Using Bayesian modeling in frequentist adaptive enrichment designs

Area of Science:

• Clinical Trials
• Biostatistics
• Pharmacology

Background:

• Disease mechanistic heterogeneity necessitates targeted therapeutics.
• Identifying the precise patient subgroup (target population) for targeted drugs is challenging due to pathway complexity and assay limitations.
• Adaptive enrichment trials dynamically learn and update enrollment criteria to identify the target population.

Purpose of the Study:

• To propose a novel framework for group-sequential adaptive enrichment clinical trials.
• To integrate Bayesian methods for adaptive decision-making and frequentist methods for hypothesis testing.
• To enhance the characterization of the target population and estimation of treatment effect size.

Main Methods:

• Development of a group-sequential adaptive enrichment trial framework.
• Application of Bayesian methods for optimizing enrollment decisions and characterizing the target population.
• Incorporation of a frequentist hypothesis test for final trial analysis, building upon Simon & Simon (2013).

Main Results:

• The proposed framework combines Bayesian decision-making with frequentist hypothesis testing.
• Bayesian methods facilitate adaptive enrollment criteria refinement and target population characterization.
• The design preserves the studywise probability of a type I error through a formal hypothesis test.

Conclusions:

• The joint frequentist/Bayesian design offers a powerful approach for adaptive enrichment trials.
• This framework improves the efficiency and accuracy of identifying patient populations likely to benefit from targeted therapies.
• It balances adaptive learning with robust statistical inference for clinical trial evaluation.