Identification and validation of molecular subtypes' characteristics in bladder urothelial carcinoma based on autophagy-dependent ferroptosis.

ABSTRACT

Background: Nowadays, more evidences indicated that autophagy-dependent ferroptosis regulatory molecules (ADFRMs) may be closely related to various tumors. In current study, we intended to establish a prognostic ADFRMs signature and investigated its potential roles in bladder urothelial carcinoma (BLCA). Methods: Two distinct clusters were determined by consensus clustering with expression of 119 identified ADFRMs in BLCA. The tumor microenvironment was investigated through "CIBERSORT" algorithm, and enrichment analyses were utilized to seek molecular characteristics of differentially expressed genes (DEGs) between clusters. Moreover, a 2-ADFRMs prognostic signature including TRIB3 and WIPI1 was identified in TCGA cohort and further evaluated in the GSE13507 cohort. The qRT-PCR was conducted to examine the expression of prognostic genes. Further, the risk score was gained through calculating the level of TRIB3 and WIPI1 expression through the coefficient. The correlations between risk score with clinicopathologica features, tumor microenvironment, and drug sensitivity were explored. Results: Patients in TCGA-BLCA were grouped into two clusters with different expression patterns of ADFRMs. And the overall survival, tumor microenvironment and biological functions were significant different between two clusters. Moreover, a 2-ADFRMs model was constructed, and patients were separated into a low-risk and high-risk group. Survival analysis indicated patients with low risk promised a good prognosis, suggesting the risk score determined with ADFRMs signature exhibited an acceptable capacity for survival prediction in BLCA. Correlation analysis demonstrated risk score had close ties with age, stage, and tumor microenvironment. In vivo, the expression of prognostic genes was identified to be up-regulated in BLCA cell line T24. Conclusion: The constructed 2-ADFRMs signature was a promising model to predict prognosis and correlated with tumor microenvironment, which had latent clinical value in the intervention for BLCA.