Abstract PO016: Spatial Transcriptomic Profiling to Characterize the Tumor-Vascular Interactome of Hepatocellular Carcinoma

Abstract BACKGROUND: The hepatocellular carcinoma (HCC) tumor microenvironment (TME) is composed of a complex ecosystem dominated by cancer cells and the endothelial cells that line tumor blood vessels. Although many genomic drivers have been identified and at least three transcriptional subsets have been proposed, these efforts have not yet led to novel therapies or otherwise significantly impacted management options. Single cell transcriptional profiling has generated deep insights into the multiple heterogeneous cell types within tissues, but the spatial context of these data is lost during single cell processing. Spatial transcriptomic approaches aim to bridge the gap between dissociative single cell technologies and in situ histopathological characterization.METHODS: To gain insight into potential in situ cancer-endothelial crosstalk interactions, we utilized the Nanostring GeoMx spatial transcriptomics platform with the Cancer Transcriptome Atlas ~1800 gene oligonucleotide probe panel to generate tumor (Arginase+) and blood vessel (CD31+) areas of interest (AOI) gene expression profiles from formalin-fixed, paraffin-embedded archival tissue specimens obtained from HCC resection specimens. Oligonucleotides released from each microscopic AOI were then captured, processed by DNA sequencing, and analyzed using custom computational pipelines.RESULTS: Using the 119 ROI containing data from both tumor and vessels that passed quality control filters, we performed unbiased hierarchical clustering of both the tumor and vessel areas of interest (AOI) within each ROI using the most highly variable genes for each AOI set and identified at least 3 clusters within each AOI type (tumor and vessel). Based on gene ontology analysis of the tumor AOIs, the two subsets were distinguished by unique immune and inflammatory-related genes. Analogous ontology-based characterization of the vessel AOIs demonstrated two groups: 1) an interferon-activated, inflamed progenitor, and immune checkpoint-associated cluster; and 2) a TGF-beta and oxidative stress-associated cluster. Notably, both vessel clusters also contained significant numbers of leukocyte genes, concordant with the intimate relationship of the vasculature and immune system. Canonical correlation analysis (CCA) utilizing both the most variable genes within each AOI set showed significant correlated gene sets within tumor AOIs and vessel AOIs, implying biologically significant interactions in multiple signaling pathways.CONCLUSIONS: Spatial transcriptomic profiling enables an understanding of cell-cell interactions in situ that can uncover biologically distinct tumor and blood vessel niches within the HCC microenvironment. Subsequent efforts will be focused on functionally assessing the spatially linked cancer and endothelial cell phenotypes with the goals of developing improved prognostic and predictive biomarkers and generating novel drug targets. Citation Format: Joseph W Franses, Michael J Raabe, Amaya Pankaj, Bidish Patel, Avril Coley, Irun Bhan, Martin Aryee, David T Ting. Spatial transcriptomic profiling to characterize the tumor-vascular interactome of hepatocellular carcinoma [abstract]. In: Proceedings of the AACR Special Conference: Advances in the Pathogenesis and Molecular Therapies of Liver Cancer; 2022 May 5-8; Boston, MA. Philadelphia (PA): AACR; Clin Cancer Res 2022;28(17_Suppl):Abstract nr PO016.