With the advent of spatial-omics, our ability to quantitatively map tumor organization in situ at the level of cell states, cell types, and genetic subclones has improved dramatically. We use detailed tumor spatial maps to identify patterns that repeat robustly across tumors. Knowing which cell states and cell types tend to associate with each other and how they interact within a community can potentially be leveraged as a therapeutic vulnerability. We integrate spatial information across molecular, cellular, and tissue scales in order to understand the principles of tumor organization. Our initial focus is on CNS tumors.
Cancer cells often mimic aberrant versions of normal developmental expression programs such as epithelial-to-mesenchymal transition or senescence. It is unknown to what extent they may also mimic spatial associations found in the tissue of origin. In addition to searching for spatial patterns that repeat between tumors, we can also look for spatial patterns that repeat between a tumor and its tissue of origin in the adult or during development.Â
On one hand, a hallmark of cancer is lack of responsiveness to environmental cues; on the other hand, even in advanced, aggressive disease, tumors can exhibit spatial organization suggestive of responsiveness to long-range environmental signals. We are exploring this paradox in order to better understand the dichotomy of structure vs. disorganization in cancer and how it relates to tumor evolution and plasticity.
What are the regulators and drivers of tumor organization? Are there essential interactions -- interactions that when perturbed, would fundamentally alter the overall organization of the tumor? Could modulating or perturbing tumor organization have therapeutic benefit in certain contexts?
By identifying recurring spatial patterns and associations across scales, we can make predictions about regulators and drivers of tumor organization. However, while our computational analysis serves as an important tool for making predictions, we then need to test these predictions experimentally. Therefore, we aim to identify ex vivo models of tumor organization and develop an experimental toolkit for testing these predictions.
Many existing imaging methodologies provide us with rich spatial data at different scales, sometimes with functional readouts. We believe that bridging these classical techniques with spatial-omics will enable us to gain insight into tumor organization from new angles. In particular, we are interested in understanding tumor vasculature - immune dynamics from both a transcriptional and functional perspective in order to engineer more effective immunotherapies.