Using quantitative magnetic resonance imaging to track cerebral alterations in multiple sclerosis brain: A longitudinal study

Abstract

Introduction: Quantitative MRI quantifies tissue microstructural properties and supports the characterization of cerebral tissue damages. With an MPM protocol, 4 parameter maps are constructed: MTsat, PD, R1 and R2*, reflecting tissue physical properties associated with iron and myelin contents. Thus, qMRI is a good candidate for in vivo monitoring of cerebral damage and repair mechanisms related to MS. Here, we used qMRI to investigate the longitudinal microstructural changes in MS brain.

Methods: Seventeen MS patients (age 25-65, 11 RRMS) were scanned on a 3T MRI, in two sessions separated with a median of 30 months, and the parameters evolution was evaluated within several tissue classes: NAWM, NACGM and NADGM, as well as focal WM lesions. An individual annual rate of change for each qMRI parameter was computed, and its correlation to clinical status was evaluated. For WM plaques, three areas were defined, and a GLMM tested the effect of area, time points, and their interaction on each median qMRI parameter value.

Results: Patients with a better clinical evolution, that is, clinically stable or improving state, showed positive annual rate of change in MTsat and R2* within NAWM and NACGM, suggesting repair mechanisms in terms of increased myelin content and/or axonal density as well as edema/inflammation resorption. When examining WM lesions, qMRI parameters within surrounding NAWM showed microstructural modifications, even before any focal lesion is visible on conventional FLAIR MRI.

Conclusion: The results illustrate the benefit of multiple qMRI data in monitoring subtle changes within normal appearing brain tissues and plaque dynamics in relation with tissue repair or disease progression.

Keywords: longitudinal analysis; multiple sclerosis; quantitative MRI; relaxometry.

FIGURE 1

Chart flow of data creation and processing (see text). MPM maps were created with the hMRI toolbox, and FLAIR images were directly acquired for both sessions (T0 and T1). A preliminary mask was constructed based on T0 FLAIR. All images (MPM and FLAIR, T0 and T1) were coregistered to the MPM T0 space. Segmentation using USwL allowed to isolate the different tissue classes.

FIGURE 2

Schematic illustration of the NAWM and 3 lesions‐related areas: focal FLAIR lesion (dark gray area), initial peripheral lesion detected at T0 (medium gray area), later peripheral lesion detected at T1 (dashed, left, and light gray, right, area).

FIGURE 3

Line plots illustrating individual ARoCs for MTsat (left) and R2* (right) in NAWM. Each line corresponds to one subject. Dotted lines represent increasing rates.

FIGURE 4

Violin plots of significant change rates in microstructure with respect to $X_{status}$. From left to right: MTsat in NAWM, MTsat in NACGM, R2* in NAWM. * $P<.05$.

FIGURE 5

Microstructural parameters in NAWM and the 3 lesion‐related areas, for each scanning time T0 and T1. p Values were obtained with post hoc tests on the tissue area effect. * $P<.05$.