Clinical and genomic characterization of brain metastasis in bladder cancer.

PURPOSE: Bladder cancer brain metastasis (BrM) remains a poorly characterized clinical entity in neuro-oncology. The current study combines clinicopathologic and genomic sequencing data to identify prognostic factors in bladder cancer BrM. METHODS: BrM patients were identified from a single-center retrospective cohort study at an NCI-designated Comprehensive Cancer Center consisting of 1,847 patients who were diagnosed with bladder cancer and received a brain magnetic resonance imaging between January 1, 2000 and December 31, 2024. The study cohort did not consist of an unselected population of all patients with bladder cancer. Genomic data was obtained from the Memorial Sloan Kettering-Integrating Mutation Profiling of Actionable Cancer Targets clinical sequencing platform. A multivariable Cox proportional hazards regression model was constructed by including baseline covariates that were associated with overall survival from univariable analysis at p < 0.05. RESULTS: 133 patients (median age at BrM diagnosis = 69.5 years [range 34.8-92.7], 99 [74.4%] male, 34 [25.6%] female) were identified for this study. The median OS from BrM diagnosis was 3.85 months (95%CI = 2.79-4.87 months). 1- and 3-year OS rates were 19.04% (95%CI = 12.63-26.44%) and 11.21% (95%CI = 6.20-17.87%), respectively. Small cell bladder cancers represented 17.3% of bladder cancer BrMs. In multivariable analysis, presence of ≥ 4 parenchymal brain metastasis (HR = 2.01, 95%CI = 1.32-3.05, p = 0.001) and immune checkpoint inhibitor (ICI) therapy prior to BrM diagnosis (HR = 2.26, 95%CI = 1.45-3.53, p = 0.0003) were associated with increased risk of mortality, with prior ICI therapy potentially serving as a surrogate for late-stage disease or prior treatment failure. Comparison of genomic alterations between non-brain metastatic bladder cancers (n = 1,966) and bladder cancer BrMs (n = 17) identified genomic events in MSH6, RBM10, PTPRT, and FAT1 to be significantly enriched in BrMs. A subsequent analysis of 8 matched primary and BrM samples revealed acquired missense mutations in one or more of RBM10, SF3B1, ARID1A, and FAT1, which were not present in the primary tumor. Of these, FAT1 alterations were most frequently acquired in BrM samples (n = 3). CONCLUSION: There is an urgent need for improved treatment in bladder cancer BrM given the poor outcomes and limited prognostic variables described in our large cohort.