A0747
Title: Spatial omics driven computational discovery of prognostic cellular neighborhoods in tumor ecosystems
Authors: Lulu Shang - MD Anderson (United States) [presenting]
Abstract: Recent advancements in spatial omics have deepened the understanding of tissue microenvironments and cellular neighborhoods in tumors and other diseases. Existing neighborhood detection methods lack a universal, interpretable framework, yielding ambiguous neighborhoods that cannot be aligned across patients, disease stages, or platforms, hindering clinical translation. The aim is to present SCOPE (SystematiC prOfiling of disease-relevant cellular Ecotypes), which extracts functional spatial ecotypes directly from single-cell spatial data and scales to thousands of tissue sections and millions of cells. SCOPE is benchmarked on both simulated and real datasets, showing that it can identify ecotypes with known prognostic patterns more accurately than existing methods. Ecotype enrichment is found across patient subtypes, revealing mutually exclusive immune-active versus immune-suppressive niches, and ecotype-remodeling trajectories are traced, enhancing survival and therapy response prediction beyond standard clinical markers. SCOPE provides a more comprehensive view of tissue microenvironments and shows great potential in improving patient survival prediction and therapy response assessment.
