Robust Estimators for Missing Covariate Parameters for Right-Censored Data

Providing guidance on alternative treatments for diffuse large B-cell lymphoma patients in partial remission

Full Project Name:Robust Estimators for Missing Covariate Parameters for Right-Censored DataPrincipal Investigator:Soyoung Kim, PhD, Institute for Health and EquityCo-Investigator:Kwang Woo Ahn, PhD, Institute for Health and Equity; Mehdi Hamadani, MD, MedicineAward Amount:$200,000

Award Date

July2022

Project Duration:24 months

Project Description Narrative:

Non-Hodgkin's Lymphoma is one of the most common cancers in Wisconsin. Traditional lymphoma treatments include chemotherapy, radiation therapy, and hematopoietic cell transplantation (HCT). Diffuse large B-cell Lymphoma (DLBCL) is the most common non-Hodgkin's Lymphoma and has dismal outcomes with multi-agent chemotherapies.

The FDA recently approved chimeric antigen receptor (CAR)-T cell therapy for DLBCL patients who relapsed or did not achieve remission after chemotherapy. Because CAR-T cell therapy is new and offers another treatment option for DLBCL patients, there is much interest in comparing it with traditional treatments, particularly with HCT.

This project will give clear guidance on a choice of treatments between auto-HCT and CAR-T cell therapy for DLBCL patients in partial remission. In addition, researchers will develop statistical methods that can be applied to general survival data analysis in oncology studies.

Project Updates:

Developed a novel statistical method to handle missing covariates in time-to-event data using multiple imputation
Created an efficient estimation scheme for left-truncated competing risks data under the case-cohort study design and successfully applied it to real coronary heart disease data
Conducted a comprehensive study of the multivariate Cox model with missing covariates' statistical properties and assessed its performance through simulation studies and real data analysis