and are potentially important biomarkers for early diagnosis and treatment of esophageal squamous cell carcinoma: based on WGCNA and machine learning analysis.

BACKGROUND

Esophageal squamous cell carcinoma (ESCC) does not have distinct and highly sensitive biomarkers, making its diagnosis difficult. Consequently, identifying dependable biomarkers is critical, as these indicators can facilitate accurate ESCC diagnosis and enable effective prognostic evaluation.

METHODS

ESCC datasets (GSE29001, GSE20347, GSE45670, and GSE161533) were sourced from the GEO, and the Limma package identified differentially expressed genes (DEGs). To characterize co-expression network, weighted gene co-expression network analysis (WGCNA) was performed, allowing for the identification of relevant co-expression modules. To assess the biological pathways of intersecting genes, we performed pathway enrichment analysis using Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO). The Support Vector Machine Recursive Feature Elimination (SVM), along with Least Absolute Shrinkage and Selection Operator (LASSO) regression, was applied to identify clinical biomarkers. Finally, the differences of immune cell infiltration were also detected.

RESULTS

1,019 genes were derived by integrating DEGs with co-expressed module genes. KEGG and GO revealed a strong association between these genes and processes such as chemotaxis and IL-17 signaling pathways. Two hub genes ( and ) were selected through LASSO regression and SVM. Additionally, ROC curve analysis confirmed their potential for reliable diagnostic performance. Furthermore, differences in immune cell infiltration were observed.

CONCLUSION

Collectively, and emerged as promising candidate biomarkers, offering novel insights to enhance early detection and guide targeted treatment strategies for ESCC.