Background and purpose: Intracranial arterial stenosis increases flow velocities on the upslope of the Spencer's curve of cerebral hemodynamics. However, the velocity can decrease with long and severely narrowed vessels. We assessed the frequency and accuracy for detection of focal and diffuse intracranial stenoses using novel diagnostic criteria that take into account increased resistance to flow with widespread lesions.
Methods: We evaluated consecutive patients referred to a neurovascular ultrasound laboratory with symptoms of cerebral ischemia. Transcranial Doppler mean flow velocities were classified as normal (30 to 99 cm/s), high and low. Pulsatility index >or=1.2 was considered high. Focal intracranial disease was defined as >or=50% diameter reduction by the Warfarin Aspirin in Symptomatic Intracranial Disease criteria. Diffuse disease was defined as stenoses in multiple intracranial arteries, multiple segments of one artery, or a long (>1 cm) stenosis in one major artery on contrast angiography (CT angiography or digital subtraction angiography) as the gold standard.
Results: One hundred fifty-three patients (96 men, 76% white, age 62+/-15 years) had previous strokes (n=135) or transient ischemic attack (n=18). Transcranial Doppler detection of focal and diffuse intracranial disease had sensitivity 79.4% (95% CI: 65.8% to 93%), specificity 92.4% (95% CI: 87.7% to 97.2%), positive predictive value 75.0% (95% CI: 60.9% to 89.2%), negative predictive value 94.0% (95% CI: 89.7% to 98.3%), and overall accuracy 89.5% (95% CI: 84.5% to 94.4%). After adjustment for stroke risk factors, transcranial Doppler findings of low mean flow velocities and high pulsatility index in a single vessel were independently associated with angiographically demonstrated diffuse single vessel intracranial disease, whereas low mean flow velocities/high pulsatility index in multiple vessels were related to multivessel intracranial disease (OR: 19.7, 95% CI: 4.8 to 81.2, P<0.001).
Conclusions: Diffuse intracranial disease may have a higher than expected frequency in a select stroke population and can be detected with noninvasive screening.