Gene and isoform transcriptomics to uncover distinct immune and signaling programs across sarcoma subtypes.

Abstract

e23538 Background: Sarcomas are heterogeneous malignancies arising from diverse mesenchymal lineages. While transcriptomic profiling has revealed distinct signaling dependencies, the contribution of alternative splicing and transcript isoform usage to sarcoma heterogeneity remains poorly defined. We investigated how gene-level and isoform-level transcriptomics jointly shape the molecular and immune landscapes of angiosarcoma (AS, n = 26) and leiomyosarcoma (LMS, n = 11). Methods: Bulk RNA sequencing from a multi-sarcoma cohort was analyzed at both gene and isoform resolution. Unsupervised clustering was performed, followed by cluster annotation using: (i) protein–protein interaction (PPI) network analysis, (ii) immune cell deconvolution using xCell 2.0. Cluster-level molecular programs were integrated to define functional and immune states. Results: Three molecular clusters with distinct functional, immune, and signaling characteristics were identified. Cluster 1 was enriched for AS of the liver, breast, and heart; Cluster 2 comprised head and neck AS; and Cluster 3 was specific to LMS. PPI analysis revealed cluster-specific biological programs: Cluster 1 was enriched for complement activation, endothelial development, and chemokine signaling; Cluster 2 for keratinization-related structural processes; and Cluster 3 for muscle differentiation and stress-response programmes. Immune profiling demonstrated divergent tumor immune microenvironments. Cluster 1 showed enrichment of PD-1–high CD8⁺ T cells and plasma cells, consistent with an immune-infiltrated but exhausted phenotype. Cluster 2 exhibited broad enrichment of adaptive immune populations, including CD4⁺ T cells, Th1/Th17 cells, B cells, and regulatory T cells, indicative of an immune-inflamed and functionally active state. In contrast, Cluster 3 demonstrated minimal immune cell enrichment, consistent with an immune-cold or immune-excluded phenotype. Isoform-level analysis revealed organ-specific patterns within AS. Each primary site demonstrated higher expression of distinct isoforms within its corresponding cluster. Notably, breast and cardiac AS in Cluster 1 preferentially expressed canonical oncogenic isoforms associated with DNA repair and genome stability, cell-cycle regulation, apoptosis, and transcriptional control, compared with liver AS. Conclusions: This integrative gene- and isoform-level transcriptomic analysis identifies functionally distinct sarcoma subtypes driven by differences in immune contexture and signaling pathway activity. Isoform-specific expression patterns further highlight potential therapeutic vulnerabilities, particularly in primary breast and cardiac AS. These findings provide a framework for functional stratification of sarcomas and support the incorporation of isoform-level analyses into future translational and biomarker-driven studies.