The Distributional Characteristics of Multiple Sclerosis Lesions on Quantitative Susceptibility Mapping and Their Correlation With Clinical Severity

Background: Multiple sclerosis (MS) patients have a wide spectrum of severity and responses to therapy; the personalization of treatment relies on sensitive and specific biomarkers. Previous studies have suggested that susceptibility contrast in demyelinated plaques is associated with iron-related pathology in multiple sclerosis which may indicate clinical severity. The aims of this study were to characterize the spatial distribution of MS lesions with different iron patterns by using quantitative susceptibility mapping and to explore neuroradiological findings that correlate with poor clinical outcome.Methods: Twenty-six patients with relapsing–remitting MS [14 men, 12 women; mean age, 29 ± 8 (standard deviation) years; age range, 21–52 years] were included in this study. Differences in lesion number, T2 volume, and susceptibility were compared among lesions subcategorized by location and by the presence or absence of a hyperintense rim on quantitative susceptibility mapping. Associations between these imaging features and clinical outcomes including Expanded Disability Status Scale scores and annual relapse rates were investigated.Results: A total of 811 unifocal MS lesions were included, and their QSM patterns were nodular hyperintensity with no rim (rim–, 540, 67%) or with a hyperintense rim on the edge (rim+, 172, 21%) and nodular isointensity (99, 12%). Rim+ lesions had significantly larger volume (115 ± 142 vs. 166 ± 185 mm3, p < 0.001) and lower susceptibility (4 ± 15 vs. 8 ± 16 ppb, p < 0.05) than rim– lesions. More rim+ lesions were found in periventricular areas [median, 45%; interquartile range (IQR), 36%], whereas a larger proportion of rim– lesions were distributed in juxtacortical (median, 32%; IQR, 21%) and deep white matter (median, 38%; IQR, 22%) areas. The annual relapse rate was positively correlated with the proportion of periventricular rim+ lesions (p < 0.001, r = 0.65) and the proportion of subtentorial rim+ lesions (p < 0.05, r = 0.40). Additionally, a significant association was found between the burden of periventricular rim+ lesions (β = 0.64, p < 0.001) and the burden of subtentorial rim– lesions (β = 0.36, p < 0.05).Conclusions: A high number or lesion burden of periventricular rim+ lesions or subtentorial lesions is associated with frequent clinical relapses