Wei Tu (Queen's University)
Date
Friday March 8, 20242:30 pm - 3:30 pm
Location
Jeffery Hall, Room 234Math & Stats Department Colloquium
Friday, March 8, 2024
Time: 2:30 p.m. Place: Jeffery Hall, Room 234
Speaker: Wei Tu (¾ÅÐãÖ±²¥)
Title: Informative Censoring in Oncology Clinical Trials
Abstract: Clinical trials remain the cornerstone for evaluating new cancer treatments, with the selection of primary endpoints being crucial to the trial’s design and potential to change medical practice. Progression-free survival (PFS) is frequently chosen as an endpoint in oncology trials due to the advantage of shorter study durations, particularly in studies of early-stage diseases. Nevertheless, informative censoring presents a challenge in trials that use PFS as the primary endpoint. This type of censoring occurs when the projected outcomes for individuals who leave the study early differ from those who continue, which can skew the trial’s findings. In this talk, I will discuss the consequences of informative censoring and the statistical methods used to mitigate its effects, using data from the Canadian Cancer Trials Group (CCTG) MA.17R trial as an example. To assess informative censoring’s effect on estimated treatment benefits, we evaluate the use of competing risk assessments and gamma imputation methods for handling non-breast cancer mortality within the dataset. Moreover, we utilize inverse probability censoring weighting (IPCW) to mitigate its impact, assigning weights to participants based on shared characteristics to accurately represent those lost to on-breast cancer causes.
Bio: Dr. Tu is a Senior Biostatistician with the Canadian Cancer Trials Group and an Assistant Professor in the Department of Public Health Sciences at Queen’s University. He completed both his MSc and PhD in Statistics at the University of Alberta. His research interests encompass clinical trial design and analysis, statistical machine learning, data privacy, and genomic data science. His work is supported by the Natural Sciences and Engineering Research Council of Canada, Canadian Institutes of Health Research and Canadian Cancer Society.