All Times EDT
Keywords: Randomized Phase II Trial, Quantitative Decision Making, Predictive Probability of Success
In oncology Phase II clinical trials, as overall survival (OS) becomes gold standard endpoint for evaluating clinical efficacy, many Phase II trials use OS as primary endpoint. In general, for a classic randomized Phase II trial in oncology, with a concurrent control arm, the trial is usually designed to incorporate a one-sided type I error rate of 5% - 10% and a type II error rate of 10% - 20%. A trial will be considered success if the p-value of the test for OS (typically log-rank test) is less than the prespecified type I error rate. Unlike classic randomized Phase II trial using p-value as success criteria, we calculated predictive probability of success (PPOS) for a hypothetical future Phase III trial and use PPOS as the guidance criteria to decide whether the experimental agent can proceed to Phase 3 for further development. The PPOS based on Bayesian framework is calculated using the posterior predictive distribution of future Phase III log-transformed hazard ratio (log(HR)), given the observed log(HR) from Phase II, assuming that the normal approximation is valid for both the observed log(HR) and the prior for mean of log(HR). If the PPOS is greater than a pre-specified cutoff value, e.g. 50% or other values, then the experimental regimen will be considered for further development.