|
Abstract:
|
In recent years, Bayesian estimation methods have come into use for drug dose determination as part of Phase I oncology studies. These methods promise to provide more accurate estimation than traditional escalation schemes, which typically use only dose-limiting toxicity (DLT) information from the most recent cohort of patients. Starting from the Continual Reassessment Method for a single-arm trial based on DLTs, modifications, extensions, and competing methods have been developed. A partial list includes methods for multiple arms, overdose estimation, and joint toxicity-efficacy modeling. Implementation issues also exist, such as the choice of cohort size and stopping rule. The advantages and disadvantages of these methods will be examined and their relative merit discussed against a context of traditional methods and examples from Novartis Oncology.
|
