The continual reassessment method for dose-finding studies: a tutorial

The Continual Reassessment Method (CRM), along with other adaptive dose-finding study designs, has gained popularity since its proposal by O'Quigley. Several of the reasons it has been embraced by clinical trialists is that it tends to incur fewer toxic events, and more accurately estimate the maximum tolerated dose as compared to the standard Phase I dose escalation designs. Many variations have been published and discussed in the statistical literature, but there has not been as much practical advice for choosing design parameters and implementing the CRM. As a result, the CRM has not been as widely utilized as it could be for dose-finding studies. The goal of this paper is to provide a tutorial for those unfamiliar with the CRM who are either statisticians considering using the CRM for the first time, or investigators with some statistical background. This paper presents the original CRM, and then some of its modified versions. It also explains the specifications that define a CRM design, along with simulated examples of CRMs and standard designs, for illustration.