Abstract:
|
We propose a toxicity and efficacy probability interval (TEPI) design for dose finding in phase 1 oncology trials. This approach incorporates efficacy outcomes to inform dosing decisions to optimize efficacy within safety constraints. The TEPI design is an extension over the modified toxicity probability interval (mTPI) design, which uses only dose-limiting toxicity (DLT) data. The TEPI method is motivated by recent promising data from adoptive cell therapy, for which response or a biomarker for response can be observed in the same interval as the DLT. TEPI is based on a two-way decision table constructed by combining the target probability intervals of DLT and efficacy. It uses a beta/binomial model to compute the posterior probabilities of intervals for toxicity and efficacy. The dosing decision depends on the joint unit probability mass (JUPM) of toxicity and efficacy data, which follows the Bayes rule under independent beta prior distributions. TEPI pre-calculates all dosing actions prior to a trial start. It is simple, transparent, and easy to implement. Simulation studies have been performed to evaluate the performance.
|