Bayesian sample size determination for phase IIA clinical trials using historical data and semi-parametric prior's...

Area of Science:

• Biostatistics
• Clinical Trial Design
• Pharmacoeconomics

Background:

• Simon's two-stage design is standard for phase IIA trials, optimizing sample size.
• High uncertainty in pre-trial response rate beliefs necessitates alternative approaches.
• Bayesian methods naturally incorporate distributional information, addressing uncertainty.

Purpose of the Study:

• To explore Bayesian counterparts of Simon's two-stage design.
• To develop methods for constructing prior distributions in Bayesian clinical trial designs.
• To evaluate the impact of different priors on sample size determination.

Main Methods:

• Utilized the predictive version of the single threshold design.
• Specified distinct analysis and design prior distributions.
• Derived prior distributions using linear combinations of B-splines, moving beyond traditional beta priors.
• Applied to a phase IIA trial for an anti-HER2 DNA vaccine in breast cancer.

Main Results:

• Demonstrated a method for constructing clinical priors from summarized data, accommodating high uncertainty.
• Evaluated the influence of various prior distributions on sample size calculations.
• Provided a flexible Bayesian framework for two-stage trial designs.

Conclusions:

• Bayesian two-stage designs offer a robust alternative to Simon's design when prior response rate uncertainty is high.
• The B-spline approach provides a flexible way to define priors in Bayesian clinical trial planning.
• This methodology enhances the precision of sample size determination in early-phase oncology trials.