Multiple sclerosis is a demyelinating disease, but brain parenchymal and meningeal inflammation and chronic cytokine exposure also affect neuronal metabolism and survival. This leads to brain atrophy, fatigue, cognitive loss, and neurologic abnormalities. The course of multiple sclerosis can be broken down into 3 phases:
Immunity underlying the CNS pathology in multiple sclerosis. The initiating event for the first exacerbation is unknown. Genetics and environment both play a role (Page et al 1993).
Multiple sclerosis plaques are formed after invasion of inflammatory T cells and monocytes. Immune activation is a multi-step process. Before T cells are activated by brain antigens, they may be alerted to a CNS antigen and then “licensed” by the innate immune system, based on models of exposure to viral or microbial antigens through CpG oligonucleotides and toll receptor-9 (TLR-9) or pertussis toxin that cause experimental allergic encephalomyelitis (Darabi et al 2004). During development, it is possible that an autoimmune cascade starts with thymic presentation of alternately spliced golli-myelin basic protein in the context of abnormal costimulatory molecules (Maimone and Reder 1991), or later exposure to a viral antigen. Following peripheral activation, circulating T cells adhere to post-capillary venules in the brain and spinal cord, or penetrate through the choroid plexus. The T cells pass through the endothelial cells and migrate into perivascular brain parenchyma. Equivalent numbers of monocytes and T cells are present in plaques at early stages. Brain dendritic cells can emigrate through cervical lymph nodes and educate peripheral T cells. These T cells may then home back to the brain.
In the plaque, the cellular infiltrate is associated with destruction of the inner myelin lamellae and dysfunction of oligodendroglia, and with diffuse effects such as fatigue and slowed cognition. Early on, gemistocytic astrocytes have high levels of GFAP and when stimulated by IL-9, produce CCL20, which attracts Th17 cells. Astrocytes also produce potentially beneficial trophic factors, BDNF, TrK receptors, and VEGF (Ludwin 2006).
Inflammation, based on the presence of Gd-enhancing MRI lesions, resolves in 2 to 8 weeks. However, some immune cells remain in plaques and are poised for activation, with continued low-grade inflammation, chronic axonal loss, and demyelination.
Immune activation and dysregulation. Immune activation in peripheral blood precedes neurologic problems and MRI activity. Several weeks before attacks, there are increases in Concanavalin A-stimulated IFN-gamma and TNF-alpha production (Beck et al 1988), IFN-gamma levels in serum (Dettke et al 1997), IFN-gamma-induced [Ca++] influx in T cells (Martino et al 1995), and secretion of prostaglandins by monocytes (Dore-Duffy et al 1986). Excessive numbers of cytokine-secreting cells are seen early in multiple sclerosis, even in acute monosymptomatic optic neuritis. Cytokines such as IFN-gamma, osteopontin, and IL-2 activate immune cells, Th17 cells, and endothelial cells, and induce costimulatory molecules that further enhance T cell proliferation and activation (Prat et al 2000a).
During active multiple sclerosis, Th1 cell-mediated inflammation increases. Lymphocytes express excessive levels of the activating zeta chain of the T cell receptor on CD4 T cells (Khatibi and Reder 2008), activation proteins (HLA-DR and CD71) and costimulatory molecules on B cells (CD80, also called B7-1) (Genc et al 1997a), and chemokine receptors (CCR5 and CXCR3) on Th1 cells (Balashov et al 1999). Elevated IFN-gamma receptors on Th17 cells in multiple sclerosis allow IFN-gamma to inhibit these pro-inflammatory cells. Inflammatory cytokines and messenger ribonucleic acid, and cytokine-secreting cells (eg, IL-2, IL-15, IL-17, IL-23, and IFN-gamma), are elevated in mononuclear cells (Trotter et al 1991; Rieckmann et al 1994; Byskosh and Reder 1996; Lu et al 1993). IL-1, IL-6, and IL-15 and TNF-alpha are present in the CSF (Maimone et al 1991a; Kivisakk 1998). These Th1-like cytokines and monokines amplify immune responses. IFN-gamma "therapy" and granulocyte colony-stimulating factor (G-CSF) infusions trigger attacks of multiple sclerosis, despite preventing experimental allergic encephalomyelitis. IFN-gamma, a proinflammatory cytokine, is toxic to actively remyelinating oligodendroglia, and it activates monocytes and microglia. However, it inhibits proliferation of Th1 cells and can cause apoptosis of activated T cells (Ahn et al 2004), and it is protective for mature oligodendroglia (Lin et al 2007). Thus, timing, location, and degree of inflammation are all affected by cytokines.
Control of inflammation is lost during attacks of multiple sclerosis, when concanavalin A-induced suppressor cell function drops (Antel et al 1986). During progressive multiple sclerosis, excessive IL-12 production induces IFN-gamma (Balashov et al 1997). Low production of IL-10 removes another brake on Th1 cells (Soldan et al 2004). IL-15 levels rise in blood and, to a lesser extent, in CSF monocytes during attacks and progression. These changes could lead to delayed-type hypersensitivity (Th1-type) immune reactions and enhanced CD8 T cell cytotoxicity.
The Th1/Th2 dichotomy is too simplistic, however:
Th17 cells are a subset of CD4 cells that amplify autoimmune CNS inflammation and may be important in multiple sclerosis. IL-6 plus transforming growth factor-beta generate IL-17-producing cells from naïve CD4 cells. IL-23 maintains this population and also induces IL-17 in memory CD4 cells. The inflamed blood-brain barrier and monocytes that have transformed into dendritic cells help polarize naïve T cells into Th17 cells (Ifergan et al 2008). IL-4, IL-27, IFN-gamma, and IFN-beta all inhibit IL-17 production.
IL-17-expressing cells increase during exacerbations and are higher in multiple sclerosis plaques and CSF than in serum (Matusevicius et al 1999; Durelli et al 2009). IL-17 receptor is also elevated on brain endothelial cells. Th17 cells rise in optico-spinal multiple sclerosis (Ishizu et al 2005) and likely in some Devic variants of multiple sclerosis. IL-17 is produced by CD4 and CD8 cells and oligodendrocytes in perivascular areas of active multiple sclerosis lesions (Tzartos et al 2008). Cells simultaneously secreting IFN-gamma plus IL-17 are also increased in multiple sclerosis (Th1 and Th17). CSF levels of IL-17 and IL-8 correlate with the length of spinal cord lesions.
The aryl hydrocarbon receptor (AhR) is bound by dioxin, breakdown products of aromatic amino acids (eg, tryptophan), prostaglandins, and compounds in cigarette smoke. The AhR induces both Th17 and regulatory T cells (Tregs). Dioxin inhibits hematopoietic stem cell expansion, but aryl hydrocarbons induce lymphoid follicles in the intestine. Smoke inhibits NF-kB activation, reduces IFN-alpha and IFN-beta production, and increases virus infections (Mian et al 2009). Smoking increases risk of rheumatoid arthritis 1.5-fold, but with variants of HLA and protein tyrosine phosphatase nonreceptor variant 22 (PTPN22), risk is 20-fold. Smoking and inflammation induce homocitrulline, triggering antibodies to it in severe rheumatoid arthritis. Aryl hydrocarbons induce Epstein-Barr virus reactivation and increase risk of Sjogren syndrome. Effects are possible on Devic disease, which is related to Sjogren syndrome. Effects on multiple sclerosis are likely to be complex. AhR ligands in brain germinal-center-like follicles in multiple sclerosis have not been studied, nor have effects of beauty salon vapors or diesel fumes. These fumes suppress NF-kB and interferon-beta production. Effects on multiple immune cell populations and varied culture conditions may explain published differences in Th17 function. Commonly used RPMI culture media has low levels of AhR ligands, but Iscove’s media has high levels and is much more conducive to Th17 cell induction (Veldhoen et al 2009). Conversely, nicotine inhibits immunity (Nouri-Shirazi and Guinet 2012).
CD2 is a costimulatory T cell molecule that binds CD58 (LFA-1). The conformation of CD2 is altered because there is a marked fall in avid rosette-forming cells (CD2 on T cells binds CD58 on RBC) and other antibodies do not bind normally (Reder et al 1991). Although expression of the usually measured epitope of CD2 is normal on CD4 and CD8 cells, stimulation through CD2 is reduced in progressive multiple sclerosis. An allele of CD58 that increases CD58 mRNA is protective against multiple sclerosis (odds ratio = 0.82), and CD58 mRNA is elevated 1.2 times versus normal in exacerbations and 1.7 times in remissions (De Jager et al 2009). There may be a reciprocal relationship between multiple sclerosis state-specific low CD2 function and CD58 expression. Activation through CD2 increases regulatory CD4 cells and CD4 suppressor function; effects on CD8 cells are unknown.
Cytolytic CD8 cells and monocytes in plaques directly damage neurons and axons more than CD4 cells do. CD8 cells that produce Th1-like cytokines are elevated in optico-spinal multiple sclerosis (Ochi et al 2001). Expanded CD8, but not CD4, clones appear in blood, CSF, and multiple sclerosis plaques. Multiple sclerosis therapies tend not to target these cells.
CD8+,CD28- suppressor cell function may be the most important component of immune suppression in multiple sclerosis. The process that regulates these suppressor cells is unknown. When induced by the non-specific mitogen, concanavalin A, suppressor function drops during attacks of multiple sclerosis (Antel et al 1986; Karaszewski et al 1991; Correale and Villa 2008). In an extensive series of experiments, Antel and colleagues showed that the T cell population in multiple sclerosis that suppresses immune reactions is predominantly CD8+CD28- (Antel et al 1979; Crucian et al 1995). CD8 cells had much more potent suppressor effects than CD4 cells. CD8 suppressor cells form a 3-way bridge with monocytes and destroy pathogenic CD4 cells that express HLA-E (mouse Qa-1) (Tennakoon et al 2006; Correale and Villa 2008). CD8+,CD28-,FoxP3+ suppressor cells also induce tolerogenic ILT3 and ILT4 molecules on endothelial cells (Manavalan et al 2004) and on antigen-presenting cells. During exacerbations, high levels of IL-15 and likely IFN-gamma induce expression of the inhibitory NG2A protein on CD8 cells, and their suppressor function falls (Correale and Villa 2008). In mice, similar CD8,CD122 regulatory cells produce IL-10 to inhibit proliferation and IFN-gamma production by CD8 cytotoxic cells. In humans, IL-10 induces these suppressor cells, and so do some multiple sclerosis therapies (below).
Transfer of neuroantigen-reactive CD8 cells inhibits experimental allergic encephalomyelitis (York et al 2010). In CD8 knockout mice, attacks resolve, but later relapses still occur. This would suggest that CD8 cells do not terminate the inflammation in mice but do prevent recurrent attacks. Generalizations across species are suspect, however. The major suppressor cell subpopulation in mice consists of CD4+CD25+ T regulatory cells, but in multiple sclerosis, the more potent subset is CD8+CCD28-.
The defect in mitogen-induced CD8 suppressor cell function in multiple sclerosis is unexplained, but it correlates highly with clinical activity (r = 0.79) (Antel et al 1979), far better than MRI correlates with clinical disease (r = 0.25). MRI also correlates poorly with serum cytokine levels (Kraus et al 2002). This suppressor defect in multiple sclerosis is corrected with IFN-beta, glatiramer acetate, beta2-adrenergic agonists, and Fc receptor ligands. Monitoring of CD8 expression, suppressor cell function, CD80 expression, or specific Th1, Th2, and Th17 markers could predict impending attacks of multiple sclerosis, could differentiate multiple sclerosis attacks from transient fever-induced worsening, and could mirror early therapeutic responses to drugs.
Tr1 CD4 suppressor cells secrete 6 times less inhibitory IL-10 in multiple sclerosis. Target multiple sclerosis cells are also resistant to IL-10 compared to normal controls (Martinez-Forero et al 2008). CD56bright NK suppressor cells (Takahashi et al 2004) and CD4+,CD25++,(CD39+),FoxP3+ T regulatory cells (Treg) are also involved in immune regulation in multiple sclerosis, and the latter have reduced function in multiple sclerosis. Memory Tregs return to normal levels in progressive disease (Venken et al 2008). Treg development requires IL-2, IL-7, vitamin A, TGF-beta, and indoleamine dioxygenase (induced by IFN-beta). These cells are induced by cAMP agonists. The environment in the eye is immuno-suppressive; very small amounts of retinal antigens create CD4,CD25+ cells that inhibit immunity in mice. The CNS may behave similarly.
Myeloid suppressor cells are precursors of macrophages, dendritic cells, and granulocytes. They increase in number during exacerbations (140/million mononuclear cells) versus normal controls (5/million), fall during stable disease (15/million), and have potent function in multiple sclerosis (Ioannou et al 2012).
Thymic export of new T cells is reduced in multiple sclerosis, so blood T cells have fewer T cell receptor excision circles (Trec). Recent thymic emigrant cells, which include naïve T cells and Tregs, are reduced in relapsing-remitting multiple sclerosis (Haas et al 2007). Using this measure, the immune system in multiple sclerosis shows premature aging, and it is 30 years older than in healthy controls (Hug et al 2003).
B cells reflect the abnormal T cell immunity. They also have direct effects on immune regulation and brain destruction (Meinl et al 2006). B cells secrete IL-6, IL-10, TNF-alpha, and chemokines. IL-6 can enhance generation of destructive Th17 cells. In contrast, lipopolysaccharide-activated B cells produce potential beneficial nerve growth factor and brain-derived neurotrophic factor. Nerve growth factor is a survival factor for memory B cells.
B cells are activated in multiple sclerosis. Compared to healthy controls, B cells secrete half as much inhibitory IL-10 after stimulation with anti-CD40 (a model of bystander T cell activation) or B cell receptor plus anti-CD40 (a model of B cell plus T cell activation) (Duddy et al 2007). B cells in multiple sclerosis blood express high levels of costimulatory molecules (CD80). B cells are potent antigen-presenting cells because they are exquisitely focused against specific antigens (Genc et al 1997b). B cells are activated by B cell activating factor (BAFF), made by myeloid cells. CSF BAFF and CXCL13, a B cell attracting chemokine, increase during relapses and in secondary progressive multiple sclerosis (Ragheb et al 2011). CSF BAFF correlates with IL-6 and IL-10 levels, suggesting that all of these factors amplify B cell function and CSF antibody production. CSF CXCL13 is elevated in all forms of multiple sclerosis and correlates with CSF white cells, B cells, IgG synthesis, and MRI activity. The number of CSF B cells and plasmablasts correlates with active MRI lesions in the brain.
High CSF immunoglobulin synthesis and antibody titers to measles virus were reported in the 1950s. CSF IgG and oligoclonal bands are present in more than 95% of patients. In clinically isolated syndromes, clonal expansion is reflected by rearranged mRNA and certain heavy chains (VH4 or VH2) and is more likely to lead to multiple sclerosis, yet these antibodies do not predominantly react against myelin (Bennett et al 2008). There are CSF and serum antibodies to unknown antigens, viruses, myelin proteins, axons (triose-phosphate isomerase), and DNA (ANA). Over 50% of brain plaques contain antibodies plus complement, although the antibodies and oligoclonal bands have not been shown to cause demyelination (Lucchinetti et al 1999). Some anti-brain antibodies enhance remyelination in mice. In progressive multiple sclerosis CSF and brain, B cells clonally expand and are present in germinal center-like areas in the meninges.
Chemokines attract immune cells. Monocytes secrete excessive CXCL8 (IL-8) in multiple sclerosis serum, and presumably CNS, to attract other monocytes and potentially polymorphonuclear neutrophils. However, polymorphonuclear neutrophils are not seen in multiple sclerosis CSF. In contrast, in Japanese optico-spinal multiple sclerosis, there is increased IL-8 and IL-17 as well as both Th1 (IFN-gamma) and Th2 (IL-4 and IL-5) cytokines. In a subset of patients with this Japanese Devic-like variant, IL-8 in CSF and neutrophils in lesions correlate with spinal cord lesion formation (Ishizu et al 2005). IFN-beta decreases IL-8.
Multiple sclerosis CSF and plaques contain CCR7+ dendritic cells; T cells express CCR7 only in the CSF. T cells in plaques have downregulated CCR7, a receptor needed for migration, so they are unable to leave the CNS (Kivisakk et al 2004).
Monocytes and microglia present antigens and amplify immune responses. They communicate with cells hundreds of microns away through tunneling nanotubes that transmit calcium ions and antigens. They over-express receptors for immunoglobulins and are activated by low levels of serum receptor for advanced glycation end-products (RAGE). Inhibitory molecules expressed by monocytes (HLA-G, ILT3) are reduced in multiple sclerosis, but expression is upregulated by IFN-beta (Mitsdoerffer et al 2005; Jensen et al 2010) and vitamin D, IL-10, and suppressor CD8 T cells. Peripheral monocytes produce excessive nitric oxide, which is neurotoxic and damages oligodendroglia, but also destroys activated T cells. Microglia in the brain release nitric oxide, oxygen radicals, complement, proteases, and cytokines. CSF nitric oxide metabolites correlate with gadolinium-enhanced MRI lesions, clinical activity, and progression of multiple sclerosis. Nitric oxide also modifies brain proteins to form nitrotyrosine. This creates neoantigens in the brain and generates antibodies to S-nitrosocysteine in the CNS (Boullerne et al 2002). Even though activated macrophages are generally toxic to CNS cells, they may have positive effects too. (See “Recovery from relapses,” below.)
IFN-alpha-secreting plasmacytoid dendritic cells are more frequent in early multiple sclerosis in some studies. However, the multiple sclerosis cells produce less IL-10 and less IFN-alpha and are defective as antigen-presenting cells (Stasiolek et al 2006) and more likely to induce Th17 cells. In contrast, myeloid dendritic cells in secondary progressive multiple sclerosis are activated and proinflammatory (Karni et al 2006).
Trauma and stress have been proposed as causing multiple sclerosis or triggering exacerbations (McAlpine et al 1972; Poser 1986; Buljevac et al 2003; Li et al 2004). Stress and exacerbations are sometimes difficult to define, and studies conflict. Links of exacerbations to stress and trauma are nonexistent when stress, trauma, and concomitant clinical manifestations of multiple sclerosis are carefully analyzed (Sibley 1988; 1993; Siva et al 1993), even though there is a slight increase in new MRI lesions (Mohr et al 2000). Stress in the home and physical abuse during childhood appear to prevent multiple sclerosis. Gunshot wounds and SCUD missile attacks actually seem to protect against exacerbations according to some reports (Sibley 1988; Nisipeanu and Korczyn 1993), but another war report suggests increased exacerbations (Golan et al 2008). Local irradiation of the brain can increase lesions of multiple sclerosis within the radiation field, possibly by disruption of the blood-brain barrier (Murphy et al 2003).
The hypothalamus regulates autonomic functions, body temperature, sleep, and sexual activity. Hypothalamic corticotrophin releasing hormone (CRH) controls an endocrine cascade to adrenocorticotropic hormone and then to cortisol. Serum cortisol and exogenous steroids turn down corticotrophin secretion. Endocrine activity also affects immunity.
Hypothalamic plaques are common in multiple sclerosis and disrupt endocrine regulation (Huitinga et al 2004). Surviving myelin bundles are next to HLA class II positive microglia. Inflammation in the hypothalamus may explain the high number of double-positive corticotrophin and arginine-vasopressin neurons that are unique to multiple sclerosis, especially in disease of long duration. Arginine-vasopressin potentiates the action of corticotrophin on adrenocorticotropic hormone release. The resultant elevation in cortisol could be beneficial because high numbers of corticotrophin-releasing factor/arginine-vasopressin neurons correlate with low hypothalamic lesion load. Similarly, rats with high corticosterone are protected against experimental allergic encephalomyelitis.
The hypothalamic-pituitary-adrenal (HPA) axis is hyper-responsive to corticotrophin-releasing hormone, especially in primary progressive multiple sclerosis (Then Bergh et al 1999). Chronic HPA axis overactivity may render cells insensitive to glucocorticoids and allow them to escape from immune restraint. Levels of cortisol, adrenocorticotropic hormone, dehydroepiandrosterone, and cells secreting corticotropin releasing hormone are increased most in progressive and active forms of multiple sclerosis (Ysrraelit et al 2008). Glucocorticoids plus antidepressants normalize the HPA axis in multiple sclerosis.
Acute and chronic inflammation induces high serum cortisol levels that cause systemic and local steroid resistance. IL-1alpha, produced by activated macrophages, inhibits glucocorticoid receptor translocation to the cell nucleus (Pariante and Miller 2001). High levels of tumor necrosis factor and IL-1 and IL-6 correlate with hypothalamic-pituitary-adrenal axis (HPA) activation and with fatigue. In parallel during active multiple sclerosis, the hypothalamic-pituitary-adrenal axis is hyporesponsive to dexamethasone feedback, and so are immune cells ex vivo (Reder et al 1987). Conversely, cyclic adenosine monophosphate (cAMP) agonists (prostaglandins, beta-adrenergic agonists such as terbutaline, and some antidepressants) enhance steroid receptor translocation and could potentiate glucocorticoids. Weak response to steroids correlates with high CSF white blood cell counts and enhancing lesions on MRI (Fassbender et al 1998). Mechanisms for resistance to steroids include (1) downregulation from chronic high cortisol (mildly increased in multiple sclerosis), possibly from adrenocorticotropic hormone released by immune cells (Reder 1992; Reder et al 1994c; Lyons and Blalock 1997); (2) a mutation in the steroid receptors; and (3) interaction with other signaling pathways.
Recovery from relapses. Immune regulation causes inflammation to wane. As clinical symptoms resolve, there is a rise in inhibitory Th2 cytokines, immunoglobulins, and glucocorticoids and suppression of inflammation (Reder et al 1994a). Axonal sodium channels redistribute in ravaged, but surviving axons, and there is remyelination and rewiring of the brain (compensatory adaptation; functional reorganization of neurons and synapses).
Inflammation is turned off by apoptosis and suppression of activated immune cells. Apoptosis of Th1 cells is mediated by steroids (endogenous or therapeutic), IFN-gamma (Furlan et al 2001; Ahn et al 2004), TNF-alpha, and nitric oxide. IFN-beta causes apoptosis of Th17 cells, which express high levels of the type I interferon receptor (Durelli et al 2009). Some of these same compounds are toxic to neurons and oligodendroglia (TNF-alpha, glutamate, nitric oxide, and others). However, subnormal suppressor T cell function in clinically active multiple sclerosis may prolong inflammation.
Macrophages secrete some compounds that are neuroprotective, suggesting there is a balance between destruction and repair during inflammation. Macrophages produce platelet-derived growth factor (PDGF), epidermal growth factor (EGF), transforming growth factor beta (TGF-beta), insulin-like growth factor 1 (IGF-1), neural growth factor (NGF), brain-derived neurotrophic factor (BDNF), and neurotrophin-3 (NT3). BDNF is expressed in lesions by T cells, macrophages, microglia, and astrocytes. Immune cells secrete more BDNF during relapse, but levels fall with progression. After relapses, other neurotrophic factors rise, including glial cell-line derived neurotrophic factor (GDNF), NT3, NT4, NGF, and possibly ciliary neurotrophic factor (CNTF) and leukemia inhibitory factor (LIF). Foamy macrophages, after ingesting myelin, secrete anti-inflammatory IL-4, IL-10, and prostaglandin (PGE). Staph A-activated monocytes produce 10 times more IL-10 (3000 pg/ml) than B plus T lymphocytes (Hamamcioglu and Reder 2007). IL-4, IL-10, and granulocyte-macrophage colony-stimulating factor (GM-CSF) in turn induce anti-inflammatory properties in microglia and monocytes.
IFN-beta and IFN-gamma cause macrophages to produce indoleamine 2,3 dioxygenase, an anti-inflammatory compound that helps induce regulatory T cells. Glatiramer acetate activates type II monocytes, which induce Th2 cells and regulatory CD4 T cells, and these inhibit experimental allergic encephalomyelitis (Weber et al 2007a).
The central nervous system is normally hostile to immune activation (immune privilege). The blood-brain barrier prevents access of white blood cells and cytokines to the brain because of tight junctions, reduced lymphocyte adhesion molecules, less endothelial transcytosis, and potent extrinsic pumps (PgP) and active transporters. Glia secrete transforming growth factor-beta, platelet-derived growth factor, and prostaglandin E that inhibit lymphocyte proliferation (Reder et al 1994b). As an example of CNS immune deviation, brain tumors that produce inhibitory cytokines are more aggressive (Whelan et al 1989). Neurons secrete factors that prevent induction of apoptosis. Brain antigens draining into the cervical lymphatics provoke strong antibody responses and Th2 immunity that can block Th1-mediated inflammation.
Evolution of the progressive clinical course: many possible causes. Repeated exacerbations and remissions change to a progressive clinical course approximately 10 years after disease onset. Clinical remissions wane, but constant low-grade immune activation continues or may change in character. In progressive multiple sclerosis, there is cumulative loss of oligodendroglia and neurons, with increasing demands on surviving, yet compromised, cells.
A theory for the evolution from relapsing-remitting to progressive multiple sclerosis is that early subclinical neurodegeneration eventually becomes noticeable. Surprisingly, however, patients who continue to have frequent relapses after the first 2 years are less likely to become progressive (Scalfari et al 2010). This suggests that there is a distinctive transformation from relapsing-remitting to progressive multiple sclerosis. The mechanism for this gradual failure of immune regulation and CNS repair is unknown (Maimone and Reder 1991), but there are many changes that could provoke the transition from relapses to progression:
(1) Spontaneous and activation-induced apoptosis are impaired in T cells during clinically active multiple sclerosis (Zipp et al 1999; Sharief 2000), so autoimmune cells are not eliminated.
(2) Interferon signaling becomes subnormal in mononuclear cells from patients with progressive multiple sclerosis--but not in relapsing-remitting disease (Feng et al 2002a).
(3) Weak suppressor T cell function, seen intermittently during exacerbations, becomes continuous with progression (Reder and Arnason 1985; Antel et al 1986). As a result, autoimmune T cells accumulate over time, and there is loss of peripheral immune tolerance.
(4) T cell clones from patients with progressive multiple sclerosis are insensitive to steroids. Steroids only weakly inhibit proliferation, and there is little apoptosis, a “pre-leukemic state” (Correale et al 1996).
(5) Adhesion molecules shed from the lymphocyte surface (low expression), leading to high levels in serum in relapsing-remitting and secondary progressive disease. Serum adhesion molecule levels are normal in primary progressive multiple sclerosis (Duran et al 1999). This suggests that T cell-endothelial cell adhesion is important in relapsing disease, but there is less T cell-endothelial activation in primary progressive multiple sclerosis.
(6) Gadolinium-enhancing MRI lesions decrease in frequency, possibly from a change in the makeup of inflammatory CNS cells or in endothelial cell activation.
(7) Nonetheless, patients who have primary progressive multiple sclerosis and high MRI T2-weighted lesion volume have excessive IFN-gamma production and rapid T cell migration through endothelial cells (Prat et al 2000b).
(8) Monocyte and microglial changes indicate excessive innate immunity, with a proinflammatory profile. Monocytes produce 5- to 10-fold more IL-12 and IL-18, and more IL-23, in progressive multiple sclerosis (Balashov et al 1997). These interleukins induce IFN-gamma, along with a change in the character of dendritic cells (Karni et al 2006), to activate pro-inflammatory Th1 cells. In contrast, protective IL-10, PGE, and BDNF decrease. A fall in IL-10 is associated with more disability and more MRI lesions. Immune cells in secondary progressive multiple sclerosis also secrete high levels of kallikreins and destructive serine proteases; both are associated with more disability.
(9) Plaques contain more monocytes and fewer T cells in chronic disease. Perhaps related to a shift to innate immunity, progressive MS does not respond clinically to potent anti-T cell therapy.
(10) Germinal center-like areas appear under the meninges. These organized sites of chronic B cell activation suggest there is a loss of CNS immune control.
(11) Neurotrophic abilities are lost as disability progresses, both in the central nervous system and in peripheral blood cells. Nerve growth factor produced by endothelial cells drops with increasing disability (Biernacki et al 2005) as does BDNF produced by T cells (Hamamcioglu and Reder 2007). Chronic defeat stress, and possibly the stress of brain inflammation, reduces BDNF levels through di-methylation of the BDNF gene (a DNA-repressive modification, not demethylation) (Tsankova et al 2007).
(12) Inertia. Non-inflammatory damage from cuprizone, followed by completed remyelination, is clinically stable for some time. After 6 months, however, slow axonal degeneration crosses a threshold and mice lose clinical function (Manrique-Hoyos et al 2012). In multiple sclerosis, residua from old lesions, plus ongoing inflammation, may cause similar loss of function and atrophy.
(13) Regional brain atrophy is periventricular in relapsing multiple sclerosis, but atrophy is cortical in progressive multiple sclerosis (Pagani et al 2005), and the gray matter is more abnormal on MRI (Pulizzi et al 2007; Fisniku et al 2008). The rate of atrophy in gray matter accelerates at the transition to progressive multiple sclerosis, and gray matter loss is the main contributor to total brain atrophy (Fisher et al 2008). Deep cortical invaginations are most affected and abut the most germinal center-like inflammation (Lassmann 2008).
(14) Cord atrophy and neuronal loss appears in all forms of multiple sclerosis. Atrophy correlates with disability and progression.
(15) Axonal damage markers in the CSF predict conversion to definite multiple sclerosis after a first attack (Brettschneider et al 2006). Tau and neurofilaments (40%) are slightly better than MRI (34%) as predictors. Both increase further in secondary progressive multiple sclerosis, reflecting damage.
(16) Destruction of oligodendrocytes continues at a faster pace in secondary progressive multiple sclerosis than in relapsing-remitting multiple sclerosis, and there is more myelin basic protein-like material in urine.
(17) Oligodendrocyte precursor cells (OPC) are lost as plaques accumulate. OPC are also not recruited to the lesions and fail to differentiate, possibly from inhibitory factors in plaques such as hyaluronan, a glycosaminoglycan, or from inhibitory molecules on demyelinated axons such as polysialated neural cell adhesion molecule (PSA-NCAM) (Franklin and Ffrench-Constant 2008).
(18) In chronic plaques, premyelinating progenitor oligodendrocytes often extend processes to axons, but importantly, do not wrap around axons. The axons that spurn the oligodendrocyte remain dystrophic and swollen and vulnerable to insults (Chang et al 2002).
(19) Activated astrocytes inhibit extension of oligodendrocyte processes. Normal central nervous system astrocytes express beta2-adrenergic receptors, which reduce major histocompatibility complex class II and adhesion molecule expression, inhibit immune responses, and prompt secretion of trophic factors and lactate, an energy source for axons and oligodendrocytes. Adrenergic receptors are absent from astrocytes in multiple sclerosis and in Alzheimer disease (De Keyser et al 1999). Loss of these receptors may overcome immune privilege (above) and allow overactive immune responses in the central nervous system.
(20) A gradual increase in plaque burden may eliminate neural pathways that regulate immunity. Autonomic responses are frequently abnormal in progressive forms of multiple sclerosis. A "strategic hit" to central autonomic pathways may interfere with the immuno-inhibitory tone from spinal sympathetic fibers that innervate the spleen (Karaszewski et al 1990). A form of denervation supersensitivity appears in immune cells—beta2 adrenergic receptors are overexpressed on CD8 T cells in progressive multiple sclerosis. These cells have exaggerated cyclic AMP responses. This change may affect suppressor CD8 T cell function and possibly innate immunity. Oral terbutaline, a beta2-adrenergic agonist, does not enhance cytokine expression (IL-10 and IL-12) in whole blood cells activated with T cell mitogens in multiple sclerosis (especially with higher EDSS), but controls have strong responses (Heesen et al 2002). Oral terbutaline does increase CD8 suppressor function, however (Reder and Arnason 1991, unpublished).
(21) Cortisol levels rise, adrenals increase in size, and feedback inhibition of the hypothalamic-pituitary-adrenal axis by cortisol is abnormal in progressive multiple sclerosis, perhaps from excessive production of AVP/CRH in the plaque-filled hypothalamus. Glucocorticoid therapy no longer suppresses multiple sclerosis symptoms. High glucocorticoid levels, endogenous and therapeutic, are linked to hippocampal atrophy. Hypercortisolemia may be corrected with antidepressants.
(22) Therapies directed at T and B cells, which are effective in relapsing or transitional multiple sclerosis, do not work in progressive multiple sclerosis. These include alemtuzumab, glatiramer, IFN-beta, IVIG, mitoxantrone, and rituximab. Riluzole had MRI benefit in a small study. A large fingolimod trial is ongoing.
Clinical characteristics interact with the above factors. Progression and poor prognosis is more likely in older patients and men, and involved sphincter, motor, and multi systems and poor recovery after attacks.
Pathology of the lesions. Multiple sclerosis plaques are found in both gray and white matter throughout the brain and spinal cord. Lesions are random--but have predilection for some brain regions. Periventricular and periaqueductal sites are most likely to suffer, and optic nerves are almost always involved. Cord lesions are often subpial. “Normal-appearing white matter” is abnormal on magnetic resonance spectroscopy and on histology where there is significant axonal loss, especially near plaques.
A plaque is a well-demarcated area with myelin loss, inflammatory cells, gliosis, and relative but partial preservation of axons and neurons. Demyelination usually predominates, but in some cases axonal loss is severe (Trapp et al 1998). Swollen, hypertrophied astrocytes at plaque edges contain dense core particles and sometimes endocytosed oligodendroglia. Astrocytes are initially hypertrophic or gemistocytic. Months to years later they become fibrillary and form fibrous scars (sclerosis; gliosis).
The periventricular location of multiple sclerosis plaques has been explained by multiple theories: effects of CSF toxins or cytokines; regional variation in microglia or capillary pericytes; and slow blood flow in the post-capillary venules that facilitates T cell adhesion. Local toxins are unlikely because abluminal molecules diffuse throughout the brain within 6 minutes, facilitated by arterial pulses (Rennels et al 1985). Rapidly diffusing cytokines should activate pericytes or endothelial cells throughout the central nervous system. MRI in multiple sclerosis suggests that a slow flow rate (less than 50% of normal) through periventricular veins allows immune cells time to attach to endothelium (Law et al 2004). All plaques are perivenular on MRI (Tan et al 2000), and areas with high vein density are most frequently affected.
The blood-brain barrier consists of specialized endothelial cells connected by tight junctions. A few blood-brain barrier areas lack tight junctions but do not have increased plaque activity, suggesting that activated endothelial cells in the post-capillary venules actively attract immune cells into the restricted areas.
The endothelium is abnormal in multiple sclerosis. In 1872, Rindfleisch described abnormal blood vessels in all lesions (Kirk et al 2004; Ludwin 2006). Endothelial is also abnormal in experimental allergic encephalomyelitis (Muller et al 2005). Active plaques appear on MRI because of Gd uptake by activated endothelial cells and/or leakage though the barrier. Lack of Gd+ lesions in progressive multiple sclerosis would suggest there is less breakdown of the blood-brain barrier in progressive multiple sclerosis, yet serum proteins extrude through the blood-brain barrier in progressive multiple sclerosis more than in other forms (Leech et al 2007). Tight junctions may be compromised, even with less apparent inflammation, as fibrin is increased in the perivascular space in chronic plaques (Kwon and Prineas 1994; Claudio et al 1995; Leech et al 2007). Many active demyelinating lesions are missed on MRI, suggesting a spectrum from profound inflammation in classical plaques, to moderate inflammation in slowly expanding plaques, to mild or no inflammation in inactive plaques (Lassmann 2008).
Choroid plexus cells are activated in multiple sclerosis, with HLA-DR and VCAM-1 expression by macrophages, dendritic cells, and epiplexus cells (Vercellino et al 2008). It may be an important site for antigen presentation and for the earliest lymphocyte entry into the CNS.
Adhesion molecules on endothelial cells bind ligands on T cells. The T cells then penetrate directly through the endothelial cells (Astrom et al 1968), not necessarily through the tight junctions. The self-amplifying loop between activated endothelial cells and activated T cells can be blocked by natalizumab and interferon. After monocytes cross the blood-brain barrier, they pile up in the perivascular space. The basement membrane/basal lamina is breeched in a second step by matrix metalloproteases. Interferons decrease matrix metalloproteases. The combination of natalizumab and IFN-beta could synergistically decrease leukocyte traffic into the CNS.
The initial inflammatory lesion is a cuff of macrophages and CD4 T lymphocytes that surround vessels lined with endothelial cells that express major histocompatibility complex class II proteins (Traugott et al 1985; Raine 1991), similar to the lesions of experimental allergic encephalomyelitis. Very early, the margin may be indistinct (Raine 1991). Sensitive magnetization transfer ratio scans can detect abnormalities weeks prior to Gd+ enhancement. The cellular infiltration is minimal in some acute plaques, suggesting a direct insult to oligodendrocytes that is similar to the Lucchinetti type III lesions described below (Barnett and Prineas 2004). These authors argue that in some early cases, oligodendroglial damage precedes immune infiltration; then macrophages arrive, followed by T cells. This frequently debated proposition suggests multiple sclerosis is a primary degenerative disorder. In another series of very early plaques, there were clusters of activated microglia, a few CD8 > CD4 cells, activated complement but not on myelin, and mild to moderate demyelination, all near Virchow-Robin spaces filled with CD4 and B cells (Gay et al 1997).
Acute plaques typically spread out from the post-capillary venules. Clonally expanded (“oligoclonal”) CD8 cells begin to outnumber CD4 cells at the plaque margins (Booss et al 1983; Babbe et al 2000). These margins are often abrupt and thin, suggesting a battle between the spread of inflammation and endogenous resistance to the intruders. “Immune privilege” in the brain, from many factors, contributes to the immune cloaking of brain cells. Neurons express immune inhibitory transforming growth factor-beta. Neuron-T cell contact converts encephalitogenic CD4 T cells to regulatory CD4 T cells that inhibit experimental allergic encephalomyelitis (Liu et al 2006). Nonetheless, CD8 cells can damage neurites, axons, and oligodendroglia. Mitochondrial number and protein expression are increased in axons and astrocytes of active and inactive lesions.
Plaques are of various ages in multiple sclerosis, unlike the monophasic lesions of postinfectious and postvaccinal encephalomyelitis. Some old multiple sclerosis plaques show ongoing demyelination and infiltrating macrophages that phagocytose compact myelin.
Gray matter is affected, as it does contain some myelinated fibers. There is neuronal and synaptic loss in the cerebral cortex, as well as death in up to 35% of thalamic neurons (Cifelli et al 2002). Although T2 lesions are rare on MRI, cortical demyelination can be extensive, and atrophy correlates with fatigue, cognitive loss, and physical disability. Deep gray matter (thalamus, putamen, caudate) also atrophies and has reduced blood flow. Extensive neuronal loss in the hypothalamus is common (Huitinga et al 2004) and may explain circadian rhythm disruption and alterations in cortisol regulation, sexual function, depression, and even poor sleep (Javed and Reder 2006).
Cortical lesions can be contiguous with expanding subcortical lesions (type 1), can be confined to small perivascular areas of the cortex (type 2), or can extend from pia to cortical layer 3 or 4, usually in progressive disease (type 3) (Peterson et al 2001). These multiple sclerosis-specific, subpial lesions are mainly found in cerebellum, hippocampus, and deep invaginations—in the cortex of the insula, cingulate, and deep occiput and fronto-and temporal-basal areas (Lassmann 2008). Cortical plaques are 60% of all brain lesions, but are easily missed on histopathology when a long Luxol Fast Blue destaining step is used to detect white matter demyelination.
Cortical lesions are less inflammatory than white matter lesions. There are 10 to 40 times fewer T cells and 6 times fewer macrophages and microglia, plus little edema compared to the adjacent white matter plaque. Nonetheless, activated microglia ensheath neurites, and apoptotic neurons appear. Lesions are associated with lymphoid germinal center-like B cell inflammation. Type 1 cortical lesions have less GAP43 protein, suggesting that atrophic cortex has lost neurons (average 10% fewer) and glia (36%). (MRI appearance of plaques is discussed in MRI versus histopathological subtypes.)
The oligodendrocyte is the major target. Each of these cells maintains myelin on up to 50 axons and has an extraordinary metabolic demand. They are easily damaged, yet contain plentiful protective mechanisms. Oligodendrocyte progenitor cells (OPC) are numerous, widely distributed, and can remyelinate naked axons. Triggers for OPC differentiation and survival include IL-11 and chemokine CXCL2. Fibroblast growth factor enhances OPC recruitment but inhibits differentiation. Myelin basic protein inhibits OPC differentiation; IFN-gamma inhibits remyelination. (See Recovery from relapses, above.)
Monocytes and macrophages destroy neurons and oligodendroglia, and they may increase in later stages of multiple sclerosis. Oligodendrocyte loss correlates with the number of macrophages, but not with T cells or plasma cells in histological sections (Lucchinetti et al 1999). NK, gamma/delta T cells, and CD8 T cells also damage oligodendrocytes through the NKG2D protein and other targets. CD4 cells do not directly damage oligos, but secrete cytokines that activate CD8 cells and macrophages. Supernatant from resting multiple sclerosis B cell cultures is more toxic to oligodendrocytes than those from B cells of normal controls. Astrocytes and B cells secrete IL-15, which enhances function of CD8 cytolytic cells.
Monocytes and microglia activate NOS and also cause lipid peroxidation, tyrosine nitrosylation, and DNA stand breaks (Zeis and Schaeren-Wiemers 2008). Within active lesions next to dystrophic axons, macrophages express high levels of glutaminase, involved in glutamate synthesis (Werner et al 2001). Macrophages release IL-1, which induce glutamate to cause synaptic hyperexcitability. Glutamate is toxic to neurons and oligodendroglia. Inflammation increases p53 in oligodendrocytes, increasing susceptibility to apoptosis. IFN-beta blocks secretion of anti-inflammatory, protective IL-10 in activated macrophages and possibly in microglia, contrary to its effect in the periphery where it increases T cell IL-10 (Feng et al 2002b). This suggests therapy should be tailored to the inflammatory makeup of the brain as the disease evolves.
Partial remyelination takes place during recovery from relapse, especially in early multiple sclerosis, but also occurs in progressive multiple sclerosis (Patrikios et al 2006). Many fibers are thinly myelinated within acute plaques or at the edge of chronic plaques during and after active myelin breakdown. Remyelination is more extensive in the cortex than in the white matter (Albert et al 2007). Moderate remyelination by hyperplastic oligodendroglia sometimes gives the appearance of a "shadow plaque." There is severe demyelination in the center but thin myelin sheaths in the shadowy periphery of the plaque (Prineas et al 1993). Remyelination is enhanced by some cytokines and gliotrophic factors secreted by immune cells, including macrophages. Low levels of inflammatory cytokines may be able to trigger protective oligodendroglial genes such as HIF-1alpha and HSP70; oligodendrocytes also can produce growth factors such as NGF, IGF-1, and TGF-beta (Zeis and Schaeren-Wiemers 2008). N-acetyl-aspartate (NAA), synthesized in the mitochondria of neurons, is reduced in lesions. As the brain recovers, NAA in plaques increases most in patients with better clinical outcome (Ciccarelli et al 2010).
Oligodendroglia precursor cells (OPC), expressing the anti-apoptotic protein Bcl-2, arise in the plaque or may have migrated from out of the subventricular zone. These cells proliferate and migrate to demyelinated lesions. In plaque subtypes I and II (Lucchinetti et al 2000), oligo precursors are preserved and can form remyelinating shadow plaques. Macrophage products, plus platelet-derived growth factor (PDGF) and fibroblast growth factor-2 (FGF-2), increase the number of OPC.
There are premyelinating oligodendrocyte precursors in chronic plaques, yet few mature oligodendrocytes and surprisingly little remyelination of adjacent bare axons (Chang et al 2002). Loss of axonal receptivity for remyelination and lack of remyelination could be from multiple factors:
Perineuronal oligodendrocytes are non-myelinating progeny of OPC that have a unique membrane and RNA profile (Szuchet et al 2011). These cells support neurons and protect themselves and neurons from apoptosis, and are located in the cortex. They are unexplored in multiple sclerosis. These cells respond to oligo-preserving therapies in multiple sclerosis and amyotrophic lateral sclerosis.
With each CNS insult and during chronic multiple sclerosis, more oligodendrocyte precursors and mature cells die, and the ability to remyelinate decreases. With chronic progressive disease, normal-appearing white matter has diffuse reduction of myelin, infiltration of T cells, and microglial activation (Lassmann 2008). Reactive oxygen species, glutamate, proteases, viruses, and immune cell products (nitric oxide, IL-1-beta, IFN-gamma, and tumor necrosis factor alpha) also damage myelin. IFN-gamma protects mature oligos against oxidative stress, but damages immature oligos (Balabanov et al 2007). However, some patients exhibit remyelination that trumps the effect of age. Some shadow plaques show only small areas of remyelination at plaque margins, but others show extensive remyelination, including 2 elderly patients with longstanding disease (Patrikios et al 2006). Loss of trophic support from oligodendroglia and myelin potentially damages axons, even in the absence of inflammation.
Axons are damaged in multiple sclerosis. During the earliest stages, there are abundant axonal “ovoids,” ends of transected axons ballooning from ongoing anterograde transport (Charcot 1850; Trapp et al 1998). There are 10,000 axonal spheroids/mm3 in multiple sclerosis plaques (indicating transection months or years earlier), but only 2/mm3 in healthy controls (Trapp et al 1998). Early active lesions have 10% to 20% axonal loss. Even in normal-appearing multiple sclerosis white matter, axons are half as numerous as in control brains. In chronic progressive multiple sclerosis, two thirds of the axons are lost; one half of the axons in many long tracts and the corpus callosum disappear (Bjartmar and Trapp 2001). Unexpectedly, there is little correlation between plaque load and axonal loss, suggesting different types of inflammation will provoke the axonopathy. The atrophy rate on MRI in multiple sclerosis is approximately 1% per year; this rate is higher than in healthy controls (0.12%), but less than in Alzheimer disease (3%).
Axonal damage predominates in selected pathways such as the optic nerve, corpus callosum, and spinal cord. In the cervical cord, up to 65% of the axons can be lost. In the corpus callosum, transcallosal bands of Wallerian degeneration predict poor prognosis. Axonal loss correlates with clinical disability and with central nervous system atrophy (Trapp et al 1998). The damage predominates in small-sized axons, in corticospinal axons at all levels, and in sensory axons largely in the cervical cord (DeLuca et al 2006). Axonal loss can be severe enough to cause clinical diaschisis and elevation of neurofilament light protein in the CSF, Wallerian degeneration on MRI, low magnetization transfer ratio, and a contralateral decrease in N-acetyl aspartate (NAA) on MR spectroscopy.
Circulating anti-ganglioside antibodies in progressive multiple sclerosis reflect axonal damage. They are present in 50% of primary or secondary progressive patients compared to only 3% of relapsing-remitting patients (Sadatipour et al 1998). T cells proliferate excessively to GM3 and GQ1b in primary progressive disease. Some antibodies are myelin-protective.
Demyelination and toxic cytokines increase the energy demand of impulse conduction by axons. As inflammation subsides, axonal sodium channels redistribute diffusely in demyelinated axons, away from the former nodes of Ranvier. More of these diffused K+ and Na+ channels are activated per unit length of axon. This greater Na+ influx is accompanied by a greater Ca++ influx (one tenth of the Na+ flux), so compromised axons must contend with sequestration of potentially toxic Ca++ (Rosenbluth et al 1999). Axonal mitochondria are dysfunctional, also leading to Ca++-mediated axonal degeneration. Low axonal ATP is akin to a state of chronic hypoxia with mitochondrial dysfunction, high Na influx, and high Ca++ influx (Trapp and Stys 2009).
Altered ion channels affect function. Ten molecularly distinct subtypes of sodium channels control timing and duration of axon potentials. A transcriptional channelopathy can arise when new types of sodium channels appear at high density in demyelinated axons (Waxman 2002). There is also robust expression of sodium channels on activated microglia and monocytes. Blockade of Na channels with phenytoin decreases inflammation (Craner et al 2005), and phenytoin and flecainide inhibit experimental allergic encephalomyelitis. Abrupt withdrawal of phenytoin or carbamazepine provokes exacerbation of experimental allergic encephalomyelitis, although this is not reported in multiple sclerosis. Lamotrigine, a Na channel blocker, slightly enhanced walking speed, but was linked to more brain atrophy (Kapoor et al 2010).
Many mechanisms subvert immune privilege of the brain. Myelin basic protein, tau, neurofilaments, and 14-3-3 proteins in CSF reflect neuronal and glial damage. CNS antigens drain through cervical lymphatics to secondary lymphoid organs where B cells are educated and produce antibodies. Presumably sensing these brain antigens, deep cervical lymph nodes produce large amounts of antibodies and educated T cells that return to the CNS. The Virchow-Robin spaces in the brain contain extracellular matrix proteins, facilitating migration of MHC-expressing macrophages that are well-placed to interact with T and B cells plus activated B cells. In chronic plaques, astrocytes hypertrophy and express B7-1 and B7-2 costimulatory molecules, possibly allowing them to present antigens. B cells also act as antigen-presenting cells, co-stimulate T cells, and secrete cytokines (Cross et al 2001).
Antibodies are produced within the CNS itself. B cells in the CNS undergo local clonal expansion, activation, “receptor editing,” and hypermutation and develop an activated memory cell phenotype (Monson et al 2004). Editing of surface immunoglobulin in response to antigens improves affinity but ordinarily reduces proclivity to autoimmune disease. These clonal cells produce immunoglobulin in an oligoclonal band pattern.
Lymph node medulla- and spleen germinal center-like areas appear in the perivascular spaces of some old multiple sclerosis plaques (Prineas 1979) and seem to be restricted to secondary progressive multiple sclerosis. Similar structures are seen in affected tissues in rheumatoid arthritis, Sjögren syndrome, Crohn disease, and Hashimoto thyroiditis. This organized inflammation appears in the meninges in 40% to 54% of autopsy cases. In the presence of TNF-alpha, B cells proliferate in the meninges and form germinal center-like areas. Lymphotoxin-alpha, essential in formation of tertiary lymphoid tissue, is elevated in multiple sclerosis CSF. Germinal center-like areas in the meninges contain follicular dendritic cells, proliferating B cells, and plasma cells (Serafini et al 2004). The B cell follicle-like areas are associated with subpial demyelination and cortical atrophy. These ectopic B cell follicles in some cases are major sites of Epstein-Barr virus persistence, possibly driving antibody production (Serafini et al 2007). T cells, which affect B cell function, are present in the spinal meninges and near activated microglia in the normal-appearing white matter of spinal cords. T cells correlate with 25% atrophy in multiple sclerosis spinal cords (Androdias et al 2010). These regions are potentially difficult to reach with multiple sclerosis therapies. An alternate explanation for these germinal center-like areas is that CNS injury itself triggers systemic autoimmunity and local B cell activation (Ankeny et al 2006).
As the plaque ages, its inflammation and edema partially resolve. The relative number of B cells, CD8 cells, and monocytes increases, compared to CD4 cells (Booss et al 1983; Lassmann et al 1994). In chronic inactive plaques, CD8 cells are 10 times more numerous than CD4 T cells. There is a distinct plaque margin, with a residue of occasional inflammatory cells, myelin-laden macrophages, a glial scar, and damaged and demyelinated axons. In late chronic plaques, inflammation is minimal and comparable to other neurologic disease controls, macrophages predominate, and plasma cells and mast cells are present.
Mast cells are typically missed by usual histological stains, but with specific stains they are seen in chronic active lesions (Ibrahim and Reder 1996). Mast cells may be a consequence of any kind of chronic inflammation and not specific to multiple sclerosis. Mast cells could enhance migration through the blood-brain barrier, activate Th1 cells, reduce Treg function with secreted histamine, and release destructive or neuroprotective molecules. On microarrays, there is high expression of mRNA for “allergic” molecules such as prostaglandin D synthase, histamine receptors, immunoglobulin Fc-epsilon receptor, tryptase, and chemokines (CCL5, stem cell factor) (Pedotti et al 2003; Couturier et al 2008).
RNA profiling of plaques, usually post-mortem, shows differences from normal white matter. The signature is of neuroprotection, anti-oxidative stress (inflammatory and anti-inflammatory), and mitochondrial de-activation, with nearby glial and astrocytic activation. Proteomic analysis shows activation of tissue factor and other coagulation molecules. Conclusions are difficult because of varying damage to a mix of cells and heterogeneity of lesion activity and age.
Brain pathology in primary progressive multiple sclerosis differs from that in relapsing-remitting disease. Spinal cord lesions predominate and cause gradual paraparesis. Less inflammation is reflected by fewer Gd-enhancing lesions. However, myelin is pale (“dirty” MRI) in the “normal-appearing” white matter. Cortical atrophy and demyelinated plaques in deep gyri of the cerebral cortex, insula, cingulate, limbic circuit, and cerebellar cortex is much more severe than in relapsing-remitting multiple sclerosis (Lassmann et al 2007). Extensive hippocampal demyelination in chronic multiple sclerosis is likely to interfere with cognition (Geurts et al 2007). Subpial germinal center-like areas may contribute to cortical damage. N-acetyl aspartate levels are low in the cortical gray (Sastre-Garriga et al 2005). Urine myelin basic protein-like material is lower than in secondary progressive multiple sclerosis, suggesting a slower rate of destruction in primary progression.
The brain pathology in multiple sclerosis is not stereotypical. The MRI ranges from small lesions in the white matter to huge plaques that are sometimes mistaken for gliomas. Large solitary demyelinating lesions in the centrum semiovale are often biopsied. These large lesions, even if associated with multiple plaques, sometimes have good prognosis (Kepes 1993).
Distinct patterns in different brains, but similar within a given brain, appear in biopsies of large lesions and at autopsy (Lucchinetti et al 2000). Pathological subtypes depend on the degree of inflammation, myelin destruction, and oligodendroglial preservation. In each case, the number of macrophages is 10-fold greater than T cells, that are themselves 10-fold more numerous than B cells.
Four pathological subtypes are described in this table:
This table shows the 4 subtypes of multiple sclerosis pathology in the brain. (Contributed by Dr. Anthony Reder.)
I. T cell and macrophage-mediated demyelination (18% of 201 patients)
II. T cell and macrophage, plus antibody-induced or complement-mediated demyelination (56%)
III. Oligodendrocyte dystrophy and apoptosis with myelin protein dysregulation (24%)
IV. Primary oligodendrocyte degeneration with features similar to viral, ischemic, or toxic oligodendrocyte damage (2%)
Patterns I and II are seen in acute, early active multiple sclerosis. An intense perivenous immune reaction causes a sharply demarcated area of demyelination and destruction of oligodendroglia, astrocytes, and axons. There is preservation of oligodendrocytes and significant remyelination (shadow plaques) and less expression of multiple myelin proteins, without (pattern I) or with deposition of activated complement and IgG (pattern II). Many myelin proteins are decreased, but myelin-associated glycoprotein is not lost. Oligodendrocytes die at the plaque edge, but they reappear in the plaque center.
Antibody and complement-facilitated pattern II is the most common. Most antibodies in multiple sclerosis plaques are “nonsense” antibodies to unknown determinants and their relevance is unknown. They usually do not react with myelin antigens. Some, especially IgM, may stimulate remyelination. Others are probably pathogenic, ie, antibodies to gangliosides (above) and IgM against myelin and S-nitrosocysteine (from nitric oxide reactants, some directed against myelin-associated glycoprotein on oligodendrocytes) (Boullerne et al 2002). Complement binding to antigens increases their immunogenicity. Nonetheless, myelin damage in pattern II appears to be macrophage-mediated. Plasma exchange appears to benefit pattern II, but not patterns I and III.
Patterns I and II are similar to the lesions of experimental allergic encephalomyelitis in which there is an autoimmune attack against myelin. Pattern I resembles destruction of myelin by macrophage products (tumor necrosis factor-alpha and reactive oxygen species). Pattern II is similar to experimental allergic encephalitis induced by myelin oligoglycoprotein, mediated by T cells interacting with anti-myelin oligodendrocyte glycoprotein antibodies.
Patterns III and IV exhibit primary oligodendroglial dysfunction with subsequent demyelination. Pattern III consists of an inflammatory infiltrate of macrophages, microglia, and T cells, but no immunoglobulin. Ill-defined, non-perivenous areas of demyelination (preservation of oligodendroglia near venules) and limited remyelination are seen, sometimes with concentric rings of demyelination reminiscent of Balo concentric sclerosis, “dying back” destruction and apoptosis of oligodendrocytes, and a marked fall in myelin-associated glycoprotein compared to other myelin proteins. Myelin-associated glycoprotein is needed for myelin attachment to axons, and possibly in remyelination, and is located in distal periaxonal oligodendrocyte processes (Chang et al 2002). This pattern of demyelination resembles acute white matter hypoxia and suggests a virus or toxin such as nitric oxide that could interfere with mitochondrial energy production.
Pattern IV consists of an inflammatory perivenous plaque with a sharp border of destruction and apoptotic loss of oligodendroglia with little remyelination. This rare pattern is seen only in some patients with primary progressive multiple sclerosis. It may reflect an underlying dysfunction in oligodendroglia (oligo-opathy) (Gulcher et al 1994).
Patterns I, II, and III are seen in acute, relapsing-remitting, and secondary progressive multiple sclerosis. Patterns I, II, and IV are seen in progressive multiple sclerosis.
All active plaques throughout a given brain have a similar histopathological subtype, suggesting a consistent style of immune and brain response at the time of autopsy or biopsy. The pattern of MRI lesions is also similar in a given brain (Lucchinetti et al 2000). Differences in pathology between patients suggest there is heterogeneity in the pathogenesis of the disease, a fundamental difference in the mechanism and targets of demyelination, and probably in therapeutic responses. For instance, agents that modify cellular immunity (eg, interferons) are theoretically best for subtype I. Plasmapheresis or intravenous immunoglobulin might be of benefit in antibody-mediated subgroup II. Growth factors for oligodendroglial progenitors or actual transplants are potential therapies for types III and IV.
In a smaller series, there were combinations of different categories in the same brain, such as pattern IV plus other plaques with remyelination (ie, patterns I or II) (Barnett and Prineas 2004). They also suggest in some cases oligodendroglial apoptosis may precede inflammation. In contrast, in late “established” multiple sclerosis all lesions show complement and antibodies associated with macrophages in areas of active demyelination, suggesting that heterogeneity disappears over time (Breij et al 2008).
MRI versus histopathological subtypes, clinical symptoms, and behavior in trials. MRI is important in determining extent of brain and cord lesions, presence of new lesions, atrophy, and possibly responses to therapy (Lassmann 2008).
Perivascular inflammation is associated with gadolinium-enhancing lesions on MRI. Smaller nodular lesions enhance from the center outward; ring-shaped lesions enhance centripetally over 30 minutes (Gaitan et al 2011). Ring-enhancing lesions on MRI are areas of new inflammation, consisting largely of a sharp border of macrophages that secrete TNF-alpha, some T cells, oligodendroglia with DNA fragmentation, and axonal loss. This ring surrounds older lesions and is characterized by protein leakage (blood-brain barrier breakdown), isointense T1, and hyperintense T2 MRI (Bruck et al 1997). T2 activity persists for 10 weeks after contrast enhancement, suggesting it measures degeneration and repair. Small T2 lesions are disproportionately more damaging than large ones (Meier et al 2007). Ring-enhancing lesions correspond to pattern I and II lesions described above. Demyelinated or remyelinating lesions have less inflammation, significant axonal loss, and modest blood-brain barrier breakdown. They are hypointense on T1 (less so with remyelination) and hyperintense on T2, and are variably enhancing (Bruck et al 1997). T2 signal is from edema; T1 hypointensity is from axonal loss, myelin loss, edema, and widening of the extracellular space.
Although MRI films are a dramatic way to show CNS lesions to patients, there are caveats for using MRI as a biological marker for multiple sclerosis. The T2 edema signal alone can’t differentiate between demyelinated and partially myelinated lesions. Two of nine T2 MRI lesions show no demyelination on postmortem analysis (Barkhof et al 2003). Lesions in many parts of the brain are clinically silent. Correlation between T2 lesions and clinical symptoms is poor (r = 0.2 to 0.3; less than 6% of the variance) (“clinico-tomographic dissociation”). In 1354 placebo-treated relapsing-remitting patients from 45 clinical trials and natural history databases, T2 total lesion load did not predict change in disability from baseline to trials’ end (Daumer et al 2009). There was a small predictive effect of total lesion load on disability in secondary progressive multiple sclerosis (r = 0.21). Gd+ lesions did not predict relapses. T1 black holes, a measure of lost axons, correlate well with spinal cord atrophy and also with clinical deterioration in secondary progressive multiple sclerosis (r = 0.8) (Barkhof 1999), but not in relapsing-remitting multiple sclerosis (r = 0.3) (Simon et al 2000). Glatiramer acetate and IFN-beta reduce the chance that black holes will become permanent (Filippi et al 2001; 2011), and IFN-beta prevents them ab initio.
Gd+ lesions are most common early in the course of multiple sclerosis and in relapsing forms of multiple sclerosis compared to primary progressive disease. Multiple new and reactivated old Gd+ lesions appear in concert during disease activity. Occasional new T2 lesions may arise without enhancement, especially in periventricular areas (Lee et al 1999). Two or more Gd+ lesions strongly predict the development of multiple sclerosis (96%) after an isolated clinical attack (Group CHAMPS 2002). Long-duration Gd+ lesions are most likely to evolve into a hypointense T1 MRI lesion. However, Gd-enhancing lesions are only modest predictors of a worse clinical course. Changes in therapy must be made in the context of clinical patterns and not simply based on Gd+ lesions. After Gd-positive scans in untreated patients, the number of contrast-enhancing lesions falls at 3 months by 4%, at 6 months by 29%, and at 9 months by 48% (Zhao et al 2008). Cholesterol increases by 4.4% for each gadolinium-enhancing MRI lesion (Pantano et al 2001).
Brain regions differ in Gd enhancement. Cortical gray matter lesions are difficult to see on T2-weighted MRI, likely because immune responses and edema are influenced by less myelin, little water, and 10 times fewer inflammatory cells than in white matter lesions (Peterson et al 2001). This may explain why plaques enhance in subcortical U fibers but do not extend into the gray matter on T1 MRI. A single plaque often enhances in the white mater portion but not in the gray matter, forming an “open ring”—highly specific for a demyelinating lesion. Cortical lesions are rare on T2 MRI, but are sometimes seen with FLAIR (fluid-attenuated inversion recovery) (Bakshi et al 2001). Double inversion recovery MRI is more sensitive and detects cortical lesions in over 80% of primary progressive multiple sclerosis brains (Calabrese et al 2009). MRI with 8 Tesla magnets easily demonstrates gray matter plaques (Bakshi et al 2005). Even with minimal inflammation, cortical neurons are injured, contributing to motor, sensory, and cognitive losses and possibly fatigue.
Lesion location sometimes determines clinical symptoms. Patients with “benign multiple sclerosis” and those with severe disability can have similar brain atrophy and N-acetyl aspartate content. The disabled group often has significant atrophy at the second cervical cord level (Brass et al 2004). Some patients with primary progressive multiple sclerosis have only diffuse MRI abnormalities in brain and cord (“dirty white matter”) (Zwemmer et al 2008). This represents ongoing inflammation and significant axonal pathology.
Functional MRI shows activation of wide areas of primary cortex and supplementary motor cortex. This technique measures blood flow to areas of brain involved in various tasks. The enlarged cortical area on functional MRI is presumably less efficient because plaques have disrupted normal connections, forcing cortical reorganization or unmasking of less efficient latent pathways.
The rate of brain atrophy is increased up to 10-fold in multiple sclerosis. Atrophy is caused by loss of neurons and axons, with some contribution from damaged oligodendroglia and myelin. Dehydration for 16 hours reduces brain volume by 0.55% and can confound measurement of atrophy. Studies do not account for possible diuretic effects of interferons. Gd+ lesions often do not predict brain atrophy (Saindane et al 2000), but are more predictive of future atrophy when they are ring-enhancing with central contrast pallor (Leist 2001) and when they are present at onset of multiple sclerosis (Simon et al 2000). T2 lesions do not predict cord atrophy (Bergers et al 2002). The intercaudate nucleus distance correlates better with loss of clinical function (brain atrophy versus disability, r = 0.67; versus cognitive function, r = -0.42). Cord atrophy correlates best with clinical disability and poor walking; deep gray atrophy correlates with slowed cognition. In early relapsing-remitting multiple sclerosis, gray matter atrophy on MRI is twice that of normal controls (Tiberio et al 2005). Gray matter atrophy includes cortex (especially in deep sulci), thalamus, and hippocampus (CA1 and subiculum). Hippocampal volume loss is associated with high cortisol and depression in multiple sclerosis. Cigarette smoking correlates with lower brain volume and trends with faster progression (Durfee et al 2008). Atrophy is potently slowed by IFN-beta, glatiramer, and natalizumab therapy.
Magnetic resonance spectroscopy detects constituents of neurons and glial cells. N-acetyl-aspartate is part of an osmoregulatory molecular water pump, synthesized by neurons (Baslow 2002). It is a marker of neuronal and axonal function but not necessarily of neuronal loss. N-acetyl aspartate concentrations are high in mast cells and present in oligodendroglia; this could confuse magnetic resonance spectroscopy readings of presumed neuron and axon integrity. Early in multiple sclerosis, “normal-appearing white matter” as well as thalamic and cortical gray matter have decreased N-acetylaspartate (Chard et al 2002; Filippi et al 2003). Decreased N-acetylaspartate correlates with the number of clinical relapses over the preceding 2 years (Parry et al 2003), suggesting that N-acetylaspartate forecasts prognosis even before T2 lesions are visible, and before detectable inflammation. Loss of N-acetyl aspartate precedes atrophy and strongly correlates with disability (Bjartmar and Trapp 2001), fatigue, lateralized cognitive dysfunction, and abnormal visual evoked potentials. A fall in cortical N-acetyl aspartate correlates with disability in primary progressive multiple sclerosis (Sastre-Garriga et al 2005). Periventricular N-acetylaspartate falls with progression and is lowest in secondary progressive multiple sclerosis (Matthews et al 1996), more than in primary progressive disease. Levels continue to fall in untreated patients, but rise back toward normal in affected areas after 6 months of IFN-beta therapy (Narayanan et al 2001). Thus, metabolic disturbances and axonal shrinkage may be reversible.
MRI using ultra-small particles of iron oxide (USPIO) can trace macrophage activity. Oligodendrocyte progenitor and hematopoietic stem cells can also be labeled and traced with these nanoparticles. Benzodiazepine receptor-labeled microglia on PET scans (with PK1195) (Banati et al 2000) show lesions not detectable on regular MRI.
Cytokines, chemokines, autoantibodies, and Th1/Th2/monocyte ratios vary between patients and over time, possibly explaining some of the differences in course, pathology, or MRI (Hickey 1999). For instance, IFN-gamma production by activated mononuclear cells is increased in multiple sclerosis patients compared to controls (Balashov et al 1998), but this does not correlate with MRI lesions (Killestein et al 2002b). Correlation of data on immune function, urine myelin basic protein (Whitaker et al 1995), and MRI subtypes versus clinical responses in drug trials is essential to define whether these subtypes can be used to determine prognosis or the best drug therapy.