Joint Modelling of Efficacy and Toxicity in the Dose Escalation Phase I Studies

Most Phase I oncology trials are primarily concerned with establishing the safety profile of a new treatment and focus on toxicity alone to determine the maximum tolerated dose (MTD) defined as the highest dose with the probability of toxicity less than a pre-specified target toxicity rate.When additional data are available, there is an interest in selecting a recommended dose based on PK, PD, efficacy data, etc.We propose a method that uses modeling of both toxicity and efficacy to further guide the estimation of the recommended dose(s) by finding an optimal dose or range of doses that maximizes the efficacy while safety is controlled.The toxicity model is a Bayesian Logistic Regression Model (BLRM) assessing the dose-toxicity relationships.The efficacy model is a polynomial logistic regression model describing the dose-response relationships.This model generalizes the monotonic dose-response relationship and allows for different dose-response shapes.In addition, the association between toxicity and efficacy is included in the modelling using global cross-ratio method.All analyses are performed in the Bayesian framework.The proposed method is evaluated by intensive simulation analyses and operating characteristics are provided.The design identifies adequately the range of the recommended doses while safety is controlled and potentially shortens the duration of a trial by enrolling fewer patients.