Background
5-year follow-up of the phase 3, global, open-label CheckMate 648 trial; N=970, randomized 1:1:1. Previously untreated unresectable advanced, recurrent, or metastatic esophageal squamous cell carcinoma (ESCC). Primary endpoints: OS and BICR-assessed PFS in tumor-cell PD-L1 ≥1% (49% of pts). Minimum follow-up 5 years (median 71.5 mo; cutoff Jan 13, 2025).
Results
Interventions and follow up: Arm A: Nivolumab + chemotherapy (fluorouracil + cisplatin)
Arm B: Nivolumab + ipilimumab
Arm C: Chemotherapy (fluorouracil + cisplatin)
Primary endpoint: OS and PFS (BICR) in tumor-cell PD-L1 ≥1%
mFollow up: 71.5 mo (minimum 61 mo)
Results: OS, PD-L1 ≥1%: nivo+chemo 15.0 vs 9.1 mo, HR 0.62 (95% CI 0.48–0.79); nivo+ipi 13.1 vs 9.1 mo, HR 0.62 (95% CI 0.48–0.80)
5-y OS, PD-L1 ≥1%: 12% (nivo+chemo) and 18% (nivo+ipi) vs 7% (chemo)
OS, all randomized: 13.2 (nivo+chemo) and 12.7 (nivo+ipi) vs 10.7 mo; HR 0.77 (95% CI 0.65–0.92) for both; 5-y OS 13% and 16% vs 9%
PFS, PD-L1 ≥1%: nivo+chemo 6.8 vs 4.4 mo, HR 0.67 (95% CI 0.51–0.88); nivo+ipi 4.0 vs 4.4 mo, HR 1.03 (95% CI 0.78–1.35), curves cross ~6.5 mo with late plateau (5-y PFS 9% vs 0%)
ORR, PD-L1 ≥1%: 53% (nivo+chemo) and 35% (nivo+ipi) vs 20% (chemo); CR 17% and 18% vs 4%
Arm B: Nivolumab + ipilimumab
Arm C: Chemotherapy (fluorouracil + cisplatin)
Primary endpoint: OS and PFS (BICR) in tumor-cell PD-L1 ≥1%
mFollow up: 71.5 mo (minimum 61 mo)
Results: OS, PD-L1 ≥1%: nivo+chemo 15.0 vs 9.1 mo, HR 0.62 (95% CI 0.48–0.79); nivo+ipi 13.1 vs 9.1 mo, HR 0.62 (95% CI 0.48–0.80)
5-y OS, PD-L1 ≥1%: 12% (nivo+chemo) and 18% (nivo+ipi) vs 7% (chemo)
OS, all randomized: 13.2 (nivo+chemo) and 12.7 (nivo+ipi) vs 10.7 mo; HR 0.77 (95% CI 0.65–0.92) for both; 5-y OS 13% and 16% vs 9%
PFS, PD-L1 ≥1%: nivo+chemo 6.8 vs 4.4 mo, HR 0.67 (95% CI 0.51–0.88); nivo+ipi 4.0 vs 4.4 mo, HR 1.03 (95% CI 0.78–1.35), curves cross ~6.5 mo with late plateau (5-y PFS 9% vs 0%)
ORR, PD-L1 ≥1%: 53% (nivo+chemo) and 35% (nivo+ipi) vs 20% (chemo); CR 17% and 18% vs 4%
Adverse events
Grade 3–4 TRAEs: 49% (nivo+chemo) vs 33% (nivo+ipi) vs 37% (chemo)
Any-grade TRAEs: 96% vs 80% vs 90%
Treatment-related deaths: 2% in each arm
New signals with extended follow-up: none
Any-grade TRAEs: 96% vs 80% vs 90%
Treatment-related deaths: 2% in each arm
New signals with extended follow-up: none
Conclusions
At 5 years' minimum follow-up, both nivolumab + chemotherapy and nivolumab + ipilimumab maintain clinically meaningful OS benefit over chemotherapy in 1L advanced ESCC, with 5-year OS of 12–18% vs 7% (PD-L1 ≥1%) and durable responses — long-term confirmation of the practice-changing 2022 primary analysis.
Key Limitations
Open-label design. Benefit concentrated in tumor-cell PD-L1 ≥1%; effect attenuated in PD-L1-low disease. Chemotherapy backbone (5-FU/cisplatin) and lack of head-to-head data vs pembrolizumab + chemotherapy (KEYNOTE-590) leave regimen choice to cross-trial inference. Nivo+ipi shows no median PFS benefit (early crossing curves) — a chemo-free option best suited to selected patients; PFS2 analyses are exploratory.
Clinical Context
FDA approved both nivolumab + chemotherapy and nivolumab + ipilimumab for 1L advanced ESCC (May 2022) based on this trial; both remain NCCN/ESMO-endorsed 1L options alongside pembrolizumab + chemotherapy (KEYNOTE-590). These 5-year data are the longest randomized IO follow-up in advanced ESCC and reinforce PD-L1 ≥1% enrichment for benefit.