4093

Background: KEYNOTE-966 demonstrated that adding pembro to gem/cis provided a statistically significant, clinically meaningful improvement in OS as first-line therapy for BTC. After a median follow-up (ie, time from randomization to data cutoff) of 25.6 months (mo), median OS was 12.7 mo for pembro + gem/cis vs 10.9 mo with placebo + gem/cis (hazard ratio [HR] 0.83, 95% confidence interval [CI] 0.72-0.95; P = 0.0034; data cutoff date December 15, 2022). We present updated efficacy and safety data after 11 mo of additional follow-up. Methods: Key eligibility criteria included age ≥18 years; previously untreated, histologically confirmed metastatic or unresectable locally advanced BTC measurable by RECIST v1.1; and ECOG PS 0 or 1. Pts were randomized 1:1 to pembro 200 mg or placebo Q3W for ≤35 cycles plus gem 1000 mg/m² on days 1 and 8 Q3W until PD and cis 25 mg/m² on days 1 and 8 for <8 cycles.Pts were stratified by region (Asia vs non-Asia), stage (locally advanced vs metastatic), and site of origin (gallbladder vs intrahepatic vs extrahepatic). The primary end point was OS; secondary end points were PFS, ORR, and DOR assessed per RECIST v1.1 by blinded independent central review, and safety. The data cutoff date for this analysis was November 14, 2023. Results: 1069 pts were randomized to pembro + gem/cis (n = 533) or placebo + gem/cis (n = 536). At a median follow-up of 36.6 mo (range, 29.2-49.4), the OS benefit of pembro + gem/cis was maintained (HR 0.86, 95% CI 0.75-0.98). Median OS was 12.7 mo (95% CI 11.5-13.6) for the pembro arm vs 10.9 mo (95% CI 9.9-11.6) for the placebo arm. At 24 mo, the OS rate was 24.6% (95% CI 21.0-28.3) in the pembro arm vs 19.2% (95% CI 16.0-22.6) in the placebo arm. OS in key subgroups continued to favor the pembro arm. Median PFS was 6.5 mo (95% CI 5.7-6.9) for the placebo arm vs 5.6 mo (95% CI 4.9-6.5) for the pembro arm (HR 0.85, 95% CI 0.75-0.97). ORR was 28.7% (n = 153) in the pembro arm vs 28.7% (n = 154) in the placebo arm. DOR was 8.3 mo (range: 1.2+ to 44.3+) for the pembro arm vs 6.9 mo (1.1+ to 41.1+) for the placebo arm; at 18 mo, estimates of ongoing response were 24% in the pembro arm vs 14% in the placebo arm. 378 (71.5%) out of 529 treated pts in the pembro arm had grade 3-5 treatment-related adverse events (AEs) vs 370 (69.3%) out of 534 treated pts in the placebo arm. 31 (5.9%) pts in the pembro arm died vs 50 (9.4%) in the placebo arm. There were no additional treatment-related deaths since the final analysis. Conclusions: With a median follow-up of 36.6 mo, the clinically meaningful improvement in OS with pembro + gem/cis was maintained, with no new safety signals, compared with placebo + gem/cis in pts with unresectable or advanced BTC. These data support pembro + gem/cis as first-line treatment for advanced BTC.