Aim To evaluate the protocol-specified final analysis of overall survival (OS) in the KEYNOTE-002 study (NCT01704287) of pembrolizumab versus chemotherapy in patients with ipilimumab-refractory, advanced melanoma. Methods In this randomised, phase II study, eligible patients had advanced melanoma with documented progression after two or more ipilimumab doses, previous BRAF or MEK inhibitor or both, if BRAFV600 mutant-positive. Patients were randomised to pembrolizumab 2 mg/kg or 10 mg/kg every 3 weeks or investigator-choice chemotherapy. Crossover to pembrolizumab was allowed following progression on chemotherapy. The protocol-specified final OS was performed in the intent-to-treat population. Survival was positive if p < 0.01 in one pembrolizumab arm. Results A total of 180 patients were randomised to pembrolizumab 2 mg/kg, 181 to pembrolizumab 10 mg/kg and 179 to chemotherapy. At a median follow-up of 28 months (range 24.1–35.5), 368 patients died and 98 (55%) crossed over to pembrolizumab. Pembrolizumab 2 mg/kg (hazard ratio [HR] 0.86, 95% confidence interval [CI] 0.67–1.10, p = 0.117) and 10 mg/kg (0.74, 0.57–0.96, p = 0.011) resulted in a non-statistically significant improvement in OS versus chemotherapy; median OS was 13.4 (95% CI 11.0–16.4) and 14.7 (95% CI 11.3–19.5), respectively, versus 11.0 months (95% CI 8.9–13.8), with limited improvement after censoring for crossover. Two-year survival rates were 36% and 38%, versus 30%. Progression-free survival, objective response rate and duration of response improved with pembrolizumab versus chemotherapy, regardless of dose. Grade III–V treatment-related adverse events occurred in 24 (13.5%), 30 (16.8%) and 45 (26.3%) patients, respectively. Conclusion Improvement in OS with pembrolizumab was not statistically significant at either dose versus chemotherapy.
- Programmed cell death-1