Statistical considerations for the next generation of clinical trials

Abstract

In this article, we address some of the statistical issues associated with the next generation of oncology clinical trials. Specifically, we focus on two critical aspects of oncologic clinical trials: study endpoints and study design. In our discussion of study endpoints, we provide an overview of endpoints relevant to each phase of clinical trials (phases I, II, and III) and discuss some of the trends in selecting endpoints in recent years. For phase I designs, the traditional assumption that increasing dose will always lead to increasing efficacy, appropriate for cytotoxic agents, is not applicable to novel therapeutics such as in colorectal carcinoma. We emphasize the role of surrogate endpoints in modern clinical trials. In our discussion of study design, we are particularly interested in the essential role of randomization, and discuss recently developed randomized phase II designs. We consider the role biomarkers play in the design of clinical trials and introduce study designs for biomarker evaluations. Finally, we briefly discuss the application of adaptive designs in clinical trials.