PhD Public Seminar - Jessie Medeiros
PhD Public Seminar - Jessie Medeiros
Supervisor: Dr. John Dick
Title: Using Stemness Features to Guide the Development of Prognostic Biomarkers in Leukemia and Beyond
Date: Wednesday, April 3rd , 2024 at 10:00 AM
Location: MSB 3287
Abstract: Significant advances in our understanding of the leukemogenic process have emerged over the last decade. We now know that there is a long history of clonal evolution prior to a patient ever receiving a clinical diagnosis. Mutations present up to ten years prior to diagnosis are predictive of future leukemia development with similar mutations found to originate as far back as the developing fetus. From the acquisition of the first mutation to clonal hematopoiesis, preleukemia, and overt malignancy, this continuum emerges through the properties collectively possessed by a pool of heterogenous stem cells. As these cells are the primary unit of selection, both in the natural history of disease emergence and in the context of therapy resistance, their features should be used to guide effective diagnostics and eventually, therapeutics.
Using single-molecule molecular inversion probe technology, we developed a targeted sequencing panel and complementary informatic pipeline to assess the most recurrently mutated genes in clonal hematopoiesis and myeloid malignancies. Our approach outperformed the current gold standard for clinical testing with sensitive capture of low allele frequency mutations. We deployed our methodology and demonstrated the prognostic potential of capturing these variants across clinically relevant contexts, including patients experiencing blood and cardiac related complications.
To gain insight into the earliest stages of leukemogenesis, we used myeloproliferative neoplasm (MPN) as a naturally occurring and accessible biological model of leukemic progression. We identified rare patients with concurrent myeloid and lymphoid diseases and used high-resolution cell sorting and genetic approaches to reveal complex clonal origins even in a small subset of patients. These findings provided the impetus to move beyond mutational profiling toward assessing the transcriptomic landscape of a large cohort of MPN patients. We used machine learning approaches informed by gene sets capturing stem cell heterogeneity to train and validate expression signatures predictive of both overall survival and risk of leukemic transformation.
Collectively, our work demonstrates the power of using stemness as a guiding principle in the development of diagnostic tools. Importantly, we developed these tools to be high-throughput, scalable and cost-effective, making them amenable to clinical adoption and enabling eventual precision medicine approaches.