Pediatric Acute Myeloid Leukemia (AML) is the most lethal hematologic malignancy in childhood, with a probability of 5-year survival at only 60%. Most children diagnosed with AML initially respond well to standard chemotherapy; however, nearly 40% eventually develop relapsed disease, which responds poorly to treatment and is fatal in the majority of patients. Although age at diagnosis, response to induction chemotherapy, and cytogenetic status have been identified as coarse prognostic factors in pediatric AML, it is still unclear what molecular features lead certain patients to relapse over others. Thus, developing an enhanced understanding of the mechanistic drivers underlying relapse in pediatric AML represents a significant area of clinical need. Many reports indicate that there are rare, hematopoietic stem cell-like subpopulations in AML patients that resist chemotherapy and drive relapse. However, the exact characteristics of these relapse-associated cells?often called ?leukemic stem cells? (LSCs)?are a matter of contention, with reported phenotypes spanning much of the known hematopoietic developmental continuum and differing significantly between patients and throughout the course of disease. As such, the identity and importance of these relapse-associated cells as well as their relationship to normal hematopoietic developmental processes remain mysterious. The proposed project will examine the relationship between single-cell AML phenotypes, clinical outcomes, and normal myeloid development in 60 clinically-annotated primary samples from pediatric AML patients in order to identify relapse-associated cellular subtypes. To achieve this, we will leverage the versatility of mass cytometry, a 40-parameter single-cell proteomics platform, and machine learning in simultaneously studying the complex surface and signaling phenotypes of millions of leukemic cells from patients? diagnostic and relapse bone marrow samples relative to healthy controls. Central hypothesis: We hypothesize that high-dimensional molecular profiling of primary AML cells will reveal consistent, functional phenotypes associated with relapse-driving subpopulations that computationally align with particular stages of healthy hematopoietic development and represent points of future therapeutic intervention.
Aim 1 : Develop methods to computationally align high-dimensional, single-cell AML phenotypes with their most analogous developmental state along the healthy myeloid continuum.
Aim 2 : Utilize predictive modeling to determine the surface, signaling, and functional phenotype of AML subpopulations predicting relapse and functionally validate these characteristics in vitro and in vivo.
For reasons that remain poorly understood, nearly 40% of the roughly 500 children diagnosed with acute myeloid leukemia (AML) each year in the United States will experience relapse after initially responding to treatment. To better characterize why some AML patients relapse while others do not, the proposed project will use state-of-the-art molecular profiling approaches to characterize millions of primary AML cells and their computational ?alignment? with healthy hematopoietic development. These data will be used to construct a method of predicting at diagnosis which patients have a high risk of relapse in order to identify prognostically important cell populations that represent future therapeutic targets in the management of pediatric AML.
