Background
Phase III open-label RCT (CheckMate 651). N=947 patients with recurrent/metastatic HNSCC, any PD-L1 expression, ECOG PS 0–1, no prior systemic therapy for R/M disease. Enrolled 2016–2019 at 168 global sites. Tested first-line nivolumab (anti-PD-1) + ipilimumab (anti-CTLA-4) dual checkpoint blockade vs EXTREME (cetuximab + platinum + 5-FU or paclitaxel), the then-standard first-line R/M HNSCC regimen. Hierarchical testing: first tested TPS ≥20%, then TPS ≥1%, then all-comers.
Interventions and follow up
Arm A: Nivolumab 360 mg q3wk + ipilimumab 1 mg/kg q6wk until PD or toxicity
Arm B: EXTREME (cetuximab + cisplatin or carboplatin + 5-FU or paclitaxel × 6 cycles → cetuximab maintenance)
Primary endpoint: OS (hierarchical: TPS ≥20% → TPS ≥1% → all patients)
mFollow up: 35.6 month
Arm B: EXTREME (cetuximab + cisplatin or carboplatin + 5-FU or paclitaxel × 6 cycles → cetuximab maintenance)
Primary endpoint: OS (hierarchical: TPS ≥20% → TPS ≥1% → all patients)
mFollow up: 35.6 month
Results
OS (PD-L1 TPS ≥20%): HR 0.78 (95% CI 0.59–1.03), P=.0469 — prespecified α boundary was 0.0240; did not cross threshold; hierarchy failed
Adverse events
Main adverse events: Grade ≥3 AEs: 56.2% (nivo+ipi) vs 67.9% (EXTREME). Immune-related grade ≥3: 15.9% vs 4.0%. Treatment-related discontinuation: 17.6% vs 8.6%. EXTREME had more grade 3–4 anemia, neutropenia, and nausea.
Conclusions
Nivolumab + ipilimumab did not formally meet its primary endpoint of superior OS vs EXTREME in PD-L1 TPS ≥20% R/M HNSCC due to hierarchical testing failure (P=.0469 vs required P=.0240). Although HR 0.78 was consistent in all biomarker subgroups and the treatment was better tolerated, the trial is classified as negative by its prespecified analysis. EXTREME or pembrolizumab monotherapy (KEYNOTE-048) remain the preferred first-line standards.
Key Limitations
Key Limitations: Hierarchical testing with an extremely stringent alpha (0.0240 for TPS ≥20%) meant that even an HR of 0.78 was insufficient — the design required a larger effect in the TPS ≥20% stratum to cascade testing to the broader population. KEYNOTE-048 (pembrolizumab monotherapy or pembro + chemo) was published concurrently, showing superior OS vs EXTREME — making EXTREME a weaker comparator going forward. The dual IO combination may not be optimal in this setting; better synergy may exist with other combinations.
Clinical Context
CheckMate 651 joined KEYNOTE-412 as the second major negative or formally non-significant IO trial in HNSCC. The first-line standard for R/M HNSCC is now pembrolizumab monotherapy (CPS ≥1) or pembrolizumab + platinum-5FU (any CPS), per KEYNOTE-048. The negative CheckMate 651 reinforced that dual IO checkpoint blockade (nivo+ipi) does not clearly outperform platinum-based chemotherapy in unselected HNSCC, and the field has shifted toward IO + chemotherapy combinations for PD-L1 low patients.
References
References: Haddad RI et al, JCO 2023 (primary)