Identification of a prognostic signature for bladder cancer associated with the epithelial-mesenchymal transition hallmark pathway using the TCGA Database

• Two muscle-invasive bladder cancer subtypes (C1, C2) identified via NMF using TCGA data. • Identified subtypes differed in prognosis, clinical and immunological features. • C2 subtype enriched in ECM/structure organization pathways, key EMT drivers. • Four-gene signature (SIGLEC6, NEFL, NTNG1, FOXC2) identified by regression analyses. • EMT four-gene signature effectively predicted bladder cancer prognosis. Bladder cancer is a common urological malignancy with poor prognosis, underscoring the need for robust prognostic biomarkers. This study investigated gene expression profiles of 194 epithelial-mesenchymal transition (EMT) hallmark genes in 411 muscle-invasive bladder cancer patients from The Cancer Genome Atlas (TCGA). Non-negative matrix factorization identified two subtypes (C1 and C2) with distinct prognostic, clinical, and immunological characteristics. C2 was enriched in EMT-related pathways, including extracellular matrix and structure organization. Differential expression analysis revealed 1,514 prognosis-associated genes. A four-gene prognostic signature— SIGLEC6, NEFL, NTNG1, and FOXC2 —was constructed using univariate, LASSO, and multivariate Cox regression models. Patients were stratified into high- and low-risk groups, with the signature demonstrating predictive value in survival analysis and ROC curves. External validation was performed using GEO data. This EMT-based four-gene signature offers a promising prognostic tool for bladder cancer patient stratification and potential therapeutic targeting.