ABSTRACT
This chapter presents methodology for identifying optimal dose-schedule combinations in early-phase oncology trials. In oncology, several preclinical and clinical studies have established a relationship between toxicity, as well as efficacy, and schedule. Taking into account the subset of combinations for which the practitioners know the toxicity order, the partial-order continual reassessment method aims to formulate possible simple orders of the toxicity profile. Li developed a method based on a strong assumption that a prolonged administration of the study drug would not increase the dose-limiting toxicity probability in patients, an assumption that may not always hold. The method of Guo is a phase I-II method that does not rely on such a strong ordering assumption. The proposed Bayesian isotonic transformation of Li is an extension of Dunson and Neelon to the matrix order setting. The overall allocation strategy in Guo is to assign patients to dose-schedule combinations that are safe and efficacious.
