Abstract

Dose finding clinical trials are fundamentally important for new drug development. Based on the key assumption that both toxicity probability and efficacy probability are increasing in dose level, a primary goal is to estimate the maximum tolerated dose (MTD), which is the maximum dosage with an acceptable level of toxicity. This objective is particularly important for cytotoxic drugs against cancer, which have potentially strong or even lethal toxicity. There are both nonparametric and parametric approaches to adaptive design of dose finding clinical trials. The continual reassessment method is most popular for the parametric approach, and we will discuss and compare the performance of different functions for depicting the toxicity probability curve.