Accounting for Biomarker-Based Subpopulation Selection in Phase II Trials in POS Computations

• Activity Number: 606
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Guochen Song*+ and Ilya Lipkovich and Alex Dmitrienko
• Companies: Quintiles and Quintiles and Quintiles
• Keywords: subgroup ; probability of success ; Bayesian ; multiplicity ; biomarker

Abstract

The goal of this research is to simultaneously select a "promising subgroup" in a phase 2 trial from a relatively small number of candidate subgroups (e.g. each based on a single biomarker selected from p=5-10 candidate binary biomarkers) and to assess the probability of success (POS) in a phase 3 trial based on selected subgroup while properly adjusting the subgroup effect sizes for multiplicity inherent in subgroup selection. Our approach combines Bayesian and frequentist ideas. First the selection bias is estimated via resampling data under the null hypothesis of no treatment effect across all subgroups (in excess of the treatment effect in overall population) . The prior for the selected subgroup then is adjusted using this estimated bias by the amount that ensures that the posterior for the subgroup treatment effect under the null would be centered at the treatment effect observed for the overall population. The subgroup effects indentified in the phase 2 trial are then integrated (via adjusted priors) into POS computation for a future Phase 3 trial.