Optimal Conditioning Regimen Intensity for Myelodysplastic Syndromes

Developing statistical methods for optimal CRI identification for MDS patients

Full Project Name:Optimal Conditioning Regimen Intensity for Competing Risks Data in Myelodysplastic SyndromesPrincipal Investigator:Kwang W. Ahn, PhD, Institute for Health and EquityCo-Investigator:Soyoung Kim, PhD, Institute for Health and Equity; Wael Saber, MD, MS, MedicineAward Amount:$200,000

Award Date

January2018

Project Duration:24 months

Project Description Narrative:

Myelodysplastic Syndromes (MDS) are a group of diverse bone marrow disorders in which the bone marrow does not produce enough healthy blood cells. MDS primarily affects patients over 60 years of age, but can affect younger patients as well.

To treat MDS, the optimal conditioning regimen intensity (CRI) must be identified. This requires serious consideration of the characteristics of the individual patients with MDS. To accomplish the identification of a therapy regimen that uses less intense chemotherapy and radiation to control the disease, but decreases treatment-related mortality without increasing relapse, the research team will aim to develop statistical methods to find a rule that assigns the optimal CRI to each patient.

Outcomes & Lessons Learned:

Developed a doubly robust statistical method for personalized medicine under the competing risks data setting that can be used to choose the optimal conditioning regimen intensity for myelodysplastic syndrome patients and the optimal graft source for patients with acute myelogenous leukemia, acute lymphoblastic leukemia, chronic myelogenous leukemia, and myelodysplastic syndrome, improving stem cell transplantation outcomes
Developed a variable selection procedure for group variables such as categorical variables, which is a significant and novel contribution to the field, and makes this tool more informative and useful to clinicians and patients
Importantly, simulation studies confirmed the statistical method identified the best treatment given the patient's characteristics for competing risks outcomes, even with some misspecification in the model under the simulation settings
Submitted five grants to continue the research
Submitted two manuscripts to statistics journals to share the findings
Presented work thus far at the ENAR 2019 Spring Meeting