The Genomic Landscape of Prostate Cancer Brain Metastases

Lethal prostate cancer commonly metastasizes to bone, lymph nodes, and visceral organs but with more effective therapies, there is an increased frequency of metastases to the brain. Little is known about the genomic drivers of prostate cancer brain metastases (PCBM). To address this, we conducted a comprehensive multi–regional, genomic, and targeted transcriptomic analysis of PCBM from 28 patients. We compared whole–exome and targeted RNA sequencing with matched primary tumors when available (n = 10) and with publicly available genomic data from non–brain prostate cancer metastases (n = 416). In addition to common alterations in TP53, AR, RB1, and PTEN, we identified highly significant enrichment of mutations in NF1 (25% cases (6/28), q = 0.049, 95% CI = 2.38 — 26.52, OR = 8.37) and RICTOR (17.9% cases (5/28), q = 0.01, 95% CI = 6.74 — 480.15, OR = 43.7) in PCBM compared to non–brain prostate cancer metastases, suggesting possible activation of the druggable pathways RAS/RAF/MEK/ERK and PI3K/AKT/mTOR, respectively. Compared to non–brain prostate cancer metastases, PCBM were almost three times as likely to harbor DNA homologous repair (HR) alterations (42.9% cases (12/28), p =0.016, 95% CI = 1.17 — 6.64, OR = 2.8). When considering the combination of somatic mutations, copy number alteration, and Large-–scale State Transitions, 64.3% of patients (18/28) were affected. HR alterations may be critical drivers of brain metastasis that potentially provide cancer cells a survival advantage during re-establishment in a special microenvironment. We demonstrate that PCBM have genomic dependencies that may be exploitable through clinical interventions including PARP inhibition.