Voxel-Based quantitative MRI reveals spatial patterns of grey matter alteration in multiple sclerosis

Abstract

Despite robust postmortem evidence and potential clinical importance of gray matter (GM) pathology in multiple sclerosis (MS), assessing GM damage by conventional magnetic resonance imaging (MRI) remains challenging. This prospective cross-sectional study aimed at characterizing the topography of GM microstructural and volumetric alteration in MS using, in addition to brain atrophy measures, three quantitative MRI (qMRI) parameters-magnetization transfer (MT) saturation, longitudinal (R1), and effective transverse (R2*) relaxation rates, derived from data acquired during a single scanning session. Our study involved 35 MS patients (14 relapsing-remitting MS; 21 primary or secondary progressive MS) and 36 age-matched healthy controls (HC). The qMRI maps were computed and segmented in different tissue classes. Voxel-based quantification (VBQ) and voxel-based morphometry (VBM) statistical analyses were carried out using multiple linear regression models. In MS patients compared with HC, three configurations of GM microstructural/volumetric alterations were identified. (a) Co-localization of GM atrophy with significant reduction of MT, R1, and/or R2*, usually observed in primary cortices. (b) Microstructural modifications without significant GM loss: hippocampus and paralimbic cortices, showing reduced MT and/or R1 values without significant atrophy. (c) Atrophy without significant change in microstructure, identified in deep GM nuclei. In conclusion, this quantitative multiparametric voxel-based approach reveals three different spatially-segregated combinations of GM microstructural/volumetric alterations in MS that might be associated with different neuropathology.

Keywords: atrophy; demyelination; gray matter; multiple sclerosis; quantitative MRI; voxel-based analysis.

FIGURE 1

Example of MPM quantitative maps for a specific MS patient. From left to right: 3 MPM quantitative maps (MT, R1, R2*), standard FLAIR sequence image and FLAIR image overlaid with the estimated lesion mask. MT, magnetization transfer saturation; R1, longitudinal relaxation rate (1/T1); R2*, effective transverse relaxation rate (1/T2*); Lesion mask, posterior probability map of lesion tissue thresholded at 90%

FIGURE 2

VBM and VBQ results superimposed on the group mean MT map. Green: Average WM lesion probability map of MS patients, thresholded at 90% (a); Voxels showing a significant difference between MS and HC (thresholded at cluster level, p < .05 FWER‐corrected): decreased gray matter (GM) volume in red (b), MT reduction in blue (c), R1 reduction in yellow and R2* reduction (circled) in violet (d). The right most column (MERGE) overlays the maps displayed in columns b, c, and d columns, same color scheme, and highlights the three different patterns discussed in the main text (1 = Primary Neocortical Regions, 2 = Hippocampus, 3 = Deep Gray Matter Nuclei). Images are shown in neurologic convention and the X/Y/Z coordinates indicate the slice position in millimeter in MNI space

FIGURE 3

Illustration of the distribution of each GM parameter (GM volume, MT, R1, R2*) extracted from four different voxels, across MS and HC subjects in four brain regions of interest: Left thalamus (x = −15, y = −27, z = 14); Left hippocampus (x = −31, y = −27, z = −7); Left Heschl's gyrus (x = −37, y = −28, z = 10); Left precentral gyrus (x = −34, y = −6, z = 46). All voxels coordinates expressed in MNI space and chosen as the local statistical maximum in the ROI. Statistical significance (*) set at p < .05 FWER‐corrected for the whole GM volume