Abstract
Objective
Paramagnetic rim lesions (PRLs) are a biomarker of chronic active lesions (CALs), and an important driver of neurological disability in multiple sclerosis (MS). The reason subtending some acute lesions evolvement into CALs is not known. Here we ask whether a relatively lower oxygen content is linked to CALs.
Methods
In this prospective cross‐sectional study, 64 people with multiple sclerosis (PwMS), clinically isolated syndrome and radiologically isolated syndrome underwent a 7.0 Tesla (7 T) brain magnetic resonance imaging (MRI). The scanning protocol included a T2‐w fluid‐attenuated inversion recovery (FLAIR), and a single echo gradient echo from which susceptibility‐weighted imaging (SWI) was derived. WM lesions were identified on the T2‐w‐FLAIR whilst PRLs were identified on the SWI sequence. T2‐lesions were classified as PRLs and rimless lesions (PRLs‐). We registered a universal vascular atlas to each subject's T2‐w‐FLAIR and classified each T2‐lesions according to its location into watershed‐ (ws), non‐watershed‐ (nws), and mixed‐lesion (m). Ws‐lesions were defined as lesions that were fully located in a region between the territories of two major arteries.
Results
Out of 1,975 T2‐lesions, 88 (4.5%) were PRLs. Ws‐regions had a higher number (p = 0.005) and proportion (p < 0.001) of PRLs‐ compared to nws‐regions. Ws‐PRL‐ were larger compared to nws‐ones (p = 0.009). The number (p = 0.043) and proportion (p < 0.001) of PRLs was higher in ws‐regions compared to nws‐ones. Ws‐PRLs were not significantly larger than nws‐ones (p = 0.195).
Interpretation
We propose the novel concept of a link between arterial vascularization and chronic activity in MS by demonstrating a preferential localization of CALs in ws‐territories.
Introduction
Chronic inflammation trapped within an intact blood–brain barrier is a critical pathological element of multiple sclerosis (MS) and an important driver of neurological disability. 1 Although the pathobiological components of chronic inflammation are not fully understood, it is established that chronically activated microglia and macrophages have a central role in perpetuating compartmentalized inflammation. 2
Chronic active lesions (CALs), also known as mixed active‐inactive or smoldering lesions, are an important manifestation of chronic inflammation. 3 CALs are featured by a core of demyelination surrounded by several microglia and astrocyte populations, among which iron‐loaded CD68+ microglia. 2 , 3 , 4 The presence of this intracellular iron makes CALs detectable as paramagnetic rim lesions (PRLs) on susceptibility‐based magnetic resonance imaging (MRI). 3 , 5
It is unclear why some white matter (WM), 5 juxtacortical, 6 and gray matter (GM) 7 lesions evolve into CALs. They develop from fresh lesions, which diffusely contain heme iron and are present in different sites of the brain. 8 , 9 Whether heme iron accumulation is related to specific properties of the vasculature or a specific mechanism of tissue injury is still unknown. Along with heme iron, that released from injured oligodendrocytes is likely a component of PRLs. 10 PRLs may be identified as early as the stage of radiologically isolated syndrome (RIS) and clinically isolated syndrome (CIS). When present, PRLs tend to be associated with worse outcome. 11 , 12
Histopathological evidence supports the notion that arterial vascularization is related to WM lesions, in that watershed (ws) vascular regions of the brain harbor a higher concentration of MS plaques. 13 These findings are attributed to the role of relative hypoxia present in ws‐regions relative to non‐ws‐ (nws) regions. 14 A neuropathological analysis of the brains of people with MS (PwMS) who died in the progressive stages of the disease, showed that lesions location in ws regions is associated with an increase in the average demyelination by two to three times and in the extent of axonal or neuronal damage by five to seven times in the cortex, WM, and the deep GM. 15 As virtual hypoxia, due to inflammation, microglia activation, oxidative injury, and subsequent mitochondrial damage, appears to play a major role in the induction of tissue injury in MS lesions, 14 it may be predicted that areas of low oxygen tension, as it is the case in ws areas of the brain, are more susceptible to chronic tissue injury and inability of tissue repair. 14
In line with histopathology data, recent work in our group has shown that people with newly diagnosed CIS (PwCIS), RIS (PwRIS), and PwMS harbor a higher concentration of WM‐lesions in ws‐regions, relative to other vascular brain territories. We also showed that ws‐T2‐lesions have a higher degree of tissue injury as measured using quantitative magnetic resonance imaging (MRI) metrics compared to non‐ws‐ (nws) T2‐lesions.
Here we leverage and expand on our previous work by studying the topography of CALs in relation to arterial vascularization. We hypothesize that due to a tissue environment less favorable to repair, CALs tend to localize more frequently in ws‐territories. We used ultra‐high field susceptibility‐sensitive MRI to: (i) detect PRLs; (ii) map their topography relative to major arterial territories; and (iii) assess the associations between PRLs and WM lesions located in different vascular territories and people's neuro‐cognitive disability.
Subjects/Materials and Methods
Study design and cohort
We adhered to the Reporting guidelines for observational studies (STROBE). 16 Our prospective, cross‐sectional study was approved by the Vanderbilt University Medical Center Institutional Review Board and was carried out under the Declaration of Helsinki criteria. A signed consent form was obtained from each participant. Sixty‐four newly diagnosed PwCIS, 17 PwRIS, 18 and PwMS 17 were consecutively enrolled. Each participant underwent a 7.0 Tesla (7 T) brain MRI and a clinical assessment using the expanded disability status scale (EDSS), 19 the timed 25‐foot walking test (T25‐FW), 20 the 9‐hole peg test (9‐HPT), and the minimal assessment of cognitive function in MS (MACFIMS) battery. 21 Due to COVID restrictions, only 43/64 PwMS underwent cognitive evaluation. Patients were not included if they had: contraindication to MRI; history of ischemic or hemorrhagic stroke; other systemic or central nervous system autoimmune, neoplastic, or infectious illnesses; uncontrolled hypertension, diabetes, hyperlipidemia; or cardiac diseases. Once enrolled, participants reporting any clinical change between the enrolment date (which coincided with the clinical assessment), and the research procedures were also excluded. All patients were treatment naïve apart from one who had started Interferon beta 1a and desired to be part of the study.
MRI acquisition protocol
MRIs were performed using a 7 T whole body Achieva scanner (Philips Healthcare, Best, The Netherlands) equipped with a volume transmit and 32‐channel receive head coil (NOVA Medical, Wilmington, MA). The MRI protocol included the following sequences: a magnetization prepared two rapid gradient echoes (MP2RAGE), a T2‐w fluid‐attenuated inversion recovery (FLAIR, Fig. 1A), and a single echo gradient echo (SE‐GRE) from which susceptibility‐weighted imaging (SWI, Fig. 1B) was computed. Except for the MP2RAGE which was acquired on the sagittal plane, all scans were acquired axially. Pulse sequence parameters, previously reported, 22 are described in Table 1. All subjects were assessed for contrast‐enhancing lesions (CELs). This was done using a post‐gadolinium diethylenetriamine penta‐acetic acid (Gd‐DTPA) MP2RAGE at 7 T on the first 27 subjects. On the remaining 37 subjects, we assessed for CELs using a clinical scan or a 3 T T1‐w MPRAGE research scan. Irrespective of the acquisition protocol, Gd‐DTPA was administered intravenously (IV) at the dose of 0.1 mmol/kg of body weight.
Figure 1.
Anatomical and vascular maps. T2‐weighted fluid‐attenuated inversion recovery showing four white matter T2‐lesions, pointed by the arrows (A). The susceptibility‐weighted image in (B) shows that of the four lesions, one is a paramagnetic rim lesion (PRL, white arrow). In (C) one can see lesion localization relative to arterial vascularization with the blue arrows pointing toward the rimless‐T2‐lesions (PRLs‐) and the white one pointing toward the PRL. Maps of arterial vascularization across different regions of the brain (D–F). In E, red and pink contours delineate watershed and non‐watershed areas, respectively. A and B indicate the right and left hemispheres; 1 = middle cerebral artery territory 2 = anterior cerebral artery territory and 3 = posterior cerebral artery territory. In this map, the red, pink, and dark red circles represent hypothetical lesions of watershed (ws), non‐watershed (nws), and mixed (m) zones, respectively.
Table 1.
Pulse sequence parameters.
MP2RAGE | T2‐weighted FLAIR | GRE | |
---|---|---|---|
Coverage | Whole brain | Whole brain | Whole brain |
Scan time (min: sec) | 7:33 | 5:07 | 10:44 |
Field of view (mm) | 190 × 230 × 180 | 192 × 180 × 120 | 256 × 256 × 110 |
Echo time (ms) | 2.5 | 342 | 15 |
Inversion time (ms) | 1011/3311 | 1650 | – |
Repetition time (ms) | 6 | 4800 | 50 |
Acquired voxel size (mm) | 0.9 isotropic | 1 isotropic | 0.5 × 0.5 × 2 |
Reconstructed voxel size (mm) | 0.9 isotropic | 0.86 × 0.85 × 1 | 0.4 × 0.4 × 2 |
Orientation | Sagittal | Axial | Axial |
Flip angle (degrees) | 4 | 90 | 16 |
FLAIR, fluid‐attenuated inversion recovery; GRE, gradient echo; mm, milimeters; MP2RAGE, magnetization‐prepared two rapid acquisition gradient echo; ms, miliseconds.
MRI post‐processing
MP2RAGE and SWI computation
We combined T1‐w GRE and proton density images to obtain MP2RAGE data. The T1‐w imaging pipeline outlined by O'Brien et al 23 was used to reduce the background noises of the image.
For SWI maps, MATLAB was used to convert straightened magnitude data to a single or double‐precision floating‐point. Thereafter, we loaded, scaled, and shifted the straightened phase data so it could be wrapped between −pi and +pi. Next, the FSL (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki) brain extraction tool (BET) was used to create a binary mask from the skull‐stripped brain, with the mask being set to DATA.img = ~0. As described by Haacke and collaborators, the data underwent high‐pass filtering using a Hanning filter to improve its quality. 24 Finally, using a phase threshold = 1, and linear negative phase that moved from 1 → 0, an SWI mask was created where phase = 0 → −pi, with threshold = 0. To create the SWI images, the SWI mask was multiplied by the magnitude data.
Flow territories
The flow territories were defined using Digital 3D Brain MRI Arterial Territories Atlas. 25 The atlas was expanded to include ws‐regions of the brain using a previously validated method. 26 The following three arterial territories were considered; the middle cerebral artery (MCA), anterior cerebral artery (ACA), or posterior cerebral artery (PCA). Ws‐regions were defined as any voxel within 10 mm of at least one other primary arterial territory. Overall, four ws‐regions were considered, including the ACA‐MCA ws‐region the ACA‐MCA‐PCA ws‐region, the ACA‐PCA ws‐region, and the MCA‐PCA ws‐region (Fig. 1).
Image registration
T2‐w FLAIR images were rigidly registered to the SWI using FLIRT/FSL and the mutual information cost function whilst the MP2RAGE were registered to the T2‐w FLAIR image co‐registered to the SWI using the Advanced Normalization Tools (ANTs) software package and the Mattes metric as the cost function. 27 The N4BiasFieldCorrection (https://github.com/ANTsX/ANTs/wiki/N4BiasFieldCorrection) by ANTs was used to correct the high level of heterogeneity present at 7 T T2‐w FLAIR images. To co‐register the cerebrovascular flow territory atlas to the subjects' T2‐w‐FLAIR MRI, we followed two steps. First, using the brain extraction tools from the FSL suite, we skull‐stripped the subjects' T2‐w‐FLAIR MRI. Second, the 1 mm3 isotropic T1‐w MRI template in the Montreal Neurological Institute (MNI) 152 space 28 was registered to each subject's skull‐stripped T2‐w‐FLAIR MRI using linear (rigid + affine) registration tools from ANTs. 27 We saved the resulting transformation matrix and applied it to the cerebrovascular flow territory atlas in the MNI 152 space 28 (Fig. 1C–F) to create a flow territory atlases in each subject's native T2‐w‐FLAIR. The quality of individual segmentations and registrations was visually inspected by the first author.
Image analyses
Image analysis was performed using visualization, graphic, and statistical tools available in the Medical Imaging Processing, Analysis, and Visualization (MIPAV) software version 7.3 (https://mipav.cit.nih.gov/). First, CELs were identified and excluded from all subsequent analyses. Thereafter, T2‐lesions were delineated, and individual sizes were computed. PRLs were assessed following the recently published guidelines. 3 PRLs were identified by AAT and HFK and reviewed as a group (FB, HFK, AAT, BH, KC, ZR). The final PRLs map was reviewed and approved by the senior author (FB). Accordingly, T2‐lesions were classified into rimless T2‐lesions (PRLs‐) and PRLs. Periventricular (PV) lesions were defined as lesions with a portion of their border attached to any of the ventricles without normal appearing WM in between. 29
Each PRL‐ and PRL mask was overlaid onto the vascular map and classified into ws‐, nws‐, or mixed‐ (m) lesion. Ws‐lesions extended fully in a region between the territories of two major arteries; nws‐lesions located fully in a region that is supplied by a major artery, and m‐lesions encompassed both territories (Fig. 1).
Statistical analyses
Chi‐squared (χ 2) test was used to assess the proportions of PRLs‐ and PRLs in ws‐regions or PV regions relative to other regions. Within each subject, paired samples t‐tests were used to evaluate differences in the number of PRLs‐ and PRLs between the following pair comparators: ws‐ versus nws‐regions, ws‐ versus m‐regions, and m‐ versus nws‐regions. The results of these analyses are reported as mean difference (MD) and its associated 95% confidence intervals (95% CIs).
Generalized linear mixed models for continuous outcomes were used to assess differences in lesion size between PRLs‐ or PRLs situated in different vascular zones. These models were chosen as they take into consideration possible inter and intrasubject differences. In these models, the MRI metrics (PRL‐ volume/PRL volume) were the variable of interest; the vascular classification (ws‐, nws‐, and m‐regions) was included as a fixed effect factor and the subject as a random effect factor. The results are reported using the beta coefficient (β) and its 95% CIs along with the Akaike Information Criterion (AIC).
Spearman's correlation coefficients were used to assess associations between MRI and clinical variables.
For all statistical analyses, a p‐value <0.050 was considered significant. Statistical analyses were performed using the Statistical Package for Social Sciences (SPSS) Statistics for Windows, Version 24.0. Armonk, NY: IBM Corp.
Results
Table 2 depicts the demographic, clinical, and MRI characteristics of the study cohort. Of the 64 initially enrolled subjects, two were eliminated from the analyses. The first subject did not complete the 7 T MRI protocol and did not have the T2‐FLAIR sequence. The second subject had a several large confluent T2‐lesions and a few tumefactive CELs 30 that would have biased the analyses. A total of 2,091 T2‐lesions were identified in the remaining 62 subjects. Of these, 12 were CELs and 104 had poor visibility on the SWI making the PRL assessment not possible in a reliable manner. As a result, a total of 1,975 T2‐lesions were included in the analyses. Of these, 88 were PRLs and 1,887 were PRL‐.
Table 2.
Demographic and clinical characteristics of the study cohort (n = 62).
Age (years) | 38.64 ± 9.78 (21–66) |
Sex (female) (self identified: number, proportion) | 37, 59.7% |
Ethnicity (self‐identified: number, proportion black / white people) | 3, 4.8%, 59/95.2% |
Body mass index | 28.15 ± 5.14 (19–41) |
Normal/overweight/obese people (n) | 19/22/21 |
Comorbidities (number, % of subjects) | |
Dyslipidemia | 7, 11.3% |
Hypertension | 12, 19.2% |
Diabetes mellitus | 2, 3.1% |
Heart diseases | 1, 1.6% |
Smoking | 7, 11.2% |
Migraine | 6, 9.7% |
Disease phenotype (number of PwRIS/CIS/MS) | 5/7/50 |
Days from post‐Gd DTPA MRI a | 26 [15–52] |
Days from diagnosis a | 30 [14.5–63.5] |
Days from last steroids use | >90 days in 58/62 people |
Months since first symptom a | 24 [6–72] |
Expanded Disability Status Scale score a | 1 [0–2] |
Timed 25‐foot walk test score b (sec) a | 4.73 [4.10–5.97] |
9‐hole peg test left b (t score) | 21.42 ± 4.24 (14.28–32.86) |
9‐hole peg test right b (t score) | 22.12 ± 4.35 (15.32–39.19) |
Elevated IgG index (number, % of subjects) | 28, 45.2% |
Oligoclonal bands (number, % of subjects) | 44, 70.1% |
Number of T2‐lesions | 33.31 ± 28.61 |
Volume of T2‐lesions (millimeters cubic) | 3,721.53 ± 5,464.39 |
Data are expressed in mean ± standard deviation (minimum‐maximum values) unless otherwise stated.
CSF, cerebrospinal fluid; IgG, immunoglobulin G; MRI, magnetic resonance imaging; PwCIS, people with clinically isolated syndrome; PwMS, people with multiple sclerosis; PwRIS, people with radiologically isolated syndrome.
Median (1st quartile‐3rd quartile).
Averaged across two trials.
Rimless T2 ‐lesions analysis
We report details of PRLs‐ distribution in relation to arterial vascular zones in the Supplementary Results. Briefly, the number and proportion of PRLs‐ was higher in ws‐ compared to nws‐regions of the brain (p ≤ 0.004). The volume of PRLs‐ was larger in ws‐ and m‐ compared to nws‐regions of the brain (p ≤ 0.009) (Fig. 2A–C).
Figure 2.
Sizes of lesions located in different vascular territories. Mean lesions size of watershed (red), non‐watershed‐ (pink), and mixed‐lesions (dark red). The box represents the interquartile range, the horizontal lines inside the box represent the median, the vertical lines outside the box represent the 95% confidence interval, and the dots represent the lesions. We present rimless‐T2‐lesions (PRLs‐) (A), non‐periventricular PRLs‐ (B), PRLs‐ in patients without comorbidities (C), paramagnetic rim lesions (D). *Signifies statistically significant results at p‐value <0.050.
PRLs analyses
A total of 88 PRLs were identified in 30 subjects (48.4%), corresponding to 4.5% of the evaluated T2‐lesions.
Incidence and size of PRLs in relation to vascular topography
A higher proportion of PRLs was seen in the ACA‐MCA ws‐territory (p < 0.001). Table 3 details the topographical distribution of PRLs.
Table 3.
Frequency of paramagnetic rim lesions (PRLs) in different regions of the brain.
PRL number (%) | |
---|---|
Vascular classifications | |
Watershed (ws) | 35 (39.8%) |
Non‐watershed (nws) | 20 (22.7%) |
Mixed (m) | 33 (37.5%) |
Vascular territories | |
ACA | 6 (6.8%) |
MCA | 8 (9.1%) |
PCA | 6 (6.8%) |
ACA‐MCA | 59 (67.0%) |
PCA‐MCA | 9 (10.2%) |
ACA‐PCA | 0 (0.0%) |
MCA‐ACA‐PCA | 0 (0.0%) |
Relation to the ventricles | |
Periventricular (PV) | 11 (12.5%) |
Non‐PV | 77 (87.5%) |
Vascular territories and relation to the ventricles | |
m‐PV | 4 (4.5%) |
m‐nPV | 29 (33.0%) |
ws‐PV | 7 (8.0%) |
ws‐nPV | 28 (31.8%) |
nws‐PV | 0 (0.0%) |
nws‐nPV | 20 (22.7%) |
ACA, anterior cerebral artery; MCA, middle cerebral artery; PCA, posterior cerebral artery;
The number of PRLs was higher in ws‐regions compared to nws‐ones (MD = 0.61; 95% CI: 0.02–1.21, p = 0.043) but not m‐ones (MD = 0.08; 95% CI: −0.663 to 0.817, p = 0.834). The proportion of PRLs was higher in ws‐ and m‐regions compared to nws‐regions (p < 0.001). PRLs in m‐regions were larger relative to nws‐PRLs (AIC = 2459.7; m: 279.65; 95% CI: 84.09–475.20; p < 0.001, Fig. 2D). ws‐PRLs tended to be larger than nws‐ones (AIC = 2459.7; ws: β = 125.31; 95% CI: −66.44 to 317.06; p = 0.195) but this difference was not statistically significant.
Differences between PV‐PRLs and nPV‐PRLs
Out of the 88 PRLs, 11 were PV (12.5%) and 77 were nPV (87.5%) (p = 0.009). Of the 11 PV‐PRLs, seven (63.6%) were in ws‐regions and four (36.4%) in m‐regions. This distribution may account for the fact that most PV regions are located in ws‐territories.
Of the 77 nPV‐PRLs, 28 (36.4%) were in ws‐, 20 (25.9%) were in nws‐, 29 (37.7%) were in m‐regions.
Although larger, PV‐PRL size was not statistically different from that of nPV‐PRLs (AIC = 1,110.27; ws: β = 149.58; 95% CI: −60.94 to 360.09; p = 0.195, Fig. 3A). Similarly, ws‐PV‐PRLs (β = 154.03; 95% CI: −127.54 to 435.61; p = 0.279, Fig. 3B), and ws‐nPV‐PRLs (β = 109.92; 95% CI: −90.41 to 310.26; p = 0.278, Fig. 3B) were not larger than nws‐nPV‐PRLs. Differences were instead seen between the volume of m‐PV‐PRLs (β = 560.16; 95% CI: 209.26–911.07; p = 0.002) and m‐nPV‐PRLs (β = 233.78; 95% CI: 34.76–432.80; p = 0.022) compared to nws‐nPV‐PRLs (AIC = 1,066.84, Fig. 3B).
Figure 3.
The volume of paramagnetic rim lesions. We depict the mean size of non‐periventricular (nPV, light green box) and periventricular paramagnetic rim lesions (PV‐PRLs, dark green box) (A) and the mean size of nPV‐non‐watershed (PV‐nws, light blue), nPV‐watershsed (nPV‐ws, sky blue), PV‐ws (cornflower blue), nPV‐mixed (nPV‐m, dark blue), and PV‐m PRLs (midnight blue). The box represents the interquartile range, the horizontal lines inside the box represent the median, the vertical lines outside the box represent the 95% confidence interval, and the dots represent the lesions. *Signifies statistically significant results at p‐value <0.050.
Clinical‐MRI associations
There were no associations between any of the MRI metrics and age, disease duration, EDSS, T25‐FW, and 9HPT. On the contrary, the number of PRLs‐ in ws‐ (r = −0.406, p < 0.05), nws‐ (r = −0.462; p < 0.01), and m‐regions (r = −0.404; p < 0.05) was associated with the CVLT‐scores. Associations were also seen between the following pairs: (i) number of m‐PRLs‐ and BVMT‐R‐Total‐Recall scores (r = −0.369, p < 0.05); (ii) mean volume of m‐PRLs‐ and CVLT scores (r = −0.462, p < 0.01); (iii) ws‐PRLs‐ size and SDMT (r = −0.406, p < 0.05) and DFKES‐contrast measure (r = −0.384, p < 0.05). We did not find significant associations between the number of PRLs in different vascular regions and clinical disability measures.
Discussion
We previously reported an association between arterial vascularization and T2‐lesion frequency and severity in a cohort of newly diagnosed PwMS, PwCIS, and PwRIS imaged at 3 T. 31 In this patient group, that is a sub‐cohort of the one herein studied, we showed that T2‐lesions and chronic black holes occur more frequently in WM ws‐regions of the brain and that lesions in ws‐regions exhibit a higher degree of tissue injury as measured using quantitative magnetization transfer imaging. 31
We here confirm our previous results on a larger group of newly diagnosed PwMS, PwCIS, and PwRIS imaged at ultra‐high field MRI and add the knowledge that CALs as well are related to arterial vascularization in MS. Differently than previously done, we classified T2‐lesions into PRLs‐ and PRLs. We confirm that PRLs‐ preferentially localize in ws‐regions and that ws‐PRLs‐ are larger than nws‐PRL‐. These findings persisted after excluding people with vascular comorbidities and when considering the possible effect of ws‐PV PRL‐. Thus, we confirm the histopathology‐driven hypothesis 15 , 32 that regions with a diminished degree of oxygen levels, for example, ws areas, are associated with an increased likelihood of PRL‐ formation as early as the time of the disease diagnosis. 32
The novel findings associated with the current report are that (1) PRLs as well are more likely to form in ws‐ or m‐regions and that (2) m‐PRLs are larger compared to the ones situated in nws‐regions. We also confirmed that PV regions did not harbor higher concentrations of PRLs.
It is established that PRLs are surrogates of CALs, 3 thus representing areas of focal compartmentalized inflammation, primarily featured by macrophages and microglia activation. The reason subtending the evolution of some acute lesions into CALs is unknown. The persistence of chronic inflammation is however an obstacle to repair. Here we observed that PRLs occur more frequently in ws‐ or m‐territories, indicating a potential role of oxygen levels in fostering the perpetuation of chronic inflammation. This novel finding raises the question if lesion evolution into CAL is in part the result of a failed attempt to quiet down inflammation and promote repair. It is known that hypoxia can initiate inflammation in tissues by causing a vascular leak that starts the inflammation cycle. 33 Hypoxia in inflamed tissues is also not a bystander as it promotes further inflammation and activation of the immune system. 33 , 34 Studies showed the role of the hypoxia‐induced factor alpha in activating nuclear factor kappa B and tumor necrosis factor alpha which amplify the stimulation of the immune system. 34 These hypoxia‐related factors are usually found in lesions of PwMS with more severe disease exacerbations. 14 This environment created in the lesion and its surrounding tissue might resist lesion repair by continuously destroying newly repaired myelin and axons. 35 Thus, hypoxia‐induced inflammation may be one of the factors that foster lesion formation and amplify progressive tissue damage in MS. Other factors certainly synergistically contribute to the PRL/CAL expansion as stroke lesions, known to be triggered by hypoxia, are less dynamic than MS plaques 36 and do not chronically expand, despite persistent inflammation and microglia activation. 37
Previous studies asked whether PV regions are more vulnerable than non‐PV ones to PRL development. The proximity to the cerebrospinal fluid (CSF) has been indeed found to be associated with larger lesion volumes in general as well as to a higher severity of MS lesions and slowly expanding lesions in MS. 38 When looking at CALs / PRLs specifically, discordant findings were seen. Wittayer and collaborators used SWI maps to identify PRLs in PwMS and demonstrated that PRLs form equally across the WM. 39 Miscioscia and collaborators confirmed these findings by showing that both PRLs‐ and PRLs had a similar WM distribution gradient with the highest concentration around the ventricles. 40 Thus, while confirming PV areas' vulnerability, they did not observe a higher incidence of PRLs in these regions. 40 On the contrary, Guo and collaborators demonstrated that PRLs are more likely to form in PV areas while PRLs‐ are more likely to form in juxtacortical and deep WM areas of the brain. 41 The results of our study mirror the ones reported by Miscioscia 40 and Wittayer 39 in a cohort of younger people at the time of disease onset and confirm the notion that the vicinity of the CSF does not necessarily promote an environment that is more favorable to CALs formation.
Correlation with clinical disability
The scope of our paper was that of assessing the relationship between PRLs / PRL‐ with arterial vascularization in MS and to assess if lesion stratification based on arterial vascularization would impact clinical‐MRI associative analyses. Correlations between PRLs and other clinical / radiological metrics of disease are the scope of separate and ongoing work in our group and will be therefore not discussed here.
Germane to this study, only m‐ and ws‐PRL‐ volumes were correlated with disability scores at the SDMT and CVLT, respectively. Additionally, we did not find significant correlations between PRLs stratified by vascular locations and clinical disability. Together, our data testify to the importance of T2‐lesion burden in impacting cognitive impairment early in the disease diagnosis. 42
Limitations and Conclusions
A few limitations need to be acknowledged before concluding. First, our cohort of newly diagnosed people with a narrow range of changes measured on physical and cognitive disability scales limits our MRI‐clinical correlative analyses. However, including newly diagnosed people provides an opportunity to assess the importance of the arterial vascularity of the brain in MS while minimizing the contamination arising from vascular deep white matter injury (leukoarayosis) or the potentially confounding impact of disease‐modifying agents. Also, early in the disease, delineating and classifying lesions is more reliable due to the low prevalence of confluent lesions.
Second, the higher signal‐to‐noise ratio achievable at 7 T may lead to increased conspicuity of WM MS and but also non‐MS lesions. We attempted to mitigate the effect of vascular lesions by repeating the analyses in people without a known history of vascular comorbidities and obtained similar results.
Third, PRLs but also PRLs‐ were more frequently seen in the ws‐ and m‐areas. One can speculate that the higher frequency of PRLs in these areas may represent a statistical phenomenon, in that ws‐areas may be zones with higher vulnerability to larger lesion occurrence in general and as such, PRLs as well. However, statistical considerations do not nullify biological inferences. This is to say, the predominance of PRLs in these areas testify not only for an increased likelihood of lesion formation but also for a more pathological aggressive course of these lesions as indicated by their larger volumes, 31 lower degree of myelin integrity 31 and the presence of a microglia rim as we demonstrate in this report. We note that the presence of paramagnetic rim indicates microglial involvement which represents a different pathological process than that leading to PRLs‐.
The incidence of PRLs in our cohort was slightly lower than those reported in similar PwCIS 43 , 44 , 45 and PwRIS 46 , 47 cohorts. This may be due to technical and methodological factors in that, different from previous studies, we used the most updated recommendations 3 for the definition of PRLs which might have been more stringent. The overall small number of PRLs might have affected the outcome of some of our analyses especially when PRLs were classified based on the vicinity to the ventricles.
Despite these limitations, we believe our study confirms that as early as the time of disease diagnosis, an association is present between arterial vascularization and WM lesions concentration. It also adds to the current knowledge by providing evidence of a higher concentration of PRLs in ws‐ and m‐regions compared to nws‐regions of the brain. This observation raises the question of whether the development of CALs is in part the result of failing reparative mechanisms which tend to be affected by the degree of oxygenation of the brain tissue.
Funding Information
This study was primarily supported by the National MS Society (RG‐1901‐33190: AAT, JW, HFK, BH, TV, CG, ZR, CK, FB). Additional support includes the Veterans Health Administration (I01CX002160‐01A1: AAT, JJE, TV, FB), the Voros Innovation and Impact Fund (FB and JW), and the National Institutes of Health (NIH) (6S10OD012297‐02: VUIIS, Dr. John Gore).
Conflict of Interest
None of the authors has any competing interest to declare.
Author Contributions
FB, MJD and AAT were involved in Conceptualization; AAT, JJE, JW, HFK, BH, TV, CG, ZR, CK, JJE, MJD, and FB were involved in Data curation; AAT, JW, HFK, BH, MAC, RC, ZR, CK were involved in Formal analysis; AAT, MJD, and FB were involved in Investigation, and Methodology, FB was involved in Project administration, Resources, and Softwares; AAT and FB were involved in Writing the original draft; MJD and FB were involved in Supervision and JJE, JW, HFK, BH, TV, CG, ZR, CK, JK, JE, and MJD were involved in Reviewing & Editing the manuscript.
Supporting information
Data S1.
Acknowledgments
We are grateful to our patients and their families who agreed to participate in this study. We thank Dr. Karina Ciccone, Mr. Reece Clarke, and Mr. Keejin Yoon for their invaluable help in setting up the study, recruiting subjects' and preparing IRB documentation. We are grateful for the assistance of the VUIIS Center for Human Imaging. We thank the following collaborators for invaluable earlier help although not involved with the specific work presented in this publication. These collaborators include Dr. Matthew Cronin who set up the SWI acquisition and post‐processing pipelines; Drs. Seth A. Smith, Kristin O'Grady, Colin McKnight, and Baxter Rogers for initial fruitful scientific discussions. We would like to express our gratitude to Dr. Hans Lassmann for his inspiring work leading to this project and for his insights on this manuscript. We are very grateful for the generous donation of the Voros Innovation and Impact Fund. This manuscript is dedicated to the sweet memory of Mr. Oscar Castro.
Funding Statement
This work was funded by the Veterans Health Administration grant I01CX002160‐01A1; National Multiple Sclerosis Society grant RG‐1901‐33190; the Vorus Innovation and Impact Fund; the National Institutes of Health grant 6S10OD012297‐02.
Data Availability Statement
Data are available to share upon reasonable request to the corresponding author.
References
- 1. Ffrench‐Constant C. Pathogenesis of multiple sclerosis. Lancet. 1994;343(8892):271‐275. [DOI] [PubMed] [Google Scholar]
- 2. Prineas JW, Parratt JD. Oligodendrocytes and the early multiple sclerosis lesion. Ann Neurol. 2012;72(1):18‐31. [DOI] [PubMed] [Google Scholar]
- 3. Bagnato F, Sati P, Hemond CC, et al. Imaging chronic active lesions in multiple sclerosis: a consensus statement. Brain. 2024;147:2913‐2933. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Masuda T, Sankowski R, Staszewski O, et al. Spatial and temporal heterogeneity of mouse and human microglia at single‐cell resolution. Nature. 2019;566(7744):388‐392. [DOI] [PubMed] [Google Scholar]
- 5. Bagnato F, Hametner S, Yao B, et al. Tracking iron in multiple sclerosis: a combined imaging and histopathological study at 7 tesla. Brain. 2011;134(Pt 12):3602‐3615. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Galbusera R, Bahn E, Weigel M, et al. Characteristics, prevalence, and clinical relevance of Juxtacortical paramagnetic rims in patients with. Mult Scler. 2024;102(3):e207966. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Pitt D, Boster A, Pei W, et al. Imaging cortical lesions in multiple sclerosis with ultra‐high‐field magnetic resonance imaging. Arch Neurol. 2010;67(7):812‐818. [DOI] [PubMed] [Google Scholar]
- 8. Clark KA, Manning AR, Chen L, et al. Early magnetic resonance imaging features of new paramagnetic rim lesions in multiple sclerosis. Ann Neurol. 2023;94(4):736‐744. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. 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] [PubMed] [Google Scholar]
- 10. Gillen KM, Mubarak M, Nguyen TD, Pitt D. Significance and in vivo detection of iron‐laden microglia in white matter multiple sclerosis lesions. Front Immunol. 2018;9:255. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Luchetti S, Fransen NL, van Eden CG, Ramaglia V, Mason M, Huitinga I. Progressive multiple sclerosis patients show substantial lesion activity that correlates with clinical disease severity and sex: a retrospective autopsy cohort analysis. Acta Neuropathol. 2018;135(4):511‐528. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Fischer BH, Marks M, Reich T. Hyperbaric‐oxygen treatment of multiple sclerosis. A randomized, placebo‐controlled, double‐blind study. N Engl J Med. 1983;308(4):181‐186. [DOI] [PubMed] [Google Scholar]
- 13. Gottlieb SF, Smith JE, Neubauer RA. The etiology of multiple sclerosis: a new and extended vascular‐ischemic model. Med Hypotheses. 1990;33(1):23‐29. [DOI] [PubMed] [Google Scholar]
- 14. Aboul‐Enein F, Lassmann H. Mitochondrial damage and histotoxic hypoxia: a pathway of tissue injury in inflammatory brain disease? Acta Neuropathol. 2005;109(1):49‐55. [DOI] [PubMed] [Google Scholar]
- 15. Haider L, Zrzavy T, Hametner S, et al. The topograpy of demyelination and neurodegeneration in the multiple sclerosis brain. Brain. 2016;139(Pt 3):807‐815. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Cuschieri S. The STROBE guidelines. Saudi J Anaesth. 2019;13(Suppl 1):S31‐S34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Thompson AJ, Banwell BL, Barkhof F, et al. Diagnosis of multiple sclerosis: 2017 revisions of the McDonald criteria. Lancet Neurol. 2018;17(2):162‐173. [DOI] [PubMed] [Google Scholar]
- 18. Okuda DT, Mowry EM, Beheshtian A, et al. Incidental MRI anomalies suggestive of multiple sclerosis: the radiologically isolated syndrome. Neurology. 2009;72(9):800‐805. [DOI] [PubMed] [Google Scholar]
- 19. Kurtzke JF. Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS). Neurology. 1983;33(11):1444‐1452. [DOI] [PubMed] [Google Scholar]
- 20. Cutter GR, Baier ML, Rudick RA, et al. Development of a multiple sclerosis functional composite as a clinical trial outcome measure. Brain. 1999;122(Pt 5):871‐882. [DOI] [PubMed] [Google Scholar]
- 21. Benedict RH, Cookfair D, Gavett R, et al. Validity of the minimal assessment of cognitive function in multiple sclerosis (MACFIMS). J Int Neuropsychol Soc. 2006;12(4):549‐558. [DOI] [PubMed] [Google Scholar]
- 22. Clarke MA, Cheek R, Kazimuddin HF, et al. Paramagnetic rim lesions and the central vein sign: characterizing multiple sclerosis imaging markers. J Neuroimaging. 2024;34(1):86‐94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. O'Brien KR, Kober T, Hagmann P, et al. Robust T1‐weighted structural brain imaging and morphometry at 7T using MP2RAGE. PLoS One. 2014;9(6):e99676. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Haacke EM, Xu Y, Cheng YC, Reichenbach JR. Susceptibility weighted imaging (SWI). Magn Reson Med. 2004;52(3):612‐618. [DOI] [PubMed] [Google Scholar]
- 25. Liu CF, Hsu J, Xu X, et al. Digital 3D brain MRI arterial territories atlas. Scientific Data. 2023;10(1):74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Jones RS, Ford AL, Donahue MJ, et al. Distribution of silent cerebral infarcts in adults with sickle cell disease. Neurology. 2024;102(10):e209247. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Avants BB, Tustison NJ, Song G, Cook PA, Klein A, Gee JC. A reproducible evaluation of ANTs similarity metric performance in brain image registration. NeuroImage. 2011;54(3):2033‐2044. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Fonov VS, Evans AC, McKinstry RC, Almli CR, Collins DL. Unbiased nonlinear average age‐appropriate brain templates from birth to adulthood. NeuroImage. 2009;47:S102. [Google Scholar]
- 29. Casini G, Yurashevich M, Vanga R, et al. Are periventricular lesions specific for multiple sclerosis? Journal of Neurology & Neurophysiology. 2013;4(2):150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Ikonomidou VN, Richert ND, Vortmeyer A, et al. Evolution of tumefactive lesions in multiple sclerosis: a 12‐year study with serial imaging in a single patient. Mult Scler. 2013;19(11):1539‐1543. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Toubasi AA, Xu J, Eisma JJ, et al. Watershed regions are more susceptible to tissue microstructural injury in multiple sclerosis. Brain Commun. 2024:fcae299. doi: 10.1093/braincomms/fcae299 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Brownell B, Hughes JT. The distribution of plaques in the cerebrum in multiple sclerosis. J Neurol Neurosurg Psychiatry. 1962;25(4):315‐320. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Eltzschig HK, Carmeliet P. Hypoxia and inflammation. N Engl J Med. 2011;364(7):656‐665. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Chen L‐W, Egan L, Li Z‐W, Greten FR, Kagnoff MF, Karin M. The two faces of IKK and NF‐κB inhibition: prevention of systemic inflammation but increased local injury following intestinal ischemia‐reperfusion. Nat Med. 2003;9(5):575‐581. [DOI] [PubMed] [Google Scholar]
- 35. Trapp BD, Stys PK. Virtual hypoxia and chronic necrosis of demyelinated axons in multiple sclerosis. Lancet Neurol. 2009;8(3):280‐291. [DOI] [PubMed] [Google Scholar]
- 36. Okuda DT, Moog TM, McCreary M, et al. Utility of shape evolution and displacement in the classification of chronic multiple sclerosis lesions. Sci Rep. 2020;10(1):19560. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Zrzavy T, Machado‐Santos J, Christine S, et al. Dominant role of microglial and macrophage innate immune responses in human ischemic infarcts. Brain Pathol. 2018;28(6):791‐805. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. 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(Pt 5):1239‐1246. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Wittayer M, Weber CE, Platten M, Schirmer L, Gass A, Eisele P. Spatial distribution of multiple sclerosis iron rim lesions and their impact on disability. Mult Scler Relat Disord. 2022;64:103967. [DOI] [PubMed] [Google Scholar]
- 40. Miscioscia A, Treaba CA, Barletta VT, et al. White matter paramagnetic rim and non‐rim lesions share a periventricular gradient in multiple sclerosis: a 7‐T imaging study. Mult Scler. 2024;30:166‐176. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Guo Z, Long L, Qiu W, et al. The distributional characteristics of multiple sclerosis lesions on quantitative susceptibility mapping and their correlation with clinical severity. Front Neurol. 2021;12:647519. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Bagnato F, Salman Z, Kane R, et al. T1 cortical hypointensities and their association with cognitive disability in multiple sclerosis. Mult Scler. 2010;16(10):1203‐1212. [DOI] [PubMed] [Google Scholar]
- 43. 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] [PMC free article] [PubMed] [Google Scholar]
- 44. Altokhis AI, Hibbert AM, Allen CM, et al. Longitudinal clinical study of patients with iron rim lesions in multiple sclerosis. Mult Scler. 2022;28(14):2202‐2211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Meaton I, Altokhis A, Allen CM, et al. Paramagnetic rims are a promising diagnostic imaging biomarker in multiple sclerosis. Mult Scler. 2022;28(14):2212‐2220. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Oh J, Suthiphosuwan S, Sati P, et al. Cognitive impairment, the central vein sign, and paramagnetic rim lesions in RIS. Mult Scler. 2021;27(14):2199‐2208. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. George IC, Rice DR, Chibnik LB, Mateen FJ. Radiologically isolated syndrome: a single‐center, retrospective cohort study. Mult Scler Relat Disord. 2021;55:103183. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1.
Data Availability Statement
Data are available to share upon reasonable request to the corresponding author.