Point estimation following two-stage adaptive threshold enrichment clinical trials

Area of Science:

• Biostatistics
• Clinical Trial Design
• Translational Medicine

Background:

• Biomarker-based subpopulations are increasingly used in clinical trials.
• Statistical methods often prioritize type I error and power for these designs.
• Estimating treatment effects in selected subpopulations presents unique statistical challenges.

Purpose of the Study:

• To develop and evaluate point estimators for treatment effects in two-stage adaptive enrichment threshold designs.
• To address the statistical complexities of estimating treatment effects in biomarker-defined subpopulations.
• To provide robust estimation strategies for adaptive clinical trial designs.

Main Methods:

• Development of unbiased and shrinkage estimators for treatment effects.
• Derivation of estimators that estimate and subtract bias from naive estimates.
• Simulation studies to compare estimator performance based on bias and mean squared error.
• Proposal of a two-stage adaptive enrichment threshold design framework.

Main Results:

• Several novel point estimators were derived for adaptive enrichment trials.
• One unbiased estimator was recommended, but no single estimator dominated all simulation scenarios.
• Performance varied based on bias and mean squared error across different scenarios.
• A hybrid estimator strategy, dependent on the selected subpopulation, was suggested.

Conclusions:

• The developed estimators offer improved treatment effect estimation in adaptive enrichment trials.
• The choice of estimator may depend on the specific subpopulation identified.
• A pre-trial simulation study is recommended to select the optimal hybrid estimator strategy.
• These methods contribute to more precise and efficient clinical trial designs using predictive biomarkers.