Abstract / Summary
Abstract Purpose Head and neck squamous cell carcinoma (HNSCC) remains a major global health burden, with cervical lymph node metastasis (CLNM) being a key determinant of recurrence and survival. Accurate identification of occult nodal disease remains challenging, highlighting the need for reliable biomarkers. Methods Candidate genes associated with CLNM and epithelial–mesenchymal transition in HNSCC were identified through literature mining and evaluated in GSE65858 and TCGA datasets. Genes consistently associated with CLNM were used to derive a three-gene score, which was subsequently validated by quantitative RT-PCR in 63 treatment-naïve FFPE HNSCC samples. Associations with clinicopathological parameters were assessed using appropriate statistical tests. Results Literature mining identified 135 genes associated with lymph node metastasis. In GSE65858 (n = 240), 22 genes were significantly associated with nodal status, including 6 upregulated and 16 downregulated genes. TCGA validation (n = 515) identified 11 significantly dysregulated genes. IL15, BIRC5, and HNRNPA2B1 were consistently upregulated and associated with CLNM in both datasets. The three-gene score derived by computing the average gene expression of the 3 genes was significantly associated with CLNM and stage in GSE65858 (p = 0.002) and CLNM and tumor grade in TCGA (p = 0.001). In the institutional cohort, the score was associated with CLNM (p = 0.006), tumor size (p = 0.016), and pathological stage (p < 0.0001). ROC analysis demonstrated an AUC of 0.703 (95% CI, 0.573–0.832; p = 0.002) for discriminating node-positive from node-negative cases. Conclusion A three-gene score comprising IL15, BIRC5, and HNRNPA2B1 showed reproducible associations with CLNM across independent datasets and an institutional cohort. Larger studies are warranted to validate its utility for predicting nodal metastasis and improving risk stratification in HNSCC.