Article Text

This article has a correction. Please see:

Download PDFPDF

Paraneoplastic ophthalmoplegia and subacute motor axonal neuropathy associated with anti-GQ1b antibodies in a patient with malignant melanoma
  1. L Kloos1,
  2. P Sillevis Smitt1,
  3. C W Ang2,
  4. W Kruit3,
  5. G Stoter3
  1. 1Department of Neurology, Erasmus University Medical Centre, Rotterdam, Netherlands
  2. 2Department of Medical Microbiology and Infectious Diseases, Erasmus University Medical Centre
  3. 3Department of Medical Oncology, Erasmus University Medical Centre
  1. Correspondence to:
 Dr Peter Sillevis Smitt, Department of Neurology, Erasmus University Medical Centre, Dr Molewaterplein 40, 3015 GD Rotterdam, Netherlands; 


A 68 year old woman developed oculomotor paresis shortly after metastatic progression of her melanoma was discovered. She was then immunised with the tumour antigen MAGE-3 in combination with an immunological adjuvant. During immunisation her symptoms worsened and she developed severe, predominantly proximal axonal motor neuropathy and became bedridden. IgM antibodies against gangliosides GM2, GD3, and GQ1b were detected in serum obtained two weeks before and nine weeks after the onset of symptoms. Immunohistochemically, the patient’s IgM reacted with the tumour and co-localised with GQ1b. She improved neurologically following steroid treatment and became ambulatory.

  • paraneoplastic syndrome
  • ophthalmoplegia
  • malignant melanoma
  • motor 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.

Malignant melanoma is a potentially immunogenic tumour, and the development of melanoma vaccines has been an important line of research. Tumour specific antigens used as targets of melanoma immunotherapy are often surface glycoproteins that share immunogenic similarities with glycoproteins on the surface of normal melanocytes and cells in the central and peripheral nervous systems. The immune response against these melanoma antigens may also cross react with normal melanocytes or neurones, resulting in several clinical symptoms such as vitiligo, inappropriate secretion of antidiuretic hormone (SIADH), and chronic inflammatory demyelinating polyneuropathy (CIDP). We describe a patient with malignant melanoma who developed oculomotor paresis followed by a subacute motor axonal neuropathy associated with antiganglioside antibodies suggesting cross reactivity (“molecular mimicry”) between melanoma and peripheral nerve antigens.


The patient had been diagnosed with a malignant melanoma on her left foot in 1983. This was surgically excised (Breslow thickness 1.9 mm). In March 2000, at the age of 68, she complained of night sweating and weight loss of 17 kg over the previous six months. Computed tomography (CT) showed extensive retroperitoneal and left iliac lymphadenopathy. A CT guided biopsy of an iliac lymph node was undertaken and pathological examination showed malignant melanoma. Reverse transcriptase polymerase chain reaction (RT-PCR) showed tumour expression of the tumour antigen MAGE-3, and the patient was included in an immunotherapy trial. The treatment consisted of three intramuscular vaccinations at weekly intervals with 300 μg MAGE-3 recombinant protein in combination with immunological adjuvant.

In June, one week before the first vaccination, she developed double vision. Two weeks later she was referred to the neurology clinic. Examination revealed fluctuating external ophthalmoplegia. Motor and sensory examination was normal apart from absent deep tendon reflexes.

One month later (five weeks after the first vaccination), she developed rapidly progressive and predominantly proximal motor weakness: MRC muscle strength grade 3 in the deltoids and hip flexors, grade 4 in biceps and triceps, and grade 5 distally; weakness in the neck flexors was MRC grade 2. Sensory examination remained normal. The patient was admitted to our hospital for further investigation.

Cranial and spinal magnetic resonance imaging (MRI) before and after gadolineum administration was normal. The CSF was acellular with raised protein (1.9 g/l) and repeatedly negative cytology (on six occasions). Other laboratory studies were unremarkable including creatine kinase, calcium, and thyroid function studies. The erythrocyte sedimentation rate was 68 mm/hour. Antibodies to acetylcholine receptor and voltage gated calcium channels were negative and no antibodies were detected against the paraneoplastic Hu, CV-2, Ri, Yo, or amphiphysin antigens. A neostigmine provocation test produced no improvement in her symptoms. Electrophysiological studies showed normal motor and sensory nerve conduction velocities with normal or only mildly decreased motor amplitudes. There were no signs of dispersion or conduction block. Needle electromyography showed widespread acute denervation changes in the proximal muscles and mild denervation in the distal muscles. Low frequency (3 Hz) and high frequency (20 Hz) repetitive nerve stimulation did not change the compound muscle action potentials. Because of the proximal distribution of the weakness, a deltoid muscle biopsy was taken. Pathological examination did not show myelitis or myopathy. Sural nerve biopsy showed signs of axonal degeneration without inflammatory cells or immunoglobulin deposits.

The MAGE-3 vaccinations were stopped and she was treated with intravenous immunoglobulin (0.4 g/kg/d for five days). Despite this, her weakness deteriorated and she developed dysphagia and dysarthria. Subsequent steroid treatment (20 mg dexamethasone daily) resulted in a remarkable improvement in strength and bulbar function within two days. She was discharged from hospital on 12 mg dexamethasone a day with a gradual taper. There was continuing improvement in her symptoms and after six weeks she was able to walk again without help. Because of relapsing diplopia she was treated with five plasma exchanges, resulting in subjective improvement. She has remained steroid dependent (dexamethasone 6 mg/d).


Serum samples were obtained from the patient two weeks before the onset of symptoms (three weeks before the vaccination was started) and nine weeks after the onset of symptoms (eight weeks after the vaccination) . Both samples were analysed for the presence of antiganglioside antibodies against GM1, asialo-GM1, GM2, GM3, GD1a, GD1b, GD3, GT1b, and GQ1b with an enzyme linked immunosorbent assay (ELISA) and confirmed with thin layer chromatography as described before.1 Serum was also tested for the presence of IgM and IgA antibodies against Campylobacter jejuni, cytomegalovirus, Epstein-Barr virus, and Mycoplasma pneumoniae, as described.2 Indirect immunofluorescent examination was undertaken on formalin fixed, paraffin embedded sections of the primary melanoma and the metastasis.3 In brief, the sections were prepared for double immunofluorescent labelling by microwave preparation (five minutes at 900 W) in citric acid (pH 6) followed by five minutes of 0.1% pronase treatment at 37°C. The patient’s serum was diluted 1:200 and polyclonal rabbit anti-GQ1b, R2327E4 diluted 1:100 was added, followed by rhodamine and fluorescein labelled secondary antibodies. Antibody specificity was determined by incubation with a non-immune serum and by leaving out the primary serum. Images were analysed and photographed using a confocal fluorescent microscope.


Thin layer chromatographic overlay of the patient’s serum to multiple purified gangliosides revealed strong and specific binding of IgM with GM2 (titre 200), GQ1b (titre 400), and GD3 (titre 200). The patient’s serum did not react with the gangliosides asialo-GM1, GM1, GM3, GD1a, and GT1b. Weak background reactivity with GD1b was observed. The titres of the pre- and postvaccination samples (obtained two and nine weeks after the onset of symptoms, respectively) were the same. There was no IgG reactivity with any of the gangliosides tested. The patient’s serum did not contain monoclonal bands as determined by immunoelectrophoresis. No evidence for recent infection with Campylobacter jejuni, cytomegalovirus, Epstein-Barr virus, or Mycoplasma pneumoniae could be detected.

On indirect immunofluorescence, the patient’s IgM reacted with many of the tumour cells, as shown in fig 1A (fluorescein filter). Antiserum to GQ1b reacted with many of the same cells (fig 1B), as visualised with the rhodamine filter. Co-localisation is confirmed with double exposure in yellow (fig 1C).

Figure 1

The patient’s IgM immunoreacts with many of the tumour cells visualised in green with fluorescein filters (panel A). Anti-GQ1b antiserum co-localises with most of the same cells as shown in red with rhodamine filters (panel B). Co-localisation is confirmed in yellow (panel C) by double exposure from the superimposition of red and green.


The patient presented with external ophthalmoplegia in combination with areflexia and a high CSF protein concentration, with a normal CSF cell count and anti-GQ1b antibodies. She subsequently developed a severe subacute motor axonal neuropathy which made her bedridden within two months and was accompanied by involvement of the lower cranial nerves.

The patient’s neuropathy developed shortly after the diagnosis of recurrent melanoma and before vaccination, suggesting a paraneoplastic aetiology. Neither ophthalmoplegia nor subacute motor axonal neuropathy has previously been reported as paraneoplastic syndromes associated with melanoma.5 In this patient, other cancer related causes such as leptomeningeal metastases and direct invasion of the peripheral nerves by the tumour were excluded.

Most paraneoplastic neurological syndromes are considered autoimmune disorders caused by an immune response directed against antigens in the tumour which subsequently (cross) react with the same or similar epitopes in the nervous system. The patient had IgM autoantibodies in her serum which reacted with the gangliosides GM2, GD3, and GQ1b. All three antigens are immunogenic and are expressed on melanoma tumours and cell lines, suggesting that the tumour may have triggered the production of antibodies. Indeed, the patient’s IgM also reacted with the melanoma tissue, co-localising with GQ1b. An immune aetiology of the neuropathy is further suggested by the remarkable, albeit partial, response to steroid treatment.

The importance of each of the target glycolipids for the pathophysiological mechanism remains unclear. Anti-GQ1b reactivity is strongly associated with oculomotor symptoms in patients with immune mediated neuropathies.6 Moreover, in 20% of patients with Guillain-Barré syndrome and anti-GQ1b antibodies, the EMG changes are mainly axonal, as in our patient.7 In vitro models have clearly shown that anti-GQ1b antibodies disrupt the neuromuscular junction, resulting in breakdown of axonal terminals; this suggests that they can be pathogenic in vivo.8 Anti-GM2 IgM antibodies can also affect neuromuscular transmission.9 A pathogenic role of these antibodies in our patient is less likely, because anti-GM2 antibodies are associated with demyelinating neuropathies and their specificity is questioned. Anti-GD3 antibodies most probably result from cross reactive epitopes with GQ1b.

Other forms of neuropathy have been described in association with melanoma, including five reported cases of chronic inflammatory demyelinating polyneuropathy (CIDP).3,10 A further pathophysiological relation between melanoma and neuropathy is suggested by reports of demyelinating neuropathy following vaccination with melanoma lysates.11 In contrast to the previously reported cases, the neuropathy in our patient was strictly motor, axonal, and accompanied by upper and lower cranial nerve involvement.

The relation between the vaccinations and worsening of the neuropathy remains unclear. A direct relation with immunity directed against MAGE-3 is highly unlikely because MAGE-3 is a tumour specific antigen that is not expressed in the nervous system.12 Furthermore, the neuropathy and antiganglioside antibodies were clearly present before vaccination, and the titres of the antibodies were not influenced by the vaccinations. It is difficult to conclude whether the evolution of symptoms was spontaneous, or whether concomitant vaccination played an indirect role. Clearly, stimulation of the immune system—whether through an infectious process or through vaccination—could theoretically promote autoimmune phenomena by causing a systemic increase of inflammatory cytokines. Although there is no evidence that MAGE-3 vaccination caused the worsening of the neuropathy, our observation suggests the need for caution when inducing immune responses in patients with ongoing autoimmune symptoms.


We gratefully acknowledge technical assistance by Esther Hulsenboom, Hans Vuik, and Adriaan Houtsmuller.



  • L Kloos is presently at the Neurology Department, Leyenburg Ziekenhuis, Den Haag, The Netherlands

  • Competing interests: none declared

Linked Articles

  • Correction
    BMJ Publishing Group Ltd