Identification of Small Extracellular Vesicle (sEV)-Related Genes andthe Establishment of Diagnostic Models for Esophageal Cancer

Introduction: Esophageal cancer (EC) remains a global health challenge with high incidence and mortality. While sEV-related genes (ERGs) are emerging as key players in the tumor microenvironment, their diagnostic and pathogenic roles in EC are still poorly characterized. Methods: This study employed integrated bioinformatics approaches to identify ERGbased diagnostic biomarkers and molecular mechanisms in EC. Four GEO datasets were analyzed to identify differentially expressed genes (DEGs), which were cross-referenced with ERGs from GeneCards and the literature. Functional enrichment analyses (GO/KEGG), Cytoscape network analysis, and machine learning algorithms (LASSO, SVM, random forest) were used to prioritize diagnostic markers. Immune infiltration patterns were assessed via ssGSEA, and key genes were validated by qRT‒PCR in EC and normal cells. Results: We identified 456 DEGs, including 34 sEV-related ERDEGs enriched in bacterial defense response, secretory granule lumen, calcium-dependent protein binding, and pathways such as proteoglycans in cancer and IL-17 signaling. GSEA highlighted cell cycle, ECM-receptor interaction, and DNA replication as central mechanisms. Discussion: Machine learning pinpointed EpCAM and RUVBL1 as robust diagnostic biomarkers, validated by their elevated expression in EC cell lines. Immune infiltration analysis revealed correlations between these genes and 20 immune cell types, underscoring their relevance to the tumor microenvironment. Conclusion: EpCAM and RUVBL1 are promising diagnostic biomarkers for EC, linking sEV-mediated pathways to disease progression. These findings provide insights into EC pathogenesis and highlight potential therapeutic targets.