Article Text

Download PDFPDF

Nodal and paranodal antibody-associated neuropathies
  1. Janev Fehmi1,
  2. Tom Vale1,
  3. Stephen Keddie2,
  4. Simon Rinaldi1
  1. 1 Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, UK
  2. 2 Department of Molecular Neuroscience, National Hospital for Neurology and Neurosurgery, MRC Centre for Neuromuscular Diseases, London, UK
  1. Correspondence to Dr Janev Fehmi, Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford OX3 9DU, UK; janev.fehmi{at}


Within the last decade, antibodies targeting the node and paranode of myelinated peripheral nerves have been increasingly identified in patients with acquired immune-mediated neuropathies, commonly termed ‘nodo-paranodopathies’. Crucially, these patients often present with additional clinical features not usually seen with the most common immune-mediated neuropathies, Guillain-Barré syndrome and chronic inflammatory demyelinating polyneuropathy, and respond poorly to conventionally used immunomodulatory therapies. Emerging evidence that these are pathologically distinct diseases has further prompted the use of more targeted treatment, such as the B cell depleting monoclonal antibody rituximab, which has been reported to significantly improve functional outcomes in this subset of patients. We provide an overview of the emerging clinical and serological phenotypes in patients with specific nodal/paranodal antibodies, the practicalities of antibody testing and current evidence supporting the use of non-standard therapies.

  • immunology
  • neuroimmunology
  • neuropathy
  • Guillain-Barre syndrome

Statistics from

Request Permissions

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.



The so called ‘inflammatory’ neuropathies have historically been classified on clinical and electrophysiological grounds, though this categorisation does not necessarily reflect common underlying pathogenic mechanisms or predict response to treatment. Antibodies targeting the node and paranode of myelinated peripheral nerves predominantly occur in patients initially meeting diagnostic criteria for Guillain-Barré syndrome (GBS) or chronic inflammatory demyelinating polyneuropathy (CIDP), often in combination with additional clinical features and poor response to conventional therapies. There is increasing evidence that these ‘nodo-paranodopathies’ may better be considered as pathologically distinct disease entities. They may have limited evidence of inflammation or demyelination, and more targeted immunotherapies, such as the B cell depleting monoclonal antibody rituximab, may be of greater benefit. Increasing knowledge of the origins and immunopathogenic mechanisms of these antibodies will undoubtedly provide insight into the determinants of clinical phenotype and response to treatment in these disabling but treatable disorders.

Molecular targets of the node of Ranvier

The node and paranode are highly specialised domains of the peripheral nerve. Here, a dense array of communicating proteins facilitate adhesion of myelin to axon and help maintain the correct localisation of voltage-gated sodium and potassium channels (figure 1), thus promoting effective saltatory conduction. Predominantly immunoglobulin G4 (IgG4) antibodies, specifically targeting nodal neurofascin 186 (NF186), or the paranodal molecules neurofascin 155 (NF155), contactin-1 (CNTN1) and contactin-associated protein-1 (Caspr1) have recently been identified in phenotypically distinct groups of patients.

Figure 1

Key axo-glial target molecules at the node and paranode.37 (A) Overview of the specialised domains of a myelinated peripheral nerve—each node is flanked by paranodes, where terminal myelin loops form a tight junction with the axon. (B) NF186 at the node binds to gliomedin attached to Schwann cell microvilli, helping to maintain structural integrity of VGSCs. CNTN1 and Caspr1 form a complex with NF155 at the paranode to restrict movement of ion channels between domains. Adapted by permission from BMJ Publishing Group Limited (Nodes, paranodes and neuropathies, Fehmi J et al, 0:1–11, 2017). Caspr1, contactin-associated protein-1; CNTN1, contactin-1; MAG, myelin-associated glycoprotein; NF186/155, neurofascin 186/neurofascin 155; VGSCs, voltage-gated sodium channels; PSD, postsynaptic density.

The practicalities of testing for paranodal/nodal antibodies

Who and when to test

We suggest that it may be more appropriate to base therapeutic choices in patients presenting with a rapidly progressive neuropathy on suspicion or identification of the likely pathological cause, rather than solely relying on ‘time to nadir’, as is usually the case in distinguishing GBS (within 4 weeks) from CIDP (over 8 weeks). The umbrella terms GBS and CIDP encompass numerous subgroups and ‘atypical’ variants, in many cases reflecting pathologically distinct processes. Identifying these as early as possible should enable the prompt initiation of the most appropriate treatment with the aim of preventing further disability. Serological testing for paranodal/nodal antibodies (PNAbs) is increasingly accessible, and these assays now run weekly in our laboratory. For these reasons, we advocate testing patients early (ideally before treatment) and not to restrict testing to those with specific clinical features.

Suggestive clinical phenotypes

Although paranodal/nodal antibody positive patients (PNAb+) are distinct from those with seronegative (PNAb-) inflammatory neuropathies at the group level, no single clinical or paraclinical feature accurately predicts serological status. Nevertheless, various findings do increase the likelihood of detecting a paranodal/nodal antibody (table 1). Most reported patients with PNAbs are adults, although there have been cases in children,1–3 and our cohort has a 2.5:1 excess of male versus female cases. Seropositive individuals commonly present with a severe and symmetrical, motor-predominant and distal-predominant polyneuropathy, with a more rapidly progressive disease than those with seronegative CIDP. This is reflected by the fact that an initial clinical diagnosis of GBS was suspected significantly more frequently in PNAb+ compared with PNAb- patients from our cohort (24/51 vs 43/190, p=0.0013, OR: 3.04, 95% CI 1.56 to 5.87). Peak clinical severity was also significantly greater, with 80% (44/55) of PNAb+ patients versus 40% (74/185) of PNAb- patients being immobile or worse at clinical nadir (modified Rankin score - mRS ≥4) (p<0.0001, OR: 6, 95% CI 2.93 to 12.53).

Table 1

Characteristics suggesting a paranodal/nodal antibody-positive (PNAb+) neuropathy

With a subsequent chronically progressive or relapsing–remitting disease course, and accompanying electrodiagnostic features, patients typically later meet the European Federation of Neurological Societies/Peripheral Nerve Society (EFNS/PNS) criteria for ‘definite CIDP’.4 However, PNAb+ patients more frequently have additional features, and are less often associated with seronegative CIDP, in particular severe sensory ataxia and tremor.5–7 Other specific features occur preferentially in specific antibody groups (table 2). For example, tremor and cerebellar or sensory ataxia more commonly occur in anti-NF155 seropositive patients, while an aggressive-onset neuropathy associated with cranial neuropathies, autonomic dysfunction and respiratory insufficiency may occur in patients with antibodies that cross react with both neurofascin isoforms (NF155 and NF186—‘pan-NF’).

Table 2

Typical characteristics in patients with individual nodal/paranodal antibodies

Nephrotic syndrome is increasingly identified as an associated condition, particularly in those with CNTN1 antibodies.7–12 In the largest case series published to date of patients with CNTN1 antibody positive,11 we identified CNTN1-containing immune complex deposits within renal glomeruli of seropositive patients, and saw a temporal correlation of the clinical course of both neuropathy and nephropathy, pathologically implicating CNTN1 antibodies in this cross-system disorder. In our cohort, nephrotic syndrome was significantly more common in patients with PNAbs, compared to those without (14/51 vs 5/150, p<0.0001, OR: 10.97, 95% CI 3.80 to 28.64). Nephrotic syndrome is also a frequent comorbidity in patients with pan-NF antibodies (6/19 in our cohort). Additionally, we have found underlying haematological disorders (Hodgkin’s lymphoma, chronic lymphocytic leukaemia and myeloma) in pan-NF patients, in our cohort all characterised by the presence of an IgG-lambda paraprotein.

This group is further defined by an incomplete response to first-line therapies, intravenous Ig in particular, and a more sustained improvement from alternative immunomodulatory therapy such as rituximab.

Typical electrophysiological features

Similarly, there are no electrophysiological features that are unequivocally diagnostic of PNAb seropositivity. Features suggesting demyelination and/or axonal degeneration can be seen, with no strong or predictable association to serological status. Slowed conduction velocities, prolonged distal motor and F-wave latencies, and conduction block are frequent findings, often interpreted to represent demyelination. Indeed, these features may be prominent enough to meet electrodiagnostic criteria for ‘primary demyelinating’ GBS, or CIDP, despite no or little pathological evidence of segmental demyelination.13–16 ‘Axonal’ conduction block without temporal dispersion of compound muscle action potentials, characteristic of paranodal/nodal pathology, can also develop. This may either rapidly resolve (reversible conduction failure) or progress to axonal degeneration. Even in these cases, clinical recovery can be complete, contrary to the general perception that axonal damage is associated with poor prognosis. Nevertheless, there is a strong rationale to suggest that early treatment aimed at reversing conduction block and preventing axonal degeneration should improve outcomes.

How to test

Venous blood, collected in a serum separator tube, is preferred for testing. Ideally, this should be taken before treatment to avoid contamination with an external source of IgG (from intravenous Ig), or dilution/removal or suppression of the antibody (by plasma exchange or immunosuppression). At present, we recommend requesting testing for antibodies against CNTN1, Caspr1, NF155 and NF186, in parallel, as phenotypic variability within and overlap between antibody cohorts makes identifying the specific seropositive syndrome in an individual patient (before testing) challenging. Pan-NF antibodies appear to have different clinical, prognostic and mechanistic associations when compared with NF155-monospecific antibodies. This distinction cannot be made without testing against both isoforms. Furthermore, other clinically relevant antibodies preferentially target the CNTN1/Caspr1 complex, and may not be detected unless using an assay in which both proteins are co-presented.17

The value of testing cerebrospinal fluid (CSF) has not yet been determined. In studies of patients with combined central and peripheral demyelination, NF155 antibodies were largely detected in both serum and CSF,18 though one patient had only CSF antibodies.19 In a series of pan-NF antibody positive patients, antibodies were detected in CSF of one of three patients tested, though only after 6 months from disease onset, suggesting poor clinical utility, at least early in the disease process.20

How testing is done

In the UK, PNAb testing is currently performed by the diagnostic neuroimmunology laboratory in Oxford. Detailed sample requirements and a request form can be found at A live cell-based assay (CBA), in which cells are transiently transfected to overexpress the target protein on their surface, is used for the initial screen (figure 2). Positivity is rated on an ordinal scale, using fluorescence microscopy, based on cellular membrane fluorescence intensity and where possible, co-localisation with a commercial antibody targeting the same protein. A positive result is indicated by a clear cellular membrane fluorescent signal, which co-localises with the commercial antibody, and the absence of non-specific staining of control cells. Positive samples undergo repeat testing with subclass specific secondary antibodies and serial doubling dilutions from 1:100 to 1:6400 to determine the end-point titre. Positive and equivocal samples are also tested using an ELISA, whereby the target protein is fixed to a solid surface, exposed to patient sera and analysed for bound antibodies.

Figure 2

Cell-based assay for paranodal/nodal Abs. (A) Illustration showing how HEK cells are transfected with plasmid vectors encoding the desired nodal/paranodal protein target, that is, NF155, which is then expressed within the cell membrane. Subsequent exposure to patient sera allows anti-paranodal/nodal Abs (green), if present, to bind to the target antigen. Concurrent exposure to a commercial Ab targeting the same protein (red) can both be visualised independently using fluorescence markers, and should co-localise if the patient is PNAb+ (merge). (B) Fluorescence microscopy images (×63 magnification) taken using either NF155 Ab+ or pan-NF Ab+ patient sera incubated with NF155 or NF186 transfected HEK cells. Human IgG (green) from both patients bind to NF155 transfected cells, and co-localises with the commercial NF Ab (red). However, only IgG from the patient with pan-NF Abs also binds to NF186 transfected cells, emphasising the importance of testing for reactivity to both antigens individually. DAPI (blue) identifies the cell nuclei. Abs, antibodies; CNTN1, contactin-1; HEK, human embryonic kidney; IgG, immunoglobulin G; NF, neurofascin.

Most centres around the world use similar approaches, although some employ fluorescence-activated cell sorting or western blot. It should be noted that the established clinical associations were made using CBAs, with or without ELISA and confirmation on teased nerve fibres, and as such the sensitivity, specificity and clinical implications of results generated by other methods may not be comparable. This is a particular concern with western blot, which uses denatured, linearised proteins, thus altering the conformational epitopes present in cell-based assays, and exposing cytoplasmic domains to which pathologically irrelevant antibodies might bind. In contrast, cell-based assays should more accurately simulate the in vivo configuration of the extracellular epitopes that pathogenic antibodies are thought to target.

Variability in the range of antigens tested against and assay method used at least in part accounts for the variable frequencies of antibody detection observed in published cohorts of patients, with differences in the profiles of the patients tested likely also contributing.

Diagnostic significance of antibody isotype and subclass

Of the five antibody isotypes, IgG is the most abundant in human serum, and can be further divided into subclasses 1–4. These subclasses have different constant regions and vary in their ability to induce specific humoral or cellular effector functions. For example, IgG1 and IgG3 are more potent activators of complement than IgG2 and IgG4.

Antibody isotype, subclass and titre are likely relevant to pathogenicity, but can be influenced by timing of sera sampling in relation to stage of disease and whether prior treatment has been administered. IgG4 antibodies are most frequently detected in PNAb+ patients and appear to be highly specific for the presence of an immune-mediated neuropathy phenotypically distinct to seronegative CIDP. IgG1–3 subclass antibodies are often found concurrently at lower titres.

Patients with only non-IgG4-subclass PNAbs have been reported to be clinically indistinguishable from seronegative patients,7 and may have a more favourable response to intravenous Ig. There is, however, emerging evidence that non-IgG4 subclass antibodies can have pathogenic effects.16 IgG3 pan-NF antibodies have been found in a small number of patients with a distinct, rapidly progressive, GBS-like neuropathy, characterised by severe tetraparesis, cranio-respiratory and autonomic dysfunction.20 In our series, most pan-NF antibodies have been exclusively or predominantly IgG1 (17/19), but have again been found in patients with this phenotype. Whether the subclass differences between these studies reflect technical variability in the laboratories, or a true difference in the populations being tested, remains to be seen. This should be answered by an Europe-wide, inter-laboratory validation study, which is shortly due to start. We have only identified a further five patients with exclusively non-IgG4-subclass antibodies (predominantly IgG1) targeting NF155. However, these seem to be less specific, and one patient with exclusively IgG1, low titre, NF155 antibodies was ultimately diagnosed with motor neurone disease.

Patients with IgM antibodies, mainly targeting NF (without concurrent IgG or a history of monoclonal gammopathy of undetermined significance - MGUS), have also been reported. These IgM antibodies appear to be less specific than IgG for an immunotherapy-responsive neuropathy.2 In five patients with low titre IgM anti-NF155 antibodies,21 the clinical features were in keeping with those described for patients with IgG NF155 antibodies, although those with IgM alone responded well to intravenous Ig. However, their clinical significance and specificity are not yet known, and currently we do not routinely test for IgM PNAbs in our laboratory.

If IgG4 PNAbs are detected in patients with an aggressive onset neuropathy and/or additional atypical features, they should be considered reliable diagnostic biomarkers for an immune-mediated neuropathy distinct from seronegative CIDP. As this group responds poorly to one or a combination of corticosteroids, intravenous Ig and plasma exchange, this should prompt early consideration of a different treatment strategy. If other antibody subclasses predominate, with the probable exception of IgG1/3 in pan-NF positive patients, the clinical relevance is less clear, and the response to standard treatments may be more favourable.

A negative test result is less informative, and most patients with clinical diagnoses of GBS or CIDP are seronegative. Repeat testing can be useful if sera are initially negative or equivocal, but clinical suspicion remains high, in particular if the sample was taken during a period of remission or immediately after treatment, when antibody concentrations may be at their lowest.


Seropositive patients appear less responsive to first-line treatments for ‘inflammatory’ neuropathies, and it would seem appropriate to consider escalation therapies (such as rituximab) early in the disease course, even if there is a transient response to initial therapies, likely indicating an ongoing and potentially treatment-responsive immune process.

There are many challenges in accurately determining the relative efficacy of specific treatments in these patients. The current evidence largely consists of small, uncontrolled, retrospective cases series or case reports, in which multiple treatments were administered in quick succession, driven by rapid clinical deterioration. It is, therefore, difficult to ascribe recovery to a single agent, the temptation being to attribute success to the last tried. Second, there has been no standardised way of reporting efficacy of treatments, so with each study or report providing variable levels of clinical detail. Reports of ‘partial’ or ‘good’ recovery do not allow for reliable comparisons between cohorts.

With this is mind, there has been a consistent observation that intravenous Ig is often, at best, transiently effective. One proposed explanation is that its anti-inflammatory action involves suppression of complement activity, whereas IgG4-subclass antibodies do not fix complement and must exert their pathogenic effects by other mechanisms.22 Indeed, this might explain why intravenous Ig has provided transient relief in the acute stages of illness, where complement-activating IgG subclasses (IgG1–3) may be more likely to predominate.23 We have recently shown that repeated treatment with immunoadsorption or plasma exchange may be required to induce sustained suppression of antibody levels and clinical improvement.24 Perhaps counterintuitively, in our series, plasma exchange was less often judged to be effective in seropositive patients compared with seronegative patients. In a single patient, following immunoadsorption alone, we observed a rapid rebound in NF155-antibody titres on serial testing, and this partial and transient suppression of antibody levels was not associated with any improvement in disability. It could be that more persistent suppression of antibody levels is required to induce a clinically meaningful response, or simply that in some cases, more time is needed for peripheral nerve repair and recovery. Nevertheless, some patients have clinically improved and remained in long-term remission following corticosteroid treatment alone, in parallel to suppression of antibody titres. Whether this represents spontaneous or steroid-induced recovery is not clear.

The suggestion that the anti-CD20 monoclonal antibody rituximab may be more effective if given early in the disease course7 is in keeping with the notion that rituximab mainly targets immature CD20+ B cells, rather than mature plasma cells. Immature CD20+ B cells are responsible for generating short-lived antibody-secreting cells (plasmablasts), which may be primarily responsible for nodal/paranodal antibody production early on. However, more indirect mechanisms remain possible. Rituximab has a favourable and enduring effect in most PNAb+ patients refractory to multiple immune therapies (~31 patients at the time of writing). In 77% (24) of cases, the authors concluded that patients had between a ‘good’ and ‘complete’ response. Although there is not yet a consensus on optimal timing of starting non-standard treatment, patients treated with escalation therapy, including rituximab, within 1 year of disease onset showed a sustained good response.7 25 A few patients showed some response to rituximab started many years after disease onset.26

Responses can occur within weeks to months of starting treatment, and can persist for at least 1–2 years.25 Antibodies are often undetectable after treatment, in line with clinical remission,11 and remained low at 6 months following treatment in four patients in one study.25 Although most antibody-secreting B cells are CD20−, there is some evidence that autoreactive plasmablasts are more often CD20+, and may be selectively depleted by such therapy, thus reducing autoantibody titres, while preserving overall antibody concentrations,27 which are largely maintained by CD20− plasma cells in the bone marrow.

One recent study has shown near full clinical remission in 3 patients within 6 months using a low-dose rituximab regimen (100 mg/week for 4 weeks, 100 mg/month for 2 months), concurrent with a reduction in corticosteroid dosage.28 In rheumatoid arthritis, lower dose treatment (2×500 mg) is considered as effective as, and safer than, standard dosing (2×1000 mg),29 which carries a 8%–10% annual risk of significant infection.30 It remains to be seen whether cheaper, lower dose regimens are as effective in PNAb+ patients.

In our cohort, 57% of the 44 treated with corticosteroids and 83% of the 23 treated with rituximab were judged to have had at least a ‘partial’ response (figure 3). Conversely, most patients treated with intravenous Ig (67% of 46), plasma exchange or immunoadsorption (66% of 29) responded ‘equivocally’ or worse. Of the 12 patients who received non-standard immunosuppressive treatment other than rituximab, 8 (67%) responded at least partially.

Figure 3

Physician reported treatment responses of the 62 patients with PNAb+.

Rituximab has been previously categorised as ‘not for routine commissioning’ for CIDP by NHS England. A commissioning review is underway for paranodal/nodal antibody-associated neuropathies more specifically and is expected to report soon. As things stand, obtaining rituximab for this indication in England invariably relies on departmental or institutional level discretionary funding. We are not aware of any success following individual funding requests to NHS England.


From existing data, we know that disability at clinical nadir is high in PNAb+ patients, but crucially they have the potential to achieve long-lasting clinical remission, most commonly with non-standard immunosuppressants. This is perhaps best illustrated by those with pan-NF antibodies, who often present with a fulminant neuropathy requiring intensive care support,20 yet have the potential to return to independent living, in some cases without any residual deficit. This indicates that specific antibodies may in themselves be prognostic, in addition to diagnostic, markers. Furthermore, with increasing data linking falling antibody titres to clinical and electrophysiological remission,1 2 8 11 21 31–35 they show promise as serological markers for disease monitoring, facilitating more aggressive escalation of therapy if required. Indeed, prolonged suppression of antibody titres has invariably correlated with long-term clinical remission in patients with N/PNAb+ identified via our diagnostic laboratory.11 24 36 Conversely, other patients have achieved clinical remission, while remaining seropositive. Whether they have a greater risk of subsequent relapse remains to be seen. Five PNAb+ patients from our series became seronegative after a single cycle of rituximab, in parallel with significant clinical improvement. Only 2 of 23 rituximab-treated patients have so far relapsed clinically, both around 18 months after treatment. In both cases, clinical relapse was associated with a return of seropositivity for the same antigen.

However, we cannot yet predict clinical outcome based purely on serological status. The longer-term functional outcome (>1 year) of PNAb+ patients also needs to be documented, along with the need for repeated immunosuppressive treatment and the impact of adverse events. Only then can we begin to understand the factors that determine favourable outcome in individual patients.


Patients with PNAbs have characteristic clinical, electrophysiological and histological features that differentiate them from ‘typical’ CIDP and GBS. Although they frequently meet electrodiagnostic criteria for CIDP, the underlying neuropathies may not be inflammatory, demyelinating or even chronic, and we believe they should be considered as distinct, serologically defined disorders.

The clinical implications of PNAbs are significant, and we recommend their prompt detection in patients presenting with suspected immune-mediated neuropathies. Their presence is highly specific in identifying patients with a rapidly progressive and severe autoimmune polyneuropathy distinct from seronegative CIDP. Seropositivity is associated with resistance to standard first-line CIDP immunotherapies, and treatment escalation is often required, probably with increased chances of meaningful recovery if administered early in the disease course.

Currently, detecting specific PNAbs does not dictate treatment choice. With increasing understanding of the pathogenic mechanisms driving both antibody induction/production and neural injury, even more specific targets for therapeutic intervention may emerge.

Key points

  • Patients with paranodal/nodal antibodies (PNAbs) have a distinct clinical phenotype, with different electrophysiological and histological characteristics compared with seronegative chronic inflammatory demyelinating polyneuropathy (CIDP).

  • PNAb+ patients appear less responsive to standard first-line treatments for CIDP/Guillain-Barré syndrome (GBS) (in particular intravenous immunoglobulin), and are likely to respond to rituximab clinically and serologically, although this is currently based on low-quality evidence.

  • Patients with an acute, subacute or ‘atypical’ (including treatment resistant) presentation of suspected GBS or CIDP should be prioritised for PNAb testing, and we advocate early testing of all patients with suspected immune-mediated neuropathies.

  • Prolonged suppression of antibody titres may be associated with sustained clinical remission, but the reverse is not necessarily true; we need more data to establish the prognostic implications and clinical utility of serial antibody measurements.

Further reading

  • Pascual-Goñi E, Martín-Aguilar L, Querol L. Autoantibodies in chronic inflammatory demyelinating polyradiculoneuropathy. Curr Opin Neurol 2019;32(5):651–7.

  • Uncini A, Vallat JM. Autoimmune nodo-paranodopathies of peripheral nerve: The concept is gaining ground. J Neurol Neurosurg Psychiatry 2018;89(6):627–35.

  • Manso C, Querol L, Mekaouche M, et al. Contactin-1 IgG4 antibodies cause paranode dismantling and conduction defects. Brain 2016.

  • Manso C, Illa I, Devaux JJ, et al. Anti-neurofascin-155 IgG4 antibodies prevent paranodal complex formation in vivo. 2019;129(6):2222–36.

Ethics statements

Ethics approval

Patient samples and clinical data were collected under Research Ethics Committee approval number 14/SC/0280.



  • Twitter @Neuroconference

  • Contributors JF and SR wrote the original manuscript. TV and SK reviewed and contributed to subsequent revisions of the manuscript.

  • Funding The funding was provided by the Medical Research Council (Clinician Scientist Fellowship awarded to Simon Rinaldi, MR/P008399/1) and the GBS/CIDP Foundation International Benson Fellowship awarded to Janev Fehmi (1709HM001/SB17).

  • Competing interests None declared.

  • Provenance and peer review Provenance and peer review. Commissioned. Externally peer reviewed by Hugh Willison, Glasgow, UK.

Linked Articles

  • Editors’ commentary
    Phil E M Smith Geraint N Fuller

Other content recommended for you