The continual reassessment method for multiple toxicity grades: a Bayesian quasi-likelihood approach.

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Oncology

Background:

• Phase I cancer trials often use dose-limiting toxicity to guide treatment escalation.
• Dichotomizing toxicity grades may oversimplify severe or irreversible toxicities (e.g., renal, liver, neurological).
• Existing methods may not adequately capture the nuances of graded toxicity data.

Purpose of the Study:

• To propose a novel method, the Quasi-Continual Reassessment Method (Quasi-CRM), for Phase I clinical trials.
• To enhance dose allocation by incorporating detailed toxicity grade information.
• To improve upon standard methods when dealing with severe or graded toxicities.

Main Methods:

• Developed the Quasi-Continual Reassessment Method (Quasi-CRM) as an extension of the standard CRM.
• Converted toxicity grades into numerical scores reflecting their impact on dose allocation.
• Utilized a quasi-Bernoulli likelihood to integrate toxicity scores into the CRM framework.

Main Results:

• Simulation studies showed Quasi-CRM outperforms the standard CRM in dose allocation.
• Quasi-CRM demonstrated comparable performance to existing advanced methods like the univariate Bekele and Thall method.
• Sensitivity analyses confirmed the robustness of Quasi-CRM with respect to toxicity score variations.

Conclusions:

• The Quasi-CRM offers a statistically sound and practical approach for Phase I trials with graded toxicity data.
• This method enhances patient safety by providing more precise dose escalation decisions.
• Quasi-CRM represents a valuable advancement in clinical trial design for severe diseases.