Clinical Correlations of Brain Lesion Location in Multiple Sclerosis: Voxel-Based Analysis of a Large Clinical Trial Dataset

There is a limited correlation between white matter (WM) lesionload as determined by magnetic resonance imaging and disability in multiple sclerosis (MS). The reasons for this so-called clinico-radiological paradox are diverse and may, at least partly, relate to the fact that not just the overall lesionburden, but also the exact anatomical location of lesionspredict the severity and type of disability. We aimed at studying the relationship between lesiondistribution and disability using a voxel-based lesionprobability mapping approach in a very large dataset of MS patients. T2-weighted lesionmasks of 2348 relapsing-remitting MS patients were spatially normalizedto standard stereotaxic space by non-linear registration. Relations between supratentorial WM lesionlocations and disability measures were assessed using a non-parametric ANCOVA ( Expanded Disability Status Scale[EDSS]; Multiple Sclerosis Functional Composite, and subscores; Modified Fatigue Impact Scale) or multinomial ordinal logistic regression (EDSS functional subscores). Data from 1907 (81%) patients were included in the analysis because of successful registration. The lesionmapping showed similar areas to be associated with the different disability scales: periventricular regions in temporal, frontal, and limbic lobeswere predictive, mainly affecting the posterior thalamic radiation, the anterior, posterior, and superior parts of the corona radiata. In summary, significant associations between lesionlocation and clinical scores were found in periventricular areas. Such lesionclusters appear to be associated with impairment of different physical and cognitive abilities, probably because they affect commissuraland long projection fibers, which are relevant WM pathways supporting many different brain functions.