Association of enlarged perivascular spaces and anticoagulant-related intracranial hemorrhage

• Published in: Neurology Vol. 95; no. 16; p. e2192
• Main Authors: Best, Jonathan G, Barbato, Carmen, Ambler, Gareth, Du, Houwei, Banerjee, Gargi, Wilson, Duncan, Shakeshaft, Clare, Cohen, Hannah, Yousry, Tarek A, Al-Shahi Salman, Rustam, Lip, Gregory Y H, Houlden, Henry, Brown, Martin M, Muir, Keith W, Jäger, Hans Rolf, Werring, David J
• Format: Journal Article
• Language: English
• Published: United States 20.10.2020
• Subjects: Aged; Aged, 80 and over; Anticoagulants - adverse effects; Basal Ganglia - pathology; Female; Glymphatic System - pathology; Humans; Intracranial Hemorrhages - chemically induced; Male; Proportional Hazards Models; Prospective Studies; Risk Factors; White Matter - pathology
• ISSN: 1526-632X
• DOI: 10.1212/WNL.0000000000010788

To investigate whether enlarged perivascular spaces (PVS) within the basal ganglia or deep cerebral white matter are risk factors for intracranial hemorrhage in patients taking oral anticoagulants (OACs), independent of established clinical and radiologic risk factors, we conducted a post hoc analysis of Clinical Relevance of Microbleeds in Stroke (CROMIS-2) (atrial fibrillation [AF]), a prospective inception cohort study. Patients with atrial fibrillation and recent TIA or ischemic stroke underwent standardized MRI prior to starting OAC. We rated basal ganglia PVS (BGPVS) and centrum semiovale PVS (CSOPVS), cerebral microbleeds (CMBs), white matter hyperintensities, and lacunes. We dichotomized the PVS rating using a threshold of >10 PVS in the relevant region of either cerebral hemisphere. The primary outcome was symptomatic intracranial hemorrhage (sICH). We identified risk factors for sICH using Cox regression. A total of 1,386 participants with available clinical and imaging variables were followed up for a mean of 2.34 years; 14 sICH occurred (11 intracerebral). In univariable analysis, diabetes, CMB presence, lacune presence, and >10 BGPVS, but not CSOPVS, were associated with sICH. In a multivariable model incorporating all variables with significant associations in univariable analysis, >10 BGPVS (hazard ratio [HR] 8.96, 95% [CI] 2.41-33.4, = 0.001) and diabetes (HR 3.91, 95% CI 1.34-11.4) remained significant risk factors for sICH. Enlarged BGPVS might be a novel risk factor for OAC-related ICH. The strength of this association and potential use in predicting ICH in clinical practice should be investigated in larger cohorts.