Selection in inflammation-susceptible cancer subtypes

How can we leverage information about unperturbed (treatment naïve) tumor ecology and evolution? Chronic Myelomonocytic Leukemia (CMML) provides an opportunity to study pre-leukemic dynamics and progression: patients can be longitudinally followed and typically remain treatment-naïve before they progress or transform to aggressive disease. This progression is potentially due to inflammatory changes/cytokine release syndrome (CRS), which act as time-dependent adaptive pressures. During progression, the genomic heterogeneity changes little, but cell phenotypes diversify. We seek to identify novel measures of cell phenotypic diversity and models of selection, to understand the importance of the rich heterogeneity and immune response of these leukemias.

Phenotypic subpopulations in patient samples can often only be identified via single cell analyses. We perform computational analyses of single cell data to inform quantitative models of selection leading to progression. We investigate how transcriptional cellular diversity gives rise to stem-like or inflammatory subtypes and leads to stochastic progression. For these investigations we leverage individual patient samples, at multiple times, using bone marrow derived single cell RNA sequencing, to determine transcriptional/phenotypic states using machine learning. We also use matched flow cytometry data to identify immune-, stem cell-markers and cytokine receptors. By genotyping the same samples we can link phenotypic states to cancer driver mutations, especially those associated with changes in transcription, such as IDH1/2, TET2 and others. These studies can reveal how certain genotypes contribute to phenotypic diversification, and can be used to inform stochastic models of selection during inflammatory episodes, to can explain progression and aggressiveness of disease.