Of 237 families referred to the Institute for Neuroscience and Muscle Research, 83 had LGMDs that were diagnosed using standard techniques of immunohistochemistry (IHC), Western blotting (WB), candidate gene sequencing, and screening for type 2 myotonic dystrophy (DM2) and type 1 facioscapulohumeral muscular dystrophy (FSHD1). Based on the clinical indication, IHC was performed for merosin, α-dystroglycan, α-sarcoglycan, γ-sarcoglycan, β-sarcoglycan, δ-sarcoglycan, dysferlin, caveolin-3, desmin, myotilin, and type VI collagen, and/or WB was performed for dystrophin, α-dystroglycan, lamin A/C, emerin, caveolin-3, dysferlin, and calpain-3. Candidate genetic testing was then performed for LGMD1A-related MYOT mutations, LGMD1B-related LMNA mutations, LGMD1C-related CAV3 mutations, LGMD1E-related DES mutations, LGMD2A-related CAPN3 mutations, LGMD2B-related DYSF mutations, LGMD2C-related γ-sarcoglycan mutations, LGMD2D-related α-sarcoglycan mutations, LGMD2E-related β-sarcoglycan mutations, LGMD2F-related δ-sarcoglycan mutations, LGMD2I-related FKRP mutations, LGMD2K-related POMT1 mutations, LGMD2M-related FKTN mutations, LGMD2N-related POMT2 mutations, FSHD1, and DM2. Consent and DNA samples were obtained from 60 of the families with undiagnosed LGMDs for WES. Diagnosis was achieved for 27 families, including variants in genes not typically considered LGMD-related genes. Neurogenetic subexomic supercapture (NSES) confirmed all the sequence variants identified by WES and did not reveal any new pathogenic variants. There were 2 families for whom pathogenicity could not be confirmed in known myopathy-related genes.
eTable 1. List of known myopathy genes
eTable 2. Results of 27 families with likely pathogenic changes in candidate genes
eTable 3. Variants of uncertain significance in two families
Customize your JAMA Network experience by selecting one or more topics from the list below.
Ghaoui R, Cooper ST, Lek M, et al. Use of Whole-Exome Sequencing for Diagnosis of Limb-Girdle Muscular DystrophyOutcomes and Lessons Learned. JAMA Neurol. 2015;72(12):1424–1432. doi:10.1001/jamaneurol.2015.2274
Copyright 2015 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
To our knowledge, the efficacy of transferring next-generation sequencing from a research setting to neuromuscular clinics has never been evaluated.
To translate whole-exome sequencing (WES) to clinical practice for the genetic diagnosis of a large cohort of patients with limb-girdle muscular dystrophy (LGMD) for whom protein-based analyses and targeted Sanger sequencing failed to identify the genetic cause of their disorder.
Design, Setting, and Participants
We performed WES on 60 families with LGMDs (100 exomes). Data analysis was performed between January 6 and December 19, 2014, using the xBrowse bioinformatics interface (Broad Institute). Patients with LGMD were ascertained retrospectively through the Institute for Neuroscience and Muscle Research Biospecimen Bank between 2006 and 2014. Enrolled patients had been extensively investigated via protein studies and candidate gene sequencing and remained undiagnosed. Patients presented with more than 2 years of muscle weakness and with dystrophic or myopathic changes present in muscle biopsy specimens.
Main Outcomes and Measures
The diagnostic rate of LGMD in Australia and the relative frequencies of the different LGMD subtypes. Our central goals were to improve the genetic diagnosis of LGMD, investigate whether the WES platform provides adequate coverage of known LGMD-related genes, and identify new LGMD-related genes.
With WES, we identified likely pathogenic mutations in known myopathy genes for 27 of 60 families. Twelve families had mutations in known LGMD-related genes. However, 15 families had variants in disease-related genes not typically associated with LGMD, highlighting the clinical overlap between LGMD and other myopathies. Common causes of phenotypic overlap were due to mutations in congenital muscular dystrophy–related genes (4 families) and collagen myopathy–related genes (4 families). Less common myopathies included metabolic myopathy (2 families), congenital myasthenic syndrome (DOK7), congenital myopathy (ACTA1), tubular aggregate myopathy (STIM1), myofibrillar myopathy (FLNC), and mutation of CHD7, usually associated with the CHARGE syndrome. Inclusion of family members increased the diagnostic efficacy of WES, with a diagnostic rate of 60% for “trios” (an affected proband with both parents) vs 40% for single probands. A follow-up screening of patients whose conditions were undiagnosed on a targeted neuromuscular disease–related gene panel did not improve our diagnostic yield.
Conclusions and Relevance
With WES, we achieved a diagnostic success rate of 45.0% in our difficult-to-diagnose cohort of patients with LGMD. We expand the clinical phenotypes associated with known myopathy genes, and we stress the importance of accurate clinical examination and histopathological results for interpretation of WES, with many diagnoses requiring follow-up review and ancillary investigations of biopsy specimens or serum samples.
The limb-girdle muscular dystrophies (LGMDs) are a heterogeneous group of disorders characterized by progressive weakness that primarily affects the proximal muscles of the upper and lower extremities.1-5 To date, 27 different genetic forms of LGMD (types and subtypes) have been described and are grouped according to inheritance pattern; type 1 LGMD refers to dominantly inherited forms, whereas type 2 LGMD refers to recessively inherited forms.4,6 Comprehensive guidelines to assist clinicians in managing and diagnosing LGMDs7 have been recently published because predicting the genetic form of LGMD based on clinical examination or histological appearances alone is difficult, and many patients’ conditions remain undiagnosed. In Australia, the rate of genetic diagnosis of LGMD remains low at approximately 24%, despite traditional approaches of histopathological findings to triage genetic screening of known LGMD-related genes.8 Families with undiagnosed LGMD face uncertainty about recurrence risks for relatives, the health surveillance required, and the long-term prognosis.
With the advent of next-generation sequencing (NGS), patients can now be screened using custom DNA capture, via neuromuscular disease–related gene panels9,10 (Centre for Medical Research, University of Western Australia) or by whole-exome sequencing (WES).11,12 Our study explores the use of WES for the diagnosis of a cohort of 60 Australian patients with LGMD for whom protein-based analyses and targeted Sanger sequencing had failed to identify the genetic cause of their disorder. Although WES results require further validation for clinical diagnosis, this approach held the capacity to screen all known LGMD-related genes, as well as provide an additional opportunity to identify new disease-related genes. Our central goals were to improve the diagnostic rates of LGMD in Australia, to investigate whether the WES platform provides adequate coverage of the known LGMD-related genes, and to identify new LGMD-related genes. A defined genetic diagnosis is vital to guide patient clinical care and informed genetic counseling and is often required for inclusion of patients in clinical trials. We demonstrate that NGS technologies can be readily transferred from a research setting to neuromuscular clinics, and we identify valuable lessons for others intending to incorporate WES into their diagnostic pipeline.
The present study was approved by the Human Research Ethics Committee of the Children’s Hospital at Westmead (Institute for Neuroscience and Muscle Research Biospecimen Bank approval CHW/10/45), and all participants provided written informed consent.
We ascertained 237 patients with LGMD retrospectively through the Institute for Neuroscience and Muscle Research Biospecimen Bank (between 2006 and 2014), which receives referrals from all states in Australia and New Zealand but consists largely of samples from New South Wales encompassing a diverse ethnic population. Therefore, our cohort is considered a representative sample for these rare disorders within an Australasian population. All patients (except families 7 and 8, which were referred through other centers) had normal dystrophin immunohistochemistry results and, based on clinical indications, were subject to immunohistochemistry for merosin, α-dystroglycan, α-sarcoglycan, γ-sarcoglycan, β-sarcoglycan, δ-sarcoglycan, dysferlin, caveolin-3, desmin, myotilin, and type VI collagen and/or Western blotting for dystrophin, α-dystroglycan, lamin A/C, emerin, caveolin-3, dysferlin, and calpain-3. Based on clinical indication and results from protein-based screening, candidate genetic testing was performed for LGMD1A-related MYOT mutations, LGMD1B-related LMNA mutations, LGMD1C-related CAV3 mutations, LGMD1E-related DES mutations, LGMD2A-related CAPN3 mutations, LGMD2B-related DYSF mutations, LGMD2C-related γ-sarcoglycan mutations, LGMD2D-related α-sarcoglycan mutations, LGMD2E-related β-sarcoglycan mutations, LGMD2F-related δ-sarcoglycan mutations, LGMD2I-related FKRP mutations, LGMD2K-related POMT1 mutations, LGMD2M-related FKTN mutations, LGMD2N-related POMT2 mutations,13,14 type 1 facioscapulohumeral muscular dystrophy, and type 2 myotonic dystrophy15 (Table 1).
We retrospectively reviewed the results of 237 patients with LGMD enrolled through the Institute for Neuroscience and Muscle Research Biospecimen Bank between 2006 and 2014 to determine the diagnostic rate and frequencies of the different subtypes of LGMDs in our Australian population. We included the previously characterized cohort by Lo et al8 from 2008. Because LGMDs have historically been considered a distinct entity from the much more common X-linked dystrophinopathies, which include Duchenne muscular dystrophy and Becker muscular dystrophy,1,4 this group of LGMDs was not included in our analysis. Based on protein-based testing, followed by candidate genetic testing as already described, a genetic diagnosis was achieved for 83 of 237 families (35%) (Figure). The most common cause of dominant LGMD in Australia was due to mutations in LMNA (related to subtype LGMD1B in 16 of 237 families [6.8%]), whereas mutations in CAPN3 (related to subtype LGMD2A in 14 of 237 families [5.9%]) and DYSF (related to subtype LGMD2B in 10 of 237 families [4.2%]) were the most common recessive LGMDs. Type 1 facioscapulohumeral muscular dystrophy (in 10 of 237 families [4.2%]) and type 2 myotonic dystrophy (in 2 of 237 families [0.8%]) accounted for myopathies with an overlapping phenotype referred to as LGMD. A total of 154 of 237 families (65.0%) with a LGMD phenotype remained without a genetic diagnosis using protein analysis and candidate genetic sequencing (Table 1).
This project encompassed funding for 100 exomes. Of the 154 families with undiagnosed LGMDs, 62 that satisfied the inclusion criteria were contacted. Sixty patients and 40 family members agreed to participate in the project, and 2 of the 62 families contacted were not willing to participate (Figure).
Patients who presented after 2 years of age with muscle weakness and when dystrophic or myopathic changes were present in muscle biopsy specimens were included in our study.1,16 All male patients who were included had negative dystrophin multiplex ligation–dependent probe amplification test results and/or normal results from dystrophin staining and Western blotting.
Patients whose age at onset was 2 years or younger and who had a confirmed diagnosis of Duchenne muscular dystrophy, Becker muscular dystrophy, type 1 facioscapulohumeral muscular dystrophy, or a type 1 or type 2 myotonic dystrophy triplet repeat disorder were excluded from our study.15
Whole-exome sequencing was performed in 2014 (at the Broad Institute of the Massachusetts Institute of Technology and Harvard University) on trios of affected patients and unaffected parents (12 families). For some patients, DNA samples from only 1 parent were available (3 families), or parental DNA samples were not available (45 patients). In families for which DNA samples and consent were available, siblings and grandparents were also included. Exome libraries were captured using hybridization with the Agilent SureSelect V2, and WES performed on an Illumina HiSeq 2000 as previously described.17 FASTQ sequencing files were processed using Picard and jointly called with more than 2000 control samples using the GATK 3.0 Haplotype Caller. The data received had an average read depth of more than 80 and 20-fold coverage in more than 90% of the targeted regions.
Filtering of variants was performed using xBrowse (https://atgu.mgh.harvard.edu/xbrowse), initially screening known myopathy-related genes (eTable 1 in the Supplement). First, sequence variants were filtered based on population frequency in the Exome Variant Server (http://evs.gs.washington.edu/EVS/ [March 2013]), 1000 Genomes,18 and internal databases, with variants showing more than 1% frequency in any population discarded. Second, we considered only variants with a predicted functional impact on coding regions (predicted missense, nonsense, and essential splice site single-nucleotide polymorphisms and insertions or deletions). When a candidate neuromuscular disease–related gene was not identified, we searched for variants in all genes based on inheritance pattern. For trios with a dominant inheritance pattern, we searched for de novo dominant variants in the proband.
The Polymorphism Phenotyping version 2 pathogenicity prediction program was used to perform in silico analysis of missense variants (http://genetics.bwh.harvard.edu/pph2/). Variants were also correlated with patient phenotype, results of clinical investigations, and/or, if previously reported, as pathogenic in the Leiden Muscular Dystrophy pages (http://www.dmd.nl/). Sanger sequencing was used to confirm all variants identified by WES and to assess familial segregation.
Patients who remained without a diagnosis for their condition by use of WES were tested further using Neurogenetic Subexomic Supercapture (NSES; Centre for Medical Research, University of Western Australia). This involves NGS (Ion Proton; Life Technologies) of 336 genes, including 254 disease-related genes listed in the Gene Table of Neuromuscular Disorders (http://www.musclegenetable.fr [December 2012]), a number of unpublished and candidate neuromuscular disease–related genes, and 59 cardiomyopathy disease–related genes. Alignment and variant calling were performed using Torrent Suite Software (Life Technologies). Variants were annotated and filtered in Bench Lab NGS (Cartagenia).
Sixty families that did not receive a genetic diagnosis following our routine diagnostic testing (see Methods section) were asked to consent to WES (Figure). Based on family pedigrees, 5 families had a likely autosomal dominant pattern of inheritance, 54 families were typical of an autosomal recessive inheritance pattern, and 1 family showed X-linked inheritance. The age range of disease onset was between 2 and 63 years, with a mean age of 20.9 years. There were 32 male probands and 28 female probands. By use of WES, pathogenic mutations were identified in 27 of 60 families (45.0%). Clinical characteristics, age at onset, maximum creatine kinase level, and biopsy findings of the families that received a diagnosis are shown in eTables 2 and 3 in the Supplement. In families for which 1 or more parental samples were included for WES, we identified pathogenic mutations in 9 of 15 families (60.0%) compared with 18 of 45 of families (40.0%) for which only DNA samples from the proband were available.
Of the 60 families that did not receive a genetic diagnosis, 12 possessed mutations in a known LGMD-related gene. The LGMD subtypes and the genes identified are detailed in the Figure and Table 2, with clinical information summarized in eTable 2 in the Supplement. There were 2 families with autosomal dominant LMNA mutations and 10 families with mutations in recessive LGMD-related genes, including 1 family with variants within the newly described TOR1AIP1 gene19 associated with contractures, rigid spine, dilated cardiomyopathy, and restrictive lung disease.20 In family 12, the proband developed LGMD-like weakness and severe dilated cardiomyopathy at 12 years of age, requiring a heart transplant at 16 years of age. Providing a genetic diagnosis for patients with LMNA and TOR1AIP1 mutations is crucial in terms of ensuring targeted, potentially lifesaving cardiac surveillance.
Importantly, 14 of the 60 families had mutations in genes typically associated with other forms of inherited myopathy, highlighting the diagnostic challenge with overlapping clinical presentation among patients with an LGMD phenotype (Figure, Table 3, and eTable 2 in the Supplement). Mutations in known congenital muscular dystrophy (CMD)–related and collagen (Bethlem) myopathy–related genes were the most common overlapping myopathies identified in our cohort of families with LGMD. Two families had mutations in the CMD α-dystroglycanopathy–related genes GTDC2 (family 13) and GMPPB (family 14). GTDC2 (glycosyltransferase-like domain-containing protein 2) is typically associated with severe congenital Walker-Warburg syndrome,21 and GMPPB has been associated with a range of phenotypes ranging from severe CMD to a mild LGMD.10,22 Families 13 and 14 displayed an atypical presentation for α-dystroglycanopathy in that cognitive impairment was the presenting feature, with muscle weakness not evident until adulthood. The proband in family 13 had adult-onset, mild proximal weakness and significant learning difficulties. Although the variants found in GTDC2 have not previously been reported, the reduced α-dystroglycan staining in biopsy specimens was supportive of the diagnosis (Table 3 and eTable 2 in the Supplement). Two families had mutations in LAMA2 (families 15 and 16), which is typically associated with CMD and white matter T2-weighted hyperintensity.23 Both families had later-onset weakness with no reported cognitive impairment. Subsequent cranial magnetic resonance imaging (MRI) scans showed subcortical white matter changes supportive of the diagnosis (Table 4).
Bethlem myopathy due to mutations in COLVI genes was also common in our cohort (families 17, 18, 19, and 20). A subsequent clinical review identified follicular hyperkeratosis and keloid scarring in 2 families, missed on initial examination, emphasizing the value of thorough clinical phenotyping for interpretation of WES results. Moreover, the results of muscle ultrasonography for patients in family 18 showed a characteristic “central cloud” pattern in both rectus femori.24,25 Collagen staining was also performed on all biopsy specimens available after the WES results confirmed reduced and abnormal staining (Table 4).
By use of WES, we also identified 2 families with a metabolic myopathy as a cause of the LGMD phenotype. It is atypical for fixed weakness to be present in patients with McArdle disease and CPT2 deficiency, although there are reported cases.26-29 In family 21, a homozygous sequence variant in PYGM was associated with the absence of phosphorylase staining, supporting the diagnosis of McArdle disease (Table 4). Family 22 had a homozygous variant in CPT2, with CPT2 deficiency subsequently confirmed after direct enzymology testing.
The less-common myopathies identified included those found in family 23 with DOK7 mutations. Dystrophic changes were reported in muscle biopsy specimens obtained from the patient at 10 years of age (hence, the LGMD diagnosis), although fatigability was a prominent feature and consistent with myasthenia. Family 24 had likely pathogenic changes in ACTA1 (Table 3); the late onset of weakness represents an atypical presentation for ACTA1, usually associated with a severe congenital myopathy,30 but the diagnosis was supported by the presence of nemaline rods in biopsy specimens. STIM1-related tubular aggregate myopathy was found in family 25, and the presence of thrombocytopenia in the proband supports the diagnosis of Stormorken syndrome.31 Lastly, myofibrillar myopathy due to an FLNC mutation was identified in family 26 and was previously reported as pathogenic in Leiden Muscular Dystrophy pages and consistent with biopsy findings.
Family 27 presented with a novel, mild phenotype in the CHARGE syndrome spectrum associated with a novel heterozygous variant in CHD7 that we think is very likely to be pathogenic.32 Classical CHARGE syndrome is associated with notable malformations, such as colobomas, microphthalmia, choanal atresia, cranial nerve palsies, deafness, and external ear abnormalities.32 However, in family 27, hypoplasia of trapezii demonstrated on MRI scans of muscle caused noticeable scapular winging and mild proximal upper limb weakness, considered most likely due to a mild muscular dystrophy. The proband had small ears but was not noted to be dysmorphic. A cranial MRI scan subsequently showed hypoplasia of the semicircular canals (Table 4), present in more than 95% of individuals with CHD7 mutations.33 The CHD7 mutation was confirmed to have arisen de novo in the proband.
For 2 families, variants in MSTN and ANO5 were identified by both WES and NSES but could not be confirmed as causal (eTable 3 in the Supplement). Family 28 had a homozygous MSTN substitution (c.209C>G, p.P70R) predicted to be deleterious by 5 in silico prediction algorithms, conserved from zebrafish to humans and not identified in healthy control DNA data sets (1000 Genomes, the Exome Variant Server, or the Exome Aggregation Consortium). We could not confirm segregation because the proband was adopted and DNA samples from the parents were unavailable. Muscle samples were also unavailable for myostatin antibody staining. The proband’s clinical phenotype did not correlate with previously reported phenotypes of patients with MSTN variants.34 Family 29 had 2 ANO5 variants, a missense change (c.240T>G, p.I80M) and an extended splice site change (c.1120-3C>T). The missense change p.I80M has not been reported in healthy control DNA data sets and was predicted by use of Polymorphism Phenotyping version 2 to be “probably damaging” (with a pathogenicity score of 0.996, where 1 is the highest score that can be achieved, and 0 is the lowest). However, 2 splice site prediction tools did not predict deleterious consequences of the second variant, and therefore it was considered not to be pathogenic. Muscle biopsy specimens were unavailable for RNA sequence analysis.
In 33 families for which we did not identify a pathogenic mutation in a known myopathy disease–related gene by use of WES, targeted supercapture of neuromuscular disease–related genes was performed for these patients. By use of NSES, we confirmed all sequence variants identified by WES but did not detect any new pathogenic variants.
We evaluated the utility of WES for genetic diagnosis with regard to 60 Australian families with LGMD. The patients included in our study represent a challenging group for diagnosis because they had previously been extensively investigated using traditional protein biochemistry and sequencing approaches8 and remained without a diagnosis for their conditions. Because many common genetic neuromuscular disorders that overlap with LGMDs are routinely missed when using NGS technologies owing to DNA expansions/deletions, we screened and excluded all families with type 2 myotonic dystrophy and type 1 facioscapulohumeral muscular dystrophy prior to inclusion in WES.
Among the 60 families with undiagnosed LGMDs, we achieved a diagnostic success rate of 45.0% (27 of 60 families), comparing favorably with our previous 35% diagnostic success rate using a triaged approach of IHC, Western blot analyses, and candidate gene sequencing—acknowledging that the latter was achieved cumulatively over a decade after extensive and laborious pursuit of several genes for some patients. The improved WES diagnostic rate with additional family members (60% diagnosis for trios vs 40% for single probands) was due to increased capacity to filter and stratify identified variants based on familial segregation with disease. Families that remained undiagnosed were subject to targeted capture using a neuromuscular panel, to cross-check the coverage of known neuromuscular genes by use of WES (all patients could not be screened by both methods owing to budget constraints). Indeed, with the exception of FKRP, which did not show good coverage with either WES or NSES, and required Sanger sequencing of the whole gene in 1 family to identify the second variant, WES achieved good coverage of known neuromuscular genes.
In the remaining undiagnosed families, disease-causing mutations may be located in noncoding regions of genes, such as regulatory or deep intronic regions not captured and sequenced by use of WES. Mutations in coding regions of known myopathy-related genes may have been missed by both methods of NGS because 5% to 10% of coding exons are poorly captured by a range of NGS methods.11,35 Other limitations may result from the presence of repetitive regions that obscure the specific copy to which the variant maps or from deletions that remove entire exons.12,35-37 In accordance with American Academy of Neurology/American Association of Neuromuscular & Electrodiagnostic Medicine guidelines,7 it is very important to exclude common genetic neuromuscular disorders due to DNA expansions/deletions (ie, type 1 facioscapulohumeral muscular dystrophy, spinal muscular atrophy, type 1 myotonic dystrophy, type 2 myotonic dystrophy, Duchenne muscular dystrophy, and Becker muscular dystrophy) because these are routinely missed when using NGS technologies. Type 2 facioscapulohumeral muscular dystrophy due to dominant mutations in SMCHD1 can be screened for by use of WES,38 although we found no likely disease-causing variants in our cohort. In our remaining undiagnosed families, RNA sequencing39 and/or whole-genome sequencing40 will be performed to identify additional candidate disease–causing variants.
Whole-exome sequencing provided an efficacious approach to establishing a genetic diagnosis for 45.0% of our difficult-to-diagnose patients. In most of these cases, review of clinical phenotype, follow-up investigations of biopsy specimens, serum enzyme assays, and/or MRI were required to support the genetic diagnosis identified by use of WES. Thus, we stress the continued reliance on accurate clinical assessment and established diagnostic investigations to augment and validate NGS findings. Our results emphasize the range of phenotypes associated with mutations in genes originally identified in patients with CMD and Bethlem myopathy, their overlap into the LGMD spectrum, and the increasingly phenotypic heterogeneity associated with neuromuscular disease–related genes as a result of WES studies. Thus, we recommend open consideration of all known myopathy-related genes when undertaking WES (or targeted screening of a neuromuscular disease–related gene panel), taking care to consider the possibility of a milder phenotype for a gene currently implicated in congenital-onset myopathy, as well as a potentially novel recessive presentation vs typical dominant inheritance, or vice versa.
Moreover, the costs for WES (approximately A$800 per exome) are comparable to research-based costs for biochemistry and sequencing of 1 large muscle-related gene (ie, dysferlin; A$850 for Western blot, complementary DNA isolation and sequencing, and genomic DNA confirmation and familial segregation). Transferring NGS to clinical practice is crucial not just to facilitate diagnosis but also to improve optimal health outcomes of patients by targeted health surveillance. In our study, the xBrowse bioinformatics interface was easy to use with limited training of clinical and academic researchers. Thus, transferring NGS technologies such as WES and whole-genome sequencing is feasible for many groups and will greatly improve the success rate of diagnosis for patients with neuromuscular disease.
Accepted for Publication: July 20, 2015.
Corresponding Author: Roula Ghaoui, FRACP, Institute for Neuroscience and Muscle Research, Kid’s Research Institute, Children’s Hospital at Westmead, Hawkesbury Road, Westmead, New South Wales, Australia 2145 (email@example.com).
Published Online: October 5, 2015. doi:10.1001/jamaneurol.2015.2274.
Author Contributions: Dr Ghaoui had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Ghaoui, Cooper, Jones, Sue, North, Clarke.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Ghaoui, Cooper.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: MacArthur.
Obtained funding: Cooper, Sue, Laing, North, MacArthur, Clarke.
Administrative, technical, or material support: Ghaoui, Cooper, Lek, Reddel, Kaur, Ong, Davis, Sue, Laing, MacArthur, Clarke.
Study supervision: Cooper, Jones, Corbett, Needham, Waddell, Nicholson, Sue, Laing, North, Clarke.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was supported by Australian National Health and Medical Research Council grant APP1074954 (Dr Ghaoui), grants APP1031893 and APP1022707 (Drs North, Laing, and Clarke), grant APP1035828 (Dr Clarke), grants APP1048814 and APP1048816 (Dr Cooper), and grant APP108433 (Dr Sue); by the Muscular Dystrophy New South Wales (Dr Ghaoui); and by the National Human Genome Research Institute of the National Institutes of Health (Medical Sequencing Program grant U54 HG003067 to Eric S. Lander, PhD, director of the Broad Institute and principal investigator of the National Human Genome Research Institute grant).
Role of the Funder/Sponsor: The National Human Genome Research Institute of the National Institutes of Health supported the design and conduct of the study and collection of the data, but had no role in the management, analysis, or interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: We thank the study patients and their families and the referring clinicians, without whose participation this work would not have been possible.