Sample size re-estimation in Phase 2 Dose-Finding: Conditional power vs. Bayesian predictive power

TL;DR

This paper compares frequentist and Bayesian unblinded sample size re-estimation methods in Phase II dose-finding trials, demonstrating their advantages and error control through simulation studies.

Contribution

It introduces two-stage SSR designs using conditional and Bayesian predictive power, with rigorous type I error control in multi-arm trials.

Findings

Bayesian SSR effectively incorporates prior knowledge.

Unblinded SSR improves trial efficiency in multi-arm settings.

Type I error is rigorously controlled in proposed designs.

Abstract

Unblinded sample size re-estimation (SSR) is often planned in a clinical trial when there is large uncertainty about the true treatment effect. For Proof-of Concept (PoC) in a Phase II dose finding study, contrast test can be adopted to leverage information from all treatment groups. In this article, we propose two-stage SSR designs using frequentist conditional power and Bayesian posterior predictive power for both single and multiple contrast tests. The Bayesian SSR can be implemented under a wide range of prior settings to incorporate different prior knowledge. Taking the adaptivity into account, all type I errors of final analysis in this paper are rigorously protected. Simulation studies are carried out to demonstrate the advantages of unblinded SSR in multi-arm trials.