Background: Despite the increase in the use of direct oral anticoagulants (DOACs) for cancer-associated thrombosis, limited data exists regarding the safety of DOACs in patients with brain metastases or primary brain tumors. The purpose of this retrospective cohort study is to describe overall tolerability of DOACs compared to low-molecular weight heparin (LMWH) for the treatment of venous thromboembolism (VTE) or for stroke prevention in atrial fibrillation (AF), in patients with brain tumors or CNS metastases. Methods: This is a single center, institutional review board approved retrospective cohort study at New York University Langone Health of patients with primary brain tumors or CNS metastasis that received therapeutic dosed anticoagulation with either a DOAC or LMWH (enoxaparin) between December 1, 2018 and August 19, 2022. Electronic medical records were reviewed for pertinent data. The primary outcome was tolerability of anticoagulation, which included continuation of initial anticoagulation therapy and absence of any switches in anticoagulation, dose adjustments or temporary interruptions. Reasons for discontinuations included bleeding events, intolerance/patient preference (such as the need for subcutaneous administration or cost of therapy), provider preference, or a change in clinical status such as transition to hospice or unable to determine based on chart review. Secondary outcomes included the incidence of anticoagulation discontinuations secondary to an intracerebral hemorrhage (ICH). Results: A total of 153 patients were included (DOAC n = 105, 69%; enoxaparin n = 48, 31%). The study population was predominantly White (74%), male (59%), with a median BMI≥30 kg/m2 in 40% of the cohort and age of 65 (IQR 56-73) years; however, the enoxaparin group was younger than the DOAC group at 59.5 years vs. 67 years (p <0.001), respectively at the time of anticoagulation initiation. Primary brain tumors were the most prevalent in the total cohort at 89%, with enoxaparin most commonly prescribed at 71% versus a DOAC at 47%, p = 0.005. Other primary malignancies included anaplastic astrocytoma, glioma, anaplastic oligodendroglioma, and tanycytic ependymoma. Non-small cell lung cancer was the most common malignancy that resulted in CNS metastasis (7.2%). The most common indication for anticoagulation was for the treatment of VTE at 70%, followed by stroke prevention in AF at 28%, with the remainder 2% as off-label. The primary outcome of remaining on initial anticoagulant chosen without any anticoagulant therapy changes occurred in only 52% of the cohort. There were more changes in therapy in the enoxaparin cohort at 29, 60% vs. 44, 42% for the DOAC cohort, p = 0.033, driven by provider preference in 48% of the cohort and patient intolerance in 29% of the cohort. More patients in the enoxaparin cohort were switched to a DOAC (n = 17) and more patients in the DOAC cohort had their initial anticoagulation discontinued for non- ICH reasons (n = 14). The median time to any anticoagulation switch was 2.12 months (IQR 0.35-4.27 months). Discontinuations in patients who were still alive was primarily driven by completion of therapy, deemed by the treatment team, intolerance or patient preference, and provider preference (Table 1). Though the reason for provider preference was not always clearly elucidated upon chart review, this trend supports the movement towards increasing DOAC use. The secondary outcome of the incidence of ICH was also lower in the DOAC cohort compared to enoxaparin at 3.8 vs 14.6% (p = 0.037). (Figure 1) Conclusions: Our study demonstrates real world prescribing practice of anticoagulants for the treatment of VTE or stroke prevention in AF in patients with primary brain tumors or metastases. Although initial anticoagulants chosen were either discontinued or switched to an alternative agent, we observed a shift towards a preference for DOACs, similar to the shift towards DOACs for cancer associated thrombosis. We also observed a lower incidence of ICH with DOACs compared to LMWH. However, these results should be confirmed in a randomized controlled trial to account for the inherent selection biases we could not account for, which may have favored the initial anticoagulant chosen, the reason for switch, and overall tolerability.