Screening designs for drug development

Area of Science:

• Biostatistics
• Clinical Trial Design
• Pharmacology

Background:

• Phase II clinical trials are critical for drug development, evaluating treatment efficacy and safety.
• Current screening designs may not optimally handle the dynamic nature of treatment evaluation, including the emergence of new therapies and the discontinuation of others.
• A need exists for adaptive and statistically rigorous methods to manage multiple treatment arms in early-phase drug screening.

Purpose of the Study:

• To propose novel drug screening designs utilizing a Bayesian decision-theoretic framework.
• To facilitate simultaneous evaluation and adaptive management of multiple treatments in phase II studies.
• To provide a structured approach for making terminal decisions on treatment progression or abandonment.

Main Methods:

• Employs a Bayesian decision-theoretic approach for sequential stopping decisions.
• Utilizes decision boundaries in the space of marginal posterior moments for treatment evaluation.
• Implements a Monte Carlo simulation algorithm for practical application.

Main Results:

• The proposed method allows for dynamic adjustments, including the addition of new treatments and removal of underperforming ones.
• Provides a clear framework for making go/no-go decisions for treatments entering phase III studies.
• Demonstrates the feasibility of the approach through simulation.

Conclusions:

• The Bayesian decision-theoretic approach offers a robust framework for optimizing phase II drug screening.
• The adaptive design enhances efficiency by enabling timely decisions on multiple treatment candidates.
• An R library is available to facilitate the implementation of this advanced methodology.