Bayesian designs for phase I clinical trials in cancer
In phase I clinical trials in cancer, patients often receive successive cycles of an experimental drug. Dose limiting toxicity may occur during any of the cycles. The aim of the project is to investigate the use of an interval-censored survival model in dose-escalation procedures for cancer treatments. In this approach, a generalised linear model will be used to represent the relationship between the probability of experiencing a Dose Limiting Toxicity (DLT) during cycle j of the treatment conditional on there being no DLT prior to cycle j on the dose level and other covariates such as a biomarker. A Bayesian approach will be used by first creating a prior distribution for the parameters associated with cycle, dose and possibly a biomarker to enable a pre-defined Maximum Tolerated Dose (MTD) to be computed for the first cohort of patients entering the first cycle of treatment. Once this cohort has completed the cycle the distribution of these parameters is updated and the resulting posterior distribution used to compute the MTD for the patients entering the next cycle of treatment and also for the next cohort of patients entering the study during the first cycle.
- Sinclair, K., Whitehead, A., ( 2014 ). A Bayesian approach to dose finding studies for cancer therapies: incorporating later cycles of therapy . Statistics in Medicine. 33:2665-2680.