ABSTRACT
In situations where the clinical investigators believe the MTD is among the higher doses but are not comfortable treating patients at a high dose without first testing its lower doses, one may adopt a two-stage CRM that starts a trial according to an initial sequence {xi,0} with x1,0 = d1. This chapter discusses the characterization and the choice of the initial sequence. To be specific, Section 10.2 introduces some basic concepts about the ordering of dose sequences. Section 10.3 presents the calibration approach for {xi,0}, along with numerical recommendations in some common clinical scenarios. Practical issues such as sample size considerations related to the use of a two-stage CRM will be addressed in Section 10.4. The calibration techniques will be illustrated in Section 10.5 in the context of the NeuSTART.
