Article Text

Download PDFPDF

Original research
Clinical characterisation of sensory neuropathy with anti-FGFR3 autoantibodies
  1. Yannick Tholance1,2,
  2. Christian Peter Moritz2,
  3. Carole Rosier2,3,
  4. Karine Ferraud3,
  5. François Lassablière2,
  6. Evelyne Reynaud-Federspiel2,
  7. Marcondes C França Jr4,
  8. Alberto R M Martinez4,
  9. Jean-Philippe Camdessanché2,3,
  10. Jean-Christophe Antoine2,3
  11. anti-FGFR3 antibody Study Group
    1. 1 Laboratory of Biochemistry, CHU Saint-Etienne, Saint-Etienne, France
    2. 2 Synaptopathies et autoanticorps (synatac), Institut Neuromyogène, Saint-Priest-en-Jarez, France
    3. 3 Department of Neurology, CHU Saint-Etienne, Saint-Etienne, France
    4. 4 Department of Neurology, University of Campinas (UNICAMP), Campinas, São Paulo, Brazil
    1. Correspondence to Dr Yannick Tholance, Laboratory of Biochemistry, CHU Saint-Etienne, Saint-Etienne 42055, France; yannick-tholance{at}


    Objective Sensory neuropathies (SNs) are often classified as idiopathic even if immunological mechanisms can be suspected. Antibodies against the intracellular domain of the fibroblast growth factor receptor 3 (FGFR3) possibly identify a subgroup of SN affecting mostly the dorsal root ganglion (DRG). The aim of this study was to identify the frequency of anti-FGFR3 antibodies and the associated clinical pattern in a large cohort of patients with SN.

    Methods A prospective, multicentric, European and Brazilian study included adults with pure SN. Serum anti-FGRF3 antibodies were analysed by ELISA. Detailed clinical and paraclinical data were collected for each anti-FGFR3-positive patient and as control for anti-FGFR3-negative patients from the same centres (‘center-matched’).

    Results Sixty-five patients out of 426 (15%) had anti-FGFR3 antibodies, which were the only identified autoimmune markers in 43 patients (66%). The neuropathy was non-length dependent in 89% and classified as sensory neuronopathy in 64%, non-length-dependent small fibre neuropathy in 17% and other neuropathy in 19%. Specific clinical features occurred after 5–6 years of evolution including frequent paresthesia, predominant clinical and electrophysiological involvement of the lower limbs, and a less frequent mixed large and small fibre involvement. Brazilians had a higher frequency of anti-FGFR3 antibodies than Europeans (36% vs 13%, p<0.001), and a more frequent asymmetrical distribution of symptoms (OR 169, 95% CI 3.4 to 8424).

    Conclusions Anti-FGFR3 antibodies occur in a subgroup of SN probably predominantly affecting the DRG. Differences between Europeans and Brazilians could suggest involvement of genetic or environmental factors.

    • anti-FGFR3 autoantibodies
    • autoimmune diseases
    • fibroblast growth factor receptor 3 (FGFR3)
    • sensory neuronopathies
    • sensory neuropathies

    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.


    Sensory neurons of the peripheral nervous system are the target of many disorders including toxic, genetic, viral and inflammatory diseases.1 However, in about half of the patients with sensory neuropathy (SN), no cause is found after extensive workup.2 Depending on the targeted population of neurons, patients develop ataxia, painful small fibre neuropathy (SFN), or a mixed painful and ataxic disorder. Disorders affecting the sensory neuron cell body in the dorsal root ganglion (DRG) are called sensory neuronopathies (SNNs). A subgroup of sensory neuron diseases are immune mediated, such as paraneoplastic SNN or SNN/SFN associated with Sjögren syndrome (SS), and probably an unknown proportion of idiopathic cases.1 Anti-Hu and anti-CV2/collapsin response mediator protein 5 (CRMP5) antibodies are highly specific of paraneoplastic SNN3 while until recently, no antibodies were known in the other conditions. We have reported that an antibody reacting with the intracellular domain of the fibroblast growth factor receptor 3 (FGFR3) identifies a subgroup of patients with SN that mostly fulfilled the criteria of SNN.4 Interestingly, in 60% of the patients with anti-FGFR3 antibodies, these antibodies were the only sign of an autoimmune context. However, this study resulted from a monocentre and retrospective analysis of only 16 anti-FGFR3-positive cases out of 106 SN patients. Here, we report the results of a prospective, comparative, multicentric and international study on 65 anti-FGFR3-positive patients out of 426 included SN revealing the pattern of the neuropathy. We also show that the frequency of anti-FGFR3 antibodies may be different in Europe and Brazil.


    Standard protocol approvals, registrations and patient consents

    All participants provided written informed consent.

    Study design and patients

    Between 3 February 2015 and 14 June 2018, we conducted a prospective multicentric international study in 36 neuromuscular centres in Europe (France, Switzerland and UK) and one in Brazil. The aim was to include at least 50 patients with anti-FGFR3 antibodies and 50 paired negative controls over 18 years of age and a clinically pure SN. Because we previously found that anti-FGFR3 antibodies may occur in some patients not fulfilling the criteria of SNN,4 the present study included SNN, SFN, sensory chronic inflammatory demyelinating polyneuropathy (CIDP), length-dependent neuropathy (LDN) and other SN. All the patients should have a significant development of their neuropathy before entry in the study. Exclusion criteria were clinical motor involvement, neuropathies with established etiological diagnosis (diabetic, genetic or toxic origin) or neuropathies associated with known autoantibodies (onconeural, anti-myelin-associated glycoprotein and anti-ganglioside antibodies). The patients were investigated using conventional methods with neurological clinical examination and paraclinical investigations including at least electroneuromyography (ENMG) and a biological workup with autoantibody testing (table 1). Inclusion in the study required the availability of serum and a form collecting the SNN score, the final diagnosis retained by the neurologist following the patient in the centre, and information about the autoimmune context of the neuropathy (table 1). A second form was sent to the clinicians to collect detailed clinical and paraclinical data for each anti-FGFR3-positive patient (table 1) and as control for one randomly chosen anti-FGFR3-negative patient from the same centre (‘center-matched’). We used this procedure of matching to characterise the anti-FGFR3-positive patients in terms of age, sex and type of neuropathies and thus avoid selection bias related to the type of patients recruited in each centre. This second form collected information on symptoms and signs, modified Rankin (M-Rankin) score, ENMG, skin biopsy or laser-evoked potentials for SFN, biological workup, associated diseases, treatment and appreciation of its effect according to the clinician’s opinion (table 1). Two evaluations were requested at recruitment. The first described the status of the patient at full development of the neuropathy and the second retrospectively collected data concerning the neuropathy onset. A third evaluation obtained at least 6 months later was left to the discretion of the investigators. The final diagnosis of the neuropathy as defined by referring physicians was checked against the diagnostic criteria of SNN,5 CIDP6 and SFN7 by the authors JCA and JPC.

    Table 1

    Description of the data requested in the study forms

    Interpretation of ENMG data

    ENMG was performed according to the usual recommendations in current practice.8 9 Motor and sensory conduction velocities were analysed according to the normal values of each laboratory and expressed as a percentage of the lower limits of the normal (LLN). Amplitudes of sensory nerve action potentials (SNAPs) for median, ulnar, radial, superficial peroneal and sural nerves were collected. The compound muscle action potentials and motor conduction velocities (MCVs) between the wrist and the elbow were recorded in the median and ulnar nerve and between the ankle and knee for the tibial and peroneal nerves. The number of normal, abolished and reduced SNAPs was calculated. For each motor nerve, three abnormal patterns were considered, as previously published10: a demyelinating pattern, an axonal pattern and a mixed axonal/demyelinating pattern (see online supplementary table 1).

    Supplemental material

    Detection of serum anti-FGFR3 antibodies and IgG isotypes

    Detection of anti-FGFR3 antibodies was performed by using a previously described indirect ELISA.11 We used the human recombinant intracellular domain of FGFR3 (Invitrogen, Carlsbad, California, USA) for the coating. For each serum, the signals were normalised for the serum-specific background noise by calculating the difference between the optical density of the FGFR3-coated well and that of the non-coated well (see supplementary methods S1). The test was considered positive when the difference was 3 SD above the average signal of 58 apparently healthy blood donors (z-score >3). To determine IgG isotypes, indirect ELISA was repeated with secondary anti-human antibodies specific against IgG1, IgG2, IgG3 or IgG4 (see supplementary methods S1).

    Statistical analysis

    Categorical data were analysed by the χ² test and continuous data by the Mann-Whitney test or Student’s t-test depending on their distribution (Kolmogorov-Smirnov test). Quantitative results were expressed as median (IQR) or mean±SD and categorical data were presented as count (percentage). A p value of ≤0.05 was used to determine statistical significance after Bonferroni correction for multiple comparisons. Logistic regression adjusted for age and sex was used to characterise the clinical and paraclinical data of anti-FGFR3-positive patients. Parameters with p value <0.2 in univariate analysis were entered into the logistic regression model by backward and forward method. Age and sex were introduced afterwards in the logistic regression together with the significant parameters of the first model. Data were expressed as OR, 95% CI. To evaluate the logistic regression model, we used a receiver operating characteristic curve analysis. Missing data were excluded from the analysis both in univariate and multivariate analysis. Statistical analysis was performed with MedCalc Statistical Software V.18.2.1.


    Included population and screening of anti-FGFR3 antibodies

    Four hundred and thirty-five patients were enrolled, 393 (90.3%) from European centres and 42 (9.7%) from Brazil. Of those enrolled, 9 did not undergo serum analysis due to exclusion criteria (figure 1), yielding 426 patients with anti-FGFR3 antibody results. Of these 426 patients, the diagnostic repartition was 240 SNN, 86 SFN, 31 CIDP, 25 LDN and 15 other neuropathies (CIDP, LDN and other neuropathies were grouped as ‘other SN’ (OSN)). For 29 patients, no final diagnosis was communicated. Of these 426 patients, 65 patients (15.3%) had anti-FGFR3 antibodies (figure 2), thereof 40 women and 25 men (ratio 1.6). Median age of onset was 53 (40–64). Fifteen patients (23%) came from Brazil and 50 from Europe (figure 2). The median concentration of anti-FGFR3 antibodies was 6.6 (5.0–12.3) µg/mL with a range of values between 3.8 and 745.6 µg/mL. In terms of IgG subtypes, IgG1 was the most frequent followed by IgG4 (see supplementary figure S1). We analysed a second sample of a later blood sampling for 38 patients (33 anti-FGFR3-negative and 5 anti-FGFR3-positive patients) and the conclusion about the presence or not of anti-FGFR3 antibodies was always the same (see supplementary figure S1).

    Figure 1

    Study profile. FGFR3, fibroblast growth factor receptor 3; OSN, other SN; SFN, small fibre neuropathy; SNN, sensory neuronopathy.

    Figure 2

    Detection of anti-FGFR3 antibodies via indirect ELISA. The FGFR3-specific reactivity of each serum is represented by the optical density (OD) difference between FGFR3-coated wells and non-coated well (serum-specific background noise normalisation). A z-score ≥3 above the apparently healthy blood donors’ mean value defines positivity (dotted line). These z-score ≥3 corresponds to an OD difference of ≥0.279 between FGFR3-coated wells and non-coated well. FGFR3, fibroblast growth factor receptor 3; N, number; OSN, other SN; SFN, small fibre neuropathy; SNN, sensory neuronopathy.

    The clinical characteristics of the 65 anti-FGFR3-positive patients

    Recruitment for the study occurred after a median delay of 2.9 (1.0–6.0) years from neuropathy onset. Anti-FGFR3 antibodies were the only identified autoimmune markers in 43 patients (66%). Forty-two patients (64.6%) were classified as SNN of whom 40 (95.2%) fulfilled the SNN diagnostic criteria,5 11 (16.9%) had SFN of whom 9 (82%) with a non-length dependent distribution and 12 (18.5%) had OSN which was non-length dependent in 9 (table 2). Among patients with OSN, two patients fulfilled the criteria of CIDP6 (one of them also met those of SNN5); three patients had ENMG changes suggesting demyelination in one nerve but the criteria of CIDP were not fulfilled; three patients had sensory abnormalities limited to the lower limbs (LL) with abolished Achilles reflexes suggesting LDN; the last four patients could not be classified: one had a pure multifocal painful neuropathy with perivascular cuffs of lymphocytes on nerve biopsy and a non-length dependent fibre loss on skin biopsy; two patients had a severe ataxic neuropathy with normal MCVs and normal or subnormal SNAPs but abnormal sensory evoked potentials in the four limbs indicating a proximal peripheral nerve disorder; the last patient had diffuse areflexia, gait instability and normal sensory examination.

    Table 2

    Clinical and paraclinical data of anti-FGFR3-positive patients at onset and at full development of the neuropathy (n=65)

    The neuropathy onset was mostly chronic (42/64, 66%) and limited to the LL (27/65, 42%) or similarly affected the four limbs (33/65, 51%) and the distribution was asymmetrical in one third of patients (table 2). At recruitment, the neuropathy was non-length dependent in 58 patients (89%). The disorder was restricted to the LL in 18 (28%) and asymmetrical in 25 (39%). Pain was reported in half of the cases. Ataxia was more frequent in the LL (39/65, 60%) than in the UL (25/65, 39%). Autonomic nervous system disorders were rare (11/65, 17%) and included constipation, bladder dysfunction, orthostatic hypotension and abnormal pupil reaction indicating involvement of the sympathetic and parasympathetic system (table 2). A third evaluation was obtained in 53 patients (82%, figure 3A) with a median delay of 5.9 (3.9–11.3) years from onset. The disorder was limited to the LL in a minority of cases (14/53, 26%) with a global symmetrical distribution (32/53, 60%). Pain was observed in 29 patients (56%), LL ataxia in 38 patients (72%) and dysautonomia in 15 patients (28%).

    Figure 3

    Distribution of sensory symptoms and clinical signs at the last clinical exam in all anti-FGFR3-positive patients (A), and in European (B) and Brazilian (C) anti-FGFR3-positive SNN patients. The distribution of symptoms is indicated by the frequency of involvement and by a colour code summarising all the data (the darker the colour, the higher the frequency of involvement). DSS, deep and superficial sensation (myelinated sensory fibres); FGFR3, fibroblast growth factor receptor 3; LL, lower limbs; N, number; Prox, proximal; PTS, pain and temperature sensation (unmyelinated sensory fibres); UL, upper limbs.

    The ENMG recording was performed with a median delay of 3.7 (1.8–8.2) years after onset (table 2). In total, SNAPs were abnormal in the LL for 50/62 patients (81%) and in the UL for 22/62 (37%). Concerning MCVs, MCVs were strictly normal in the four limbs in 33% of patients (19/58), in the UL in 71% of patients (41/58) and in the LL in 47% (29/62). Cerebrospinal fluid (CSF) was analysed in 40 patients and total CSF protein levels were elevated in seven cases (18%).

    Thirty-five patients (54%) received immune-modulatory treatments. Multivariate logistic regression found no treatment associated with improvement or stabilisation, while steroids (OR 12.11 (0.9 to 160.0)) or intravenous Ig (OR 12.8 (1.2 to 141.8)) were significantly associated with deterioration.

    Comparison of the neuropathy between anti-FGFR3-positive and anti-FGFR3-negative patients

    To determine whether there exists a pattern of neuropathy specific to anti-FGFR3 antibodies, we compared the 65 anti-FGFR3-positive patients with the 66 centre-matched anti-FGFR3-negative patients. The number of missing data was similar in the two groups. The median delays from onset to the first or the last clinical evaluation or ENMG recording were not different between the two groups (p=0.21, p=0.87, p=0.12). In univariate analysis (see supplementary table S2), the two populations did not show major differences regarding type of neuropathy, clinical manifestation at onset, first and last examination, associated diseases and paraclinical investigations including ENMG. After Bonferroni correction, only age was significantly different, the anti-FGFR3-positive patients being older than controls (52.6 (39.7 to 64.0) vs 48.0 (36.2 to 56.6), respectively, p=0.02). Multivariate logistic regression adjusted for age, sex and for Brazilian origin found no difference in the two populations regarding the clinical and paraclinical data originating from the first examination. At the last examination, in contrast, patients with anti-FGFR3 antibodies had significantly more frequent paresthesia (OR 4.9, 95% CI 1.2 to 19.3, p=0.02), a higher number of abnormal SNAPs in LL (OR 1.7, 95% CI 1.1 to 2.5, p=0.01), a higher number of SNAPs>50% of the LLN in UL (OR 1.6, 95% CI 1.0 to 2.6, p=0.04), less frequent mixed alteration of myelinated and unmyelinated sensory fibres (OR 0.2, 95% CI 0.07 to 0.77, p=0.02) and less frequent trunk involvement (OR 0.2, 95% CI 0.04 to 0.94, p=0.04). The area under the curve of the logistic regression model was 0.82 (0.73–0.89) suggesting that, with time, the clinical and paraclinical pattern of the anti-FGFR3-positive patients became clearly distinct from that of the anti-FGFR3-negative one (table 3).

    Table 3

    Multivariate analysis by logistic regression with clinical and paraclinical data

    Anti-FGFR3 antibodies in Brazilians and comparison of the neuropathy between Brazilian and European anti-FGFR3-positive patients

    A comparison of the whole European and Brazilian population showed that 15 out of 42 (36%) Brazilian patients, all fulfilling the SNN criteria, had anti-FGFR3 antibodies against only 50 out of 381 (13%) European patients (p<0.001). Whatever the presence of anti-FGFR3 antibodies, Brazilians were younger than Europeans (40 (35–50) vs 54 (42–63), p<0.001) and had more frequent autoimmune contexts (16/32 (50%) vs 70/327 (21%), p<0.001) with a different distribution of their disorders (p<0.001). In detail, autoimmune hepatitis was only observed in Brazilians (n=4 cases) while SLE (6), inflammatory bowel-disease (2), rheumatoid arthritis (7), uveitis (1), sarcoidosis (2) and Biermer’s disease (2) were only present in the Europeans. Moreover, the frequency of SS was higher in Brazilians than in Europeans. Multivariate logistic regression showed that after adjustment for age, sex, SNN score and presence of an associated autoimmune disease, a Brazilian status was still significantly associated with detection of anti-FGFR3 antibodies (OR 7.3, 95% CI 3.0 to 18.0, p<0.001; table 3). The same results were obtained when comparing the frequency of anti-FGFR3 antibodies in Europeans and Brazilians with SNN only.

    When comparing the detailed clinical and paraclinical data of the Brazilian (n=15) versus European (n=27) anti-FGFR3-positive SNN (table 4), we found the same differences in term of age and frequency of autoimmune disease as when comparing the whole populations. The clinical pattern of the SNN was different in the two populations with a more severe sensory impairment with time and a very different distribution of sensory disorders in Brazilians. At onset and during evolution, in Brazilians, sensory symptoms more frequently involved UL and LL proximally and were more frequently asymmetric comparatively to Europeans (figure 3B,C and table 4). In multivariate analysis (table 3), age of onset and asymmetrical distribution were significantly associated with a Brazilian origin. Comparatively to onset, the differences persisted at first clinical investigation and in addition, sensory symptoms and ataxia were more frequent in the UL in Brazilians (table 4). In multivariate analysis, a higher frequency of proximal involvement in UL and a higher asymmetrical distribution of symptoms were significantly associated with a Brazilian origin. At the last examination (figure 3B,C), the difference of symptoms distribution persisted and global areflexia and dysautonomia were more frequent in Brazilians (table 4). There was no difference concerning ENMG data.

    Table 4

    Significant clinical and paraclinical differences between the Brazilian (n=15) and European (n=27) SNN patients with anti-FGFR3 antibodies


    This prospective multicentric controlled study specifies the pattern of patients bearing anti-FGFR3 antibodies. Overall, 15% of the patients with SN enrolled in this study had anti-FGFR3 antibodies. About two-thirds of them fulfilled the diagnostic criteria of SNN, 17% had SFN and 19% another form of SN. Interestingly, the neuropathy had a non-length dependent distribution in about 89% of cases including SFN suggesting that the DRG were the main target of the disorder. This is supported by the electrophysiological study which shows that on average 77% of the SNAPs studied in a given patient were abnormal with 30% of them abolished in the UL. However, motor nerve conduction was abnormal in 26% of nerves in a given patient. The most frequent patterns were mild axonal or mild mixed axonal and demyelinating changes. Similar alterations are frequent in SNN as a whole2 and more particularly in paraneoplastic SNN12 13 or with SS.14 The autonomic nervous system was involved in 28% of patients at the last clinical exam, another feature shared with paraneoplastic and SS-associated SNN.15 16 Three patients fulfilled the ENMG criteria of a primary demyelinating disorder and were classified as CIDP by the referring clinician. Similarly, clearly demyelinating changes occasionally occur in patients with the onconeural anti-Hu and CV2/CRMP5 antibodies.12 13 In only three patients, the distal part of sensory nerves in LL was affected both clinically and electrophysiologically. Another feature in patients with anti-FGFR3 antibodies, also shared with paraneoplastic17–19 or SS-associated SNN14 is a predominant involvement of large or small sensory neurons sometimes restricted to one or the other type,20 explaining that pain or ataxia may prevail in some patients. These variations may depend on the fact that the expression of FGFRs including FGFR3 occurs both in large and small sensory neurons and is not restricted to their cell body since it is also present is axons and Schwann cells including non-myelinating Schwann cells.21–25

    The clinical pattern of patients with anti-FGFR3 antibodies showed particular features, some of them already underlined in our previous study.4 In detail, a mean age around 50 years, a high frequency of paresthesia, a less frequent mixed involvement of sensory myelinated and unmyelinated fibres, a more severe impairment of sensory conduction in LL and a less impaired sensory conduction of UL characterise the anti-FGFR3 antibody-related neuropathy. Interestingly, these features probably take several months or years to become apparent since they were statistically significant only at the last examination. This indicates a specific evolution of the neuropathy in patients with anti-FGFR3 antibodies which with time tends to be less severe in the UL than in the LL and predominate on either large proprioceptive or small pain-related sensory neurons. These findings could justify the research of these autoantibodies in routine practice.

    Another important result is the high prevalence of anti-FGFR3 antibodies in the Brazilian population, almost three times that of Europeans. This difference applies regardless of whether we compare the two populations of neuropathies or only SNN in Europeans and Brazilians since Brazilians had SNN only, which is explained by the particular interest of the Brazilian centre in this disorder. The Brazilian patients were younger than the Europeans and have more frequently an associated autoimmune disease. In addition, all of them had typical SNN. However, after adjustment for these confounding factors, Brazilian origin was still highly associated with anti-FGFR3 antibodies. The clinical presentation of anti-FGFR3-positive SNN patients also differed between the Brazilian and European populations, although the proportion of patients fulfilling the SNN diagnostic criteria was not different between the two populations. Thus, the distribution of sensory symptoms among the Brazilian anti-FGFR3-positive patients was more frequently asymmetrical, proximal and affecting the UL. As the number of tested Brazilian patients is relatively limited, these results including the prevalence of anti-FGFR3 antibodies and the clinical pattern of the neuropathy need to be confirmed on a larger multicentric cohort.

    Several factors may explain the higher prevalence of anti-FGFR3 antibodies in Brazil, among them environmental factors acquired early in life or a specific genetic background. The former may explain the younger age of onset and the latter the striking difference in the frequency and type of associated autoimmune diseases in the two populations and some differences in the pattern of the SNN. One particular difference about SNN with autoimmune context was the more frequent association of anti-FGFR3 antibodies with an underlying SS context in SNN patients from Brazil versus Europe, although the same diagnostic criteria of SS were used by the different centres.26 This could be explained by a possible higher overall incidence of SS in Brazil than in Europe or by a more common association of SNN with SS in Brazil than in Europe.27

    Anti-FGFR3 antibodies were the only markers of an autoimmune reaction in 66% of cases. At present, it is not possible to determine whether they play a pathogenic role or are only biomarkers of a dysimmune process as it is the case, for example, with most of the onconeural antibodies, particularly anti-Hu antibodies.28 29 However, anti-FGFR3 antibodies recognise the intracellular domain of the protein4 which contains the tyrosine kinase domains whose activation is responsible for signal transduction and regulation of many cell functions including survival.30 Whether the antibody may enter the neurons and interfere with signal transduction remains at present speculative and needs further work. Anti-FGFR3 antibodies were mostly IgG1, associated or not with IgG3, as it is frequent in autoimmune diseases. These IgGs can activate the complement cascade or induce antibody-dependent toxicity by macrophages. Thirty per cent of the patients had IgG4 which may have a quite different action by inducing functional blocking of receptor proteins.31 However, the low number of identified cases does not allow any conclusion as whether IgG4 antibodies are associated with a specific clinical profile or response to treatment. In contrast to another publication,32 we found the concentration of anti-FGFR3 antibodies to not differ significantly over time in anti-FGFR3-positive patients, but this needs to be confirmed in a larger number of cases.

    This study was not designed to appreciate the effect of treatments on the neuropathy in patients with anti-FGFR3 antibodies. Only half of them received immunomodulatory treatments in an open uncontrolled way. Steroids or IVIg were associated with deterioration in anti-FGFR3-positive patients. However, one should keep in mind that many patients were probably treated too late in the evolution of their disease, while early treatment is crucial in these disorders to prevent irrevocable neuron cell death.33

    ELISA was the technique used for antibody screening. The use of a single technique may potentially be a source of misclassification. For example, it is well-accepted that anti-myelin oligodendrocyte glycoprotein autoantibodies can be specifically detected via cell-based assays (CBA), but cause false positives in ELISA.34 However, we have shown the concordance between ELISA and CBA in our previous study.4 Furthermore, to reduce the risk of misclassification, we systematically added an antigen-free well to take serum-specific background noise into account, which is a major cause of false-positive results in ELISA.11

    In conclusion, our prospective, controlled and international study confirms that anti-FGFR3 antibodies are associated with a subgroup of SN which mostly affects the DRG. Anti-FGFR3-positive neuropathies had clinical features that could help to distinguish them from other SN which can justify the research of these autoantibodies in clinical practice. The prevalence of anti-FGFR3 antibodies appears to be higher in Brazilian compared with European patients, suggesting that environmental or genetic factors may lead to an autoimmune reaction against FGFR3.


    We would like to acknowledge all the care teams who collaborated in our study and all the patients who donated blood samples for this work.



    • Collaborators Collaborators anti-FGFR3 antibody Study Group: David Adams, MD, PhD, Cécile Cauquil, MD (FILNEMUS, Department of Neurology, Kremlin Bicêtre hospital, AP-HP, University of Paris-Sud, Bicêtre, France, site investigator); Shahram Attarian, MD, PhD, Emilien Delmont, MD (Reference Center for neuromuscular disorders and ALS, University Hospital La Timone, Aix-Marseille University, Marseille, France, site investigator); Genevieve Blanchet-Fourcade, MD (Department of Neurology, Hospital Center of Narbonne, Narbonne, France, site investigator); Céline Callias, MD (Department of Neurology, Central institute of hospitals (ICH), Valais Hospital, Sion, Switzerland, site investigator); Julien Cassereau, MD (Department of Neurology, ALS Center, University Hospital of Angers, Angers, France, site investigator); Ariane Choumert, MD (Department of Neurology, University hospital of the Reunion, SaintPierre, France, site investigator); Pierre Clavelou, MD, PhD (Department of Neurology, University hospital of Clermont-Ferrand, Clermont-Ferrand, France, site investigator); Alain Créange, MD, PhD (Department of Neurology, CHU Créteil, Henri Mondor University Hospital, Créteil, France, site investigator); Gabriella Di Virgilio, MD (Department of Neurology, RivieraChablais Hospital, Vevey, Switzerland, site investigator); Andoni Echaniz-Laguna, MD (Department of Neurology, University hospital of Strasbourg, Strasbourg, France, site investigator); Benoit Faucher, MD (Department of Multidisciplinary Medicine, Toulon-La Seyne-sur-Mer Intercommunal Hospital Center, Toulon, France, site investigator); Ian Galea, MD (Wessex Neurosciences Centre, University Hospital Southampton NHS Foundation Trust, Southampton, UK, site investigator); Steeve Genestet, MD (Department of Neurological Functional Explorations, University Hospital of Brest, Brest, France, site investigator); Antoine Guéguen, MD (Department of Neurology, Fondation Ophtalmologique Adolphe de Rothschild, Paris, France, site investigator); Ioana M Ion, MD (Department of Neurology, University Hospital of Nîmes, Nîmes, France, site investigator); Raul Juntas-Morales, MD, Guillaume Taieb, MD (Department of Neurology, University hospital of Montpellier, Montpellier, France, site investigator); Thierry Kuntzer, MD, PhD (Department of Neurology, Lausanne University Hospital CHUV, Lausanne, Switzerland, site investigator); Emmeline Lagrange, MD (Department of Neurology, University Hospital of Grenoble, Grenoble, France, site investigator); Jean-Marc Léger, MD, PhD, Tanya Stojkovic, MD, PhD (National Referral Center for Neuromuscular Diseases, University Hospital Pitié-Salpétrière, Paris, France, site investigator); Maud Lepetit, MD (Department of Neurology, Cornouailles Hospital Center, Quimper, France, site investigator); Laurent Magy, MD, PhD (Department of Neurology, Reference Center for Rare Peripheral Neuropathies, University Hospital of Limoges, Limoges, France, site investigator); Julie Mas, MD (Department of Neurology, St. Jean Hospital, Perpignan, France, site investigator); Maud Michaud, MD (Department of Neurology, University Hospital of Nancy, Nancy, France, site investigator); Capucine Mouthon-Reignier, MD (Department of Neurology, Sud francilien Hospital Center, Corbeil-Essonnes, France, site investigator); Jean-Philippe Neau, MD, PhD (Department of Neurology, Poitiers University Hospital, Poitiers, France, site investigator); Canan Ozsancak, MD (Department of Neurology, Regional Hospital Center of Orléans, Orléans, France, site investigator); Yann Péréon, MD, PhD (Center for Neuromuscular Diseases, Hôtel-Dieu Hospital, Nantes, France, site investigator); Paul Perrotte, MD (Department of Neurology, Hospital Center of Le Havre, Le Havre, France, site investigator); Angela Puma, MD (Peripheral Nervous System, Muscle and ALS Department, University of Nice and Côte d'Azur (UCA), Nice, France, site investigator); Yusuf A Rajabally, MD, PhD (Regional Neuromuscular Clinic, Queen Elizabeth Hospital, University Hospitals of Birmingham, Birmingham, UK, site investigator); Simon Rinaldi, MD, PhD (Oxford Autoimmune Neurology Group, Nuffield Department of Clinical Neurosciences, University of Oxford, John Radcliffe Hospital Oxford, Oxford, UK, site investigator); Violaine Rouaud, MD (Department of Neurology, Bretagne-Atlantique Hospital Center, Vannes, France, site investigator); Guilhem Sole, MD (Department of Neurology, University hospital of Bordeaux, Bordeaux, France, site investigator); Céline Tard, MD (Department of Neurology and Movement Disorders, Lille University Medical Center, Lille, France, site investigator); Anne-Evelyne Vallet, MD (Department of Neurology, Hospital Center of Vienne, Vienne, France, site investigator) and Christophe Vial, MD (Neuromuscular Reference Center, Hospices Civils de Lyon, Bron, site investigator).

    • Contributors YT and J-CA had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: J-PC and J-CA. Acquisition, analysis and interpretation of data: all authors. Drafting the manuscript: YT, CPM, J-PC and J-CA. Critical revision of the manuscript for important intellectual content: all authors. Statistical analysis: YT and J-CA.

    • Funding This study was supported by University hospital of Saint-Etienne (NCT02539329). CPM was funded by the German Research Foundation (DFG; MO 3240/1-1:1) during the study. MCFJ has a research grant on SNN funded by Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP, 2013/01766-7). ARMM is supported by PhD scholarship from FAPESP (2013/26410-0).

    • Competing interests YT reports personal fees from The Binding Site and Alexion, outside of the submitted work. CPM received travel grants from CSL Behring (France), outside of the submitted work. J-PC reports personal fees for lectures, consulting, writing of articles, or training courses from Akcea, Alnylam, Biogen, CSL Behring, Genzyme, Laboratoire Français des Biotechnologies (LFB), Merck, Novartis, Pfizer, Pharmalliance, Teva, Editions Scientifiques L&C, Edimark, Expression Santé, Natus, Scien, SNF-Floerger, outside of the submitted work. J-CA received honoraria for scientific counseling from Pfizer and from a license on diagnostic test for the detection of anti-CRMP5 antibodies, and travel grants from LFB. J-CA and J-PC also participated to a patent on the detection of FGFR3 antibodies. CR, KF, FL, ER-F, MCFJ and ARMM report no disclosure relevant to the manuscript.

    • Patient consent for publication Not required.

    • Ethics approval Ethical approval for the study was provided by the ethical committee of the University Hospital of Saint-Etienne (Clinical trial #NCT02539329) and by the committee for the protection of persons (n°2014-27).

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

    • Data availability statement Data are available upon reasonable request.

    Linked Articles