Adaptive Group Sequential Designs with Population Enrichment and Endpoint Assessment in Phase 3 Randomized Controlled Trials

Failure rates of phase 3 randomized clinical trials are estimated to be between 50 to 60 percent. The failure rates include studies which failed to show expected benefits, stopped because of unexpected harm, etc. It is claimed that the costs associated with new drug development and obtaining approval for marketing ranges on average from 802 million US dollars to 1.7 billion US dollars. When a treatment has a substantial differential effect on subgroups, a higher than expected efficacy in one subgroup and a lower than expected in another may show on average lower than expected effect size in the overall population, and consequently may lead to an unsuccessful trial. The importance of finding new strategies for decreasing heterogeneity in study populations has already been recognized by the US Food and Drug Administration (FDA). FDA encouraged more research on population enrichment methods. Enrichment refers to use of patients characteristics such as age, sex, diabetes , etc, to select a population where the treatment has a higher chance of showing a beneficial effect compared to that of the unselected populations.^ Use of multiple endpoints or co-primary endpoints in randomized trials to assess an intervention’s multidimensional effect is not uncommon. This allows the investigators to capture an experimental intervention’s multidimensional effect more comprehensively. There are a few diseases for which the regulatory agencies require demonstration of a treatment’s statistically significant benefit on multiple endpoints before accepting the treatment efficacy. For example, for approval of a treatment for Alzheimer\u27s disease, the regulators recommend that a treatment has to show benefit on two endpoints: Alzheimer\u27s Disease Assessment Scale-cognitive subscale (ADAS- cog) and Clinician\u27s Interview-Based Impression of Change (CIBIC).^ The primary focus of this research is on the implication of heterogeneity of treatment effects in a study population, evaluation of two co-primary endpoints, and use of population enrichment in the success of a phase 3 confirmatory trial. We proposed two population enrichment designs, provided the operating characteristics of these designs, and demonstrated how the designs can be used using selected data from The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT). We also commented on how implementation of such designs can be challenging when the outcomes of interest are delayed outcomes.