Homologous recombination deficiency signature (HRDsig) in adenocarcinoma of the bladder (ACB): Implications for PARP inhibitor (PARPi) treatment.

Abstract

853 Background: Pure and predominant non-urachal adenocarcinoma histologies in primary bladder carcinomas are uncommon. Given the interest in expanding the clinical utility of PARPi to new tumor types, we conducted a study of ACB to evaluate their genomic landscape stratified on HRDsig status. Methods: 328 clinically advanced ACB underwent hybrid capture-based comprehensive genomic profiling (CGP) to assess all classes of genomic alterations (GA). All patients had clinically advanced disease and were confirmed by central pathology review to be either pure or predominant adenocarcinoma histology at the time of CGP. Genomic ancestry, microsatellite instability (MSI) status, tumor mutational burden (TMB), HRDsig and cosmic trinucleotide signature were analyzed from the CGP data. PD-L1 expression was determined at percentage of tumor proportion score (%TPS) using the Dako 22C3 system. Statistical analysis utilized the Fisher Exact system with the False discovery rate (FDR) corrected using the Benjamini/Hochberg adjustment. Results: Of the 328 ACB cases, 13 (4.0%) were HRDsig-positive (HRDsig+), while 315 (96.0%) were HRDsig-negative (HRDsig-). There were more men with HRDsig+ ACB (92.3% vs 59.7%; NS) and the median age was relatively similar between the groups (median 63 vs 67). Genomic ancestry distribution was similar with a trend towards higher frequencies of EUR ancestry in the HRDsig+ ACB (84.6% vs 66.2%; NS) and AFR ancestry in the HRDsig- ACB (19.7% vs 7.7%; NS). MSI high status was extremely rare in both groups (0.0% vs 1.3%) and the median TMB levels were also similar (3.8 vs 3.5 mutations/Mb) in HRDsig+ and HRDsig-, respectively. An MMR signature was more frequent in the HRDsig+ ACB (15.4% vs 5.4%; NS). HRDsig- ACB over-expressed PD-L1 more frequently than HRDsig+ ACB (15.2% vs 0.0%; NS). Higher frequencies of GA associated with HRDsig+ status were identified in the HRDsig+ ACB but did not reach statistical significance: ATM (15.4% vs 6.0%), BRCA1 (7.7%vs 1.0%), BRCA2 (7.7% vs 1.9%), RAD21 (10.0% vs 4.8%), RAD51C (7.7% vs 0.0%). ERBB2 were uncommon (7.6% in HRDsig- and 0.0% in HRDsig+; NS). KRAS (28.9% vs 0.0%) and TP53 (81.3% vs 76.9%) GA were more frequent in the HRDsig- ACB (NS), whereas GA in PTEN (23.1% vs 4.8%) and MET (7.7% vs 2.2%) were more frequent in the HRDsig+ ACB. Conclusions: At a 4.0% frequency, HRDsig+ status is uncommon in ACB but associated with trends towards particular genomic features, which have the potential to guide future clinical trial designs testing PARPi and other agents in this challenging histology. Limitations include the retrospective nature, lack of clinical outcomes annotation, selection and confounding biases.