(P 399) CRITICAL MOLECULAR PATHWAYS ASSOCIATED WITH DIFFERENT HISTOPATHOLOGY FEATURES AND DETAILED PATTERNS OF IMMUNE CELL INFILTRATES IN EPITHELIOID HEMANGIOENDOTHELIOMA (EHE)

Objective: While the presence of specific gene rearrangements is pathognomonic for EHE, its histopathology is heterogeneous and affects prognosis. Critical molecular pathways associated with prognostic histopathological features of EHE remain elusive. Furthermore, understanding the context of the tumor immune microenvironment in EHE is also limited. We intended to uncover the critical molecular pathways based on histopathology features and the pattern of immune cell infiltrates in EHE.

Methods: The samples of EHE with different primary sites and gene rearrangements were selected. The spatial transcriptome of each FFPE sample was analyzed using the Visium slide. Integration was performed on four spatial transcriptome datasets, and graph-based clusters were identified. To calculate the pathway expression intensity of each spot, AUCell was applied to the Hallmark gene-set from MSigDB. Additionally, XCell was utilized to estimate the cell compositions of each spot.

Results: Four EHE patients (F:M 2:2, the median age at diagnosis 41) with two fusion types (CAMTA1 fusion in 3, TFE3 fusion in 1) and three different primary sites (liver in 2, extremity, pleura) were included. The mean profiled genes was > 18,000, and the spots covered by tissues ranged from 5,887 to 14,334. CD31 and CD34 were homogeneously expressed in the tumor parts, while ERG distribution was patchy. Among 22 identified spot clusters (Fig. 1), three clusters (supercluster A, SA) represented solid sheet pattern EHE tumor cells, and two clusters (supercluster B, SB) were in accordance with the locations of EHE tumor cells with papillary features and/or necrosis. Pathways associated with mTORC1 (Fig. 2), KRAS/MEK, angiogenesis, hypoxia, inflammation, interferon-alpha response, TGF-B were upregulated in SB compared to SA. NOTCH upregulation in SB was only found in CAMTA1-rearranged EHE. Between the sub-clusters of SB, cluster 6 was more commonly associated with necrosis and significantly upregulated pathways of mTORC1, angiogenesis, STAT3/5, epithelial-mesenchymal transformation, and KRAS/MEK pathways. Regarding immune cell infiltrates, B cells were the most prominent immune cell observed. While in CAMTA1-rearranged EHEs, the B cells were located at the invasive border of the tumor, diffuse B cell infiltration was observed in the liver EHE with TFE3 rearrangement (Fig. 3). More study details will be presented at CTOS.

Conclusion: Spatial transcriptomics showed that different histopathology components of EHE have different molecular pathway activation. EHE with different gene arrangements may have differential downstream molecular pathway activations and immune cell infiltration patterns. The differentially activated pathways or immune cells associated with specific prognostic features provides novel insights into selecting future treatments for EHE patients.