Neuroendocrine mesenchymal transition associated biomarkers EPAS1 TGFBR2 and PVT1 have prognostic significance in metastatic non-small cell lung cancer

This study aimed to identify neuroendocrine–mesenchymal transition (NMT)-associated core biomarkers in metastatic-stage non-small cell lung cancer (MS-NSCLC) and explore their mechanisms and prognostic significance. Transcriptomic data from TCGA-LUAD and TCGA-LUSC were integrated. Cases with AJCC-M stage M1, M1a, or M1b were defined as MS-NSCLC. DESeq2 and WGCNA identified MS-NSCLC-related genes, intersected with NMT-related genes from GeneCards. GO, KEGG, and PPI analyses were performed. Five machine learning algorithms (univariate logistic regression, random forest, LASSO, XGBoost, Boruta) screened core biomarkers, validated externally in GSE30219. Kaplan–Meier and single-gene GSEA assessed prognosis and mechanisms. 9,127 DEGs were identified; the turquoise module was most MS-NSCLC-associated. Intersection yielded 30 NMT candidates, enriched in Hippo, PD-L1/PD-1, cell migration, and EMT. Five algorithms identified EPAS1, TGFBR2, and PVT1 as core biomarkers. EPAS1 and TGFBR2 were downregulated, while PVT1 was upregulated in training and showed similar trends in validation. In the MS-NSCLC subgroup with available survival data, EPAS1 and TGFBR2 did not show statistically significant prognostic associations, whereas exploratory analysis in the full integrated NSCLC cohort showed protective associations. GSEA suggested EPAS1 involvement in extracellular matrix remodeling, drug metabolism, and immune functions; TGFBR2 in amino acid metabolism and redox homeostasis. EPAS1, TGFBR2, and PVT1 are potential NMT-related core biomarkers in MS-NSCLC. EPAS1 and TGFBR2 showed relatively stable expression validation, but their prognostic relevance in MS-NSCLC requires cautious interpretation and further validation.