To explore superficial white matter (SWM) integrity in early multiple sclerosis (MS) patients.

SWM is composed of small diameter, intracortical association fibers resting just beneath the cortex. These structures are among the last parts of the brain to myelinate and hence are likely more vulnerable to early damage. To our knowledge, this is the first investigation of SWM in MS to date.

33 relapsing-remitting patients (mean age=34.7 years, mean EDSS score=0.92, 21 female) without visible juxtacortical lesions on T2 images; 31 age and sex-matched healthy controls (HC) from a larger cohort (RADIEMS) were included. Fractional anisotropy (FA) and mean diffusivity (MD) values from 100 SWM bundles across the cortex and mean FA and MD of 18 major DWM were extracted. In addition, thickness of 68 cortical regions and resting-state functional-connectivity (RSFC) among them were calculated. Wilcoxon rank sum test was used to compare the distribution of structural and functional metrics between two groups. Using a machine learning method (logistic regression with ridge regularization), 5 models were built to classify MS patients and HC based on: 1-SWM, 2-DWM, 3-cortical thickness and 4-RSFC measures. In model 5, all features from previous models were concatenated. Area under the receiver operating characteristic curve (AUC) values were calculated to evaluate classification performance of each model.

MS patients had higher MD in SWM bundles in insula, lateral-orbitofrontal, superior-temporal, pars orbitalis, medial-orbitofrontal and rostral middle-frontal cortices (p<0.05-corrected). No other MRI/fMRI differences were found between two groups. SWM model provided higher AUC (0.67) compared to DWM (0.61), RSFC (0.61) and all-features (0.61) models (p<0.005 for all comparisons-corrected). Cortical thickness provided the best classification, AUC=0.78.

Our results reveal SWM damage at early stages of MS that are mainly in language-related cortical regions. Future studies are warranted to explore the association of SWM integrity and cognitive/behavioral performance.