Bridging the Gap Between Preclinical and Clinical Evaluation of Therapeutic Candidates

Session 5

Innovative Statistical Designs for Clinical Trials

Donald A. Berry, Ph.D.
University of Texas MD Anderson Cancer Center

Adaptive clinical trials have been changing the face of drug and medical device development. The idea is that various aspects of the trial’s conduct can be modified depending on the accumulating results. The oldest type of adaptation is stopping a trial early for efficacy or futility. More recent adaptations include: extending the trial’s accrual (or not); assigning doses to maximize information about dose-response (including dropping or adding doses); seamless phases in which a trial morphs from phase I into phase II or phase II into phase III without stopping or pausing accrual; adaptively randomizing to treat patients in the trial effectively (while preserving the trial’s scientific aspects); homing in on responding populations (such as based on biomarkers); and ramping up accrual (or not), possibly in a seamless-phase fashion. The goals include more efficient, more cost-effective, and more rapid clinical development. Modern adaptive designs have been or are being used in practically every type of drug indication: from oncology to migraine, from lupus to stroke, from diabetes to spinal cord injury. I will describe how the Bayesian approach to statistics is being used to build efficient adaptive designs, with examples from some of these indications.

Up to Top