Background
Phase III double-blind RCT, N=1579, previously untreated locally advanced or metastatic HER2-negative gastric or GEJ adenocarcinoma. 33 countries; ~46% Asian. Tested pembrolizumab + investigator-choice chemotherapy (FP or CAPOX) vs placebo + chemotherapy. PD-L1 CPS assessed at enrollment; primary endpoint hierarchically tested in CPS ≥10 → CPS ≥1 → ITT populations.
Interventions and follow up
Arm A: Pembrolizumab 200 mg q3wk (up to 35 cycles) + investigator-choice chemotherapy: cisplatin 80 mg/m² d1 + 5-FU 800 mg/m²/day d1–5 q3wk (FP), or oxaliplatin 130 mg/m² d1 + capecitabine 1000 mg/m² bid d1–14 q3wk (CAPOX)
Arm B: Placebo + FP or CAPOX (same schedule)
Primary endpoint: Overall survival (hierarchical: CPS ≥10, then CPS ≥1, then ITT)
Follow-up: 31.0 mo
Arm B: Placebo + FP or CAPOX (same schedule)
Primary endpoint: Overall survival (hierarchical: CPS ≥10, then CPS ≥1, then ITT)
Follow-up: 31.0 mo
Results
mOS (ITT): 12.9 vs 11.5 mo, HR 0.78, 95% CI 0.70–0.87, P<.0001
mOS (CPS ≥1): 13.0 vs 11.4 mo, HR 0.74, 95% CI 0.65–0.84, P<.0001
mOS (CPS ≥10): 15.7 vs 11.8 mo, HR 0.65, 95% CI 0.53–0.79, P<.0001
mOS (CPS ≥1): 13.0 vs 11.4 mo, HR 0.74, 95% CI 0.65–0.84, P<.0001
mOS (CPS ≥10): 15.7 vs 11.8 mo, HR 0.65, 95% CI 0.53–0.79, P<.0001
Adverse events
Grade ≥3 hematologic: anemia 12% vs 10%, decreased neutrophil count 10% vs 8% (pembro vs placebo).
Serious treatment-related: 23% (pembro) vs 19% (placebo).
Treatment-related deaths: 8 (pembro) vs 16 (placebo).
Serious treatment-related: 23% (pembro) vs 19% (placebo).
Treatment-related deaths: 8 (pembro) vs 16 (placebo).
Conclusions
Pembrolizumab + chemotherapy significantly improved OS across all PD-L1 subgroups in previously untreated HER2-negative gastric/GEJ adenocarcinoma, with the greatest benefit in CPS ≥10 (HR 0.65). KEYNOTE-859 provides robust confirmatory evidence across a broad gastric cancer population.
Key Limitations
Benefit is graded by PD-L1 CPS, but even CPS <1 patients showed a numeric advantage (not formally tested). OS gains are modest in absolute terms in ITT (1.4 mo). HER2-positive patients were excluded (addressed by KEYNOTE-811). No clear OS separation by chemotherapy backbone (FP vs CAPOX). Multiple competing trials (CheckMate 649, KEYNOTE-811) create complex treatment selection.
Clinical Context
FDA approved pembrolizumab + chemo for HER2-negative advanced gastric/GEJ adenocarcinoma; KEYNOTE-859 complements CheckMate 649 (nivolumab + chemo) as first-line IO + chemo standards. For CPS ≥10, pembrolizumab or nivolumab + chemo both provide ~HR 0.65 OS benefit. In CPS <1, CheckMate 649 showed no OS benefit. For HER2+ patients, trastuzumab + pembrolizumab + chemo (KEYNOTE-811) is preferred. ESMO endorses CPS-guided IO selection.