Determining the spatial contexts that modulate immune cell phenotypes in colorectal cancer

Stanford University, Stanford CA

Investigators

Abstract

PROJECT SUMMARY Colorectal cancer is the fourth deadliest cancer in the US and the second deadliest in persons under the age of 50, with a five-year survival rate less than 66%. Immune checkpoint inhibitors have become the standard of care for patients with microsatellite-instability high and mismatch- repair-deficient colorectal cancers, but these represent only ~7% of patients. Within the other 93% of patients, some respond to immune checkpoint inhibitors, but without predictive biomarkers to identify those who will benefit, the survival benefit is not enough to justify clinical use. Interacting cellular communities in the tumor microenvironment comprised of cancer, immune, and other cells can promote or restrain the antitumor immune response. These communities, termed ecotypes, can be described by the transcriptional states of their constituent cells. Work in other cancers has identified ecotypes that maintain functional or hypofunctional T cell states and predict response to immunotherapy. The central goal of this proposal is to identify spatially organized ecotypes in colorectal cancers that govern antitumor T cell activity and predict patients’ response to immunotherapy across all forms of colorectal cancer. This project will use a transcriptomic atlas of more than 200 colorectal cancer patients’ tumors curated from published data, including more than 170 pre-treatment samples from patients treated with immune checkpoint inhibition. A novel deep learning algorithm designed for the proposal will be used to identify spatial ecotypes in the colorectal cancer microenvironment from transcriptomics data. In the long term, robust predictive markers of response to immunotherapy would expand the population of patients eligible for immunotherapy by identifying the microsatellite-stable cancer patients most likely to benefit. Defining these predictive markers in terms of spatially organized multicellular communities will identify the cell-cell interactions that promote or restrain antitumor immunity, which could ultimately identify new drug targets to augment or improve upon existing immunotherapies.