Online Program

Return to main conference page
Wednesday, September 25
Wed, Sep 25, 9:45 AM - 10:30 AM
Marriott Foyer
Poster Session

Bayesian Dose-Finding Design for Time-to-Event Data with Multiple Toxicity Grades (300884)

Yuan Ji, The University of Chicago 
*Meizi Liu, The University of Chicago 

Keywords: Bayesian adaptive design, Late-onset toxicities, Multiple toxicities, Time-to-Event toxicity, Phase I clinical trial, toxicity grades

A well-designed and conducted dose escalation study lays the foundation for successful clinical development. With patient centric drug development, the focus of dose finding trials switches from targeting the maximum tolerated dose (MTDs) to more comprehensive evaluation of the drug effects on patients, such as toxicity grades summarized as a toxicity burden. In addition, accounting for time-to-event outcomes may allow faster enrollment speed, leading to shorter trials. We propose a Bayesian design to incorporate time-to-event data of multiple toxicity grades in dose finding. The design allows rolling patient enrollment based on the assessment of toxicity burden. We show that the proposed design can speed up the trial process and better characterize the toxicity profile of each dose. Through simulation study, we report the operating characteristics of the design in terms of duration of the trial and the success in finding the most promising dose with acceptable toxicity profile for later drug development.