Introduction
Charcot-Marie-Tooth (CMT) disease, also known as hereditary motor and sensory neuropathy, is a common, clinically heterogeneous group of inherited peripheral neuropathies with an estimated prevalence of one in 2,500 individuals (Wiszniewski et al., 2013; Di Vincenzo et al., 2014).
CMT is clinically, neurophysiologically, and genetically heterogeneous. It is most commonly characterized by sensory loss that starts in the lower limbs and progresses slowly in a length-dependent manner. This produces progressive distal muscle atrophy, weakness, distal sensory loss, foot deformities, and depressed tendon reflexes (Shy et al., 2005; Rossor et al., 2013).
The clinical classification is based on age at onset, distribution of muscle weakness, sensory loss, walking difficulties, and foot deformities (Rossor et al., 2013). Neurophysiology allows subdividing the disease into a demyelinating (CMT1) and axonal (CMT2) forms depending on whether the median motor nerve conduction velocity (NCV) is below or above 38 m/s, respectively. A third form, intermediate CMT, has both demyelinating and axonal features and NCV between 25 and 45 m/s (Reilly and Shy, 2009; Berciano et al., 2012).
The duplication of PMP22 is the most common cause of CMT, with a prevalence up to 40% in some populations (Reilly and Shy, 2009; Abe et al., 2011; Saporta et al., 2011; Murphy et al., 2012; Østern et al., 2013; Sivera et al., 2013). Then, approximately 100 different genes have been linked to CMT-like phenotypes which are associated with related conditions involved in axonal transport, myelin structure, and membrane metabolism that have been found in multiple unrelated families or confirmed by functional studies (Saifi et al., 2003; Saporta et al., 2011; Azzedine et al., 2012; Vallat et al., 2013; IPNMDB, 2014; OMIM, 2014; Washington University, 2014).
Therefore, the large spectrum of genetically identifiable disease alleles complicates the molecular diagnosis. Genetic heterogeneity is associated with a wide spectrum of phenotypes, complicated by the fact that mutations in the same gene cause different phenotypes (Reilly and Shy, 2009; Azzedine et al., 2012; OMIM, 2014). Furthermore, sporadic cases of CMT are not uncommon due to autosomal recessive inheritance, reduced penetrance, late-onset, small family size, and de novo mutations (Braathen et al., 2011; Saporta et al., 2011; Vallat et al., 2013).
In this study, we enrolled 67 patients and applied a diagnostic protocol including multiple ligation-dependent probe amplification (MLPA) for copy number variation (CNV) detection of PMP22 locus, and next-generation sequencing (NGS) for sequencing of 47 genes known to be associated with CMT and routinely screened in medical genetics.
The goal of this study is to identify the genetic features in the recruited cohort of patients, highlighting the role of rare variants in the genotype-phenotype correlation.
Discussion
In this study, 67 patients with a clinical diagnosis of CMT were selected for molecular analysis of genes known to be related to the disease. Although the genetic of CMT is featured by a strong genetic heterogeneity, most of the patients carry 17p CNV of PMP22. Among this CNV, almost all cases carry the whole gene duplication of PMP22. The prevalence of the duplication varies among populations, ranging from 15% in Norwegian studies, up to more than 50% in Italy and Spain (Abe et al., 2011; Saporta et al., 2011; Murphy et al., 2012; Østern et al., 2013; Sivera et al., 2013; Høyer et al., 2014). In this study, only 30% of Italian patients (20/67) carry the whole gene duplication of PMP22, thus under-representing other Italian studies though still being in line with prevalence measured in European studies (Figure 2B).
When ruling out carriers of 17p CNV in PMP22, NGS identified the causative variants in 8/41 patients, which correspond only to 12% of the patients analyzed in this study (8 out of 67), being over 50% of the total of causative variants confirming the clinical diagnosis detected in the present analysis. Thus, combining various genetic tools using a target panel of 49 genes allowed to cover to a greater extent the complex clinical setting.
Several cohort studies tried to identify the diagnostic yield in CMT patients. To this aim, we conducted a systematic review of 23 cohort analysis studies consisting of Italian and European CMT patients in the period between 1997 and 2020. This data, summarized in Figure 2B and Supplementary data, identified a diagnostic yield of 71.7% in Italy (Mostacciuolo et al., 2001; Manganelli et al., 2014; Gentile et al., 2020) and 26.7% in Europe (Bort et al., 1997; Saporta et al., 2011; Murphy et al., 2012; Gess et al., 2013; Østern et al., 2013; Sivera et al., 2013; Høyer et al., 2014; Antoniadi et al., 2015; Laššuthová et al., 2016; Lupo et al., 2016; Dohrn et al., 2017; Marttila et al., 2017; Bacquet et al., 2018; Hoebeke et al., 2018; Milley et al., 2018; Nicolas et al., 2018; Lerat et al., 2019; Vaeth et al., 2019; Cortese et al., 2020; Figure 2B). Therefore, the prevalence of genetic cases of CMT provided by the present data represents an average compared with the general prevalence measured in Italy and Europe. The differences in the diagnostic rate may be explained by the differences in the specific features of the cohorts being analyzed, the number of demyelinating CMT cases enrolled, and the heterogeneous inclusion criteria considering the common causative, which were genes adopted by the previous MLPA and Sanger sequencing.
In our cohort, NGS approach identified variants in eight different genes: MFN2, MPZ, GDAP1, SH3TC2, HSPB1, KIF5A, MTMR2, and KIF1A (Table 2), five responsible for demyelinating CMT, and three for axonal or intermediate CMT. These NGS-identified variants appeared as sporadic (3/8) or familial (5/8).
A frequency of sporadic or de novo mutations is reported in 28–34% (Marques et al., 2005; Ando et al., 2017), although this percentage may vary and some studies report up to a half of the patients as sporadic (Ando et al., 2017). In this study, variants considered as sporadic are reported in three out of eight patients with a NGS genetic diagnosis (37.5%). This percentage, in line with other studies, may represent an overestimation since this trend cannot be verified in a few families where a full familial anamnesis is lacking. This point is crucial since it demonstrates how a careful, multifaceted genetic approach with a complete familial anamnesis and genetic segregation analysis could provide a reduction of sporadic CMT.
Among sporadic variants, the proband Family ID 564 (Figure 1A) is a carrier of a whole gene mosaic in the PMP22 gene. This is consistent with the high instability of the PMP22 gene which undergoes de novo PMP22 duplications in up to 90% of sporadic CMT1 (Boerkoel et al., 2002; Marques et al., 2005).
The other probands carry sporadic variants in the gene MFN2, MPZ, and HSPB1. Mutations in MFN2 cause CMT type 2A by altering mitochondrial fusion and trafficking along with the axonal microtubule system (Guerriero et al., 2020). Pathogenic variants in MFN2 are typically inherited as autosomal dominant and are usually missense variants (Di Vincenzo et al., 2014).
The sporadic variant identified, Family ID 883 (Figure 1B), consists of a single base insertion in the coiled-coil domains producing a stop mutation p.(Gln754AlafsTer9). This could be a de novo variant. De novo MFN2 mutations are regularly found in patients with a classical CMT2 phenotype (Østern et al., 2013), and some MFN2 mutations have also been reported as de novo in several patients (Züchner et al., 2004; Chung et al., 2006; Verhoeven et al., 2006; Cho et al., 2007). The early onset of the proband, at 7 years old, is in line with data reporting that mutations in MPZ, MFN2, or NEFL are the most frequent disease causes of patients with infantile-onset CMT (Hsu et al., 2019).
Thanks to NGS studies, several data are broadening the phenotype spectrum produced by mutations in some CMT genes. For example, mutations in the N-terminal motor domain of KIF5A are responsible for hereditary spastic paraplegia (SPG10 MIM 604187) and CMT type 2 (CMT2), clustered in the switch regions SWI (199–204) and SWII (232–237) necessary for microtubules interaction (Filosto et al., 2018). On the other hand, mutations in the C-terminal cargo-binding tail domain (Tail domain) are related to a specific ALS phenotype (ASL MIM 617921), with an early disease onset but a longer survival compared with typical amyotrophic lateral sclerosis (ALS; Brenner et al., 2018; Nicolas et al., 2018). Beyond these main phenotypes, other complex and overlapping phenotypes are emerging, which demonstrates that KIF5A mutations may be responsible for a wide range and heterogeneous disease spectrum (Schüle et al., 2008; Tessa et al., 2008). For example, in a patient, a mixed slowly progressive disease was described resembling ALS as well as HSP and axonal neuropathy. This is caused by a mutation within the terminal region of the stalk domain (Cho et al., 2007). Additionally, the patient Family ID 1141 (Figure 1H) reported in this paper represents the first case of a familial mutation, the new heterozygous deletion p.Leu957del, falling in the tail domain of KIF5A producing spastic paraplegia 10 with autosomal dominant neuropathy (MIM 604187).
Taken together, these results broaden the phenotypic spectrum of KIF5A mutations and confirm the importance of cytoskeletal defects in the pathogenesis of both ALS and CMT.
In line with this, the phenotype related to heat shock protein 27 (HSP27) is broadening. For instance, mutations in the HSPB1 gene have been reported to cause autosomal dominant CMT with minimal sensory involvement (CMT 2F MIM 606595). Then, two independent studies reported different HSPB1 mutations in two sporadic cases and one consanguineous family with ALS, suggesting that the disease spectrum of HSP27 may not be limited to CMT2/dHMN (Scarlato et al., 2015; Capponi et al., 2016; Adriaenssens et al., 2017). In this study, we identified the known heterozygous missense mutation p.Gln190H in Family ID 402 (Figure 1D), which falls in the C-terminal domain involved in the control of its chaperone-like activity, responsible for CMT, axonal, and type 2F autosomal dominant (MIM 606595; Lelj-Garolla and Mauk, 2012). This variant falls in the C-terminal domain involved in the control of chaperone-like activity (Lelj-Garolla and Mauk, 2012; Capponi et al., 2016). It has been previously reported in a sporadic ALS patient with onset at 58 yo, and axonal neuropathy with an intermediate conduction velocity (38 m/s; Capponi et al., 2016). Therefore, considering this study in which the same variant has been identified in a CMT patient, the genotype-phenotype correlation is very complex. Functional characterization of different mutations located in the C-terminal domain did not show a clear correlation between the location of the mutation and associated cellular phenotype. This suggested that the genomic variant, more than its location, determines the cell pathology and phenotype, However, the present study contradicts such a conclusion which is not consistent with the present data which indicate how the same substitution has been identified in patients owing to a different disease (ALS in one case CMT in another; Capponi et al., 2016). This suggests the importance of how specific genetic backgrounds may alter the expression of the genetic variants while explaining how rare variants, like VoUS may be responsible for the disease.
In this study, seven VoUS were identified in six (7%) patients. These variants, all in heterozygous status, all fall in one of those genes associated with autosomal dominant and recessive CMT: AGRN, SCP2, DNM2, MED25, DYNC1H1, and BSCL2. This high number of VoUS is in line with the previous studies reporting a higher number of VoUS in neuropathy-associated genes, including single mutations in autosomal recessive CMT genes compared with the control population, with some patients presenting more than one VoUS in different CMT genes (Gonzaga-Jauregui et al., 2015). For instance, here, we report multiple VoUS in the proband of Family ID 721 (Figure 3A) characterized by two VoUS in AGRN and SCP2. A recent hypothesis suggests that a combinatorial effect of rare variants contributes to the disease burden in CMT and partly explains its various phenotypes. This confirms what was demonstrated by using in vivo experimental models in zebrafish (Cortese et al., 2020). Therefore, the interpretation of VoUS remains a key diagnostic challenge in the current NGS era.