Continual Reassessment Method In Phase I Oncology Clinical Trial:Comparison Of Stopping Rules

The main goal of phase I oncology trial is to find the investigational drug’s maximum tolerated dose (MTD) rapidly and accurately, therefore to proceed quickly with a potentially promising treatment to the phase II stage. In phase I oncology clinical trial, it is often assumed that when the dose of the investigational drug increases, the probability of toxicity and potential efficacy for the investigational drug correspondently escalates. Therefore, toxicity is usually used as an important indicator of the investigational drug’s therapeutic use, and the toxic response of the investigational drug usually means the dose limiting dose toxicity (DLT), which is a set of unacceptable adverse events and is also the main factor to restrict the escalation of the investigational drug’s dose. And the investigational drug’s MTD is the dose where the toxicity probability is mostly close to target toxicity probability among all defined doses in the trial.O’ Quigley et al proposed the continual reassessment method (CRM) to find the investigational drug’s MTD in phase I oncology clinical trial. It assumed a dose-toxicity curve model and use the curve to find the dose closest an ideal toxicity rate of the investigational drug, which is the target toxicity probability.When applying the CRM, the designer of the clinical trial must face the question about how to terminate the clinical trial, and many stopping rules about CRM have been emerged recently, therefore choosing the optimal stopping rule is very important.In this paper, we summarize different derived methods of the CRM and different stopping rules, and simulate the application of the CRM with different stopping rules in the phase I oncology clinical trial, therefore to compare the performance of the CRM with different stopping rules in different scenarios and to choose the optimal method to find the MTD in phase I oncology clinical trial.OBJECTIVETo explore the application of CRM and its variants in phase I oncology trial, and the performance of the CRM adopting different stopping rules has been assessed to choose the best variant at different scenarios.METHOD SAS was used to simulate the application of the CRM when exploring the investigational drug’s maximum tolerated dose in different scenarios.RESULTThe simulation results indicated that CRM adopting different stopping rules have similar performance in general, yet each has distinct advantage at some specific scenarios.CONCLUSION(1) Compared with the original CRM, the modified CRM can reduce the risk of subjects exposed to doses with high toxicity, and shorten the trial duration.(2) The performance of CRM is mainly affected by the sample size and there is no significant difference between the influences of different stopping rules on the method itself in general; yet each stopping rule has its distinct advantage at specific scenarios.(3) Compared with approximate calculation, CRM using the integration of joint probability method seems to choose lower dose level as the drug’s MTD.