High‐Dose Pulse Glucocorticoid Treatment Prevents White Matter Spinal Cord Pseudoatrophy in Newly Diagnosed Multiple Sclerosis

Abstract

Objective

Spinal cord (SC) atrophy correlates with and predicts the underlying progressive biology in active and non‐active multiple sclerosis (MS), thereby providing a biomarker for clinical trials and patient management. Initiation of disease‐modifying therapy (DMT) may be followed by early pronounced central nervous system (CNS) volume loss due to resolution of inflammation (pseudoatrophy) and confounding the interpretation of atrophy. High‐dose glucocorticoids (HDGs) reduce inflammation and might therefore modify pseudoatrophy.

Methods

One hundred twenty‐three newly diagnosed and DMT‐naïve MS participants (relapsing–remitting, 70% female participants, median age = 36 years, Expanded Disability Status Scale [EDSS] 2.0) were followed for up to 3 years. Forty‐two participants received HDG before baseline magnetic resonance imaging (MRI; DMT‐HDG; median = 52 days, interquartile range [IQR] = 37–71), whereas 60 did not (DMT/no‐HDG). Twenty‐one participants remained untreated (no‐DMT), and 102 started DMT after baseline MRI. SC total cervical cord cross‐sectional area (TCA), gray matter area (GMA), and white matter area (WMA) and regional brain volumes were analyzed using mixed effects models.

Results

The DMT‐HDG, DMT/no‐HDG, and no‐DMT groups had similar demographic, clinical, and radiological features. Pronounced SC pseudoatrophy was observed based on more year 1 versus year 2 volume loss for DMT/no‐HDG (−2.06% vs. 0.83%; P = 0.02) but not DMT‐HDG (−0.51% vs. 0.66%; P = 0.8) and more year 1 volume loss for DMT/no‐HDG compared to DMT‐HDG (−2.06% vs. 0.51%; P = 0.02).

Interpretation

HDG preceding baseline MRI suppresses CNS white matter (WM) pseudoatrophy after DMT initiation, most conspicuously for the SC. Suppression of pseudoatrophy with HDG may improve the fidelity of clinical trials and enhance the feasibility for short‐term trials with SC and brain MRI outcomes in active MS by pretreatment with HDG. ANN NEUROL 2025;98:837–850

1.

Multiple sclerosis (MS) is a chronic disease of the central nervous system (CNS) characterized by inflammation, demyelination, and neurodegeneration, resulting in diffuse CNS volume loss. ¹ Assessing CNS volumes through quantitative magnetic resonance imaging (MRI) has assumed an increasingly important role in clinical trials evaluating the efficacy of therapeutic interventions. ² Brain atrophy is also currently used as a biomarker for patient management and can be included in the no evidence of disease activity (NEDA) formulation. ³ Spinal cord (SC) atrophy represents the strongest prognostic imaging marker for clinical disability and progressive MS, is increasingly used in clinical trials, and has the potential to be an important clinical management biomarker. ⁴ , ⁵

On MRI, CNS volumetric loss can be particularly pronounced immediately following the initiation of disease‐modifying therapies (DMTs). This has been interpreted as the result of pseudoatrophy, a phenomenon characterized by greater volume loss occurring in the first year of DMT use compared to later years or in untreated patients. ⁶ , ⁷ Pseudoatrophy is assumed to be the early volumetric changes detected after DMT initiation that are reversible and related to a decrease of the inflammatory milieu (ie, edema and immune cells) that contributes to CNS volumes. However, as both normal aging and disease‐related neurodegeneration can also occur, even early in the disease course, ⁸ , ⁹ , ¹⁰ at least some of the early volumetric changes observed on MRI may be related to ongoing CNS tissue loss. A commonly used approach to compensate for pseudoatrophy involves re‐baselining CNS volumes at 6 months. However, the true timing of pseudoatrophy is not well characterized and may extend beyond 6 months, creating challenges for new therapeutics undergoing evaluation. The volume loss associated with DMT‐related resolution of inflammation (“true” pseudoatrophy) should be more adequately quantified or suppressed in order to mitigate its impact in clinical trials aimed at detecting effects of treatments on CNS volume loss. ¹¹ , ¹² , ¹³ , ¹⁴

SC pseudoatrophy is understudied. Only a single study to date has systematically addressed SC pseudoatrophy and reported a more conspicuous percentage volume loss within the SC compared to the brain in the first year following DMT initiation. ¹⁴ Because of the increasing importance of SC imaging to MS, understanding the dynamics of early SC volumetric changes is timely and important. ¹⁵ Cervical SC atrophy, assessed through total cervical cord cross‐sectional area (TCA), is detectable early in the disease course and more robustly correlates with disability compared with brain volumetric changes. ⁴ , ¹⁵ , ¹⁶ As both silent progression/progression independent of relapses (PIRA) and conversion to secondary progressive disease are strongly predicted by and associated with cervical cord atrophy, ⁵ its increasing value as a biomarker in trials for neuroprotective drugs is expected. For SC volumetry to assume this role, however, the relationship of early volumetric changes with the initial therapeutic approaches needs to be established and taken into account.

High‐dose glucocorticoids (HDGs) can manage acute relapses by reducing inflammation. In nonhuman primates, HDG may induce fluctuations in brain volume. ¹⁷ , ¹⁸ HDG could also affect SC volume, as suggested by the use of high‐dose methylprednisolone treatment after SC injury and the use of dexamethasone for the first‐line treatment of metastatic SC compression. ¹⁹ , ²⁰ Volumetric changes related to HDG could reasonably be attributed predominantly to their action on inflammation. ²¹ , ²² Nonetheless, most of the literature on pseudoatrophy does not account for the potential contribution of HDG use to CNS volumetric changes. ¹² , ¹³ , ¹⁴ , ¹⁷ , ²³ , ²⁴ , ²⁵

Here, we investigated the effects of DMT and HDG on the development of pseudoatrophy, assessing the dynamics of treatment‐related early volume loss in persons newly diagnosed with MS (PwMS). The quantification of volume loss related to HDG administration is essential to develop a more accurate method to measure early volumetric changes that are driven by ongoing tissue loss and not to the resolution of inflammation (ie, “true” pseudoatrophy).

2. Methods

2.1. Participants

Participants with MS were retrospectively selected from 2 prospective observational cohort studies (EPIC and ORIGINS) conducted at the University of California San Francisco. MRI scans acquired from February 2013 to the end of 2022 were analyzed because the same scanner was used for the 2 studies, and only minor changes to the MRI protocol occurred (see Supplementary Material).

Inclusion criteria for participants:

1. Newly diagnosed with relapsing–remitting course (RRMS) at the time of baseline MRI.

2. DMT naïve at baseline.

3. At least 3 brain and SC MRI acquisitions during the first 3 years of enrollment.

Exclusion criteria:

1. Enhancing lesions at baseline within C2, C3, and C4 levels.

2. Ended or stopped DMTs for an extended time (> 3 months) during the observation period.

3. Unknown status for HDG or low‐dose glucocorticoid (GC) use in the 5 months before baseline.

A final cohort of 123 PwMS was selected (Fig 1), of whom 102 started DMT shortly after their baseline MRI. Of these, 42 had received a cycle (3 to 5 days) of intravenous or oral HDG within 5 months (median = 52 days, interquartile range [IQR] = 37–71) before the baseline examination (DMT‐HDG), whereas 60 had not (DMT/no‐HDG). The other 21 patients remained DMT‐naïve throughout the first year of follow‐up and were included as controls (no‐DMT). Among them, no subject was exposed to HDG before the baseline examination, whereas 2 out of 21 only received HDG right after baseline but well before the first follow‐up ensuring no significant effect of this treatment on the computed yearly change. The research protocol was approved by the Committee on Human Research at UCSF and informed consent was obtained by all participants prior to study enrollment.

FIGURE 1.

Flow chart for participants’ selection.

2.2. MRI Acquisition and Protocol

All participants were scanned on the same Siemens 3T Skyra scanner equipped with either a 20‐ or 64‐channel head–neck coil and a 32‐channel spine coil. The scan protocol included standard sagittal brain 3D T2‐fluid‐attenuated inversion recovery (FLAIR; voxel size = 1 × 1 × 1 mm³, matrix = 256 × 240, 176 slices per slab, TR/TE/TI = 5000/388/1800 ms, and parallel imaging acceleration factor [iPAT] = 3), 3D MPRAGE (voxel size = 1 × 1 × 1 mm³, matrix = 256 × 240, 176 slices per slab, TR/TE/TI = 2300/2.98/900 ms, flip angle = 9 degrees, and iPAT = 2) and a single slice 2D SC phase‐sensitive inversion recovery (PSIR) acquisition with an optimized protocol (axial in‐plane resolution 0.78 × 0.78 mm², slice thickness = 5 mm, matrix = 256 × 256, TR/TE/TI = 4000/3.22/400 ms, and flip angle = 10 degrees, 3 averages, and magnitude and phase‐sensitive images reconstructed). In the more recent years, for 48 (14%) of the total 348 examinations included in this study, an updated 2D PSIR protocol was acquired with the following parameters (axial in‐plane resolution = 0.5 × 0.5 mm², slice thickness = 12 mm, matrix = 400 × 400, TR/TE/TI = 4000/3.63/400 ms, flip angle = 10 degrees, 7 averages, and magnitude and phase‐sensitive images reconstructed). The 2D PSIR protocols were acquired at the C2 to C3 disc level. Whereas baseline acquisitions were all performed with the former protocol, 15% of the first follow‐up images and 35% of the second follow‐up images were acquired with the updated PSIR protocol. The 2 versions of the PSIR were acquired in the same scan session on a large cohort of subjects (~800 visits), and a calibration between the 2 versions of the PSIR protocol was computed to allow pooling of TCA and cervical gray matter areas (cGMAs). Details of this calibration are reported in the Supplementary Material. Sagittal T2‐weighted (in‐plane resolution = 0.7 × 0.7 mm², slice thickness = 2.4 mm, TR/TE = 3000/88 ms, flip angle = 160 degrees, and iPAT = 2) and sagittal TIRM (in‐plane resolution = 0.7 × 0.7 mm², slice thickness = 2 mm, TR/TE/TI = 5610/51/200 ms, flip angle = 150 degrees, and iPAT = 2) sequences of the whole SC were also acquired to properly assess lesion location within the SC.

2.3. Brain Volumetric Analysis

Anatomic 3D T1‐weighted MPRAGE images were used to segment 41 well‐documented neuro‐anatomic structures using the Sequence Adaptive Multimodal SEGmentation (SAMSEG) algorithm. ²⁶ , ²⁷ Volumetric data from these structures were combined and merged to define the following regions of interest (ROIs): cortical gray matter (GM), subcortical GM, total white matter (WM), and total brain volume. To improve the accuracy of the longitudinal analysis, a co‐registration was performed among the images acquired over the 3 different time points (baseline, first follow‐up, and second follow‐up) using Freesurfer's robust templating estimating tool. ²⁸ Volumetric data obtained within the different ROIs were normalized using intracranial volume.

2.4. Spinal Cord TCA, Gray and White Matter Area Computation

TCA was semi‐automatically segmented on the C2 to C3 2D PSIR acquisitions using Jim 9 (Xinapse) as previously described. ²⁹ The cervical SC gray matter area (cGMA) was automatically segmented on the C2 to C3 2D PSIR acquisitions using a deep learning method that uses a U‐net as its model structure. The deep learning method was trained with training data taken from our in‐house cord GM segmentation algorithm which segments GM areas by registering a training set of atlases to the given target. The registered atlases were fitted to a curve used to predict the probability that a given pixel belongs to the GM. ³⁰ The cervical SC white matter area (cWMA) was obtained by subtracting the cGMA from the TCA. The TCA, cWMA, and cGMA outputs were normalized to reduce non‐pathologic interindividual variance using the head size parameter v‐scale from SIENA, as previously reported. ²⁹

2.5. Data Collection and Statistical Analysis

Clinical and demographic characteristics of participants were collected by the study team. Radiographic data, such as the number of spinal cord lesions at baseline, presence of lesions at the C2 to C3 level, brain or SC enhancement, and new brain or SC lesions at follow‐ups were assessed by a neuroradiologist (author S.S.).

Differences in demographics, lesion distribution, disease severity, C2 to C3 PSIR protocol, and DMT use were compared using Wilcoxon rank‐sum or chi‐squared tests, as appropriate. Annualized TCA, cGMA, total brain volume, brain WM, brain GM, subcortical brain GM, and cortical brain GM changes were computed and compared using mixed effects models adjusted for age at baseline, sex, presence of spinal cord lesion, and baseline values. Interaction terms between pre‐baseline HDG and time were included to estimate the effect of pre‐baseline HDG on volume changes over time. Statistical analyses were conducted in JMP Pro version 16 (SAS Institute) and R software version 4.3 (the R Foundation) using the lme4 and emeans packages. The alpha level was set at 0.05, and all tests were 2‐sided.

3. Results

3.1. Clinical and Radiological Characteristics for the Whole Cohort

At the baseline scan, all 123 participants were newly diagnosed with RRMS and were DMT‐naïve, with a median disease duration of 1 year (IQR = 0–1). CNS (brain or SC) enhancement was present in 46%, 5%, and 2% of participants at baseline, and at the first and second annual follow‐up examinations, respectively. SC lesion enhancement (participants with SC enhancing lesions at C2 to C4 levels were excluded from the study) was detectable in 16%, 2%, and 0% of participants at baseline, and at the first and second follow‐up examinations, respectively. At the first follow‐up, 41% of participants showed new non‐enhancing CNS lesions, and, in particular, 15% within the SC. At the second follow‐up, 14% of the participants showed new non‐enhancing CNS lesions, with 3% localizing to the SC. One hundred two of 123 participants started DMT treatment a median of 42 days after baseline (IQR = 14–93). One‐quarter of the treated participants changed treatment during the observation period, at a median of 378 days after baseline (IQR = 244–588 days).

3.2. Comparisons of Demographic, Clinical, and Qualitative Radiological Parameters Between Treatment Groups

DMT‐treated (n = 102) and DMT‐untreated (n = 21) participants did not significantly differ in baseline demographic and most of the radiographic variables (Table 1), nor did the DMT‐HDG and DMT/no‐HDG groups differ in any of these parameters. Compared with the DMT‐treated subjects, the DMT‐untreated showed a lower proportion of enhancement within the CNS and the SC at baseline (see Table 1). The groups, however, did not differ significantly in the percentage of participants with new SC lesions at the first follow‐up, new lesions within the CNS at the first follow‐up or number of new CNS lesions at the first follow‐up. Compared to the DMT‐treated subjects, the DMT‐untreated showed a slightly higher proportion of updated C2 to C3 PSIR protocol acquired at the first follow‐up (see Table 1).

TABLE 1.

Demographic and Qualitative Radiological Comparison Between Groups of Participants

	DMT Treated (n = 102)	DMT Untreated (n = 21)	p	DMT‐HDG (n = 42)	DMT/No‐HDG (n = 60)	p
Age, median [IQR], min‐max	36 [30, 43], 18–54	32 [28, 42], 21–56	0.22 a	36.5 [31, 43], 18–54	36 [30, 43], 21–54	0.82 a
Sex (%F, number)	68% (70)	76% (16)	0.5 b	66.6% (28)	70% (42)	0.72 b
White (%, number)	84% (86)	66% (14)	0.06 b	85% (36)	83% (50)	0.74 b
EDSS at baseline, median [IQR], min‐max	2 [1.5, 2.5], 0–4.5	1.75 [1, 2.5], 0–3.5	0.18 a	2 [1.5, 2.5], 0–4.5	2 [1.5, 3], 0–6	0.32 a
Disease duration at baseline, median [IQR], min‐max	1 [0, 1], 0–14	1 [0, 1.5], 0–2	0.69 a	0 [0, 1], 0–11	1 [0, 1], 0–14	0.19 a
Number of attacks before baseline, median [IQR], min‐max	1 [1, 2], 1–4	1 [1, 1], 1–3	0.14 a	1 [1, 2], 1–4	1 [1, 2], 1–3	0.93 a
Total cord lesion at baseline, median [IQR], min‐max	2 [1, 5], 0–22	1 [0, 5], 0–20	0.45 a	2 [1, 5], 0–22	2 [1, 4], 0–19	0.75 a
Cervical cord lesion at baseline, median [IQR], min‐max	1 [0, 3], 0–10	1 [0, 3], 0–9	0.23 a	1 [0, 3], 0–10	1 [0, 3], 0–7	0.53 a
Lesions at C2 to C3 level (%, number)	26% (27)	23% (7)	0.84 b	24% (10)	28% (17)	0.6 b
Enhancement anywhere CNS at baseline (%, number)	50% (51)	23% (5)	0.03 a	45% (19)	53% (32)	0.4 b
Enhancement spine at baseline (%, number)	20% (20)	0% (0)	0.03 a	14% (6)	23% (14)	0.26 b
New lesion spine at the first follow‐up (%, number)	16.6% (17)	9.5% (2)	0.4 a	14% (6)	18% (11)	0.6 b
New lesion CNS at the first follow‐up (%, number of participants)	42% (43)	33.3% (7)	0.43 a	43% (18)	42% (25)	0.9 b
Number of new CNS lesions at the first follow‐up, median [IQR], min‐max	0 [0, 1], 0–10	0 [0, 1], 0–4	0.16 b	0 [0, 1], 0–6	0 [0, 1], 0–10	0.98 a
Updated C2 to C3 PSIR protocol at the first follow‐up (%, number)	7% (7)	23% (5)	0.02 a	7.1% (3)	7.5% (4)	0.9 a
Updated C2 to C3 PSIR protocol at the second follow‐up (%, number)	N.A.	N.A.		36% (15)	35% (21)	0.94 a

^a Wilcoxon rank‐sum test.

^b CHI‐squared test p value in bold when statistically significant.

CNS = central nervous system; EDSS = Expanded Disability Status Scale; IQR = interquartile range; N.A. = not applicable; PSIR = phase‐sensitive inversion recovery.

3.3. Comparison of DMT‐HDG and DMT/No‐HDG Groups

Among the 102 participants starting DMT, there was no statistically significant difference in the percentage of participants who began therapy with monoclonal antibodies (ocrelizumab, ozanimod, rituximab, and natalizumab) or dimethyl fumarate between those who received GC and those who did not (Tables 2 and 3). The 2 groups also did not differ in terms of the interval between baseline visit and DMT start, the percentage of participants changing DMT, or in the interval between baseline and DMT change. The only significant difference between groups was the percentage of participants receiving HDG in the first year after baseline. Thirty percent of DMT/no‐HDG received HDG in the first year compared to 7% of DMT‐HDG (p = 0.005).

TABLE 2.

Specific DMT Used Within DMT‐HDG and DMT/No‐HDG Groups (%, Number)

Therapy	DMT‐HDG % (n = 42)	DMT/No‐HDG % (n = 60)
Interferon beta‐1a	2.4% (1)	1.7% (1)
Glatiramer acetate injection	19% (8)	20% (12)
Fingolimod	14.4% (6)	15% (9)
Ocrelizumab	38% (16)	23.3% (14)
Ozanimod	0% (0)	1.7% (1)
Rituximab	7.15% (3)	6.6% (4)
Dimethyl fumarate	9.5% (4)	16.7% (10)
Natalizumab	7.15 (3)	15% (9)
Peginterferon Beta‐1a	2.4% (1)	0% (0)

DMT = disease‐modifying therapy; HDG = high‐dose glucocorticoid; No‐HDG = no high‐dose glucocorticoid received.

TABLE 3.

Comparison of Treatments Between DMT‐HDG and DMT/No‐HDG Groups

	DMT‐HDG (n = 42)	DMT/No‐HDG (n = 60)	p
Treatment type: participants on monoclonal antibodies DMTs (%, number)	52% (22)	47% (28)	0.6 a
Treatment type: participants on dimethyl fumarate (%, number)	10% (5)	17% (9)	0.3 a
Days from baseline to treatment start, median [IQR], min‐max	37.5 [4, 80], 0–180	42 [22, 105], 0–175	0.21 b
Treatment change (%, number)	21.4% (9)	28% (17)	0.43 a
Days from baseline to treatment change, median [IQR], min‐max	359 [350, 425], 196–525	455 [234, 630], 77–951	0.74 a
Participants receiving HDG in the first year (%, number)	7% (3)	28% (17) c	0.008 a
Participants receiving HDG in the first year for treatment of clinical relapses (%, number)	7% (3)	17% (10) c	0.15 a
Number of HDGs in the first year, median [IQR], min‐max	0 [0, 0], 0–1	0 [0, 1], 0–4	0.004 a

^a Chi‐squared test.

^b Wilcoxon rank‐sum test. The p values in bold indicate when statistically significant.

^c Although 17 participants in the DMT/no‐HDG group received HDG in the first year, 7 of them did not have relapses in the first year and only received HDG as part of the initial work‐up and therapeutic approach right after the baseline examination.

DMT = disease‐modifying therapy; HDG = high‐dose glucocorticoid; IQR = interquartile range; No‐HDG = no high‐dose glucocorticoid received.

3.4. DMT‐Related Spinal Cord Pseudoatrophy in the Absence of Pre‐Baseline HDG Exposure

In the DMT/no‐HDG group (n = 60), larger volumes losses were observed in the first‐year coinciding with DMT initiation than the second year within SC TCA and SC cWMA (TCA = −2.06% vs. −0.83%: difference = −1.23%, p = 0.02; and SC cWMA = −2.22% vs. −0.56%: difference = 1.66, p = 0.005; Table 4A and Fig 2A), consistent with pseudoatrophy. Further support for pseudoatrophy were larger observed volume losses in the first year for the DMT/no‐HDG group compared to the DMT untreated group (TCA = −1.84% vs. −0.66%, difference = −1.18, p = 0.04; and SC cWMA = −1.97% vs. −0.5%, difference = 1.47, p = 0.045; Table 4B).

TABLE 4.

Spinal Cord Pseudoatrophy as Suggested by (A) Greater Volume Loss in the First Year When Compared to the Second Year for all Subjects Receiving DMT (n = 102), for Subjects Receiving DMT After Recent Exposure to HDG (DMT‐HDG, n = 42) and for Subjects Receiving DMT in the Absence of Recent Exposure to HDG (DMT/no‐HDG, n = 60). (B) Effects of DMT Usage on the First Year of Spinal Cord Change

(A) All Subjects Receiving DMT (n = 102)				DMT‐HDG (n = 42)			DMT/No‐HDG (n = 60)
Region of interest	1st year % change	2nd year % change	p	1st year % change	2nd year % change	p	1st year % change	2nd year % change	P
C2‐C3 TCA	−1.43 ± 0.34	−0.85 ± 0.26	0.11 a	−0.51 ± 0.45	−0.66 ± 0.45	0.8	−2.06 ± 0.41	−0.83 ± 0.41	0.02 b
C2‐C3 WMA	−1.31 ± 0.3	−0.57 ± 0.31	0.11 a	−0.20 ± 0.53	−0.20 ± 0.53	1	−2.22 ± 0.45	−0.56 ± 0.45	0.005 b
C2‐C3 GMA	−1.44 ± 0.56	−2.17 ± 0.52	0.32	−1.88 ± 0.93	−2.59 ± 0.93	0.6	−1.13 ± 0.86	−2.0 ± 0.86	0.35
Total brain volume	−0.27 ± 0.07	−0.25 ± 0.06	0.8	−0.31 ± 0.1	−0.31 ± 0.1	0.9	−0.34 ± 0.1	−0.18 ± 0.1	0.27
Cortical GM	−0.24 ± 0.1	−0.23 ± 0.08	1	−0.23 ± 0.1	−0.21 ± 0.1	0.9	−0.25 ± 0.1	−0.26 ± 0.1	0.9

(B) Region of interest	No DMT (n = 21) % changes	All Subjects Receiving DMT (n = 102) % changes	p	DMT HDG (n = 42) % changes	p	DMT/No‐HDG (n = 60) % changes	p
C2‐C3 TCA	−0.66 ± 0.6	−1.43 ± 0.34	0.22	−0.6 ± 0.5	0.8	−1.84 ± 0.4	0.04 b
C2‐C3 WMA	−0.5 ± 0.7	−1.31 ± 0.4	0.28	−0.3 ± 0.5	0.6	−1.97 ± 0.5	0.045 b
C2‐C3 GMA	−0.2 ± 1.2	−1.28 ± 0.7	0.4	−1.58 ± 0.1	0.4	−1.18 ± 0.9	0.53
Total brain volume	−0.06 ± 0.2	−0.3 ± 0.1	0.2	−0.35 ± 0.2	0.24	−0.33 ± 0.2	0.22
Cortical GM	−0.06 ± 0.2	−0.18 ± 0.1	0.5	−0.27 ± 0.2	0.6	−0.22 ± 0.2	0.5

^a Statistically significant.

^b Trend toward statistical significance.

DMT = disease‐modifying therapy; GM = gray matter; GMA = gray matter area; HDG = high‐dose glucocorticoid; No‐HDG = no high‐dose glucocorticoid received; TCA = total cord area assessed at C2 to C3; WMA = white matter area.

FIGURE 2.

Boxplots of TCA, WMA (A), and GMA (B) measurements in the different time points within different subgroups. (A) TCA (green colors) and WMA (yellow colors) measurements within the different subgroups reveal similar volumes for the no‐DMT and DMT‐HDG groups across the different time points (baseline, first year, and second year from left to right, respectively). Only the DMT/no‐HDG shows statistically significant changes in WMA and GMA volumes (*), and only in the passage from baseline to the first year, with stability of volumes from the first year to the second year. (B) GMA (orange) measurements within the different subgroups reveal similar volumes for all groups across different time points (baseline, first year, and second year from left to right, respectively). DMT = disease‐modifying therapy; GMA = gray matter area; HDG = high‐dose glucocorticoid; TCA = total cervical cord cross‐sectional area; WMA = white matter area. [Color figure can be viewed at www.annalsofneurology.org]

3.5. Impact of Pre‐Baseline HDG on Spinal Cord Pseudoatrophy Development

In contrast to the DMT/no‐HDG group, DMT‐HDG participants (n = 42) did not show significant differences between the first and second years rate of change (TCA = −0.51% vs. −0.66%, difference = 0.15, p = 0.8; and SC cWMA = −0.20% vs. −0.20%, difference = 0, p = 1; Table 4A and Fig 2A). Suppressed pseudoatrophy was also supported by similar first‐year rates of change of DMT‐HDG and DMT untreated groups (TCA = −0.6% vs. −0.66%, difference = 0.06, p = 0.8; and SC cWMA = −0.3% vs. −0.5%, difference = 0.2, p = 0.6; see Table 4B and Fig 2A).

3.6. Comparison of the DMT/HDG and DMT/No‐HDG Groups Further Supports Suppression of Pseudoatrophy With Pre‐Baseline HDG

Over the first year, the TCA showed greater volume loss for DMT/no‐HDG when compared to DMT‐HDG (−2.06% and −0.51%; difference = −1.55 p = 0.02; see Fig 2A and Fig 3). Over the second year, TCA atrophy rates were not significantly different between the DMT/no‐HDG group and DMT‐HDG groups (respectively, −0.83% and −0.66%, p = 0.8; see Table 4). There was no effect of pre‐baseline HDG administration on SC GMA changes (Table 5, Fig 2B).

FIGURE 3.

Hypothesis. We hypothesize that a recent exposure to HDG is associated with less pronounced DMT‐related volumetric changes within the spinal cord. DMT = disease modifying therapy; DMT‐HDG (A) = received HDG up to 5 months before baseline; DMT/no‐HDG (B) = did not receive HDG in the 5 months before baseline; HDG = high dose glucocorticoid treatment; TCA = total cord area assessed at C2 to C3. [Color figure can be viewed at www.annalsofneurology.org]

TABLE 5.

Pre‐Baseline HDG Affects the First Year WM % Volumetric Changes in Both the Brain and Spinal Cord With no Effect on the GM

	Brain				Spinal cord
		DMT‐HDG (n = 42)	DMT/No‐HDG (n = 60)	P GC Subgroups Comparison	DMT‐HDG (n = 42)	DMT/No‐HDG (n = 60)	P GC Subgroups Comparison
Brain WM								Spinal cord cWMA
	First year	−0.31 ± 0.1	−0.44 ± 0.1	0.45	−0.20 ± 0.4	−2.22 ± 0.4	0.01 a
	Second year	−0.33 ± 0.1	−0.12 ± 0.1	0.24	−0.20 ± 0.4	−0.56 ± 0.4	0.97
	P First vs. second year	0.93	0.03 a		1	0.01 a
Brain subcortical GM								Spinal cord cGMA
	First year	−0.73 ± 0.7	−0.74 ± 0.1	0.9	−1.88 ± 0.7	−1.13 ± 0.8	0.83
	Second year	−0.39 ± 0.7	−0.19 ± 0.1	0.27	−2.59 ± 0.7	−2.0 ± 0.8	0.85
	P first vs. second year	0.08 b	< 0.001 a		0.6	0.35

^a Statistically significant.

^b Trend toward statistical significance.

cGMA = cervical SC gray matter area; cWMA = cervical SC white matter area; DMT = disease‐modifying therapy; GM = gray matter; HDG = high‐dose glucocorticoid; No‐HDG = no high‐dose glucocorticoid received; WM = white matter.

3.7. Relationship of Early SC Pseudoatrophy With Same‐Level SC Lesions and First Year Relapses

Although subjects with enhancing lesions at baseline at the C2 to C3 cord level were excluded from the study to avoid the impact of acute lesions on cord measurements, participants with non‐enhancing lesions at this level were included. For the entire DMT‐treated group, the TCA volume loss was greater over both the first and second year in participants with non‐enhancing lesions at C2 to C3 level (n = 27) compared with participants without C2 to C3 level lesions (n = 75, −2.35% vs. −1.05%, p = 0.02 and − 1.86% vs. −0.43%, p = 0.01 for the first and second year, respectively). However, when excluding all participants with non‐enhancing lesions at the C2 to C3 level and considering only 32 and 43 participants for DMT‐HDG and DMT/no‐HDG groups, respectively, there was still a trend toward a greater first‐year decrease in TCA in the DMT/no‐HDG group when compared to the DMT‐HDG group (−1.52% vs. −0.42%, p = 0.06; Fig 4). This same analysis is significant when specifically considering exclusively SC cWMA (−1.81% vs. −0.32%, p = 0.03). With regard to the possible influence of inflammatory relapses on first‐year volume changes, when excluding participants presenting relapses in the first year from both groups (10 subjects for DMT/no‐HDG and 3 subjects for DMT‐HDG), the DMT/no‐HDG group (n = 50) had a significantly greater TCA loss in the first year when compared to the second year (−1.81% vs. −0.37%, p = 0.009); first‐year TCA volume loss was greater in DMT/no‐HDG (n = 50) when compared to the DMT‐HDG group (n = 39, −1.81% vs. −0.57%, p = 0.015).

FIGURE 4.

The first year and second year TCA and WMA % change when excluding participants with non‐enhancing lesions at the C2 to C3 level. (A) When excluding participants with non‐enhancing lesions at the C2 to C3 level, the DMT/no‐HDG group (red; n = 43) showed a trend toward bigger first year TCA change (^) when compared to the DMT‐HDG group (blue; n = 32, p = 0.06). The TCA change is similar among groups in the second year. (B) When excluding participants with non‐enhancing lesions at the C2 to C3 level, the DMT/no‐HDG group (red; n = 43) showed a bigger first year WMA change (*) when compared to the DMT‐HDG group (blue; n = 32, p = 0.03). The WMA change is similar among groups in the second year. DMT = disease modifying therapy; DMT‐HDG (A) = received HDG up to 5 months before baseline; TCA = total cord area assessed at C2 to C3. [Color figure can be viewed at www.annalsofneurology.org]

3.8. DMT‐Related Brain Pseudoatrophy and Marginal Effect of HDG on Brain WM Volume Change

Atrophy rates were also examined in cortical and subcortical GM and WM of the brain. Rates of change within the subcortical GM in the DMT/no‐HDG and DMT‐HDG groups were similar in both the first and second year (see Table 5). Regardless of pre‐baseline HDG administration status, all participants starting DMT showed greater volume loss within the subcortical GM in the first year compared with the second year. For the DMT/no‐HDG group, this was more accentuated (−0.74% vs. −0.19%, p < 0.001), whereas for the DMT‐HDG group it was only a trend (−0.73% vs. −0.39%, p = 0.08; Table 5). When compared to the DMT untreated group (n = 21), only DMT/no‐HDG showed greater subcortical GM volume loss in the first year (−0.82% vs. −0.28, p = 0.04). For brain WM, differences between the first year and second year rates of changes were detectable only for the DMT/no‐HDG group (−0.44 vs. −0.12, difference = 0.32, p = 0.03) and not for the DMT‐HDG group (−0.31 vs. −0.33, difference 0.02, P = 0.9; see Table 5). No pseudoatrophy was observed for cortical GM or total brain volume.

4. Discussion

The SC cross‐sectional area is a promising biomarker to assess progression in MS. Radiological measures that strongly predict disability progression are required to efficiently test the efficacy of disease modifying treatments and improve clinical management, especially for patients with progression independent of relapse‐related activity. As such, understanding the dynamics of volumetric changes in the SC following the initiation of treatment is an essential prerequisite for detecting and investigating its prognostic value for the progressive biology of MS.

In this work, we analyzed the dynamics of early SC volumetric changes in newly diagnosed RMS and the relationship of those changes to treatments (both DMTs and HDG). We found that following the start of DMT, SC pseudoatrophy was detectable exclusively in those participants not recently exposed to HDG before the baseline scan. Furthermore, the presence of pseudoatrophy and its suppression by HDG following DMT initiation was mostly driven by changes within WM regions. These results represent the first evidence for the impact of HDG on SC pseudoatrophy.

The novelty of the current work is threefold. First, all patients were DMT naïve, removing the possible confound of previous treatments. Second, we were able to directly compare similar groups of participants with and without HDG administration prior to their baseline scan, providing stronger evidence for a pseudoatrophy interpretation of the observed volume loss. Third, for the first time, both GM and WM compartments of the SC were investigated. These findings have important implications for the design of MS DMT clinical trials using SC volumetry as an outcome and for any future use of SC atrophy as a clinical management tool.

The effect of HDG administration on the development of CNS pseudoatrophy could significantly affect the interpretation of the early volumetric changes observed in clinical trials in which HDGs are administered in a heterogeneous fashion before baseline. Only few studies have previously investigated the effects of HDG on CNS volumes in MS. They were focused only on the brain and in much smaller cohorts than the one presented here. ³¹ , ³² In a study of 26 participants, followed 3 months before and after HDG, brain volumes were found to be significantly reduced for 2 months after HDG administration used to treat acute relapses. ³¹ Here, we similarly observe that DMT‐related CNS early volumetric changes are significantly impacted by HDGs when they are administered up to a few months before baseline (with a median of 52 days, IQR = 37–71 days). Our data suggest that the standardized use of HDGs before the baseline MRI in clinical trials could be a potential strategy to minimize the confound of CNS pseudoatrophy after DMT initiation. Removing the effect of inflammation resolution on the early CNS volumetry, would greatly benefit the assessment of neuroprotective effects of therapies without the need for complex re‐baselining strategies of uncertain value.

This study also provides the first evidence for the effect of HDG administration on SC volumetry in MS. The TCA atrophy measures revealed a dramatic—nearly 3 times faster—rate of volume loss in the first year for DMT/no‐HDG compared with DMT‐HDG participants. Empirical evidence derived from the treatment of SC injury and metastatic compression suggests that HDG can significantly reduce edema and inflammation within the SC, promoting stabilization of vascular membranes at the compression/injury site, and resulting in clinical benefits (reductions in back pain and neurological deficits). ¹⁹ , ²⁰ The only prior study assessing pseudoatrophy in the SC did not consider the administration of HDG, but similarly found that early volumetric changes after DMT initiation were more conspicuous within the SC when compared to the brain. ¹⁴

It is not known why pseudoatrophy might be more pronounced in the SC compared to the brain. One possible explanation considers the smaller ratio between microscopic lesional and inflamed tissue to unaffected tissue in the SC compared to the brain. In the SC, a higher amount of lesional tissue, known to shrink after the start of DMT, could result in a greater propensity to show early volume loss due to inflammation resolution. In our analysis, HDG‐related TCA changes were mainly driven by changes within the SC cWMA. However, even when specifically excluding the participants with WM lesions at the level of measurement (C2 to C3), the DMT/no‐HDG group had greater loss in the SC cWMA in the first year when compared with DMT‐HDG subjects. This suggests an effect of HDG on whole WM, not only the lesional tissue. As HDG are known to act predominantly on inflammation, the early volumetric changes related to their administration are likely to represent a true pseudoatrophy phenomenon.

In contrast to the WM, similar GM volumetric changes occurred regardless of pre‐baseline HDG exposure. Prior literature on brain pseudoatrophy describes divergent findings regarding the specific anatomic compartments in which DMT‐related early volumetric changes are more pronounced. Older studies report more conspicuous volume changes after DMT initiation within the WM, ⁶ , ²³ , ³³ , ³⁴ whereas a few more recent ones highlight a predominant GM involvement. ²³ , ²⁴ Only one of these studies accounted for HDG use, and it did so for only up to 1 month before baseline. ³³ We speculate that differences in exposure to pre‐baseline HDG might be at least partially responsible for these divergent findings.

Additional considerations are related to the possible causes for which the HDG administration could result in different effects on the GM and WM compartments. As a pre‐baseline exposure to HDG is associated with significantly smaller CNS WM volumetric changes, with no effect on early GM atrophy, we speculate that early volumetric changes might have different pathological substrates within these 2 anatomic compartments. DMT‐related early volumetric changes within the WM are likely to result primarily from the stabilization of vascular membranes (ie, blood brain barrier) and resolution of inflammation, and are, therefore, likely to represent a true pseudoatrophy phenomenon. In contrast, early volume loss within the GM could result from various factors. A recent report described an age‐dependent reduction in cortical volume loss rates in RMS, suggesting more pronounced neurodegeneration in the earlier phases of the disease. ⁸ The early DMT‐related volumetric changes observed within the GM, described as pseudoatrophy in prior studies, are, therefore, more likely to reflect a combination of both inflammation resolution and ongoing tissue loss. This speculation is further supported by 2 additional observations. The first is on different possible pathologies of focal MS lesions in WM versus GM. Whereas WM lesions are associated in the acute phase with blood brain barrier disruption, GM lesions are not, and are in contrast thought to result from perivascular inflammation. ³⁵ The WM volume loss associated with HDG administration could, therefore, be related, at least partially, to a decrease in vascular permeability within focal WM lesions and resolution of focal edema. As an effect on vascular permeability is less likely for focal GM pathology, GM volumes would be less affected by HDG administration. Interestingly, a recent study assessing the effects of systemic and inhaled glucocorticoids on brain volumes and diffusion tensor imaging metrics suggests that WM integrity is more sensitive to GC administrations than GM volume. ³⁶

Unlike prior studies assessing brain pseudoatrophy, ²³ , ²⁴ we observed GM volumetric changes exclusively confined to subcortical regions. Although the DMT‐HDG and DMT/no‐HDG groups had similar first year changes within the subcortical GM, DMT/no‐HDG participants only presented significantly greater subcortical GM changes in the first year when compared to the second years (see Table 5), and this could be related to some marginal effect of the HDG within those subcortical structures constituted by a mix of GM and WM (ie, thalamus). No significant change was detectable within the brain cortical or spinal cord GM. It should be noted, however, that early brain volumetric changes in MS (median disease duration of 1 year for our cohort) are predominantly located within the deep GM, which shows the fastest rates of atrophy among all brain compartments and may also harbor higher levels of inflammation. ³⁷ , ³⁸ , ³⁹ Volume loss within the deep GM can be seen even without a decrease in total brain volume or total GM. ⁴⁰ , ⁴¹ In a recent systematic study assessing the role of the different ROIs in determining the change in total brain volume, deep GM volume change was more than twice as accelerated as total brain volume change and more than 5 times higher than the change in cortical thickness. ⁴² Methodological differences in assessing GM volumes within subcortical nuclei versus cortical folds could also be responsible, at least partially, for the different sensitivity of detecting overtime changes within these different ROIs.

Our study has several limitations. First, pseudoatrophy was assessed in PwMS treated with various DMTs, and our cohort size was not large enough to analyze volumetric and area changes separately by specific treatment. Literature on pseudoatrophy is dated, and although this phenomenon has been assessed in both lower (IFN alpha/beta ¹⁴ , ²⁴ ) and higher efficacy therapies (Natalizumab ³³ , ⁴³ and Fingolimod ⁴⁴ , ⁴⁵ ), there are few direct studies, ¹⁴ , ²⁴ , ³³ , ⁴³ whereas the others only refer to pseudoatrophy as a possible interpretation of their data. In light of this, the findings of this study are even more compelling because, for the first time, to the best of our knowledge, we are assessing pseudoatrophy in a more updated MS therapeutic landscape. A second limitation is the retrospective design. Although data were gathered prospectively, the analyses were performed post hoc and should therefore be considered hypothesis generating rather than declarative. Replicating these observations in independent datasets is of high priority for future studies. A third limitation is that HDG use was not standardized in our cohort. Differences in our clinician's practice, severity of presenting symptoms, and other potential confounders likely influence the decision to use HDG. Reassuringly, the groups receiving HDG or not prior to the baseline visit did not differ in any clinical, demographic, or qualitative radiological variable known to potentially affect early volumetric CNS changes in MS. Nonetheless, we recognize that data generated from observational studies may have unrecognized confounders that can only be controlled for using treatment randomization in adequately powered randomized controlled trials.

Most of the clinical trials in MS use MRI‐based brain volumes and thalamic volumes and are now including spinal cord volumes. The potential for pseudoatrophy following treatment initiation has been recognized but often only inferred from either immediate volume loss in the treatment versus the placebo arms or from increased loss in the first observation period compared with later intervals. The solution developed to take into account this phenomenon is to re‐baseline at 6 months or 1 year, although the exact knowledge of the timing of the influence of pseudoatrophy can only be guessed at. Suppression of pseudoatrophy through an HDG administration within the 2 months preceding the baseline MRI could potentially improve clinical trial designs and greatly reduce the need for re‐baselining or improve confidence that re‐baseline at 6 months is adequate. These considerations are particularly relevant for the design of critically important early‐phase studies that span short observation times. A decision regarding the administration of HDG across clinical trials would need, however, to take into account the well‐known side effects of this treatment, which can include altered tolerance to glucose, insomnia, mood alterations, gastritis and/or gastric ulcers, and hypertension. Administration of HDG for the sole purpose of improving the management in clinical trials without real evidence of a clinical benefit for the participants might be ethically acceptable only when no risk factor for the above‐mentioned conditions are reported. Therefore, this strategy might present the drawback of confining eligibility for such clinical trials to participants who have received HDG in the last 2 months for clinical reasons and to those who do not present known susceptibility for HDG side effects.

5. Conclusions

Recent exposure to HDG largely suppresses the development of DMT‐related early volume loss within the CNS WM (pseudoatrophy), with no significant effect on the GM compartment. As HDGs are known to act predominantly on inflammation, the differences observed in the dynamic of DMT‐related changes across GM and WM compartments suggest 2 distinctive pathological substrates for early volume loss in MS. HDG‐related effects are particularly conspicuous for the SC, where individuals without prior HDG exposure showed an almost 3 times faster rate of atrophy in the first year of observation compared with those with prior HDG exposure. These observations imply that active suppression of pseudoatrophy or otherwise controlling for pre‐baseline HDG administration will be an important consideration for clinical trials in which SC areas, as well brain and thalamic volumes, are used to measure therapeutic efficacy.

Potential Conflict of Interest

S.S., N.P., V.A.S., S.C., W.A.S., H.Z., A.B., E.C., M.D., J.J., A.A., C.C., A.B., R.G., M.H., A.S., R.M.B., J.M.G., D.S.G., A.J.G., J.R.O., E.W., M.R.W., S.S.Z., B.A.C., S.L.H., and R.G.H. have no conflict of interest directly related to the contents of this article.

Supporting information

DATA S1. Supporting Information.

Data Availability

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.