Intracerebral haemorrhage risk in microbleed-positive ischaemic stroke patients with atrial fibrillation: Preliminary meta-analysis of cohorts and anticoagulation decision schema.

INTRODUCTION: Whether ischaemic stroke patients with atrial fibrillation (AF) and cerebral microbleeds (CMBs) on MRI can be safely anticoagulated is a hotly debated topic. We performed a systematic review and meta-analysis of published aggregate data, to investigate the risk of subsequent intracerebral haemorrhage (ICH) based on CMBs presence in this stroke population, generally considered for oral anticoagulation. We also suggest a decision-making schema for anticoagulation use in this setting.

METHODS: We searched PubMed for relevant observational studies. Random effects models with DerSimonian-Laird weights were used to investigated the association between CMBs presence at baseline MRI and ICH or ischaemic stroke during follow-up.

RESULTS: Four studies, with slightly heterogeneous design, including 990 ischaemic stroke patients were pooled in a meta-analysis (crude CMBs prevalence: 25%; 95%CI: 17%-33%). The median follow-up ranged between 17 and 37months. The future symptomatic ICH rate was 1.6% (16/990), while recurrent ischaemic stroke rate was 5.9% (58/990). Baseline CMB presence was associated with increased risk of symptomatic ICH during follow-up compared to patients without CMBs (OR: 4.16; 95%CI: 1.54-11.25; p=0.005). There was no association between CMBs presence and recurrent ischaemic stroke risk.

CONCLUSION: We have shown that the presence of CMBs in cohorts of ischaemic stroke patients, most with AF on warfarin, is associated with a 4-fold increase in subsequent ICH (but not ischaemic stroke) risk (Class III evidence). These pooled estimates are useful for future trials design. We propose a simple data-driven anticoagulation schema which awaits validation and refinement as new prospective data are accumulated.