Background
Phase III double-blind RCT, N=840 (main cohort), resectable locally advanced gastric or GEJ cancer (cT2+, node-positive, or T4 any N; no prior treatment; HER2-negative). 24 countries; ~48% Asian patients. PD-L1 CPS assessed. Tested perioperative pembrolizumab + cisplatin-based chemotherapy vs placebo + chemotherapy.
Interventions and follow up
Arm A: Pembrolizumab 200 mg q3wk + cisplatin-based chemotherapy: neoadjuvant × 3 cycles → surgery → adjuvant chemo × 3 cycles → pembrolizumab maintenance × 11 cycle
Arm B: Placebo + cisplatin-based chemotherapy (same perioperative schedule)
Primary endpoint: EFS and pCR (co-primary)
mFollow up: 59.9 mo (final analysis)
Arm B: Placebo + cisplatin-based chemotherapy (same perioperative schedule)
Primary endpoint: EFS and pCR (co-primary)
mFollow up: 59.9 mo (final analysis)
Results
pCR: 13.4% vs 2.0% (main cohort, p-value not reported as significant for EFS boundary)
EFS: Pre-specified boundary not met (HR not reaching significance)
mOS: 71.8 vs 55.7 mo, HR 0.86, 95% CI 0.71–1.06 (final, P=.17, not significant)
EFS: Pre-specified boundary not met (HR not reaching significance)
mOS: 71.8 vs 55.7 mo, HR 0.86, 95% CI 0.71–1.06 (final, P=.17, not significant)
Adverse events
Overall: Grade ≥3 AEs 73% vs 69%; treatment discontinuation 20% (pembrolizumab) vs 8% (placebo)
Immune-related: Any grade 31% vs 12%; grade ≥3 6% vs 1%
Immune-related: Any grade 31% vs 12%; grade ≥3 6% vs 1%
Conclusions
Perioperative pembrolizumab + cisplatin-based chemotherapy significantly improved pCR rates but failed to achieve its co-primary endpoints of EFS or OS benefit in resectable gastric/GEJ cancer. pCR improvement did not translate to survival benefit, making KEYNOTE-585 a practice-negative trial.
Key Limitations
Key Limitations: pCR improved markedly yet did not translate to EFS/OS — challenging pCR's role as a survival surrogate in perioperative gastric cancer. Chemotherapy backbone was cisplatin-based, not FLOT; MATTERHORN (durvalumab + FLOT, positive EFS) suggests backbone choice is critical. PD-L1 CPS not used as enrollment criterion. Complex multi-phase perioperative IO delivery. Possible negative interaction between platinum-based chemo and anti-PD-1 in perioperative context.
Clinical Context
Perioperative pembrolizumab is NOT standard of care for resectable gastric/GEJ cancer. MATTERHORN (durvalumab + FLOT) demonstrated significant EFS benefit (HR 0.71, P<.001) and is FDA-approved for this setting, likely reflecting the superiority of FLOT as the backbone. In metastatic disease, pembrolizumab (KEYNOTE-859, -590) remains first-line standard. The KEYNOTE-585 vs MATTERHORN contrast underscores that not all perioperative IO combinations are equivalent.
References