ComPAS: A Bayesian drug combination platform trial design with adaptive shrinkage

Area of Science:

• Pharmacology
• Biostatistics
• Clinical Trial Design

Background:

• Drug combination therapies offer synergistic effects and overcome monotherapy resistance.
• Evaluating numerous potential drug combinations is complex, especially with evolving compounds.
• Existing trial designs struggle with flexibility and efficiency in adaptive combination studies.

Purpose of the Study:

• To propose a flexible Bayesian platform design for drug combination trials named ComPAS (Combination Platform with Adaptive Shrinkage).
• To develop an adaptive shrinkage method for efficient information borrowing across treatment combinations.
• To enable dropping futile combinations, graduating effective ones, and adding new ones during a trial.

Main Methods:

• Utilized a Bayesian framework for flexible platform trial design.
• Developed an adaptive shrinkage method for statistical analysis.
• Employed Bayesian model selection and hierarchical models for identifying efficacious combinations.
• Conducted simulation studies to compare ComPAS with existing designs.

Main Results:

• ComPAS demonstrated higher probability in identifying effective drug combinations compared to existing designs.
• The adaptive shrinkage method efficiently borrowed information across combinations.
• The platform design allows for seamless addition and removal of treatment arms.

Conclusions:

• ComPAS offers an efficient and flexible approach to accelerate drug development through adaptive combination trials.
• The proposed methodology enhances the identification of synergistic and efficacious drug combinations.
• This platform design facilitates timely and seamless progression of promising therapeutic strategies.