Prevalence and predictors of brain metastases in metastatic prostate cancer.

e17046 Background: Brain metastases (BM) in metastatic prostate cancer (mPCa), although rare, are associated with a poor prognosis. Accurate predictors of BMs in mPCa remain largely unknown. This study aims to investigate the prevalence of BMs in mPCa and identify significant clinical and pathological predictors associated with their occurrence. Methods: The National Cancer Database (NCDB) was queried for patients diagnosed with prostate cancer (PCa) from 2004-2021. Patients with cM1 PCa and without missing patient and tumor characteristics were identified. Kaplan-Meier analysis and log-rank test compared overall survival (OS) between patients with and without BMs. Univariable and multivariable logistic regression assessed significant factors associated with the presence of BMs. Results: Among 101,900 patients with cM1 PCa, 1,144 (1.1%) patients had BMs. After excluding missing data, 6,654 patients remained, with 46 (0.7%) having BMs. In chi-square analysis, patients with BMs more often had cT4 tumors (36.8% vs 24.6%, p<0.001), cN+ status (51.9% vs 42.8%, p<0.001), neuroendocrine histology (6.8% vs 1.6%, p<0.001), and higher Gleason grade group (GGG) (p=0.017). Patients with BMs more frequently had liver (17.2% vs 4.1%, p<0.001) and lung mets (27.0% vs 8.0%, p<0.001), while bone mets were less common (80.3% vs 89.8%, p<0.001). Median OS was lower for patients with BMs (15.08 months vs 32.59 months, p<0.001), despite having younger median age. In multivariable analysis, neuroendocrine histology (aOR 3.916, p=0.037), GGG≥3 (aOR 3.699, p=0.016), and liver mets (aOR 3.855, p=0.001) were associated with increased likelihood of BMs. Tumor stage, bone mets, and lung mets were not significantly associated the likelihood BMs. Conclusions: We report specific tumor characteristics associated with BMs, particularly neuroendocrine histology, higher GGG, and liver mets. This analysis is the largest to date to investigate this; however, future studies are needed for validation and further elucidation of patient, tumor, and molecular predictors of BMs in mPCa. Multivariable logistic regression for the presence of brain metastases. Variable aOR (95% CI) p-value Histology (Ref = Adeno) Neuroendocrine Other 3.916 (1.086-14.127)4.557 (0.597-34.819) 0.037 0.144 Gleason (Ref = GG<3) GG≥3 3.699 (1.279-10.699) 0.016 Bone Mets (Ref = No) Yes 0.549 (0.288-1.045) 0.068 Liver Mets (Ref = No) Yes 3.855 (1.681-8.841) 0.001 Lung Mets (Ref = No) Yes 1.833 (0.849-3.958) 0.123 Other Mets (Ref = No) Yes 2.472 (1.103-5.538) 0.028