The "central vein sign" to differentiate multiple sclerosis from migraine

Objective: To investigate, in two cohorts including patients with multiple sclerosis (MS) and migraine, (i) the prevalence of the "central vein sign" (CVS), (ii) the spatial distribution of positive CVS (CVS+) lesions, (iii) the threshold of CVS+ lesions able to distinguish MS from migraine with high sensitivity and specificity. Methods: A total of 70 patients with MS/clinically isolated syndrome and 50 age- and sex-matched patients with migraine underwent a 3-T magnetic resonance imaging scan. The CVS was evaluated according to current guidelines, excluding eight patients with migraine who did not show white matter (WM) lesions. A receiver operating characteristic curve analysis was performed to identify the best threshold in terms of proportion of CVS+ lesions and the absolute number of CVS+ lesions able to differentiate MS from migraine. Results: Lesion volume was different between CVS+ and CVS negative lesions (median 1043 vs. 176.5 mm(3) for MS cohort; median 35.1 vs. 52.2 mm(3) for migraine cohort; p < 0.001 for all). A higher proportion of CVS+ lesions was associated with a higher probability of being diagnosed as MS (adjusted odds ratio 1.09, 95% confidence interval [CI] 1.04-1.14; p < 0.001). CVS+ lesion volume and number were higher in MS with respect to migraine, both considering the whole brain and its subregions (p < 0.001). The proportion of CVS+ lesions in juxtacortical and infratentorial areas was higher in MS than in migraine (p = 0.015 and p = 0.034, respectively). The best CVS proportion-based threshold able to differentiate MS from migraine was 35.0% (sensitivity 97.1%, specificity 85.7%) with an area under the curve of 0.95 (95% CI 0.90-1.00). The "select 6" rule seemed to be preferable in terms of specificity with respect to the "select 3" rule. Conclusions: A CVS proportion-based threshold of 35.0% is capable of distinguishing MS from migraine with high sensitivity and specificity. The "select 6" algorithm may be useful in the clinical setting.