825 High-dimensional image cytometry reveals spatially organized tumor-immune microenvironment in hepatocellular carcinoma

Background Structured and spatial-nuanced interactions between components in tumor microenvironment (TME) regulates the efficacy of anti-tumor regimens. Insights into this orchestrated behavior therapeutic responders non-responders will facilitate immunotherapies. High-multiplex imaging spatial statistics enable deep profiling TMEs by simultaneous arraying cell phenotypes locations. In study, we quantified landscape from neoadjuvant cabozantinib nivolumab administered locally advanced hepatocellular carcinoma (HCC) biospecimen. Methods 14 patients with HCC were treated combination through Johns Hopkins Sidney Kimmel Comprehensive Cancer Center. Among them, 12 (5 + 7 non-responders) underwent successful margin negative resection are subjects to tissue microarray (TMAs) construction containing 37 representative region cores. Using TMAs, performed mass cytometry (IMC) a panel 27-cell lineage functional markers (figure 1). All multiplexed images then segmented generate single-cell dataset that enables (1) tumor-immune compartment analysis (2) community based on graph-embedding technology. Results these hierarchies merged response-associated biological process patterns. Image processing discriminated 59,453 cells clustered 17 types. Multi-level quantification revealed distinct TME arrangements across cores (R) (NR): showed at immune-tumor boundaries NR, PD-L1 level is significantly higher than remote regions; however, Granzyme B lower 2B). We also identified proximity CD8+ T subset macrophages – Arginase 1hi CD163- (hazard macrophage) CD4+ cells, prognostic biomarker therapy 3A 3B). In-depth extracted cell-cell interaction networks proximity. Next, hierarchical clustering grouped all similar (cell types) 8 categories (CC). correlation test, observed macrophage-enriched CC (MCC) lymphocytes-enriched (LCC) strongly communicating CC; whereas R, such communications weakened engagement MCC LCC 3C). Conclusions conclusion, employed an unbiased, quantitative determine how immune interact responding nonresponding tumors. Based our results, four immune-regulating factors derived summarized as communication 4). The proposed framework represents novel application translational medicine has potential initialization validation computational immuno-oncology models. Acknowledgements authors acknowledge financial support Bristol-Myers Squibb, Exelixis, National Institute Specialized Program Research Excellence (SPORE) Gastrointestinal Cancers (P50 CA062924), Passano Foundation, Institutes Health (Grant No. U01CA212007 R01CA138264) Emerson Collective Fund (640183). Ethics Approval studies involving human participants reviewed approved Institutional Review Board Medical Institutions. Consent Written informed consent for participation was not required study accordance national legislation institutional requirements. Abstract 825 Figure 1 A 27 used stain processed using IMC. marker names descriptions included 2 (A) Color overlays proteins covering Pan-Keratin CD45 (rainbow) (white) whole core subregions. (B) (C) Protein expression analytical strategy. For compartmentalized cores, expressions target examined adjacent border truncated treatment response criteria comparisons 3 Diagram RiskScore. Denote each its nearest hazard d1 d2, thus RiskScore formally computed taking proportion d2 combined distance d2. per-cell basis non-responders. Cell maps associated 4 synergistic immunity lymphocytes favors nivolumab; function regulators (i.e., GranB PD-L1) upregulated throughout non-responders; close away associate poorer effector