Phase III clinical trial designs incorporating predictive biomarkers: An overview

Abstract

Advances in biotechnology have revolutionized clinical trials in oncology, shifting the emphasis to co-development of molecularly targeted drugs and companion predictive biomarkers. However, the difficulty in developing and validating biomarkers in the early phases of clinical development complicates the design and analysis of definitive phase III trials that aim to establish the clinical utility of new treatments with the aid of predictive markers. This chapter provides an overview of several designs for phase III trials that incorporate predictive markers at various levels of development and credibility, the latter in terms of these markers' abilities to predict treatment responsiveness at the initiation of phase III trials. We first discuss the enrichment design and marker-stratified all-comers designs with a single binary marker. For the marker-stratified designs, multi-stage analyses for sequential testing across the subgroups and adaptive subgroup selection are provided. We also discuss other adaptive designs, including the adaptive threshold design and the adaptive signature design with some variants, in cases where the threshold for marker positivity is unclear or a single marker for use in evaluating treatment efficacy is not available at the initiation of phase III trials. Lastly, we introduce the prospective-retrospective approach that allows for the evaluation of treatment efficacy in a marker subgroup based on external evidence.