Design considerations for two-stage enrichment clinical trials

Area of Science:

• Biostatistics
• Clinical Trial Design
• Pharmacogenomics

Background:

• Predictive biomarkers enable enrichment strategies to focus clinical trials on patient subpopulations likely to benefit.
• Two-stage enrichment designs involve an initial stage for threshold estimation and a subsequent stage for a pivotal trial.

Purpose of the Study:

• To explore critical design issues in two-stage enrichment trials, including sample size determination for both stages.
• To investigate the impact of re-estimating the Stage 2 sample size after Stage 1.
• To understand how the predictive nature of a biomarker influences sample size requirements.

Main Methods:

• The study proposes a two-stage enrichment design assuming a continuous outcome and continuous biomarker, modeled using a bivariate normal distribution.
• Analysis focuses on the correlation coefficient between outcome and biomarker to understand design behavior.
• Simulations are used to illustrate the impact of model misspecification and threshold estimation accuracy.

Main Results:

• Failure to accurately estimate the biomarker threshold in Stage 1 can lead to significant negative consequences in Stage 2.
• Sample size requirements are directly dependent on the predictive strength of the biomarker.
• Model misspecification and poor threshold estimation can compromise the integrity of the enrichment trial.

Conclusions:

• The proposed two-stage enrichment design provides a framework for efficiently identifying and testing treatments in biomarker-defined subpopulations.
• Accurate threshold estimation is paramount for the success of enrichment trials, influencing both statistical power and resource allocation.
• Understanding the interplay between biomarker characteristics and trial design is essential for optimizing clinical trial development.