ABSTRACT

This chapter suggests assume that a confirmatory trial is designed for making joint inference on the biomarker-positive (BM+) subpopulation and the overall population. If it is known that there is no treatment effect in the biomarker negative subgroup, then a confirmatory trial should be carried out in the BM+ subgroup of patients only. Ultimately, the sponsor is responsible for adequately powering the trial for the population(s) of interest. There are several instances in the pharmaceutical industry, where ignoring the biomarker effect has led to a failure to reject the null hypothesis of no treatment effect in the overall population. The reason is that subjects from the whole population are too heterogeneous in their drug response. At times, both the treatment effects in the BM+ subgroup and overall population are of interest to test as co-primary endpoints.