A Bayesian dose-finding design for drug combination clinical trials based on the logistic model

Area of Science:

• Oncology
• Biostatistics
• Clinical Trial Design

Background:

• Determining maximum tolerated dose (MTD) in early-phase cancer trials is crucial.
• Drug-combination trials present unique challenges due to drug-drug interactions.
• Existing dose-finding designs often use complex statistical models unfamiliar to clinicians.

Purpose of the Study:

• To propose a simplified Bayesian dose-finding design for cancer drug-combination trials.
• To improve the practical application of dose-finding methodologies in clinical settings.

Main Methods:

• Utilized a standard logistic regression model within a Bayesian framework.
• Implemented continuous updating of posterior estimates for dose assignment and early stopping decisions.
• Extended the design to accommodate delayed toxicity assessments.

Main Results:

• Simulation studies demonstrated the proposed design's competitiveness against existing methods.
• The Bayesian approach showed superior performance in certain scenarios.
• The design effectively handles delayed toxicity data.

Conclusions:

• The proposed Bayesian logistic regression design offers a practical and effective approach for dose-finding in cancer combination trials.
• This method enhances accessibility for clinicians by simplifying complex statistical models.
• The design's ability to manage delayed toxicities adds to its clinical utility.