Development and validation of visual grading system for stenosis in intracranial atherosclerotic disease on time-of-flight magnetic resonance angiography

Although the overestimation problem of time-of-flight magnetic resonance angiography (TOF-MRA) applying the warfarin-aspirin symptomatic intracranial disease (WASID) method to assess intracranial arterial stenosis has often been suggested, no pertinent grading system for TOF-MRA has been developed. We aimed to develop and evaluate the performance of a visual grading system for intracranial arterial stenosis on TOF-MRA (MRAVICAST). This single-center cohort study analyzed prospective observational registry data from a comprehensive stroke center between January 2014 and February 2020. Patients with confirmed stenosis of the intracranial large arteries who underwent confirmative digital subtraction angiography (DSA) were included; a 4-point grading system was developed based on physical characteristics of TOF-MRA. The overall diagnostic accuracies of MRAVICAST for each grade, interobserver reproducibility, and positive predictive values for > 50% and > 70% stenoses were evaluated. We analyzed 132 segments with intracranial atherosclerotic stenosis from 71 patients (34 men and 37 women; mean age, 61.0 ± 15.25 years; range, 21–89 years). The overall diagnostic accuracy of MRAVICAST (93.9%, 124/132) was higher than that of MRAWASID (50.8%, 67/132) for each grade. The degree of stenosis did not differ significantly between MRAVICAST and DSAWASID (p = .849). Regarding reproducibility, MRAVICAST demonstrated excellent interobserver agreement (ICC, 0.989; 95% CI, 0.979–0.999). The positive predictive values of MRAVICAST for the diagnosis of > 50% and > 70% stenoses were 97.3% and 100.0%, respectively. The new intuitive grading system accurately and reliably determined the degree of stenosis in intracranial arterial atherosclerosis patients. MRAVICAST could be a versatile alternative to MRAWASID for evaluating intracranial arterial stenosis. • In this retrospective diagnostic study (sample: 132 stenotic segments), the overall accuracy of the visual grading system (MRAVICAST) was 94%, and positive predictive value for > 50% stenosis was 97%. • In the era of medical treatment for intracranial atherosclerotic stenosis, MRAVICAST could be a versatile alternative method to MRAWASID for evaluating intracranial arterial stenosis.