Assessing efficacy in important subgroups in confirmatory trials: An example using Bayesian dynamic borrowing.

Area of Science:

• Biostatistics
• Clinical Trials
• Pharmacometrics

Background:

• Traditional subgroup analyses in confirmatory trials often analyze subgroups independently, neglecting valuable data from complementary groups.
• This approach can lead to reduced statistical power and potentially missed efficacy signals within specific patient populations.

Purpose of the Study:

• To introduce and illustrate Bayesian dynamic borrowing, a novel statistical method for assessing subgroup efficacy.
• To demonstrate how this method utilizes information from complementary subgroups to enhance the analysis of the subgroup of interest.

Main Methods:

• Bayesian dynamic borrowing constructs a robust mixture prior by combining an informative prior from a complementary subgroup with a weak, non-informative prior.
• The method dynamically adjusts the influence of the complementary subgroup's data based on its consistency with the subgroup being studied.
• Tipping point analysis is incorporated to determine the necessary prior weight for establishing subgroup efficacy and to visualize evidence.

Main Results:

• The Bayesian dynamic borrowing approach allows for more efficient use of available data compared to traditional separate subgroup analyses.
• Tipping point analysis provides a clear method for quantifying the required evidence from complementary subgroups to demonstrate efficacy.
• The method was successfully illustrated in a severe asthma trial, assessing adolescent subgroup efficacy using adult data.

Conclusions:

• Bayesian dynamic borrowing offers a statistically robust and transparent framework for subgroup efficacy assessment in clinical trials.
• This method enhances power by intelligently borrowing information from related data, leading to more reliable conclusions.
• The integration of tipping point analysis further strengthens the interpretability and trustworthiness of subgroup findings.