Article Text

Download PDFPDF

Research paper
Anti-MAG antibodies in 202 patients: clinicopathological and therapeutic features
  1. Juliette Svahn1,
  2. Philippe Petiot1,
  3. Jean-Christophe Antoine2,
  4. Christophe Vial1,
  5. Emilien Delmont3,
  6. Karine Viala4,
  7. Andreas J Steck5,
  8. Armelle Magot6,
  9. Cecile Cauquil7,
  10. Aline Zarea8,
  11. Andoni Echaniz-Laguna9,
  12. Ruxandra Iancu Ferfoglia10,
  13. Antoine Gueguen11,
  14. Laurent Magy12,
  15. Jean-Marc Léger13,
  16. Thierry Kuntzer14,
  17. Karine Ferraud2,
  18. Arnaud Lacour2,
  19. Jean-Philippe Camdessanché2
  20. The Francophone anti-MAG cohort Group
  1. 1 Electroneuromyography and Neuromuscular Department, University Hospital Lyon, Lyon, France
  2. 2 Department of Neurology, University Hospital Saint-Etienne, Saint-Etienne, France
  3. 3 Department of Neurology, University Hospital Timone, Marseille, France
  4. 4 Department of Neurophysiology and Neuropathology, University Hospital Pitié-Salpêtrière, Paris, France
  5. 5 Department of Neurology, University Hospital Basel, Basel, Switzerland
  6. 6 Neuromuscular Reference Center, University Hospital Nantes, Nantes, France
  7. 7 Department of Neurology, University Hospital Bicêtre, Le Kremlin Bicêtre, France
  8. 8 Neuromuscular Competence Center, University Hospital Rouen, Rouen, France
  9. 9 Department of Neurology, University Hospital Strasbourg, Strasbourg, France
  10. 10 Electroneuromyography and Neuromuscular Disorders Unit, University Hospital Geneva, Geneva, Switzerland
  11. 11 Department of Neurology, Fondation Ophtalmologique A. de Rothschild, Paris, France
  12. 12 Department of Neurology, University Hospital Limoges, Limoges, France
  13. 13 Department of Neurology, University Hospital Pitié-Salpêtrière, Paris, France
  14. 14 Department of Clinical Neurosciences, Nerve-Muscle Unit, Neurology Service, Lausanne University Hospital (CHUV), Lausanne, Switzerland
  1. Correspondence to Dr Juliette Svahn, Electroneuromyography and Neuromuscular Department, University Hospital Lyon, 69677 Lyon, Bron Cedex – France; juliette.svahn{at}


Objective To assess the clinicopathological and therapeutic features of patients with low (≥1000 to <10 000 Bühlmann Titre Units) (BTU), medium (10 000–70 000) or high (≥70 000) anti-myelin-associated glycoprotein (anti-MAG) antibody titres.

Methods We retrospectively and prospectively analysed standardised report forms and medical records of 202 patients from 14 neuromuscular centres.

Results Mean age at onset and mean time between symptom onset to last follow-up were respectively 62.6 years (25–91.4) and 8.4 years (0.3–33.3). Anti-MAG antibody titres at diagnosis were low, medium or high in 11%, 51% and 38% of patients. Patients presented with monoclonal gammopathy of undetermined significance in 68% of cases. About 17% of patients presented with ‘atypical’ clinical phenotype independently of anti-MAG titres, including acute or chronic sensorimotor polyradiculoneuropathies (12.4%), and asymmetric or multifocal neuropathy (3%). At the most severe disease stage, 22.4% of patients were significantly disabled. Seventy-eight per cent of patients received immunotherapies. Transient clinical worsening was observed in 12% of patients treated with rituximab (11/92). Stabilisation after rituximab treatment during the 7–12-month follow-up period was observed in 29% of patients. Clinical response to rituximab during the 6-month and/or 7–12-month follow-up period was observed in 31.5% of patients and correlated with anti-MAG titre ≥10 000 BTU.

Conclusion Our study highlights the extended clinical spectrum of patients with anti-MAG neuropathy, which appears unrelated to antibody titre. Besides, it may also suggest beneficial use of rituximab in the early phase of anti-MAG neuropathy.

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.


Neuropathy associated with anti-myelin-associated glycoprotein (anti-MAG) IgM antibodies produced in the setting of a monoclonal gammopathy is typically a chronic, progressive, predominantly sensory distal demyelinating neuropathy, with ataxia and sometimes postural tremor.1–4

Caudie et al described 97% of sensitivity and 86% of specificity with diagnostic cut-off at 1000 Bühlmann Titre Units (BTU) to identify a demyelinating polyneuropathy and IgM monoclonal gammopathy.5

The clinical and electrophysiological spectrum of patients with high titre of anti-MAG antibodies ranges from paucisymptomatic patients to patients with predominantly motor non-length-dependent demyelinating neuropathy evocative of chronic inflammatory demyelinating polyneuropathy (CIDP). In contrast, some patients present the ‘typical’ electroclinical phenotype but low anti-MAG titre at diagnosis.

Concerning therapeutic management, there is actually no evidence of significant benefits from immunotherapies in anti-MAG neuropathy.6–9 Nevertheless, rituximab may provide sustained benefits in some patients.10–15

Our multicentre study focused on the clinicopathological phenotype and the response to treatment according to anti-MAG antibody titre.


Patients and study design

Anti-MAG antibodies were ascertained by ELISA from Bühlmann. Between March 2014 and April 2016, data of 202 patients with monoclonal IgM gammopathy and anti-MAG antibody titres >1000 BTU were collected retrospectively (n=63), prospectively (n=10) or ambispectively (n=129) from 14 neuromuscular centres. Forty patients were excluded because of insufficiently detailed clinical data regarding onset, evolution, clinical phenotype and post-treatment evaluation. Clinical and paraclinical data were collected by each investigator per predefined protocol and completed and updated if necessary by the lead author with the patient’s medical record or queries during the study period. Information leaflets were given or mailed to the patients.

Clinical features

Data were collected concerning anti-MAG antibody titre; age at onset, at anti-MAG antibodies detection, at last follow-up, at death; family history of neuropathy or monoclonal gammopathy; neurological symptoms at onset and follow-up; and clinical examination. Neuropathy onset was classified by each investigator as acute (progression in <1 month), subacute (progression in >1 but <6 months) and progressive (progression in ≥6 months). Data collected at the most severe disease stage concerned the last evaluation in untreated or non-responder patients and the stage before effective treatment for patients who presented with sustained clinical improvement or stabilisation. The Overall Neuropathy Limitations Scale (ONLS) score16 and the modified functional impairment scale score, ranging from 0 to 517 at the most severe disease stage were ascertained. Patients were considered as significantly disabled if the score was >2 (3=inability to perform some manual activities/walking independently with support).

‘Typical and variant’ phenotypes were defined as follows by the lead author. Typical phenotype: Patients presenting with sensory ataxic distal polyneuropathy, with or without neuropathic tremor and progressive distal motor deficits in the lower limbs (LL); patients with asymptomatic/paucisymptomatic sensory polyneuropathy at the more severe disease stage (ONLS score=0, ‘Inflammatory Neuropathy Cause and Treatment (INCAT)’ sensory sum score (ISS) ≤4), sensory or sensorimotor length-dependent polyneuropathy without ataxia.

‘Atypical’ phenotype: Patients presenting with Guillain-Barré-like syndrome, chronic sensorimotor polyradiculoneuropathy (defined clinically by LL areflexia, marked and rapidly progressive distal or non-length-dependent sensorimotor deficit and chronic evolution), small fibre neuropathy, asymmetric or multifocal neuropathy, and patients with associated motor neuron disease.


Electrodiagnostic (EDX) tests performed at diagnosis before any specific treatment were preferentially analysed. All patients underwent motor (tibial, peroneal, median, ulnar) and sensory (sural, median, ulnar, radial) nerve conduction studies. The neuropathy was considered as demyelinating according to the European Federation of Neurological Societies and Peripheral Nerve Society criteria.18 Prolonged median distal motor latency was not retained as indicating demyelination in the presence of carpal tunnel syndrome symptoms and/or abolished median sensory nerve action potential (SNAP). Terminal latency index ≤0.25 in the median and ulnar nerves indicated prominent distal slowing.19 20

Biology and histology

Cerebrospinal fluid (CSF) and nerve biopsy features when performed were analysed. Neuropathological aspects were considered as suggestive of IgM anti-MAG neuropathy in the presence of demyelination on semithin sections, associated with IgM deposits in the myelin sheaths by direct immunohistochemistry and/or widening of external myelin lamellae by electron microscopy.21 22

Titres of anti-MAG antibody at diagnosis (ELISA Bühlmann) were classified as low (≥1000 and <10 000 BTU), medium (≥10 000 and <70 000 BTU) and high (≥70 000 BTU). Waldenström’s macroglobulinemia (WM) during the disease course was diagnosed in the presence of ≥10% clonal lymphoplasmacytic cells in bone marrow. Monoclonal gammopathy of undetermined significance (MGUS) was addressed in patients without evidence of WM or related plasma cell proliferative disorder. The causal relationship between anti-MAG antibodies and neuropathies, according to anti-MAG titre, EDX features and nerve biopsy, was considered as ‘uncertain’ in cases of anti-MAG titres <10 000 BTU, absence of EDX criteria for definite demyelinating neuropathy and axonal neuropathological pattern.

Types of treatment

Treatments used (intravenous immunoglobulins (IVIg), plasma exchange, corticosteroids, rituximab, cyclophosphamide, chlorambucil and fludarabine) were inventoried. We assessed for rituximab and immunosuppressants if treatment indication was determined by neurological condition, haematological condition or both. We evaluated retrospectively therapeutic response during the 6-month and 7–12-month post-treatment periods. An objective clinical response was retained if the patient presented improvement on one or more clinical scales: ≥1 point on the modified Rankin scale (mRS),23 ≥2 points in upper limbs or ≥1 point in LL on the ONLS, ≥1 point on the ataxia score in LL which ranges from 0 (normal) to 3 (immediate falling) and ≥2 points for the most affected hemibody on the ISS.24 A composite definition of responder patients was used due to variability of functional disability score assessed after treatment in each centre (mRS or ONLS).

In patients responding to rituximab associated or not to immunosuppressants, we assessed, if available, improvement of EDX parameters at ≥6 months, according to Lunn and Nobile-Orazio’s electrophysiological criteria: reappearance of sural SNAP or compound muscle action potential (CMAP) in previously unexcitable nerves or EDX changes in at least two nerves (>20% increase in motor or sensory nerve conduction velocities or >20% decrease in motor distal latencies) compared with baseline.25

Statistical analysis

Categorical variables were compared using Fisher’s exact test while continuous variables were compared using the Mann-Whitney and Kruskal-Wallis tests. All tests of hypothesis were two-sided and performed at significant level α=0.05. Logistic regression was realised to study factors associated with rituximab response. Statistical analyses were implemented with XLSTAT-Pro.


Clinical features

In total, 202 patients, 133 males (65.8%) (sex ratio 1.9) with a mean age at symptom onset of 62.6 years (range 25–91.4, SD 11.2), were included. Mean time to diagnosis of neuropathy associated to anti-MAG antibodies was 3.1 years (0–30, SD 4.3). Mean time between symptoms onset to the most severe disease stage and last follow-up were respectively 7.2 (0.01–30, SD 6.3) and 8.4 years (0.3–33.3, SD 6.1). Six patients died (3.2%). The status living or deceased could not be assessed for 10 patients. The causes of death were amyotrophic lateral sclerosis (ALS) (one), metastatic colorectal cancer (one), respiratory infection (two) and acute myeloid leukaemia (two). Two patients were siblings and presented with WM and anti-MAG neuropathy. There was no significant difference of patient’s clinical features at onset and at the most severe disease stage according to anti-MAG titre <10 000 (n=22) or ≥10 000 BTU (n=180) (see online supplementary data table 1).

Supplementary file 1

Thirty-four patients (17%) presented with ‘atypical’ phenotype (table 1) without significant differences according to anti-MAG antibody titres.

Table 1

Clinical phenotypes according to anti-MAG antibody titre and EDX features

Evolution of Guillain Barré Syndrome (GBS)-like cases was variable: Complete recovery after IVIg in one patient and improvement of motor impairment and ataxia with persistence of mild sensory polyneuropathy in two. Six patients presented multifocal neuropathies, leading sometimes to surgical nerve decompression at carpal tunnel, fibular head and elbow. In one patient, hereditary neuropathy with liability to pressure palsies was proven by a deletion in the PMP22 gene. Another patient with MGUS presented with subacute clinical and neurophysiological sensorimotor right sciatic nerve impairment without extrinsic compression. EDX revealed in addition a diffuse demyelinating neuropathy. MRI showed focal increased hyperintensity of the right sciatic nerve in T2-weighted sequences, suggesting inflammatory oedema. One patient with a family history of ALS developed ALS. This patient presented a mild sensory polyneuropathy with demyelinating features and widening of myelin lamellae in nerve biopsy.26 At the most severe disease stage, 22.4% of patients were significantly disabled.

Anti-MAG antibodies, biology and nerve biopsy

Detection of anti-MAG antibodies was performed before any treatment in 92.1% of the patients (n=186) or after treatment in 7.9% (n=16) (table 2).

Table 2

Biological and pathological features according to anti-MAG antibody titre

Anti-MAG titre at diagnosis was low (<10 000 BTU), medium (10 000–70 000) or high (≥70 000) in 11% (n=22), 51% (n=104) and 38% (n=76) of cases, respectively, with no difference between ‘typical/variant’ and ‘atypical’ phenotype. Six untreated patients presented with low anti-MAG titres after 2 (three) or several years (three). MGUS was encountered in 68% of cases, independently of anti-MAG titres (table 2). Anti-ganglioside antibodies were positive in 21.8% of the tested patients (29/133), mostly of IgM anti-GM1 type. Nineteen patients had a nerve biopsy (table 2).


In total, 170 EDX tests (84%) were realised before any treatment. EDX data were collected during follow-up for 2 untreated patients (0.1%), during treatment for 2 patients (0.1%) and after treatment for 28 patients (13.9%). The mean delay between onset of symptoms and EDX tests was 3.3 years (0–29.2, SD 4.6). EDX studies showed a pattern of demyelinating polyneuropathy with prolonged distal motor latencies predominating in LL and median nerves (see online supplementary data table 2).

Conduction blocks were infrequent (3.9%–12.5% according to the nerve studied) but sometimes observed outside nerve entrapment sites. SNAPs were frequently reduced or abolished: 90% for the sural nerves (355 recorded nerves), 95% for the median nerves (288 nerves), 93.5% for the ulnar nerves (247 nerves) and 70% for the radial nerves (224 nerves).

Thirty-two patients (15.8%) presented a neuropathy which did not qualify as definite demyelinating neuropathy, without correlation with anti-MAG titre or clinical phenotype (table 1). The mean delay between onset of symptoms and EDX tests for these patients was increased compared with the whole population: 5.9 years (0.1–29.2, SD 7.9). Among these patients, conduction studies revealed axonal sensorimotor neuropathies (5/32, mean delay of 11.1 years), purely prolonged distal median motor latencies with reduced SNAPs (2/32, mean delay of 1 year), abolished CMAPs in all limbs (2/32, mean delay of 12.8 years) and only mild delayed distal latencies in LL and median nerves (not subscribing to our criteria of demyelination) without axonal loss (two patients with high anti-MAG titres). These two patients were investigated soon after the neuropathy onset (mean delay 0.3 year). There was a slight progression of demyelinating parameters in one of them 1 year later. Twenty-one patients presented with abolished or extremely reduced CMAPs in the LL and prolonged distal median motor latencies (mean delay of 4.8 years). One patient with unexcitable nerves in all four limbs presented with prolonged latencies at blink reflex.

Clinicopathological synthesis

The causal relationship between anti-MAG antibodies and neuropathy was considered as uncertain in five patients with low anti-MAG titres (2.5% of the whole population and 22.7% in the group with anti-MAG titre <10 000 BTU) (figure 1). Two of these patients presented with IgM anti-GM1 and IgM anti-GQ1b antibodies, respectively.

Figure 1

Clinicopathological synthesis. Patients for which causal relationship between neuropathy and anti-myelin-associated glycoprotein (anti-MAG) antibodies appeared uncertain are shown in shadow boxes. *Suggestive of anti-MAG neuropathy (NP); demyelinating NP; axonal NP. EDX, electrodiagnostic.

Treatment features

The majority of our patients were treated with immunotherapies (78.2%) (table 3). Indications for immunotherapies were mostly because of neurological condition, except for combined immunosuppressants. Some patients received up to six lines of immunotherapy (mean=1.5, (0–6], SD 1.3). Rituximab and immunosuppressants were more frequently administered in patients with malignant haemopathies (n=65) compared with MGUS (n=137) (mean lines of therapy=1.2, (0–4], SD 0.7 vs0.85, (0–3], SD 0.85, p=0.002).

Table 3

Details of immunosuppressants, clinical responses and indications

Patients with MGUS received most often IVIg courses followed by rituximab in case of no response or progression of the disease. Conversely, rituximab and immunosuppressants were frequently administered as first-line treatment in patients with malignant haemopathies.

Rituximab was the most common treatment (n=92, 45.5% of the whole population). Stabilisation during the 7–12-month post-treatment period was observed in one-third of our patients treated with rituximab monotherapy. We retained objective clinical response in 31.5% of patients during the same period, with two patients displaying only transient improvement not maintained after the 6-month follow-up period. Patients improved in the mRS score (n=18), ONLS score (n=17), ISS (n=14) and ataxia score (n=11); improvement occurred in all scores in three patients, three scores (n=6), two scores (n=10) and one score (n=10).

Eleven patients (12%) presented with transient and reversible clinical worsening during or just after rituximab monotherapy. They rapidly had progressive worsening of paraesthesias, proprioceptive ataxia and/or appearance of distal motor impairment. Four patients were treated, subsequently to clinical worsening, with IVIg courses associated with corticosteroids (three) or corticosteroids orally alone. Patients presented with progressive clinical improvement over several weeks or months to reach pretreatment clinical status. We looked for clinical, EDX, biological parameters predictors of response to rituximab monotherapy. No correlation between response to rituximab and clinical severity, EDX parameters at baseline was found. Response to rituximab did not differ according to the associated haematological condition (9/30 responders in the group with malignant haemopathies vs 20/62 responders in the MGUS group). Response to rituximab correlated with anti-MAG titre ≥10 000 BTU (30/31 responders vs 50/61 non-responders, p=0.05). At 7–12-month follow-up, responder patients presented shorter symptom duration compared with non-responders, thought non-significant after logistic regression according to age, sex and previous immunosuppression (mean=3.6 years, (0.13–16.23], SD 3.7 in 29 responders vs 5.1, (0.13–17.6], SD 4.2, in 63 non-responders, p=0.06). IVIg courses brought transient objective clinical improvement in some patients during the 6-month post-treatment period (27.9%), not maintained in the 7–12-month post-treatment period (see online supplementary data table 3).

Among responder or stabilised patients after rituximab associated or not with immunosuppressants, 36 were assessed by EDX tests at ≥6 months. In 11 patients, an improvement of EDX parameters was observed (see online supplementary data figure 1). Seven patients were treated by rituximab alone, four patients by rituximab in association with other chemotherapies (chlorambucil, cyclophosphamide, fludarabine) with a mean delay between baseline EDX and post-treatment control EDX of 3.9 years (0.6–8.5, SD 2.7).


To date, our cohort of patients with anti-MAG antibodies is the largest analysed in the literature. Our study highlights the extended clinical phenotype encountered in anti-MAG neuropathy without significant difference according to anti-MAG antibody titre as underlined in other series or case reports.27–29 Anti-MAG neuropathy may provide significant disability in a subset of patient after several years.17 Death in our cohort was usually linked to infections in patients who received immunosuppressants and/or presented with haematological malignancies. Nobile-Orazio et al reported a death rate of 32% with a higher mean neuropathy symptoms duration than in our cohort (11.8 vs 8.4 years).17

Thirty-two patients (15.8%) presented a neuropathy which did not correspond to definite CIDP EDX criteria, independently of anti-MAG titre or clinical presentation.18 In most of these cases, it depended on abolished or extremely reduced CMAP in LL or all nerves, which did not permit to assess an underlying demyelination. This proportion reached 23% in Nobile-Orazio et al’s cohort2 and 17.5% in Chassande et al’s.30

Nerve biopsies highlighted that ‘typical’ anti-MAG pathological pattern may be observed in cases with anti-MAG titre <10 000 BTU at diagnosis. It appears as a valuable tool for the diagnosis of ‘atypical’ cases.

The probability that the neuropathy was directly related to anti-MAG antibodies appeared uncertain in 2.5% of our patients. ELISA may be less specific with titre between 1000 and 10 000 BTU, possibly because of some cross-reactivity with GM1 and disialosyl gangliosides.5

Concerning rituximab monotherapy, two controlled studies failed to capture rituximab benefits in primary endpoints notably on sensory scale for the last study.6 They, however, reported significant improvement of ISS in 8/20 and 4/13 patients after rituximab (40% and 31%) in per protocol analysis, respectively.6 31 Moreover, many uncontrolled studies have reported that 30%–67% patients respond to rituximab according to the outcomes used, which appears similar to our results.10–15 32

Response to treatment was not different according to haematological condition.15 Clinical prolonged stabilisation after rituximab may be considered as a partial response in this potentially progressive disabling pathology and appears as an interesting point to focus on long-term prospective and controlled study.

As noticed by Benedetti et al, we observed a trend between shorter symptom duration before treatment and response to rituximab.33 Kawagashira et al reported correlation between clinical response to rituximab and shorter disease duration or preservation of nerve fibre density in sural nerve biopsies before treatment.34 Gazzola et al reported that response to rituximab was significantly associated with subacute evolution and proximal weakness of the LL at the onset of the disease.14 We did not find any association between the response to rituximab and the presence of an ‘atypical’ clinical phenotype or acute onset of the neuropathy. Marked improvement of nerve conduction study has been reported in some patients with anti-MAG neuropathy.13 33 35

Transient and reversible clinical worsening after rituximab monotherapy has been described in few cases.36–38 The pathogenesis of this acute exacerbation may involve IgM flare, release of proinflammatory cytokines and increased permeability of the blood–brain or blood–nerve barrier to antibodies, complement or cytokines.38 Relationship between antibody titre and response to rituximab is debated with contradictory results and different methods for anti-MAG titre assessment.15 31 33

We chose to describe clinical data at the most severe disease stage to avoid bias incurred by sustained clinical improvement after treatment. Partially retrospective collection of data and use of composite scores to assess objective clinical response to treatment represent a weakness of our study. Nevertheless, it allowed us to assess clinical follow-up for >7 years in many patients, clinical evolution and therapeutic management.

In conclusion, in our series, clinicopathological phenotype did not appear to differ according to anti-MAG antibody titre and patients with low anti-MAG titre may have a ‘typical’ phenotype. Moreover, our results may suggest beneficial use of rituximab in the early phase of anti-MAG neuropathy. These data may warrant the development of another clinical trial, with relevant endpoints taking into account the extended clinical spectrum of this neuropathy and incorporating EDX parameters.


The authors thank the other members of the Francophone anti-MAG cohort Group who provided detailed clinical information: D Adams, UH Bicêtre; S Attarian, UH Marseille; A L Bedat-Millet, UH Rouen; F Bouhour, UH Lyon; C Boutte, UH Grenoble; G Chauplannaz, UH Lyon; R Costa, UH Pitié-Salpêtrière; P Devic, UH Lyon; C Grand, UH Lyon; G Jousserand, UH Lyon; T Lenglet, UH Pitié-Salpêtrière; P Lozeron, UH Bicêtre; T Maisonobe, UH Pitié-Salpêtrière; C Muntean, UH Pitié-Salpêtrière; Y Pereon, UH Nantes; J Pouget, UH Marseille.

They are grateful to Mrs Corinne Court for English reviewing.



  • Contributors JS, PP, JCA and JPC contributed to the study concept, analysis or interpretation of data. JS and KF contributed to the acquisition of data. JS, JPC, PP, JCA, AS, KV, AEL and TK contributed to drafting/revising of the manuscript. All other authors provide clinical care to patients and detailed clinical information for the study.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval The project was approved by the Advisory Committee for Data Processing in Health Research (CCTIRS).

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement All relevant data are presented in the article. Further information is available from the corresponding author.