Association of antithrombotic therapy with cerebral microbleeds: a cross-sectional, multicentre analysis

Background Cerebral microbleeds (CMBs) are associated with an increased risk of stroke, intracranial haemorrhage, and cognitive decline. Whether antithrombotic therapy influences CMB presence remains unclear. Meta-analyses suggest a preferential association of antiplatelet therapy with CMBs, but data in patients with atrial fibrillation are scarce. Purpose We examined the associations of antithrombotic medication with CMB presence, burden, and location. Methods In 2,655 patients from the multicentre Swiss-AF cohort, CMBs were assessed by standardised brain MRI. Antithrombotic therapy was categorised as no therapy (n=419), oral anticoagulation (OAC, n=1,382), antiplatelet therapy (APT, n=622), or combination therapy (OAC+APT, n=232). CMB presence, burden, and location (lobar vs non-lobar) were analysed using logistic and proportional-odds models with covariable adjustment. Results Mean age was 73 ± 8 years, 31% were female, 29% had coronary heart disease, 8% peripheral artery disease, and 64% a history of atrial fibrillation. CMBs were present in 635 patients (23.9%), with the highest prevalence in patients receiving APT (28.7%) and OAC+APT (29.7%), compared with OAC alone (22.1%) and no therapy (19.8%). Patients receiving APT and OAC+APT had a higher CMB burden (≥3 CMBs: 7.1% and 6.0% vs 3.7% with OAC and 2.9% with no therapy). In unadjusted models, both APT and OAC+APT were associated with CMB presence (OR 1.63, P=.001; OR 1.71, P=.004) and higher burden (OR 1.68, P<.001; OR 1.73, P=.003). After multivariable adjustment for vascular risk factors, these associations were no longer statistically significant for presence (APT: OR 1.10, 95% CI 0.78–1.55, P=.576; OAC+APT: OR 1.07, 95% CI 0.67–1.71, P=.762) or burden (APT: OR 1.15, 95% CI 0.82–1.62, P=.406; OAC+APT: OR 1.09, 95% CI 0.69–1.73, P=.708). OAC was not associated with CMBs in any model (multivariable-adjusted OR 0.95, 95% CI 0.68–1.33, P=.753). When analysed by location, APT was associated with lobar CMBs in multivariable models (OR 1.55, 95% CI 1.02–2.38, P=.043) but not with non-lobar CMBs (OR 1.00, 95% CI 0.67–1.50, P=.995). Similarly, OAC+APT showed a trend towards lobar CMBs (OR 1.61, 95% CI 0.92–2.80, P=.093) without non-lobar association (OR 0.82, 95% CI 0.46–1.43, P=.478). OAC was not associated with either lobar (OR 1.23, 95% CI 0.81–1.88, P=.341) or non-lobar CMBs (OR 0.79, 95% CI 0.53–1.19, P=.257). The formal interaction between therapy group and CMB location was not statistically significant (P for interaction = .339). Conclusion In this large, multicentre cohort, patients receiving antiplatelet or antiplatelet and anticoagulation therapy had a higher prevalence and burden of CMBs compared to no therapy. After adjustment for vascular risk factors, neither antiplatelet nor combination therapy remained independently associated with overall CMB prevalence or burden. However, there might be an association of antiplatelet therapy with lobar CMBs.