A Dose Escalation Method for Dual-Agent in Phase 1 Cancer Clinical Trial using the SAS MCMC Procedure

Continual reassessment method (CRM) is a model-based dose escalation method commonly used to design a phase 1 trial in oncology when evaluating one agent. The main characteristics include the definition of a working model for dose levels, an acceptable level of toxicity (target) and a model defining the dose-toxicity relationship (for example, power or logistic function). This relationship is updated after the toxicity evaluation of each patient in the cohort and we assign to the next cohort the dose level closest to the target. This allows estimating the maximum tolerated dose (MTD). With the advance of targeted therapies era in oncology, more and more phase 1 trials aim to identify one or more MTDs from a set of available dose levels of two or more agents. Combining several agents can indeed increase the overall anti-tumor action but at the same time increase toxicity. Since the single-agent dose finding methods (algorithm-based or model-based) are not appropriate for combination therapies, several authors developed different methods. In this paper, we propose to illustrate the SAS MCMC procedure through 2 examples related to the phase 1 cancer clinical trials. The first example shows how to estimate the model parameters of Bayesian CRM. The second example presents a SAS program we developed to implement a dose escalade method for dual agent.