EPCO-46. SPATIAL TRANSCRIPTOMICS ANALYSIS OF GLIOBLASTOMA REVEALS THREE DISTINCT REGIONAL PROGRAMS OF T-CELL INFILTRATION

Abstract Glioblastoma (GBM) resists T-cell killing by developing an immunosuppressive tumor microenvironment. Transcriptomic technologies, such as single-cell RNA sequencing (scRNA-seq), have characterized immune cell phenotypes and putative cell-cell interactions in GBM. However, the tumor microenvironment surrounding T-cells is still not well characterized, motivating the need for spatial analyses. Two publicly-available GBM spatial transcriptomic datasets were analyzed, with one discovery cohort (N=18) and one validation cohort (N=12). Using cell deconvolution with a scRNA-seq reference dataset, areas of high T-cell infiltration were identified and clustered into regional programs based on transcriptomic signatures. To highlight which ligand-receptor interactions are most prominent in these regional programs, a novel spatial enrichment algorithm was applied. The impact of spatially enriched ligand-receptor interactions on overall survival was evaluated using bulk RNA-sequencing data from The Cancer Genome Atlas (TCGA). Three distinct T-cell regional programs (transitional, immunosuppressive, and infiltrative) were found across both the discovery and validation cohort. The infiltrative T-cell regional program resided on the periphery of the tumor with a neural progenitor phenotype (NRGN, VSNL1, SLC17A7). In contrast, the immunosuppressive T-cell region had a higher proportion of malignant cells and a greater mesenchymal tumor signature (ANXA1, CD44, complement pathway). Furthermore, ligand-receptor enrichment analysis showed that the immunosuppressive region had higher tumor and tumor-associated macrophage (TAM) signaling towards CD4 T-cells along the collagen, midkine, and osteopontin pathways. TCGA survival analysis showed that higher collagen and midkine signals were associated with worse overall and disease free survival. Our results identified three distinct regional programs in GBM and highlight a set of signaling pathways associated with T-cell suppression. Our findings suggest that the disruption of these pathways could reduce immunosuppressive signals from tumor and TAMs to T-cells. Subsequent validation with spatial proteomics data in glioblastoma is ongoing.