Abstract 3172: 1 Unsupervised cellular phenotypic hierarchy enables spatial intratumor heterogeneity characterization, recurrence-associated microdomains discovery, and harnesses network biology from hyperplexed in-situ fluorescence images of colorectal carcinoma

Abstract

Abstract Background: Tumor cells and their stromal cell counterparts that comprise the tumor microenvironment (TME) reciprocally coevolve to generate heterocellular communication networks. A distinctive characteristic of the functional organization of this continuously evolving ecosystem is spatial intratumoral heterogeneity (ITH), a key determinant of disease progression landmarks in multiple carcinomas that include colorectal cancer. To optimize diagnosis, prognosis, therapeutic strategies and to identify novel therapeutic targets it is important to define spatial ITH in the tumors of individual patients and determine the mechanistic underpinnings of its relationship to metastatic potential, immune evasion, recurrence, therapeutic response and drug resistance. Methods: The first step in investigating spatial ITH is to identify the cell phenotypes within the TME. This, however, is a challenging task owing not only to the diversity of well-defined cell types within the TME but also the intrinsic plasticity of many of these cell types in response to the selection pressures within their particular confines (i.e., microdomains). There has been a recent explosion of hyperplexed (> 9 fluorescence or mass spec-based) biomarker labeling and imaging modalities utilizing various reagent technologies to probe the same tissue sections with several dozens of biomarkers at cellular and subcellular resolutions. The challenge now is to accurately characterize the complex spatial and high-dimensional output of these hyperplexed techniques. Results: We propose LEAPH an unsupervised machine learning algorithm for characterizing in situ phenotypic heterogeneity in tissue samples. LEAPH builds a phenotypic hierarchy of cell types, cell states and their spatial configurations. The recursive modeling steps involve determining cell types with low-ranked mixtures of factor analyzers and optimizing cell states with spatial regularization. We applied LEAPH to hyperplexed (51 biomarkers) immunofluorescence images of colorectal carcinoma primary tumors (N=213). LEAPH, combined with pointwise mutual information (PMI), enables the discovery of phenotypically distinct microdomains, composed of spatially configured computational phenotypes. Harnessing network biology. LEAPH identified a subset of microdomains composed of cancer stem cells driving a Wnt signaling- based immunosuppressive program in patients who exhibited recurrence within 3 years of surgical resection. The LEAPH framework, when combined with microdomain discovery and microdomain-specific network biology, has the potential to provide insights into pathophysiological mechanisms, identify novel drug targets and inform therapeutic strategies for individual patients. Citation Format: Samantha Furman, Andrew Stern, Shikhar Uttam, Taylor D. Lansing, Pullara Filippo, S. CHAKRA Chennubhotla. 1 Unsupervised cellular phenotypic hierarchy enables spatial intratumor heterogeneity characterization, recurrence-associated microdomains discovery, and harnesses network biology from hyperplexed in-situ fluorescence images of colorectal carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 3172.