The principal aim of phase I oncology trials is to estimate the maximum tolerated dose (MTD) of a novel compound. Empirical procedures or algorithmic designs, such as the ‘3+3’ design, can lead to falsely identifying an MTD as well as exposing patients to toxic or pharmacologically inactive doses. Yet information on the relationship between dose and toxicity will be available from preclinical studies by the time first-in-man trials are conducted. Using preclinical data to specify a prior for the human dose-toxicity model parameter(s) can improve interim dose recommendations and precision for estimating the MTD. Such advantages, however, must be balanced against the risk that more patients will possibly be treated with excessively toxic doses in the case of a prior-data conflict. We make a proposal for how to measure and address a prior-data conflict in a sequential study with a small sample size. Animal data are translated onto the human dosing scale and incorporated via a robust two-component mixture prior for the parameters of the human dose-toxicity relationship. The weights placed on each component of the prior are chosen empirically and updated dynamically as the trial progresses and more data accrue. After completion of each cohort, we use a Bayesian decision-theoretic approach to evaluate the predictive utility of the animal data for the observed human toxicity outcomes, reflecting the degree of agreement between dose-toxicity relationships. The proposed methodology is illustrated through data examples and an extensive simulation study. To evaluate the performance of our novel approach, we also compare it with alternative Bayesian dose-finding approaches driven by weakly-informative priors.
1. To learn Bayesian methods for adaptive dose-escalation in oncology trials and to consider Bayesian decision-theoretic approach to dose-finding problems
2. To understand prior-data conflicts in a sequential study and hear about future research