Introduction: Large vessel intracranial stenosis (LVIS) is the most common stroke subtype worldwide and is associated with high risk of stroke recurrence. Current transcranial Doppler (TCD) diagnostic criteria for LVIS mainly rely on velocity measurement with unsatisfactory accuracy.
Hypothesis: A new scoring system that integrated several features of the cerebral blood flow velocity from TCD is able to offer more reliable identification of significant (≥ 50%) LVIS.
Methods: Using the TCD-Digital Subtraction Angiography (DSA) database from a previous NIH-funded trial - Stroke Outcomes and Neuroimaging of Intracranial Atherosclerosis (SONIA), the hard copy TCDs from SONIA trial were reviewed blinded to the results of DSA. A summed score was calculated for each middle cerebral artery (MCA) based on the four parameters: mean velocity (MV) (score 0: MV<80 cm/s, 1: MV=80-99 cm/s, 2: MV=100-119 cm/s, 3: MV≥120 cm/s); stenotic/pre-stenoic ratio (score 0: ratio<2, 1: ratio≥2); stenotic/contralateral MCA ratio (score 0: ratio<1.50, 1: ratio=1.50-1.99, 2: ratio ≥ 2.00); spectrum pattern (score 0: normal spectrum, 1: any pattern of turbulence). DSA results (presence of ≥ 50% stenosis) from SONIA were used as the gold standard. To define the optimal score that predict significant stenosis on DSA, predictive values (positive predictive value [PPV] and negative predictive values [NPV], and overall accuracy) with 95% CI were calculated.
Results: 110 MCAs with both TCD and DSA were available in 72 patients (50.7 % of total patients with TCD). The mean score was 1.8 (SEM 0.21), the optimal cutoff score with balanced PPV and NPV for identifying ≥50% stenosis was >4 with the PPV 76% (53-92), NPV = 84% (75-91) and overall accuracy 83% (76-90).The PPV of the new scoring system (76%) was higher than velocity-only criteria in SONIA (i.e. previously validated cutpoints from SONIA of MV=80 cm/s [32%] or 100 cm/s [37%]), while NPV remained similar between the two methods (84% for new scoring vs 86% or 85%).
Conclusions: The new TCD scoring system suggested higher diagnostic accuracy compared to the velocity-only method in diagnosis of ≥50% MCA stenosis using digital subtraction angiography as the confirmative method. Further validation is required.