|
Abstract:
|
In recently reported immune-oncology (IO) studies, the conventional RECIST based ORR and PFS do not appear to be well correlated with OS. The objective was to further evaluate correlation between treatment effects in PFS with OS (and ORR with OS), based on recently reported PD1/PDL1 randomized trials. We applied weighted linear regression to different treatment effect estimates reported for these studies. Spearman's rho are used to quantify the linear association. In addition, correlation between the three endpoints, using patient level data, were also evaluated using Bayesian joint modeling approach. We found PFS rate at 6 month is positively correlated with OS rate at 12 month; however, delta PFS6 is not well correlated with delta OS12 or OS HR. A modest trial level correlation was observed between PFS and OS HR in the overall patient population, compared to a strong correlation at the patient level. A stronger positive association was observed in the PD-L1 positive subgroup. No meaningful difference in association between ORR and PFS or ORR and OS was observed between the two populations, with a very weak correlation of ORR with OS.
|
