Diagnostic yield of exome sequencing in myopathies: Experience of a Slovenian tertiary centre

Background Our aim was to present the experience of systematic, routine use of next generation sequencing (NGS) in clinical diagnostics of myopathies. Methods Exome sequencing was performed on patients with high risk for inherited myopathy, which were selected based on the history of the disease, family history, clinical presentation, and diagnostic workup. Exome target capture was performed, followed by sequencing on HiSeq 2500 or MiSeq platforms. Data analysis was performed using internally developed bioinformatic pipeline. Results The study comprised 86 patients, including 22 paediatric cases (26%). The largest group were patients referred with an unspecified myopathy (47%), due to non-specific or incomplete clinical and laboratory findings, followed by congenital myopathies (22%) and muscular dystrophies (22%), congenital myotonias (6%), and mitochondrial myopathies (3%). Altogether, a diagnostic yield was 52%; a high diagnostic rate was present in paediatric patients (64%), while in patients with unspecified myopathies the rate was 35%. We found 51 pathogenic/likely pathogenic variants in 23 genes and two pathogenic copy number variations. Conclusion Our results provide evidence that phenotype driven exome analysis diagnostic approach facilitates the diagnostic rate of complex, heterogeneous disorders, such as myopathies, particularly in paediatric patients and patients with unspecified myopathies.


Introduction
Myopathies represent a clinically and genetically heterogeneous group of neuromuscular disorders that commonly represent a diagnostic challenge [1,2]. They are characterized by a highly variable, sometimes non-specific clinical presentation and frequently overlapping phenotypes between different diseases. Mutations over 200 genes have been implicated in myopathies, including some of the largest genes in humans [2]. The genotype-phenotype correlation is often difficult to establish; mutations in different genes can present with similar phenotypes and different mutations in a single gene may result in different phenotypes.
Clinical implementation of massive parallel sequencing technology has emerged as a new approach to overcome the diagnostic complexity of myopathies and to reduce the necessity for invasive testing such as muscle biopsy. Correct and timely genetic diagnosis is critical for appropriate management of the patient, for prognostic information, and genetic counselling of the patient and his family as well as for carrier status determination and genetic diagnostics in prenatal settings. There is no widespread consensus for clinical indications for next generation sequencing (NGS), neither standardized NGS methodological approaches. Two approaches are mainly being implemented in clinical practice: exome or mendeliome sequencing and more frequently used targeted gene panel sequencing [3]. Few studies investigated the genetic diagnosis using NGS testing in unselected cohort of patients with myopathies. In recent years, there has been an increasing amount of studies conducted on subgroups of patients with a specific myopathy pattern, including patients with congenital muscular disorders, limb girdle muscular dystrophies or distal myopathies [4][5][6][7][8][9][10][11][12][13][14]. However, far less attention has been paid to the accuracy and efficiency of sequencing technology in unselected group of patients with various myopathy diagnoses [15][16][17][18][19].
The aim of the present study was to present the experience of systematic, routine use of novel NGS-based diagnostic approach addressing both clinical and genetic heterogeneity of myopathies. It is important to emphasize that we sought to assess the diagnostic yield and clinical utility of this approach in a group of patients who were consecutively referred to our centre with myopathy diagnosis, including patients with unspecified myopathy due to overlapping and non-specific clinical and laboratory findings.

Patients
We conducted a retrospective study of patients with a clinical diagnosis of myopathy who were referred for a genetic evaluation and testing to the Clinical Institute of Medical Genetics, University Medical Centre, Ljubljana, Slovenia, in a period from 2014-2018. Patients with specific myopathic phenotype associated with specific mutation mechanisms like facioscapulohumeral muscle dystrophy, myotonic dystrophies and Duchenne/Becker muscle dystrophy were diagnosed directly by gene-targeted assays and were therefore excluded from further evaluation in this study.
UM was defined when a clinical presentation was non-specific or incomplete, or when the clinical phenotype was suggestive for a myopathy, but EMG and muscle biopsy were not performed, were normal, or unspecific. Patients with UM were eligible for exome sequencing in a case of positive family history or paediatric age of symptom onset or when no other acquired cause of myopathy could be identified by a clinical diagnostic workup.
Patients were referred for genetic evaluation by a paediatric/adult neurologist. All patients received genetic counselling and were referred for exome sequencing by a clinical geneticist. Patients' data were obtained through medical records review, including age of symptom onset, family history, clinical evaluation of muscle symptoms, and associated phenotypic features (e.g. facial dysmorphism, cardiomyopathy, cognitive delay) as well as standardised clinical diagnostic workup findings (examination by neurologist specialised in neuromuscular diseases, serum creatine kinase level, electromyography (EMG), and muscle biopsy). Phenotypic features were described according to the Human Phenotype Ontology (HPO) nomenclature [23].

Ethics statement
In this study we analysed retrospectively the results of genetic testing previously performed as a part of a routine clinical diagnostics at our institution. No genetic testing was performed solely for the purpose of this study. All patients provided written informed consent. All patients were de-identified (regarding detected genetic variants data included in the manuscript). No other individual patient data was included.

Exome sequencing, data analysis and interpretation
Sequencing was performed using a standardized series of procedures, starting with an in-solution capture of exome sequences (TruSight One, TruSight Exome, and Nextera Coding Exome capture kits, Agilent SureSelect Human All Exon v2, Agilent SureSelect Human All Exon v5 capture kits). This was followed by sequencing on Illumina MiSeq or Illumina HiSeq 2500 platform. The exome data analysis was completed using internally developed pipeline based on the combined disease and phenotype gene target definition approach, as we previously described [24,25]. Basic analysis, including SNV and indel discovery and annotation, was performed according to Genome Analysis Toolkit Best Practices workflow [26][27][28][29]. The Genome Aggregation Database (gnomAD) was employed as source of variant frequencies in worldwide populations. We also used Slovenian genomic database at our Institute which is planned to be an open database in the future. The details of diagnostic exome sequencing, analysis and interpretation were previously described [24,25]. When possible, family exome sequencing or variants segregation in families through Sanger sequencing was performed. All variants were classified according to the guidelines of the American College of Medical Genetics (ACMG) modified according to Association for Clinical Genomic Science (ACGS) recommendations as pathogenic, likely pathogenic, variant of uncertain significance (VUS), likely benign, or benign [27,28]. Evidence support level was weighted using modifiers VSTR (very strong), STR (strong), MOD (moderate) or SUP (supporting), where applicable according to ACGS recommendations [28]. Variant interpretations were submitted to the ClinVar Database (https:// www.ncbi.nlm.nih.gov/clinvar/). Pathogenic and likely pathogenic variants were classified as disease causing variants.

Gene selection
The search through OMIM (http://www.ncbi.nlm.nih.gov/omim) and Pubmed (http://www. ncbi.nlm.nih.gov/pubmed) databases enabled the creation of a 250 myopathy associated gene panel (S1 Table). The panel was supplemented with phenotype-based gene panel, which was generated using a Web tool (http://kimg.eu/generator). This tool enabled tracking for genes associated with clinical signs and symptoms using HPO nomenclature [24,25,30]. This database-based gene panel was applied for the filtration of sequence variants. In case of negative results, untargeted interpretation of all exome variants was performed by using internally developed variant analysis and interpretation pipeline.

Patients
The study included 86 probands (38 males and 48 females). There were 22 paediatric and 64 adult patients. The mean age of the paediatric patients was 8.36 years (range from 36 months to 16 years) and the mean age of symptom onset was 16 months (range from birth to 7 years). In adult patients, the mean age was 41.22 (range from 18 to 72 years). The mean age of symptom onset in adult patients was 21.40 years (range from birth to 65 years). Nineteen cases (22%) had records of family history of the disease.
EMG and muscle biopsy findings in patients with UM are presented in Table 1.

Diagnosis by exome sequencing and the compatibility of genetic variant interpretation with the referral diagnosis
The diagnostic yield in our cohort of patients is presented in Table 2. Overall, genetic diagnosis was achieved in 52% (45/86) of cases. Among patients referred with a specific clinical myopathy pattern, pathogenic variants were identified in 67% (31/46) of cases. Clinical diagnosis was confirmed in 90% (28/31) of patients and in 10% of cases (3/31) genetic diagnosis revealed that myopathic symptoms were a part of syndrome or other rare neuromuscular disease (S2 Table). The diagnostic yield among patients referred with UM was 35% (14/40) ( Table 2). Overall, among 38 cases with a genetic diagnosis of myopathy, 32 (84%) had myopathic EMG and/or muscle biopsy, 1 had normal EMG (muscle biopsy was not performed), and in 5 patients neither EMG nor muscle biopsy were performed. In 7 patients with other rare pathogenic variants, 4 (57%) had myopathic EMG and/or muscle biopsy, 1 patient had normal EMG (muscle biopsy not performed), while in 2 patients neither EMG nor muscle biopsy were performed. Out of 20 cases with likely benign, benign or no variants detected, 17 (85%) had myopathic EMG and/or muscle biopsy, 2 had normal EMG (muscle biopsy not performed) and in 1 case neither EMG nor muscle biopsy were performed. There were also 21 cases with VUS variants, whereas 17 had myopathic EMG and/or muscle biopsy and 4 had normal findings or neither of the examinations were performed.

Diagnostic rate in paediatric and adult cases
Among paediatric patients the two groups, DMD and CMyop, comprised 23% (5/22) of cases. Two patients were referred with LGMD. Almost a half (10/22, 45%) of cases were referred with UM. Overall, NGS enabled diagnosis in 64% (14/22) of paediatric cases (Fig 1); myopathy Table 2. Diagnostic yield in patients grouped according to referral diagnosis.

Classification and interpretation of identified genetic variants
Identified pathogenic /likely pathogenic variants for a myopathy are presented in Table 3. In total, 50 pathogenic/likely pathogenic variants were found in 22 nuclear genes and one variant was detected in the mitochondrial genome. Eight (15%) were novel. Two pathogenic copy number variations (CNV) were detected (duplication 22q11.2 and 17q12).
Out of 42 patients with pathogenic genetic variants in nuclear genes, 11 carried heterozygous variants resulting in dominant type of a myopathy. Among these, 3 variants were of de novo origin, whereas the remaining pedigrees were consistent with dominant inheritance. A homozygous or compound heterozygous pathogenic variant was identified in 23 probands resulting in a recessive type of a myopathy. A hemizygous pathogenic variant was detected in 8 probands resulting in X-linked type of a myopathy.
Pathogenic CNVs and pathogenic/likely pathogenic variants in genes, including PLOD1, GCH1, CHAT, RAPSN, and SYT2 enabled the diagnosis of other rare diseases with myopathic symptoms (S2 Table).

Discussion
Using a phenotype driven exome analysis diagnostic approach we demonstrated a high diagnostic yield of 52% in a cohort of patients who were consecutively referred to our centre with clinical signs and symptoms consistent with a myopathy. Our cohort of patients was not selected according to specific clinical pattern of myopathy and cases with UM were also included. To the best of our knowledge, the evaluation of clinical utility of NGS-based approach in patients with UM has not been conducted so far. A growing body of literature has been investigating the benefits and the challenges of NGS based technology implementation in the standard clinical practice of patients with inherited myopathies [5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22]. So far, however, fewer studies have been conducted on patients with unselected primary muscular disease, whereas reported diagnostic yield ranged between 16% and 36% [15][16][17][18][19]. Park et al. included only patients, who presented with a specific clinical myopathy pattern [14]. Using targeted gene panel NGS diagnostic approach they obtained the diagnostic yield of 36%. Bugiardini et al. applied focused exome sequencing to investigate complex adult myopathy patients, which were categorized based on the age of symptom onset and predominant pattern of weakness [16]. Pathogenic or likely pathogenic variants were identified in 32% of cases. Haskel et al. performed exome sequencing in 93 undiagnosed paediatric and adult patients with various neuromuscular diseases, including 31 patients with myopathy phenotype, 21 patients with neuropathy phenotype and 41 complex patients with neuropathy, myopathy, and additional phenotype [17]. They used targeted gene panel (myopathy or neuropathy) and a broad neuromuscular gene panel. Among patients with myopathy phenotype diagnostic yield was 16.1%, regardless of the used gene panel. In a group of patients with a complex phenotype, they illustrated higher diagnostic rate (9.8%) using a broader diagnostic gene panel compared with using neuropathy (4.9%) or myopathy (0%) diagnostic gene panel alone. Punetha et al. carried out targeted sequencing with a myopathy candidate gene panel in a cohort of 94 undiagnosed muscle disease patients and obtained the diagnostic yield of 35% [18]. In recent study, Thuriot et al. used gene panel approach in a large cohort of patients with suspected muscle disorders and reported an overall diagnostic yield of 15.1% (22.4% in children and 13.7% in adults) [19]. Most of other reported NGS based studies focused on one or more types of myopathy, including congenital muscular disorder [5,8,11,13], LGMD [5,6,8,10,12,14], or DM [7,12]. Using targeted NGS approach [5][6][7][8] or WES [10][11][12][13][14] the reported diagnostic rate ranged from 12% to 52%, except for the study of Chinese patients presenting with muscular dystrophy and congenital myopathy, which demonstrated the diagnostic yield of 65% [9]. Our results showed a high diagnostic rate among cases with specific clinical pattern of myopathy (67%), whereas CMyop and LGMD represented the largest groups of patients (Table 2).
However, it was particularly interesting to assess the diagnostic yield and clinical utility of our exome sequencing approach in patients with UM, since our cohort encompassed 47% of such cases. Considering that myopathies are characterized by highly variable, non-specific clinical presentation and frequently overlapping phenotypes between different diseases, it is expected that some of the patients could not be designated with certainty into any of the myopathy subgroup. We were able to detect a pathogenic variant in 35% of patients with UM, which is in the range of published diagnostic rates obtained among patients with specific clinical pattern of muscular disease.
In addition, 26% of cases in our study were paediatric patients. A separate analysis of paediatric cases with myopathies revealed a high diagnostic yield of 64%, whereas the average time to diagnosis was 7.05 years. Our diagnostic rate coincides well with the results published by Schofield et al, which was conducted on 56 paediatric patients, referred with the diagnosis of congenital muscular dystrophy or nemaline myopathy [21]. Using the traditional investigation based on muscle biopsy and protein-based studies of muscle biopsy specimens, followed by candidate gene sequencing, they established the diagnosis in 46% of referred patients. They increased their diagnostic efficacy using neuromuscular gene panel to 75% and to 79% using WES. In recent study, WES analysis was conducted on 50 patients with undiagnosed paediatric-onset neuromuscular diseases resulting with the overall diagnostic yield of 26% [22]. Diagnostic rate among patients with hereditary congenital myopathy was 17% and among patients with hereditary muscular dystrophy subgroup was 45%.
The diagnostic yield among our adult patients was also high (48%) and the average time to diagnose was 19.63 years.
Our approach also included the assessment of mitochondrial and CNV variations, which enabled the diagnosis of microduplication syndrome 17q12 in paediatric patient, referred with UM. Clinical variability of 17q12 duplication syndrome is broad and myopathy was already described in those patients [31]. Similarly, we revealed a likely causative 22q11.2 microduplication in a patient with predominant presentation of hypotonia and motor delay. As known microduplication 22q11.2 may manifest predominantly with severe muscle hypotonia (OMIM#608363). In addition, in one patient who presented with unspecific muscle symptoms, genetic diagnosis revealed Ehlers-Danlos type 4. Interestingly, our results also demonstrated that diagnostic exome sequencing may affect medical management of the patients. Genetic testing revealed Dopa-Responsive Dystonia (DRD) in a case referred as UM, as well as congenital myasthenic syndromes in 3 cases referred with UM or CMyop. Those patients started with symptomatic therapy after genetic diagnosis. These findings further confirm the added value of our exome sequencing approach in diagnostics of cases with unspecific myopathy phenotypes and further decipherment whether myopathic symptoms are a part of a syndrome or separate clinical feature.
Despite the accumulating evidence that NGS technology has high clinical utility and huge potential to become the first approach in clinical laboratories, there are no universally or widely accepted guidelines for genetic testing, regarding the patient selection or diagnostic approach in heterogeneous disorders. This is also because the selection of appropriate target genes is challenging and carries a risk of missing causative variation due to their restrictiveness, whereas time and cost efficiency of such approach is arguable [24]. In view of limitations of targeted testing, we have applied clinical exome sequencing supplemented with panel of genes associated with clinical signs/symptoms using HPO nomenclature. This enabled the assessment of patients in regard to clinical diagnostic hypothesis as well as presenting signs/ symptoms. This approach has been particularly useful in patients with complex, atypical clinical presentation that does not direct the clinical diagnosis toward a specific disease, whereas some of those patients would not even be considered for genetic testing due to unfulfilled clinical criteria for a specific myopathy. We suggest that patients with specific clinical myopathy pattern as well as patients with UM and suspected genetic aetiology should be referred for genetic testing. Genetic diagnosis is of great importance for both, patients and their families, as it can have a significant impact on patient management (particularly in cases of treatable disorders), prognosis, accurate genetic counselling, prenatal diagnostics and avoiding additional diagnostic testing. Thus, the need for widely accepted recommendations, which will improve the diagnostic implementation of exome sequencing in clinical practice of heterogeneous disorders, still remains.
The strength of our study is the patient cohort, which also included cases with UM and suspected genetic aetiology. Secondly, our approach is independent of specificity and accuracy of the clinical diagnostic hypothesis. The study's limitation is the number of patients with a particular muscular disorder, which hinders the evaluation of our exome sequencing diagnostic utility of rare disorders.
In conclusion, our results provide evidence that phenotype driven exome analysis diagnostic approach improves the diagnostic rate of complex, heterogeneous disorders, such as myopathies, particularly in patients with an unspecified myopathy.
Supporting information S1 Table. 250 inherited muscular disorders associated gene panel.