Background: Homologous recombination repair (HRR) genes are mutated in about 20% of metastatic prostate cancer (mPC) patients. Of these, ATM and BRCA2 mutations exhibit stark differences in response to PARP inhibitors (PARPi). We aimed to investigate the underlying genomic and molecular features of ATM- and BRCA2-mutated mPC and if these contribute to divergent clinical outcomes or future treatment decisions.
Methods: Building upon our prior study, we examined a novel subset of mPC 1066 mPCs after exclusion of microsatellite instable (MSI) tumors to prevent elevation of bystander non-driver BRCA2 mutation. NextGen sequencing of DNA (592-gene and/or whole exome) and RNA (whole transcriptome) from mPCs including bone, lymph nodes, and liver metastases, was performed at Caris Life Sciences (Phoenix, AZ). mPCs were stratified by ATM (n= 91) or BRCA2 (n=102) mutations, as well as HRR-deficiency (HRD) based on mutations in 23 other HRR genes (n = 237), and HRR-proficiency (HRP) based on lack of mutations in any HRR genes (n=636). High tumor mutational burden (TMB-high) was defined as ≥10 mutations/Mb. The differential expression analyses were conducted by Limma. Real-world overall survival (OS) was determined from insurance claims data and Kaplan-Meier estimates were calculated.
Results: Mutations in ATM and BRCA2 were associated with worse OS compared to the HRP group (HR 0.4, 0.5; CI 0.305-0.81, 0.339-0.845; P < 0.001). ATM or BRCA2 mutations also predicted worse response to the AR-targeted abiraterone/enzalutamide therapies (HR 0.4, 0.3; CI 0.244-1.003, 0.172 – 0.6; P = 0.047, p < 0.001). ATM-mutated tumors exhibited significantly lower levels of TP53 mutations (11.1% vs 35.1% and 40.8% in BRCA2 and HRP, p < 0.01) and RB1 loss (2.1% vs 15.5% in BRCA2, p < 0.05), and harbored genomic amplification of genes in 11q13 (~10% vs ~1.5% in HRP, p < 0.001), which includes CCND1 and genes in the FGF family. BRCA2-mutated tumors were enriched for amplifications in PDCD1 (PD-1) (3.1% vs 0% in HRP, p < 0.005), and were more frequently TMB-High (7.1% vs 0.63% in HRP, p < 0.001). At the gene expression level, BRCA2-mutated mPC exhibited robust correlations with increased KCNJ3 and decreased ALDH1A3, while ATM mutated tumors exhibited decreased correlation with HES4. BRCA2-mutated mPCs were enriched in cell cycle signaling, a potential indicator of platinum sensitivity (Normalized Enrichment Score > 2, FDR < 0.001) but AMT mutated tumors were not enriched of these profiles.
Conclusions: We found that somatic ATM and BRCA2 mutations associated with differential OS outcomes, which may help tailor treatment decisions. At the molecular level, ATM- and BRCA2-mutated tumors exhibited differences in the landscape of co-occurring genomic and transcriptional features. These features unique to ATM or BRCA2 mutations can inform rationale for divergent treatment strategies and should be investigated.