MUNICH — Edoxaban did not reduce the overall risk of stroke or systemic embolism in patients with atrial fibrillation (AF) who had previously experienced an intracranial hemorrhage (ICH), according to results of the randomized ENRICH-AF trial.

Although anticoagulation reduced ischemic stroke, this benefit was offset by an increase in hemorrhagic stroke and major bleeding, leaving no overall clinical benefit in an unselected population of ICH survivors with AF.

“Our findings do not support the use of edoxaban in unselected patients with AF after intracranial hemorrhage but highlight the need for an individualized decision-making approach,” said principal investigator and presenter Ashkan Shoamanesh, MD, of McMaster University in Hamilton, Ontario, Canada.

The findings provide much-needed randomized evidence for a challenging clinical population that has largely been excluded from landmark trials that established the benefits of oral anticoagulation for stroke prevention in AF, said Shoamanesh, who presented the findings at the European Society of Cardiology (ESC) Congress 2026.

Uncertainty After ICH

Oral anticoagulants are highly effective at preventing ischemic stroke in patients with AF. But their use after ICH poses a difficult trade-off: patients remain at high risk of thromboembolism while also facing an increased risk of recurrent intracranial bleeding.

“The ideal strategy to prevent ischemic stroke is uncertain in patients with AF and a prior intracranial hemorrhage,” Shoamanesh said.

Four previous small, randomized trials tested anticoagulation vs avoiding anticoagulation in this setting but collectively did not provide clear evidence of net benefit, he added.

ENRICH-AF was an investigator-initiated, prospective, open-label, blinded-endpoint, event-driven, randomized phase 3/4 trial conducted across 174 recruiting sites in 20 countries. 

The trial randomized 948 patients with high-risk AF and previous ICH to edoxaban (60 mg once daily, with standard dose reduction to 30 mg where indicated) or no anticoagulation. Of these, 944 patients were included in the analysis. Control patients could receive either single antiplatelet therapy or no antithrombotic treatment based on clinician discretion. Participants were a mean age of 77 years, 39% were women, and the median CHA₂DS₂-VASc score was 4. Approximately 52% of control patients received antiplatelet therapy. 

Ischemic Benefit Offset by Bleeding

Over a mean follow-up of 28 months, the primary endpoint of stroke or systemic embolism occurred in 11.8% of patients assigned edoxaban compared with 12.8% assigned to no anticoagulation (HR, 0.88; 95% CI, 0.61-1.26; P = .48). 

The individual components, however, revealed opposing effects. Ischemic stroke occurred in 6.2% of patients receiving edoxaban compared with 10.3% of controls, representing a 43% reduction (HR, 0.57; 95% CI, 0.36-0.90; P = .02). Conversely, hemorrhagic stroke occurred in 5.8% vs 1.9%, respectively, representing an approximately threefold higher risk (HR, 2.99; 95% CI, 1.41-6.37; P = .003). 

Myocardial infarction was also significantly reduced, occurring in one patient (0.2%) receiving edoxaban compared with seven (1.5%) in the control group (HR, 0.19; 95% CI, 0.02-0.88; P = .03). There were no significant differences in death, cardiovascular death, or the composite of stroke or cardiovascular death. 

International Society on Thrombosis and Haemostasis (ISTH) major bleeding occurred in 11.6% of edoxaban-treated patients compared with 5.2% of controls (HR, 2.23; 95% CI, 1.39-3.59; P < .001). Fatal bleeding occurred in 4.1% vs 0.8%, respectively (HR 4.62; 95% CI, 1.57-13.58; P = .002). 

The type of ICH appeared to be particularly important. Among patients with lobar ICH or convexity subarachnoid hemorrhage, the annualized rate of recurrent ICH was 7.4 per 100 patient-years with edoxaban compared with 3.2 with control. Among those with nonlobar ICH, corresponding rates were 2.4 and 1.0 per 100 patient-years. 

Shoamanesh cautioned that subgroup findings were hypothesis-generating. Overall, he said, edoxaban reduced ischemic stroke, but in unselected patients “this benefit was offset by excess major bleeding, including hemorrhagic stroke,” resulting in a lack of net clinical benefit. 

Individualized Decisions Remain Key

Discussing the findings, Renate B. Schnabel, MD, stressed that ICH “is not one disease,” with the risk of recurrence varying according to its underlying cause. She singled out lobar ICH, which may reflect cerebral amyloid angiopathy and carry a particularly high risk of recurrent bleeding. “We need to understand the individual risk of ischemic stroke and recurrent intracranial hemorrhage,” she said, emphasizing that these competing risks need to be weighed for each patient.

Despite ENRICH-AF being larger than all previous randomized trials in this setting combined, Schnabel said important questions remain about which patients to treat, when to start treatment, and which strategy to use. She highlighted left atrial appendage occlusion and factor XI/XIa inhibitors as approaches requiring further study, concluding that for now clinicians need to make “individualized decisions” that balance the absolute risks of thromboembolism against recurrent hemorrhage.

ENRICH-AF was an investigator-initiated trial supported by grants-in-aid from Daiichi Sankyo and the Population Health Research Institute. Shoamanesh has reported consulting/advisory, speaking, and/or research relationships with multiple pharmaceutical and biotechnology companies, including Daiichi Sankyo, Bayer, AstraZeneca, Bristol Myers Squibb/Pfizer, Servier, and others, as well as research support from governmental and nonprofit organizations. Schnabel has reported receiving research funding from the European Research Council, European Union Horizon 2020 and Horizon Europe programs, German Center for Cardiovascular Research, German Ministry of Research and Education, AFGen, Wolfgang Seefried project funding, and the German Heart Foundation. She has received lecture and advisory board fees from Bristol Myers Squibb/Pfizer and Bayer outside the submitted work.