Two-Stage Modified Toxicity Probability Interval Design for Low Target Toxicity Rate

• Activity Number: 364
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Section on Risk Analysis
• Author(s): Sheau-Chiann Chen* and Yu Shyr
• Companies: Vanderbilt University and Vanderbilt University
• Keywords: adaptive design ; dose-finding ; phase I

Abstract

In recent practical phase I trials, the modified toxicity probability interval (mTPI) design for a binary response has been increasingly used to find the maximum tolerated dose (MTD). The dose-finding spreadsheet of the mTPI design includes escalation, staying, de-escalation and unacceptable toxicity actions. But, when the target toxicity rate is 0.1 or less, there is no staying action for a small cohort size at each dose. This problem may reduce the accuracy of the identified MTD. Our aim is to improve the performance of the mTPI design through incorporating lower grade toxicity information into dose finding designs. Thus, a Dirichlet/multinomial model is used to develop a two-stage mTPI design for a polytomous response. The simulation results show that the two-stage mTPI design performs as well as or better than the mTPI design except while the moderate grade and severe grade toxicity probabilities have the same increasing trend. Also, the two-stage mTPI design performs well when all doses are too toxic. To summarize, the two-stage mTPI design is recommended for use with a target toxicity rate of 0.1 or less.