % Change in Tumor Size at 9 Weeks ctDNA+, reduction <median 20 0 -30 -50 -110 Clinical data cutoff: April 29, 2020 --- [Slide 3] Patient-Level Cycle 2/Cycle 1 ctDNA Changes by tTMB and PD-L1 Status and Treatment Arm tTMB high/PD-L1 low tTMB high/PD-L1 high tTMB low/PD-L1 low tTMB low/PD-L1 high 2.00 2.00 ctDNA Change with Treatment 1.00 ctDNA Change with Treatment 1.00 $ 0.50 0.50 0.25 0.25 0.00 0.00 Chemotherapy Pembrolizumab Chemotherapy Pembrolizumab Chemotherapy Pembrolizumab Chemotherapy Pembrolizumab n= 13 n=8 n 22 n 22 n 41 n 34 n . 26 n = 23 Treatment Treatment ITMB high, a175 mut/exome; ITMB low, <175 mut/exome, PD-L1 high, CPS >10; PD-L1 low, CPS <10. cIDNA change with treatment " ratio of C2/C1 maxVAF, v1. ITMB was obtained from available tissue and normal WES data. Clinical data cutoff: April 29, 2020.

[Slide 1] KEYNOTE-361 (NCT02853331) Study Design and ctDNA Substudy Design KEYNOTE-361 study design ctDNA substudy Key Eligibility Criteria Evaluable ctDNA at CIDNA- Representative subset baseline 19.2% Histologically or Pembrolizumab* selected for ctDNA n = 130 n 302 analysis by WES cytologically n - 132 confirmed diagnosis ctDNA+ Evaluable ctDNA at of locally advanced, 80.8% baseline and C2 unresectable or n = 115 metastatic UC of the Pembrolizumab* renal, pelvic, ureter, Baseline N = 1010 bladder or urethra Chemotherapy® ctDNA status n 349 No prior systemic therapy for advanced Evaluable ctDNA at disease CIDNA. baseline Measurable disease Representative subset 12.8% n=130 Chemotherapyb selected for CIDNA per RECIST v1.1 n 342 analysis by WES ECOG PS 0-2 n = 131 ctDNA+ Tumor tissue sample Evaluable ctDNA at 87.2% for PD-L1 assessment baseline and C2 n = 123 ctDNA substudy objectives: Primary: Determine whether baseline or on-treatment changes in ctDNA levels defined by maxVAF were associated with clinical outcomes (BOR, PFS, and OS) Secondary: Determine whether maxVAF changes from baseline to cycle 2 were associated with outcome in models adjusting for (TMB/PD-L1 subgroups, baseline prognostic factors, and radiographic response by RECIST v1.1 C2, cycle 2; ctDNA, circulating tumor DNA: maxVAF, maximum variant allele frequency: TMB, tissue tumor mutational burden. ctDNA was collected at pre-dose prior to first therapy cycle and at cycle 2 (3-week 1200 -- ow rusles). 1000 main? DC - day and dau n ow choice of when circlation TO maint are own -- carborlatio AUG maint leelin INF --- [Slide 2] Best 9-Week Percentage Radiological Change From Baseline in Tumor Size Chemotherapy Pembrolizumab 100 ctDNA Reduction Category ctDNA clearance ctDNA+, reduction >median 50 % Change in Tumor Size at 9 Weeks ctDNA+, reduction <median 20 0 -30 -50 ... --- [Slide 3] Within Arm Associations Between Cycle 2/Cycle 1 ctDNA Change and Clinical Outcomes Chemotherapy Pembrolizumab P value reflecting the association between cycle 2/cycle 1 ctDNA change and clinical outcomes* 3.00 Outcomes Chemotherapy Pembrolizumab 2.00 ctDNA Change with Treatment n 102 n = 87 1.00 BOR per RECIST v1.1 by 1.34 X 10⁴ 7.83 X 10-5 BICR 0.50 0.25 PFS per RECIST v1.1 by 1.20 X 10⁴ 1.61 X 10⁻⁵ investigator 0.00 os 0.011 0.002 CR PR SD PD CR PR SD PD n=12 n = 38 n = 39 n=11 n=15 n = 17 n = 15 n=35 "After baseline tumor size and ttmb/PD-L1 status adjustments Best Overall Response --- [Slide 4] Between Arm Comparisons of OS by Cycle 2/Cycle 1 Reductions by Treatment Arm Large Reduction Not Large Reduction Large Reduction Not Large Reduction 100 Chemotherapy (n = 78) 100 Chemotherapy (n = 24) Pembrolizumab (n = 19) Pembrolizumab (n = 72) 3.00 ctDNA Change with Treatment 75 75 2.00 Overall Survival, % 50 1.00 Overall Survival, % 50 0.50 0.25 25 25 HR = 0.25 (95% CI, 0.10-0.62) HR # 0.82 (95% CI, 0.48-1.38) 0.00 0 0 Chemotherapy Pembrolizumab Chemotherapy Pembrolizumab 0 10 20 30 40 0 10 20 30 40 n - 78 n - 19 n # 24 n - 72 Time Time

[Slide 1] KEYNOTE-361 (NCT02853331) Study Design and ctDNA Substudy Design KEYNOTE-361 study design ctDNA substudy Key Eligibility Criteria Evaluable ctDNA at ctDNA Representative subset baseline 19.2% Histologically or Pembrolizumab® selected for ctDNA n 130 n 302 analysis by WES cytologically n 132 confirmed diagnosis ctDNA+ Evaluable ctDNA at of locally advanced, 80.8% baseline and C2 unresectable or n 115 metastatic UC of the Pembrolizumab® renal, pelvic, ureter, Baseline N 1010 bladder or urethra Chemotherapyb ctDNA status n 349 No prior systemic therapy for advanced Evaluable ctDNA at disease CIDNA- baseline 12.8% Measurable disease Representative subset n 130 Chemotherapyb selected for ctDNA per RECIST v1.1 n 342 analysis by WES ECOG PS 0-2 n 131 ctDNA+ Tumor tissue sample Evaluable ctDNA at 87.2% for PD-L1 assessment baseline and C2 n 123 ctDNA substudy objectives: Primary: Determine whether baseline or on-treatment changes in ctDNA levels defined by maxVAF were associated with clinical outcomes (BOR, PFS, and OS) Secondary: Determine whether maxVAF changes from baseline to cycle 2 were associated with outcome in models adjusting for TMB/PD-L1 subgroups, baseline prognostic factors, and radiographic response by RECIST v1.1 C2, cycle 2; ctDNA. circulating tumor DNA: maxVAF, maximum variant allele frequency: TMB, tissue tumor mutational burden. ctDNA was collected at pre-dose prior to first therapy cycle and at cycle 2 (3-week postbaseline assessment). *200 mg IV Q3W (s35 cycles). Gemcitabine 1000 mg/m2 IV on day 1 and day 8 Q3W investigator's choice of either cisplatin 70 mg/m2 IV Q3W or carboplatin AUC 5 mg/mL/min IV Q3W (s 6 cycles). Assessed using tumor-informed maxVAF. Clinical data cutoff: April 29, 2020. ctDNA Measures Evaluated with GuardantOMNI assay MaxVAF (tumor-independent) Uses maximum detected variant allele frequency (VAF) of putative somatic mutations to quantify ctDNA levels MeanVAF (tumor-independent) Uses average of VAF of putative somatic mutations detected in C1 sample variants cleared at C2 (3-week postbaseline assessment) included as zeros in the average to quantify ctDNA levels Tumor-informed maxVAF Uses paired tissue and matched normal (whole blood) WES to identify somatic variants in ctDNA Guardant Health Molecular Proprietary algorithm calculating change in ctDNA levels (not Response score relevant for baseline) Data in this presentation focuses on the tumor-informed metric C1, cycle 1 (baseline). ctDNA levels were assessed using the 500-gene fixed-panel NGS-based GuardantOMNI assay. --- [Slide 2] Association Between Baseline ctDNA Levels and Clinical Outcomes by Treatment Arm Chemotherapy Pembrolizumab P value reflecting the association between baseline ctDNA levels and 75 clinical outcomes* 50 Outcomes Chemotherapy Pembrolizumab Baseline Tumor-Informed maxVAF n 125 n 125 25 BOR per RECIST v1.1 by 0.172 0.009 BICR PFS per RECIST v1.1 by 0.060 2.03 X 10⁻⁵ investigator 0 OS 0.088 6.14 X 10⁻⁵ Nonresponders Responders Nonresponders Responders n . 62 n 63 n 79 n 46 "Analysis showed similar results after adjustment for baseline tumor size and tTMB/PD-L1 status Response Association was evaluated using logistic regression (BOR) and Cox proportional hazards regression (PFS and OS), with adjustment for ECOG PS. Multiplicity-adjusted P values were calculated Significance was prespecified at a . 0.05. Bold font indicates significance. Hypothesis: negative association Clinical data cutoff: April 29, 2020. Between Arm Comparisons of OS by Baseline ctDNA Levels Baseline maxVAF <median Baseline maxVAF ≥median 100 Chemotherapy (n = 62) 100 Chemotherapy (n = 63) Pembrolizumab (n = 63) Pembrolizumab (n = 62) 75 75 Overall Survival, % 50 Overall Survival, % 50 25 25 HR = 0.65 (95% CI, 0.42-1.02) HR = 1.06 (95% CI, 0.71-1.58) 0 0 0 10 20 30 40 0 10 20 30 40 Time, months Time, months Median, 3.84. Clinical data cutoff: April 29, 2020. --- [Slide 3] Cycle 2/Cycle 1 ctDNA Changes and Clearance Rates Chemotherapy Pembrolizumab ctDNA Reduction Category 3 ctDNA clearance C2/C1 Tumor-Informed maxVAF Ratio ctDNA+, reduction >median ctDNA+, reduction <median (including increases) 2 41% 11% 1 Median across treatment groups 0 Clinical data cutoff: April 29, 2020. Best 9-Week Percentage Radiological Change From Baseline in Tumor Size Chemotherapy Pembrolizumab 100 ctDNA Reduction Category ctDNA clearance ctDNA+, reduction >median 50 % Change in Tumor Size at 9 Weeks ctDNA+, reduction <median 20 0 -30 -50 -110 Clinical data cutoff: April 29, 2020. --- [Slide 4] Within Arm Associations Between Cycle 2/Cycle 1 ctDNA Change and Clinical Outcomes Chemotherapy Pembrolizumab P value reflecting the association between cycle 2/cycle 1 ctDNA change and clinical outcomes* 3.00 Outcomes Chemotherapy Pembrolizumab 2.00 ctDNA Change with Treatment n 102 n 87 1.00 BOR per RECIST v1.1 by 1.34 X 10-4 7.83 X 10-⁵ BICR 0.50 0.25 PFS per RECIST v1.1 by 1.20 X 10.6 1.61 X 10-5 investigator 0.00 OS 0.011 0.002 CR PR SD PD CR PR SD PD n = 12 n = 38 n = 39 n = 11 n = 15 n = 17 n = 15 n = 35 *After baseline tumor size and tTMB/PD-L1 status adjustments Best Overall Response ctDNA change with treatment . ratio of C2/C1 maxVAF, v1. Association was evaluated using logistic regression (BOR) and Cox proportional hazards regression (PFS and OS), with adjustment for ECOG PS. Multiplicity-adjusted P values were calculated Significance was prespecified at a 0.05. Bold font indicates significance Hypothesis: negative association Clinical data cutoff: April 29, 2020. Patient-Level Cycle 2/Cycle 1 ctDNA Changes by tTMB and PD-L1 Status and Treatment Arm tTMB high/PD-L1 low tTMB high/PD-L1 high tTMB low/PD-L1 low tTMB low/PD-L1 high 2.00 2.00 ctDNA Change with Treatment 1.00 ctDNA Change with Treatment 1.00 : 0.50 0.50 0.25 0.25 0.00 0.00 Chemotherapy Pembrolizumab Chemotherapy Pembrolizumab Chemotherapy Pembrolizumab Chemotherapy Pembrolizumab n 13 n=8 n 22 n 22 n 41 n 34 n 26 n 23 Treatment Treatment tTMB high, >175 mut/exome; tTMB low, <175 mut/exome; PD-L1 high, CPS >10; PD-L1 low, CPS <10. ctDNA change with treatment - ratio of C2/C1 maxVAF, v1. TTMB was obtained from available tissue and normal WES data. Clinical data cutoff: April 29, 2020.