Genomic and transcriptomic profiling of high-risk bladder cancer reveals diverse molecular and microenvironment ecosystems

Abstract

Despite surgical resection, rigorous endoscopic surveillance, and immunotherapy with the Bacillus Calmette-Guerin (BCG) vaccine, 30% of high-risk bladder cancers recur, and 10% result in fatal outcomes within two years of diagnosis. The global shortage of BCG underscores the urgent need for alternative or complementary therapeutic strategies. To address this, we integrated transcriptomic profiling and targeted genomic sequencing to identify four consensus intrinsic subtypes of bladder cancer. Initially derived from bulk RNA profiling, these subtypes were further validated at the cellular and tissue-compartment levels using single-cell RNA sequencing and spatial transcriptomics. Notably, we identified a subtype of inflamed tumors with enhanced endogenous retroelement expression and increased commensal bacterial presence, which showed the highest responsiveness to BCG therapy. Additionally, we developed a machine learning-based model incorporating composite molecular features to predict recurrence risk, achieving a high accuracy (AUC = 0.90). Our findings establish a molecular precision framework for bladder cancer and nominate novel therapeutic targets to reduce reliance on BCG immunotherapy.