Effect of disease-modifying therapies on subcortical gray matter atrophy in multiple sclerosis

Abstract

Background:

The effects of disease-modifying therapies (DMTs) on region-specific brain atrophy in multiple sclerosis (MS) are unclear.

Objective:

To determine the effects of higher versus lower efficacy DMTs on rates of brain substructure atrophy in MS.

Methods:

A non-randomized, observational cohort of people with MS followed with annual brain MRI was evaluated retrospectively. Whole brain, subcortical gray matter (GM), cortical GM, and cerebral white matter (WM) volume fractions were obtained. DMTs were categorized as higher (DMT-H: natalizumab, rituximab) or lower (DMT-L: interferon-beta and glatiramer acetate) efficacy. Follow-up epochs were analyzed if participants had been on a DMT for ≥6 months and had at least one follow-up MRI while on DMTs in the same category.

Results:

A total of 86 DMT epochs (DMT-H: n=32; DMT-L: n=54) from 78 participants fulfilled the study inclusion criteria. Mean follow-up was 2.4 years. Annualized rates of thalamic (−0.15% vs. −0.81%; p=0.001) and putaminal (−0.27% vs. −0.73%; p=0.001) atrophy were slower during DMT-H compared to DMT-L epochs. These results remained significant in multivariate analyses including demographics, clinical characteristics and T2 lesion volume.

Conclusion:

DMT-H treatment may be associated with slower rates of subcortical GM atrophy, especially of the thalamus and putamen. Thalamic and putaminal volumes are promising imaging biomarkers in MS.

INTRODUCTION

The armamentarium of multiple sclerosis (MS) disease-modifying therapies (DMTs) has dramatically expanded and now includes treatments with high efficacy at reducing overt inflammatory disease activity. However, the effects of more aggressive therapy on the neurodegenerative aspects of MS and the accrual of long-term disability are unclear. Notably, the conventional magnetic resonance imaging (MRI) measures utilized extensively in clinical practice and clinical trials to assess for radiological inflammatory disease activity are modestly associated with clinical measures of disability in MS, as they do not measure neuro-axonal loss, which is considered to be the substrate of permanent disability in MS.^1-3

Whole brain volume (WBV) atrophy is thought to reflect global neuro-axonal loss in MS, is accelerated in MS relative to normal aging, and is better associated with disability in MS than conventional MRI measures.^1-3 The majority of DMTs approved for use in MS, however, have demonstrated only moderate evidence of reductions in the rate of WBV atrophy.^1,4 Notably, effects on WBV atrophy have been observed mainly after the first year of treatment, since accelerated reductions in WBV may be observed initially, a phenomenon attributed to resolution of tissue edema that has been termed “pseudoatrophy”.¹

Studies examining brain substructure volumes have shown that progressive gray matter (GM) atrophy is detectable early in the disease course and exhibits stronger correlations with disability than WM measures.^5-7 Importantly, measures of the GM may be less susceptible to pseudoatrophy effects as compared to the WM.^8,9 Furthermore, subcortical GM structures, particularly the thalamus, have been shown to be especially affected at the pre-symptomatic or earliest stages of the disease.^10,11 Subcortical GM atrophy has also been shown to correlate with cognitive impairment in MS better than whole brain GM or cortical GM atrophy.¹² However, there is a paucity of studies assessing the effects of DMTs on subcortical GM atrophy in MS.

In a single-center, longitudinal, non-randomized, observational cohort followed with annual MRI, we sought to determine the differential effects of DMTs on the rates of brain substructure atrophy in MS. We hypothesized that treatment with higher efficacy DMTs would be associated with a reduced rate of brain substructure atrophy, especially of the GM structures.

METHODS

Study Participants

Johns Hopkins University Institutional Review Board approval was obtained for the study protocol, and written, informed consent was obtained from all participants.

A prospective, observational cohort of 167 people with MS followed at Johns Hopkins Hospital with annual brain MRI was retrospectively reviewed for inclusion in the study. Participants were recruited from the Johns Hopkins MS Center by convenience sampling and were studied between August 2007 and June 2017. MS diagnosis was confirmed by the treating neurologist based on the 2005 revised McDonald criteria, and disease subtype was classified as relapsingremitting (RRMS), secondary-progressive (SPMS) or primary-progressive MS (PPMS).^13,14 Analyses were restricted to RRMS participants, given that the studied DMTs are only approved for use in this disease subtype (with the exception of off-label use of rituximab). Expanded Disability Status Scale (EDSS) was assessed by Neurostatus certified raters. Occurrence of clinical relapses prior to and during follow-up was ascertained by the treating clinician as recorded in the medical record.

DMT use was recorded, including dates of any changes in DMT. DMTs were categorized as “lower” (DMT-L: interferon-beta and glatiramer acetate) or “higher” (DMT-H: natalizumab, rituximab) efficacy, based on the effects of these DMTs on clinico-radiological disease activity in clinical trials.¹⁵ Given that oral therapies including dimethyl fumarate and fingolimod are generally considered intermediate efficacy, these were excluded from our analysis given insufficient sample size and follow-up for these to be analyzed as a separate category. Other DMTs including teriflunomide, daclizumab and alemtuzumab, were not included in this categorization as we did not identify any participants fulfilling inclusion criteria in our study on any of these DMTs. All decisions regarding DMT selection were made by the treating clinician. Follow-up epochs were included in the analysis if the following criteria were fulfilled:

• 1)A baseline MRI identified greater than 6 months after DMT change or initiation (to minimize possible confounding effects of pseudoatrophy and ensure that adequate time had elapsed for the DMT to take effect)
• 2)At least one follow-up MRI visit while continuously remaining on DMTs in the same category.

For participants with more than one follow-up epoch (referred to hereafter as “DMT epoch”) fulfilling the study inclusion criteria, all DMT epochs (including all MRI scans during that time period) were included in the analyses. Follow-up data were censored at the time of DMT category change or therapy discontinuation.

Magnetic resonance imaging

Brain MRI scans were performed with a 3T Philips Achieva scanner (Philips Medical System, Best, Netherlands). Three axial whole-brain sequences without gaps were utilized: multi-slice fluid-attenuated inversion recovery (FLAIR; acquired resolution:0.8×0.8×2.2mm; echo time [TE]:68ms; repetition time [TR]:11s; inversion time [TI]:2.8s; SENSE factor:2; averages:1); T2-weighted dual-echo turbo spin echo (DE-TSE; acquired resolution:0.8×0.8×2.2mm; TE:80ms; TR:4170ms; SENSE factor:2; averages:1); and three-dimensional [3D] magnetization prepared rapid acquisition of gradient echoes (MPRAGE; acquired resolution:0.8×0.8×1.2 mm; TE:6ms; TR:10ms; TI:835ms; flip angle:8 degrees; SENSE factor:2; averages:1). All brain MRIs were reviewed by a neuroradiologist and the presence of new T2 lesions and/or gadolinium-enhancing lesions were recorded.

Segmentation of brain substructures was performed automatically using a multi-step processing pipeline. Images were first preprocessed to remove non-uniformity artifacts.¹⁶ Next, the intracranial vault was extracted using the Multi-cONtrast brain STRipping method (MONSTR), based on co-registered MPRAGE and T2-weighted images.¹⁷ For WM lesion segmentation, Subject-Specific Sparse Dictionary Learning (S3DL) was applied and the resulting lesion maps were used to perform in-painting of the MRI volumes, using the “lesion_filling” package included in the FMRIB Software Library (FSL).^18,19 Finally, Multi Atlas Cortical Reconstruction Using Implicit Surface Evolution (MA-CRUISE) was applied, which combines multi-atlas label fusion for whole brain segmentation with a deformable model for cortical surface reconstruction.^20-22 Together, these techniques segment the brain into its component substructures while simultaneously outlining MS lesions.

Utilizing this method, volumes of the brain substructures, including the cortical GM, cerebral WM, and subcortical GM structures (thalamus, putamen, globus pallidus and caudate nucleus) were derived, as well as the total T2 lesion volume. Segmentations were reviewed (ESS, NGC, BD), and scans with segmentation errors/failures were excluded. All volumes were normalized to intracranial volume (ICV) to obtain volume fractions, thus accounting for variation in head size.²³ For bilateral structures (including the thalamus, putamen, caudate nucleus and globus pallidus), the sums of the right and left volume fractions were used for analysis. Cerebral volume fraction (CVF), which is representative of whole brain volume and analogous to the brain parenchymal fraction (BPF), was calculated by dividing the summed volume of brain substructures by the ICV.

Statistical methods

Statistical analyses were performed with Stata 14 (StataCorp, College Station, TX). Analyses were not adjusted for multiple comparisons, given the exploratory nature of the study and results should be interpreted as such.²⁴ Statistical significance was defined as p <0.05.

T2 lesion volume fractions were not normally distributed and were logarithmically transformed, yielding an approximately normal distribution. Baseline demographics, clinical characteristics, and brain substructure volume fractions were compared between DMT groups with a t-test for normally distributed quantitative variables, Wilcoxon rank-sum test for non-normally distributed quantitative variables, and chi-squared test for categorical variables. Baseline brain substructure volumes were also compared with multivariate linear regression including age, sex, race, disease duration and EDSS.

Annualized rates of brain substructure volume fraction change were calculated and compared between DMT-H and DMT-L epochs with mixed-effects linear regression models. Analyses were also performed using raw volumes (i.e. not normalized to ICV), but results did not differ from the former approach. Details regarding the mixed effects models are included in the Supplementary Methods. Briefly, this method inherently adjusts for baseline volume at both the DMT-specific level (given inclusion of the fixed effect of DMT category) and at the subject and DMT epoch-specific level (given inclusion of random intercepts). Analyses were additionally adjusted for covariates including baseline age, race, sex, disease duration and EDSS, as well as baseline T2 lesion volume fractions and rates of T2 lesion volume fraction change during the study.

RESULTS

Study Population and Baseline Demographics, Clinical Characteristics and Brain Substructure Volume Fractions

A total of 167 participants were retrospectively assessed for eligibility for the study (Figure 1). Eight-six DMT epochs (DMT-L: n=54; DMT-H; n=32) from 78 participants fulfilled inclusion criteria. Demographics and clinical characteristics are presented in Table 1. Median follow-up was 2.5 and 2.1 years for the DMT-L and DMT-H groups, respectively. Baseline EDSS was higher in the DMT-H group (median 3.0 vs. 2.0; p<0.001). Disease duration at baseline was longer in DMT-H (median 9 vs. 4 years; p=0.01). Clinico-radiological disease activity in the year preceding the studied DMT epoch did not differ between groups (relapses and new T2 and/or post-gadolinium enhancing lesions; p=0.74 and p=0.51 respectively). Time on current DMT at baseline also did not differ between groups (median time on current DMT: DMT-L 1.2 years; DMT-H 1.4 years; p=0.14). As expected, there was less clinico-radiological disease activity during follow-up in the DMT-H group, as evidenced by lower proportions of patients experiencing relapses (3% vs. 28%; p=0.005) or developing new T2 lesion and/or postgadolinium enhancing lesions (22% vs. 56%; p=0.002).

Figure 1. Study Profile.

MS: multiple sclerosis; DMT: disease modifying therapy; DMT-H: higher efficacy DMT (natalizumab, rituximab); DMT-L: lower efficacy DMT (interferon-beta, glatiramer acetate)

Table 1.

Demographics and clinical characteristics by DMT category.

	DMT-L (n=54 epochs)	DMT-H (n=32 epochs)	P-value
Age, years, mean (SD)	38.4 (9.6)	42.5 (11.7)	0.08*
Female sex, n (%)	41 (76%)	22 (69%)	0.47†
Race, n (%)
Caucasian-American	45 (83%)	26 (81%)
African-American	8 (15%)	4 (13%)	0.55†
Asian-American	1 (2%)	2 (6%)
Disease Duration, years, median (IQR)	5 (3–9)	9 (5–13)	0.01‡
Baseline EDSS, median (IQR)	2.0 (1.5–2.5)	3.0 (2.5–3.5)	<0.001‡
Relapse during 1 year before baseline	20 (37%)	13 (41%)	0.74†
New T2 lesion(s) and/or Gd-enhancement during 1 year before baseline, n(%) ^	25 (50%)	17 (57%)	0.51†
DMT at baseline, n (%)
IFN-beta	30 (56%)	-
Glatiramer Acetate	24 (44%)	-
Rituximab	-	2 (6%)
Natalizumab	-	30 (94%)
Time on current DMT at baseline, years, median (IQR)	1.4 (0.9–3.3)	1.2 (0.9–1.8)	0.14‡
Patient-years on DMT during follow-up
IFN-beta	126.6	-
Glatiramer Acetate	78.7	-
Rituximab	-	4.3
Natalizumab	-	95.1
Follow-up, years, median (IQR)	2.5 (1.4–6.5)	2.1 (1.1–4.5)	0.34‡
New T2 lesion(s) and/or Gd-enhancement during follow-up, n (%)	30 (56%)	7 (22%)	0.002†
Relapse during follow-up, n (%)	15 (28%)	1 (3%)	0.005†

IQR: interquartile range; DMT: disease modifying therapy; DMT-H: higher efficacy DMT (natalizumab, rituximab); DMT-L: lower efficacy DMT (interferon-beta, glatiramer acetate); EDSS: expanded disability status scale

^{^}Available for 50 DMT-L and 30 DMT-H epochs

^*Student's t-test

^†Chi-squared test

^‡Wilcoxon rank-sum test

Comparisons of baseline brain substructure and T2 lesion volume fractions between the DMT-H and DMT-L groups revealed higher T2 lesion and lower whole brain (CVF), WM, and subcortical GM volume fractions, including the thalamus and globus pallidus, in the DMT-H group (Table 2). In multivariate analyses including age, sex, race, disease duration and EDSS, only the thalamic and pallidal volume fractions remained significantly lower in the DMT-H group.

Table 2.

Baseline brain substructure and T2 lesion volume fractions (normalized to intracranial volume) by DMT category.

	DMT-L (n=54 epochs)	DMT-H (n=32 epochs)	P-value (Unadjusted†)	P-value (Adjusted*)
Brain Substructure Volume Fractions, mean (SD)
Cerebral Volume Fraction	0.8078 (0.0391)	0.7880 (0.0399)	0.03	0.15
Cortical Gray Matter	0.3839 (0.0179)	0.3780 (0.0164)	0.13	0.68
Cerebral White Matter	0.2854 (0.0219)	0.2754 (0.0229)	0.048	0.10
Subcortical Gray Matter	0.0233 (0.0029)	0.0217 (0.0028)	0.01	0.07
Thalamus	0.0103 (0.0016)	0.0094 (0.0016)	0.009	0.03
Putamen	0.0062 (0.0008)	0.0060 (0.0008)	0.38	0.79
Caudate Nucleus	0.0046 (0.0007)	0.0043 (0.0006)	0.06	0.21
Globus pallidus	0.0022 (0.0002)	0.0020 (0.0003)	<0.001	0.008
T2 Lesions	0.0041 (0.0042)	0.0061 (0.0044)	0.01	0.15

DMT: disease modifying therapy; DMT-H: higher efficacy DMT (natalizumab, rituximab); DMT-L: lower efficacy DMT (interferon-beta, glatiramer acetate)

^†Student's t-test

^*Multivariate linear regression including age, sex, race, disease duration and EDSS

Subcortical GM atrophy rates were slower in participants treated with higher-efficacy therapies

The annualized unadjusted rates of brain substructure atrophy and T2 lesion volume fraction change by DMT group, as well as the comparisons between DMT groups (adjusted and unadjusted), are shown in Table 3 and Figure 2.

Table 3.

Comparison of annualized percent change in brain substructure and T2 lesion volume fractions (normalized to intracranial volume) between DMT groups.

	Annualized % Change, Unadjusted β 195% CI)		Difference in Annualized % Change, β (95% CI), Unadjusted	P-value (Unadiusted†)	Difference in Annualized % Change, β (95% CI), Adjusted	P-value (Adjusted*)
	DMT-L (n=54 epochs)	DMT-H (n=32 epochs)
Cerebral Volume Fraction	−0.48% (−0.62% to −0.35%)	−0.34% (−0.56% to −0.13%)	0.14% (−0.11% to 0.39%)	0.28	0.19% (−0.10% to 0.47%)	0.20
Cortical Gray Matter	−0.56% (−0.70% to −0.41%)	−0.35% (−0.58% to −0.12%)	0.21% (−0.07% to 0.48%)	0.14	0.23% (−0.07% to 0.54%)	0.13
Cerebral White Matter	−0.36% (−0.47% to −0.26%)	−0.33% (−0.49% to −0.16%)	0.04% (−0.15% to 0.23%)	0.69	0.11% (−0.08% to 0.30%)	0.24
Subcortical Grav Matter	−0.80% (−1.0% to −0.60%)	−0.25% (−0.52% to 0.03%)	0.55% (0.21% to 0.89%)	0.001	0.56% (0.20% to 0.92%)	0.002
Thalamus	−0.81% (−1.03% to −0.58%)	−0.15% (−0.48% to 0.17%)	0.66% (0.26% to 1.06%)	0.001	0.63% (0.21% to 1.06%)	0.003
Putamen	−0.73% (−0.88% to −0.58%)	−0.27% (−0.50% to −0.04%)	0.46% (0.19% to 0.73%)	0.001	0.41% (0.12% to 0.69%)	0.005
Caudate Nucleus	−0.75% (−1.08% to −0.41%)	−0.27% (−0.76% to 0.23%)	0.48% (−0.12% to 1.08%)	0.12	0.57% (−0.09% to 1.22%)	0.09
Globus Pallidus	−0.95% (−1.31% to −0.60%)	−0.53% (−1.05% to −0.01%)	0.42% (−0.21% to 1.05%)	0.19	0.50% (−0.16% to 1.16%)	0.14
T2 Lesion Volume	6.73% (3.55% to 9.91%)	0.71% (−3.86% to 5.29%)	−6.02% (−11.59% to −0.44%)	0.03	−5.51% (−11.81 % to 0.79%)	0.09

DMT: disease modifying therapy; DMT-H: higher efficacy DMT (natalizumab, rituximab); DMT-L: lower efficacy DMT (interferon-beta, glatiramer acetate); CI: confidence intervals

^†Mixed effects linear regression model including follow-up time, DMT category and their interaction.

^*Mixed effects linear regression model including (in addition to variables included in unadjusted model) age, sex, race, disease duration, EDSS and their respective interactions with follow-up time.

Statistically significant results are in bold

Figure 2. Mean annualized percent change and 95% confidence intervals of brain substructure volume fractions by DMT category.

CVF: cerebral volume fraction; GM: gray matter; WM: white matter; DMT: disease modifying therapy; DMT-H: higher efficacy DMT (natalizumab, rituximab); DMT-L: lower efficacy DMT (interferon-beta, glatiramer acetate)

Unadjusted analyses revealed that slower annualized rates of subcortical GM atrophy occurred during DMT-H as compared to DMT-L epochs (−0.25% vs −0.80%; 69% reduction; p=0.001). Among the subcortical GM structures, reduced annualized rates of atrophy in the DMT-H vs. DMT-L epochs were observed in the thalamus (−0.15% vs −0.81%; 81% reduction; p=0.01) and putamen (−0.27% vs −0.73%; 63% reduction; p=0.001), but not in the caudate nucleus or globus pallidus. In multivariate analyses including age, sex, race, disease duration, EDSS, rates of subcortical GM, thalamic and putaminal atrophy remained significantly lower during the DMTH epochs. Rates of whole-brain (CVF), cortical GM and cerebral WM atrophy did not differ significantly between DMT-H and DMT-L epochs in unadjusted or adjusted analyses.

Annualized rate of T2 lesion volume fraction change was higher during DMT-H as compared to DMT-L epochs, in unadjusted analyses (6.73% vs 0.71%; p=0.03), but this finding was not statistically significant in analyses adjusting for age, sex, race, disease duration, and EDSS (p=0.09).

In order to investigate if differences in rates of atrophy may be accounted for by the differences in pre-existing lesion load or lesion accumulation during follow-up, analyses were additionally adjusted for baseline T2 lesion volume fractions and DMT epoch-specific annualized rates of T2 lesion volume fraction change. In these analyses, DMT-H treatment status remained independently associated with reduced rates of subcortical GM (p=0.036), thalamic (0.035) and putaminal (0.014) atrophy. The complete results for the multivariate mixed-effects models are presented in the Supplementary Results.

DISCUSSION

Our results suggest that use of higher efficacy DMTs may be associated with reduced rates of subcortical GM atrophy, an effect that was specifically observed in the thalamus and putamen. These are findings that will clearly need to be reproduced in larger prospective randomized cohorts, but they suggest that these measures show promise for use as imaging outcomes in clinical trials in MS and may improve on whole brain and/or cortical GM measures.

The mechanisms underlying subcortical GM atrophy in MS have not been fully elucidated, but there exists pathologic evidence that both diffuse neurodegeneration and focal lesions occur in the subcortical GM in MS.²⁵ Importantly, our findings were not affected by inclusion of baseline T2 lesion volume or T2 lesion volume change in our statistical models. This suggests that the reduced subcortical GM atrophy observed in our study associated with higher efficacy therapy was, at least to some extent, independent of overt radiologic inflammatory activity preceding or occurring during the course of the study. This may support the notion that higher efficacy DMTs may have effects on inflammatory activity and neurodegeneration, which are not detectable by conventional MRI, such as prevention of the development of subcortical GM lesions, which cannot be routinely visualized at 1.5T or 3T. Additionally, it is possible that the lack of specificity of total brain T2 lesion volume for lesions involving the thalamocortical connections and/or spinal cord inflammatory activity (which was not systematically assessed in our study) and consequent anterograde and/or retrograde degenerative processes involving the subcortical GM, may account for our findings.²⁶

Interestingly, we did not observe an effect in the other subcortical GM structures that were examined, namely the globus pallidus or caudate nucleus. It is possible that this may be due to differential biological effects, however the caudate nucleus and putamen are both components of the striatum and it would be expected that similar effects would be observed in both structures. Notably however, the variability of the rates of change in the globus pallidus and caudate nucleus in both DMT groups was higher than for the other studied regions. This may be attributed to technical factors, especially for the globus pallidus which exhibits relatively poor contrast on T1-weighted scans.²⁷

Importantly, data is limited regarding the effect of DMTs on subcortical GM atrophy. A prior observational study in MS, comparing the rates of brain atrophy between daclizumab and first-line DMT (interferon-beta and glatiramer acetate) treated participants, showed reduced rates of atrophy of the thalamus and caudate nucleus.²⁸ A recent large longitudinal observational study, showed no difference in rates of whole-brain, cortical, and subcortical GM atrophy between DMT-treated and untreated participants.²⁹ However, 90% of the participants with available treatment data were receiving interferon-beta or glatiramer acetate at study entry, rather than higher efficacy therapies. Subcortical GM structure volumes have only been reported as outcomes in a limited number of placebo-controlled clinical trials.^30,31 Post-hoc analyses of brain substructure volumes from completed trials, especially those with active comparators, and inclusion of these measures as exploratory outcomes in ongoing and future clinical trials, may help to further expand our knowledge regarding the differential effects of DMTs on brain substructure atrophy.

Our study has several limitations that merit discussion. Firstly, this is an observational cohort that was analyzed retrospectively. The participants were not assigned to DMTs in a randomized fashion; rather, treatments were selected at the discretion of the treating physician. The DMT groups differed significantly in regards to baseline disability and disease duration (both higher in the DMT-H group, consistent with more of an escalation approach to therapy within our center). However, multivariate analyses including demographic and clinical variables did not alter our findings. This clearly does not overcome the lack of randomization or the many provider-and patient-specific factors involved in the selection of a specific DMT for an individual patient. Thus, it is possible that characteristics that may have influenced the selection of specific DMTs are unaccounted for in our study. Furthermore, the inclusion criteria could have biased to inclusion of subjects with more stable disease. However, in our view, the potential resulting selection bias would likely have favored the DMT-L group, similar to the bias that is present when comparing treated and untreated contemporary cohorts of people with MS.³² Furthermore, brain substructure volumes at baseline were overall lower in the DMT-H group, including in the subcortical GM. Although our analysis accounted for baseline volume and disease duration and changes are presented as relative rather than absolute change, it is conceivable that loss of tissue that is more susceptible to neurodegeneration may have already occurred in these participants. However, there is evidence to support that rates of subcortical GM atrophy (specifically the thalamus) in MS are relatively constant throughout the course of disease, which may argue against this possibility.³³ Another important issue relates to the accuracy and reliability of regional brain volumes. Physiologic factors such as diurnal variation and hydration status have been reported to affect brain volume and further validation of the MRI pipeline for accurate quantification of longitudinal change is necessary, though the segmentation method utilized in our study (MA-CRUISE) appears to have similar reliability as compared to other commonly used automated segmentation methods.^2,34,35 Notably, these physiologic and technical factors would be expected to result in increased random variability of our measurements, rather in a systematic bias. Importantly, a significant strength of our study was that it was performed at a single-center with use of a single scanner for imaging of participants and uniform techniques for post-image acquisition processing and derivation of brain substructure volumes. There are currently significant challenges in utilizing volumetric MRI across centers, even with a consistent scanner manufacturer, field strength and protocol harmonization, an important issue when attempting to apply these measures in multicenter studies.³⁴ A final limitation of our study is the lack of current availability of sufficient longitudinal healthy control data at our center in order to further explore rates of brain substructure atrophy that would be expected in the context of normal aging. However, it has been clearly demonstrated from prior studies that whole brain and GM atrophy occurs at an accelerated rate in people with MS, as compared to healthy controls.^3,33 An important issue, that remains to be resolved, is the relative contribution of normal aging to brain substructure atrophy in MS. It has been suggested that this may not be uniform across the various brain substructures, nor across age groups, and further characterizing these effects of age is warranted.³⁶

In conclusion, our study suggests that use of higher efficacy DMTs is associated with reduced rates of subcortical GM atrophy in MS, especially of the thalamus and putamen. These results support the rationale for future studies to evaluate the clinical significance of these findings and to validate volumetric measures of these structures for use as outcome measures in clinical trials in MS and for clinical monitoring of DMT therapeutic efficacy.