Brain, Vol. 125, No. 12, 2591-2625,
December 2002
© 2002 Oxford University Press
Review Article |
Peripheral neuropathies and anti-glycolipid antibodies
1 Department of Neurology, Institute of Neurological Sciences, Southern General Hospital, Glasgow, UK and 2 Department of Neurology, Dokkyo University School of Medicine, Tochigi, Japan
Correspondence to: H. J. Willison, University Department of Neurology, Institute of Neurological Sciences, Southern General Hospital, Glasgow G51 4TF, UK E-mail: h.j.willison{at}udcf.gla.ac.uk
Received March 8, 2002. Revised June 18, 2002. Accepted July 5, 2002.
 |
Summary |
|---|
 Top Summary IntroductionGlycosphingolipid structure and...Detection methods for...Ganglioside localization to...Measurement of anti-glycolipid...Acute clinical syndromesMiller Fisher syndrome and...Origins and immunological...Antecedent events as inducers...Physiological models of anti...Experimental neuropathies...Therapeutic considerationsConclusionsReferences |
|---|
This review charts the progress of anti-glycolipid antibodies in neuropathy, from their original discovery 20 years ago in immunoglobulin M paraproteinaemic neuropathy through to current discoveries mapping their relationship to subtypes of Guillain–Barré syndrome. Antibodies to >20 different glycolipids have now been associated with a wide range of clinically identifiable acute and chronic neuropathy syndromes. Particular progress has been achieved in understanding the link between acute motor axonal neuropathy and antibodies to GM1, GD1a, GM1b and GalNAc-GD1a, and between the cranial, bulbar and sensory variants of GBS and antibodies to the disialylated gangliosides GQ1b, GT1a, GD1b and GD3. In addition to clinical and serological studies, the origins and measurement of anti-glycolipid antibodies and their relationships to similar carbohydrate structures on infectious organisms, particularly Campylobacter jejuni, are discussed in the context of a molecular mimicry hypothesis. The structure and nomenclature of relevant glycolipids are outlined, along with information on their localization in nerve, and the influence this has on clinical phenotypes. Major advances have been made in animal modelling of anti-glycolipid antibody-associated diseases, both in vitro and in vivo. This has advanced our understanding of the role of anti-GQ1b antibodies in Miller Fisher syndrome with particular respect to the motor nerve terminal as a potential site of injury, and led to the creation of rabbit models of anti-GD1b and anti-GM1 antibody-mediated sensory and motor neuropathy, respectively. With such information in place, it will now be possible to determine the precise mechanisms by which antibodies injure the different compartments of peripheral nerve and establish how a range of immunomodulating therapies, including current treatments, exert their therapeutic effects. Despite these very significant advances, considerable gaps in our knowledge persist, and it is likely that other pathogenic pathways operate in inflammatory neuropathy that are unrelated to glycolipid antibodies, although these are outside the scope of this review.
Keywords: autoantibody; autoimmunity; gangliosides; Guillain–Barré syndrome; peripheral neuropathy
Abbreviations: AIDP= acute inflammatory demyelinating polyneuropathy; AMAN = acute motor axonal neuropathy; AMSAN = acute motor and sensory axonal neuropathy; CANOMAD = chronic ataxic neuropathy, ophthalmoplegia, IgM paraprotein, cold agglutinins and disialosyl antibodies; CMV = cytomegalovirus; DRG = dorsal root ganglion; GBS = Guillain–Barré syndrome; Ig = immunoglobulin; LPS = lipopolysaccharide; MAG = myelin-associated glycoprotein; MFS = Miller Fisher syndrome; NMJ = neuromuscular junction; SGLPG = sulfated glucuronyl lactosaminyl paragloboside; SGPG = sulfated glucuronyl paragloboside
|
Introduction |
|---|
|
TopSummaryIntroductionGlycosphingolipid structure and...Detection methods for...Ganglioside localization to...Measurement of anti-glycolipid...Acute clinical syndromesMiller Fisher syndrome and...Origins and immunological...Antecedent events as inducers...Physiological models of anti...Experimental neuropathies...Therapeutic considerationsConclusionsReferences |
|---|
The identification and characterization of neural autoantigens of pathogenic significance has been achieved successfully for a small number of disorders, including several neuromuscular junction (NMJ) diseases, but has remained elusive in major areas of clinical neuroimmunology, particularly for demyelinating diseases. Despite the ability to generate antigen-specific animal models of peripheral demyelinating diseases, such as myelin protein-induced experimental allergic neuritis, these models in general have not led to the identification of clinically applicable disease markers, although recent progress in this area is being achieved (Hughes et al., 1999
; Gabriel et al., 2000; Ritz et al., 2000; Kwa et al., 2001) One area in which considerable progress has been made is in the relationship between anti-glycolipid antibodies and neuropathy, a field of research that arose primarily from clinical–serological observations, rather than experimental studies. Thus, the identification of anti-glycolipid antibodies in large cohorts of patients with peripheral neuropathy, and their association with particular clinical phenotypes, both topographical and fibre type specific, has uncovered a wealth of clinical–serological associations. Speculation about their role in pathophysiology provides a rationale for laboratory-based research. Ironically, the reverse problem to that traditionally encountered in neuroimmunological research now exists, i.e. the difficulty in translating the human serological data into experimental models of neuropathy induced by anti-glycolipid immune responses. Despite this very significant progress in our understanding of anti-glycolipid antibody-mediated neuropathy, both gaps and inconsistencies in our knowledge persist. Some opinion holds that other pathogenic pathways operate in inflammatory neuropathy that are unrelated or even contradictory to a model based on anti-glycolipid antibodies. Identifying such pathways remains a major challenge for investigators and, where relevant, these data are discussed in outline.
An early impetus to the search for nerve autoantigens arose when it was recognized that acquired polyneuropathies occurred in association with benign monoclonal gammopathies; this now forms an important subset of predominantly late-onset neuropathy (Kyle, 1992; Quarles, 1997; Ponsford et al., 2000). It seemed hypothetically likely that the monoclonal paraprotein might have anti-neural activity, and the first such antigen to be identified was the myelin-associated glycoprotein (MAG). Stemming from this observation, it transpired that the antigen specificities of the paraproteins frequently were directed to carbohydrate determinants present on different glycolipids distributed in neural tissue, in addition to glycoproteins such as MAG. The first clinical–serological association to be studied in detail was the immunoglobulin M (IgM) paraproteinaemic neuropathy with reactivity against MAG and the cross-reactive glycolipids, sulfated glucuronyl paragloboside (SGPG) and its higher lactosaminyl homologue, sulfated glucuronyl lactosaminyl paragloboside (SGLPG) (Latov, 1994; Chassande et al., 1998; Quarles and Weiss, 1999). Chronic motor neuropathies were then identified in association with polyclonal or monoclonal IgM antibodies directed to GM1 and other Gal (ß1–3) GalNAc-bearing glycolipids including GD1b and asialo-GM1: these are now known to be present in 
50% of cases of multifocal motor neuropathy with conduction block (Kornberg and Pestronk, 1995), depending upon the clinical definition and detection methodology used (Leger et al., 2001). In 1985, the first case of IgM paraproteinaemic neuropathy in which the paraprotein reacted with NeuAc (
2–8) NeuAc (2–3) Gal-configured disialylated gangliosides including GD1b, GD3, GD2 and GT1b was reported (Ilyas et al., 1985), and many further cases of this syndrome have now been described (Willison et al., 2001).
Although originally described in association with paraproteinaemic neuropathies, the main recent clinical and research impetus on the role of antibodies directed at glycolipid epitopes has been in the context of the acquired inflammatory neuropathy, Guillain–Barré syndrome (GBS). Anti-ganglioside antibodies were first found in cases of GBS in 1988 (Ilyas et al., 1988b). Early studies were performed on the basis that anti-glycolipid antibody had been identified in patients with chronic demyelinating polyneuropathies and multifocal motor neuropathies. Our knowledge of anti-glycolipid antibodies in chronic neuropathies is extensive, and it transpired that there were many interesting parallels with GBS. A wealth of new information covering different aspects of this area is now available. Thus, antibodies to a wide range of glycolipids including GM1, GM1(NeuGc), GM1b, GalNAc-GM1b, GD1a, GalNAc-GD1a, GD1b, 9-O-acetyl GD1b, GD3, GT1a, GT1b, GQ1b, GQ1b, LM1, galactocerebroside and SGPG have been reported in >200 papers on GBS and other inflammatory neuropathies, as case reports and in larger series.
Placing this literature into a clinical and pathophysiological framework is complex, and several points need emphasizing. Firstly, anti-ganglioside antibodies assays, on which much of these data are based, are technically capricious, are not served by a uniform supply of glycolipid reagents and have high inter-laboratory variation. Secondly, the epidemiological patterns of anti-ganglioside antibodies may vary substantially between geographic regions, according to the prevalent subtypes of GBS and their relationship to preceding infections, such as Campylobacter jejuni infection. Thirdly, host susceptibility factors controlling the immune response to glycolipid epitopes that are related specifically to individuals and/or populations with particular genetic or environmental backgrounds may be present. In much of the anti-ganglioside antibody literature, these factors are not controlled in a systematic fashion. Some of the most significant publications in this area have come from anti-ganglioside antibody analysis of sera collected as part of carefully controlled clinical studies and trials (Rees et al., 1995a; Jacobs et al., 1996; Hadden et al., 1998; Ang et al., 1999; Yuki et al., 2000a), rather than random ascertainment of sera from affected patients. Despite these and other caveats, very significant progress is being made, and some clear disease patterns have emerged, and others are still emerging.
|
Glycosphingolipid structure and nomenclature |
|---|
|
TopSummaryIntroductionGlycosphingolipid structure and...Detection methods for...Ganglioside localization to...Measurement of anti-glycolipid...Acute clinical syndromesMiller Fisher syndrome and...Origins and immunological...Antecedent events as inducers...Physiological models of anti...Experimental neuropathies...Therapeutic considerationsConclusionsReferences |
|---|
Glycosphingolipids are composed of a ceramide (N-acylated sphingosine) attached to one or more sugars (hexoses) (Ledeen and Yu, 1982
). A selection of some of the clinically relevant structures is shown in Fig. 1. The hydrophobic ceramide is immersed in the lipid membrane and when the hydrophilic carbohydrate structure is exposed extracellularly, as is the case with plasma membranes, it is capable of acting as an autoantibody target. All known neuropathy-associated antibodies target this extracellular carbohydrate structure, rather than the ceramide moiety. One of the simplest glycosphingolipid structures of clinical relevance is galactocerebroside, comprising galactose linked to ceramide (monogalactosylceramide). Sulfatide is galactocerebroside sulfated on the third carbon of galactose. The term ganglioside refers to the large family of glycosphingolipids that contain sialic acid linked to the oligosaccharide core, synthesized through addition of monosaccharides in a stepwise fashion by glycosyltransferases and sialyltransferases. Gangliosides are present throughout the body but are very highly concentrated in the nervous system (Ledeen, 1985).
|
The nomenclature proposed by Svennerholm is used most widely and accepted for gangliosides of the ganglio-series (IUPAC-IUB Commission on Biochemical Nomenclature, 1977
; Svennerholm, 1994). The designations are based on the findings that brain tissue contained four major gangliosides with a ganglio-series tetraose chain of neutral sugars (i.e. asialo-GM1) that are sialylated in different positions, although it is now recognized that there are >100 structurally distinct gangliosides. Four gangliosides, GM1, GD1a, GD1b and GT1b, were designated to belong to the G1 series, where G stands for ganglio-series ganglioside. The four major gangliosides differ with regard to the number and position of their sialic acids, where M, D and T stand for mono-, di- and tri-sialosyl groups. Thus there are two disialosylgangliosides, GD1a and GD1b. Although ‘b’ normally is used to designate gangliosides with a disialosyl group attached to the internal galactose (so-called ‘b-series’ gangliosides), the term GM1b is used for the monosialosylgangliotetraosylceramide in which the sialosyl group is attached to the terminal galactose, in contrast to GM1a (normally referred to more simply as GM1) in which the sialic acid is on the internal galactose. When three sialic acids link to the internal galactose of the ganglioside, they are designated to belong to the c-series. Now that the biosynthetic pathway of gangliosides of the ganglio-series has been elucidated in large part, it is evident that this early description of the ‘a’-, ‘b’- and ‘c’-series predicted the crucial role of the sialyltransferases in ganglioside biosynthesis. Gangliosides lacking the terminal galactose, pre-terminal galactosyl-N-acetylgalactosamine or internal galactose are assigned the number 2, 3 or 4, respectively.
One major difference between the CNS and PNS is the abundance of neolacto-series gangliosides in the PNS that are localized mainly in myelin, as recently reviewed (Ogawa-Goto and Abe, 1998). There is still no general agreement about the nomenclature of the neolacto-series gangliosides. The term LM1 is used for the sialosylneolactotetraosylceramide, which is also known as sialosylparagloboside. LM1, GM1, GM3, GM2 and sialosyllactosaminylparagloboside (also known as Hex-LM1) are the major monosialosylgangliosides in human peripheral nerve, although only some of these have been identified as dominant autoantigens in peripheral nerve diseases. SGPG and its higher homologue SGLPG have structures similar to that of LM1, except for a 3-sulfated glucuronic acid instead of sialic acid on the terminal saccharide chain. SGPG and SGLPG were discovered as a direct result of studying anti-MAG IgM paraproteins from patients with chronic polyneuropathy (Ariga et al., 1987). In man, most ganglioside sialic acid is in the N-acetyl form, as opposed to the N-glycolyl configuration that is common to many other species.
|
Detection methods for gangliosides in tissues: benefits and limitations |
|---|
|
TopSummaryIntroductionGlycosphingolipid structure and...Detection methods for...Ganglioside localization to...Measurement of anti-glycolipid...Acute clinical syndromesMiller Fisher syndrome and...Origins and immunological...Antecedent events as inducers...Physiological models of anti...Experimental neuropathies...Therapeutic considerationsConclusionsReferences |
|---|
When considering the pathogenic relationship between the presence of an antibody and neuropathy, it is clearly important to have detailed knowledge of the glycosphingolipid composition and distribution within the PNS in both humans and experimental animals and furthermore, in species from which gangliosides are purified for experimental and diagnostic use. Additionally, studies using autoimmune neuropathy sera may continue to identify additional glyco sphingolipids that have yet to be identified as autoantigens, as was the case for SGPG and SGLPG; these will be aided by further development of purification and analysis methods (Hirabayashi et al., 1988
; Taki and Ishikawa, 1997). These issues are not straightforward since the regulation of ganglioside expression, their analysis and purification are complex. Gangliosides are developmentally regulated and spatially segregated, varying between different peripheral nerve fibre types, and between different species. A complete map of the ganglioside composition of human nerves, and a comparison between species used for experimental modelling, would be a valuable resource, although this may still have limitations. As indicated above, the pattern of anti-ganglioside activity detected in a patient’s serum correlates to some extent with the clinical pattern of neuropathy, suggestive of a differential distribution of target gangliosides throughout the PNS.
The accessibility of gangliosides to circulating antibodies, being protected in their neural environment by the blood– nerve barrier and other factors, is important (Lloyd et al., 1992). One explanation for the lack of CNS involvement in anti-glycolipid antibody-associated neuropathy, despite the wide distribution of gangliosides in the CNS, would thus be the protection from autoimmune attack afforded by the blood–brain barrier. The blood–nerve barrier also needs to be overcome in order to allow antibody access to normally cryptic sites. Blood–nerve barrier injury could be mediated by anti-ganglioside antibodies, as glycolipid antigens are expressed on intraneural microvascular endothelial cells, thereby potentially mediating the destruction or malfunction of the blood–nerve barrier (Kanda et al., 1994)
In addition to localization and accessibility, ganglioside function within a given structure may influence the nature and development of antibody-mediated injury (Tettamanti and Riboni, 1993; Wu and Ledeen, 1994; Yu and Ariga, 1998). The calcium-binding properties of gangliosides have been demonstrated in several model membrane systems, and it is possible that one function is to chelate extracellular calcium that may be of relevance to nerve terminal injury. Gangliosides aggregate in glycosphingolipid- and cholesterol-rich lipid membrane microdomains, termed functional rafts, where they interact with membrane proteins and modulate events such as signal transduction and receptor function (Simons and Ikonen, 1997; Stoffel and Bosio, 1997). Thus, the pathogenic role of anti-ganglioside antibodies is likely to depend not only on how they affect the number and distribution of gangliosides but also on the extent to which the target gangliosides are intimately involved in modulating neuronal function. Additionally, activated complement components may in turn affect the normal functioning of ganglioside raft-associated proteins. The relative contribution of these factors may vary from site to site, and between anti-ganglioside antibodies of differing reactivity.
The two approaches used to establish the anatomical distribution of gangliosides each have their merits and limitations. Biochemical analysis has been useful to identify significant differences in the ganglioside composition of different nerves and can reveal subtle differences in the overall lipid composition. However, this approach is limited by the pleiomorphic composition of tissue and the loss of information about microanatomical distribution. In such circumstances, the second approach, that of immunohistology or other in situ ligand-binding studies (e.g. using ganglioside-binding bacterial toxins such as cholera toxin), can reveal fine structural detail about ganglioside distribution at the cellular and subcellular level. For these studies, high quality reagents such as affinity-purified antisera or monoclonal antibodies are essential. Many anti-ganglioside antibodies are not monospecific but may cross-react with structurally similar gangliosides and other glycoconjugate antigens, making extrapolation of results to ganglioside localization difficult.
Biochemical and immunohistological approaches may expose gangliosides that normally occupy cryptic sites (e.g. within compact myelin) by homogenization or tissue sectioning, and can thus misrepresent the ganglioside array that would be visible to circulating antibodies in physiological environments. Furthermore, gangliosides can be distributed heterogeneously within a membrane, as in functional rafts into which proteins such as growth factor receptors or ion channels are specifically included or excluded (Simons and Ikonen, 1997). The antigen density and the surrounding lipid environment can also markedly influence the ability of anti-ganglioside antibodies to bind; thus, failure to detect a ganglioside by immunohistology does not necessarily indicate its biochemical absence (Lloyd et al., 1992). A high local concentration of ganglioside may allow for good immunohistological detection, whereas a ganglioside which is evenly distributed throughout a membrane may have the same total tissue concentration in biochemical evaluation, yet not be detectable by immunohistology. Interpretation of both biochemical and immunohistological studies thus requires caution.
|
Ganglioside localization to specific nerve sites and fibre types |
|---|
|
TopSummaryIntroductionGlycosphingolipid structure and...Detection methods for...Ganglioside localization to...Measurement of anti-glycolipid...Acute clinical syndromesMiller Fisher syndrome and...Origins and immunological...Antecedent events as inducers...Physiological models of anti...Experimental neuropathies...Therapeutic considerationsConclusionsReferences |
|---|
The simplest explanation for particular anti-ganglioside antibody-associated neuropathies being confined to a motor or sensory clinical phenotype is that the two systems are composed of different gangliosides, and this issue has been addressed in several experimental studies, including immunohistological analyses (Fig. 2). A comparison of total ganglioside composition of human spinal roots showed that GM1 (associated with antibodies in motor neuropathy) is relatively enriched in the ventral roots compared with the dorsal roots (Ogawa-Goto et al., 1992
). Similarly, the cranial motor nerves supplying the extraocular muscles contain particularly high contents of GQ1b, the ganglioside antigen associated with ophthalmoplegia (Chiba et al., 1997). However, from these studies, it is also apparent that key gangliosides are also present at sites unaffected by the disease process. Thus, the absolute tissue distribution of gangliosides is an insufficient explanation for the regional localization of the clinical pathology.
|
The dorsal root ganglion (DRG) is a particularly interesting site to consider in these respects, and illustrates some of the anomalies described above. Functional subpopulations of DRG neurones can be distinguished by the expression of lacto-series and globo-series carbohydrates that correlate with their peptide and enzymatic phenotype (Dodd et al., 1984
; Dodd and Jessell, 1985). It is very likely that ganglio-series antigens are also distributed selectively and that such differences may underlie the specific nature of sensory deficits. In the rodent DRG, the GM1 ligand cholera toxin B-subunit and anti-GM1 antibodies selectively identify a subset of DRG neurones. In contrast, cholera toxin and anti-GM1 antibodies bind the majority of neurones in the human DRG (O’Hanlon et al., 1996, 1998). Thus, GM1 is clearly present in the human DRG in abundance, yet the clinical phenotype of anti-GM1-associated antibody neuropathy is strikingly devoid of sensory features. Anti-GD1b antibodies are highly associated with sensory neuropathy and bind to the vast majority of DRG neurones (Kusunoki et al., 1993b; Maehara et al., 1997). Disialosylated gangliosides (including GD1b, GT1b, GQ1b, GD3 and GD2) are prominent gangliosides in cultured DRG neurones and these can be lysed by anti-disialosyl antibodies (Ohsawa et al., 1993).
The node of Ranvier is another key site of injury in autoimmune neuropathy. Immune attack directed at antigenic determinants located at the paranodal Schwann cell surface may lead to paranodal demyelination, whereas antigens targeted on the exposed axolemma may result in axonal degeneration, both of which would result in conduction failure. Ligand-binding studies have suggested that GM1, GD1b and polysialosylated gangliosides are enriched in the paranodal myelin loops of peripheral nerve (Fig. 2). In the oculomotor nerves affected in Miller Fisher syndrome (MFS), GQ1b is particularly enriched at nodes of Ranvier (Chiba et al., 1993). With respect to the neuronal components of the node of Ranvier, GM1 is present on the cytoplasmic surface of motor neurones (Corbo et al., 1992). Toxin- and antibody-binding studies have identified gangliosides on paranodal and internodal axolemma (Ganser et al., 1983; Ganser and Kirschner, 1984; Corbo et al., 1993) and the adaxonal membrane (Molander et al., 1997). Similarly, an antibody reactive with disialosylated gangliosides has been shown to bind to internodal axolemma and/or adaxonal Schwann cell cytoplasm (Willison et al., 1996). Antibodies to GD1a are associated with pure motor axonal neuropathy, and preferentially stain ventral root axons in comparison with dorsal root axons, indicating a good correlation between ganglioside localization and phenotype in this example (Gong et al., 2001).
The presynaptic NMJ recently has been considered a potential target vulnerable to autoimmune attack in GBS, with some clinical justification and also for a variety of hypothetical reasons. First, it lacks a blood–nerve barrier, thereby readily allowing access to circulating autoantibodies. Secondly, it is the site for other paralytic antibody-mediated diseases including myasthenia gravis and Lambert–Eaton myasthenic syndrome. Thirdly, it is rich in gangliosides including GQ1b, GM1 and GD1a. Fourthly, it is the binding site for a wide range of bacterial toxins that also use gangliosides as ectoacceptors (Willison and Kennedy, 1993). In particular, cholera and tetanus toxins are readily taken up into nerve terminals, loaded into synaptic vesicles and ultimately transported back to the motor neurone cell body (Wan et al., 1982; Hirakawa et al., 1992). As a result of this property, enzymic conjugates of cholera toxin are used frequently as retrograde neuronal markers. As expected, histological analyses have demonstrated cholera toxin and anti-GM1/GD1b antibody binding to the NMJ (Latov et al., 1988; Schluep and Steck 1988; O’Hanlon et al., 1998). Antibodies reactive to polysialylated gangliosides also bind to the NMJ (Fig. 2). Some of the -series gangliosides specific to cholinergic neurones (Chol-1 antigens) are also expressed at the mature NMJ which may make them potential targets for autoantibodies (Derrington and Borroni, 1990).
|
Measurement of anti-glycolipid antibodies in serum |
|---|
|
TopSummaryIntroductionGlycosphingolipid structure and...Detection methods for...Ganglioside localization to...Measurement of anti-glycolipid...Acute clinical syndromesMiller Fisher syndrome and...Origins and immunological...Antecedent events as inducers...Physiological models of anti...Experimental neuropathies...Therapeutic considerationsConclusionsReferences |
|---|
In parallel with clinical developments, there has been a recent widespread increase in the use of anti-ganglioside antibody assays as both diagnostic and research tools for studying autoimmune peripheral neuropathies (Adams et al., 1991
; Kornberg and Pestronk, 1994; Willison, 1994; van Schaik et al., 1995; Taylor et al., 1996). Whilst this is highly desirable, it has also led to some inconsistencies in methodological approaches. Assays of anti-ganglioside antibodies present technical difficulties, with variables including antigen source and purity, timing of clinical sampling, details of assay method and definition of normal ranges for serum titres. The screening method used by most laboratories is an enzyme-linked immunosorbent assay, and many factors that influence this assay and contribute to the inter- and intra-laboratory variation have been identified (Ravindranath et al., 1994; Willison et al., 1999). Some attempts to recommend standard methodology have been made, although most laboratories have established local in-house immunodetection protocols based on elements of previously published assay methods (Ben Younes-Chennoufi et al., 1992; Bansal et al., 1994; Bech et al., 1994). In two multicentre comparative studies, in which investigators used local methodology, there was agreement on clearly positive or negative cases but variable results with intermediate titre samples (Marcus et al., 1989; Zielasek et al., 1994). An additional technique for assaying sera for anti-glycolipid antibodies is thin-layer chromatography overlay (Figs 3 and 4); however, this is only carried out in specialized laboratories and thus not routinely available. One enzyme-linked immunosorbent assay protocol designed by a pan-European consortium that we commonly use has been described previously in detail (Willison et al., 1999). Some glycolipids, including SGPG and SGLPG, are not commercially available and thus cannot be assayed for with ease, unless diagnostic laboratories have access to glycolipid purification facilities. Antibodies to MAG, with which some anti-SGPG antibodies share reactivity, are normally sought by western blot of CNS myelin.
|
|
Acute clinical syndromes |
|---|
|
TopSummaryIntroductionGlycosphingolipid structure and...Detection methods for...Ganglioside localization to...Measurement of anti-glycolipid...Acute clinical syndromesMiller Fisher syndrome and...Origins and immunological...Antecedent events as inducers...Physiological models of anti...Experimental neuropathies...Therapeutic considerationsConclusionsReferences |
|---|
Guillain–Barré syndrome
A large body of data indicates that anti-glycolipid antibodies are present in the acute phase sera of a proportion of patients with GBS. Whilst this falls far short of providing proof of pathogenic involvement for such antibodies, the strength of these associations warrants continued investigation. Such findings should not undermine the possible relevance of other classes of antigens; indeed, efforts are also underway to identify other pathogenic factors, including T cell and antibody responses to myelin proteins and glycosaminoglycans (Pestronk et al., 1998
; Hughes et al., 1999). A number of factors need to be highlighted at the outset of this discussion. First, only a proportion of patients with GBS have identifiable anti-glycolipid antibodies. The associations of specific anti-glycolipid antibodies with definite clinical forms of GBS or variants are at best statistically significant and certainly not constant, as one would ideally expect to see if a specific antibody is indeed responsible for a specific clinical or electrophysiological presentation. The one situation where a clear and constant association exists is between anti-GQ1b/GT1a antibodies and MFS; however, other GBS variants such as acute motor axonal neuropathy (AMAN), or antibody specificities such as GM1, do not live up to this standard. Explanations for the reported differences in the clinical correlates of these antibodies are discussed in the foregoing sections.
The syndrome of symmetrical, rapidly evolving flaccid paralysis and areflexia described by Guillain, Barré and Strohl in 1916 (Guillain et al., 1916) has now been subclassified on the basis of distinct patterns of axonal and demyelinating forms of the disease. GBS formerly was considered to be a relatively homogeneous entity characterized by segmental demyelination in all cases. The most frequent pattern of GBS encountered in Europe and North America is that originally described as acute inflammatory demyelinating polyneuropathy (AIDP), characterized by demyelination and a variable degree of lymphocytic infiltration (Asbury et al., 1969). In severe cases, axonal degeneration may accompany the demyelination as a ‘bystander’ event.
Less frequently encountered in North America and Europe, but common in China (McKhann et al., 1993; Griffin et al., 1996a) and Japan (Kuwabara et al., 1998b), and probably other regions of the developing world, especially where C.jejuni infections are frequent, is the axonal pattern of GBS, in which primary axonal degeneration occurs with little or no demyelination. Feasby et al. (1986) first presented evidence to support the possibility that some cases of GBS might be due to primary motor and sensory axonal degeneration without preceding demyelination and that the target antigen might lie on the axon. Collaborative Chinese–American studies firmly established the presence of primary axonal GBS, and it is now recognized that the axonal patterns can be classified further into two groups, AMAN and acute motor and sensory axonal neuropathy (AMSAN) (McKhann et al., 1993; Griffin et al., 1996a). The principal clinical method for distinguishing the AMAN, AMSAN and AIDP patterns is electrodiagnostic, and clear criteria have been formulated for separating the phenotypes. One caveat that affects interpretation of anti-ganglioside antibody studies is that some patients with inexcitable nerves often cannot be classified into axonal or demyelinating groups.
Pathological findings in autopsy cases of AMAN showed axonal degeneration of the motor axons with little demyelination or lymphocytic infiltration. Early changes at the nodes of Ranvier of motor fibres are accompanied by the presence of IgG and complement deposits on the axolemma, and macrophage recruitment, the findings being strongly suggestive of highly selective antibody-mediated attack on axonal membranes, rather that T cell-mediated disease (Griffin et al., 1996b; Hafer-Macko et al., 1996a). Macrophages insert processes into the nodal gap, penetrating the overlying basal lamina of the Schwann cell, and enter the periaxonal space of the internode. The end stage of this process is interruption of motor axons, with degeneration extending as far up as the ventral root entry zone. Patients with such extensive Wallerian-like degeneration of ventral root fibres could recover only by regeneration, a process requiring very long periods of time and unlikely to be complete. Some patients with AMAN, however, recover quite rapidly, which suggests that transient loss of motor function may be related to antibody binding to nodes of Ranvier with subsequent blocking of conduction but without axonal transection, or be due to axonal injury in the very distal part of the motor nerve (Ho et al., 1997; Kuwabara et al., 1998 a). In AIDP, in contrast, the immune attack is directed against components of the Schwann cell abaxonal membrane and is accompanied by the more characteristic features of vesicular demyelination (Hafer-Macko et al. 1996b). With respect to the current consideration on anti-glycolipid antibodies, it is clearly crucial to understand how this might be related to the presence or absence of particular glycolipid antigens at these sites or, alternatively, other as yet unidentified classes of antigens.
GBS is a self-limiting disease, occurring 1–3 weeks after infection, with muscle weakness usually reaching a nadir within 4 weeks, followed by partial or complete recovery taking place over weeks to months; the long-term prognosis is dependent upon the site and extent of axonal injury. This temporal pattern of evolution and decay is suggestive of pathophysiology centred on a primary humoral immune response, and this is also consistent with the rate of recovery being accelerated by plasma exchange or intravenous immunoglobulin. The therapeutic effect of plasma exchange presumably is related to the removal or dilution of circulating factors, and some indirect evidence suggests that the critical factors are most likely to be IgG immunoglobulins (Thornton and Griggs, 1994). Plasma concentrations of cytokines are elevated in GBS patients during the acute phase of the illness, but because their circulating half-lives are only a few hours, the effect of plasma exchange on their plasma levels is likely to be short term. Complement depletion resulting from plasma exchange is also brief, for the same reasons. The respective half-lives of IgM and IgA are 5 and 6 days. In contrast, the half-life of IgG is 21 days, except for the IgG3 subclass (7 days) that is longer than that of other plasma proteins. Plasma IgG level may be reduced for up to 5 weeks following a course of plasma exchange. Autoreactive IgG, which binds to neural components and thereby activates complement and recruits macrophages, may thus be the most important factor to be removed during plasma exchange. The mechanisms of action of intravenous immunoglobulin in ameliorating GBS are widely debated but poorly understood, although they may include antibody neutralization through a number of mechanisms (Stangel et al., 1999).
In addition to historical models of putative T cell-mediated mechanisms, antibodies directed to peripheral nerve were long believed to participate in the development of GBS, but no target molecules for the autoantibodies were found until 1988, when Ilyas et al. (1988b) first reported serum antibodies to gangliosides in five of 26 patients. IgG antibodies in one patient reacted strongly with LM1 and its hexose analogue Hex-LM1. IgG from two other patients with GBS reacted with GD1b. IgM antibodies in sera from two other patients reacted with GD1a and GT1b, which have a shared terminal carbohydrate sequence. The antibody titres in these cases decreased with clinical improvement. In the same year, Inuzuka et al. (1988) reported that a patient with GBS following Mycoplasma pneumoniae infection had IgM antibody to Hex-LM1. On the basis of these early findings, the search for anti-ganglioside antibodies in GBS accelerated rapidly throughout the 1990s.
Acute motor axonal neuropathy
The first reports of anti-GM1 antibodies in GBS appeared in the early 1990s, around the time at which concepts of GBS were emerging that led to the AMAN and AIDP subclassifications. The literature surrounding anti-GM1 antibodies that are found in both AMAN and AIDP remains confusing for a large number of reasons. First, and perhaps most unappealing in our view although maintained by others, is the possibility that anti-GM1 antibodies are irrelevant to the development of either AMAN or AIDP, but solely exist in GBS serum as bystander or secondary events. Thus they are a variable linked to the disease, either through preceding infection or as a result of a secondary immune response to nerve injury, but are independent of its pathogenesis. There are some data to support this view (Press et al., 2001). Secondly, subcategories of anti-GM1 antibodies may exist that have not been fully elucidated. For example, some anti-GM1 antibodies may be GM1 monospecific whereas others may cross-react with other gangliosides, and these subcategories of anti-GM1 antibodies may correlate with disease subgroups. Thirdly, GM1 and related epitopes may exist in both myelin and axolemmal membranes in varying concentrations or configurations that can lead to preferential binding of antibody under different circumstances in different individuals. Thus some individuals may be more susceptible to myelin injury and others to axonal injury upon exposure to a particularly subcategory or class of anti-GM1 antibody. Furthermore, this may vary during the course of the disease. Thus, at a node of Ranvier for example, axolemmal GM1 may be ‘cryptically’ disguised early in the course of the disease, but become exposed for antibody binding following paranodal demyelination induced by anti-GM1 (or other antibody) binding to paranodally sited GM1. An illness that started as AIDP could then evolve into AMAN, or AIDP with secondary axonal injury. Clearly a large number of complexities can be introduced that confound these considerations. Thus we should consider the foregoing data with an open mind and in further studies attempt to control for as many variables as possible. This applies not least to the method, definition and timing of the clinical electrophysiological analysis and the serological analyses, on which much of this discussion rests and which in some studies have only been conducted on a single occasion.
The report in 1990 of two patients with AMAN subsequent to C.jejuni enteritis in whom high titres of anti-GM1 IgG antibodies were found during the acute phase of the illness was followed by many subsequent studies (Yuki et al., 1990). Walsh et al. (1991) then reported that 14 out of 95 patients (15%) with GBS had anti-GM1 antibodies, and that the predominant immunoglobulin class was IgG rather than IgM. Kornberg et al. (1994) also reported that anti-GM1 IgG antibodies were strongly associated with AMAN. Cor relations were sought between the presence of these antibodies and the prognosis in terms of long-term disability. In early studies, neither Enders et al. (1993) nor Vriesendorp et al. (1993) found a correlation between anti-GM1 antibody titres, C.jejuni infection and the severity, type (axonal versus demyelinating) or outcome of GBS. Rees et al. (1995a) showed that patients who were anti-GM1 antibody positive were more likely to have axonal degeneration and had less sensory disturbance than anti-GM1 antibody-negative patients. In a subgroup analysis of GBS cases from another large series of patients, it was also found that patients with anti-GM1 antibodies had a more severe neuropathy with predominantly distal weakness and no sensory involvement (Jacobs et al., 1996).
These findings were supported further by an electrophysiological analysis showing that anti-GM1 antibodies are more common in the patient groups with axonal injury or inexcitable nerves (Hadden et al., 1998). Some consideration has centred on the relative contribution of IgG and IgM antibodies to these clinical features. Four of 24 anti-GM1-positive patients reported by Rees et al. (1995a) had IgM class antibody alone, but this proportion was unknown in the study of Jacobs et al. (1997a). As outlined above, failure to take into account anti-GM1 antibody fine specificities and isotypes or identify the presence of other antibodies co-occurring in the same patients (such as anti-GD1a or anti-myelin protein antibodies) may confound the relationship between electrodiagnosis and the presence of anti-GM1 antibody. For example, in a study by Kuwabara et al. (1998b), the relationship between electrodiagnosis and anti-GM1 IgG or IgM antibody was investigated directly, and a significant association between axonal dysfunction and the presence of anti-GM1 IgG antibody, but not IgM, was shown. In this analysis of 34 GBS patients, 16 were anti-GM1 IgG-positive, 12 of whom were classified by electrodiagnostic criteria at the first examination as having AMAN or AMSAN, three as having AIDP, and one was unclassified. In three patients initially diagnosed as having AIDP, conduction slowing resolved within days, and one of the AIDP cases and three AMAN cases showed rapid recovery of distal compound muscle action potential amplitudes without evidence of concomitant denervation. The time courses of conduction abnormalities were distinct from those in anti-GM1 IgG-negative AIDP patients. Rapid resolution of conduction slowing and block, and the absence of slow components indicative of remyelination, suggest that the conduction failure might be caused by impaired physiological conduction at the nodes of Ranvier rather than segmental demyelination. Thus it is possible that reversible conduction failure, in addition to or in place of axonal degeneration, could account for some of the pathophysiological events occurring in the anti-GM1 IgG-positive GBS cases. In both cases, immune-mediated attack may occur on the axolemma of motor fibres, and some experimental and human pathological findings support this, as discussed below. The extent to which these interpretations are supported by other studies is hard to unravel, because of methodological variations, but there are some parallels. For example, in the study of Hadden et al. (1998), the proportion of patients initially classified as ‘demyelinating’ who later changed to ‘axonal’ was significantly higher in those with anti-GM1 IgG antibody (five of 27, 19%) compared with those without the antibody (one of 108, 1%; P = 0.0006), but there was no such relationship with anti-GM1 IgM antibody. These types of studies underlie the rather complex considerations that need to be controlled for if we are to unravel all the elements of anti-GM1 antibody-associated neural injury.
In addition to the antibodies directed against GM1 described above, Kusunoki et al. (1996a) found that antibodies to a minor monosialosylganglioside GM1b were present in 22 out of 104 GBS cases tested and that this was highly disease specific. Their observation was confirmed in other studies (Yuki et al., 1997a). In a further study of 132 patients with GBS by Yuki et al. (2000 a), 25 (19%) patients had anti-GM1b antibodies; these were IgM class in 14, IgG class in 15, and mixed isotype in four patients. The anti-GM1b antibody-positive cases, especially those with IgG class antibodies, had a distinct clinical pattern compared with the 107 seronegative cases in this series, showing more frequent serological evidence of preceding C.jejuni infection, and a more rapidly progressive, severe and predominantly distal weakness with slow recovery. Furthermore, cranial nerve and sensory deficits were less common in the patients with anti-GM1b antibodies.
Anti-GD1a IgG antibodies have also been detected in AMAN. This was first observed in two patients with severe axonal GBS in 1992 (Yuki et al., 1992c). In a larger series of 37 patients, a significant association was found between the presence of anti-GD1a IgG antibody and a poor clinical outcome, as manifested by prolonged artificial ventilation with poor recovery at 3 months (Yuki et al., 1993d). An autopsy in one case showed severe axonal degeneration and segmental demyelination of peripheral nerves, lymphocytic infiltration and marked central chromatolysis of the lower motor neurone cell bodies. Anti-GD1a antibodies have been found to be highly specific for GBS in other studies, especially when present at high titre (Carpo et al., 1996). In a large group of Chinese patients with GBS and appropriate controls, 24% of AMAN patients and none of the AIDP patients or control subjects had high titre anti-GD1a IgG antibodies. The anti-GD1a antibody was the most specific for AMAN among other anti-glycolipid antibodies tested (GM1, GD1b, asialo-GM1 and GQ1b), and in particular indicated that anti-GD1a IgG antibody was better able to discriminate between AMAN and AIDP than anti-GM1 antibody (Ho et al., 1999). An interesting feature of this study that supports some of the methodological considerations described above was the definition of criteria for what constitutes a positive result for anti-GD1a antibodies. Thus when using a cut off titre >1 : 100, 60% of AMAN versus 4% of AIDP patients had IgG anti-GD1a antibodies, whereas when using a cut off titre >1 : 1000, 24% of AMAN patients and none of the AIDP patients had IgG anti-GD1a antibodies. Thus the identification of a relationship amongst these features depends on how one uses the serological and electrophysiological criteria, with stricter criteria leading to more specific, but less sensitive results in this example.
Kusunoki et al. (1994) found that GalNAc-GD1a is yet another target molecule for serum antibodies in the AMAN variant of GBS, being detected in six out of 50 patients (12%), a finding subsequently confirmed by Yuki et al. (1996b). In a series of 147 cases, anti-GalNAc-GD1a antibodies were found to mark a distinct clinical pattern characterized by lack of cranial nerve involvement (87% versus 38%), distal-dominant weakness (80% versus 25%) and no sensory disturbance (73% versus 22%) (Hao et al., 1999). Another recent study by Ang et al. (1999) found that anti-GalNAc-GD1a antibodies could be detected in 19 out of 132 cases (14%) and correlated with antecedent C.jejuni infection, a rapidly progressive, more severe course with predominantly distal weakness and little sensory and cranial nerve involvement. Similar findings have been reported by Kaida et al. (2000).
As described above, anti-GM1, anti-GM1b, anti-GD1a and anti-GalNAc-GD1a IgG antibodies have been demonstrated in numerous studies to have a strong association with the AMAN pattern of GBS. In a recent extensive study of 86 consecutive Japanese GBS patients by Ogawara et al. (2000), electrodiagnostic criteria showed AIDP in 36% of the patients and AMAN in 38%. The most frequent anti-ganglioside antibodies were of the IgG class and against GM1 (40%), GD1a (30%) and GalNAc-GD1a (17%), all of which showed a strong association with AMAN. These relationships between clinical phenotype, anti-ganglioside antibody and antecedent infection are summarized schematically in Tables 1 and 2 .
|
|
AMSAN cases exhibit a slower recovery than AMAN, in addition to sensory fibre involvement, but the pathologies are very similar (Feasby et al., 1986
; McKhann et al., 1993; Griffin et al., 1996a). Moreover, both conditions may follow C.jejuni enteritis. Griffin et al. (1996a) proposed that AMAN and AMSAN are part of the spectrum of a single type of immune attack on the axon. In a study to investigate whether anti-ganglioside IgG antibodies could be used as immunological markers to differentiate AMAN from AMSAN, the frequencies of anti-GM1, anti-GM1b and anti-GD1a IgG antibodies were similar (Yuki et al., 1999b). These data suggest that AMAN and AMSAN share a common immunological profile and support the view that they form a spectrum, as proposed (Griffin et al., 1996a).
Acute inflammatory demyelinating polyneuropathy
Clinically, patients with AIDP present with flaccid paralysis and areflexia and usually have some sensory loss either symptomatically or on physical examination. Electro physiological testing typically reveals increased distal motor latencies and F waves accompanied by reductions in nerve conduction velocity, and temporal dispersion. AIDP was long presumed to be predominantly a T cell-mediated disorder. This presumption was based on the lymphocytic inflammation found on nerve biopsy in many cases (Asbury et al., 1969) and by analogy with the widely studied animal model, experimental allergic neuritis. Recently, the immunopathology of early and unusually well-preserved autopsy cases was evaluated and suggests that antibody-mediated injury may be more important than previously recognized, at least in some cases (Hafer-Macko et al., 1996b). In one highly informative pathological study on a patient who died 3 days after onset of symptoms, inflammation was scanty and only a few fibres had been completely demyelinated. Staining for complement activation demonstrated complement activation products on the outermost of these fibres which had early vesicular changes in the myelin sheaths, usually beginning in the outer lamellae of the sheath. The resulting pathological picture closely resembled the appearance of experimental conditions in which nerve fibres are exposed to anti-galactocerebroside antibody in the presence of complement (K. Saida et al., 1979; T. Saida et al., 1979). Thus, an attractive reconstruction based on this pathology is that an antibody directed against antigens on the outermost surface of the Schwann cell (the abaxonal Schwann cell plasmalemma) binds complement, resulting in sublytic complement activation and the development of transmembrane pore formation. Macrophages are recruited and also participate in the removal of damaged myelin. The nature of the antigen(s) on the abaxonal Schwann cell plasmalemma and within compact myelin that may be involved in directing the immune attack in AIDP to Schwann cells remains uncertain, and elusive. Some evidence suggests that anti-myelin glycolipid antibodies may be involved, along with antibodies to Schwann cell protein or carbohydrate determinants that are also expressed at the cell surface (Hughes et al., 1999)
Complement-fixing antibodies to peripheral nerve myelin have been detected in the serum of a high proportion of patients with GBS, using a sensitive assay that detects antibody binding to isolated human peripheral nerve myelin by fixation of the first component of complement C1 (Koski et al., 1985). In attempts to identify the myelin component, it was observed that some of the anti-peripheral nerve myelin antibodies in 12 GBS sera tested bound an unidentified neutral glycolipid of human peripheral nerve myelin and cross-reacted with Forssman antigen (Koski et al., 1989). However, in a follow-up study, anti-Forssman IgM and IgG antibodies could not be detected in GBS compared with controls (Ilyas et al., 1991). Whether this Forssman-like glycolipid can be verified as an autoantigen in sera from AIDP patients requires further investigation.
Two studies have suggested that galactocerebroside could be an autoantigen in GBS (Kusunoki et al., 1995; Hao et al., 1998), although in neither was the electrodiagnosis of all the anti-galactocerebroside antibody-positive cases described. In the report of Kusunoki et al. (1995), two of four patients with anti-galactocerebroside antibody were confirmed electrophys iologically as AIDP, but electrophysiological examinations were not performed in the others (S. Kusunoki, personal communication). In another study, the association of AIDP with anti-galactocerebroside antibody could not be shown (K. Susuki et al., unpublished observations). The interesting relationship between GBS and anti-galactocerebroside antibody following M.pneumoniae infection is described below.
Unlike CNS myelin, human peripheral nerve myelin contains LM1, Hex-LM1 and SGPG (Ogawa-Goto and Abe, 1998), and these glycolipids sensibly have been screened in a number of studies as potential antigens in AIDP. Inuzuka et al. (1988) described a single GBS patient who had IgM antibody to Hex-LM1, and Ilyas et al. (1988 b) also reported one GBS patient who had IgG antibody to LM1 and Hex-LM1. A more detailed follow-up study led by Ilyas et al. (1992) showed that 23% of GBS patients had anti-LM1 IgG, but none were found with anti-LM1 IgM antibodies. Fredman et al. (1991) detected anti-LM1 antibody in 58% of the GBS patients, but also observed such antibodies in 30% of their normal controls. In a further study from Japan, five out of 96 patients (5%) with GBS had high titres of anti-LM1 IgG, all of whom had AIDP (Yuki et al., 1996a). In another Japanese study, anti-LM1 IgG antibodies were detected in seven out of 140 patients (5%) with GBS, five of whom had AIDP but the other two were not classified (Yako et al., 1999). At odds with the view that there may be an exclusive relationship between anti-LM1 antibodies and AIDP, a study of 19 patients with AIDP and 21 patients with AMAN found two from each group with anti-LM1 IgG antibodies (Yuki et al., 1999b). In another study by Susuki et al. (2002), anti-LM1 IgG antibody was detected in only one patient with AIDP, whereas it was present in seven with AMAN and in one with AMSAN. Sera from the eight IgG anti-LM1-positive patients with AMAN/AMSAN also had IgG activity against the gangliosides GM1, GM1b, GD1a, GalNAc-GD1a, GD1b or GQ1b. Anti-LM1 IgG antibodies from the AMAN/AMSAN patients cross-reacted with other gangliosides, whereas IgG antibody from the AIDP patient was monospecific against LM1. Anti-LM1 IgG antibody, therefore, cannot be a marker of AIDP. Larger studies are needed to verify whether monospecific anti-LM1 IgG antibody could be a marker of AIDP.
With respect to anti-SGPG antibodies and AIDP, few positive data have been forthcoming despite strong hypothetical grounds for such a relationship. Ilyas et al. (1992) detected anti-SGPG IgG in 9% of GBS patients studied and anti-SGPG IgM in 15%. However, Yuki et al. (1996 a) failed to detect anti-SGPG IgG in any GBS patients, with low titres of anti-SGPG IgM being found in 29% of cases. Another peripheral nerve-enriched glycolipid is GM2, and this appears to have a special relationship with preceding cytomegalovirus (CMV) infection, as discussed in detail below. The electrophysiological pattern of GBS after CMV infection is demyelinating, and anti-GM2 IgM antibody does appear in some cases of CMV-associated GBS (Visser et al., 1996; Jacobs et al., 1997c). Acute CMV infection without GBS is also associated with anti-GM2 IgM (Yuki and Tagawa, 1998). Another issue to consider is that GM2 cannot be detected in human peripheral nerves by the standard immunohistochemical techniques using anti-GM2 IgM antiserum (O’Hanlon et al., 2000). These studies raise some doubts about the pathophysiological significance of anti-GM2 IgM antibody in AIDP associated with CMV infection.
|
Miller Fisher syndrome and related conditions |
|---|
|
TopSummaryIntroductionGlycosphingolipid structure and...Detection methods for...Ganglioside localization to...Measurement of anti-glycolipid...Acute clinical syndromesMiller Fisher syndrome and...Origins and immunological...Antecedent events as inducers...Physiological models of anti... |
|---|