Development and experimental validation of a folate metabolism-related gene signature to predict the prognosis and immunotherapeutic sensitivity in bladder cancer.

Abstract

Folate metabolism is critical for the maintenance of genomic stability due to its regulatory ability to methylation, nucleotide metabolism, and reduction capabilities in cancer cells. However, the prognostic value of folate metabolism-related genes has not been clarified, especially in bladder cancer (BLCA). 91 folate metabolism-related genes were retrieved from the public database. TCGA-BLCA cohort, obtained from the Cancer Genome Atlas, was selected for training, while GSE13507, GSE31684, and GSE32894, downloaded from the Gene Expression Omnibus, and 35 BLCA samples collected from the local hospital were used for external validation. Through genomic difference detection, protein-protein interaction network analysis, LASSO regression, and Cox regression, a three-gene signature, including ATIC, INS, and MTHFD1L, was constructed. The signature was a reliable prognosis predictor across multiple independent cohorts (pooled hazard ratio = 2.79, 95% confidence interval = 1.79-4.33). The signature was associated with the BLCA malignant degree, which was validated in the local clinical samples (P < 0.01) and multiple cell lines (all P < 0.05). Additionally, the TIDE algorithm, GSE111636 cohort, and IMvigor210 cohort indicated that the signature was a promising tool to evaluate the immunotherapeutic response. Collectively, a folate metabolism-related gene signature was constructed to predict the prognosis and immunotherapeutic sensitivity in BLCA, which was verified in multiple large-scale cohorts, clinical samples, and cellular experiments, providing novel insights into the biological mechanisms.