Group-sequential methods for adaptive seamless phase II/III clinical trials

Abstract

The last 10 years have seen considerable interest in clinical trial designs that allow the seamless combination of Phases II and III in a single clinical trial. Such designs bring together the selection of the most promising of a number of treatments, as usually performed in a Phase II clinical trial, with the rigorous analysis and control of type I error rates required for a Phase III clinical trial. Statistically, the challenge is that of allowing for multiple comparisons arising both from repeated analyses of the accumulating data and from selection of a treatment on the basis of data that will be included in the final analysis. This challenge can be met in a number of different ways. This paper reviews methods based on the group-sequential methodology for monitoring of sequential clinical trials. The main focus of the paper will be a description of the methodology, including the setting in which short-term data are used for decision making at an early interim analysis.