The continual reassessment method for multiple toxicity grades: a bayesian model selection approach.

Area of Science:

• Biostatistics
• Clinical Trial Design
• Pharmacology

Background:

• The Combinatorial Randomized (CRM) model incorporates toxicity grade information in phase I trials.
• The Quasi-CRM model, while useful, is sensitive to the choice of skeleton, impacting performance.
• Existing methods lack robustness in handling toxicity grade data variability.

Purpose of the Study:

• To propose a novel Robust Quasi-CRM model for phase I clinical trials.
• To address the performance variability of the Quasi-CRM model due to skeleton selection.
• To enhance the incorporation of toxicity grade information using Bayesian model selection.

Main Methods:

• Developed a Robust Quasi-CRM model utilizing a Bayesian model selection approach.
• Integrated the concept of parallel skeletons, inspired by the BMA-CRM model.
• Conducted extensive simulation studies to evaluate model performance.

Main Results:

• The Robust Quasi-CRM model demonstrated superior performance compared to existing methods.
• Bayesian model selection effectively mitigated the pitfalls associated with skeleton choice.
• The proposed model showed improved reliability in handling toxicity grade information.

Conclusions:

• The Robust Quasi-CRM model offers a more reliable approach for phase I clinical trials.
• The method is suitable for practical application in clinical trial design.
• This approach enhances the accurate assessment of dose-limiting toxicities.