Abstract
The development of bladder cancer (BLCA) is associated with mitochondrial dysfunction and neutrophil extracellular traps (NETs); however, the relationship between mitochondrial function and NET formation in BLCA remains poorly understood. In this study, BLCA datasets, along with mitochondria- and NET-related genes, were retrieved from public databases and existing literature. Differential expression analysis, weighted gene co-expression network analysis (WGCNA), and protein-protein interaction (PPI) networks were applied to identify prognostic genes. A prognostic model incorporating six key genes (CCDC80, NIBAN1, CSPG4, PDGFRA, MAP1A, and PCOLCE2) was established through machine learning methods and univariate Cox regression analysis. This model demonstrated strong prognostic accuracy for BLCA, further validated by a nomogram exhibiting excellent predictive performance. Using the established prognostic model, patient samples were stratified into high-risk (HRG) and low-risk groups (LRG). Significant differences in immune cell infiltration-including eosinophils and 24 other immune cell types-were observed between these groups. The risk scores strongly correlated with multiple immune cells, notably natural killer cells. Furthermore, immune checkpoint analysis revealed significant upregulation of only three checkpoint genes (TNFRSF14, TNFRSF25, and VEGFA) in the LRG. Additionally, fibroblasts were identified as key cells through analysis of the GSE222315 dataset, with prognostic gene expression varying significantly during fibroblast differentiation. Experimental validation via reverse transcription-quantitative polymerase chain reaction (RT-qPCR) confirmed that all six prognostic genes were significantly downregulated in BLCA clinical samples collected for this study. Overall, the study highlights six novel prognostic biomarkers and presents a robust predictive model, providing new insights into BLCA prognosis.