Application of Bayesian predictive probability for interim futility analysis in single-arm phase II trial

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Bayesian Statistics

Background:

• Bayesian predictive probability design is gaining traction for phase II trials due to its innovative approach.
• Current methods can be complex, limiting accessibility for biostatisticians and clinicians.
• There is a need for streamlined tools to integrate advanced Bayesian designs into clinical practice.

Purpose of the Study:

• To enhance the accessibility of Bayesian predictive probability designs for phase II clinical trials.
• To develop an R package that simplifies the creation of statistical plans for Bayesian trial designs.
• To facilitate the integration of Bayesian methods by biostatisticians and clinicians.

Main Methods:

• Utilized a Bayesian framework incorporating posterior and predictive probabilities.
• Developed an R package, 'BayesianPredictiveFutility', with a graphical interface.
• The package calculates the minimum responders needed for efficacy and evaluates future success at interim stages.

Main Results:

• An R package, 'BayesianPredictiveFutility', was created, generating comprehensive statistical plans.
• The package provides detailed outputs including stopping boundaries, performance metrics (PET, type I error, power), and sensitivity analyses.
• Demonstrated utility through application in two lung cancer clinical trials.

Conclusions:

• The Bayesian predictive probability method offers a flexible clinical trial design.
• The developed statistical tool enhances the practical application of Bayesian designs.
• This approach adds significant value by broadening the use of advanced statistical methodologies in clinical research.