Antithrombotic Drug Use, Cerebral Microbleeds, and Intracerebral Hemorrhage A Systematic Review of Published and Unpublished Studies

Background and Purpose-Cerebral microbleeds (MB) are potential risk factors for intracerebral hemorrhage (ICH), but it is unclear if they are a contraindication to using antithrombotic drugs. Insights could be gained by pooling data on MB frequency stratified by antithrombotic use in cohorts with ICH and ischemic stroke (IS)/transient ischemic attack (TIA). Methods-We performed a systematic review of published and unpublished data from cohorts with stroke or TIA to compare the presence of MB in: (1) antithrombotic users vs nonantithrombotic users with ICH; (2) antithrombotic users vs nonusers with IS/TIA; and (3) ICH vs ischemic events stratified by antithrombotic use. We also analyzed published and unpublished follow-up data to determine the risk of ICH in antithrombotic users with MB. Results-In a pooled analysis of 1460 ICH and 3817 IS/TIA, MB were more frequent in ICH vs IS/TIA in all treatment groups, but the excess increased from 2.8 (odds ratio; range, 2.3-3.5) in nonantithrombotic users to 5.7 (range, 3.4-9.7) in antiplatelet users and 8.0 (range, 3.5-17.8) in warfarin users (P difference = 0.01). There was also an excess of MB in warfarin users vs nonusers with ICH (OR, 2.7; 95% CI, 1.6-4.4; P < 0.001) but none in warfarin users with IS/TIA (OR, 1.3; 95% CI, 0.9-1.7; P = 0.33; P difference = 0.01). There was a smaller excess of MB in antiplatelet users vs nonusers with ICH (OR, 1.7; 95% CI, 1.3-2.3; P < 0.001), but findings were similar for antiplatelet users with IS/TIA (OR, 1.4; 95% CI, 1.2-1.7; P < 0.001; P difference = 0.25). In pooled follow-up data for 768 antithrombotic users, presence of MB at baseline was associated with a substantially increased risk of subsequent ICH (OR, 12.1; 95% CI, 3.4-42.5; P < 0.001). Conclusions-The excess of MB in warfarin users with ICH compared to other groups suggests that MB increase the risk of warfarin-associated ICH. Limited prospective data corroborate these findings, but larger prospective studies are urgently required. (Stroke. 2010; 41: 1222-1228.)