Multinational cohort study of intracranial hemorrhage in patients with brain metastases receiving anticoagulation

Spontaneous intracranial hemorrhage (ICH) represents a common complication in patients with brain metastases.^1,2 As cardiovascular complications, including venous thromboembolism (VTE), are prevalent in cancer patients, many patients with brain metastases have an indication for therapeutic anticoagulation. The incidence rate of ICH with anti-coagulation is approximately 15% in patients with metastatic brain cancer.² Although anticoagulation increases the risk of any bleeding event, prior cohort studies suggest that the risk of ICH in patients with brain metastases is similar with or without anticoagulation.^1,2,3 While two small retrospective cohort studies suggest safety of direct oral anticoagulants (DOAC) compared to low-molecular-weight heparin (LMWH) in this population,^4,5 due to the limited sample sizes and wide Confidence Intervals (CI) of point estimates, the ICH profile of DOAC versus LWMH in patients with metastatic brain tumors remains uncertain. Data are also lacking regarding risk factors for ICH and clinical and radiological presentation of ICH, as well as outcomes following ICH including recurrent thrombosis and recurrent hemorrhage after re-initiation of anticoagulation. To address these knowledge gaps, we conducted a multinational cohort study to compare rates of hemorrhage among patients with brain metastases treated with DOAC or LMWH and evaluate outcomes including mortality, recurrent thrombosis, and hemorrhage.

The Anticoagulation in Brain Cancer (ABC) Study was a retrospective cohort study involving 12 academic and non-academic hospitals in Canada, Israel, Mexico, Switzerland, the Netherlands, and the United States. The study protocol was approved by local medical ethics committees. Informed consent was waived and the study was conducted in line with local regulations. We screened the records of all patients in the hemato-oncology or medical oncology departments at the study centers between January 1^st, 2014 and January 1^st, 2022 for eligibility. The study included adult patients with systemic solid cancer and brain metastases confirmed through pathology and imaging, respectively. Eligibility criteria were active cancer, defined as newly diagnosed or undergoing treatment, and therapeutic anticoagulation (both full dose and indicated dose reductions prescribed with therapeutic intent) with either DOAC or LMWH. Patients with ICH before the initiation of anticoagulation and those lacking any follow-up data were excluded. Repeat brain imaging during follow-up was not a prerequisite for inclusion. Study index was defined as the first day of concurrent anticoagulation and brain metastases diagnosis, and patients were followed for 12 months. An additional 90-day follow-up was conducted for patients who experienced anticoagulation-related ICH to evaluate management and outcomes. The primary outcome was spontaneous ICH confirmed by central adjudication of radiographic imaging. The cumulative incidence of any ICH, major ICH, overt ICH (i.e., any ICH excluding hemosiderin deposits and/or trace/unmeasurable ICH), VTE or stroke/ systemic arterial thromboembolism over 12 months with corresponding 95% CI was calculated, and was compared between anticoagulation groups with anticoagulation as a fixed variable for ICH. Hazard Ratios (HR) with corresponding 95% CI for any ICH and major ICH were calculated using a Cox proportional hazards model, with anticoagulation as time-dependent variable, and death and loss to follow-up as competing risks (Fine and Gray model).

A total of 505 patients were included in the study: 202 patients received DOAC and 303 received LMWH (Table 1). Lung and breast cancer were the most common primary tumors, with similar distributions between the two groups. VTE was the most frequent indication for anticoagulation, although atrial fibrillation was more common in the DOAC group. DOAC-treated patients often had brain metastases diagnosed while on chronic anticoagulation (median 278.5 days of anticoagulation before diagnosis), compared to a median of 131 days for the LMWH group. The median follow-up duration was 209 days in the DOAC group and 173 days in the LMWH group. Loss to follow-up occurred in 13.4% of DOAC patients and 12.5% of LMWH patients.

The 12-month cumulative incidence of spontaneous ICH was 11.1% (95% CI: 7.1-16.2) in the DOAC group (N=21) and 8.6% (95% CI: 5.6-12.4) in the LMWH group (N=23), yielding a HR of 1.08 (95% CI: 0.56-2.08) when anticoagulation was treated as a time-dependent variable (Figure 1). After adjustment for age, sex, cancer type, and high-bleeding-risk cancer treatments, the HR for ICH was 0.84 (95% CI: 0.41-1.70). For major ICH, the cumulative 12-month incidence was 1.7% (95% CI: 0.5-4.7) in the DOAC group and 2.7% (95% CI: 1.2-5.3) in the LMWH group, with an adjusted HR of 0.56 (95% CI: 0.14-2.24). The incidence of overt ICH was also similar: 8.5% (95% CI: 5.1-13.1) in DOAC-treated patients and 7.4% (95% CI: 4.6-10.9) in LMWH-treated patients, with a HR of 1.19 (95% CI: 0.62-2.29).

Mortality rates were high in both groups; 94 of 202 patients (46.5%) in the DOAC group and 154 of 303 patients (50.8%) in the LMWH group had died by the end of 12 months of follow-up. Three DOAC-treated patients (1.5%) and 2 LMWH-treated patients (0.7%) experienced an ICH leading to death within 24 hours. The 12-month incidence of VTE was 5.3% (95% CI: 2.6-9.4) in DOAC-treated patients and 5.5% (95% CI: 3.6-8.0) in the LMWH group. The 12-month incidence of arterial thrombosis was 1.6% (95% CI: 0.4-4.2) and 2.5% (95% CI: 1.0-5.2), respectively. The clinical presentation of ICH events was comparable between the DOAC and LMWH groups, with approximately half of the patients (12/22 [55%] and 10/20 [50%], respectively) presenting without any clinical emergency (Figure 2). The clinical course of ICH, classified according to our predefined criteria (Figure 2), demonstrated an uneventful immediate course in most patients, whether treated with DOAC (82%) or LMWH (95%). Radiological review showed that most ICH events were overt (73% in the DOAC group and 65% in the LMWH group) and primarily intra-tumoral (86% and 85%, respectively) (Online Supplementary Table S1). Following ICH, anticoagulation was continued in 19 of 42 patients (45%). In 7 (37%) of the 19 patients continuing anticoagulation immediately, the ICH was characterized by presence of hemosiderin residues only (indicative of remote prior bleeding). Approximately one-third of the patients who initially discontinued anticoagulation resumed it within 90 days (Online Supplementary Table S2). Recurrent ICH occurred in only one patient (2.4%) within 90 days post-ICH, while the cumulative incidence of VTE during this period was 19.1% (95% CI: 8.8-32.2), with no arterial thromboembolic events observed. Exploratory multivariable analyses identified several potential risk factors for ICH. The use of anticancer medication, a history of anticoagulant-associated ICH prior to the diagnosis of brain cancer, the presence of a concomitant bleeding disorder, concurrent use of aspirin, and primary tumor sites of melanoma or renal cell carcinoma were found to be significantly associated with ICH, albeit with notably wide corresponding CI (Online Supplementary Table S3).

Table 1.Patient characteristics stratified for type of anticoagulation.

The ABC Study represents the largest cohort to date evaluating ICH risk in patients with brain metastases treated with DOAC or LMWH. The study found no significant difference in ICH risk between the two anticoagulation groups, even after adjusting for confounders. These findings are consistent with previous studies^2,4,5 and a recent meta-analysis that included data from the ABC study, which found no significant difference in ICH risk between DOAC and LM-WH use in patients with metastatic brain cancer (relative risk 1.05, 95% CI: 0.71-1.56).⁶ In the ABC study, radiological and clinical outcomes were comparable between groups. Notably, patients presenting with hemosiderin residues continued anticoagulation without recurrent ICH, suggesting that such findings may not contraindicate therapeutic anticoagulation. This is consistent with prior studies showing low recurrent ICH rates in patients with minor bleeds who resumed anticoagulation.⁷ The high incidence of VTE post-ICH underscores the thrombotic risk in this population. More VTE events were observed in patients who resumed anticoagulation post ICH, which likely reflects confounding by indication, given the high baseline thrombotic risk in these patients. The observed association between aspirin use and increased ICH risk warrants further investigation, particularly in the light of contrasting data from a recent cohort study that found no increased ICH risk with anti-platelet therapy.⁸ We acknowledge the potential for residual confounding and the selection of patients with lower bleeding risk for DOAC treatment. The heterogeneity in patient characteristics, inherent to the retrospective study design, and loss to follow-up numbers are clear limitations. The current study lacks power to exclude increased risk of ICH with DOAC compared to LMWH, as funding restrictions did not permit the time it would take to reach the calculated sample size. Furthermore, we acknowledge that data on ICH presentation and course are limited by the low number of events, precluding definitive conclusions. Strengths include the analysis of anticoagulation type as a time-dependent variable, the use of a strict ICH definition, and showing radiological outcomes in the context of clinical parameters, such as presentation, course and management of ICH. In conclusion, our study indicates similar ICH profiles with LMWH and DOAC in patients with brain metastases. The variability in clinical presentation and course of ICH suggests that a subset of patients may safely continue anticoagulation after ICH. Validation of the identified clinical predictors of ICH, using standardized reporting and classification of ICH, is needed to support clinical decision-making in this setting.

Figure 1.12-month cumulative incidence of intracranial hemorrhage. Cumulative incidence of any intracranial hemorrhage (ICH) (A), major ICH (B) with direct oral anticoagulant DOAC (blue curve) or low-molecular-weight heparin (LMWH) (red curve) treatment. The shaded areas represent the 95% Confidence Interval. Major ICH was defined as volume ≥10 mL OR surgical intervention OR clinical symptoms, focal neurologic deficits or cognitive changes. Overt ICH was defined as any ICH excluding hemosiderin deposits and/or trace/unmeasurable ICH (indicative of prior intra/extra-tumoral hemorrhage).

Figure 2.Clinical presentation and course of any anticoagulation-associated intracranial hemorrhage. Severity of hemorrhage by percentage of patients at clinical presentation (A) and clinical course (B) of any spontaneous intracranial hemorrhage (ICH) in patients experiencing ICH while on direct oral anticoagulant (DOAC) (N=22) compared to those on low-molecular-weight heparin (LMWH) (N=20). Clinical severity of ICH was assessed using pre-specified criteria.⁹ None of the patients presented with an ICH event of category 4.

Footnotes

• Received November 14, 2024
• Accepted January 24, 2025

Correspondence

*JIZ and ALe contributed equally as senior authors

A. LEADER - leadera@mskcc.org

Disclosures

ALe reports honoraria from Leo-Pharma. ALu reports lecture fees and research funding from Pfizer, and fees for lectures and advisory boards from Bayer. BCO reports honoraria for lectures and advisory board paid to her institution from BMS, MSD, Merck, Ipsen, Roche, Pfizer, Novartis, Sanofi, and Janssen. GS reports personal fees for lectures and advisory boards from Bayer, Pfizer, Boehringer Ingelheim, and Sanofi. JIZ reports prior research funding from Incyte and Quercegen, consultancy fees from Calyx, Incyte, Bristol Myers Squibb (BMS), Regeneron, and the Med Learning Group, and advisory boards from Sanofi and CSL Behring. KMS reports research funding from the American Society of Hematology and the National Institute of Health not related to this study, paid to her institution. LBK reports research support from CSL Behring, Takeda, and Sanofi to her institution. MC reports grants from Pfizer and Leo Pharma paid to his institution, and personal fees from Bayer, BMS, Pfizer, Servier, Sanofi, Valeo, and Leo Pharma paid to his institution. All other authors have no conflicts of interest to disclose.

Contributions

Funding

This study was supported by a 2021 ISTH Small Grant and funded in part through the NIH/NCI Cancer Center Support Grant P30 CA008748.

Acknowledgments

We would like to thank Tzippy Shochat for her help in the statistical analyses, and the ISTH Scientific and Standardization Committee (SSC) on Haemostasis and Malignancy for supporting this project. A list of members of the ABC Study Group is available in the Online Supplementary Appendix.