时间：2019年10月28（周一）下午 16:00-17:00
地点：中北校区理科大楼A1514
题目：Probability Intervals of Toxicity and Efficacy (PITE) design for CAR-T Dose Finding Clinical Trials
报告人：林晓蕾 复旦大学大数据学院 青年副研究员
主持人：方方 副教授
摘要：
Adoptive cell therapy (ACT), such as chimeric antigen receptor (CAR) T-cell therapy, have demonstrated promising therapeutic effects in oncology patients. We consider statistical designs for dose-finding ACT trials, in which the monotonic dose-response relationship assumed in traditional oncology trials may not hold. Building upon previous Toxicity and Efficacy Probability Interval (TEPI) design (Li et al., 2017), we propose a new dose finding method - Probability Intervals of Toxicity and Efficacy (PITE), which utilizes toxicity and efficacy jointly in making dosing decisions, does not require a pre- elicited decision table and at the same time handle Ockham’s razor properly in the statistical inference. We show that optimizing the joint posterior expected utility of toxicity and efficacy under a 0-1 loss is equivalent to maximizing the marginal model posterior probability in the 2-dimensional probability space. An extensive simulation study under various scenarios are conducted and results show that PITE outperforms TEPI in patients allocation (assign more patients to optimal dose and fewer to toxic doses) and optimal dose selection. The simple and transparent features together with good operating characteristics make PITE a potential choice for dose finding designs in ACT trials.
个人简介：
林晓蕾，复旦大学大数据学院青年副研究员，2018年毕业于芝加哥大学生物统计学系，主要研究方向为生物统计学方法，纵向数据和临床试验。