|
Subgroup identification procedures are increasingly becoming part of clinical development programs. While subgroup identification helps better tailor approved drugs to existing populations, it is often used within a seamless phase2/3 program or a single phase 2 trial with an interim look to identify a subgroup of patients with maximum benefit and minimal risk from a study drug for a reasonably large subpopulation. As such groups are typically unknown at the start of the study, existing data-driven techniques for subgroup identification should be used and various methods evaluated.
We are proposing a procedure for a two-stage design that allows for evaluating and choosing a subgroup defined by one or two out of several candidate biomarkers at the interim analysis using a pre-defined utility function that may incorporate trade-offs for efficacy, safety and population size. The procedure allows for other decisions based on expected utility at the interim including stopping for early efficacy or futility. We conducted a simulation study that evaluates operating characteristics of the procedure under various choices of utility and subgroup identification procedures.
|
