ABSTRACT
In some dose-finding trials, there are several groups of patients, and the goal is to estimate a maximally tolerated dose (MTD) within each group. These groups may be defined by the patients' degree of impairment at baseline, amount of pre-treatment, or genetic characteristics. To avoid the issues with running parallel groups, a number of statistical methods have been proposed for estimating MTDs when there is heterogeneity among patients. All of these methods proposed to date for dose-finding, accounting for patient heterogeneity, are generalizations of model-based methods for single-agent trials. The methods can be broadly classified as those that incorporate heterogeneity by using additional parameters in the model, or as those based on order-restricted inference, either alone or in combination with fully or underparametrized parametric models. Yuan and Chappell propose a generalization of the continual reassessment method (CRM) that is a hybrid of the single agent-single group CRM and isotonic regression methods described by Robertson, Wright, and Dykstra.
