Introduction

Neuromuscular disorders such as congenital muscular dystrophy (CMD), Duchenne muscular dystrophy (DMD), limb-girdle muscular dystrophy (LGMD), congenital myopathy (CM) and congenital myasthenic syndrome (CMS) manifest during the neonatal and childhood periods together with the onset of hypotonia and muscle weakness. These disorders form a clinically and genetically heterogeneous group of diseases that present common features, mainly hypotonia, muscle weakness, motor retardation and abnormal posture. The broadly overlapping clinical characteristics can make diagnosis challenging.

Neuromuscular disorders have traditionally been diagnosed based on the onset period, clinical symptoms, clinical course and neurological findings. However, because of major advances in molecular genetics over the past 20 years, neuromuscular disorders can now be definitively diagnosed through molecular genetic testing. Starting with the discovery of the dystrophin gene in 1987,1 the causative genes related to numerous neuromuscular disorders have been successively identified, and these genes currently number 519.2 Today, neuromuscular disorders are diagnosed based on both characteristic pathological findings and causative genes.3, 4, 5, 6 For CMD and spinal muscular atrophy (SMA) in particular, diagnosis through molecular genetic testing without muscle biopsy is becoming the gold standard. Furthermore, although CM can be differentiated to a certain degree based on muscle computed tomography and magnetic resonance imaging (MRI) findings mainly pertaining to the distribution of affected muscles, the findings are not sufficiently specific for CM and similar disorders. Therefore, muscle biopsy and genetic testing are essential for definitive diagnosis. However, during genetic diagnoses of neuromuscular disorders, difficulties arise due to the particular features of these disorders, and a definitive diagnosis is thus not obtained in nearly 40% of affected patients.7, 8, 9 The critical genetic features of the aforementioned neuromuscular disorders include several of the causative genes being very large, broadly overlapping genotypes and clinical phenotypes, and the existence of several genes related to a single phenotype.

With regard to causative genes, many of the proteins expressed in muscle are extremely large, and their genes thus contain numerous exons. For example, the DMD gene is 2200 kb in total length and contains 79 exons, the TTN gene encoding giant titin proteins contains 363 exons, and the NEB gene encoding nemaline proteins comprises 183 exons. One example of genotype/phenotype overlap is α-dystroglycanopathy: nearly 20 genes have been identified that encode proteins involved in α-dystroglycan-related pathways, and these disorders are considered to be manifested through a common mechanism. The clinical phenotypes cover a very wide range from Walker-Warburg syndrome, which is a severe form of CMD accompanied by severe brain-structure abnormalities and ophthalmic abnormalities, to mild LGMD in the form of late-onset muscle weakness.10, 11 Lastly, an example of several genes being related to a single phenotype is nemaline myopathy (NM), the most frequent form of CM and a disease group characterized by the presence of nemaline bodies.12, 13 To date, nine genes have been identified that encode thin-filament proteins and related proteins: TPM3, NEB, ACTA1, TPM2, TNNT1, KBTBD13, CFL2, KLHL40 and LMOD3.

Because several of the aforementioned genes are large, substantial labor, time and costs are required for the individual analysis of causative genes through conventional Sanger sequencing for differential diagnosis of clinically and genetically heterogeneous neuromuscular disorders. However, the recent advent of next-generation sequencing (NGS), which enables investigators to perform rapid high-throughput analysis, has made it possible to easily reduce both costs and time. Furthermore, by combining NGS results and pathological findings from muscle biopsy, causative genes can now be identified rapidly, efficiently and reliably.14, 15 Herein, we performed NGS-based target resequencing on causative genes related to neuromuscular disorders in 42 patients in whom undiagnosed neuromuscular disorders were suspected based on the manifestations that had developed during the neonatal through childhood period.

Materials and methods

Patients

The study participants were 42 Japanese patients who had been referred to the Institute of Medical Genetics at Tokyo Women’s Medical University between 2004 and 2014. The patients had shown symptoms of muscle weakness and hypotonia neonatally and/or during childhood, and neuromuscular disorders were thus suspected based on their clinical symptoms and muscle pathology findings. Pediatric neurology specialists made diagnoses based on the presence of muscle weakness and hypotonia, other physical findings, clinical course, muscle imaging features and muscle pathology findings, and divided the 42 patients into three groups accordingly: (1) MD group, 20 patients (MD-1 to MD-20); (2) CM group, 17 patients (CM-1 to CM-17); and (3) SMA group, five patients (SMA-1 to SMA-5) (Figure 1 and Table 1).

Figure 1
figure 1

Diagnostic process of this study and three clinical categories of undiagnosed patients. CGH, comparative genomic hybridization; CK, creatine kinase; CM, congenital myopathy; CT, computed tomography; EMG, electromyogram; FISH, fluorescence in situ hybridization; MD, muscular dystrophy; MLPA, multiplex ligation-dependent probe amplification; NCV, nerve conduction velocity; SMA, spinal muscular atrophy.

Table 1 Summary of clinical information and investigations, and conventional genetic tests performed

Muscle biopsy was performed in 16 of the 42 patients. For 34 patients, a definitive diagnosis was not yielded by conventional genetic tests (DMD gene: multiplex ligation-dependent probe amplification (MLPA); and/or SMN1 gene exons 7, 8: MLPA; and/or FKTN gene: detection of ancestral founder mutation containing a 3-kb retrotransposal insertion in the 3′-non-coding region; and/or DMPK gene: CTG repeat detection). However, in patient MD-12, deletion of exons 45–53 in the DMD gene causing BMD was detected using MLPA, but this did not correspond to the patient’s clinical findings, which were thus attributed to a potentially more pathogenic mutation.

Patients with suspected genetic chromosomal disorders such as trisomy 21 and Prader–Willi syndrome, central nervous system disorders or spinal cord injuries were excluded. Among neuromuscular disorders, myotonic dystrophy is caused by the expansion of a CTG trinucleotide repeat in the non-coding region of DMPK gene. Patients with these gene expansions were excluded because the expansions cannot be detected using NGS.

This study was conducted after explaining its purpose and contents to all patients and their parents, and obtaining written consent. Moreover, this study was approved after review by the ethics committee of Tokyo Women’s Medical University (approval number: 2709).

Target resequencing

We used the SOLiD 4 system platform and the Ion PGM platform (Thermo Fisher Scientific, Waltham, MA, USA). We first used the SOLiD 4 system in 25 patients. Later, Ion PGM was introduced. Thirteen patients whose pathogenic mutations were not identified using the SOLiD 4 system and 18 patients who had been referred after May 2012 underwent testing using the Ion PGM platform.

Panel design

SOLiD

The 42-gene panels (Supplementary Table S1) were designed using the eArray tool (Agilent Technologies, Santa Clara, CA, USA) for the capture of non-contiguous target muscle-disease genomic regions that were approximately 3 Mb in total length.

Ion PGM

The 74-gene panels (Supplementary Table S1) were designed automatically using Ion Ampliseq Designer (Thermo Fisher Scientific); two pools of 3786 primers were generated for amplifying genomic regions. Coverage included the coding DNA sequence region and at least five intronic flanking nucleotides of each targeted gene to capture intron/exon splice junctions, as well as both 5′ promoter regions and 3′ untranslated regions. The amplicons covered 95.32% of the target sequence (515.15 kb).

Library preparation and sequencing

SOLiD

We used standard protocols to extract genomic DNA from blood, and then used 3 μg of the DNA to construct a library composed of adaptor-ligated randomly fragmented DNA by employing the SOLiD Fragment Library Construction Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. The adaptor-ligated DNA was captured through hybridization in solution with custom-designed cRNA oligonucleotide baits as per the manufacturer protocols. DNA sequencing was performed using Paired-End Sequencing (version 1.3) on the SOLiD 4 system.

Ion PGM

Genomic DNA was extracted from blood by using standard protocols, and 10 ng samples of the DNA were amplified and adaptor-ligated libraries were prepared by employing the Ion Ampliseq Library Kit with Ion Ampliseq Custom Primer Pool protocols according to the manufacturer’s instructions (Thermo Fisher Scientific). DNA was sequenced using the Ion PGM 200 Sequencing Kit and an Ion Torrent PGM (Thermo Fisher Scientific).

Data analysis

Data from the SOLiD and the PGM runs were processed using SOLiD BioScope Software v1.3.1 and Ion Torrent Suite 3.2 software (Thermo Fisher Scientific), respectively. Sequences were aligned to the human genome reference (hg19). The obtained single-nucleotide variants and insertions/deletions were combined. All of the variants were filtered against dbSNP132 and annotated using ANNOVAR and a custom analysis pipeline employing dbNSFP,16 which includes the relevant prediction scores such as SIFT17 and PolyPhen-2,18 conservation scores such as PhyloP19 and GERP++,20 and other related information, including allele frequencies observed in the 1000 Genomes Project data21 and the NHLBI Exome Sequencing Project ESP6500 data.22

Sanger sequencing-based confirmation of variants detected using NGS

All potential pathological variants detected using NGS were confirmed by means of Sanger sequencing performed using a 3130xl Genetic Analyzer (Thermo Fisher Scientific), according to the manufacturer’s protocol.

mRNA analysis

When aberrant splicing was suspected, total RNA was extracted from blood using a TRIzol RNA Isolation Reagents (Thermo Fisher Scientific) and reverse transcribed with an oligo(dT)20 primer using SuperScript III reverse transcriptase (Thermo Fisher Scientific). mRNA was amplified by reverse transcriptase-PCR and the amplified fragment was directly sequenced by Sanger sequencing.

Validation of identified variants

We determined variants that had been reported previously in the literature or were classified as pathogenic mutations in databases, and we classified novel nonsense mutations, insertions and deletions, and splice-site mutations as pathogenic because they alter protein structures. We determined a variant to be a novel missense variant if it was not present in any of these databases: 1000 Genomes Project data, NHLBI Exome Sequencing Project ESP6500 data, the NCBI dbSNP database23 and the Leiden Muscular Dystrophy database.24 Furthermore, we classified these identified mutations as pathogenic missense mutations if the gene corresponded to the patient’s clinical features, such as histological and ultrastructural findings from muscle biopsy or other specific and significant findings, and referred to the aforementioned prediction scores to evaluate allele frequency and determine whether the gene was pathogenic.

A disease diagnosis was considered 'definitive' if a pathogenic mutation was either detected in both alleles of a gene showing autosomal recessive inheritance, or detected in one allele of a gene showing autosomal dominant inheritance. If a patient carried two autosomal recessive mutations, we confirmed whether one gene carrying the mutation had come from each parent.

Finally, we considered a novel mutation to be a variant of uncertain significance if it was detected but could not be confirmed in either parent, or if we could not readily determine whether the mutation was pathogenic.

Results

Sequencing summary

The SOLiD sequencing yielded an output of 0.79 Gb per sample, with a median sequencing depth of 84.17 × . The Ion PGM sequencing yielded an output of 1.95 Mb per sample, with a median sequencing depth of 442.8 × . Both approaches allow reliable detection of sequence variants with high accuracy.

Identified mutations and clinical and histological findings

Causative genes were identified in 19/42 (45.2%) patients, and 25 pathogenic mutations (12 novel and 13 previously reported mutations) were identified in 12 genes. Of these, six were novel missense mutations. The novel mutations were determined to be pathogenic based on the histological and ultrastructural findings from muscle biopsy or brain imaging features specific to each. All of these mutations were found to be associated with high levels of pathogenicity based on the prediction scores and minor allele frequencies (Table 2). Variant of uncertain significance were detected in 2 of the 42 patients (patients CM-8 and -9).

Table 2 Clinical findings and prediction scores of novel missense mutations identified by NGS

MD group

In six genes in the 20 patients with the MD phenotype (FKTN, POMT2, LAMA2, COL6A2, DMD and CAPN3), 14 pathogenic mutations were identified (six novel and eight previously reported mutations), and in 12 patients (60.0%) a definitive diagnosis was obtained (Table 3).

Table 3 Mutations detected by molecular genetic testing in the MD group

Definitive diagnoses were made in 12 cases, as follows: 5 cases with dystroglycanopathies (3, Fukuyama congenital muscular dystrophy (FCMD); 2, LGMD2N); cases with merosinopathies (one each, merosin-deficient congenital muscular dystrophy (MDC) 1A and partial MDC1A); 1 case with collagenopathy (Ullrich congenital muscular dystrophy); 2 cases with dystrophinopathy (BMD); 1 case of LGMD1A; and 1 case with concurrent BMD and FCMD. In 5 of the 12 cases that received a definitive diagnosis, muscle biopsy was not performed.

Specifically, in patient MD-1, a pathogenic mutation present within the FKTN intron was identified. In MD-9, mutant alleles were detected at a high rate (70.4%), and somatic mosaicism was identified. In MD-7, we found a heterozygous mutation in LAMA2, and the other mutation is missing, the patient’s brain MRI showed diffuse white matter hypodensities in the bilateral periventricular areas (Figure 2d), so we made the diagnosis of MDC1A. In MD-12, hypotonia was severe, the patient’s neck was unstable at age 5, and brain abnormalities were identified. In this case, an in-frame deletion of exons 45–53 in the DMD gene causing BMD was detected using MLPA, and then NGS analysis identified an FKTN point mutation. Furthermore, based on FKTN mRNA analysis, a 124-bp insertion mutation was identified near exon 4. Thus, we diagnosed concurrent BMD and FCMD.

Figure 2
figure 2

Findings of imaging and muscle biopsy. (a) T1-weighted brain MRI of patient MD-4 at age 8 months, showing mild pachygyria in the frontal region. (b) T2 FLAIR-weighted brain MRI of patient MD-5 at age 15 years, showing mild pachygyria in the frontal region. (c) T2-weighted brain MRI of patient MD-6 at age 12 months, showing diffuse white matter hypodensities in the bilateral periventricular areas. (d) T2-weighted brain MRI of patient MD-7 at age 5 years, showing diffuse white matter hypodensities in the bilateral periventricular areas. (e) CT imaging of patient MD-5 at age 15 years, showing dystrophic features predominantly in the thigh and bicep muscles. (f) Immunostained muscle biopsy specimen from patient MD-5 at age 3 years, showing absence of α-dystroglycan. (g) Control specimen for merosin immunostaining. (h) Muscle biopsy specimen obtained from patient MD-6 at age 8 months, showing dystrophic changes and absence of merosin immunostaining. (i) Merosin immunostaining of muscle biopsy specimen from patient MD-7 at age 6 years, showing diminished merosin immunostaining and non-uniform sarcolemma. (j) Gomori trichrome staining of the quadriceps femoris muscle of patient CM-1 at age 6 years, showing numerous cytoplasmic nemaline bodies, without dystrophic or inflammatory changes. (k) Electron microscopy image of a muscle specimen from patient CM-1 showing intracellular high electron-dense nemaline bodies. (l) HE staining of a muscle biopsy specimen from patient CM-5 at age 7 years, showing mild variation in fiber size and the predominance of type II fibers. (m) Electron microscopy image of a neuromuscular junction in patient CM-7 at age 8 months, showing simplified postsynaptic regions lacking junctional folds (NT, nerve terminal; PF, primary folds (arrow)).

CM group

In six genes in the 17 patients presenting the CM phenotype (NEB, RYR1, ACTA1, MYH7, DNM2 and RAPSN), 11 pathogenic mutations were detected (six novel and five previously reported mutations), and in 7 patients (41.1%), a definitive diagnosis was obtained (Table 4). Two more patients harbored likely pathogenic mutations. Patient CM-8 presented clinical features of NM such as mild hypotonia and muscle weakness with a myopathic face and scoliosis. This patient was found to carry a de novo p.Ala155Val mutation in TPM2, which was not previously reported, and the level of pathogenicity was high according to the prediction scores (Polyphen-2: 1; SIFT: 0). Patient CM-9 presented severe hypotonia and was found to harbor the novel nonsense mutation p.Arg103X in CCDC78. We highly suspected that these two mutations may be pathogenic; however, we had not taken a muscle biopsy from patient CM-8, so we could not confirm the pathological findings. Meanwhile, we could not confirm whether the parents of patient CM-9 harbored the same mutation because we could not follow-up on them any longer. Therefore, we could not readily determine whether the mutations were pathogenic.

Table 4 Mutations detected by molecular genetic testing in the CM group

The following diseases were definitively diagnosed: three cases of NM (due to mutations in NEB, RYR1 and ACTA1) and one case each of centronuclear myopathy (CNM; due to a mutation in DNM2), congenital fiber-type disproportion (CFTD; ACTA1), myosin storage myopathy (MSM; MYH7) and CMS (RAPSN). Notably, in CM-2, muscle biopsy was not performed, but NM was identified because previously reported pathogenic mutations (NEB c.24580-1G>A and c.8980delA) were identified and the patient’s clinical features clearly corresponded to the phenotypes of the reported cases.25, 26

In the case of CM-5, who had childhood-onset slowly progressive muscle weakness, no specific results were obtained except for muscle biopsy findings of myogenic changes and type II fiber predominance (Figure 2l). However, a previously reported mutation (MYH7 p.His1901Leu) was detected and the patient’s clinical features were clearly consistent with those of definite cases;27 thus, MSM was definitively diagnosed.

In CM-7, who presented a characteristic fatal akinesia deformation sequence, NGS analysis revealed compound heterozygous pathogenic mutations in RAPSN, and CMS was thereby definitively diagnosed. Subsequent electron microscopy confirmed simplification of the primary fold (Figure 2m). Thereafter, therapy with anticholinesterase agents was initiated and, currently, at 3 months after initiating therapy, ptosis of the eyelids has disappeared, and seated posture maintenance and other postural muscle improvements are ongoing.

SMA group

No causative genes were identified in any of the five patients with the SMA phenotype.

Discussion

Neuromuscular disorders are genetically and clinically heterogeneous. When we performed NGS-based target resequencing for patients with undiagnosed neuromuscular disorders, we identified pathogenic mutations in 45.2% of the patients, and efficiently and quickly arrived at definitive diagnoses in these cases. Thus, NGS can be considered to demonstrate considerable utility in allowing an exhaustive analysis of neuromuscular disorders.

CMD disease types have been categorized according to evidence of unique protein deficiencies identified based on characteristic clinical symptoms and immunostaining results. However, in recent years, the identification of causative genes through a CMD diagnostic algorithm has emerged as the gold standard.28 We obtained an MD diagnosis in 12 of the 20 patients (60.0%) with clinically suspected MD, and in 5 of these patients, muscle biopsy was not performed. Seven of the 17 patients (41.1%) in whom CM was suspected were diagnosed, and a definitive genetic diagnosis was obtained in 1 patient (CM-2) without performing a muscle biopsy. Furthermore, in CM-5, a lack of specific muscle biopsy findings precluded diagnosis, but a definitive genetic diagnosis was obtained based on the reported pathogenic mutations identified using NGS. Thus, if NGS is applied to the identification of pathogenic mutations, invasive muscle biopsies may be avoided in certain cases.

FCMD is the result of mutations in FKTN, and it is the most frequent form of CMD in Japan: 98% of FCMD patients in Japan carry homozygous or heterozygous founder mutations, and the cases that involve heterozygous founder mutations tend to be more severe than the homozygous cases.29, 30, 31 We identified point mutations in the FKTN gene in four patients. In MD-1, we identified an intronic mutation that is specific to the Korean population.32 In these four patients, severe hypotonia had been present since birth, and the clinical courses were more severe than in typical cases. Thus, if FCMD is suspected but is more severe than what is typically observed, then point mutations might be efficiently detected using NGS. Interestingly, in MD-12, we diagnosed concurrent FCMD and BMD due to mutations in FKTN and DMD, and, furthermore, the FCMD did not involve a founder mutation. Previously, Kondo et al.33 reported a case of concurrent FCMD and Leigh syndrome. We must be aware of possible occurrence of cases in which pathogenic mutations in multiple genes have been identified together with overlapping neuromuscular disorders.

DMD, the causative gene for dystrophinopathy, is extremely large (genome size, 2200 kb), and approximately 30% of DMD contains microdeletions, insertions and point mutations.34 When NGS was applied in this study, point mutations were efficiently identified in two patients (MD-9 and 10). Currently, considerable effort is being devoted to developing treatments such as exon-skipping therapy, read-through therapy and other forms of DMD molecular therapy.35, 36, 37 Identification of genetic mutations in DMD would be extremely beneficial not only for definitive diagnosis, female-carrier diagnosis and prenatal diagnosis, but also when considering indications for the mutation-specific therapies that are currently under development.

Especially, phenotype–genotype correlations of merosinopathy (LAMA2) and Ullrich congenital muscular dystrophy (COL6A1, COL6A2 and COL6A3) have been clear.38, 39, 40

LGMD is a disease group that shows a high degree of heterogeneity. To date, 31 genes (autosomal dominant: 8 genes; autosomal recessive: 23 genes) have been identified as causative genes of LGMD.5 Furthermore, LGMD differs in its complications and degree of progression depending on the causative gene. Thus, we were able to provide detailed and accurate information about the possible complications and future disease progression for each diagnosed patient.

CM is a disease group that shows marked heterogeneity, and many of the causative genes contain numerous exons (e.g., NEB, 183 exons, and RYR1, 65 exons). Herein, we efficiently and quickly made definitive diagnoses by employing NGS. Specifically, in CM-1, NM was attributable to a mutation in NEB and nemaline bodies were detected in muscle biopsy specimens, but the patient’s clinical phenotype was extremely mild, with no difficulties in daily life being reported. The patient was even able to ride a bicycle. Thus, even among previous descriptions of NM, this case can be considered to present a particularly mild clinical picture. In CM-3, a compound heterozygous RYR1 mutation was identified as a cause of severe NM, and, furthermore, RYR1 was identified as a new candidate gene for NM with external ophthalmoplegia and fiber-type disproportion. We reported this case previously.41 RYR1 mutations carry a risk of malignant hyperthermia, and we thus could provide genetic counseling to the patient and the parents carrying the heterogeneous mutant alleles to be mindful of malignant hyperthermia when undergoing general anesthesia.

CMS is phenotypically very similar to CM, and a definitive diagnosis is thus considered unobtainable in many of the patients with this disease. Herein, in CM-7, CMS was diagnosed when a fatal akinesia deformation sequence was suspected, which necessitated the initiation of treatment with anticholinesterase agents. Subsequently, improvements were observed, including the disappearance of ptosis and maintenance of a seated posture. Effective therapies for CMS exist in the form of treatment with anticholinesterase agents, acetylcholine receptor antagonists and 3,4-diaminopyridine.6, 42 Symptoms generally improve with treatment, and early diagnosis of the disease would thus be extremely valuable.

No causative genes were identified in any of the five patients in the SMA group. This is because the sequence of SMN1, which is a causative gene of SMA, is almost identical to that of its adjacent gene, SMN2, with the difference between the genes being only five nucleotides.43 Thus, designing a primer pair exclusively for SMN1 is extremely challenging; in our panel, the coverage rate of SMN1 was 2.44%, and most of this gene could not be analyzed. Moreover, our panel included only three genes related to SMA. We speculate that additional genes related to SMA exist and remain to be identified. Several SMA patients harbor an exon 7 deletion or gene conversion in SMN1. These can be detected when MLPA is used for SMN1 exon 7, but 2–5% of the cases are compound heterozygous for point mutations44, 45 and these are not detectable by MLPA performed for SMA. Recently, Kubo et al.46 successfully isolated SMN1 and SMN2. Thus, if no SMN1 exon 7 deletion is detected but SMA is strongly suspected based on symptoms, use of the long-range PCR developed by Kubo et al. or exome sequencing is recommended.

Conventional Sanger sequencing would be highly time-consuming, expensive and labor-intensive, were it to be used for analyzing each conceivable causative gene one-by-one. However, NGS allows multiple genes to be analyzed concurrently, the amount of data obtained and the number of detected single-nucleotide variants are both large. However, patient information such as clinical and histopathological features remains extremely helpful. In addition to using bioinformatics to effectively identify causative genes, it is crucial to perform various types of filtering on large lists of candidate variants. In this study, we used multiple annotation databases to efficiently pinpoint the causative genes.

In NGS analysis, 2–4 × 105 single-nucleotide variants are reportedly detected per person in exomes, 4–9 in disease-related variant of uncertain significance, and 300–600 in variant of uncertain significance unrelated to disease.47 The bioinformatics load is extremely high, and the possibility of detecting incidental findings exists.48 The advantages of target resequencing are that, as compared with exome sequencing, it is inexpensive, the bioinformatics burden is low and there is no risk of incidental findings. Conversely, NGS cannot detect long deletions and insertions, and detects repeat sequences and copy number variations only weakly. As noted earlier in this section, 98% of Japanese patients with FCMD were found to carry the founder mutation, the size of which is about 3 kb. In the case of DMD, approximately two-thirds of the mutations are long deletions or insertions overlapping with multiple exons. Moreover, most SMA patients harbor an SMN1 exon 7 deletion, but the sequences of SMN1 and SMN2 are nearly the same, which makes primer design challenging. NGS cannot be used for detecting these mutations, and MLPA is considerably more effective than NGS in such cases.

Another concern is that several more undiscovered causative genes are assumed to exist than have thus far been identified. In target resequencing, only known genes are covered, but untranslated region and regions other than the coding DNA sequence, which involve non-coding RNAs, are occasionally the disease sources.49 Thus, the diagnostic capabilities of NGS are limited.

Herein, we combined conventional genetic testing and NGS and obtained a definitive diagnosis through rapid genetic testing for 19 of our 42 patients (45.2%) with undiagnosed neuromuscular disorders, in whom disease onsets had been during the neonatal and childhood periods. These results are almost identical to the 48.8% diagnosis rate obtained previously from target resequencing of 579 genes related to early-onset neuromuscular diseases.50 Although fewer target genes were examined in our study than in the study by Chae et al., we were able to obtain a similarly high diagnosis rate because we carefully examined each patient’s clinical symptoms, the onset period and the clinical course, and investigated the neurological findings. Subsequently, we classified the cases into three categories based on their clinical diagnoses. Recently, projects have been initiated in several countries aimed at clarifying undiagnosed genetic diseases (nearly 50% of these patients have neurological disorders) through exome sequencing, but the diagnosis rate is only approximately 25%.47, 51, 52 We suggest that as compared to simply analyzing all such cases by employing exome sequencing, a more useful approach would be to first carefully evaluate patient phenotypes, and then perform target resequencing when a neuromuscular disorder is suspected; subsequently, if pathogenic mutations are not detected through target resequencing, the next step could be exome sequencing.

Accurate molecular genetic diagnosis not only makes it possible to predict the progression of a disease, obtain information on possible future complications and implement appropriate health management, but also aids in providing appropriate genetic counseling, determining the odds of subsequent children being born with the same mutations and providing accurate prenatal diagnosis. Furthermore, because effective therapies exist for diseases such as CMS, efficient identification of causative genes, especially in the early stage, would be extremely useful. The cost of performing NGS is decreasing annually, raising the possibility of researchers being able to use NGS for clinical application in the near future.