Using Machine Learning With Single Cell Multi-modal Bone Marrow Data to Understand the Mechanisms ..
Wesley V Wilson 12, Zoey Kline 2, Fei Mao 2, Alfred L Garfall 34, Dexiu Bu 5, Elena J Orlando 5, Jennifer L Brogdon 5, Adam D Cohen 34, Michael C. Milone 14 1 Center for Cellular Immunotherapies, University of Pennsylvania, Philadelphia, Pennsylvania. 2 Perelman School of Medicine at the University of Pennsylvania 3 Division of Hematology-Oncology, Department of Medicine, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania. 4 Abramson Cancer Center, University of Pennsylvania, Philadelphia, Pennsylvania. 5 Novartis Institutes for Biomedical Research, Cambridge, Massachusetts.
Poster Not on DIsplay
Multiple myeloma (MM) is a hematological malignancy characterized by the clonal proliferation of plasma cells in the bone marrow. Despite significant advances in treatment, MM remains an incurable disease for the majority of patients, necessitating a deeper understanding of its underlying biology to improve therapeutic strategies. In this study, we employed single cell RNA sequencing, scBCR sequencing, and flow cytometry to investigate the transcriptome dynamics , composition, and clonal relationships of MM cells and the tumor micro-environment over time in patients enrolled in a dual BCMA-CAR / CD19-CAR T cell clinical trial. Specifically, we aimed to identify and compare the gene expression profiles of relapse patients versus long-term responders, shedding light on the molecular mechanisms associated with treatment resistance and favorable outcomes. We also looked at the tumor microenvironent including immune cells and their profiles over time of treatment. To capture the temporal heterogeneity within the MM population, we implemented spline models, a powerful computational tool for modeling and analyzing dynamic changes over time. Leveraging this approach, we interrogated the transcriptome data obtained from single MM cells at multiple time points from bone marrow biopsies during the clinical trial. We then applied a machine learning model to exploring the entire tumor micro-environment overtime to find correlations that lead response and resistance. Our findings revealed pronounced alterations in gene expression patterns over time in specific populations of cells, highlighting the dynamic nature of MM cellular heterogeneity. These results provide valuable insights into the molecular basis of MM progression and highlight potential therapeutic targets for overcoming treatment resistance. Ultimately, this study paves the way for the development of precision medicine strategies tailored to individual patients based on their unique molecular profiles, with the aim of improving clinical outcomes in multiple myeloma.