An adaptive two-stage design with treatment selection using the conditional error function approach.

Area of Science:

• Clinical Trial Design
• Biostatistics
• Pharmaceutical Research

Background:

• Combining Phase II (dose-finding) and Phase III (pivotal) trials is crucial for efficient drug development.
• Existing adaptive designs may not optimally integrate treatment selection with significance testing.
• Controlling Type I error in adaptive designs with treatment selection requires careful statistical methodology.

Purpose of the Study:

• To propose a two-stage adaptive design that integrates dose-finding and pivotal trial phases.
• To control the Type I error, specifically the probability of selecting an indifferent treatment and claiming significance.
• To evaluate the design's statistical properties, including expected sample size and power.

Main Methods:

• A two-stage adaptive design is proposed, selecting the best treatment in Stage 1 and testing its significance in Stage 2.
• The design utilizes a conditional error function to adjust Stage 2 error rates based on Stage 1 data.
• Properties like expected sample size and Stage 2 power are examined under various hypotheses.
• A method for optimizing the conditional error function is presented.

Main Results:

• The proposed adaptive design effectively combines trial phases while controlling Type I error.
• Analysis of expected sample size and Stage 2 power demonstrates the design's performance.
• Optimal conditional error functions were derived for specific hypothesis configurations.
• The approach was illustrated using a hypothetical clinical trial example.

Conclusions:

• The developed two-stage adaptive design offers an efficient strategy for clinical trials.
• This methodology provides robust control of statistical errors in adaptive trial settings with treatment selection.
• The design facilitates improved decision-making in drug development by integrating early and late-phase trial objectives.