Human brain atlas-based multimodal MRI analysis of volumetry, diffusimetry, relaxometry and lesion distribution in multiple sclerosis patients and healthy adult controls: implications for understanding the pathogenesis of multiple sclerosis and consolidation of quantitative MRI results in MS

Abstract

Multiple sclerosis (MS) is the most common immune-mediated disabling neurological disease of the central nervous system. The pathogenesis of MS is not fully understood. Histopathology implicates both demyelination and axonal degeneration as the major contributors to the accumulation of disability. The application of several in vivo quantitative magnetic resonance imaging (MRI) methods to both lesioned and normal-appearing brain tissue has not yet provided a solid conclusive support of the hypothesis that MS might be a diffuse disease. In this work, we adopted FreeSurfer to provide standardized macrostructure or volumetry of lesion free normal-appearing brain tissue in combination with multiple quantitative MRI metrics (T(2) relaxation time, diffusion tensor anisotropy and diffusivities) that characterize tissue microstructural integrity. By incorporating a large number of healthy controls, we have attempted to separate the natural age-related change from the disease-induced effects. Our work shows elevation in diffusivity and relaxation times and reduction in volume in a number of normal-appearing white matter and gray matter structures in relapsing-remitting multiple sclerosis patients. These changes were related in part with the spatial distribution of lesions. The whole brain lesion load and age-adjusted expanded disability status score showed strongest correlations in regions such as corpus callosum with qMRI metrics that are believed to be specific markers of axonal dysfunction, consistent with histologic data of others indicating axonal loss that is independent of focal lesions. Our results support that MS at least in part has a neurodegenerative component.

Figure 1

Illustration of the FreeSurfer generated deep and cortical brain regions. The cortical (e.g. frontal cortex subdivisions) or deep (e.g. corpus callosum subdivisions) were pooled or volume-averaged as needed to reduce the number of comparisons.

Figure 2

Group mean comparisons of regional normal-appearing qMRI values between controls and RRMS (subcortical, cortical and lobar white and gray matter) (a) absolute volume in mL, (b) volume-to-ICV percentage or VOLp (c) relative proton density, and (d) T₂ relaxation time

Figure 3

Group mean comparisons of regional normal-appearing qMRI values between controls and RRMS (subcortical, cortical and lobar white and gray matter) (a) fractional anisotropy, (b) mean diffusivity, (c) axial diffusivity, and (d) radial diffusivity.

Figure 4

Representative illustration of age-dependence of qMRI metrics in both RRMS and controls using scatter plots and linear regression (a) volume percentage of the frontal cortex (b) volume percentage of the cingulate cortex (c) mean diffusivity of the frontal cortex and (d) mean diffusivity of the cingulate cortex. Note the rapid decrease in cortical gray matter volume with age in both controls and RRMS patients.

Figure 5

Visual illustration of regional qMRI and fusion with lesion probability maps in RRMS. The upper multi-plane view shows the percentage ICV-normalized normal-appearing volume difference (significant atrophy RRMS < Controls) fused with the lesion map (lower multi-view). Note that largest normal-appearing tissue atrophy is in deep white matter where lesion probability map is largest. Note that lesions in our RRMS cohort were least frequent in the thalamus yet the volume difference is significant. The color map (minimum dark blow) in the upper views corresponds to the percentage (maximum ~ 28% (bright red) in periventricular white matter and corpus callosum isthmus; see Fig 2b for the corresponding group difference p values. The color map in the middle lower view corresponds to the percentage of patients with lesions.