Background Immunological mechanisms are suspected in sensory neuropathy (SN) occurring with systemic autoimmune diseases and in some idiopathic cases, but so far there are no antibodies (Abs) identifying these neuropathies.
Methods In the search for such specific antibodies, serum samples were collected from 106 patients with SN of these 72 fulfilled the diagnosis criteria of sensory neuronopathy (SNN) and 211 control subjects including patients with sensorimotor neuropathies, other neurological diseases (ONDs), systemic autoimmune diseases and healthy blood donors.
Results In the first step, a protein array with 8000 human proteins allowed identification of the intracellular domain of the fibroblast growth factor receptor 3 (FGFR3) as a target of Abs in 7/16 SNN and 0/30 controls. In the second step, an ELISA method was used to test the 317 patients and controls for anti-FGFR3 Abs. Abs were detected in 16/106 patients with SN and 1/211 controls (p<0.001). Among the 106 patients with SN, anti-FGFR3 Abs were found in 11/38 patients with autoimmune context, 5/46 with idiopathic neuropathy and 0/22 with neuropathy of other aetiology (p=0.006). The only control patient with anti-FGFR3 Abs had lupus and no recorded neuropathy. Sensitivity, specificity, and positive and negative predictive values of anti-FGFR3 Abs for a diagnosis of idiopathic or dysimmune SN were 19%, 99.6%, 94.1% and 77.3%, respectively. A cell-based assay confirmed serum reactivity against the intracellular domain of FGFR3. The neuropathy in patients with anti-FGF3 Abs was non-length dependent in 87% of patients and fulfilled the criteria of probable SNN in 82%. Trigeminal nerve involvement and pain were frequent features.
Conclusions A anti-FGFR3 Abs identify a subgroup of patients with SN in whom an underlying autoimmune disorder affecting sensory neurons in the dorsal root and trigeminal nerve ganglia is suspected.
Statistics from Altmetric.com
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.
Peripheral sensory neuron disorders are characterised by involvement of sensory neuron cell bodies or their axons.1 ,2 The term sensory neuronopathies (SNN) is restricted to cases in which the sensory neuron cell body is the target of the disease. Presence of an inflammatory cell reaction in dorsal root ganglia (DRG) in paraneoplastic SNN,3–5 in SNN with HIV infection,6 ,7 Sjögren's syndrome8 ,9 or unclassified connective diseases, and in some idiopathic cases suggests that a subgroup of SNN is driven by an immune response.10–14 Even if clinical criteria help to differentiate SNN from other neuropathies, they do not allow an aetiological diagnosis.15 ,16 Thus, there is a need for biological markers that can distinguish immune-mediated SNN from other aetiologies. Autoantibodies reactive with sensory neurons have been identified in paraneoplastic SNN.15 ,17 ,18 So far, studies in patients with idiopathic or Sjögren's syndrome-associated SNN have given inconclusive results.19–23 We present a study of diagnostic accuracy of serum antibodies against the intracellular domain of fibroblast growth factor receptor 3 (FGFR3; anti-FGFR3 Abs) to identify a subgroup of patients who have developed a sensory neuron disorder.
Study design and patients
We retrospectively studied serum samples from 106 eligible patients referred for the diagnosis of clinically pure sensory neuropathy (SN) between 1 January 2000 and 1 January 2012 (figure 1 and table 1). Seventy-two patients met our published SSN diagnosis criteria15 (see online supplementary materials) and 34 did not. Among SNN, the neuropathy was paraneoplastic in 6, toxic in 10 and genetic/familial in 4, associated with a dysimmune context in 20 and idiopathic in 32 patients. Among the 34 patients with non-SNN SN, the neuropathy was idiopathic in 14, associated with a dysimmune context in 18 and toxic in 2 (chemotherapy). The dysimmune context in patients with the SN included Sjögren's syndrome, systemic lupus erythematosus (SLE), lupus anticoagulant, monoclonal gammopathy and inflammatory rheumatism or bowel diseases. The control group consisted of 41 patients with sensorimotor neuropathy (see table 1 for details), 59 with other neurological diseases (ONDs), 51 patients with systemic autoimmune diseases (mostly SLE, inflammatory rheumatism or Sjögren's syndrome) and 65 healthy blood donors. All the control sera were collected over the same period of time from eligible patients and then selected randomly among our serum collection.
The clinical and electrophysiological data (see online supplementary materials) of patients with anti-FGFR3 Abs and SNN controls were collected by reviewing the case records. Thirty-six previously reported patients with SNN and anti-Hu antibody15 were used for comparison of the clinical pattern of the neuropathy. The study was approved by the ethical committee of the University Hospital of Saint-Etienne. Written consent for the studies was obtained from the patients and for those who were lost for follow-up, the ethical committee agreed for the use of their serum sample and clinical data.
Anti-FGFR3 antibody identification and detection
Sera diluted 1:500 were tested on a human ProtoArray V.4.2 (Invitrogen, Carlsbad, New Mexico, USA); bound IgGs were revealed with Alexa Fluor 647-labelled goat antihuman IgG antibodies by using appropriate laser activation (see online supplementary materials). These microarrays contained 8000 GST-tagged human proteins expressed in Sf9 insect cells (for details see online supplementary table S1). The slides were scanned and the images analysed using ProtoArray Prospector V.5.0 software. The Chebyshev inequality p value and the z-score were used to identify serum reactivity specific to the SNN patients, from controls and background noise.
Plates were coated overnight at 4°C with 1 µg/mL of the purified recombinant intracellular domain of human FGFR3 (Invitrogen) or full-length FGFR1 or FGFR2 (Novus Biological) with blocking buffer. Serum samples diluted 1:50 were incubated overnight at 4°C and bound IgGs were revealed with horseradish peroxidase-labelled antihuman IgG antibodies. In each plate, controls included blank wells containing the reaction products without human serum and the secondary antibody, and a panel of 10 healthy blood donor serum samples. The z-score was used to identify serum positivity. As sera were tested several times, the mean value of the z-score for the different tests was used for the analysis. A mean z-score >5 was considered positive.
Sera diluted 1:20 were incubated overnight at 4°C with HEK293 cells transfected with the wild-type plasmid or plasmids coding for EGFP-tagged full-length human FGFR3, the extracellular domain, the intracellular domain or the TRK1 or TRK2 subdomains of the intracellular domain of FGFR3 (see online supplementary figure S1) fixed 4 min with 5% paraformaldehyde and revealed with TRITC-labelled goat antihuman IgG antibodies (Dako).
Immunocytochemical study on sensory neurons
Cultivated rat DRG sensory neurons were incubated with human purified IgGs diluted 1/100 and a rabbit antibody reacting with the intracellular domain of FGFR3 or CRMP5, and revealed with appropriate secondary antibodies.
The distribution of anti-FGFR3 Abs in the different groups of patients was analysed by the χ2 test and Fisher's exact test. Logistic regression was used to characterise the clinical data of patients with SN and anti-FGFR3 Abs.
In the first step, serum samples from 16 patients with non-paraneoplastic SNN (4 with associated dysimmune disorders) and 30 controls (15 healthy blood donors, 8 patients with anti-Hu antibody-associated SNN and 7 with sensorimotor neuropathies) were screened using human ProtoArray V.4.2. Using the stringent criteria of reactivity restricted to the SNN group, a level of reactivity >1010 relative fluorescence units (RFU) in the SNN group and <900 RFU in the control group and a z-score >5, antibodies binding to the intracellular domain of FGFR3 (NCBI Reference Sequence: NP_000133.1) distinguished SNN from the control samples and were present in 7/16 patients with SNN (figure 2A).
In the second step, to confirm these results we tested the sensitivity and specificity of anti-FGFR3 Abs in the 317 patients and controls using an ELISA with plates coated with the FGFR3 intracellular domain (figures 1 and 2B). Anti-FGFR3 Abs were detected in 16/106 patients with SN and 1/211 controls (p<0.001 Fisher's exact test). All the seven sera positive by protoarray were also positive by ELISA, and all the nine sera negative by protoarray were negative by ELISA. The proportion of patients with FGFR3 Abs was 9/72 among patients fulfilling the SNN criteria and 7/34 among patients with SN not fulfilling the SNN criteria (p=0.406, χ2 test). Among the 106 patients with SN, anti-FGFR3 Abs were found in 11/38 with an autoimmune context, in 5/46 with idiopathic neuropathy and in 0/22 with another origin of the neuropathy (p=0.006, χ2 test; figure 2C). Among the 211 controls, the only patient with anti-FGF3 Abs had SLE and no recorded neuropathy. Sensitivity, specificity, and positive and negative predictive values of anti-FGFR3 Abs were calculated among the 317 patients using patients with idiopathic or dysimmune neuropathy as the positive condition, and patients with other causes of SN and controls as the negative condition (see online supplementary table S2). These were 19%, 99.6%, 94.1% and 77.3%, respectively.
To compare the sensitivity and specificity of the ELISA method and the immunohistochemical method using HEK293 for expressing the FGFR3 intracellular domain (cell-based assay), sera from 12/17 patients with anti-FGFR3 Abs directed against the FGFR3 intracellular domain in the ELISA and from 10 patients without these antibodies (patients with SNN, patients with SLE and blood donors) were analysed. The five remaining sera positive by ELISA were not tested because they had been exhausted by the previous experiments. The results showed immunoreactivity in 5/12 patients with ELISA-detected anti-FGFR3 Abs and in 0/10 without these antibodies. Among the 22 sera tested by cell-based assay, positive sera had a higher level of antibody by ELISA (mean z-score 10.7±4.2) than those that were negative (mean z-score 4.7±6.2; p=0.007 Mann-Whitney U test). As a whole, these results indicate a greater sensitivity of the ELISA method and show that the two methods have the same specificity.
To test serum reactivity against the different subunits of FGFR3 and to check the localisation of the epitope on the FGFR3 protein, serum samples from four patients with anti-FGFR3 Abs detected by ELISA and cell-based assay from four healthy blood donors were tested on HEK293 cells transfected with plasmids coding for different FGFR3 constructs linked to EGFP. Transfection with the different plasmids induced variable shape modifications of HEK293 cells (figure 3A). In addition, TRK1 and TRK2 were frequently cytotoxic and prevented cell growth. See online supplementary table S3 and figure S3A that summarise the results for the four patients and four controls. None of the control sera reacted with any of the transfected cells, while three of the four SNN sera reacted with the full-length protein and four with the intracellular domain, 1 of which also reacted with the TRK1 and TRK2 subdomains. All four sera reacted with the intracellular domain of FGFR3, in agreement with the ELISA results. None of them reacted with the extracellular domain.
To test the specificity of serum reactivity by the cell-based assay, one of the positive sera was adsorbed overnight with the intracellular domain of FGFR3. Immunoadsorption extinguished the reactivity (figure 3B).
Cross-reactivity of anti-FGFR3 antibody-containing sera with other FGFR proteins
FGFR3 belongs to a family of four homologous proteins (FGFR1–4). FGFR4, which was included in the protein array, was not recognised by sera from patients with SNN with anti-FGFR3 Abs. To test reactivity against FGFR1 and FGFR2, an ELISA was performed using serum samples from 10 patients with anti-FGFR3 Abs, 7 patients without anti-FGFR3 Abs and 7 blood donors using recombinant full-length FGFR1 and FGFR2. None of the 10 sera with anti-FGFR3 Abs reacted with FGFR1, while 2 cross-reacted with FGFR2. None of the control sera reacted with FGFR1 or FGFR2.
Expression of FGFR3 by sensory neurons
Expression of FGFR3 by sensory neurons has seldom been studied. We examined whether FGFR3 was expressed by rat DRG sensory neurons in an immunohistochemical study and found that rabbit antibodies directed against the intracellular domain of rat FGFR3 bound to sensory neurons in paraformaldehyde-fixed cultures of rat sensory neurons, DRG and developing trigeminal nerve (see online supplementary figures S3 and S4). Both large and small sensory neurons expressed FGFR3.
IgG reactivity of patients with anti-FGFR3 Abs on sensory neurons
Purified IgGs of a patient with FGFR3 Abs reacted with the cytoplasm of sensory neurons shortly permeabilised with 0.1% Triton X, as identified by double labelling with anti-CRMP5 antibody and co-localised with FGFR3 (figure 4) while a control serum gave no specific labelling. Although FGFR3 was expressed in the nucleus and cytoplasm, IgGs only labelled the cytoplasm suggesting that they did not reach their target in the nucleus.
Clinical pattern of the SN in patients with anti-FGFR3 Abs
SN was present in 16/17 (94%) patients with anti-FGFR3 Abs (table 2). The last patient had SLE and no recorded peripheral neuropathy, but was not examined by the authors. The 16 patients were 10 females and 6 males aged 18–73 years (median: 47). Onset was acute in 2, subacute in 4 and progressive in 10. The neuropathy had a non-length-dependent distribution in 13/15 patients with available information and fulfilled the criteria of probable SNN in 9/11 evaluated patients. Pain was present in seven patients and was prominent in five; the disorder was limited to neuropathic pain and dysesthesia in the face in one patient and in the thorax in another. Ataxia occurred in nine patients. Patient 12 developed dysautonomia. The Rankin's score ranged from 2 to 5 (median 2).
The cerebrospinal fluid (CSF) was abnormal in 5/9 patients. Electroneuromyography (ENMG) was consistent with SNN in 10/11 patients and normal in the last patient who had small fibre neuropathy. Nerve biopsy in six patients showed moderate to severe myelinated fibre loss without regenerating clusters (see online supplementary figure S2B,C). One patient had CD3-positive T-cell infiltration around epineurial vessels (see online supplementary figure S2A). At the time the serum samples were taken, 10/16 patients had no clinically overt systemic autoimmune disease or ANA, SSA or SSB antibodies. One of them had HIV infection. In 3 of these 10 patients an autoimmune systemic disorder appeared months or years later so that at the end a systemic autoimmune disease was present in 9/16 patients.
Logistic regression comparing the 16 patients with anti-FGFR3 Abs and SN with 41 patients with non-paraneoplastic SNN and no anti-FGFR3 Abs, and with 36 patients with SNN and anti-Hu Abs (see online supplementary table S4) showed that patients with anti-FGFR3 Abs were younger and tended to be more frequently female. Onset tended to be progressive, with more frequent face involvement. At maximum development of the neuropathy, the upper limbs and deep sensations were less frequently involved. There were no differences in the ENMG. The available information did not allow comparison of the efficacy of immunomodulatory treatments in the two groups.
The protein array used in this study contains about 8000 human proteins, including protein kinases, transcription factors, membrane or nuclear proteins, proteins involved in signal transduction, cell communication and metabolism or cell death originating from the Ultimate ORF collection. More than 50 of these are growth factor receptor proteins. Protoarray, ELISA, cell-based assay and purified IgG reactivity on sensory neurons showed that FGFR3 was a target of antibodies in a subgroup of patients with sensory neuron disorder. A full-length FGFR consists of an extracellular region, composed of three immunoglobulin-like domains, a single hydrophobic membrane-spanning segment and a cytoplasmic tyrosine kinase (TK) domain, which is formed of two subdomains named TRK1 and TRK2. The FGFR3 intracellular kinase domain used in this study consisted of amino acids 399–806. The cell-based assay confirmed that anti-FGFR3 Abs reacted with the TK domain, although one also reacted with TRK1 or TRK2. By ELISA, a few anti-FGFR3 Abs cross-reacted with FGFR2 and none with FGFR1 or FGFR4. The ELISA was more sensitive than the cell-based assay to detect anti-FGFR3 Abs. As in addition, the FGFR3 subunits modified the morphology and viability of transfected cells, a cell-based assay is probably not the choice technique for anti-FGFR3 Abs detection. Epitope mapping on the intracellular domain gave no conclusive results because it was not possible to test all the patients due to the limited amount of each sample serum.
Fgfr mutations induce defects that mostly affect bone formation, resulting in cranial synostosis or dwarfism, while amplification or deletion of FGFR3 is involved in carcinogenesis or tumour growth.24 FGFs and FGFRs are variously involved in neural development.25 ,26 In the human adult, these are expressed in the central and peripheral nervous system.27 We showed that, in adult rat DRG, FGFR3 was expressed in small and large sensory neurons and their satellite glial cells. We also observed FGFR3 expression in the developing Gasserian ganglion. In adults, FGFR proteins are upregulated in sensory neurons after nerve crush,28–30 and FGFR1 and FGFR2 are involved in Schwann cell–axon interaction in unmyelinated fibres.31
Ninety-four per cent of patients with anti-FGFR3 Abs had SN and several lines of evidence suggest that the sensory neuron cell bodies were probably the target of the disease since (1) the distribution of the sensory symptoms was non-length dependent in 87% of patients and fulfilled the diagnosis SNN criteria in 82% of the evaluable patients. (2) A patient with Sjögren's syndrome had a bilateral trigeminal neuropathy, which, in this context, is usually ascribed to a lesion in the trigeminal nerve ganglion.32 (3) Nerve biopsy consistently found fibre loss without regenerating clusters, which is more indicative of neuronopathy than axonopathy. Finally, the ENMG results were similar to those seen in patients with SNN.
However, the patients with anti-FGFR3 Abs had clinical characteristics which distinguished them from both anti-Hu-associated SNN and anti-FGFR3 Ab-seronegative SNN. Overall, they were younger and the neuropathy had a progressive course, although a subacute, and sometimes an acute, evolution was seen. Striking clinical features were frequent trigeminal nerve involvement and pain intensity in several patients. Although the neuropathy generally involved both large and small fibres and had a widespread somatosensory distribution; in some patients, it was restricted to the trigeminal nerve or small unmyelinated fibres. As a whole, our data indicate that anti-FGFR3 Abs should be searched in patients with a non-length-dependent SN which may or may not fulfil the SNN criteria, particularly if pain and trigeminal nerve involvement are present.
SNN occur with different systemic autoimmune diseases which frequently appear months or years after the onset of the neuropathy. In this study, 10/17 (59%) patients with anti-FGFR3 Abs developed another autoimmune disease, suggesting that the neuropathy was immune mediated. This was corroborated by the observation that CSF findings were abnormal in half of the nine cases tested, showing a mild elevation of protein concentration and sometimes cellular reaction or oligoclonal bands. In addition, of the seven patients without associated autoimmune disease, one had T-cell infiltration around epineurial vessels, confirming that the disorder was inflammatory. The presence of similar changes, although unusual, has already been reported in patients with inflammatory ganglionopathy9 indicating that the inflammatory reaction may extend into the peripheral nerve. Importantly, at the time the serum samples were taken, an autoimmune disease was not apparent in 62.5% of patients with anti-FGFR3 Abs and peripheral neuropathy, and only appeared after several years of follow-up in 30% of patients; thus, anti-FGFR3 antibody was the only marker of immunity in a significant proportion of patients.
In our study, the prevalence of anti-FGFR3 Abs was 19% in patients with an idiopathic SN or an SN associated with an autoimmune disorder. It is of the same order of magnitude of the prevalence of each of the different autoantibodies which occurs in patients with autoimmune encephalitis33 and suggests that other antibodies could be detected in patients with sensory neuron disorders.
SSN depends on a variety of aetiologies but a majority of patients appear idiopathic after a careful initial workup although several of them probably have an underlying inflammatory process and with time may develop an overt systemic autoimmune disease.16 Biopsy of DRG is the only way of identifying these cases but cannot be recommended as a routine investigation. Detection of anti-FGFR3 Abs may help to recognise these patients.
The authors express grateful thanks to all the contributors who sent serum samples and clinical data for the study: Andoni Echnaniz-Laguna, MD, PhD (Reference Centre for Neuromuscular Diseases CHU de Strasbourg France), Alain Créange, MD, PhD (Reference Centre for Neuromuscular Diseases Hôpital de Creteil, APHP Paris, France), Emilien Delmont, MD (Reference Centre for Neuromuscular Diseases CHU de Nice, France), Jerôme Franques, MD (Reference Centre for Neuromuscular Diseases CHU Marseille, France), Thierry Kuntzer, MD (Department of Neurology CHUV Lausanne, Switszerland), Emeline Lagrange, MD (Reference Centre for Neuromuscular Diseases CHU de Grenoble, France), Stephane Mathis, MD (Department of Neurology CHU de Poitiers, France), Guillaume Nicolas, MD, PhD (Reference Centre for Neuromuscular Diseases Hôpital de Garches, APHP Paris, (France) and François Ochsner, MD (La Chaux-de-Fond, Switzerland).
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Contributors J-CA conceived the project, checked the biological and clinical results, and wrote the manuscript; NB performed the protoarray analysis, and supervised the ELISA and the cell-based assay; FL performed the ELISA and the immunohistochemistry on rat tissue; ER performed the cell-based assay; VR constructed the plasmids for the cell-based assay; SP provided the sera from patients with autoimmune diseases; KF reviewed the patients’ clinical data; JH reviewed and discussed the manuscript and the study; J-PC contributed to the project conception and realisation, discussed the results and reviewed the manuscript.
Funding This study was supported by the French Ministry of Health (PHRCN 2005-0501104 and PHRCIR-API 28-01) and by la Région Rhône-Alpes (Cluster 11 5-2010-1).
Competing interests None.
Patient consent Obtained.
Ethics approval The Ethical Committee Loire Sud.
Provenance and peer review Not commissioned; externally peer reviewed.