The original study sample size of 750 patients was determined on the basis of the primary end point of best overall response rate but was revised with the new primary end point of overall survival. We estimated that with 385 events (deaths) among a total of 500 patients randomly assigned to the ipilimumab-plus-gp100 and the gp100-alone groups, the study would have at least 90% power to detect a difference in overall survival, at a two-sided alpha level of 0.05, with the use of a log-rank test. A total of 481 events were required in all three groups (assuming that the events were distributed in a 3:1:1 ratio in the ipilimumab-plus-gp100, ipilimumab-alone, and gp100-alone groups, respectively). Therefore, all patients who were randomly assigned in the study were to be followed until at least 481 events had occurred in the study. Enrollment was completed on July 25, 2008, when more than 650 patients had been enrolled. A post hoc power analysis showed that the 219 events observed among a total of 273 patients randomly assigned to the ipilimumab-alone and gp100-alone groups provided at least 80% power to detect a difference in overall survival between the two groups, at a two-sided alpha level of 0.05, with the assumption that ipilimumab alone has the same treatment effect as the combination regimen of ipilimumab plus gp100.
Survival was defined as the time from randomization to death from any cause, and progression-free survival as the time from randomization to documented disease progression or death. Event-time distributions were estimated with the use of the Kaplan–Meier method. Cox proportional-hazards models, stratified according to metastasis status and receipt or nonreceipt of previous interleukin therapy, were used to estimate hazard ratios and to test for significance of the timing of events. All reported P values are two-sided, and confidence intervals are at the 95% level. Survival rates were based on Kaplan–Meier estimation, and confidence intervals were calculated with the use of the bootstrap method. Descriptive statistics were used for adverse events.
Survival was defined as the time from randomization to death from any cause, and progression-free survival as the time from randomization to documented disease progression or death. Event-time distributions were estimated with the use of the Kaplan–Meier method. Cox proportional-hazards models, stratified according to metastasis status and receipt or nonreceipt of previous interleukin therapy, were used to estimate hazard ratios and to test for significance of the timing of events. All reported P values are two-sided, and confidence intervals are at the 95% level. Survival rates were based on Kaplan–Meier estimation, and confidence intervals were calculated with the use of the bootstrap method. Descriptive statistics were used for adverse events.
