Background
Phase III open-label RCT (KEYNOTE-689). N=714, newly diagnosed resectable stage III–IVA/B HNSCC with high-risk pathologic features (ECE and/or positive resection margins) after surgery. Enrolled 2019–2023, 170 global sites. First trial of perioperative pembrolizumab (neoadjuvant + adjuvant) in resected high-risk LA-HNSCC. Stratified by tumor site, planned post-op therapy (RT vs CRT), and PD-L1 status.
Interventions and follow up
Arm A: Neoadjuvant pembrolizumab 200 mg q3wk × 2 → surgery → post-op RT or CRT + pembrolizumab, then adjuvant pembrolizumab (up to ~15 cycles)
Arm B: Surgery → standard adjuvant RT or CRT alone (no pembrolizumab)
Primary endpoint: EFS
mFollow up: 38.3 month
Arm B: Surgery → standard adjuvant RT or CRT alone (no pembrolizumab)
Primary endpoint: EFS
mFollow up: 38.3 month
Results
EFS (all patients): HR 0.73 (95% CI 0.58–0.91), P=.0036 — significant
EFS (PD-L1 CPS ≥1): HR 0.67 (95% CI 0.51–0.87), P=.0014
EFS (PD-L1 CPS ≥10): HR 0.59 (95% CI 0.41–0.85)
mEFS: not reached (pembro) vs 31.2 months (standard)
OS: immature at interim
EFS (PD-L1 CPS ≥1): HR 0.67 (95% CI 0.51–0.87), P=.0014
EFS (PD-L1 CPS ≥10): HR 0.59 (95% CI 0.41–0.85)
mEFS: not reached (pembro) vs 31.2 months (standard)
OS: immature at interim
Adverse events
Grade ≥3 AEs: 71.7% (pembro arm) vs 67.5% (standard)
Immune-mediated grade ≥3 AEs: 14.8% vs 1.6%
Discontinuation: 17.4% vs 7.8%
Surgical complications: no excess with neoadjuvant dosing; no new safety signals
Immune-mediated grade ≥3 AEs: 14.8% vs 1.6%
Discontinuation: 17.4% vs 7.8%
Surgical complications: no excess with neoadjuvant dosing; no new safety signals
Conclusions
Perioperative pembrolizumab significantly improved EFS (HR 0.73) in high-risk resected LA-HNSCC, with greater benefit in PD-L1 CPS ≥1 (HR 0.67) and CPS ≥10 (HR 0.59). First positive perioperative IO trial in resectable LA-HNSCC, establishing a new potential standard of care for high-risk patients with ECE and/or positive margins.
Key Limitations
OS data immature — EFS is a surrogate; OS benefit must be confirmed. Benefit appears PD-L1 enriched; CPS <1 subgroup HR 1.05 (CI 0.60–1.85) — no apparent benefit in PD-L1 negative patients. Perioperative regimen is complex (neoadjuvant dosing, surgery timing, long adjuvant tail), adding immune toxicity and treatment burden. Trial mixed post-op RT-only and CRT patients, limiting interpretability. Relative contribution of neoadjuvant vs adjuvant components unknown.
Clinical Context
A paradigm shift for high-risk resected HNSCC, potentially adding pembrolizumab to the adjuvant cisplatin + RT backbone established by RTOG 9501 and EORTC 22931. FDA submission and guideline updates anticipated. Unlike the negative KEYNOTE-412 (IO added to definitive CRT), the perioperative strategy in resected disease proved effective — likely because neoadjuvant priming and adjuvant maintenance without concurrent chemotherapy enable a more favorable immune environment. PD-L1 testing will be critical for patient selection.