Keywords

FormalPara Key Points
  1. 1.

    Multifocal demyelinated lesions occur in both the white and gray matter of the central nervous system (CNS), are present early in the disease course, and accumulate over time.

  2. 2.

    Active lesions are associated with demyelination, axonal injury, gliosis, oligodendrocyte destruction, microglial/macrophage activation, and infiltration of immune cells including macrophages and T- and B-cells.

  3. 3.

    Axonal transection has historically been described as “relative sparing,” although in some chronic lesions up to 80% of axons are lost.

  4. 4.

    Secondary pathogenic mechanisms including oxidative energy and mitochondrial deficits may drive neurodegeneration.

  5. 5.

    Although magnetic resonance imaging (MRI) is a useful tool in the diagnosis of multiple sclerosis (MS), limitations exist in discriminating the underlying pathology of lesions, e.g., only approximately 55% of T2-only lesions are demyelinated, subpial and intracortical lesions are poorly visualized on MRI, and individuals with myelocortical MS have cerebral white matter lesions on MRI without significant cerebral white matter demyelination (but do have cortical and spinal cord demyelination).

Introduction

Pathological characterization of multiple sclerosis (MS) postmortem and biopsy tissue has led to an improved understanding of inflammatory lesions and neurodegeneration and shed light on potential therapeutic targets. Since Charcot’s recognition of MS as a distinct disease in the nineteenth century [1, 2], several seminal concepts have emerged. Demyelination is a key pathological feature of MS associated with inflammation that is widespread, involving the white matter (WM), cortical and deep gray matter (GM), and spinal cord [3]. Focal inflammatory lesions are associated with neuronal injury and axon/neurite transection [4, 5]. Neurodegeneration, either as a secondary consequence to focal inflammatory lesions, mediated by oxidative injury and energy failure [4], or perhaps independent of demyelination, contributes to clinical disability more significantly than WM lesions. Magnetic resonance imaging (MRI), while an invaluable tool, may not reflect pathology as accurately as we presume; contrary to popular belief, only 55% of T2-hyperintense lesions in patients with MS are demyelinated [6]. Here, we detail our current understanding of MS pathology, correlations with MRI where applicable, its clinical relevance, and gaps in our understanding.

Lesions

A hallmark of MS is the presence of focal lesions characterized by demyelination and neuronal/axonal injury in the brain and spinal cord [7]. Myelin sheaths, produced by oligodendrocytes enwrapping axons to promote rapid saltatory conduction and trophic support to axons, can be extensively lost in demyelinated lesions. A higher proportion of myelin is composed of lipids (70–85%weight/weight), with the remainder being proteins—mainly myelin basic protein (MBP), proteolipid protein (PLP), and myelin oligodendrocyte glycoprotein (MOG). Myelin loss is readily detected by immunohistochemistry using monoclonal antibodies to these proteins. Chemical staining techniques such as Luxol fast blue are limited in that they are poorly able to visualize GM demyelination and are not specific to demyelination as they can also reflect lipid perturbations. Focal demyelinating lesions have clearly demarcated margins [8] and occur either in perivenular or subependymal/subpial areas. Areas with a greater predilection for lesions include the hippocampus, lateral ventricles, and superficial cortical layers [8].

A second prominent feature of MS lesions is neurodegeneration associated with neuronal injury (dystrophic neurons and reduced neuronal size), axonal pathology (transection and swelling), and synaptic/dendritic changes. “Relative sparing” of axons can be misleading as axonal loss in chronically demyelinated lesions can reach up to 80% [9]. The extent of neuronal loss associated with demyelination has been mixed [10,11,12], but neuronal pathology in lesions includes neurite transection, shrinking of the neuronal soma, and synaptic stripping [5, 12, 13].

Axonal injury, which has been recognized since the early 1900s, is associated with clinical disability worsening and is noted in both demyelinated and remyelinating lesions with high variability between individuals. Smaller caliber axons are more vulnerable [14], and pathology includes transection and proximal spheroids [9, 15]. Direct mechanisms of injury may include proteases and reactive oxygen/nitrogen species (ROS/RNS) from activated macrophages/microglia (mØ) in close contact and granules released by cytotoxic T-cells. Secondary means of injury could be by impaired axoplasmic membrane permeability and intra-axonal energy failure due to oxidative tissue injury and accumulation of respiratory-deficient neurons [16,17,18,19,20]. Axonal loss in eloquent areas such as the pyramidal tract can cause significant clinical disability [21]. Together these processes comprise mechanisms of tissue injury that contribute to physical disability.

White Matter Lesions

White matter lesions (WMLs) gather more attention as they are readily identifiable as hyperpigmented plaques on gross pathology and associated with demyelination, inflammation, and axonal loss with immunohistochemistry [5, 15]. They are commonly located in corpus callosum, centrum semiovale, periventricularly, juxtacortically/cortically, and in the spinal cord [7, 8]. The extent of inflammatory cell infiltration with mØ and T-cells, demyelination, and axonal loss varies with the stage of the lesion noted below [22].

Histologically, earlier WMLs are more inflammatory than later stage lesions. The majority of CD3+ cells (76%) are major histocompatibility complex class I (MHC-I) restricted CD8+ T-cells and express markers associated with tissue-resident memory cells [23]. In active lesions, up to a third of CD8+ T-cells are granzyme B positive, suggesting cytotoxic activity against a yet unidentified self-antigen [23, 24]. B-cells (CD20+), in contrast, are predominantly in perivascular spaces with a lower number infiltrating the lesion parenchyma [23]. These observations are common in either relapsing or progressive disease courses and even in acute fulminant lesions. MHC-I is expressed on inflammatory cells and neuroglia [25], and MHC-II is observed in activated mØ [26, 27]. The inflammatory milieu of the active lesion changes when chronic active, characterized by a hypocellular lesion center with a rim of activated mØ [23].

Axonal pathology in the WML includes disruption of axonal transport with spheroids and transection [9]. Proposed mechanisms of acute transection include oxidative injury, cytokines, and proteolytic enzymes [4, 28]. Chronically demyelinated axons likely degenerate due to intra-axonal energy failure, altered sodium-potassium (Na+/K+) ATPase function, and accumulation of intra-axonal calcium levels [6, 16, 29].

While MRI has been paramount in the diagnosis of MS and monitoring for disease activity, there are clear limitations in detecting and staging lesions. Lesions suggestive of MS on MRI are defined by location (periventricular, juxtacortical/cortical, infratentorial, and spinal cord) and the presence of gadolinium enhancement (GdE) associated with a breakdown in the blood-brain barrier seen at the onset of lesion formation. Both the presence of GdE lesions and new/enlarging T2 lesions are useful in establishing the diagnosis of MS and surveillance of disease activity. GdE aids in the identification of new lesions on conventional T1 post-contrast images despite being relatively insensitive to blood-brain barrier disruption [30]. It is important to recognize that the presence of an MRI lesion does not necessarily equate to pathologically identifiable demyelination. In one analysis, 45% of T2-only (T2-hyperintense) and 17% of T2T1MTR (T2-hyperintense, T1-hypointense, and reduced magnetization transfer ratio) were not demyelinated with postmortem validation [29]. T2T1MTR lesions are more likely to be chronic inactive (68%) and have reduced axonal density [29]. T2-only changes do not always reflect demyelination, as they can be due to mØ activation and non-demyelinating axonal pathology [31]. T2 lesions that are not apparent on gross pathology can include preactive and active demyelinating lesions, while grossly visible lesions are often chronic inactive [32]. Remyelinated lesions cannot be discriminated by T2, but using T1 and MTR may improve sensitivity [33]. As lesions track venous vasculature, a “central vein sign” (imaged using T2*-echo planar imaging [34, 35]) can be useful to discriminate MS lesions from mimics such as microvascular disease, migraine, vasculitis, and aquaporin 4-positive neuromyelitis optica spectrum disorder (NMOSD) [36,37,38].

White Matter Lesion Classification/Staging

Histologically, focal lesions can be classified by the presence and distribution of inflammation (mØ) and the temporal pattern of demyelination. Staging lesions with in situ MRI-histopathology postmortem correlations provides insight into the efficacy of therapeutics—as one would expect minimal active lesions in individuals on highly effective therapies. More importantly, validating conventional and advanced imaging modalities with pathology lends to better clinical translation with in vivo imaging of myelin and neurodegeneration [39]. The lexicon for delineating lesion types has evolved over time to be more descriptive and is summarized in Table 8.1 [9, 22, 27, 40,41,42]. The use of “early” and “late” activity can be noted by mØ-containing myelin degradation products: cyclic nucleotide phosphodiesterase (CNP), myelin oligodendrocyte glycoprotein (MOG), or myelin-associated protein (MAG) in early and MBP and PLP in late [22]. Features of more recent lesion characterization include a description of the extent of inflammation, typically with mØ, and the presence of ongoing demyelination as evidenced by myelin degradation product inclusions in mØ.

Table 8.1 Lesion types characterized by immunohistochemistry for the presence of inflammation and myelin protein inclusions
Full size table

The simplest classification (Bö/Trapp) characterizes lesions as active, chronic active, and inactive based on cellularity, predominantly by mØ [27]. The Lucchinetti/Brϋck/Lassmann classification, derived from biopsy and autopsy samples, includes myelin degradation products to discriminate early and late active lesions [26, 42]. Recently, Kuhlmann proposed including activity without demyelination (myelin degradation products) and “mixed active/inactive” ± demyelination; smoldering/slowly expanding lesions may be included in the latter category [22]. The De Groot/van der Valk modification introduces the concept of a preactive/early active lesion type with mØ clusters in normal-appearing regions without demyelination [39]. The Vienna consensus identifies six types based on inflammation as well as demyelination by the presence of myelin degradation products but is uncommonly utilized. For most purposes, the Bö/Trapp and Lucchinetti/Brϋck/Lassmann methods are sufficient to interpret pathological findings. See Fig. 8.1.

Fig. 8.1
figure 1

Classification of white matter demyelinating lesions using 1994 Bö/Trapp staging (see Table 8.1). Areas of perivascular demyelination were identified with proteolipid protein (PLP) staining (a, c, e) in 30 μm free floating sections of MS thalamic tissue. Staining for activated microglia/macrophages (MHC class II; b, d, f) is useful for identifying cellularity. Active lesions (a, b) are hypercellular, chronic active lesions (c, d) have a rim of hypercellularity (d), and chronic inactive lesions are hypocellular (e, f)

Full size image

Gray Matter Lesions

GM pathology correlates with physical disability, disability worsening, and cognitive impairment more significantly than WM lesion burden but is greatly underappreciated due to limitations with gross examination and MRI [43,44,45,46]. GM demyelination involves the cortex, deep GM (hippocampus, thalamus, caudate, putamen, amygdala, hypothalamus, and substantia nigra), and the spinal cord [47,48,49,50]. Grossly, cortical GM demyelination does not exhibit characteristic hyperpigmentation from marked myelin loss as in WM lesions. GM atrophy correlates with clinical disability, can precede WM atrophy, is associated with GM lesions, and is detectable in individuals even with low cerebral WM lesion burden [51,52,53,54].

Cortical GM lesions are classified by their anatomic location: leukocortical (type I) lesions span the GM/WM interface, intracortical (type II) lesions are limited to the cortical GM, subpial (type III) lesions are on the surface of the cortex and extend to layers III/IV, and “type IV” lesions are sometimes used to describe subpial lesions spanning the entire width of the cortex and are relatively infrequent compared to other lesion types [5]. The degree and nature of inflammation and injury is variable between these lesion types. In early lesions characterized by biopsy specimens, cortical lesions are highly inflamed and suggested to precede WM demyelination [55]. Leukocortical lesions are seen earlier in the disease and typically start from subcortical WM (with greater inflammation) and extend into the cortex (with lesser inflammation). Intracortical lesions project radially from vessels and have less inflammation than leukocortical and WMLs [56]. Subpial lesions are the most common type of cortical lesion and are unique to MS. These lesions can be extensive, involving entire gyri, frequently occur in deep sulci regions, and often have clearly demarcated demyelinated lesion borders that sharply halt at layer III/IV [9, 50]. Similar to WML, subpial lesion borders frequently contain a line of activated mØ.

Subpial lesions in deep sulci and WMLs adjacent to the lateral and third ventricle suggest a role of cerebrospinal fluid (CSF) factors mediating demyelination. Consistent with this hypothesis are findings of subpial cortical demyelination and neurodegeneration correlating to the presence of meningeal follicle-like structures in a cohort of postmortem tissue from the UK Multiple Sclerosis Tissue Bank from donors with an aggressive disease course [57, 58]. In patients with a secondary progressive (SP) course, a gradient of neuron, astrocyte, and oligodendrocyte loss associated with follicle-like structures in low-flow sulci regions was observed, suggesting the role of yet unidentified soluble factors in mediating neurodegeneration [11]. Additionally, subpial demyelination was noted in association with persistent leptomeningeal enhancement (LME) in biopsy specimens from two subjects with MS [59]. Others have failed to replicate these findings, however, and found no correlation between meningeal inflammation and subpial lesions, nor cortical demyelination and neuronal loss [44, 60, 61]. In one report, meningeal inflammation was similar across myelinated and demyelinated associated subpial regions [60]. There is considerable debate as to the cellular organization of leptomeningeal inflamed regions and what degree aggregates of immune cells constitute a “follicle-like” structure. It has been suggested that such discrepancies could be due to cohort differences and methods of analysis [62]. LME on MRI was noted in 17% (40/240) of individuals with MS [59]. Furthermore, the presence of LME in a longitudinal study was associated with GM and cortical atrophy, and subjects with LME had a longer disease duration and greater disability [63, 64].

A clear unmet need is reliable imaging of cortical pathology. Ultrahigh field 7 Tesla (7 T) MRI and sequences such as 3D T1 magnetization-prepared 2 rapid gradient echo (MP2RAGE) have improved detection of cortical lesions. Using 7 T MRI with postmortem validation, 100% of leukocortical (type I), 11% intracortical (type II), 32% of subpial extending partially through the cortex (type III), and 68% of subpial extending the length of the cortex (type IV) lesions are detected [65].

Atypical Demyelinating Lesions

Some atypical lesions in patients with MS include Balo’s concentric sclerosis and tumefactive lesions. Balo’s is characterized by WMLs with alternating concentric rings of demyelination and non-demyelinated regions associated with T-cell and mØ inflammation and hypoxia-like tissue injury [8, 66, 67]; no GM lesions have been noted [68]. Tumefactive lesions are greater than 2 cm with perilesional edema and/or ring enhancement mimicking a malignant glioma or cerebral abscess [69] and are hypercellular with atypical reactive astrocytes and mitotic figures.

Neurodegeneration

Clinical symptoms such as cognitive impairment and fatigue as well as the accumulation of disability, particularly in progressive courses, are frequently attributed to neurodegeneration including neuronal injury and axonal transection. For example, deep GM atrophy, particularly thalamic, occurs at a greater rate attributable to MS earlier in the disease course, and later the contribution of age-related atrophy matches or exceeds MS-related atrophy [70]. While its precise etiology remains elusive, neurodegeneration (neuronal and axonal injury) can be directly mediated by cytokines and ROS/RNS or indirect consequences of chronic demyelination such as mitochondrial dysfunction and antero-/retrograde degeneration [16, 71, 72]. This is suggested by observations in pediatric-onset MS of WML burden (T2-lesion volume) correlating with thalamic atrophy [73] and limited brain growth [74]. Primary neurodegeneration with loss of neurons and axons independent of demyelination is another proposed mechanism.

The spectrum of axonopathy in WML ranges from impaired axonal transport of proteins/organelles (amyloid precursor protein accumulation and spheroid structures) and impaired fast axonal transport (nonphosphorylated neurofilament heavy chain/SMI-32 immunoreactivity) to irreparable axonal transection and the presence of terminal axonal ovoids distal to site of injury [9, 75]. The degree of axon loss in WMLs correlates with the extent of peripheral immune or resident glial inflammation [9, 15]. Anterograde axonal loss by Wallerian degeneration can occur in normal-appearing WM either adjacent or distal to the lesion and occurs early in MS as evidenced by pathologic and MRI studies [76,77,78,79].

GM normal-appearing and demyelinated areas are associated with neurite transection, synaptic and dendritic loss, and reduced neuron size without clear evidence of neuronal loss [13]. Up to 79% of hippocampi in cohort exhibited demyelinating lesions with reduced synaptic density, neuronal proteins essential for synaptic plasticity, and reduced glutamate neurotransmission, without a significant decrease in neuronal count [80, 81]. These observations are consistent with altered hippocampal MRI measures correlating with memory dysfunction [82].

Myelocortical MS

Contrary to dogma, both WM and GM exhibit areas of demyelination [71, 83], and only approximately half of MRI T2-only lesions in patients with MS are demyelinated [29]. Recently, a new subtype of MS, termed myelocortical MS (MCMS), has been identified with the application of a rapid postmortem in situ MRI with immunohistochemistry [61]. Both patients with typical MS (TMS) and MCMS had similar appearing cerebral WM T2-weighted fluid-attenuated inversion recovery (FLAIR)-lesion burden on MRI (Fig. 8.2a–d) and had severe disability (mean expanded disability status scale [EDSS] > 8.0). However, subjects only had demyelination of the spinal cord and cerebral cortex without any significant cerebral WM demyelination observed grossly (Fig. 8.2f) and histologically. MRI regions of interest were selected for greater demyelination and axonal/neuronal injury—T2T1MTR rather than T2-only areas [29]. These T2T1MTR regions in patients with MCMS however lacked significant demyelination and did not have any T- or B-cell infiltration but had swollen myelinated axons.

Fig. 8.2
figure 2

A novel subtype of multiple sclerosis (MS), termed “myelocortical,” is characterized by cortical and spinal cord demyelination without significant cerebral white matter demyelination. Distribution of MRI FLAIR lesions is comparable between typical (a, c) and myelocortical (b, d) MS, but no apparent demyelination is noted on gross examination (e and f, respectively) or with immunohistochemistry for myelin using PLP (not shown). MRI images are shown in radiological convention (left side of screen represents right hemisphere); both axial (a, b) and coronal (c, d) orientations for the subjects shown in gross images are depicted to appreciate lesion distribution. (e, f: Reprinted with permission from Trapp et al. [61])

Full size image

Compared to MCMS, TMS had more WM and GM atrophy and spinal cord demyelination. A striking observation from this study was the reduction in neuronal densities in five cortical regions not directly connected to the spinal cord in layers III, V, and VI in MCMS compared to control; significant differences were only observed between TMS and control in layer V. In all layers, neuronal densities trended lower in MCMS compared to TMS. Confirming previous work, cortical neuronal loss did not correlate with subpial cortical demyelination in either MCMS or TMS [12]. Neuronal loss in MCMS was not explained by cerebral WM or cortical demyelination, supporting the concept that cortical neurodegeneration can occur independent of cerebral WM demyelination [29].

Oxidative Tissue Injury and Mitochondrial Deficits

Oxidative damage and mitochondrial dysfunction can mediate axonal pathology and neurodegeneration (reviewed in [16, 84]). Oxidative injury, iron accumulation, energy failure, and impaired ion channels are all potential mechanisms of axonal and neuronal injury, which could occur independently or due to inflammatory lesions.

An example of oxidative damage is the production of ROS by upregulation of the nicotinamide adenine dinucleotide phosphate (NADPH) oxidase 2 catalytic subunit (NOX2). NOX2 upregulation is observed on mØ in preactive lesions, diffusely in active demyelinating lesions, and in the rim of chronic active lesions [85, 86]. The release of iron from demyelinating axons, demonstrated by iron deposition and changes in iron homeostasis, is another example [87, 88]. Oxidative injury can be amplified by liberation of ferric iron (Fe3+) from myelin loss and oligodendrocyte death [87], reduction to ferrous iron (Fe2+) by a superoxide anion, and production of highly reactive hydroxyl radicals after combination with hydrogen peroxide (Fenton reaction) [89].

Collectively, ROS/RNS cause metabolic stress, deoxyribonucleic acid (DNA) alkylation, and peroxidation of phospholipids and proteins [90]. Oxidative injury is extensively observed in oligodendrocytes, neurons, myelin, and axons [4], and markers of oxidative insult coincide with upregulation of antioxidant enzymes [4, 28].

A functional consequence of oxidative stress and tissue injury is mitochondrial dysfunction and consequent disease progression [6, 91,92,93,94]. Reduced energy production occurs by disruption of the mitochondrial respiratory chain which involves oxidative phosphorylation of adenosine diphosphate to adenosine triphosphate (ATP) [95]. Due to the lack of protective histones, mitochondrial DNA (mtDNA) is particularly susceptible to oxidative stress leading to deletions and point mutations that may be propagated during clonal expansion. Lesions have neurons functionally deficient in mitochondrial respiratory chain complex IV and mtDNA deletions throughout the GM [6, 91]. Furthermore, deceased ATP production causes dysfunction of the Na+/K+ ATPase, which leads to a reversal of the axolemmal Na+/Ca2+ exchanger, increased intracellular calcium, and consequently axonal degeneration. Demyelinated axons in chronic lesions have a 50% reduction in Na+/K+ ATPase expression [16, 96]. Histotoxic hypoxia, an impaired ability to uptake oxygen, has also been described in lesions as a consequence of mitochondrial injury and a potential contributor to neurodegeneration [96].

Distinctions Between Relapsing and Progressive MS

The clinical course of MS is heterogeneous, and there are several patterns of disease phenotypes characterized based on progression of disability [97]. Of MS patients, 85% start with a relapsing remitting (RR) disease course, and half of those individuals will transition to a SP course after 10–15 years. From onset, 15% of MS patients have a primary progressive (PP) course. Postmortem analysis of 185 subjects with PP and SP with 3188 tissue blocks with 7562 lesions shows similar lesion activity and lesion load [98]. While no differences in lesion pathology were noted, a greater lesion burden and a greater proportion of mixed active/inactive lesions at time of death was associated with a faster clinical disability worsening (time to EDSS 6). Consistent with these findings, subjects with RR had lower lesion load and a greater proportion of lesions with remyelination.

Conclusion

Pathological analysis of postmortem and biopsy tissue has fundamentally shaped our understanding of MS and has provided valuable insight to advance therapeutic discovery and disease monitoring. A postmortem donation program for the procurement of tissue with in situ and post-fixed co-registered MRI, as well as clinical characterization, is vital in advancing our knowledge of MS pathology and clinical disability worsening. Postmortem validation of advanced MRI sequences striving for quantification of myelin and axonal perturbations will be necessary as biomarkers of neurodegeneration for clinical in vivo applications and outcome measures in clinical trials targeting inflammation and neurodegeneration.