Introduction
Multiple sclerosis (MS) is a chronic disease of the central nervous system (CNS) associated with focal inflammatory demyelinating lesions in the white and grey matter [
14]. Some lesions remyelinate early after the demyelinating event [
16] and evolve into remyelinated shadow plaques, which protect against axonal degeneration [
23]. Other lesions remain chronically demyelinated. Chronic demyelination fosters persistent low-degree neurodegeneration in the form of axonal transections [
24]. A subset of lesions with inactive demyelinated centers maintains continuous myelin breakdown at the edge, which has led to the pathological concept of the slowly expanding lesion [
36]. Pathologically, the edge of slowly expanding lesions is featured by a rim of activated microglia/macrophages harboring occasional myelin degradation products [
15], few T cells [
14,
36], and a considerable amount of axonal transections [
15].
MS typically starts with relapsing-remitting course which progresses into secondary progression in 70% of patients. About 10% of patients begin the disease with a primary progressive course [
32]. The common clinical feature of progressive MS is a continuous neurological decline in the absence of new and contrast-enhancing lesions and clinical relapses. There is currently no approved treatment to reduce disability accrual in progressive MS, which is partly due to our incomplete understanding of the pathobiological mechanisms underlying progression [
32]. In a pathological survey on 2,476 WM plaques, slowly expanding lesions were predominantly found in progressive MS [
15] and were suggested to indicate progressive disease activity [
15,
36]. Focal T-cell-mediated CNS inflammation [
4] causing relapses seems to differ from the typical microglia-mediated wide-spread inflammation that features progressive MS [
26]. Mitochondrial DNA deletions, oxidative stress [
20], and iron accumulation and its liberation during demyelination [
21] are thought to be key factors of neurodegeneration in progressive MS [
28].
Iron accumulation has been described within microglia/macrophages at the edges of slowly expanding [
5,
21] and some inactive lesions, but was not observed around shadow plaques [
21]. Questions remain whether iron accumulation surrounds other lesion types, whether it differs between slowly expanding and inactive lesions, and whether it is, indeed, absent from edges of shadow plaques. Based on proper pathological characterization, edge-related iron accumulation might, therefore, separate expanding non-remyelinating lesions [
3] from those with increased remyelination probability, and ultimately become a useful imaging biomarker for disease activity in stages of MS, where contrast enhancement of lesions is rare or absent.
Magnetic resonance imaging (MRI) and post-mortem studies showed that edge-related iron accumulation is captured by a rim-shaped signal around chronic WM lesions when using phase [
3,
7,
22,
30,
39], susceptibility-weighted imaging (SWI) [
18], multi-echo gradient echo
R
2* [
40], or quantitative susceptibility mapping (QSM) [
10,
12]. These rims were seen in patients with either relapsing-remitting or secondary progressive MS [
30,
40], but were absent from supratentorial neuromyelitis optica (NMO) lesions, which render iron rims potentially helpful for distinction between MS and NMO lesions [
10]. Longitudinal analyses of persistent rim lesions in MS thus far showed lack of expansion over 2.5 years [
7] or slight shrinkage within the first 12 months after gadolinium enhancement resolution [
3], challenging the notion that iron rims surround slowly expanding lesions in vivo.
To gain knowledge on the pathological and in vivo features of rim lesions, we examined the association between rim-shaped iron accumulation at the edges and the pathological stages of lesions in a sample of 28 post-mortem MS cases. We then examined whether lesions encircled by a rim-shaped signal in SWI are more likely to expand over a period of 3.5 years than those without rims in a prospective longitudinal study in seven patients with MS using a fluid attenuated inversion recovery/SWI fusion sequence (FLAIR–SWI) at 7 Tesla (7 T) [
2,
11,
17].
Discussion
We report four key observations in our present study: (1) An iron rim at the edge of an MS lesion is predominantly seen in slowly expanding lesions, much less frequently in inactive lesions, not in active and hardly in remyelinated lesions. (2) The iron containing cells in the rim are in their vast majority microglia/macrophages with a pro-inflammatory activation status, while iron-positive astrocytes are sparse or absent. (3) Direct 7 T MRI—pathology correlation shows that the iron rim, defined by pathology, can be reliably visualized by magnetic resonance imaging. (4) Lesions with an iron rim on average expand very slowly, while non-rim lesions show a tendency to shrink. Although expansion of rim lesions has already been observed in individual MS lesions [
2,
3], we have statistically proven this expansion for the first time in multiple MS patients and lesions. However, it is important to note that some non-rim lesions expanded and some rim lesions shrinked. Our data, therefore, indicate that an observed rim lesion does not necessarily expand over time, but has a higher probability to do so, when compared with non-rim lesions. Using post-mortem pathology-imaging correlations, we confirm this rim to be due to the presence of iron inside pro-inflammatory activated microglia/macrophages [
3,
5,
22]. In line with pathological [
5,
21,
30,
34] and in vivo [
3,
39,
40] studies, this pattern of iron accumulation was restricted to edges of chronically demyelinated lesions, not present in active lesions [
1], and hardly observed in fully remyelinated shadow plaques [
21].
Pathologically, we observed the most pronounced iron accumulation at the rims of slowly expanding lesions. The empirically derived threshold for iron-laden microglia/macrophages sufficient to decrease the signal in post-mortem SWI revealed a significant association of slowly expanding lesions with a rim in SWI, when compared with inactive lesions. However, translating this threshold to the in vivo situation needs to be exerted with caution, given the differences in imaging resolution and tissue properties between in vivo and post-mortem imaging. One explanation for the absence of iron rims in classical active lesions could be the dynamics of lesion formation. In active lesions, myelin fragments are mainly taken up by cells with a macrophage phenotype (round, no processes). These are mobile, disperse within the lesion, and finally accumulate in perivascular spaces. Conversely, in slowly expanding lesions, tissue debris is taken up by cells with microglia phenotype (branched cells with processes). Microglia may remain stationary at lesion edges for prolonged time periods, which may form the basis for the persistent iron rim. Lack of iron accumulation in the vast majority of astrocytes found in iron rims could be related to the elevated expression of the iron exporter ferroxidase ceruloplasmin in astrocytes at edges of MS lesions [
21], indicating active iron efflux by astrocytes.
Microglia and macrophages in iron rims highly expressed the pro-inflammatory markers CD86 and p22phox, while anti-inflammatory CD206 (mannose receptor) expression was rare and, in line with prior results [
38], mainly expressed by perivascular macrophages. In the mentioned study [
38], CD206 in active MS lesions was also expressed by 70% of myelin-laden foamy macrophages which expressed M1 markers. Thus, the authors proposed an intermediate activation status of the majority of macrophages in active MS lesions. This situation is different for edges of slowly expanding lesions, as reported here, where the majority of microglia/macrophages show a clear pro-inflammatory activation status without CD206 expression. A similar concept has been proposed by Pitt and collaborators [
30], who also showed that CD206 was mainly expressed by lipid-laden macrophages in MS lesions, while iron-laden microglia/macrophages at edges of lesions did not express CD206. We have additionally shown the expression of pro- and anti-inflammatory markers in double-labelings with iron in a sample of ten well-characterized slowly expanding lesions. Expression of pro-inflammatory markers in microglia/macrophages was independent from iron accumulation in these cells in our study. Therefore, the iron rim indicates microglia/macrophages with a pro-inflammatory activation status, but iron itself does not seem to induce pro-inflammatory activation, as determined by the markers we have included (CD86, p22phox). Our interpretation is that pro-inflammatory microglia/macrophages at edges of slowly expanding lesions either accumulate iron or not, but iron rims specifically indicate microglia/macrophages with a pro-inflammatory activation status.
Lack of iron rims around the vast majority of shadow plaques is particularly informative in comparison with inactive lesions. This feature might indicate that remyelination is restricted to lesions which have never undergone a stage of edge iron accumulation, because otherwise traces of long-standing edge iron in a subset of remyelinated shadow plaques would be expected. However, extended follow-up studies of rim lesions will be able to investigate whether they do not remyelinate.
Our data are in line with another study showing lack of remyelination in five lesions of a single progressive MS autopsy case displaying phase rims and pathologically confirmed iron accumulation [
3]. Based on in vivo data, the authors conclude from a progressively lower
T1 intensity between 3 and 12 months of observation seen in 7/10 lesions with persistent phase rim versus 7/26 lesions without phase rim that the phase rim lesions showed failure of early tissue repair and possibly remyelination failure, as opposed to non-rim lesions. This is in line with our observation in 74 shadow plaques of 15 MS autopsy cases, showing the absence of remyelination in iron rim lesions and the absence of iron rims in remyelinated lesions. While
T2-weighted or FLAIR images are unable to separate remyelinated from demyelinated lesions [
6], SWI could, therefore, help to distinguish them.
We do not think that remyelination might explain the observed shrinking of hyperintense non-rim lesions in our survey, since remyelinated lesions were as
T2-hyperintense as completely demyelinated lesions in a post-mortem correlation study [
6]. Conversely, we ascribe the -10% baseline volume of non-rim lesions to lesion-specific gliosis and neurodegeneration, which leads to tissue retraction and shrinkage [
28].
The frequency of rims around WM lesions in our in vivo data (15.3% of all observed WM lesions) is similar to that of another in vivo study, where authors reported 10.1% of all MS WM lesions to display a rim indicative of iron accumulation [
10]. These authors furthermore noted that the majority (83.3%) of MS lesions lacking MRI signs suggestive of iron accumulation were ill-defined and showed faint margins. In comparison, 84.5% of our non-rim lesions were ill-defined and confluent and, therefore, not included in volumetric analysis, which is crucially dependent on a reliable depiction of the lesion margin.
Rim lesions were significantly larger than non-rim lesions at each timepoint, which is in line with other data [
3]. First, larger size of rim lesions could be the result of their expansion together with shrinkage of non-rim lesions over time. Second, lesions which are larger at baseline could be more likely to end up as slowly expanding. The latter possibility is suggested by data from Absinta et al. [
3]. In this study, newly forming and enhancing lesions, which later developed a persistent phase rim, were already larger at baseline (timepoint of gadolinium enhancement) than lesions which did not develop a persistent phase rim.