Identification and validation of a 9-RBPs-related gene signature associated with prognosis and immune infiltration in bladder cancer based on bioinformatics analysis and machine learning.

Background: Bladder cancer (BLCA) is the most common type of malignancy affecting the urinary tract, characterized by high recurrence rates, propensity for progression, metastatic potential, and multidrug resistance, all of which ultimately contribute to an unfavorable prognosis. RNA-binding proteins (RBPs) play a critical role in cancer development and have been associated with the progression and prognosis of the disease. However, comprehensive investigations into the biological functions and molecular mechanisms of RBPs in BLCA remain limited. The study aims to explore the relationship between RBPs and prognosis in BLCA, and to develop and validate an RBPs-based prognostic signature, providing new insights for the diagnosis and treatment of BLCA. Methods: Clinical data and RBPs expression profiles of BLCA patients were sourced from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO). A systematic bioinformatics analysis was conducted to identify differentially expressed RBPs and assess their prognostic significance. The optimal predictive model was selected by integrating multiple machine learning algorithms, enabling the identification of hub genes associated with BLCA prognosis and developing an RBP-related gene signature. To evaluate the prognostic signature's efficacy, survival curves and receiver operating characteristic (ROC) curves were generated. A nomogram was constructed and validated to predict the survival of BLCA patients at 1, 3, and 5 years. Furthermore, analyses of immune infiltration and gene set enrichment analysis (GSEA) were conducted to explore the roles of RBPs in immune cell interactions and elucidate underlying biological pathways. Results: A prognostic signature was effectively developed using nine RBPs (OAS1, MTG1, DUS4L, IGF2BP3, NOL12, PABPC1L, ZC3HAV1L, TRMT2A and TRMU), represented as risk score, through the integration of 13 combinatorial machine learning algorithms. Kaplan-Meier analysis revealed that the high-risk group exhibited a significantly poorer overall survival (OS) probability compared to the low-risk group. The areas under the ROC curves for the risk score model at 1, 3, and 5 years were 0.661, 0.655, and 0.676, respectively. The nomogram, which integrated clinical characteristics and risk scores, demonstrated robust prognostic accuracy. Furthermore, single-sample gene set enrichment analysis (ssGSEA) demonstrated significant correlations between both the risk score model and hub RBPs with the immune status of BLCA patients. GSEA indicated that major signaling pathways enriched in the high-risk group included extracellular matrix (ECM) components and interaction, as well as cytokine and receptor interaction. Conclusions: This study successfully identified and developed a prognostic signature based on nine RBPs, accompanied by a nomogram for predicting survival probability in BLCA patients. Our findings demonstrate that these nine RBPs function as significant biomarkers for forecasting the prognosis and immune status in BLCA, suggesting their potential as therapeutic targets for BLCA.