ARID1A-deficiency in urothelial bladder cancer: No predictive biomarker for EZH2-inhibitor treatment response?

Bladder cancer therapy relies on aggressive treatments highlighting the need for new, targeted therapies with reduced side effects. SWI/SNF complexes are mutated in ~20% across human cancers and dependency of SWI/SNF-deficient tumors on EZH2 has been uncovered recently. To systematically dissect the frequency of genetic alterations in SWI/SNF complexes potentially contributing to their inactivation, mutations and copy number variations in 25 SWI/SNF subunit genes were analyzed making use of publicly available sequencing data for 408 muscle-invasive bladder carcinoma samples. ARID1A truncating mutations were identified as the by far most common alterations of SWI/SNF complexes in urothelial bladder cancer. As current ARID1A protein expression data in bladder cancer are inconsistent and incomplete we examined if the frequency of truncating ARID1A mutations translates into a similar frequency of cases showing ARID1A protein loss. We applied a validated ARID1A antibody conducting a comprehensive immunohistochemistry-based expression analysis in urothelial bladder cancer (n = 362) including carcinoma in situ (CIS) cases. While observing increased median ARID1A protein levels in all carcinoma subgroups compared to normal urothelial controls (n = 21), the percentage of cases showing ARID1A protein loss was positively correlated with increasing stage and grade culminating in a rate of 14.1% in muscle-invasive disease. ARID1A-depletion did neither increase EZH2 protein or trimethylated H3K27 levels in vitro nor did ARID1A expression correlate with EZH2 or H3K27me3 amounts in human bladder carcinomas. Importantly, ARID1A-deficiency was neither associated with enhanced sensitivity towards inhibition of EZH2 enzymatic activity nor depletion of EZH2 protein. In summary, ARID1A truncating mutations, potentially translating into ARID1A protein loss in a subset of high-grade bladder cancers, are the most common SWI/SNF genetic alterations in bladder cancer. Our data do not support ARID1A-deficiency as predictive biomarker for EZH2-inhibitor treatment response in bladder cancer underlining the need for future bladder cancer-specific, drug screens for successfull discovery of ARID1A-deficiency-based targeted drugs.