Framing the statistical criteria for accelerated approval of oncology drugs: A pathway for front runners

FDA has recently proposed a one-trial approach in Project FrontRunner for both accelerated approval and regular approval in the same trial. The goal of this short communication is to make the regulatory criteria for decision-making on accelerated approval based on the interim analysis more explicit. Two related high-level statistical issues are addressed under simplified yet representative scenarios. The first is at what Type I error level should the interim analysis be tested? To conform with the conventional two-trial approach, it may be tested at the 1.0% level without paying penalty for regular approval. This criterion is more relaxed than current practice. The second is what may be considered as adequate totality of evidence for accelerated approval. It turns out that, with the first criterion met, the observed treatment effect of the clinical endpoint intended for regular approval needs to be close to the target effect for the study. The criterion helps make the regulatory decision more transparent and objective.