Statistical methods for seamless phase II/III design

Abstract

The seamless phase II/III design combining phases II and III into a single trial has been shown growing interest for improving the efficiency of drug development, becoming the most frequent adaptive design type. It typically consists of two stages, the trial objectives being often different in each stage. The primary objectives are to select optimal experimental treatment group(s) in the first stage and compare the efficacy between the selected treatment and control groups in the second stage. In the final analysis, appropriate statistical methods should be applied to avoid increasing the type I error rate and selection bias. This paper reviews several statistical methods that can be applied to adaptive seamless phase II/III designs. Especially, the most representative methods such as the group sequential design approach, combination test approach, and conditional error function approach, are organized in a unified framework and the characteristics of the methods are compared. As related topics, statistical methods for the case where endpoints are different between stages 1 and 2, and the method of sample size calculation will also be discussed. Examples of actual applications are also presented.