EHS
EHS

Association of Slowly Expanding Lesions on MRI With Disability in People With Secondary Progressive Multiple Sclerosis


Abstract

Background and Objective To explore the relationship between slowly expanding lesions (SELs) on MRI and disability in secondary progressive multiple sclerosis (SPMS).

Methods We retrospectively studied 345 patients with SPMS enrolled in the MS-SMART trial. They underwent brain MRI at baseline and at 24 and 96 weeks. Definite SELs were defined as concentrically expanding T2 lesions, as assessed by nonlinear deformation of volumetric T1-weighted images. Associations of SEL volumes with other MRI metrics and disability were assessed through Pearson correlations and regression analyses.

Results Averaged across patients, 29% of T2 lesions were classified as being definite SELs. A greater volume of definite SELs correlated with a higher total baseline T2 lesion volume (r = 0.55, p < 0.001) and percentage brain volume reduction (r = −0.26, p < 0.001), a higher number of new persisting T1 black holes (r = 0.19, p < 0.001), and, in a subset of 106 patients, with a greater reduction in magnetization transfer ratio (adjusted difference 0.52, p < 0.001). In regression analyses, a higher definite SEL volume was associated with increasing disability, as assessed by the Expanded Disability Status Scale (β = 0.23, p = 0.020), z scores of the Multiple Sclerosis Functional Composite (β = −0.47, p = 0.048), Timed 25-Foot Walk Test (β = −2.10, p = 0.001), and Paced Auditory Serial Addition Task (β = −0.27, p = 0.006), and increased risk of disability progression (odds ratio 1.92, p = 0.025).

Discussion Definite SELs represent almost one-third of T2 lesions in SPMS. They are associated with neurodegenerative MRI markers and related to clinical worsening, suggesting that they may contribute to disease progression and be a new target for therapeutic interventions.

Glossary

EDSS=
Expanded Disability Status Scale;
FLAIR=
fluid-attenuated inversion recovery;
JE=
Jacobian expansion value;
MS=
multiple sclerosis;
MS-SMART=
MS-SMART: Multiple Sclerosis-Secondary Progressive Multi-Arm Randomisation Trial;
MSFC=
Multiple Sclerosis Functional Composite;
MTR=
magnetization transfer ratio;
NBV=
normalized brain volume;
NHPT=
Nine-Hole Peg Test;
PASAT=
Paced Auditory Serial Addition Test;
PBH=
persistent black hole;
PBVC=
percent brain volume change;
PPMS=
primary progressive multiple sclerosis;
pu=
percent units;
RRMS=
relapsing-remitting multiple sclerosis;
SDMT=
Symbol Digit Modalities Test;
SEL=
slowly expanding lesion;
SPMS=
secondary progressive multiple sclerosis;
T25FW=
Timed 25-Foot Walk

Recent studies have shown that imaging chronic active (sometimes referred to as smoldering) lesions may be particularly relevant in progressive multiple sclerosis (MS).1 Pathologically, they are characterized by the presence of activated peripheral iron-rich macrophages/microglia associated with myelin breakdown,2,3 reflecting chronic inflammatory activity and incomplete remyelination, and substantial axonal injury. Chronic active lesions are more prevalent in progressive MS phenotypes and are associated with the accrual of disability.2,4

On conventional MRI, active lesions can be identified when they newly appear (or expand) on T2-weighted scans or show gadolinium enhancement on T1-weighted scans.5,6 Such activity is clinically associated with relapses.7,8 Conversely, persistent T1 hypointense lesions (or black holes) are associated with axonal injury,9,,11 and conventionally thought to be indicative of the end stage of lesion evolution.

Chronic active lesions are referred to in histopathologic literature as smoldering, slowly expanding, or mixed active–inactive as opposed to early active, inactive, and remyelinated lesions.12,13 Slowly expanding (or evolving) lesions (SELs) have been proposed as a novel MRI marker of chronic active lesions, and they can be identified using nonrigid longitudinal registration (based on local deformations when aligning consecutive scans) to look for local volume changes in individual lesions.14 They were initially described in pooled research trials involving relapsing-remitting MS (RRMS) and primary progressive MS (PPMS) (n = 1,334, n = 555, respectively),14 with SELs observed in both groups, but noticeably more so in PPMS (11.3% vs 8.6% of the total T2 lesion burden). Compared with non-SELs, SELs rarely showed gadolinium enhancement, whereas T1 intensity within SELs was lower at baseline and showed a greater decrease over time than in non-SELs.14 In a subsequent study in PPMS,15 a higher T1 lesion volume in SELs predicted clinical progression, thereby suggesting the possibility that SELs could be in vivo predictors of axonal loss observed in chronic active lesions.13

The aim of this study was to investigate the associations of SELs with physical and cognitive disability scores in secondary progressive MS (SPMS). We also performed a structural analysis of magnetization transfer ratio (MTR) in order to explore the development of tissue damage within SELs in a subset of patients. We completed our investigation with a descriptive radiologic analysis of SELs, including their relationship with other conventional MRI inflammatory and neurodegeneration markers, such as T2 lesion volume change, manually detected new or enlarging T2 lesions, new persistent black holes (PBHs), and brain atrophy.

Methods

Participants, Clinical Assessments, and MRI Acquisitions

This is a retrospective analysis of 345 patients with SPMS who were enrolled in the multicenter, phase 2b MS-SMART (MS-SMART: Multiple Sclerosis-Secondary Progressive Multi-Arm Randomisation Trial; NCT01910259; details reported previously).16 Inclusion criteria for this study were availability of MRI scans (of sufficient quality for SEL analysis) and clinical assessments at all 3 time points (baseline, week 24, and week 96). All patients were scanned using 1.5T or 3T MRI scanners at baseline, week 24, and week 96, with the following acquisitions: 3D isotropic T1-weighted (T1); 2D proton density and T2-weighted (T2); and 2D fluid-attenuated inversion recovery (FLAIR).16 A subset of 106 patients scanned in London and Edinburgh also had 3D MTR imaging at baseline and week 96.17 Details of the MRI scanners and acquisition parameters are shown in eTable 1 (links.lww.com/WNL/B869).

Clinical data included the Expanded Disability Status Scale (EDSS),18 Symbol Digit Modalities Test (SDMT),19 and Multiple Sclerosis Functional Composite (MSFC),20 the latter calculated as the composite of z scores of the 3 subcomponents: Nine-Hole Peg Test (NHPT), Timed 25-Foot Walk (T25FW), and Paced Auditory Serial Addition Test (PASAT). Disability progression was defined as a binary outcome by 1-point increase in EDSS (considering the EDSS change from baseline to week 96) if the baseline score was ≤5.0, or a 0.5-point increase if the baseline score was >5.0.21,22 No assessments were undertaken beyond the 96 weeks of the trial; therefore, it was not possible to evaluate confirmed disability progression.

Standard Protocol Approvals, Registrations, and Patient Consents

Written informed consent was obtained for all the participants and the study was approved by the local research ethics committee. Fully anonymized clinical and MRI data were analyzed at Queen Square MS Centre, University College London.

T2 Lesion Segmentation, Registration, and Tissue Segmentation

T2 hyperintense lesions were manually identified on the T2/proton density/FLAIR baseline images using a semi-automated edge finding tool (JIM v7.0; Xinapse Systems) and T2 lesion volumes were determined. We defined a participant as an MRI outlier when the total T2 volume was outside of 2 SDs of the mean, either above or below. New/enlarging T2 lesions were manually identified using a subtraction of the proton density/T2 images at baseline and 24/96 weeks. New PBHs were similarly manually selected and reported as the number of new T2 lesions at 24 weeks being persistently T1 hypointense at 96 weeks. The original T2 images acquired in 2D with a voxel resolution of 1 × 1 × 3 mm3 were resampled into a 1-mm isotropic space and lesions were coregistered to the 3D T1 images using a pseudo-T1 image generated by subtracting the proton density from the T2-weighted image23; they were then transformed from native space to 3D T1 space using nearest-neighbor interpolation.

For brain extraction, tissue segmentation, and parcellation, geodesical information flows was used on the lesion-filled 3D T1 scans,24 providing the following metrics: normalized brain volume (NBV), white matter (WM), cortical gray matter, and deep gray matter; lesion filling was used in this step using a multi–time point patch-based method to avoid segmentation bias.25 Percent brain volume change (PBVC) from baseline to week 24 and from baseline to week 96, as a measure of brain atrophy, was calculated using the SIENA method.26

SEL Detection

To identify SELs, we developed an in-house version of the pipeline proposed by Elliott et al.,27 using a nonlinear registration analysis of T2-defined lesions on volumetric T1 images in a 2-stage process (SEL detection algorithm). First, candidate SELs were identified from the baseline T2 lesion masks propagated to subsequent scans. A lesion with a positive Jacobian expansion value (JE; i.e., the determinant of the nonlinear deformation field) and size of at least 10 mm3 was classified as a candidate SEL, as per the method of Elliott et al.14 The remainder of the lesions with a negative JE were classified as non-SELs. The second step identified definite SELs, through a further subselection from the candidate SELs. This was based on both the constancy over time and concentricity of their expansion (heuristic score ≥0). Constancy was measured by determining the least-squares linear fit of JE over time, taking into account all the intermediate time points. For concentricity, the voxels within each lesion were subdivided into concentric bands from the central core, then mean JE values in each band were plotted against the distance from the edge, allowing calculation of the slope of the least-squares linear fit. SEL candidates not satisfying the full 2-stage criteria were designated as possible SELs. We refer to SEL-derived volumes to describe the lesion volumes at baseline. Lesion probability maps were obtained separately for definite, possible, and non-SELs after registering all subjects to a common Montreal Neurological Institute anatomical atlas.

Structural Analysis of MTR Within Lesion Types

MTR, as percent units (pu), was computed at baseline and week 96, and the difference between baseline and week 96 was also calculated. For each participant, the average MTR across all T2 lesions was calculated. MTR was also analyzed at the single lesion level in definite, possible, and non-SELs. To account for registration inaccuracies, MTR values greater or less than 2 SDs from the mean were excluded.

Statistical Analysis

Descriptive Statistics

First, we evaluated the distribution and the normality assumptions of all the clinical, demographic, and MRI variables. Differences in EDSS from baseline to the last follow-up (week 96) were assessed using the Wilcoxon signed-rank test. Longitudinal changes in clinical scores were calculated by subtracting baseline from week 96 values. Differences in mean T2 lesion volume between baseline and week 96, and changes in mean PBVC between weeks 24 and 96, were compared with changes in PBVC between baseline and week 96 using paired t tests. Lesion counts were assessed at the patient level by calculating the total number and volume of definite, possible, and non-SELs. The distribution of SEL-derived volumes was positively skewed, so they were log-transformed (using logarithm on base 10 of the value + 1) to normalize the data.

Preliminary Analysis of Associations Between SELs and Demographic and Other MRI Measures

To assess the magnitude and direction of associations, Pearson and partial correlations were assessed between log-transformed SEL volumes, other MRI metrics, and clinical scores. Simple linear regressions, with log-transformed SEL volumes as predictors and demographic and clinical features (age, sex, disease duration, progression duration, EDSS, MSFC, NHPT, T25FW, PASAT, and SDMT) as outcome variables, were performed (eFigure1, links.lww.com/WNL/B869). Through this association analysis, an evaluation of the relationship between the clinical outcomes and the MRI metrics was carried out with direct acyclic graphs in order to identify confounders to include in the final statistical models (eFigure 2).

Association Between MTR and SELs

The volumetric and structural MTR analysis was performed at the single lesion level by applying mixed-effects regression models to take into account within-subject variability and also adjusted for age, sex, and center (to account for scanner differences).

Association Between Clinical Disability Outcomes and SELs

Multiple linear regression models were run to explore whether SEL-derived volumes could independently predict disability outcomes. In order to evaluate the added value of SEL-derived volumes when compared with conventional MRI measures, a backward stepwise selection process was undertaken. All models were adjusted for age and sex, T2 lesion volume change, and PBVC, keeping the variables in the models if statistically significant, always forcing age and sex into the models and using a bootstrap approach, including robust standard errors. The final model included the clinical measure at final follow-up (week 96) as the dependent variable, adjusting for the clinical measure at baseline, and the SEL-derived volumes as independent variable, to assess the ability of SEL-derived volumes to predict longitudinal clinical changes. The stability of the final model was confirmed in a forward selection process. In addition, logistic regression models were built to assess the ability of SEL-derived volumes to predict the development of disability progression. All the model residuals were checked for normality. In order to validate the relationship between SEL and the clinical variable measurements across trial time points, repeated-measures mixed-effects models were performed, where the dependent variable was the value of the clinical variable (one at a time) at each time point, and the explanatory variables included the time point, SEL-derived volumes, and an interaction term between them. Whenever the interaction terms were significant, we assumed there was a significant association between the clinical variable and the SEL-derived volume, for the time point explored. To take into account the multicenter structure, all the mixed-effects models were nested at the center level. Analysis was performed with STATA version 13.1 and all the actual p values were reported.

Data Availability

Fully anonymized data are available after review by the sponsor (University College London). An application form detailing specific requirements, rationale, and proposed use should be completed, followed by a data-sharing agreement. Requested data may be made available along with supporting documentation (e.g., data dictionary) on a secure server to appropriate and approved investigators.

Results

Clinical-Demographic and Conventional MRI Metrics

From the full MS-SMART trial cohort (n = 445), 345 patients fulfilled eligibility criteria. Patients were excluded because of missing scans (n = 93) or because the T2 lesion volume was an MRI outlier (n = 7). Clinical characteristics at baseline and MRI measures at baseline and follow-up are reported in Table 1. Clinical and radiologic characteristics of the patients excluded from the analysis are shown in eTable 2 (links.lww.com/WNL/B869). No differences between treatment arms were observed in terms of counts and volumes of T2 lesions or any of the SEL-derived categories (eTable 3). In the retained study cohort, EDSS significantly increased from baseline to the final follow-up (Wilcoxon signed rank test, p < 0.001) and 36.5% of the patients developed disability progression. Mean T2 lesion volume increased significantly from baseline to week 96 (12.54 mL and 12.78 mL, respectively, paired t test, p < 0.001). Mean NBV was 1,421 mL at baseline and PBVC from baseline to week 96 was greater than PBVC from week 24 to week 96 (−1.35% and −0.92%, respectively, paired t test, p < 0.001).

Table 1

Demographic, Clinical, and Radiologic Characteristics of the Patients Whose Scans Contributed to the SEL Study

Descriptive Analysis of SEL-Derived Metrics

A descriptive analysis of the lesion types at the patient level is shown in Table 2. A total of 340 of the 345 patients (99%) had at least 1 definite SEL. The average number of T2 lesions per patient was 67.2, of which 19.5 (29%) were categorized as definite SELs. Mean T2 lesion volume of definite SELs was 4.4 mL, which accounts for 36% of the overall T2 lesion volume. At the single lesion level, definite SELs were significantly larger than non-SELs (0.25 mL [95% CI 0.18 to 0.31] vs 0.14 mL [0.07 to 0.20], respectively, p = 0.019 from mixed-effects model to account for within-subject variability). The mean annualized change in individual lesion volumes (as determined by the Jacobian values) was 3% (SD 2.9) for definite SELs, 1.5% (SD 3) for possible SELs, and 1.5% (SD 2.2) for non-SELs. Visual inspection of the lesion probability maps revealed no regional differences between definite, possible, and non-SELs, although the latter were more prevalent overall (Figure 1).

Table 2

SEL-Derived Metrics at the Patient Level (n = 345)

Figure 1
Figure 1 Lesion Probability Maps

From left to right, lesion probability maps for definite SEL, possible SEL and non-SELs. Red indicates a lower probability (starting at 3%) and yellow a higher probability (>10%).

Association Between SEL-Derived Volumes and Conventional MRI Metrics

Significant positive correlations were found between greater definite SEL volume and greater T2 lesion volume change (r = 0.24, p < 0.001), the number of new/enlarging T2 lesions (manually obtained) at final follow-up (r = 0.26, p < 0.001), and higher number of new PBHs at final follow-up (r = 0.19, p < 0.001). A positive correlation was found between definite SEL log volume and total baseline T2 lesion volume (r = 0.55, p < 0.001) in partial correlations after accounting for the effect of number of new/enlarging T2 lesions and new PBHs at final follow-up. A higher definite SEL log volume correlated with higher percentage of brain volume reduction over time (r = −0.26, p < 0.001). An example of a patient from this study with a high SEL ratio (relative to total lesion count), worsening of disability, and high PBVC is shown in Figure 2.

Figure 2
Figure 2 Example of High Number of Slowly Expanding Lesions

From left to right: T1 at baseline, T1 at week 96, and registered T1 with Jacobian maps overlaid (red refers to prevalence of positive Jacobian values of expansion; blue is related to volume stability). Out of 27 total T2 lesions identified, 16 were definite slowly expanding lesions (SELs) (59%). Expanded Disability Status Scale at baseline was 5.5 and at week 96 was 8. Percent brain volume change from baseline to week 96 was −2.5%.

MTR Analysis Within SEL-Derived Lesion Types

The mean MTR in definite SELs was significantly lower compared to non-SELs both at baseline and at week 96 (p < 0.001, Table 3). In the longitudinal analysis, the difference between MTR change over time between definite SELs and non-SELs was significant, with a higher rate of MTR reduction from baseline to week 96 found in the definite SELs when compared with the non-SELs (mixed-effects linear regression models [mean adjusted difference 0.52, 95% CI 0.38 to 0.67, p < 0.001]).

Table 3

MTR at Baseline and 96 Weeks Follow-up and MTR Changes Over Time in the Different Lesion Types

Association Between SELs and Demographic and Clinical Features

A higher log volume of definite SELs at baseline correlated with higher increase in EDSS over time (Pearson r = 0.18, p < 0.001). Similarly, when the MSFC and its subtests were analyzed, a higher definite SEL log volume correlated with increasing disability over time, as assessed by changes in the z scores of the MSFC, T25FW, and PASAT (Pearson r ranging from −0.18 to −0.22, p < 0.001). No significant associations were found between definite SEL volumes and any of the demographic features assessed (age, sex, disease duration, and progression duration).

SELs and Clinical Disability Outcomes at Final Follow-up

In stepwise multiple linear regression models, SEL-derived volumes correlated with deterioration of clinical scores at the end of the trial (Table 4). In particular, for each unit increase in definite SEL log volume (mL), there was a 0.23 (95% CI 0.04 to 0.43) increase in EDSS at follow-up (p = 0.020, adjusted R2 = 0.56). Similarly, a unit increase in definite SEL log volume (mL) was associated with a decrease of 0.47 (95% CI −0.98 to −0.03) in MSFC z score units at follow-up (p = 0.048, adjusted R2 = 0.38). In both models, T2 volume change and PBVC were included as covariates. Whereas T2 volume change was not independently associated with change in the clinical measures, PBVC remained significantly associated with worsening in EDSS and MSFC z score. In the logistic regressions, an increase in the definite SEL log volume was associated with an increased risk of developing disability progression (OR 1.92 [1.08, 3.39], p = 0.025, pseudo-R2 = 0.03).

Table 4

Multiple Linear/Logistic Regressions Between SEL-Associated Log Volumes and Clinical Scores

The analysis was extended to all the subcomponents of the MSFC and the SDMT (Table 4): an increase in the definite SEL log volume was associated with a worsening of T25FW z scores (β = −2.10 [−3.43 to −0.85], p = 0.001, adjusted R2 = 0.20) and PASAT z scores (β = −0.27 [−0.50 to −0.10], p = 0.006, R2 = 0.66). Neither non-SEL nor possible SEL log volumes were significantly associated with any of those clinical scores. No significant associations between the SEL-derived measures and changes in the NHPT z score were found. For the SDMT only, increase in non-SEL and possible SEL log volumes were associated with a worsening cognitive score (Table 4).

SELs and Longitudinal Clinical Disability Outcomes

The relationship between SEL volume and longitudinal changes in clinical disability were further confirmed through repeated-measures mixed-effects models across time points (baseline to week 24 and baseline to week 96) adjusted for covariates (age at baseline, sex, total baseline lesion volume, PBVC between baseline and last time point; Table 5). In particular, an association between increased SEL-derived log volumes and greater worsening in the clinical outcome over time was again found for nearly all the explored measures in the interval from baseline to last time point (week 96). For the EDSS case, the association between SEL-derived volumes and clinical changes over time could only be confirmed for the first time interval (between baseline and week 24). Regarding SDMT, there was a decrease of the performance from baseline to final time point associated with an increase in all the SEL-derived volumes. The SEL-derived volumes and the other MRI and clinical measures were highly reproducible and not influenced by the study center, as all the models took into account the multicenter structure (confirmed by intraclass correlation coefficients computation; eTable 4, links.lww.com/WNL/B869).

Table 5

Association Between SEL-Derived Volumes and Clinical Outcomes Over Time Using Mixed-Effects Regression Models

Discussion

SELs are a new in vivo MRI marker of chronic active lesions that have been recently investigated in RRMS and PPMS cohorts. In this study, we assessed SELs in a large group of people with SPMS. We found that SEL-derived metrics were associated with more severe lesional damage (as measured by T1 hypointensity and MTR) and predict physical and cognitive progression in SPMS.

Patients in this cohort had a substantial T2 lesion burden that significantly increased over time, consistent with other clinical trials in SPMS.28 The descriptive analysis showed that the proportion of patients with at least 1 SEL was remarkably high (99%), and greater than that observed in PPMS or RRMS (72% and 68%, respectively).29 The mean number of definite SELs (19.5) in the current SPMS cohort was also higher than in PPMS and RRMS (6.3 and 4.6, respectively).29 From the total T2 lesion volume per patient, the fraction of definite SELs was considerable (36%), indicating that they account for a substantial proportion of lesions. The annualized volume change of definite SELs was on average 3% per year in this SPMS population (previous studies of PPMS and RRMS have not assessed this). Comparison of our SEL observations in SPMS with those in RRMS suggest that chronic inflammatory activity accumulates over the course of the disease, although differences in techniques may influence the absolute numbers derived from different studies (as discussed below).

On a lesion level, definite SELs were significantly larger than non-SELs. This suggests that there is a greater tendency for ongoing lesion expansion in larger lesions, but also may reflect prior lesion enlargement. Previous imaging studies have mainly focused on the spatial location of SELs rather than the morphologic or dimensional features. SELs have been described to be preferentially located in the periventricular areas.29 We also observed this (Figure 1), but noted no difference in lesion distribution among no, possible, and definite SELs.

In a subsample of the MS-SMART study cohort (n = 345), we observed a correlation between definite SEL volume and change in overall T2 lesion volume (r = 0.24, p < 0.001), suggesting that SELs might be a significant contributor to total lesion burden. These findings are in line with pathologic studies, where chronic active lesions are associated with a higher lesion load.4 In addition, we found a significant correlation between higher SEL volume and new PBH (r = 0.18, p < 0.001), in line with previous findings of association of lower and more rapidly decreasing T1 hypointensity within SELs,15,29 in turn reflecting chronic axonal loss in MS.9,10 A recent study involving 52 patients with RRMS30 reported a correlation between SELs, normalized brain volume and PBVC (as markers of neurodegeneration), and disability accrual.31,,34 In our study, we extend this observation to SPMS, finding that a higher definite SEL (but not possible or non-SEL) volume was associated with greater brain atrophy (r = −0.26, p < 0.001). These results support the hypothesis that SELs contribute substantially to the neurodegenerative process in SPMS.

To further investigate lesion damage, we studied MTR, a marker of myelin loss, and axonal density reduction histopathologically in a subsample of 106 patients.10,11 As expected, MTR within SELs was lower than in non-SELs at baseline. In addition, over time, a greater decline in MTR was found in the definite SELs, compared to non-SELs. Similarly, a previous pilot study in RRMS (n = 52)30 found a lower baseline MTR in SELs and an increase in MTR in non-SELs after 24 months follow-up.

We found a clear association between SELs and disability in SPMS. Elliott et al.15 previously found that SELs were able to explain 12-week confirmed disability progression as measured by EDSS and a ≥20% increase in T25FW and NHPT in a PPMS trial cohort (n = 732). In our study, we confirm and extend these findings to SPMS, finding that SEL-derived volumes could significantly explain a proportion of EDSS worsening and the development of disability progression (based on EDSS change from baseline to week 96). Similarly, SEL-derived volumes were associated with worsening of MSFC z score and an increased odds for disability progression. SEL volumes in isolation explained clinical progression in both the MSFC subcomponents assessing walking and cognitive functions (i.e., T25FW and PASAT), although no significant association with hand function (NHPT) was found. Finally, SDMT worsening, another measure assessing cognitive function, was associated with increases in all the SEL-derived volumes in the mixed-effects models but no associations with the definite SEL volumes were found in the multiple linear regressions.

There are some study limitations worth mentioning. SEL analysis can be influenced by many factors, such as image resolution and field strength, the number of time points used, registration and deformation algorithms used, and the definitions (e.g., size, rate of growth) of lesion subtypes. Regarding the SEL definition, we applied a volume threshold of 10 mm3 as per Elliott et al.,27 recognizing that the computation of nonlinear deformations in smaller spatial areas reduces reliability. However, in contrast to Elliott et al., we did not threshold lesions based on rates of expansion. Also in contrast to Elliott et al.,29 we did not try to disentangle confluent lesions as, on careful review of such lesions, we could not distinctly separate merging lesions.

Postcontrast T1-weighted scans were not available in this study, so we could not assess the relationship between SEL and contrast-enhanced lesions. However, in SPMS, the frequency of gadolinium-enhancing lesions is low (10% as reported in SPMS trials),21 and a previous study showed that contrast enhancement is not a common feature of SELs.29

The magnitude associations and effect sizes in our analyses were to some extent small or borderline significant. However, given the nature of this exploratory study, analyzing the effect of a novel MRI marker, any sign of association with the disability measures, even if weak, has been considered valuable.

One hundred patients had to be excluded due to incompatibility with the inclusion criteria (i.e., missed MRI scans), not allowing us to perform the SEL analysis with our pipeline. However, the robustness of the results of the multiple linear regression including the clinical outcome variables has been accounted for by using a multiple imputation model for the missing data (eTable 5, links.lww.com/WNL/B869). Moreover, an added value of the SEL pipeline used in this study is that it is highly reproducible across centers, using common pipelines and conventional MRI sequences (proton density/T2-weighted and T1-weighted).

SELs have been observed over periods of 2–3 years, but it is not clear if SELs remain active perpetually or eventually become quiescent as they have not been investigated over a longer time period. In contrast to SELs, there is also evidence that over decades, some lesions may shrink or disappear.35,36 In addition to the T1 and MTR signature, other microstructural and cellular properties of SELs could be studied, using advanced quantitative MRI or targeted PET techniques, providing greater insights into the pathobiology of SELs. The presence of a lesion rim on susceptibility-weighted MRI has been proposed as an alternative imaging marker for chronic active MS lesions,37,,39 albeit also imperfect,40 but it is unclear how these relate to SELs. Retrospective volumetric analysis has provided evidence that rim-positive lesions have a tendency to expand.3,41 As with SELs, a higher number of rim-positive lesions appears to be associated with clinical severity,42 and persistence of rim-positive lesions is associated with a worse prognosis,41 although the temporal dynamics of rim appearance and persistence are not entirely clear. Furthermore, using quantitative susceptibility mapping, as with SELs, hyperintense rims appear to be more common in progressive MS and in patients with higher levels of disability.43

SELs are a common feature in SPMS and make up a substantial fraction of the T2 lesion volume. They correlate with other markers of neurodegeneration and relate to disability progression. As a marker of ongoing (smoldering) activity in the absence of overt new activity, they may shed light on elusive mechanisms of progression, and represent a target for anti-inflammatory treatments behind a closed blood–brain barrier.44

Study Funding

The authors report no targeted funding.

Disclosure

A. Calvi is supported by an ECTRIMS-MAGNIMS fellowship (2018), Guarantors of Brain “Entry” clinical fellowship (2019), and the UK MS Society PhD studentship (2020). F. Prados Carrasco received a Guarantors of Brain fellowship 2017–2020 and is supported by the National Institute for Health Research (NIHR) Biomedical Research Centre initiative at University College London Hospitals (UCLH). C. Tur has received an ECTRIMS Postdoctoral Research Fellowship in 2015; and honoraria and support for traveling from Merck Serono, Sanofi, Roche, TEVA Pharmaceuticals, Novartis, Biogen, Bayer, and Ismar Healthcare. D. Chard is a consultant for Biogen and Hoffmann-La Roche and is supported by the NIHR Biomedical Research Centre initiative at UCLH, the International Progressive MS Alliance, and the UK MS Society. F. De Angelis, N. John, T. Williams, A. Doshi, J. Stutters, and D. MacManus declare no conflicts of interest with respect to this work. R.S. Samson receives research funding from the MS Society (77), CureDRPLA, and Ataxia UK. C. Gandini Wheeler-Kingshott receives research funding from the MS Society (77), Wings for Life (169111), Horizon2020 (CDS-QUAMRI, 634541), BRC (RC704/CAP/CGW), UCL Global Challenges Research Fund (GCRF), and MRC (MR/S026088/1); and is a shareholder in Queen Square Analytics Ltd. O. Ciccarelli received research funding from the NIHR Biomedical Research Centre initiative at UCLH, UK and National MS Societies, and Rosetrees trust; serves as consultant for Novartis, Roche, and Teva; and is an Associate Editor for Neurology®. J. Chataway has received support from the Efficacy and Evaluation (EME) Programme, a Medical Research Council (MRC) and NIHR Biomedical Research Centre initiative at UCLH partnership and the Health Technology Assessment (HTA) Programme, the UK MS Society, the US National MS Society, and the Rosetrees Trust; has been a local principal investigator for a trial in MS funded by the Canadian MS society and a local principal investigator for commercial trials funded by Actelion, Biogen, Novartis, and Roche; has received an investigator grant from Novartis; and has taken part in advisory boards/consultancy for Azadyne, Biogen, Celgene, Janssen, MedDay, Merck, Novartis, and Roche. F. Barkhof is supported by the NIHR Biomedical Research Centre initiative at UCLH; serves on the editorial boards of Brain, European Radiology, Journal of Neurology, Neurosurgery & Psychiatry, Neurology, Multiple Sclerosis, and Neuroradiology; and serves as consultant for Bayer Schering Pharma, Sanofi-Aventis, Biogen-Idec, TEVA Pharmaceuticals, Genzyme, Merck-Serono, Novartis, Roche, Synthon, Jansen Research, and Lundbeck. The MS-SMART trial, an investigator-led project sponsored by University College London, was funded by the Efficacy and Mechanism Evaluation program as project number 11/30/11. This independent research is awarded by and funded by the MRC, the UK MS Society, and the National MS Society and is managed by the NIHR on behalf of the MRC–National Institute for Health partnership. Additional support came from the University of Edinburgh; the NIHR UCLH Biomedical Research Center and University College London; and the NIHR Leeds Clinical Research Facility (Dental Translational and Clinical Research Unit). Riluzole was provided without charge by Sanofi-Genzyme, which was not involved in the trial design, running of the trial or analysis. Go to Neurology.org/N for full disclosures.

Acknowledgment

The authors thank the MS-SMART study participants and their families and carers; the MS-SMART Investigators; the research staff of the NMR trial unit at Queen Square MS Centre, Department of Neuroinflammation, Institute of Neurology, Faculty of Brain Sciences, University College London: Marios Yiannakas (radiographer), Jonathan Steel (IT manager), Philippa Bartlett (data analyst), Carolina Crespo (data analyst), Virginia Santana (data analyst), Almudena Garcia-Gomez (data analyst), and Teresa Alfaro-Vidal (data analyst); and Sanofi-Genzyme for providing riluzole without charge.

Appendix 1 Authors

Table

Appendix 2 Coinvestigators

Table
  • Received March 7, 2021.
  • Accepted in final form January 18, 2022.



Source link

EHS
Back to top button