A comparison of digital subtraction angiography and noninvasive testing in the diagnosis of cerebrovascular disease

Among the 1,892 patients who underwent cerebrovascular digital subtraction angiography at our hospital over the past 18 months, there was a subgroup of 34 patients (65 carotid arteries) for whom noninvasive cerebrovascular test results and standard cerebral arteriograms were also available. These patients were reviewed retrospectively and the ability of both methods to detect hemodynamically significant lesions, defined as a greater than 50 percent reduction in the diameter of the carotid artery, was determined using the arteriograms as the "gold standard." Noninvasive cerebrovascular tests had a sensitivity of 81 percent, a specificity of 95 percent, a positive predictive value of 92 percent, a negative prediction value of 88 percent, and an overall accuracy of 89 percent. Digital subtraction angiography had a sensitivity of 84 percent, a specificity of 92 percent, a positive predictive value of 88 percent, a negative predictive value of 89 percent, and an overall accuracy of 89 percent. If the four cases of hemodynamically significant stenosis of the carotid siphon not detected by digital subtraction angiography had been considered as false-negatives, its sensitivity would have been reduced to 72 percent. In patients with hemispheric cerebral ischemia, we found noninvasive cerebrovascular tests neither necessary nor cost-effective. Digital subtraction angiography, on the other hand, often provided definitive diagnostic information in such patients if the intracranial circulation was well defined and the extracranial lesion corresponded to the patients' symptoms. Noninvasive cerebrovascular testing was the safest and most cost-effective technique for screening patients with asymptomatic bruits, atypical, nonhemispheric cerebral symptoms, and those who have undergone carotid endarterectomy. If the noninvasive cerebrovascular test result was positive or equivocal, digital subtraction angiography was performed to localize the responsible lesion and exclude carotid occlusion.