Intracranial hemorrhage after ischemic stroke in patients on direct oral anticoagulants: results from a prospective observational study

Approximately 10% of ischemic strokes occur in patients on oral anticoagulants (OAC), yet prospective data on hemorrhagic complications after recanalization therapies remain limited. We aimed to assess whether prior use of OAC increases the risk of intracranial hemorrhage compared to no anticoagulation in clinical routine. The prospective, multicenter RASUNOA-Prime study (clinicaltrials.gov, NCT02533960) enrolled patients with atrial fibrillation and acute ischemic stroke who were receiving a direct OAC (DOAC), a vitamin K antagonist (VKA), or no anticoagulant (non-OAC) at stroke onset. The primary endpoint was symptomatic intracranial hemorrhage (sICH) on follow-up imaging (≤ 120 h after admission), adjusted for thrombolytic therapy. Neuroimaging was centrally reviewed, blinded to treatment group. Among 2,737 patients (median age 79 years, 49% female) from 46 stroke centers (DOAC n = 1,066; VKA n = 695; non-OAC n = 976), stroke severity was lower in DOAC thanin non-OAC patients (median NIHSS 4 interquartile range (IQR) 2–9 vs. 6 3–13). Among those presenting within 4.5 h, thrombolysis was used less often in DOAC than in non-OAC patients (10.0% vs. 58.9%, p < 0.001), with longer door-to-needle time (+ 19 min). The proportion of patients with any hemorrhagic transformation on admission was similar between groups. Follow-up imaging and clinical data on sICH were available for 1,813 patients (66%). Symptomatic ICH occurred in 0.59% (95% CI 0.01–1.17) of DOAC, 1.09% (95% CI 0.14–2.03) of VKA and 1.18% (95% CI 0.37–1.99) of non-OAC patients. After adjusting for thrombolysis, the odds of sICH were comparable across anticoagulation groups (adjusted OR 1.46, 95% CI 0.36–5.92, p = 0.6 for DOAC; adj. OR 1.43, 95% CI 0.43–4.70, p = 0.56 for VKA). The risk of sICH was not increased with DOAC use compared to no anticoagulation after ischemic stroke, with or without thrombolysis. Although event rates were low and confidence intervals wide, the findings suggest non-inferiority but cannot exclude a modest increase in bleeding risk. Randomized trials are warranted to confirm safety in this population.