(P 88) SPATIAL TRANSCRIPTOMICS OF EWING SARCOMA: EVOLUTION OF TUMOR MICROENVIRONMENT ASSOCIATED EXTRACELLULAR MATRIX FROM LOCALIZED TO METASTASIS-ASSOCIATED PRIMARY TUMORS

Objective: To date, the response rate of immunotherapy in Ewing sarcoma (EwS) has been poor. The vast majority of children with metastatic or relapsed EwS die despite current intensive, multimodal therapy. The tumor immune microenvironment (TME) has a vital role in cancer survival and progression with implications in drug resistance and immune escape. However, there is a lack of understanding of the TME of EwS due to the lack of immunocompetent animal models of EwS. Although recent single-cell RNA sequencing (scRNAseq) of EwS and immune cells have revealed a heterogenous transcriptional landscape with distinct subsets of immune cell populations in EwS, little is known about the spatial organization and spatial microenvironmental niches of the TME of EwS.

Methods: We implemented a genome-wide Spatial Transcriptomics platform (Visium; 10X Genomics) on 16 primary pre-treatment biopsied EwS tumors (10 patients with localized EwS (L-EwS); 6 patients with metastatic EwS (M-EwS); median age 8.5 years) (Table 1). We performed deconvolution with a EwS scRNAseq dataset as a reference to estimate the immune spatial composition of EwS and estimated the cellular microenvironmental signaling.

Results: A total of 265,560 spatial features from all 16 primary pre-treatment EwS tumors were analyzed using two different open-source R data processing pipelines. Integrated analysis revealed inter- and intra-tumoral heterogeneity. We inferred the stromal regions of tumors transcriptionally and noted that tumors from L-EwS had significantly more stromal signature than M-EwS (p-value < 2.2e-16). Gene set enrichment analysis (GSEA) using Hallmark gene sets revealed that L-EwS tumors were enriched in epithelial mesenchymal transition (EMT) and inflammatory signatures (adjusted p-value < 0.05) and M-EwS tumors were enriched in proliferative signatures (adjusted p-value < 0.05) (Figure 1A). We identified a set of 46 unique genes enriched in L-EwS that overlapped with genes in EMT hallmark pathways and were predominantly extracellular matrix related (ECM-r) genes such as COL3A1, COL1A2, ITGB1, MMP2, ACTA2, TNC, TGFB1. These ECM-r genes spatially overlapped with the stromal regions of individual tumors (Figure 1B). Furthermore, gene composite analysis demonstrated spatial colocalization of macrophages and cancer-associated fibroblasts (CAFs) within the ECM-r regions of the tumors. Spatially informed cell type deconvolution revealed EwS tumor regions that correlated well with histology. Moreover, majority of macrophages/monocytes in L-EwS tumors colocalized with CAFs. Deconvolution analysis also uncovered previously unreported B cell infiltrates and CD8 T cell infiltrative tumors. Spatial cellular microenvironment was estimated using NICHES (Niche Interactions and Cellular Heterogeneity in Extracellular Signaling) and revealed immune microenvironmental signaling throughout all tumors and unique extracellular matrix (ECM) microenvironmental signaling predominantly in L-EwS compared to M-EwS (p-value < 0.05) (TGF-B, cell adhesion and ECM remodeling signals).

Conclusion: L-EwS tumors are enriched in ECM and ECM-related microenvironmental signals compared to M-EwS. Spatial cellular microenvironmental signals reveal that components of the ECM may serve as anchors for tumors to eventually metastasize, nutrients for EwS to proliferate, and most importantly as a communication network with host immune cells. Further studies will be performed on a novel immunocompetent zebrafish genetic EwS model to identify conserved mechanisms of host-tumor interactions and further dissect the role of ECM in EwS.