An adaptive power prior for sequential clinical trials - Application to bridging studies

Area of Science:

• Biostatistics
• Clinical Trial Design
• Pharmacometrics

Background:

• Drug development often involves leveraging existing clinical trial data from different populations.
• Efficiently utilizing historical data can optimize resource allocation and trial timelines.
• Bridging studies, particularly between Caucasian and Asian populations, require robust statistical methods for data integration.

Purpose of the Study:

• To develop and evaluate an adaptive power prior method for incorporating historical or external information in clinical trials.
• To apply this method to bridging studies, specifically focusing on sequential adaptive allocation designs.
• To provide a flexible approach for borrowing information based on data conflict or similarity.

Main Methods:

• Developed an adaptive power prior with a commensurability parameter for flexible information borrowing.
• Employed a two-step weighting process: effective sample size for maximum information sharing and a commensurability parameter based on distribution distance.
• Focused on sequential adaptive allocation designs for bridging studies between Caucasian and Asian populations.

Main Results:

• The proposed adaptive power prior method allows for full, no, or tuned borrowing of historical data based on observed conflicts.
• The method addresses elicitation and computational challenges associated with traditional Empirical Bayes approaches.
• An extensive simulation study evaluated the robustness and sensitivity of the proposed method to prior choices.

Conclusions:

• The adaptive power prior offers a flexible and efficient statistical framework for utilizing historical data in drug evaluation.
• This method is particularly relevant for bridging studies, enhancing data integration between diverse populations.
• The proposed approach provides a computationally feasible alternative to existing methods for handling external information.