Abstract
Diffusion magnetic resonance imaging can reveal quantitative information about the tissue changes in multiple sclerosis. The recently developed multi-compartment spherical mean technique can map different microscopic properties based only on local diffusion signals, and it may provide specific information on the underlying microstructural modifications that arise in multiple sclerosis. Given that the lesions in multiple sclerosis may reflect different degrees of damage, we hypothesized that quantitative diffusion maps may help characterize the severity of lesions “in vivo” and correlate these to an individual’s clinical profile. We evaluated this in a cohort of 59 multiple sclerosis patients (62% female, mean age 44.7 years), for whom demographic and disease information was obtained, and who underwent a comprehensive physical and cognitive evaluation. The magnetic resonance imaging protocol included conventional sequences to define focal lesions, and multi-shell diffusion imaging was used with b-values of 1000, 2000 and 3000 s/mm2 in 180 encoding directions. Quantitative diffusion properties on a macro- and micro-scale were used to discriminate distinct types of lesions through a k-means clustering algorithm, and the number and volume of those lesion types were correlated with parameters of the disease. The combination of diffusion tensor imaging metrics (fractional anisotropy and radial diffusivity) and multi-compartment spherical mean technique values (microscopic fractional anisotropy and intra-neurite volume fraction) differentiated two type of lesions, with a prediction strength of 0.931. The B-type lesions had larger diffusion changes compared to the A-type lesions, irrespective of their location (P < 0.001). The number of A and B type lesions was similar, although in juxtacortical areas B-type lesions predominated (60%, P < 0.001). Also, the percentage of B-type lesion volume was higher (64%, P < 0.001), indicating that these lesions were larger. The number and volume of B-type lesions was related to the severity of disease evolution, clinical disability and cognitive decline (P = 0.004, Bonferroni correction). Specifically, more and larger B-type lesions were correlated with a worse Multiple Sclerosis Severity Score, cerebellar function and cognitive performance. Thus, by combining several microscopic and macroscopic diffusion properties, the severity of damage within focal lesions can be characterized, further contributing to our understanding of the mechanisms that drive disease evolution. Accordingly, the classification of lesion types has the potential to permit more specific and better-targeted treatment of patients with multiple sclerosis.
Highlights
Multiple sclerosis (MS) is a chronic inflammatory autoimmune disease of the central nervous system (CNS) that is characterised by the presence of focal lesions, and damage to the normal-appearing white matter (NAWM) and the grey matter (Lassmann et al, 2007)
The changes in lesions and in the NAWM can be visualised through conventional magnetic resonance imaging (MRI), yet they are poorly associated with the clinical phenotype and physical disability (Barkhof, 2002), partly reflecting the failure to characterise the pathological nature of tissue injury in MS
We want to highlight here that clustering techniques may create artificial groups of data that may not be replicated in new data. We only considered those sets of diffusion MRI measurements that led to clusters that were independently and consistently replicated in new data for periventricular, juxtacortical, brainstem, cerebellar and deep WM MS lesions, as defined by a prediction strength > 0.8 (Tibshirani and Walther, 2005)
Summary
Multiple sclerosis (MS) is a chronic inflammatory autoimmune disease of the central nervous system (CNS) that is characterised by the presence of focal lesions, and damage to the normal-appearing white matter (NAWM) and the grey matter (Lassmann et al, 2007). There is substantial heterogeneity in the pathological changes among MS lesions, with different patterns of demyelination (Lucchinetti et al, 2000) and a variable degree of neuroaxonal damage having been described (Ludwin, 2006). Chronic active lesions are associated with a more aggressive disease evolution (Absinta et al, 2019; Lucchinetti et al, 2000) and differences in the severity of demyelination, remyelination and neuroaxonal damage could explain why some patients recover completely from relapses yet in others, their disability deteriorates more rapidly. The changes in lesions and in the NAWM can be visualised through conventional magnetic resonance imaging (MRI), yet they are poorly associated with the clinical phenotype and physical disability (Barkhof, 2002), partly reflecting the failure to characterise the pathological nature of tissue injury in MS. DTI findings are strongly influenced by a complex intravoxel fibre architecture, which limits the ability to accurately estimate the different pathophysiological features of the disease (Rovaris et al, 2005; Filippi and Rocca, 2011)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
