Proliferation- and cytotoxic immune signatures identify chemotherapy-responsive bladder tumors in a molecular subtype dependent manner.

Neoadjuvant cisplatin-based chemotherapy (NAC) followed by radical cystectomy (RC) has been the standard care for muscle-invasive bladder cancer (MIBC) for two decades. One third of NAC-treated patients achieve pathologic complete response (pT0N0), a proxy for improved survival after RC. Predicting response already at transurethral resection of bladder tumor (TUR-BT) would enable selective use of NAC, minimizing unnecessary toxicity and exposure to ineffective therapy. We aimed to identify tumor mRNAs predicting response across multiple transcriptomic studies, prioritizing subsequent biomarkers for validation. Three NAC-treated and one RC-only cohort with tumor transcriptomic profiles were included. Differential mRNA-expression analysis and subtype classification according to the Lund Taxonomy were performed. Within each cohort and molecular subtype, genes were ranked by differential expression, and ranks were integrated across cohorts into a meta rank-score. Its predictive value for treatment response was assessed by resampling. Survival associations of top genes in the NAC- and RC-only cohorts were compared to select candidate biomarkers for protein level validation. Proliferation/late cell-cycle gene expression predicted NAC response within the Urothelial-like subtype and expression of cytotoxic T- and NK-cell-related genes predicted response in Basal/Squamous tumors. These findings were validated by immunostainings for CCNB1 and NKG7, respectively. This integrative framework suggests a complex picture in which two main NAC-predictive signals, the proliferation and the T/NK-cell signatures, identify responders in a subtype dependent manner. Meta rank-scores additionally suggest new candidate biomarkers in each Lund Taxonomy subtype. The applied framework can be updated as new datasets become available, providing dynamic exploration of NAC-predictive biomarkers in MIBC.