Statistics Forum: Sarah Simpson
Thursday 12 June 2014, 1200-1230
A54, Postgraduate Statistics Centre Lecture Theatre
Adaptive Crossover Designs for Phase II Dose-Finding Trials
Finding the optimal dose of a new medicine is an important part of drug development. Adaptive dose-finding procedures use accumulating data to determine which doses should be allocated to each new cohort of patients recruited to the trial, with the aim of obtaining an accurate estimate of the target dose. In this presentation, we explore adaptive crossover designs for estimating the dose which provides a proportion, π, of the maximum effect of a drug, i.e., the ED100π. We restrict attention to designs where each patient receives placebo and three active doses of the drug in a sequence determined by a Williams square.
Bayesian optimal adaptive procedures are considered which recommend that each new cohort of patients receives the combination of doses that minimises the variance of the posterior modal estimate of ED100π. Prior opinion about the dose-response relationship is represented as pseudo-data. It is assumed that the dose-response relationship follows an Emax model. However, fitting Emax models can be challenging due to problems of non-convergence. With this in mind, when the Emax model fails to converge at an interim analysis we will investigate Bayesian procedures which use a cubic approximation to the Emax model for the purposes of making dose recommendations.
An algorithmic procedure is also considered for dose-finding, which assumes only that the dose-response relationship is monotonic when making interim dose recommendations. Simulation is used to compare the algorithmic procedure and Bayesian optimal design for estimating the ED100π, using a non-adaptive incomplete block design as a benchmark for comparison.