Quasi Continual Reassessment Method for Dose-escalation Studies Based on Toxicity Grades
Departments of Statistics and Biostatistics
University of Wisconsin - Madison
January 21, 2005, 12 - 1 pm in room 3265 Medical Sciences Center (1300 University Ave.)
Grade information in phase I clinical trials is important to determine the maximum tolerated dose (MTD) for severe toxicity responses. Because the numbers of subjects are generally small, incorporating grade information requires parsimonious and practical methods. In this article we introduce an equivalent toxicity score to quantify the impact of each grade on dose allocation, and fine-tune the definition of the MTD using all toxicity grades. We then propose the quasi continual reassessment method (Quasi-CRM) to incorporate the score using simple quasi-likelihood method. The asymptotic properties of Quasi-CRM are also established. Simulations indicate that Quasi-CRM allows more cautious escalation, and performs significantly better than the original CRM. We also discuss simple extensions of the tradional or up-and-down designs which accommodate grades.