Dose-Finding Using Hierarchical Modeling for Multiple Subgroups

• Activity Number: 633
• Type: Topic Contributed
• Date/Time: Thursday, August 13, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Kristen Cunanan* and Joe Koopmeiners
• Companies: University of Minnesota and University of Minnesota
• Keywords: Bayesian Adaptive Design ; Dose-Finding Study ; Hierarchical Modeling

Abstract

The primary goal of Phase I clinical trials is to determine a new treatment's highest dosage with an acceptable toxicity rate, defined as the maximum tolerated dose (MTD), via a dose-finding study. In oncology trials, often we have a heterogeneous population, i.e. subpopulations, defined by different standards of care or tumor histologies. Given the success of hierarchical modeling and the success of the continual reassessment method, we consider a design combining the two methods to facilitate borrowing of strength across multiple subgroups in a dose escalation study. We propose a Phase I Bayesian design that shares dose-response information across subgroups to improve and quicken dose finding within a subgroup, while allowing the flexibility to drop subgroups if all doses are overly toxic. Traditionally, patients are enrolled in cohorts and treated at the updated MTD. However to account for staggered enrollment between subgroups, we propose multiple approaches for dose-escalation. In a simulation study, we investigate three dose-response hierarchical models. For comparison, we investigate the three models' performance when each subgroup's dose-response is modeled independently.