Abstract
Background.
Paramagnetic rim white matter (WM) lesions (PRL) are thought to be a main driver of non-relapsing MS progression. It is unknown whether CSF–soluble factors diffusing from the ventricles contribute to PRL formation.
Objectives.
To investigate the distribution of PRL and non-rim brain WM lesions as a function of distance from ventricular CSF, their relationship with cortical lesions, the contribution of lesion phenotype and localization to neurological disability.
Methods.
Lesion count and volume of PRL, non-rim WM, leukocortical (LCL), and subpial/intracortical lesions were obtained at 7T. The brain WM was divided into 1-mm-thick concentric rings radiating from the ventricles to extract PRL and non-rim WM lesion volume from each ring.
Results.
Sixty-one MS patients with ≥1 PRL were included in the study. Both PRL and non-rim WM lesion volumes were the highest in the periventricular WM and declined with increasing distance from ventricles. A CSF distance-independent association was found between non-rim WM lesions, PRL, and LCL, but not subpial/intracortical lesions. Periventricular non-rim WM lesion volume was the strongest predictor of neurological disability.
Conclusions.
Non-rim and PRL share a gradient of distribution from the ventricles toward the cortex, suggesting that CSF proximity equally impacts the prevalence of both lesion phenotypes.
Keywords: Paramagnetic rim lesion, chronic MS lesion, CSF-gradient, cortical lesion, 7-Tesla phase, smoldering MS
Introduction
Smoldering compartmentalized inflammation represents a major factor contributing to neurodegeneration and disability worsening in MS.1 Paramagnetic rim lesions (PRL) are thought to reflect compartmentalized intrathecal inflammation and are emerging as one of the most relevant imaging markers for predicting non-relapsing MS disease progression.2,3 Paramagnetic rim lesions are characterized by the presence of activated iron-laden microglia at the lesion border, detectable in vivo as a paramagnetic rim on susceptibility-sensitive sequences at 3-T4 and 7-T MRI.1,5
In MS, previous imaging studies have demonstrated a ‘surface-in’ gradient of pathological changes as a function of distance from CSF, possibly triggered by CSF inflammatory soluble factors.6–9 Specifically, innate immune cell-mediated chronic neuroinflammation6 and myelin abnormalities7–9 seem to follow a gradient of distribution, with the highest pathological changes occurring in the immediate periventricular white matter (WM). Additionally, a subset of WM T2-hyperintense lesions have been found to be most destructive and expanding in the closest proximity to the ventricles.10 These slowly expanding lesions (SELs), however, only partially overlap with PRL11 and not all PRL expand over time.12
Multiple sclerosis lesions are typically distributed proximally to ventricles and cortex, and their load correlates with clinical disability.2,13,14 Whether CSF proximity might influence the evolution in either non-rim or PRL, and to what extent, remains unexplored.
In this study, we analyzed the distribution relative to the CSF-ventricle surface of PRL and non-rim WM lesions in a cohort of 61 MS patients presenting PRL at 7-T phase MRI. Specifically, we aimed to: i) investigate differences between PRL and non-rim WM lesion volume in different WM bands surrounding lateral ventricles towards the cortex, in order to indirectly test the spatial contribution of CSF inflammatory soluble factors to PRL development compared to the non-rim; ii) assess the relationship between cortical and WM lesions as a function of distance from CSF; iii) evaluate which localization among non-rim and PRL has the strongest association with clinical disability.
Methods
Standard Protocol Approvals, Registrations, and Patient Consents
The study was approved by the Mass General Brigham Institutional Review Board (#2007P001274) and written informed consent in accordance with the Declaration of Helsinki was obtained from all participants before study enrolment.
Study design and participants
Between 2009 and 2019, 111 MS patients (74 with relapsing-remitting MS [RRMS], 37 with secondary progressive MS [SPMS]) were consecutively enrolled in this study. Inclusion and exclusion criteria are reported in Figure 1. Eleven participants’ data were discarded due to motion artifacts. Since the main purpose of our study was the assess the contribution of the CSF proximity to non-rim and PRL distribution, sixty-one/100 patients (61%) having at least 1 PRL at baseline were included in the final cohort.
Figure 1.
Study flowchart
Study flow diagram showing the inclusion and exclusion criteria.
Clinical Assessment
All patients underwent a neurological examination at study entry and were scored with the Expanded Disability Status Scale (EDSS).15 The EDSS evaluation was also performed at follow-up (mean SD years 3.2 ± 2.4) in 47/61 patients.
MRI Protocol
All study participants underwent two imaging sessions within a week on a 7-T and a 3-T MRI human scanners (Siemens, Erlangen, Germany) using 32 channel coils to acquire: (i) 7-T two-dimensional fast low-angle shot T2*-weighted spoiled gradient-echo images to cover the supratentorial brain (repetition time/echo time [TR/TE] 1/4 1700/21.8 msec, 0.33 × 0.3 × 1 mm3 resolution) yielding magnitude and phase images, ii) 7-T 3D dual magnetization-prepared rapid gradient echo 7-T (MP2RAGE) (TR/TE 5000/2.93 msec, 0.75 mm isotropic resolution), and (ii) 3-T 3D magnetization-prepared rapid acquisition with multiple gradient-echo sequence (MP-RAGE) (TR/inversion time [TI]/TE 1/4 2530/1200 msec, 0.9 × 0.9 3 × 0.9 mm3 resolution).
MRI Data Analysis
Lesion segmentation.
Cortical and WM lesions were segmented on magnitude images from 7-T T2* scans, with the aid of MP2RAGE images, by consensus of two experienced raters, using a semi-automated tool in 3D Slicer version 4.2.0 (http://www.slicer.org). Consistently with our previous studies,2,16 cortical lesions extending for at least 3 voxels across two consecutive slices were classified as subpial/intracortical if subpial (type III/IV) or within the cortex (type II), and leukocortical (LCL) if they also involved the WM (type I).17,18 Type II and type III/IV cortical lesions were grouped together since, based on our previous and current investigations at 7-T19 as well as neuropathological examinations,17 type II lesions are exceedingly rare. Paramagnetic rim lesions were assessed by the study radiologist (CAT) and defined as lesions having a T2-hyperintense core, surrounded by a paramagnetic rim on phase images, along at least 2/3 of the lesion perimeter, and either lacking gadolinium enhancement or present in another MRI scan occurring at least 3 months apart. According to this criterion, WM lesions were classified as either non-rim or PRL. Lesion count and volumes were quantified using fslstats (FMRIB Software Library, FSL, v. 5.0, Oxford, UK, http://www.fmrib.ox.ac.uk/fsl). Prior to assessing lesion volumes, lesion masks from 7-T T2* were registered onto the 3-T anatomical Freesurfer reconstructions using a boundary-based registration method as previously detailed.20
Generation of periventricular concentric rings
Pial and WM surfaces reconstruction were performed using FreeSurfer (v5.3.0, 2013, http://surfer.nmr.mgh.harvard.edu) on the 3-T 3D MP-RAGE image. The topological defects in cortical surface reconstruction caused by the WM and leukocortical lesions were accurately corrected using the in-painting method in all patients. For each subject, WM and lateral ventricles masks were extracted from FreeSurfer segmentation. Manual corrections were performed (by A.M.) when necessary to ensure anatomic accuracy. Distance maps from lateral ventricles CSF toward the cortex in the WM were calculated using FSL software package (http://www.fmrib.ox.ac.uk/fsl). According to the 3-T MRI resolution, 42 1 mm-rings were identified, covering the whole area from the lateral ventricles surface to the cortex (Figure 2a). Similar distance maps were created on PRL and non-rim WM lesion masks, obtaining the lesion load volume distributed into 42 periventricular rings. To minimize partial volume effect, the first 1 mm-ring close to ventricles and the last 1 mm-ring close to the WM/cortical interface were excluded from the analysis. The remaining 40 rings were divided into four 10 mm-bands radiating from the ventricles toward the cortex: i) periventricular; ii) paraventricular; iii) 21–30 mm from ventricle; iv) 31–40 mm from ventricle band (Figure 2b). Volumes of PRL and non-rim WM lesions were extracted from each band for every subject (Figure 2c–d).
Figure 2.
Periventricular WM ring-band mapping and rim-based lesion classification
(a) Periventricular distance map composed of 40 1 mm-rings, covering the whole white matter (WM) area from the surface of the lateral ventricles to the cortex. (b) Clustering of the rings into four 10 mm-bands, radiating from the ventricles toward the cortex: i) periventricular (yellow); ii) paraventricular (green); iii) 21–30 mm from ventricle (blue); iv) 31–40 mm from ventricle band (white). T2*magnitude (c) images show 3 paramagnetic rim lesions (PRL) (red arrows) which are mapped with 1-mm distance rings from the lateral ventricles (d). Other 6 WM lesions are mapped, but do not show a paramagnetic rim in phase image, therefore are classified as non-rim WM lesions. Leukocortical lesions (LCL, blue arrow) and subpial/intracortical lesions (3 green arrows) were excluded from the mapping.
Statistical analysis
Statistical analyses were performed using SPSS 22.0 (StataCorp LP, College Station, TX, USA). Normality in measurements was tested graphically and using Kolmogorov-Smirnov test. Nonparametric tests were used for non-normal or skewed data and parametric tests for normally distributed data. Differences between groups (i.e. RRMS versus SPMS) were analyzed using the chi-squared test for categorical variables, the independent-samples t test for parametric continuous variables, and the Mann–Whitney test for nonparametric continuous variables. To evaluate the relationship between PRL or non-rim WM lesion volume and the distance from the ventricular CSF, we used a linear mixed-effects model, including the lesion volume in each ring as the dependent variable, and ring distance from CSF as independent variable. In this model, participants were considered as random effect to account for multiple rings in the same subjects. Comparisons between lesion volumes in consecutive bands (e.g. PRL periventricular vs PRL paraventricular) or correspondent bands (e.g. PRL periventricular vs non-rim WM lesion periventricular) were tested through Wilcoxon signed rank test. We used Spearman correlations to assess whether PRL and non-rim WM lesions as function of distance from CSF are related to any type of cortical lesion. To investigate the association between lesion type and EDSS, we used i) Spearman correlations between EDSS and either lesion count or volume, ii) a stepwise regression model to evaluate which localization and type of WM lesion is the strongest predictor for the EDSS at follow-up. It included EDSS as dependent variable and the following predictors: periventricular, paraventricular, 21–30 mm from ventricle, 31–40 mm from ventricle PRL and non-rim WM lesion volumes. A p-value of 0.05 was considered statistically significant. Benjamini-Hochberg correction was applied for multiple correlation analysis.
Data Availability
The data sharing depends on Massachusetts General Hospital and Institutional Review Board policy and on the purpose of sharing the data (profit vs. nonprofit).
Results
Study population
The demographic, clinical, and radiological characteristics of our cohort are summarized in Table 1.
Table 1.
Demographics, clinical and radiological characteristics
| MS | RRMS | SPMS | RR vs SPMS p-value | |
|---|---|---|---|---|
| Subjects | 61 | 46 | 15 | |
| Age, years, mean (SD) | 41.9 (9.0) | 40.9 (8.2) | 45.0 (9.5) | 0.102a |
| Female, n (%) | 45 (73.8%) | 35 (76%) | 10 (67%) | 0.343b |
| Baseline EDSS, median (IQR) | 2.5 (2.0) | 2.0 (0.5) | 6.0 (2.0) | <0.001 c |
| EDSS at FU*, median (IQR) | 2.5 (4.0) | 2.0 (2.0) | 6.5 (1.0) | <0.001 c |
| Disease duration, years, mean (SD) | 8.5 (7.7) | 5.9 (6.4) | 15.9 (6.4) | 0.001 a |
| Individual trajectory (clinically stable/clinically worsening), n | 33/13 | 28/6 | 6/7 | 0.007 b |
| DMT (off/LMET/HET), n | 9/37/15 | 7/27/12 | 2/10/3 | 0.854 b |
| LCL volume (mm3), mean (SD) | 788 (1737) | 345 (607) | 2048 (3021) | 0.002 c |
| LCL count, mean (SD) | 11.8 (25.8) | 4.8 (7.2) | 33.3 (45.2) | 0.003 c |
| Subpial/intracortical lesion volume (mm3), mean (SD) | 524 (705) | 402 (618) | 896 (841) | 0.022 c |
| Subpial/intracortical lesion count, mean (SD) | 9.8 (10.4) | 7.6 (8.3) | 16.8 (13.3) | 0.010 c |
Significance testing:
2-tailed t test on means
Chi-squared test
Mann-Whitney test
Bold indicates a statistically significant difference with a p-value<0.05
Follow-up of 47/61 patients (34 RRMS, 13 SPMS), years, mean (± SD) = 3.2 (± 2.4)
Abbreviations: EDSS: Expanded Disability Status Scale; DMT: Disease-modifying treatment; FU: follow-up; HET: high efficacy treatment; IQR: inter-quartile range; LCL: leukocortical lesions; LMET: low/moderate efficacy treatment; MS: multiple sclerosis; RRMS: relapsing-remitting MS; SD: standard deviation; SPMS: secondary progressive MS
Paramagnetic rim and non-rim lesions show a similar periventricular gradient of distribution
Overall, 3276 non-rim WM lesions and 238 PRL (93.2% and 6.8% respectively of the total WM lesion load) were identified on 7T T2* magnitude and phase images. The volume of PRL was highest in close proximity to the ventricles (intercept 78.8 ± 8.6 mm3), and gradually decreased of −2.5 ± 0.2 mm3 for each 1 mm of distance from the ventricles (t=−14.7, p=<0.001). Non-rim WM lesions showed an analogous pattern of spatial distribution (intercept 235 ± 16.7 mm3, slope −6.5 ± 0.3 mm3, t=−25.9, p=<0.001). These results remained unchanged adjusting for disease duration (PRL: intercept 71 ± 12.2 mm3, slope −2.5 ± 0.2 mm3, t=−14.7, p=<0.001; non-rim WM lesions: intercept 151 ± 16.3 mm3, slope −6.8 ± 0.3 mm3, t=−27.2, p=<0.001). Moreover, intra-subject comparison between consecutive bands revealed a significant reduction of volume, in both PRL and non-rim WM lesions as a function of distance from ventricles (Figure 3). Specifically, lesion volume was distributed largely in the periventricular band (67% PRL, 56% non-rim WM lesions), and became gradually smaller in the paraventricular (22% PRL, 24% non-rim WM lesions), 21–30 mm from ventricle (9% PRL, 15% non-rim WM lesions), and 31–40 mm from ventricle band (2% PRL, 5% non-rim WM lesions) (Table 2). After dividing lesion volume in each band by the lesion volume in all bands (separately for non-rim and PRL), PRL fraction was found to be significantly higher in the periventricular band (p=0.021), and lower in the 21–30 mm from ventricle and 31–40 mm from ventricle bands (p<0.001 for both comparisons) compared to non-rim WM lesion fraction. Although both types of lesions were more abundant in SPMS than RRMS, the difference between the two phenotypes reached significance only for non-rim WM lesion count and volume, namely in the periventricular, paraventricular, and 21–30 mm from ventricle bands. Relative to the total WM lesion load, the periventricular non-rim WM lesion fraction resulted, however, significantly higher in SPMS than in RRMS (33.7% RRMS vs 56.1% SPMS, p=0.016).
Figure 3.
PRL and non-rim WM lesion volume distribution
(a) Bar graph showing the mean volume of PRL and non-rim WM lesions in each band. Comparison between consecutive bands is tested through Wilcoxon signed rank test. *indicates p-value=<0.001. (b) Line graph of the lesion distribution (expressed in mean volume, mm3) in RRMS (blue line) and PMS (red line) for every periventricular ring from ventricles to the cortex.
Abbreviations: PMS: progressive MS; PRL: paramagnetic rim lesions; RRMS: relapsing-remitting MS; WM: white matter
Table 2.
PRL and non-rim WM lesion spatial distribution and comparisons
| Lesion type/total WM lesion | Lesion type in a band/Lesion type in all the bands | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| MS | RRMS | SPMS | RR vs SPMS p-value | MS | RRMS | SPMS | RR vs SPMS p-value | Band/tot ratio | Band/tot ratio PRL vs non-rim, p-value | ||
| PRL | Count, mean (SD) | 3.9 (6.5) | 2.6 (2.9) | 7.7 (11.6) | 0.066 | 0.087 (0.079) | 0.092 (0.081) | 0.072 (0.073) | 0.280 | - | - |
| Tot volume (mm3), mean (SD) | 1092 (2474) | 772 (1593) | 2074 (4089) | 0.157 | 0.220 (0.216) | 0.237 (0.223) | 0.165 (0.187) | 0.221 | - | - | |
| Periventricular volume (mm3), mean (SD) | 803 (1969) | 535 (1002) | 1624 (3526) | 0.180 | 0.158 (0.192) | 0.170 (0.200) | 0.121 (0.167) | 0.541 | 0.67 (0.34) | 0.021 | |
| Paraventricular volume (mm3), mean (SD) | 207 (554) | 164 (543) | 336 (585) | 0.219 | 0.038 (0.055) | 0.040 (0.057) | 0.033 (0.048) | 0.912 | 0.22 (0.28) | 0.213 | |
| 21–30 mm from ventricle volume (mm3), mean (SD) | 66 (201) | 59 (195) | 91 (226) | 0.166 | 0.018 (0.048) | 0.021 (0.055) | 0.010) (0.018) | 0.422 | 0.09 (0.23) | <0.001 | |
| 31–40 mm from ventricle volume (mm3), mean (SD) | 14 (69) | 12 (64) | 21 (84) | 0.969 | 0.006 (0.030) | 0.007 (0.035) | 0.001 (0.005) | 0.969 | 0.02 (0.10) | <0.001 | |
|
Non-rim WM
lesion |
Count, mean (SD) | 53.7 (55.4) | 39.5 (31.9) | 97.33 (84.8) | 0.003 | 0.912 (0.079) | 0.908 (0.081) | 0.928 (0.073) | 0.280 | - | - |
| Tot volume (mm3), mean (SD) | 3362 (4208) | 2403 (3308) | 6302 (5332) | <0.001 | 0.780 (0.216) | 0.763 (0.223) | 0.834 (0.187) | 0.221 | - | - | |
| Periventicular volume (mm3), mean (SD) | 2196 (2980) | 1502 (2420) | 4324 (3578) | <0.001 | 0.437 (0.212) | 0.337 (0.202) | 0.561 (0.197) | 0.016 | 0.56 (0.21) | - | |
| Paraventricular volume (mm3), mean (SD) | 691 (930) | 511 (701) | 1246 (1301) | 0.013 | 0.187 (0.121) | 0.192 (0.126) | 0.170 (0.106) | 0.558 | 0.24 (0.13) | - | |
| 21–30 mm from ventricle volume (mm3), mean (SD) | 378 (423) | 292 (315) | 643 (589) | 0.034 | 0.121 (0.094) | 0.130 (0.102) | 0.092 (0.057) | 0.375 | 0.15 (0.11) | - | |
| 31–40 mm from ventricle volume (mm3), mean (SD) | 95 (132) | 98 (144) | 88 (84) | 0.663 | 0.036 (0.045) | 0.044 (0.049) | 0.012 (0.010) | 0.052 | 0.05 (0.06) | - | |
Significance testing:
RR vs SPMS: Mann-Whitney test
Band/tot ratio PRL vs non-rim WM lesions: Wilcoxon signed rank test between correspondent bands Bold indicates a statistically significant difference with a p-value<0.05
Abbreviations: MS: multiple sclerosis; PRL: paramagnetic rim lesions; RRMS: relapsing-remitting MS; SD: standard deviation; SPMS: secondary progressive MS; WM: white matter
Association between white matter lesion localization and cortical lesions
Paramagnetic rim lesions and non-rim WM lesion volumes correlated to each other (r=0.420, p=0.001). We also found an association between LCL (count and volume) and WM lesions, regardless of their paramagnetic properties or their distance from ventricles (except for 31–40 mm from ventricle WM lesion volume which did not reach the significance, Table 3). Intracortical lesion load was associated only with periventricular non-rim WM lesion volume as well as their total count, but not with PRL.
Table 3.
Relationship between WM and cortical lesions
| LCL volume | Subpial/intracortical lesion volume | ||||||
|---|---|---|---|---|---|---|---|
| r | p | r | p | ||||
| PRL volume | Total volume | 0.394 | 0.002 | 0.024 | 0.857 | ||
| Periventricular | 0.396 | 0.002 | −0.030 | 0.819 | |||
| Paraventricular | 0.333 | 0.009 | −0.003 | 0.984 | |||
| 21–30 mm from ventricle | 0.277 | 0.031 | 0.077 | 0.553 | |||
| 31–40 mm from ventricle | 0.254 | 0.048 | 0.121 | 0.352 | |||
| Non-rim WM lesion volume | Total volume | 0.601 | <0.001 | 0.341 | 0.007 | ||
| Periventricular | 0.616 | <0.001 | 0.323 | 0.011 | |||
| Paraventricular | 0.463 | <0.001 | 0.246 | 0.056 | |||
| 21–30 mm from ventricle | 0.441 | <0.001 | 0.242 | 0.060 | |||
| 31–40 mm from ventricle | 0.219 | 0.090 | 0.064 | 0.621 | |||
| LCL count | Subpial/intracortical lesion count | ||||||
| r | p | r | p | ||||
| PRL count | 0.302 | 0.018 | 0.156 | 0.229 | |||
| Non-rim WM lesion count | 0.621 | <0.001 | 0.311 | 0.015 | |||
Spearman correlations
Abbreviations: LCL: leukocortical lesions; PRL: paramagnetic rim lesions; WM: white matter
Periventricular non-rim WM lesion volume has the strongest EDSS predictive value
Total volumes and counts of all lesion types (PRL, non-rim WM lesions, LCL, and subpial/intracortical lesions) showed a positive correlation with both EDSS at baseline and follow-up, except for PRL volume that was not associated with baseline EDSS (Table 4). The stepwise regression model (R2=0.26) revealed periventricular non-rim WM lesion volume (β=4.33e-4; 95% confidence interval [CI]: 2.08e-4, 6.58e-4, p<0.001) as the strongest predictor for EDSS at follow-up.
Table 4.
Relationship between EDSS and total lesion load for different lesion types
| EDSS at baseline | EDSS at follow-up | |||
|---|---|---|---|---|
| r | p | r | p | |
| PRL count | 0.270 | 0.035 | 0.394 | 0.007 |
| PRL volume | 0.118 | 0.366 | 0.309 | 0.037 |
| Non-rim WM lesion count | 0.451 | <0.001 | 0.436 | 0.002 |
| Non-rim WM lesion volume | 0.451 | <0.001 | 0.472 | <0.001 |
| LCL count | 0.308 | 0.016 | 0.460 | 0.001 |
| LCL volume | 0.254 | 0.049 | 0.502 | <0.001 |
| Subpial/intracortical lesion count | 0.304 | 0.017 | 0.405 | 0.005 |
| Subpial/intracortical lesion volume | 0.335 | 0.008 | 0.428 | 0.003 |
Spearman correlations
Abbreviations: EDSS: Expanded Disability Status Scale; LCL: leukocortical lesion; PRL: paramagnetic rim lesion; WM: white matter
Discussion
Our study assessed, in a heterogeneous MS cohort, the spatial distribution of PRL and non-rim WM lesions obtained from 7-T scans as a function of distance from the ventricles, demonstrating a shared periventricular gradient of the two WM lesion phenotypes.
In line with increasing evidence that chronic, intrathecal compartmentalized inflammation represents an important mediator of CSF–related tissue damage,6,21 our findings support the view that the proximity to CSF equally impacts the prevalence of both PRL and non-rim WM lesions. In fact, besides the higher proportion of non-rim than rim WM lesions, PRL and non-rim lesion volume was similarly distributed in the WM, being markedly higher closer to the ventricles and gradually lower with increasing distance from the ventricles.
Although evidence from translocator protein (TSPO) and myelin ligand-PET studies6,7 provides support for the hypothesis that inflammatory/cytotoxic mediators released from the CSF would induce a gradient of innate immune cell activation, demyelination, and remyelination failure in WM surrounding the ventricles, the factors that determine WM lesion evolution as a non-rim rather than PRL are still obscure.
Paramagnetic rim lesions can be seen at any stage of MS as well as in radiologically isolated syndrome, clinically isolated syndrome, and pediatric MS,22–24 and tend to remain stable or enlarge over time compared to WM rimless lesions, which otherwise remain stable or shrink.12 Longitudinal studies reported that paramagnetic rims arise at the time of lesion formation and present a dynamic course over time. Phase rims have been shown to disappear by one year in 45% of PRL showing a centripetal pattern of gadolinium enhancement at the time of the lesion birth, and the persistence of a lesion phase rim is a negative prognostic factor for the long-term lesion evolution.5 Moreover, quantitative susceptibility values of the phase rim have been found to increase in the first two years of lesion development and to gradually decrease after the fourth year.25 Some authors have reported a decrease in iron-rim MRI signal intensity over time, especially in progressive MS.12 This might suggest that most of PRL have a life span of only a few years before iron loss and probable transition to a chronic inactive state or glia scar. Accordingly, since we found a similar distribution of non-rim and PRL in the WM, we might speculate that the two types of lesions represent a continuum, and the iron-laden rim characterizes a transitory life stage of some non-rim WM lesions. We also found that the fraction of periventricular PRL volume (67%) was significantly higher, compared to the fraction of non-rim lesions in the same band (56%). This finding suggests that, even if the closeness to the CSF-ventricle surface is thought to promote the formation of both lesion types, this aspect is particularly important for PRL that might be present in higher percentage in the periventricular area due to higher concentrations of CSF soluble factors possibly inducing persisting microglia-mediated inflammation.26,27 Finally, we did not find a gradient from the cortex to the ventricles, meaning that the subarachnoid CSF might have a role for the subpial cortical lesion formation, thought to be driven by CSF-mediated soluble factors diffusing from areas of meningeal inflammation,21,28,29 but not for either rim or non-rim WM lesions.
As expected, our data showed that both lesion types were more abundant in SPMS than RRMS, especially in periventricular WM, becoming similar towards the subcortical WM. Although the significance was not reached for PRL, this result might mirror the pattern of lesion accumulation that occurs throughout the course of the disease, following a specific ventricle-to-cortex trend. Relative to the total WM lesion load, SPMS patients had a higher non-rim lesion fraction in the periventricular band compared to RRMS. This might be due to either a higher rate of non-rim lesion accumulation than PRL throughout the disease course, or to the conversion over time of pre-existing PRL into non-rim WM lesions.
Consistently with previous studies,30 we found that WM (both rim and non-rim) lesions were associated with LCL, but not with subpial/intracortical lesions, regardless of their distance from ventricles. Previous pathological studies observed differences between purely cortical (type II and III/IV) and mixed WM/cortical lesions (type I), the latter resembling characteristics of WM lesions.31,32 In fact, similarly to WM lesions, LCL are associated with T lymphocyte infiltrations as well as microglia. Conversely, subpial/intracortical lesions show a significant low density of T cells, being formed mostly by microglia.17,31 Furthermore, WM lesions and LCL share a distribution around small cerebral veins,32,33 while the localization of the extensive/subpial lesions cannot be explained by the distribution of any single cortical vein. Taking these findings altogether, it is conceivable that MS WM lesions (PRL, non-rim, and LCL) might share common pathological mechanisms (possibly microglia/macrophages and T cells-mediated, with various density of microglia at the lesion rim depending on the grade of the chronic lesion activation),34 differently from subpial/intracortical lesions (mostly microglia-mediated). However, since subpial/intracortical lesions and periventricular non-rim WM lesion loads were associated, the proximity to CSF and its cytotoxic factors diffusing from the periventricular and meningeal compartments might have a role in both pathological manifestations. Longitudinal and mechanistic studies will be important to explore this hypothesis. In this regard, emerging combined pathological-MRI studies are highlighting the role of organized meningeal inflammatory infiltrates on cortical lesion formation both in early and chronic progressive MS patients.32,35
In our previous machine-learning study2 in a larger MS cohort that included a subset of subjects included in the current study, the normalized subarachnoid CSF volume, PRL and LCL volumes were identified as the main predictors of disability progression. Here, we aimed to investigate which WM localization among PRL and non-rim WM lesions most affected neurological disability in patients with PRL at baseline. Interestingly, periventricular non-rim WM lesion load was the most impactful predictor on EDSS at follow-up (over a mean of 3.2 years). This finding might be related to the significantly higher volume of non-rim over PRL (73% vs 27%) in the periventricular band, and this supports the importance of a pathological periventricular gradient as a main driver of disability,6,9 in which non-rim and PRL might be in a continuum.
Our work has some limitations. Neurological disability as measured by EDSS was the only clinical outcome measure in our study. Although EDSS includes assessment of cognitive and visual status, the overall score is heavily weighted towards motor function, mainly reflecting physical disability. Future studies will explore the contribution of lesion distribution on cognition.
In conclusion, our study demonstrates that non-rim and PRL have a similar distribution in the WM of MS patients, which follows a gradient from the ventricles toward the cortex. This finding indicates that the proximity to the ventricular CSF equally impacts the prevalence of rim and non-rim lesions.
Acknowledgment
We would like to thank Mary Foley for the technical assistance with MRI scanning, all the subjects who participated, and the hospital staff for their help in making this study possible.
Study Funding
This work was supported by grants from the National Multiple Sclerosis Society (NMSS 4281-RG-A1, NMSS RG 4729A2/1 and NMSS RG 1802-30468), National Institutes of Health R01NS078322-01-A1, and United States Army W81XWH-13-1-0122.
Footnotes
Disclosure
Dr. Miscioscia has nothing to disclose; Dr. Treaba has received research support from Genentech/Roche; Dr. Barletta has nothing to disclose; Dr. Herranz has received research support by the NMSS fellowship FG-1507-05459; Dr. Sloane has consulted for Novartis, Biogen, Cellgene, Genentech and received research support from Genentech/Roche; Dr. Barbuti has nothing to disclose; Dr. Mainero has received research support from Genentech/Roche.
References
- 1.Dal-Bianco A, Grabner G, Kronnerwetter C, et al. Slow expansion of multiple sclerosis iron rim lesions: pathology and 7 T magnetic resonance imaging. Acta Neuropathol 2017;133(1):25–42. doi: 10.1007/s00401-016-1636-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Treaba CA, Conti A, Klawiter EC, et al. Cortical and phase rim lesions on 7 T MRI as markers of multiple sclerosis disease progression. Brain Communications 2021;3(3):fcab134. doi: 10.1093/braincomms/fcab134 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Ransohoff RM. Multiple sclerosis: role of meningeal lymphoid aggregates in progression independent of relapse activity. Trends in Immunology 2023;44(4):266–275. doi: 10.1016/j.it.2023.02.002 [DOI] [PubMed] [Google Scholar]
- 4.Maggi P, Sati P, Nair G, et al. Paramagnetic Rim Lesions are Specific to Multiple Sclerosis: An International Multicenter 3T MRI Study. Ann Neurol 2020;88(5):1034–1042. doi: 10.1002/ana.25877 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Absinta M, Sati P, Schindler M, et al. Persistent 7-tesla phase rim predicts poor outcome in new multiple sclerosis patient lesions. Journal of Clinical Investigation 2016;126(7):2597–2609. doi: 10.1172/JCI86198 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Poirion E, Tonietto M, Lejeune FX, et al. Structural and Clinical Correlates of a Periventricular Gradient of Neuroinflammation in Multiple Sclerosis. Neurology 2021;96(14):e1865–e1875. doi: 10.1212/WNL.0000000000011700 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Tonietto M, Poirion E, Lazzarotto A, et al. Periventricular remyelination failure in multiple sclerosis: a substrate for neurodegeneration. Brain 2023;146(1):182–194. doi: 10.1093/brain/awac334 [DOI] [PubMed] [Google Scholar]
- 8.Liu Z, Pardini M, Yaldizli Ö, et al. Magnetization transfer ratio measures in normal-appearing white matter show periventricular gradient abnormalities in multiple sclerosis. Brain 2015;138(5):1239–1246. doi: 10.1093/brain/awv065 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Brown JWL, Pardini M, Brownlee WJ, et al. An abnormal periventricular magnetization transfer ratio gradient occurs early in multiple sclerosis. Brain 2017;140(2):387–398. doi: 10.1093/brain/aww296 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Klistorner S, Barnett MH, Graham SL, Wang C, Klistorner A. The expansion and severity of chronic MS lesions follows a periventricular gradient. Mult Scler 2022;28(10):1504–1514. doi: 10.1177/13524585221080667 [DOI] [PubMed] [Google Scholar]
- 11.Calvi A, Clarke MA, Prados F, et al. Relationship between paramagnetic rim lesions and slowly expanding lesions in multiple sclerosis. Mult Scler 2023;29(3):352–362. doi: 10.1177/13524585221141964 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Dal-Bianco A, Grabner G, Kronnerwetter C, et al. Long-term evolution of multiple sclerosis iron rim lesions in 7 T MRI. Brain 2021;144(3):833–847. doi: 10.1093/brain/awaa436 [DOI] [PubMed] [Google Scholar]
- 13.Filippi M, Preziosa P, Banwell BL, et al. Assessment of lesions on magnetic resonance imaging in multiple sclerosis: practical guidelines. Brain 2019;142(7):1858–1875. doi: 10.1093/brain/awz144 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Vellinga MM, Geurts JJG, Rostrup E, et al. Clinical correlations of brain lesion distribution in multiple sclerosis. J Magn Reson Imaging 2009;29(4):768–773. doi: 10.1002/jmri.21679 [DOI] [PubMed] [Google Scholar]
- 15.Kurtzke JF. Rating neurologic impairment in multiple sclerosis: An expanded disability status scale (EDSS). Neurology 1983;33(11):1444–1444. doi: 10.1212/WNL.33.11.1444 [DOI] [PubMed] [Google Scholar]
- 16.Treaba CA, Granberg TE, Sormani MP, et al. Longitudinal Characterization of Cortical Lesion Development and Evolution in Multiple Sclerosis with 7.0-T MRI. Radiology 2019;291(3):740–749. doi: 10.1148/radiol.2019181719 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Peterson JW, Bö L, Mörk S, Chang A, Trapp BD. Transected neurites, apoptotic neurons, and reduced inflammation in cortical multiple sclerosis lesions: Cortical Lesions in MS. Ann Neurol 2001;50(3):389–400. doi: 10.1002/ana.1123 [DOI] [PubMed] [Google Scholar]
- 18.Bø L, Vedeler CA, Nyland HI, Trapp BD, Mørk SJ. Subpial Demyelination in the Cerebral Cortex of Multiple Sclerosis Patients. J Neuropathol Exp Neurol 2003;62(7):723–732. doi: 10.1093/jnen/62.7.723 [DOI] [PubMed] [Google Scholar]
- 19.Mainero C, Benner T, Radding A, et al. In vivo imaging of cortical pathology in multiple sclerosis using ultra-high field MRI. Neurology 2009;73(12):941–948. doi: 10.1212/WNL.0b013e3181b64bf7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Greve DN, Fischl B. Accurate and robust brain image alignment using boundary-based registration. NeuroImage 2009;48(1):63–72. doi: 10.1016/j.neuroimage.2009.06.060 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Magliozzi R, Howell OW, Reeves C, et al. A Gradient of neuronal loss and meningeal inflammation in multiple sclerosis. Ann Neurol 2010;68(4):477–493. doi: 10.1002/ana.22230 [DOI] [PubMed] [Google Scholar]
- 22.Suthiphosuwan S, Sati P, Absinta M, et al. Paramagnetic Rim Sign in Radiologically Isolated Syndrome. JAMA Neurol 2020;77(5):653. doi: 10.1001/jamaneurol.2020.0124 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Clarke MA, Pareto D, Pessini-Ferreira L, et al. Value of 3T Susceptibility-Weighted Imaging in the Diagnosis of Multiple Sclerosis. AJNR Am J Neuroradiol 2020;41(6):1001–1008. doi: 10.3174/ajnr.A6547 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Margoni M, Preziosa P, Storelli L, et al. Paramagnetic rim and core sign lesions in paediatric multiple sclerosis patients. J Neurol Neurosurg Psychiatry Published online March 29, 2023:jnnp-2022-331027. doi: 10.1136/jnnp-2022-331027 [DOI] [PubMed] [Google Scholar]
- 25.Zhang S, Nguyen TD, Hurtado Rúa SM, et al. Quantitative Susceptibility Mapping of Time-Dependent Susceptibility Changes in Multiple Sclerosis Lesions. AJNR Am J Neuroradiol 2019;40(6):987–993. doi: 10.3174/ajnr.A6071 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Aram J, Francis A, Tanasescu R, Constantinescu CS. Granulocyte-Macrophage Colony-Stimulating Factor as a Therapeutic Target in Multiple Sclerosis. Neurol Ther 2019;8(1):45–57. doi: 10.1007/s40120-018-0120-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Carrieri PB, Provitera V, Rosa TD, Tartaglia G, Gorga F, Perrella O. Profile of Cerebrospinal Fluid and Serum Cytokines in Patients with Relapsing-Remitting Multiple Sclerosis. A Correlation with Clinical Activity. Immunopharmacology and Immunotoxicology 1998;20(3):373–382. doi: 10.3109/08923979809034820 [DOI] [PubMed] [Google Scholar]
- 28.Howell OW, Reeves CA, Nicholas R, et al. Meningeal inflammation is widespread and linked to cortical pathology in multiple sclerosis. Brain 2011;134(9):2755–2771. doi: 10.1093/brain/awr182 [DOI] [PubMed] [Google Scholar]
- 29.Magliozzi R, Howell O, Vora A, et al. Meningeal B-cell follicles in secondary progressive multiple sclerosis associate with early onset of disease and severe cortical pathology. Brain 2006;130(4):1089–1104. doi: 10.1093/brain/awm038 [DOI] [PubMed] [Google Scholar]
- 30.Beck ES, Maranzano J, Luciano NJ, et al. Cortical lesion hotspots and association of subpial lesions with disability in multiple sclerosis. Mult Scler 2022;28(9):1351–1363. doi: 10.1177/13524585211069167 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Bø L, Vedeler CA, Nyland H, Trapp BD, Mørk SJ. Intracortical multiple sclerosis lesions are not associated with increased lymphocyte infiltration. Mult Scler 2003;9(4):323–331. doi: 10.1191/1352458503ms917oa [DOI] [PubMed] [Google Scholar]
- 32.Gh Popescu BF, Lucchinetti CF. Meningeal and cortical grey matter pathology in multiple sclerosis. BMC Neurol 2012;12(1):11. doi: 10.1186/1471-2377-12-11 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Kidd D, Barkhof F, McConnell R, Algra PR, Allen IV, Revesz T. Cortical lesions in multiple sclerosis. Brain 1999;122(1):17–26. doi: 10.1093/brain/122.1.17 [DOI] [PubMed] [Google Scholar]
- 34.Kamma E, Lasisi W, Libner C, Ng HS, Plemel JR. Central nervous system macrophages in progressive multiple sclerosis: relationship to neurodegeneration and therapeutics. J Neuroinflammation 2022;19(1):45. doi: 10.1186/s12974-022-02408-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Bhargava P, Hartung HP, Calabresi PA. Contribution of B cells to cortical damage in multiple sclerosis. Brain 2022;145(10):3363–3373. doi: 10.1093/brain/awac233 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data sharing depends on Massachusetts General Hospital and Institutional Review Board policy and on the purpose of sharing the data (profit vs. nonprofit).