Adjuvant Pembrolizumab versus IFNα2b or Ipilimumab in Resected High-Risk Melanoma

Abstract

We conducted a randomized phase III trial to evaluate whether adjuvant pembrolizumab for one year (647 patients) improved recurrence-free survival (RFS) or overall survival (OS) in comparison with high-dose IFNα-2b for one year or ipilimumab for up to three years (654 patients), the approved standard-of-care adjuvant immunotherapies at the time of enrollment for patients with high-risk resected melanoma. At a median follow-up of 47.5 months, pembrolizumab was associated with significantly longer RFS than prior standard-of-care adjuvant immunotherapies [HR, 0.77; 99.62% confidence interval (CI), 0.59-0.99; P = 0.002]. There was no statistically significant association with OS among all patients (HR, 0.82; 96.3% CI, 0.61-1.09; P = 0.15). Proportions of treatment-related adverse events of grades 3 to 5 were 19.5% with pembrolizumab, 71.2% with IFNα-2b, and 49.2% with ipilimumab. Therefore, adjuvant pembrolizumab significantly improved RFS but not OS compared with the prior standard-of-care immunotherapies for patients with high-risk resected melanoma.

Significance: Adjuvant PD-1 blockade therapy decreases the rates of recurrence, but not survival, in patients with surgically resectable melanoma, substituting the prior standard-of-care immunotherapies for this cancer. See related commentary by Smithy and Shoushtari, p. 599. This article is highlighted in the In This Issue feature, p. 587.

Figure 1.. Enrollment, randomization, and follow-up.

All eligible patients who were randomized were included in the intention-to-treat population. The safety population included all patients who received at least one dose of treatment that they were randomly assigned. In total, 24 patients in the pembrolizumab group and 20 patients in the standard of care group were found to have major eligibility violations and were excluded from analyses.

Figure 2.. Kaplan-Meier estimates of main time-to event endpoints.

A) recurrence-free survival (as assessed by local investigators), B) OS in the intention-to-treat population, and C) exploratory analysis of post-recurrence OS. Cox regression models were stratified by randomization stratification factors: PD-L1 status, intended standard of care regimen choice, and stage of disease. Hazard ratios report pembrolizumab versus standard of care (reference). In the intention-to-treat analysis for RFS there were 524 events (252 in the pembrolizumab group, and 272 in the control group). The HR for recurrence was 0.77 (99.62% confidence interval [CI], 0.59 to 0.99). OS analysis was completed at the protocol specified time of 3.5 years with 57% of events. There was not a statistically significant difference between the pembrolizumab group and the control group (0.82, 96.3% CI 0.61, 1.09). The post-recurrence analysis of OS was not a protocol-specified endpoint; it is included to provide information to evaluate post-recurrence outcomes in both study groups.

Figure 3.. Forest plot for recurrence-free survival according to subgroup in the overall population.

An unstratified Cox regression model was used to estimate the hazard ratios of recurrence or death in the pembrolizumab group as compared to the standard of care group among all the patients. A stratified Cox regression model stratified by the randomization stratification factors (PD-L1 status, control arm choice, and stage of disease) is reported for the overall population. 95% confidence intervals are presented.