Abstract
Background: Although neoadjuvant chemotherapy (NAC) has become the standard treatment option for muscle invasive bladder carcinoma (MIBC), its application is still limited because of the lack of biomarkers for NAC prediction. Methods: We conducted a territory multicenter real-world study to summarize NAC practice in China and its associated clinicopathologic variables with NAC response. Then, we developed and validated a robust gene-based signature for accurate NAC prediction using weighted correlation network analysis (WGCNA), the least absolute shrinkage and selector operation (LASSO) algorithm, a multivariable binary logistic regression model, and immunohistochemistry (IHC). Results: In total, we collected 69 consecutive MIBC patients treated with NAC from four clinical centers. The application of NAC in the real world was relatively safe, with only two grade Ⅳ and seven grade Ⅲ AEs and no treatment-related deaths being reported. Among these patients, 16 patients gave up surgery after NAC, leaving 53 patients for further analysis. We divided them into pathological response and non-response groups and found that there were more patients with a higher grade and stage in the non-response group. Patients with a pathological response could benefit from a significant overall survival (OS) improvement. In addition, univariate and multivariate logistic analyses indicated that tumor grade and clinical T stage were both independent factors for predicting NAC response. Importantly, we developed and validated a five-gene-based risk score for extremely high predictive accuracy for NAC response. Conclusion: NAC was relatively safe and could significantly improve OS for MIBC patients in the real-world practice. Our five-gene-based risk score could guide personalized therapy and promote the application of NAC.