Detection of the siphon internal carotid artery stenosis: Transcranial doppler versus digital subtraction angiography

Background Transcranial Doppler (TCD) is widely applied to evaluate the intracranial large artery stenosis. We aimed to evaluate the accuracy of TCD in detecting siphon internal carotid artery (SICA) stenosis against digital subtraction angiography (DSA). Methods A total of 145 patients with TCD and DSA performed were recruited. The SICA and other intracranial arteries were evaluated through transorbital and transtemporal acoustic window by TCD. We defined the best cutoff value for significant SICA stenosis by receiver-operating characteristic (ROC) curve analyses and calculated the sensitivity, specificity, and positive and negative predictive values (PPV and NPV). Resluts Mean age of the patients was 65.7 years (range 28-88 years) and 75.1% (109/145) were men. The peak systolic velocity (PSV) = 120 cm/second had the largest area under the ROC curve (area under the ROC curve =.868) compared with the MFV = 70 cm/second and MFV = 80 cm/second (area under the ROC curve =.822 and.845). So the criteria for SICA stenoses were defined as PSV = 120 cm/second plus additional parameters (abnormal spectrum, circumscribed velocity changes, or side-to-side difference). Twenty-nine patients were diagnosed with SICA stenosis or occlusion on TCD. Fourteen patients had abnormal TCD findings that were not confirmed by DSA. One patient with normal TCD but DSA showed mild (40%) stenoses. Accuracy parameters for TCD were as follows: sensitivity = 96.7% (95% confidence interval [CI] = 80.9-99.8), specificity = 93.9% (95% CI = 89.9-96.4), PPV = 65.9% (95% CI = 50.0-79.1), NPV = 99.6% (95% CI = 97.2-1.0). Most false-positive patients (11 of 15 patients) had other significant intracranial large arteries lesions. Conclusions The criteria have limited but acceptable sensitivity and specificity in detecting SICA stenoses. Abnormal findings of siphon carotid on TCD require further clarification. © 2009 by the American Society of Neuroimaging.