Construction and Validation of a 15-Top-prognostic-gene-based Signature to Indicate the Dichotomized Clinical Outcome and Response to Targeted Therapy for Bladder Cancer Patients

The clinical outcome of heterogeneous bladder cancer (BCa) is impacted by varying molecular characteristics and clinical features, and new molecular classification is necessary to recognize patients with dichotomized prognosis. We enrolled a total of 568 BCa patients from the TCGA-BLCA and GSE13507 cohorts. A total of 107 candidate genes, which were mostly involved in the extracellular matrix-associated pathway, were first selected through the consensus value of the area under the receiver operating characteristic curve (AUC). Furthermore, absolute shrinkage and selection operation regression analysis was implemented to reveal the 15 genes and establish the prognostic signature. The newly defined prognostic signature could precisely separate BCa patients into subgroups with favorable and poor prognosis in the training TCGA-BLCA cohort (p < 0.001, HR = 2.41, and 95% CI: 1.76-3.29), as well as the testing GSE13507 cohort (p < 0.001, HR = 7.32, and 95% CI: 1.76-3.29) and external validation E-MTAB-4321 cohort (p < 0.001, HR = 10.56, 95% CI: 3.208-34.731). Multivariate Cox analysis involving the signature and clinical features indicated that the signature is an independent factor for the prediction of BCa prognosis. We also explored potential targeted therapy for BCa patients with high- or low-risk scores and found that patients with high risk were more suitable for chemotherapy with gemcitabine, doxorubicin, cisplatin, paclitaxel, and vinblastine (all p < 0.05), but anti-PD-L1 therapy was useless. We knocked down HEYL with siRNAs in T24 and 5,637 cells, and observed the decreased protein level of HEYL, and inhibited cell viability and cell invasion. In summary, we proposed and validated a 15-top-prognostic gene-based signature to indicate the dichotomized prognosis and response to targeted therapy.

Automated summary

By studying the genetic characteristics and clinical features of bladder cancer patients, we have established a prognostic signature to predict patients' prognosis and response to targeted therapy. This signature was validated in different cohorts and showed independent predictive ability. We also identified potential targeted therapy options to meet the needs of patients with different risk scores.