Objectives To analyse and describe the clinical and genetic spectrum of Charcot-Marie-Tooth disease (CMT) caused by mutations in the neurofilament light polypeptide gene (NEFL).
Methods Combined analysis of newly identified patients with NEFL-related CMT and all previously reported cases from the literature.
Results Five new unrelated patients with CMT carrying the NEFL mutations P8R and N98S and the novel variant L311P were identified. Combined data from these cases and 62 kindreds from the literature revealed four common mutations (P8R, P22S, N98S and E396K) and three mutational hotspots accounting for 37 (55%) and 50 (75%) kindreds, respectively. Eight patients had de novo mutations. Loss of large-myelinated fibres was a uniform feature in a total of 21 sural nerve biopsies and ‘onion bulb’ formations and/or thin myelin sheaths were observed in 14 (67%) of them. The neurophysiological phenotype was broad but most patients with E90K and N98S had upper limb motor conduction velocities <38 m/s. Age of onset was ≤3 years in 25 cases. Pyramidal tract signs were described in 13 patients and 7 patients were initially diagnosed with or tested for inherited ataxia. Patients with E90K and N98S frequently presented before age 3 years and developed hearing loss or other neurological features including ataxia and/or cerebellar atrophy on brain MRI.
Conclusions NEFL-related CMT is clinically and genetically heterogeneous. Based on this study, however, we propose mutational hotspots and relevant clinical–genetic associations that may be helpful in the evaluation of NEFL sequence variants and the differential diagnosis with other forms of CMT.
- charcot-Marie-Tooth disease
- neurofilament light polypeptide
- whole-exome sequencing.
Statistics from Altmetric.com
Charcot-Marie-Tooth disease (CMT) comprises a heterogeneous group of inherited neuropathies clinically characterised by progressive, distal-predominant weakness, amyotrophy and sensory loss. CMT is classically divided into demyelinating (CMT1) and axonal (CMT2) subtypes based on upper limb motor nerve conduction velocities (MNCV) below or above 38 m/s, respectively, although intermediate forms of CMT with MNCV of 25–45 m/s are also recognised.1 2 While CMT1 is frequently caused by mutations affecting peripheral myelin proteins and CMT2 by mutations in genes related to neuronal and axonal functions, there is significant genetic overlap between CMT subtypes.1 3
Neurofilaments are neuron-specific intermediate filaments essential for the radial growth of axons during development and the maintenance of the axonal diameter.4 Like other intermediate filaments, neurofilaments are formed by polymerisation of subunits, the neurofilament light (NFL), medium and heavy polypeptides, all of which contain three major domains: globular N-terminal head, alpha-helical central rod and C-terminal tail. The rod domain is required for co-assembly between subunits, while the head and tail domains are involved in regulation of assembly, axonal transport and radial growth.4–7
Mutations in the NEFL gene encoding NFL were first identified as a cause of autosomal dominant CMT in 2000,8 and subsequent reports have confirmed the association of NEFL mutations with different phenotypes including CMT1 (designated CMT1F), CMT2 (designated CMT2E), dominant intermediate CMT and autosomal recessive CMT.9–18 Despite the growing number of reports about NEFL-related CMT published in the past 15 years, it remains a rare form of inherited neuropathy, accounting for <1% of all CMT cases in large cohorts,19 20 and obvious genotype–phenotype correlations have not been established so far.21 Here we report five new patients with NEFL-related CMT and review all previously published cases to describe the clinical and genetic spectrum of the disease.
Patients and clinical investigations
Five new patients with a diagnosis of NEFL-related CMT were identified from those attending neuropathy clinics at the National Hospital for Neurology and Neurosurgery, Queen Square, London, UK, and the Department of Neurology, Carver College of Medicine, University of Iowa, Iowa, USA. All patients had undergone clinical and investigational assessments during the routine diagnostic process and were under follow-up at the time of the study. Skeletal muscle strength was manually tested and scored according to the Medical Research Council (MRC) grading system ranging from 0 (no movement) to 5 (normal strength). The CMT Examination Score, a composite subscale of the CMT Neuropathy Score (version 2) based on symptoms and signs, with scores ranging from 0 (normal) to 28 (worst score), was obtained in four of the new patients.22 Laboratory tests, neurophysiological studies, MRI scans and nerve and muscle biopsy specimens were performed and analysed according to standard procedures.
In four of the five new patients, NEFL mutations were identified by bidirectional Sanger sequencing during the diagnostic process or as part of a cohort-based study to determine the frequency of CMT subtypes. In one patient, a pathogenic NEFL mutation was identified through whole-exome sequencing followed by bidirectional Sanger sequencing validation. Sanger sequencing of parents’ samples was performed for three cases. Further methodological details are provided in the online supplementary material and supplementary table 1.
Additional published cases
A systematic literature search to identify all previously published cases of NEFL-related CMT until 1 July 2016 was performed in PubMed using the terms: ‘NEFL’, ‘neurofilament light’, ‘neuropathy’ and ‘Charcot-Marie-Tooth’. These cases and the patients from our cohort were combined to determine the clinical and genetic spectrum of NEFL-related CMT. All cases with at least a simple description of the phenotype (CMT subtype and familial vs individual case) were included in the analysis of mutation frequencies. Only cases with a diagnosis of CMT and a more detailed clinical and neurophysiological description were included in the phenotypic analysis.
Internal whole-exome data
Whole-exome sequencing data from 202 patients with suspected hereditary CNS disorders investigated at the UCL Institute of Neurology, London, UK, were screened for non-synonymous NEFL sequence variants with a minor allele frequency ≤1% in the 1000 Genomes (http://www.1000genomes.org/) and the Exome Variant Server (http://evs.gs.washington.edu/EVS/) datasets.
All human studies were approved by the ethics committee of the National Hospital for Neurology and Neurosurgery and UCL Institute of Neurology, London, UK, and the Institutional Review Board of the Carver College of Medicine, University of Iowa, Iowa, USA. All patients and their relatives provided written informed consent prior to genetic testing.
Five unrelated patients with diagnosis of NEFL-related CMT were identified at the participating institutions. Clinical and investigational data are summarised in table 1 and a more detailed clinical description is provided in the online supplementary material.
Patients 1, 2 and 3 (figure 1) carried the heterozygous pathogenic mutation N98S. All had a similar phenotype consisting of motor developmental delay, hypotonia and walking difficulties before age 3–4 years, bilateral hearing loss requiring hearing aids between ages 5 and 8 years, progressive distal greater than proximal motor impairment within the first two decades and marked proprioceptive sensory deficit with positive Romberg’s test and gait ataxia. Patients 1 and 2, who had the longest follow-up, also developed proximal lower limb weakness and scapular winging, appendicular ataxia and cerebellar features supported by neuro-otological examination or brain MRI (figure 2). Additional features included tongue hemiatrophy in patient 1, mild cervical spinal cord volume loss in patients 1 and 2 and optic nerve hypoplasia with reduced central vision in patient 3. Neurophysiological studies were consistent with a length-dependent motor and sensory neuropathy with reduced or absent compound muscle action potential (CMAP) amplitudes and upper limb MNCV ranging from 23 to 35 m/s (corresponding CMAPs 1.5 and 0.8 mV, respectively; see online supplementary table 2). In patients 1 and 2, ulnar nerve CMAP waveforms were dispersed or complex on proximal stimulation. Cosegregation analyses indicated that the heterozygous mutation N98S occurred de novo in the three probands. Biological paternity and maternity was confirmed by short-tandem repeat analysis for patients 1 and 2 (results not shown; patient 3 was not tested).
Patient 4 carried the heterozygous pathogenic mutation P8R. He developed progressive distal motor and sensory impairment from the second decade along with restless legs syndrome and periodic limb movements disorder. Neurophysiological studies in his 40s revealed a length-dependent motor and sensory neuropathy with reduced or absent CMAP amplitudes and upper limb MNCV ranging from 31 to 44 m/s (corresponding CMAPs 1.7 mV; see online supplementary table 2).
Patient 5 carried the novel heterozygous missense variant L311P. He presented in the sixth decade with hand tremor followed with gait ataxia and muscle weakness and atrophy in the lower limbs. Neurophysiological studies confirmed a length-dependent motor and sensory neuropathy with reduced CMAP amplitudes in the lower limbs, upper limb MNCV ranging from 43 to 55 m/s (corresponding CMAPs 6.6 and 5.4 mV) and dispersion of CMAP waveforms in two nerve segments (see online supplementary table 2). Pathogenicity of L311P was suggested by its absence in an in-house exome database of 138 healthy controls and population databases and the results of the in silico analyses (see online supplementary table 3).
Sural nerve biopsies from patients 2 and 4 at ages 24 and 46, respectively, showed severe reduction in myelinated fibre density with complete absence of large myelinated fibres (figure 3A,B). Immunostaining for neurofilament proteins showed reduced density of axons, mainly of large axons. There was no evidence of active axonal degeneration. Definite regeneration clusters were observed in patient 4. Both biopsies contained irregularly shaped fibres. Disproportionately thin myelin sheaths and onion bulbs, confirmed by electron microscopy imaging, were seen in patient 2 (figure 3C) but not in patient 4 (figure 3D).
A systematic review of the literature revealed published data on 62 kindreds with NEFL-related CMT.8–18 23–47 These cases and the five unrelated patients from our cohort were combined for analysis.
A total of 30 different NEFL mutations have been reported to date (table 2 and online supplementary table 3).8–18 23–47 For descriptive reasons, the mutations c.22_23delCCinsAG and c.23C>G, causing the same amino acid substitution (P8R), were classified as one single mutation.
For two mutations (E186* and Q334P), there was no clinical data available.40 43 R421* has been reported in a family with an overlapping myopathic–neurogenic phenotype39 and also in association with CMT40 but no clinical details were available for the latter.
Of the remainder 27 mutations, 17 were private to a single kindred and 10 were observed in two or more kindreds. Twenty-four mutations were clustered in six regions of the NFL polypeptide (figure 4): the initial segment of the head domain (amino acid positions 8–22), the junction between the head and the rod domains (90–98), coil 2A (265–268), the mid portion of coil 2B (311–336), the end portion of coil 2B (384–396) and the tail subdomain A (421–440). Mutations in the head domain and the two ends of the rod domain accounted for 75% of kindreds and four common mutations within these regions (P8R, P22S, N98S and E396K) were observed in 55% of kindreds. Only three mutations (E140*, A149V and E210*) were located in coil 1B. No mutations were observed in linker regions of the rod domain or the tail subdomain B.
A detailed analysis of all mutations, including conservation and pathogenicity prediction scores, is provided in the online supplementary table 3. Pathogenicity for 22 of the 30 mutations was supported by cosegregation studies, existence of other mutations in the same amino acid position, detection in more than one kindred or functional studies (in vitro or animal models). This information was not available for the remaining seven mutations (E140*, E186*, L311P, L333P, Q334P, L336P, F439I and R421*); of these, only F439I is described in the ExAC dataset, being present in 13 alleles (minor allele frequency of 0.01%). According to the ACMG/AMP guidelines,48 nine mutations were classified as variants of uncertain significance (see online supplementary table 4).
Of the 67 kindreds with NEFL mutations, 27 correspond to individual cases and 40 correspond to families or familial cases. Of the latter, 29 had an autosomal dominant inheritance based on cosegregation analysis. Cosegregation was not reported or was not available in nine additional families with probable autosomal dominant inheritance; only two of them, however, had private mutations (L311P reported in this study and P8Q). The inheritance pattern was autosomal recessive in two consanguineous families in which all affected individuals had homozygous nonsense mutations located in coil 1B (E140* and E210*).13 14
Of the 27 individual cases analysed, 1 was adopted and 26 were reported as isolated or sporadic. Eight of these patients had de novo mutations based on cosegregation analysis. Of note, 11 out of 15 patients with substitutions of amino acids E90 and N98 (E90K and N98S/T) were isolated/sporadic and 6 of them had confirmed de novo mutations.10 14
Although descriptions were not uniform across the literature, a neurophysiological classification of the peripheral neuropathy was reported for 60 individual cases or families with the following distribution: CMT1 or demyelinating neuropathy in 18 kindreds (30%), CMT2 or axonal neuropathy in 25 kindreds (42%) and intermediate CMT or neuropathy with axonal and demyelinating features in 17 kindreds (28%).8 9 11–18 23–33 35 36 38 41 42 44 45 47
Upper limb nerve conduction data were available for 115 patients from 53 kindreds: 44 patients (38%) had all reported upper limb MNCV ≥38 m/s, 24 patients (21%) had upper limb MNCV ranging from <38 to ≥38 m/s and 47 patients (41%) had all reported upper limb MNCV <38 m/s.10 11 13–18 23 24 26–32 34 35 37 38 45 46 Upper limb CMAP values were available in 35 patients from the latter group, of whom 15 had amplitudes ≥ 75% of lower limit of normal.
Analysis of the same upper limb MNCV data using 25 and 45 m/s as cut-off values revealed that 12 patients (10%) had all reported values>45 m/s, 74 patients (64%) had all reported values between 25 and 45 m/s and 9 patients (8%) had all reported values<25 m/s; the remaining 20 patients had upper limb MNCV in overlapping ranges.
Median and/or ulnar MNCV<20 m/s were detected in eight patients, five of them with recessive mutations (E140* and E210*) and three with dominant mutations (P8L, E90K and N98S).10 13 14 18 Of the four common NEFL mutations, only N98S was consistently associated with upper limb MNCV<38 m/s although MNCV in proximal nerve segments varied between 30.1 and 47.8 m/s in one kindred (classified as dominant intermediate CMT).10 14 18 24 34 46
Sural nerve histopathology
Twenty-one sural nerve biopsies of patients with NEFL-related CMT, including two from our cohort, have been reported to date (table 3).10 11 13 14 16 25 27 28 31 33 38 Loss of large myelinated features was a uniform feature. Onion bulbs and/or thin or irregularly folded myelin sheaths were observed in 14 biopsies (67%) and giant axons, with or without thin myelin sheaths, were found in six (29%).
To date, a total 173 individuals with NEFL-related CMT have been reported as clinically examined.8–18 23 24 27–38 44–46 The majority developed a classical CMT phenotype but some of them had additional features including hearing loss, tremor, cerebellar features and pyramidal tract signs (tables 3 and 4).
Early onset with delayed motor milestones. Twenty-five patients from 18 kindreds had onset of symptoms at or before age 3 years.10 13 14 18 31 34 35 37 38 46 Of them, 18 patients were reported as having delayed walking or motor milestones with or without additional features such as hearing loss (11 patients), hypotonia (8 patients), intellectual impairment (3 patients), growth retardation (2 patients) or optic nerve hypoplasia (1 patient). Four dominant mutations (P8L, E90K, N98S and E396K) and one recessive mutation (E210*) were associated with both early-onset and motor delay. With one exception, all patients with substitutions of amino acids E90 and N98 (E90K and N98S/T) had onset at age ≤3 years.
Pyramidal tract signs. A variable combination of brisk reflexes, extensor plantar responses and lower limb spasticity (sometimes simply described as upper motor neuron involvement), with or without evidence of prolonged central motor conduction times, was observed in 13 patients from seven kindreds.14 16 17 30 37 44 Except for the mutation Y265C, all mutations associated with pyramidal tract signs were located in end portion of the rod domain (Y389C and E396K) and the tail subdomain A (F439I and P440L).
Cerebellar features. Limb or gait ataxia (sensory, cerebellar, episodic or described as gait ataxia or wide-based gait) was reported in 22 patients but distinction between sensory or cerebellar ataxia based on the available data was not always possible.11 14–18 27–29 Seven patients carrying the mutations P8R, N98S, L311P, C322_N326del and E396K were initially diagnosed with or tested genetically for spinocerebellar ataxia or Friedreich’s ataxia, four of them with dysarthria and/or nystagmus. Cerebellar atrophy on brain MRI was observed in four patients with the mutations N98S and E396K (table 3).14 18 46
Raised creatine kinase (CK) levels and proximal weakness. Raised CK levels up to 1414 IU/L have been detected in 14 patients carrying the mutations N98S, Q332P and E396K.17 36 38 Of these patients, seven had an element of proximal weakness in the upper or lower limbs and/or a waddling gait. This gait pattern is also described in patients with early-onset disease due to N98S and E396K.14 Myopathic features on EMG or muscle biopsy, however, have only been reported in two individuals with the E396K mutation and in one family with a myopathic-neurogenic phenotype carrying the mutation R421*.38 39
Screening for additional mutations
Based on the observation that patients with NEFL mutations may develop CNS features, we retrospectively reviewed whole-exome sequencing data from patients with pure or complicated hereditary spastic paraplegia (27 cases), early-onset ataxia (100 cases) and complex CNS syndromes (75 cases with a variable combination of ataxia, spasticity, epilepsy, extrapyramidal features and cognitive involvement), searching for rare or novel NEFL variants. Two heterozygous missense NEFL variants were identified: Q537R (c.1610A>G, rs377121179) and L336V (c.1006C>G, rs551896980). Both variants are present in population databases (minor allele frequency of 0.03% and 0.02%, respectively, in the ExAC dataset) and predicted to be tolerated or benign by SIFT and PolyPhen2. No other rare variants in NEFL were found.
In this study, we have reported five new cases of NEFL-related CMT and performed an analysis of the entire genetic and clinical spectrum of the disease based on data from the literature.
We observed that NEFL mutations are clustered in terms of frequency and location in six regions of the NFL protein, predominantly in the initial segment of the head domain, the junction between the head and rod domains and the end portion of the rod domain. The pathogenic relevance of these three regions is also supported by in vitro and animal studies. Substitution of amino acids P8 and P22 of NFL disrupts its ability to self-assemble into a filamentous network,49–52 affects the axonal transport of neurofilaments and mitochondria51 53 54 and impairs motor neuron viability in vitro.55 Transient transfection studies also indicate that T21Afs*83, E90K and N98S result in NFL assembly defects.12 50 In addition, mice models of the mutations N98S and E396K, located at the two ends of the rod domain, display an abnormal hindlimb posture together with muscle atrophy and reduced MNCV in the hNFLE396K mice and tremor in the NeflN98S/+ mice.56 57
Mutations affecting three other regions of NFL (coil 2A, mid portion of coil 2B and tail subdomain A) were less common. Six mutations involved the mid region of coil 2B, including Q332P, which has been detected in two families and causes similar defects in neurofilament assembly and transport as P8R in vitro.49 51 53 For four mutations located in this region (L333P, Q334P, L336P and L311P reported in this study), however, there was no cosegregation or functional data available. Three mutations were located in the tail subdomain A but cosegregation data were not reported for the family with CMT and R421* or for F439I, which has a minor allele frequency of 0.01%. Therefore, further research is needed to determine if these regions of NFL are mutational hotspots.
NEFL-related CMT presented with a broad clinical and neurophysiological phenotype. The mode of transmission was usually autosomal dominant although 40% of kindreds in our study were classified as sporadic or isolated cases and de novo mutations were confirmed in eight of them. Disease onset varied between the first and sixth decades of life, with 13% of patients having an early-onset presentation (≤3 years) often associated with delayed walking or motor milestones. Neurophysiologically, NEFL-related CMT was classified as axonal, demyelinating or mixed/intermediate in 42%, 30% and 28% of kindreds, respectively; descriptions were not uniform across the literature, but our analysis of published data also suggests that neuropathies with MNCV in the demyelinating or intermediate ranges are overall more frequent than those with upper limb MNCV>38 m/s. Although the reduced velocities may well be mainly explained by preferential loss of large-myelinated fibres confirmed in 21 sural nerve biopsies, additional features suggestive of demyelination and remyelination (onion bulbs and/or thin myelin sheaths) were observed in 67% of them.
Establishing definite genotype–phenotype correlations in patients with NEFL-related neuropathies is difficult given the significant number of private mutations and low prevalence of the disease. Pyramidal tract signs might be preferentially associated with mutations located towards the end of the NFL protein (end of rod domain and tail subdomain A) but this should be confirmed in future studies. Ataxia was described in association with the four common mutations (P8R, P22S, N98S and E396K), and raised CK levels were more frequently described in association with the mutation E396K.
We identified, however, an important clinical association with the mutations E90K and N98S (n=14). Most patients with these mutations were isolated or sporadic (71%), six with confirmed de novo mutations (three from our cohort); presented before age 3 (93%) with delayed walking or motor milestones (86%); had all reported upper limb MNCV in the demyelinating range (86%); and developed additional features such as hearing loss (64%), ataxia with or without nystagmus or dysarthria (36%), hypotonia (29%) and less frequently intellectual impairment, tongue hemiatrophy, signs of vestibular dysfunction and optic nerve hypoplasia; cerebellar atrophy on brain MRI was confirmed in three cases. Patients carrying E90K and N98S, therefore, presented with a rather uniform syndrome for which we propose the acronym ‘DECADDE’ (demyelinating neuropathy based on distal upper limb MNCV with cerebellar signs and/or ataxia, deafness and delayed motor milestones), which also highlights the early onset of the disease. Although only 5 out of 14 patients conformed to the full syndrome, its recognition may aid in the differential diagnosis with other autosomal recessive cerebellar ataxias.18 Of note, NFL E90K or N98S mutants share a particular phenotype in vitro, different from that observed with substitutions of amino acid P8,50 and NeflN98S/+ mice display an early-onset phenotype with tremor and pathological changes in the CNS including neurofilament aggregates in the spinal cord and cerebellum.56
Beyond the above genotype–phenotype associations, identification of uncommon clinical features such as hearing loss, pyramidal tract signs or ataxia may be useful in the differential diagnosis of NEFL-related CMT with other forms of CMT (see online supplementary table 5). In our cohort of 202 patients with CNS syndromes, we found no other potential NEFL mutations, which supports that these are unlikely to present with a predominantly CNS phenotype.
We acknowledge this study’s limitations related to the amount, completeness and heterogeneity of the data in the literature. Missing data may indeed reflect true negative data (ie, absence of a clinical feature), false negative data due to lack of assessment or unreported data. Also, cases with atypical clinical characteristics may be overrepresented in the literature because of their particular interest. We believe, however, that the results of the present study may be helpful in the evaluation of novel NEFL sequence variants and the differential diagnosis with other CMT subtypes. Also, our study highlights potential aspects of interest to be evaluated in future studies.
Contributors AH was involved in the study design, the acquisition, analysis and interpretation of data and drafted the manuscript. ML, ZJ, NUJ, MAG, JMP, RP, JCB, Y-TL, SW, CB, MPTL, HM, MGH, SB and HH were involved in the acquisition and interpretation of data and reviewed the manuscript. SZ and MS were involved in the study design, acquisition and interpretation of data and reviewed the manuscript. MMR supervised the study and was involved in the study design, acquisition and interpretation of data and reviewed the manuscript. All authors approved the final manuscript.
Competing interests None declared.
Ethics approval National Hospital for Neurology and Neurosurgery and UCL Institute of Neurology Ethics Committee and Institutional Review Board of the Carver College of Medicine, University of Iowa.
Provenance and peer review Not commissioned; externally peer reviewed.
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.