Longitudinal microstructural MRI markers of demyelination and neurodegeneration in early relapsing-remitting multiple sclerosis: magnetisation transfer, water diffusion and g-ratio

Abstract

AbstractIntroductionQuantitative microstructural MRI, such as myelin-sensitive magnetisation transfer ratio (MTR) or saturation (MTsat), axon-sensitive water diffusion Neurite Orientation Dispersion and Density Imaging (NODDI), and the derived aggregate g-ratio, may provide more specific markers of white matter integrity than conventional MRI for early patient stratification in relapsing-remitting multiple sclerosis (RRMS). The aim of this study is to determine the sensitivity of such markers to longitudinal pathological change within cerebral white matter lesions (WML) and normal-appearing white matter (NAWM) in recently diagnosed RRMS.MethodsSeventy-nine people with recently diagnosed RRMS, from the FutureMS longitudinal cohort, were recruited to an extended MRI protocol at baseline and one year later. Twelve healthy volunteers received the same MRI protocol, repeated within two weeks. Ethics approval and written informed consent were obtained.3T MRI included magnetisation transfer, and multi-shell diffusion-weighted imaging. NAWM and whole brain volume (WBV) were segmented from 3D T1-weighted MPRAGE, and WML from T2-weighted FLAIR. MTR, MTsat, NODDI isotropic (ISOVF) and intracellular (ICVF) volume fractions, and aggregate g-ratio were measured within WML and NAWM. G-ratio was calculated from MTsat and NODDI data.Longitudinal change in brain volume and microstructural metrics was assessed with paired t-tests (α=0.05) and linear mixed models, corrected for confounding factors and multiple comparisons. Longitudinal changes were compared with test-retest Bland-Altman limits of agreement from healthy volunteer control white matter. The influence of varying MTsat and NODDI parameters on derived g-ratio was explored in silico.ResultsIn NAWM, paired t-tests show a significant increase in g-ratio (p=0.012) and ICVF (p=0.025), and a decrease in MTsat (p=0.033) over one year, but no change in MTR, or ISOVF. Linear mixed models show effects in NAWM remain significant after adjusting for covariates and after False Discovery Rate (FDR) correction for multiple comparisons, and no significant change in WBV. In WML, MTsat and NODDI metrics increase significantly over one year (linear mixed models, FDR-corrected p<0.05). In silico simulations show that increase in g-ratio may result from a decrease in MTsat or ISOVF, or an increase in ICVF.DiscussionG-ratio and MTsat changes in NAWM over one year indicate subtle myelin loss in early RRMS, which were not apparent with NAWM MTR or in WBV. Increases in NAWM and WML NODDI ICVF were not anticipated, and raise the possibility of axonal swelling or morphological change. Increases in WML MTsat may reflect myelin repair. Changes in NODDI ISOVF are more likely to reflect alterations in water content. Competing MTsat and ICVF changes may account for the absence of g-ratio change in WML. Longitudinal changes in microstructural measures are significant at a group level, however detection in individual patients in early RRMS is limited by technique reproducibility.ConclusionMTsat and g-ratio are more sensitive than MTR to early pathological changes in RRMS, but complex dependence of g-ratio on NODDI parameters limit the interpretation of aggregate measures in isolation. Improvements in technique reproducibility and validation of MRI biophysical models across a range of pathological tissue states are needed.