Doppler embolic signals in cerebrovascular disease and prediction of stroke risk: a systematic review and meta-analysis.

BACKGROUND AND PURPOSE Asymptomatic embolic signals (ES) detected using transcranial Doppler have been reported in patients with potential cerebral embolic sources. They may be useful in risk stratification and in assessing therapies. First, it is essential to show whether they predict stroke risk. METHODS A systematic review and meta-analysis was performed to determine the prognostic value of ES in different potential cerebral embolic sources. Studies were identified that used transcranial Doppler to detect ES and included prospective stroke/TIA follow-up. Numbers of ES-positive and ES-negative patients were extracted with stroke/TIA and stroke alone outcomes. RESULTS ES are most frequent in large artery disease, less frequent in cardioembolic stroke, and infrequent in lacunar stroke. Data relating ES to future stroke risk were available for acute stroke, large artery disease, and the perioperative period of carotid endarterectomy. For symptomatic carotid stenosis, ES predicted stroke alone (OR, 9.57; 95%CI, 1.54 to 59.38; P=0.02) and stroke/TIA (OR, 6.36; 95% CI, 2.90-13.96; P<0.00001). For asymptomatic carotid stenosis, ES predicted stroke alone (OR, 7.46; 95% CI, 2.24-24.89; P=0.001) and stroke/TIA (OR, 12.00; 95% CI, 2.43-59.34; P=0.002) but with heterogeneity (P=0.004). In acute stroke ES predicted stroke alone (OR, 2.44; 95% CI, 1.17-5.08; P=0.02) and stroke/TIA (OR, 3.71; 95% CI, 1.64-8.38; P=0.002). A high frequency of ES immediately after carotid endarterectomy predicted stroke alone (OR, 24.54; 95% CI, 7.88-76.43; P<0.00001) and stroke/TIA (OR, 32.04; 95% CI, 11.36-90.39; P<0.00001). CONCLUSIONS ES predict stroke risk in acute stroke, symptomatic carotid stenosis, and postoperatively after carotid endarterectomy; in asymptomatic carotid stenosis, data are less robust. In these conditions ES may be useful in risk stratification and in assessing therapeutic efficacy. For other embolic sources, further prospective data are required.