Peripheral neuropathies are classified as primarily demyelinating or axonal. Microstructural alterations of the nodal region are the key to understand the pathophysiology of neuropathies with antibodies to gangliosides and the new category of nodo-paranodopathy has been proposed to better characterise these disorders and overcome some inadequacies of the dichotomous classification. Recently, the research in autoimmune neuropathies has been boosted by reports of patients carrying immunoglobulin G4 antibodies against paranodal axo–glial proteins with distinct phenotypes and showing loss of transverse bands, terminal myelin loop detachment, nodal widening and axonal loss. These patients have been classified up to now as chronic inflammatory demyelinating polyradiculoneuropathy but, in our opinion, better fit into the nodo-paranodopathy category because nerve injury is due to dismantling of the paranode, segmental de-remyelination is absent and the pathogenic mechanism is not inflammatory. Evidence from nerve conductions and electron microscopy studies in patients and mutant animal models can reconcile the apparent contrast between the electrophysiological ‘demyelinating’ features, explainable just by the paranodal involvement and the axonal pathology. These patients broaden the autoimmune nodo-paranodopathy category and re-emphasise the usage of the term that pointing to the site of nerve injury reminds specific pathophysiological mechanisms, reconciles contrasting electrophysiological and pathological findings, and avoids misdiagnosis and taxonomic confusion. In our opinion, the nodo-paranodopathy term more adequately classifies the peripheral nerve disorders due to an autoimmune attack directed and limited to the nodal region integrating the traditional classification of peripheral neuropathies.
- neurophysiol, clinical
- guillain-barre syndrome
Statistics from Altmetric.com
Peripheral nerve disorders are traditionally classified, on the basis of pathology and electrophysiology, as primary demyelinating or axonal. Microstructural alterations limited to the nodal region explain the pathophysiology of some peripheral neuropathies and the category of nodo-paranodopathy was initially introduced to describe the neuropathies with antigangliosides antibodies and with the intent to go beyond the dichotomous classification.1 This categorisation was later extended to include neuropathies of different aetiology (autoimmune, inflammatory, ischaemic, nutritional and toxic) in which the involvement of the nodal region is thought to be determinant in the pathogenesis.2
Recently, the research in autoimmune neuropathies has been boosted by reports of patients carrying immunoglobulin (Ig) G4 antibodies against proteins of paranodal junctions and specific phenotypes.3–5 These patients have been classified up to now as chronic inflammatory demyelinating polyradiculoneuropathy (CIDP) but, in our opinion, better fit into the nodo-paranodopathy category because the site of nerve injury is focused and limited to the paranode, true de-remyelination is absent and the pathogenic mechanism is non-inflammatory.
In this personal view, we consider the autoimmune neuropathies in which findings from neurophysiological studies, pathology and animal models indicate the involvement of the nodal region in the pathogenesis re-emphasising the usage of the nodo-paranodopathy concept.
The nodal region: from physiology to pathology
The myelinated axons are organised in distinct domains each characterised by specific molecular arrangements: nodes of Ranvier, paranodes, juxtaparanodes and internodes (figure 1A). The nodal region is a crucial evolutionary structure of the nervous system ensuring, by saltatory conduction, a fast transmission of impulses with the least expenditure of energy. At the node of Ranvier, the myelin is interrupted and the axolemma presents the highest density of voltage-gated sodium (Nav) channels. The nodes are flanked by paranodal junctions composed of three major proteins: contactin 1 (CNTN1) and contactin-associated protein 1 (Caspr1) on the axonal side, and neurofascin 155 (NF155) on the terminal myelin loops. The paranodal junction functions as an electrical and biochemical barrier restricting the mobility of ions and membrane proteins between the node and internode. At the juxtaparanodes, voltage-gated potassium (Kv) channels are anchored and clustered by contactin-associated protein 2 and transient axonal glycoprotein 1. Ganglioside GM1 and gangliosides with Gal(b1–3)GalNAc moieties are more concentrated on axolemma of node and paranode, Schwann cell microvilli and abaxonal myelin.6 7 It took almost eight decades, since the original description by Ranvier in 1871,8 to demonstrate that the nodes are the places where ionic currents generated action potentials for saltatory conduction,9 and only in the last two decades, the nodal region has been recognised as a possible site of specific autoimmune attack in peripheral neuropathies. Figure 1B summarises the autoimmune peripheral nerve disorders with antibodies against gangliosides and axo–glial proteins of the paranodal junctions along with the current evidence that these antibodies induce dysfunction/disruption limited to the nodal region.
Autoimmune nodo-paranodopathies: the premises
In the early 90s, the connection among an acute, motor, primarily axonal subtype of Guillain-Barré syndrome (GBS) with antecedent Campylobacter jejuni infection and antiganglioside antibodies was established.10 11 Pathological studies of patients with acute motor axonal neuropathy (AMAN) showed Wallerian-like degeneration with no demyelination and inflammation, and with the earliest changes consisting in lengthening of the node of Ranvier.12 13 Most importantly it was found that IgG and complement deposition at the nodes of Ranvier preceded the development of Wallerian-like degeneration.14 It was proposed that the binding of antibodies could be sufficient to impair nerve conduction resulting in paralysis with little structural changes and with potential rapid recovery.14
Nonetheless, for a long time, AMAN was assumed to be typified simply by axonal degeneration and its electrodiagnosis was based on the lack of demyelinating findings and low amplitudes of compound muscle action potential (CMAP).15 16 During the years, patients with GBS with anti-GM1 and -GD1a antibodies showing conduction block (CB) and/or conduction slowing that promptly resolved without the development of excessive temporal dispersion of CMAPs characteristic of de-remyelination were reported (figure 2A,B).17 18 This rapid recovery of CB and conduction slowing were supposed to be due by a temporary failure of conduction at the nodes of Ranvier possibly due to loss of Nav channels and named reversible conduction failure (RCF) to distinguish it from the classical demyelinating CB.17 In the common belief, slow conduction velocity is assumed to be a characteristic of a demyelinating process and in axonal neuropathies, conduction velocity is thought to be normal or slightly slow. However, block of Nav channels in humans by intravenous infusion of lidocaine or tetrodotoxin intoxication reduces conduction velocity, even reaching the demyelinating range,19 20 conceivably by increasing the rise time of the action potential and the internodal conduction time. The report of patients with AMAN with conduction failure progressing to axonal degeneration or with RCF and axonal degeneration in the same or different nerves21–23 ultimately established that AMAN is electrophysiologically identified by reversible failure of conduction and axonal degeneration and both could be ascribable to the same immune attack to the nodal axolemma.
At this point, the foundations for the concept of autoimmune nodo-paranodopathy were laid.
The contribution of animal models
AMAN is associated with anti-GM1 and anti-GD1a antibodies and in mutant mice lacking these gangliosides, the paranodal loops fail to attach to the axolemma with widening of the node, Nav channel clusters are broadened and juxtaparanodal Kv channels are mislocated to the paranode.24 Rabbits immunised with GM1 showed weakness; anti-GM1 antibodies that bound to the nodes of Ranvier and activated complement resulted in the formation of the membrane attack complex, destruction of Nav channels, paranodal detachment of myelin loops and widening of the nodes.25 26 In an ex vivo mouse model, it was shown that an anti-GD1a antibody bound to the nodes of Ranvier of distal motor axons.27 Complement binding was associated with loss of nodal proteins, including Nav channels, and loss of inward Na+ and outward K+ currents. Pathological and electrophysiological alterations were prevented by eculizumab, a monoclonal humanised mouse monoclonal antibody that impedes the development of the membrane attack complex, whereas inhibition of the protease calpain preserved the immunostaining profile of node of Ranvier but did not prevent electrophysiological abnormalities.
These studies suggest that in AMAN, in the early stage, there is an immune attack to the node with destruction of Nav channel cluster and nerve conduction failure which may still be reversible. If the attack progresses, calcium enters into the axon and protease activation induces axonal damage and Wallerian-like degeneration.1 The re-evaluation of experimental models with antibodies to GM1, GD1a and GD1b indicated in all a common pathophysiological mechanism characterised by complement=mediated dysfunction/disruption at the node of Ranvier.28
The crucial role of complement demonstrated in animal models prompted two ongoing randomised controlled trials, ‘Inhibition of complement activation in Guillain-Barré syndrome: the ICA-GBS study’ and the ‘Japanese eculizumab trials for Guillain-Barré syndrome (JET-GBS)’, to find out whether eculizumab, given in addition to intravenous Ig, is beneficial in GBS.29 30 The preliminary results of the Japanese trial showed a greater percentage of patients able to run 24 weeks after GBS onset in the eculizumab-treated group compared with placebo (74% vs 18%) (S. Kuwabara, personal communication).
Acute autoimmune nodopathies: the proposal of the concept
The recognition of RCF led to reconsider the diagnostic accuracy of the criteria sets employed in the electrodiagnosis of GBS subtypes.31 In Italian and Far East GBS cohorts, 24%–38% of patients changed the initial electrodiagnostic classification after serial recordings.21 32–34 The majority of shifts were from acute inflammatory demyelinating polyradiculoneuropathy (AIDP) and unclassifiable diagnosis to axonal GBS, and the principal motivation was the identification of RCF as manifestation of axonal involvement. RCF is not rare as it is detectable by serial studies in distal, intermediate or proximal segments in at least two motor nerves of 46.6% of patients with axonal GBS (figure 2).35 Not considering the possibility of distal RCF may induce, on the basis of a single test showing reduced distal CMAP amplitudes, to erroneously formulate a poor prognosis. Moreover, by a single test, patients with AMAN and RCF might be fallaciously classified as AIDP. These observations also explain why antibodies to gangliosides were believed to be related not only to AMAN but also to AIDP.15 36
RCF is an ‘a posteriori’ diagnosis as it can be recognised only by serial studies and is not included in the currently employed electrodiagnostic criteria.15 16 This questioned the usage of nerve conduction studies in distinguishing the GBS subtypes by a single test in the early stages of disease.31 32 However, a recently proposed criteria set considering proximal/distal CMAP amplitude ratio <0.7, without other features of demyelination, as indicative of axonal GBS and additionally evaluating CMAP duration and sural sparing pattern as well as an innovative statistical–mathematical method of classification greatly increased the electrodiagnostic accuracy of GBS subtypes at the first electrophysiological test.35
RCF is not only detectable in motor fibres. RCF in sensory nerves as well as lengthened nodes in the dorsal root fibres were described in patients with acute motor and sensory axonal neuropathy.37 38 RCF in motor and sensory fibres has been also described in patients with the pharyngeal–cervical–brachial subtype of GBS.39 40 Finally, RCF limited to sensory fibres has been shown in patients with acute sensory ataxic neuropathy and in Miller-Fisher syndrome (figure 2).41 42 The first ever reported sensory nerve biopsy of a patient with Miller-Fisher syndrome and anti-GQ1b antibodies revealed elongated nodes, myelin splitting at the paranode and macrophage invasion of the internodal axon without any features of segmental demyelination similar to the pathological findings found in motor fibres of patients with AMAN.43
At last, the description of a patient with the clinical picture of GBS and severe neuropathic pain, RCF and IgG3 anti-Caspr1 antibodies activating complement suggested that reactivity to paranodal Caspr1, with a mechanism similar to antiganglioside antibodies, could be also involved in acute autoimmune neuropathies.5 The dichotomous classification of neuropathies into axonal or demyelinating might be inadequate in the diagnosis of GBS subtypes. AMAN is categorised as an axonal neuropathy. However, the word axonal may be confusing as in the common view it is strictly associated with Wallerian-like degeneration, suggests poor prognosis and not the possibility of a prompt recovery. On the other hand, although the pathogenic mechanism is focused on the node, some ‘demyelinating-like’ features such and detachment of myelin terminal loops and widening of the nodes (but never true segmental demyelination) have been shown in AMAN. To overcome these nosological difficulties, GBS subtypes with antigangliosides antibodies could be better categorised as acute nodopathies because of the common pathophysiology at the node with the key initial involvement of Nav channels and a continuum from RCF to axonal degeneration (figure 1B). The advantages are that the term nodopathy points directly to the site of nerve injury reminding the specific pathophysiological mechanism, circumvents the apparent paradox that axonal neuropathies may be promptly reversible and have a good prognosis, and reconciles the contrast between electrophysiology and morphology, thereby avoiding the confusing situation that the same patients, in spite of the common site of nerve damage and pathophysiological mechanism, might be electrophysiologically classified as demyelinating or axonal.
Chronic autoimmune paranodopathies: expanding the concept
CIDP is a clinically heterogeneous disorder thought to have an autoimmune basis and its hallmark is inflammatory-mediated demyelination.44 45 The pathological criteria for diagnosis require unequivocal evidence of demyelination or remyelination with at least five demyelinated fibres in semithin or ultrathin sections or >12% of teased fibres (minimum of four internodes each) showing de-remyelination.46 47 Supportive criteria include mononuclear cell infiltration and presence of onion bulbs.
The resemblance of CIDP with experimental allergic neuritis, obtained in rats by immunisation with the proteins of compact myelin, suggested that P0, P2 and PMP 22 could be autoantigens. However, after many years there is little evidence for a pathogenic role of a humoral response to proteins of compact myelin in patients with CIDP.48 A new avenue was introduced by Devaux and coworkers in 2012 reporting that IgG from 30% of patients with CIDP bound at the nodal region of rat nerve.49 These findings re-boosted the investigation in the field and patients with antibodies targeting the axo–glial component of paranodal junctions were soon described. Querol and coworkers showed that 6.5% of 45 Spanish patients with CIDP carried IgG to CNTN1 or CNTN1 and Caspr1.3 These patients had a common phenotype characterised by advanced age, aggressive onset, severe axonal damage, motor predominance and poor response to intravenous Ig. Anti-CNTN1 antibodies were detected also in 2.4% of 500 Japanese patients with CIDP presenting subacute onset, sensory ataxia and poor response to intravenous Ig and in 8% of 53 German patients showing acute onset, prevalently motor neuropathy and tremor.50 51 The Spanish group reported also four patients with antibodies to NF155 presenting severe distal sensorimotor neuropathy, disabling tremor and poor response to intravenous Ig.4 This observation was confirmed in 7% of 533 Japanese patients with CIDP showing ataxia, tremor, poor response to intravenous Ig and, in 3 patients, central nervous system demyelination.52 Finally, antibodies directed exclusively to Caspr1 were described in one patient with CIDP with prominent neuropathic pain.5 Overall, the patients harbouring antibodies against proteins of paranodal junctions account for about 10% of all patients classified as CIDP.
The axonal cell adhesion molecules CNTN1 and Caspr1, and the glial NF155 represent an essential complex in the formation and stability of paranodal junctions and responsible, through the transverse bands, of the adhesion of terminal myelin loops to axolemma. Paranodal junctions act also as a fence for the juxtaparanodal segregation of Kv channels and overall contribute to the saltatory conduction in myelinated fibres (figure 1).
Interestingly antibodies to axo–glial proteins are predominantly or exclusively of the IgG4 isotype that has been originally considered to be immunomodulatory as it is unable to activate complement and bind to Ig Fc receptor.53 In allergies, IgG4 antibodies dampen the inflammatory response and induce tolerance to allergens after repeated challenge.54 Up to now, IgG4 antibodies have been reported only in a few non-allergic diseases as pemphigus and myasthenia gravis with antibodies to muscle-specific tyrosine kinase. In these disorders, IgG4 antibodies are thought to be pathogenic by blocking critical functions of the target antigen.53 Specifically, IgG4 anti-CNTN1 prevented, in an in vitro model, the binding of the CNTN1/Caspr1 complex to NF155 and disrupted the node of Ranvier structure in the absence of complement and inflammatory cells.55 After intraneural injection in vitro, IgG4 anti-CNT1 diffused into the paranode and induced, when passively transferred in rats immunised with P2, progressive deterioration of the experimental neuropathy.56
Electrophysiological studies of patients with antibodies to axo–glial proteins of the paranodal junction showed prolonged distal motor latencies and conduction slowing in the demyelination range, CB and some temporal dispersion but also reduced amplitude distal compound muscle action and spontaneous activity at needle electromyography indicative of substantial axonal degeneration.3–5
Immunohistochemistry studies of dermal myelinated fibres of patients with anti-CNTN1 and anti-Caspr1 antibodies revealed loss of paranodal Caspr1 and neurofascin immunoreactivity and lengthened nodes.5 51 Optic microscopy of a sural biopsy of patients carrying antibodies to CNTN1, NF155 and Caspr1 showed axonal degeneration and loss with regenerating fibres, few fibres with thin myelin sheaths but no inflammatory infiltrates or onion bulbs.5 51 52 57 A teased fibre study of nine patients with antibodies to NF155 showed widening of the nodes and ‘apparent demyelination’ in three patients but the percentages of de-remyelinated fibres in the whole group was no different from normals.58 At last, ultrastructural studies in patients with anti-NF155 antibodies displayed detachment of terminal myelin loops due to lack of transverse bands, abnormal widening of periaxonal space with in some instances Schwann cell processes interposed between myelin loops and axon, and widened nodes (figure 3).57 58 It should be underlined that the above alterations do not fulfil the pathological criteria for diagnosis of CIDP, and an axonal neuropathy with no features of de-remyelination was the final pathological diagnosis in three patients with anti-CNTN1 and in one with anti-Caspr antibodies.5 51 In the mutant mice models of CNTN1, NF155 and Caspr1, myelin thickness and the g-ratio are not different from the wild-type mice; the structure of compact myelin is normal and the ultrastructural alterations found at paranode are identical to those reported in patients.59–61 Kv channels are aberrantly expressed at the paranode, whereas Nav channels retain their nodal distribution.61 From the electrophysiological point of view in the mutant animals, there is a twofold to threefold decrease in motor nerve conduction, the threshold of nerve stimulation and the refractory period are increased, whereas the CMAP amplitude is decreased.59–61 In both patients and mutant animals, we are facing a condition in which the electrophysiology is clearly ‘demyelinating’ whereas pathology shows only paranodal dismantling without evidence of true de-remyelination. This is only an apparent contradiction. The involvement of paranodal junctions with loss of transverse bands induces terminal myelin loop detachment and retraction resulting in widening of the node with consequent increased nodal capacitance and dilution of the capacitive current over a larger surface (figure 4). Myelin loop detachment and increased periaxonal space may reduce the paranodal transverse resistance with increased current leakage, radial shunting and backflow of the current to the paranode instead of longitudinal progression to the following node (figure 4).59 Moreover, the abnormal displacement of Kv channels at the paranode may shifts membrane polarisation to more negative values. All these changes increase the time necessary to depolarise the next node producing slow conduction and, if the current is insufficient to reach the threshold, failure of impulse transmission. Overall, these considerations well explain the electrophysiological ‘demyelinating’ features in the presence of only paranodal disruption and dysfunction. Patients with IgG4 antibodies to proteins of paranodal junctions show acute–subacute onset, usually a severe course with axonal degeneration and poor response to intravenous Ig.3 4 50 As major mediators of intravenous Ig effect are inhibition of complement pathway and binding to the inhibitory Ig receptor, the peculiar characteristics of IgG4 antibodies may explain the poor response to intravenous Ig in these patients.62–64 The paranodal junction is thought to play a role in molecular communication between the axon and the Schwann cell. Nodal and paranodal axoplasm of NF155 and Caspr1 mutant mice contain large mitochondria and abnormal accumulation of cytoplasmic organelles indicating an altered axonal transport and suggesting that the disrupted axo–glial interaction may induce, in a still unknown way, axonal degeneration.60 61 B cell depletion therapy, as rituximab, resulted to be effective in patients not responding to conventional treatment, above all in patients with short disease duration possibly because of less axonal damage.5 51 65 Therefore, identifying the target antigens in these autoimmune neuropathies is important to understand their pathogenic mechanisms, but also to correctly diagnose and treat the affected patients. Because of the location and role of the axo–glial proteins and the ultrastructural alterations restricted at the paranode without de-remyelination, the neuropathies with antibodies to paranodal junction components should be better classified, in our opinion, as chronic paranodopathies (figure 1B). The advantages are that the term, focusing on the primary site of nerve injury, reconciles the contrasting electrophysiological and microscopical findings and separates these autoimmune neuropathies from CIDP as there is no evidence of inflammatory infiltrates, complement activation and of de-remyelination (namely the I and D of the CIDP acronym). Very recently, five patients classified as CIDP and harbouring antibodies against NF186 and NF140, the nodal isoforms of neurofascin, have been reported.66 Nerve conductions showed axonal features in two patients, and CB and demyelinating features in three patients although, after normalisation of the electrophysiological findings, only one patient fulfilled the European Federation of Neurological Societies/Peripheral Nerve Society electrodiagnostic criteria for definite CIDP. A nerve biopsy in this patient showed only mild axonal loss without demyelination, onion bulbs or cellular infiltration. Four patients had predominantly IgG4 antibodies and one presented IgG3 antibodies that in in vitro activated complement. Intravenous Ig and corticosteroids were effective in three patients and one patient remitted following treatment with rituximab. Remission was associated with autoantibody depletion and with recovery of CB and distal CMAP amplitudes. Although reactivity to nodal NF186 needs to be confirmed in further studies and there are no available data about ultrastructural alterations, we think that these patients could be better described as chronic nodo-paranodopathy than CIDP. Multifocal motor neuropathy (MMN) is characterised by asymmetrical, predominantly distal limb weakness, no sensory loss, motor CB (often persistent) with or without temporal dispersion of CMAP, occurrence of IgM to GM1 in about 50% of patients and response to intravenous Ig in up to 90% of patients suggesting an immune-mediated aetiology.67 In addition to CB, MMN is characterised by axonal degeneration which is the main determinant of the progressive course.68 It is controversial whether MMN is a primary demyelinating or an axonal disorder and the results of the few pathological studies are contradictory.69 70 An active immunisation model is lacking but injection of human sera containing IgM anti-GM1 antibodies and fresh complement into the rat sciatic nerve caused CB, binding of Ig at the nodes of Ranvier, nodal lengthening and some paranodal disruption.71 72 High-concentration anti-GM1 sera applied without complement to single rat myelinated axons increased K+ current, whereas in the presence of complement, Na+ current decreased and leakage current increased.73 These findings suggested that anti-GM1 antibodies, by themselves, can uncover Kv channels in the paranodal region while anti-GM1 antibodies bound to the nodal membrane in the presence of complement may block Nav channels and disrupt the membrane at the node of Ranvier. However, it should be mentioned that a following study did not confirm these effects.74
Anti-GM1 IgM antibodies from patients with MMN, activate complement in vitro and efficient complement activating properties are determinant of weakness and axonal loss.75–77 By analogy with AMAN, it has been hypothesised that IgM anti-GM1 antibodies induce complement-mediated disruption of ion channels and paranodal structures compromising nerve conduction and eventually causing axonal degeneration.67
Studies of antibodies against nodal proteins have generated conflicting results in MMN. One study found that 62% of patients had IgG autoantibodies against gliomedin or NF186 in combination with anti-GM1 IgM or alone in 10% of patients.78 In the same study, antibodies to gliomedin or NF186 were found in only 1% of patients with CIDP. A following study failed to replicate the association between MMN and anti-NF186 antibodies and did not find anti-NF155 or anti-CNTN1 antibodies either.79 Although the discrepancy of results may be explained by different proteins used for ELISA, the potential association between MMN and an antibody response against nodal proteins may be worthwhile of further investigation.
In conclusion, although the pathological evidence is still scanty and the pathophysiology not fully understood, we think that MMN could be better categorised as a chronic autoimmune nodo-paranodopathy than as a demyelinating neuropathy (figure 1B).
The nodo-paranodopathy concept seems appropriate to several acute and chronic neuropathies with antibodies to gangliosides and to paranodal axo–glial proteins. It has the advantage to focus on the site of primary nerve injury reminding specific pathophysiological mechanisms, reconciles contrasting morphological and electrophysiological findings finally avoiding misdiagnosis and taxonomic confusion. We think that the category of the autoimmune peripheral nodo-paranodopathy that we propose better systematises the neuropathies characterised by an autoimmune attack targeting and limited to the nodal region and integrates the traditional classification of peripheral neuropathies.
Contributors AU: conceived the review and wrote the first draft. J-MV: contributed with the electron microscopy studies and critically reviewed all the versions of the manuscript.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.