Abstract / Summary
Abstract Background Bladder cancer (BLCA) is a malignant tumor characterized by a high progression rate and a high incidence of muscle-invasive disease, with over 50% of patients experiencing recurrence or progression following treatment. Given that recurrence poses a major threat to the prognosis of BLCA, improving the precision of clinical management and identifying effective therapeutic targets is crucial. To address this, we construct a molecular framework based on tumor recurrence-related genes (TRGs) to achieve precise molecular subtyping and prognostic prediction for BLCA patients, while providing potential targets for clinical therapy. Methods Transcriptomic and clinical data were obtained from the TCGA and GEO databases. Differential expression analysis and weighted gene co-expression network analysis (WGCNA) were used to identify specific tumor recurrence-related genes (TRGs) between primary and recurrent BLCA samples. Potential BLCA molecular subtypes were determined through clustering analysis based on these genes, and the associations between different subtypes, patient prognosis, and immune cell infiltration were evaluated. Subsequently, multiple combined machine learning algorithms and multivariate Cox regression analysis were applied to construct a prognostic model (tumor recurrence index, TRI), followed by an assessment of the relationships between TRI and clinicopathological features, immune microenvironment, and immunotherapy response. Finally, the function of the key gene EIF1 in the model was validated through both in vivo and in vitro experiments. Results We initially identified 280 key genes associated with bladder cancer recurrence. Clustering analysis divided BLCA into two TRG-based molecular subtypes, among which the R2 subtype exhibited poorer overall prognosis, a more immunosuppressive tumor microenvironment, and enhanced oncogenic pathway activation. The prognostic model TRI was confirmed as an independent prognostic indicator for BLCA, demonstrating superior predictive accuracy compared with other clinical features and previously published prognostic models, and showing stable performance across multiple independent cohorts. Patients with high TRI scores displayed worse overall survival, a more immunosuppressive microenvironment, and unfavorable responses to immunotherapy. Functional experiments further validated that EIF1 was upregulated in recurrent BLCA tissues and cancer cell lines. Moreover, silencing EIF1 markedly inhibited BLCA cell proliferation and migration, as well as tumor growth in vivo. Conclusion This study developed and validated a novel prognostic model that integrates tumor recurrence–related processes and offers significant prognostic value in BLCA management. Moreover, the identification of EIF1 as a novel biomarker associated with bladder cancer recurrence offers a potential therapeutic target for managing recurrence.