Adaptive Designs for Clinical Trials

Abstract

andomized clinical trials serve as the standard for clinical research and have contributed immensely to advances in patient care.Nevertheless, several shortcomings of randomized clinical trials have been noted, including the need for a large sample size and long study duration, the lack of power to evaluate efficacy overall or in important subgroups, and cost.These and other limitations have been widely acknowledged as limiting medical innovation. 1 Adaptive trial design has been proposed as a means to increase the efficiency of randomized clinical trials, potentially benefiting trial participants and future patients while reducing costs and enhancing the likelihood of finding a true benefit, if one exists, of the therapy being studied. 2 Adaptive designs are applicable to both exploratory and confirmatory clinical trials.Adaptive designs for exploratory clinical trials deal mainly with finding safe and effective doses or with dose-response modeling.The emphasis is on strategies that will assign a larger proportion of the participants to treatment groups that are performing well, reduce the number of participants in treatment groups that are performing poorly, and investigate a dose range that is larger than ranges in corresponding trials with nonadaptive designs, in order to select effective doses for the confirmatory stage of investigation.Control of the type I error rate is less of an issue.In Table 1, various types of adaptive designs for exploratory clinical trials are classified into categories that reflect the time sequence in which they would be performed in the drug-development process.In confirmatory trials, the adaptive nomenclature refers to making prospectively planned changes to the future course of an ongoing trial on the basis of an analysis of accumulating data from the trial itself, in a fully blinded or unblinded manner, without undermining the statistical validity of the conclusions. 3However, modifications of randomized clinical trials that are performed in an unblinded manner are subject to closer regulatory scrutiny than those performed in a blinded manner.They require careful attention to statistical techniques and operational procedures to ensure that the implementation is scientific, ethical, and free from bias.In Table 1, different types of adaptations for confirmatory trials are classified into four major categories -seamless phase 2-3 designs, sample-size reestimation, group sequential designs, and population-enrichment designs -and the strengths and weaknesses of each type are identified in relation to corresponding nonadaptive designs.There is some overlap among the different categories.For example, sample-size reestimation could be implemented on its own or incorporated into group sequential, dose-selection, or population-enrichment designs.In this review, we focus on adaptive designs of confirmatory clinical trials.We discuss the benefits and limitations of such designs, using four case studies that highlight the statistical and operational considerations that are the prerequisites