December 9 @ 4:00 pm - 5:00 pm
Title: Clinical Trial Designs for Precision Medicine
Presenter: Yeonhee Park Assistant Professor Biostatistics and Medical Informatics UW Madison
Abstract: Precision medicine is revolutionizing medical research and changing the way physicians treat patients based on the fact that there are subgroups of patients who are sensitive to or respond differently to treatments. In this talk, I will present two interesting clinical trial designs for the precision medicine. First, I developed a group sequential enrichment design based on adaptive regression of response and survival time. This design starts by enrolling patients under broad eligibility criteria. At each interim decision, submodels for regression of response and survival time on a possibly high dimensional covariate vector and treatment are fit, variable selection is used to identify a covariate subvector that characterizes treatment-sensitive patients and determines a personalized benefit index, and comparative superiority and futility decisions are made. Enrollment of each cohort is restricted to the most recent adaptively identified treatment-sensitive patients. Group sequential decision cutoffs are calibrated to control overall type I error. A simulation study shows that the proposed designs accurately identifies the sensitive subpopulation if it exists and yields substantially higher power compared to a conventional all-comer group sequential design. Second, I proposed a Bayesian adaptive design using avatar information to guide the choice of a personalized salvage treatment. I jointly model for avatar’s data and patient’s data under the latent-class model framework, where a patient is expected to be avatar congruent or avatar incongruent based on the patient covariates. Based on accumulating data for avatars and patients, the expected mean utility of each investigational treatment is calculated, and the next patient is assigned to treatment by adaptive randomization of the estimated utility probability. A simulation study shows that the proposed design performs well compared to conventional clinical trial without using avatar information for any type of patient and avatar-driven trial using the avatar with the best performance for the incongruent patients.