Abstract: Objective To identify the differentially expressed circular RNAs(circRNAs)and construct a circRNAs-mediated ceRNA network and a prognostic assessment model. Methods Data were downloaded from Gene Expression Omnibus(GEO)and The Cancer Genome Atlas(TCGA)database. Differential expressed analysis was performed to explore the differentially expressed RNAs in bladder cancer tissues with the standard of |log2 fold change(log2FC)|>1, P<0.05. Based on multiple detabases, such as TCGA and CircInteractome, the circRNA-miRNA-mRNA ceRNA network was constructed, and functional enrichment analysis of mRNAs was conducted. Subsequently, the relationship of mRNAs in ceRNA network was analyzed through the String database. The mRNAs related to prognosis were screened with univariate Cox regression model, and a prognostic assessment model was constructed. A ROC curve was drawn with “time ROC” package to evaluate the effectiveness of the model. Results Differential expressed analysis screened out 4 circRNAs, including hsa_circ_0102787, hsa_circ_0103114, hsa_circ_0102402 and hsa_circ_0104387. A ceRNA network was constructed, which contained 4 circRNAs, 23 miRNAs and 86 mRNAs. Functional enrichment analysis showed those mRNAs participated in several cancer-related pathways. The interaction between mRNAs in the ceRNA network was visualized by PPI network. Survival analysis revealed that 6 mRNAs were related to the prognosis of bladder cancer. Then, a prognostic assessment model was constructed based on “the risk score=(-0.356 9)×ABRACL+(-0.264 4)×MAP3K8.” Kaplan Meier analysis showed patients with high risks had a poor prognosis(P<0.001), and the AUC of the five-year overall survival rate was 0.744. Conclusion This study identififed 4 differentially expressed circRNAs, and constructed a circRNA-miRNA-mRNA network and a prognostic assessment model, which provides a new insight for the exploration of the occurrence and development of bladder cancer.

Key words: Bladder cancer, circRNAs, ceRNA, Prognosis, Cox regression model