Background and purpose: Previous T2*-weighted magnetic resonance imaging (MRI) studies have used white matter lesion (WML) central veins to distinguish multiple sclerosis (MS) from its mimics. To be clinically applicable, the "central vein sign" needs to be detectable across different T2* sequences. Our objective was to determine if the central vein sign is reliably present in MS and absent in patients with ischemic small vessel disease (SVD) across different T2* sequences at 3T MRI.
Methods: Ten patients with MS and 10 with SVD were each scanned on a 3 T Philips and GE scanner. The MRI protocol included 3-dimensional (3D) T2* GRE, T2* with high echo planar imaging (EPI) factor and susceptibility-weighted angiography (SWAN). Total WML numbers, central vein numbers, and proportion of WMLs with central veins were calculated using each sequence. Three blinded raters identified a subset of six WMLs with central veins to diagnose MS or SVD.
Results: Irrespective of the sequence, MS patients were identified based on a higher proportion of WMLs with central veins. This proportion was dependent on the T2* sequence used. T2* with high EPI allowed the highest median proportion (69.6%) in MS patients; 6.1% in SVD patients (P < .0004). Rater reproducibility varied depending on the T2* sequence used. T2* with high EPI produced good agreement with the clinical diagnosis (Cohen's kappa range; .78-.89), as did SWAN imaging with some raters; ĸ = .69.
Conclusions: The central vein sign can diagnose MS in the clinical setting of modern 3T scanners. However, variations in the T2* sequences need to be considered when defining a threshold for diagnosis.
Keywords: MRI; MS; Multiple sclerosis; central vein; diagnosis.
Copyright © 2016 by the American Society of Neuroimaging.