Author + information
- Received December 20, 2017
- Revision received March 18, 2018
- Accepted April 24, 2018
- Published online July 16, 2018.
- Richard D. Bagnall, PhDa,b,
- Jodie Ingles, MPH, PhDa,b,c,
- Marcel E. Dinger, PhDd,e,
- Mark J. Cowley, PhDd,e,
- Samantha Barratt Ross, BMedScia,b,
- André E. Minoche, PhDd,
- Sean Lal, MBBS, PhDb,c,
- Christian Turner, MBBSf,
- Alison Colley, MBBS, MMedScg,
- Sulekha Rajagopalan, MBBSg,
- Yemima Berman, MBBS, PhDh,
- Anne Ronan, MRCP, MEpidi,j,
- Diane Fatkin, MBBSe,k,l and
- Christopher Semsarian, MBBS, PhD, MPHa,b,c,∗ (, )@CSHeartResearch
- aAgnes Ginges Centre for Molecular Cardiology, Centenary Institute, Sydney, New South Wales, Australia
- bSydney Medical School, University of Sydney, Sydney, New South Wales, Australia
- cDepartment of Cardiology, Royal Prince Alfred Hospital, Sydney, New South Wales, Australia
- dGarvan Institute of Medical Research, Sydney, New South Wales, Australia
- eSt. Vincent’s Hospital Clinical School, University of New South Wales, Sydney, New South Wales, Australia
- fThe Sydney Children’s Hospital, Westmead, New South Wales, Australia
- gDepartment of Clinical Genetics, Liverpool Hospital, Liverpool, New South Wales, Australia
- hClinical Genetics Department, Royal North Shore Hospital, Sydney, New South Wales, Australia
- iHunter Genetics Unit, Newcastle, New South Wales, Australia
- jUniversity of Newcastle, Newcastle, New South Wales, Australia
- kMolecular Cardiology and Biophysics Division, Victor Chang Cardiac Research Institute, Sydney, New South Wales, Australia
- lCardiology Department, St. Vincent’s Hospital, Sydney, New South Wales, Australia
- ↵∗Address for correspondence:
Dr. Christopher Semsarian, Agnes Ginges Centre for Molecular Cardiology, Centenary Institute, Locked Bag 6, Missenden Road, Newtown, NSW 2042, Australia.
Background Whole genome sequencing (WGS) is a comprehensive genetic testing approach that reports most types of nucleotide variants.
Objectives This study sought to assess WGS for hypertrophic cardiomyopathy (HCM) in which prior genetic testing did not establish a molecular diagnosis, and as a first-line genetic test.
Methods WGS was performed on 58 unrelated patients with HCM, 14 affected family members, and 2 unaffected parents of a severely affected proband. The authors searched for nucleotide variants in coding regions of 184 candidate cardiac hypertrophy genes. They also searched for nucleotide variants in deep intronic regions that alter RNA splicing, large genomic rearrangements, and mitochondrial genome variants. RNA analysis was performed to validate splice-altering variants.
Results The authors found a pathogenic or likely pathogenic variant in 9 of 46 families (20%) for which prior genetic testing was inconclusive. Three families had variants in genes not included in prior genetic testing. One family had a pathogenic variant that was filtered out with prior exome sequencing. Five families had pathogenic variants in noncoding regions, including 4 with deep intronic variants that activate novel splicing, and 1 mitochondrial genome variant. As a first-line genetic test, WGS identified a pathogenic variant in 5 of 12 families (42%) that had never received prior genetic testing.
Conclusions WGS identified additional genetic causes of HCM over targeted gene sequencing approaches. Extending genetic screening to deep intronic regions identified pathogenic variants in 9% of gene-elusive HCM. These findings translate to more accurate diagnosis and management in HCM families.
Hypertrophic cardiomyopathy (HCM) is a primary myocardial disorder characterized by unexplained left ventricular hypertrophy (1). HCM is the most common genetic heart disease with a prevalence of up to 1 in 200 people (2). Clinical heterogeneity is a hallmark of the disease, ranging from asymptomatic individuals to development of arrhythmias, heart failure, and sudden death (1,3). HCM is caused by variants in genes primarily encoding sarcomere proteins (4). Genetic testing of 8 key sarcomere genes, and additional genes that cause diseases that can be misdiagnosed as HCM, gives an overall diagnostic yield of up to 40%; however, this increases up to 70% with a family history of HCM (5).
Accurate diagnosis of diseases that mimic HCM is important, because they can have additional clinical features, alternate inheritance modes, variable treatment options, and a different prognosis. Classical diseases that can be misdiagnosed as HCM include Fabry disease, Danon disease, and PRKAG2-related cardiomyopathies. Developmental disorders such as Noonan syndrome and Costello syndrome can also present with a variable cardiomyopathy phenotype associated with cardiofacial dysmorphology and musculoskeletal abnormalities (6). Similarly, mitochondrial disorders can often present with cardiac hypertrophy, and a wide associated clinical spectrum ranging from isolated organ involvement to multisystem disease (7).
Whole genome sequencing (WGS) covers all regions of the genome, including the mitochondrial genome, and thus represents the most comprehensive genetic testing approach. We sought to assess the incremental value of genetic testing with WGS for HCM in which prior genetic testing did not establish a genetic cause, that is, gene-elusive HCM. We also assessed WGS as a first-line genetic test in patients with HCM who had never undergone prior genetic testing.
The study population comprised 58 unrelated patients diagnosed with HCM, 14 affected family members, and 2 unaffected parents, who were referred to the National Centre for Hypertrophic Cardiomyopathy, Royal Prince Alfred Hospital, Sydney, Australia. HCM was diagnosed on echocardiographic demonstration of a hypertrophied and nondilated left ventricle, with a wall thickness ≥15 mm, or ≥13 mm for people with a family history of HCM, in the absence of another cardiac or systemic disease capable of producing a similar magnitude of hypertrophy (8). We selected patients with childhood onset HCM (age <18 years); and/or surgical intervention, including heart transplantation, myectomy, septal ablation, and/or had an implantable cardioverter-defibrillator in situ; and/or those with a family history of HCM, and/or sudden cardiac death; and/or a maximal left ventricular wall thickness ≥30 mm. Informed consent was obtained from all participants, or from parents in the case of children. Sydney South West Area Health Service ethics committee approved the study.
Genome sequencing, data processing, and variant analysis
Genomic DNA was isolated from venous blood using a Qiagen Mini Blood kit (Qiagen, Hilden, Germany). Genomes were sequenced to a mean read depth of 43, and on average, >99% of coding exons of 25 established, or associated, HCM genes were sequenced at least 15 times (Online Tables 1 and 2). DNA variants were called using the Genome Analysis Tool Kit (9), version 3.3, best practice workflow and annotated using Ensembl Variant Effect Predictor, version 89 (Online Appendix).
A search for potentially pathogenic variants focused on 184 candidate cardiac hypertrophy genes, comprising 58 autosomal dominant genes, 12 X-linked genes, and 114 autosomal recessive genes (Online Table 3). Genes were retrieved from a search of the Online Mendelian Inheritance in Man database, the Genetic Testing Registry, and a review of recent published reports.
We restricted analysis to variants with a general population allele frequency <4.0 × 10−5 for dominant autosomal and X-linked genes and assuming 50% penetrance (10), that is, an allele count <10 in the Exome Aggregate Consortium data of 60,207 individuals (11), and an allele frequency <0.01 for autosomal recessive genes. We focused on variants causing a missense or nonsense change, or that alter the canonical splice signal AG or GT dinucleotides, or lead to in-frame or frameshift insertions or deletions, and cosegregate with disease in affected family members, where available. Rare variants of interest were Sanger verified and classified for pathogenicity using the American College of Medical Genetics and Genomics guidelines (12).
Large genomic rearrangements
We analyzed WGS data for large DNA deletions, duplications, and inversions that spanned any coding sequence of 184 candidate cardiac hypertrophy genes using a combination of LUMPY (13) and CNVnator (14).
Splice gain variants
Intronic variants that create new AG or GT dinucleotide sequences, and with an allele count of <10 in over 15,000 whole genome sequences of the gnomAD Consortium data (11), were scored for a potential splice donor or acceptor signal sequence using MaxEntScan (15). We focused our search on 8 common HCM genes (MYBPC3, MYH7, TNNT2, TNNI3, MYL2, MYL3, TCAP, and ACTC1) plus 3 genes for which loss of function variants cause cardiac hypertrophy (LAMP2, PRKAG2, and ALPK3). Variants with a MaxEntScan score greater than the median value of donor or acceptor sequences of canonical exons of the relevant gene were evaluated for causing novel splicing with amplification of messenger RNA isolated from fresh venous blood, induced pluripotent stem cell–derived cardiomyocytes (16), or myectomy tissue.
Mitochondrial genome analysis
Mitochondrial genome variants were annotated using Mitomaster (17). We focused on rare variants found in <10 of 37,545 publicly available mitochondrial genomes, and excluding synonymous variants and variants in families with male transmission of HCM.
Cohort 1 included 46 unrelated gene-elusive probands with HCM who had previously undergone massively parallel sequencing-based analysis of a minimum of 46 cardiomyopathy genes (n = 40 probands, 87%), or Sanger sequencing of a minimum of 7 established HCM genes (n = 6 probands, 13%), depending on when the test was performed. Cohort 1 additionally included 14 family members with HCM from 12 families, plus both unaffected parents of 1 proband, that is, a family trio. Cohort 2 comprised 12 unrelated probands with HCM who had never received prior genetic testing (Central Illustration). The demographics, clinical characteristics, and previous genetic testing of cohorts 1 and 2 are summarized in Table 1 and shown in full in Online Table 4. Probands of cohorts 1 and 2 (n = 58) were predominantly male (n = 44, 76%), with a family history of cardiomyopathy and/or sudden cardiac death (n = 39, 67%), a maximal interventricular wall thickness of 23 ± 8 mm (range 11 to 43 mm), asymmetric septal hypertrophy (n = 44, 76%), and had an implantable cardioverter-defibrillator in situ (n = 30, 52%).
Genetic findings in Cohort 1
Protein coding variants
We found 4 (9%) families with a pathogenic or likely pathogenic variant in protein-coding sequences. In 3 families, a variant was found in genes not included in previous genetic tests, whereas in 1 family, the variant was filtered out of prior exome sequencing (Table 2). A well-described pathogenic variant in a Noonan syndrome gene, PTPN11 (c.1403C>T, p.Thr468Met), was found in the male proband of family OR (OR1). The Thr468Met variant has been reported in multiple probands with Noonan syndrome with multiple lentigines and zebrafish studies provide evidence that the variant may impact on protein function (18). The proband was diagnosed at 18 years of age with moderate nonobstructive HCM with a left ventricular wall thickness of 22 mm. Further clinical evaluation found the patient to have relatively wide-spaced eyes, pectus excavatum, and significant lentigines consistent with Noonan syndrome with multiple lentigines. He was recommended to undergo a hearing test and ophthalmology review due to the associated risk of deafness and cataracts in later life.
The male proband of family T was diagnosed with mild nonobstructive HCM at 45 years of age. He had nonsustained ventricular tachycardia on Holter monitoring and an implantable cardioverter-defibrillator was inserted. His son died suddenly at age 14 years while playing soccer, and postmortem examination revealed an interventricular septal wall thickness of 20 mm. At age 63 years, the proband was referred to a clinical geneticist for consideration of a possible diagnosis of Noonan syndrome. He measured 163 cm (3rd to 10th percentile for an adult male), with a low posterior hairline, relatively wide neck, low-set ears, and unilateral undescended testis, consistent with Noonan syndrome. WGS revealed a previously reported likely pathogenic RIT1 variant (c.265T>C, p.Tyr89His), which was shown to enhance Elk1 transactivation in in vitro studies (19), in the proband and his deceased son, genetically confirming a clinical diagnosis of Noonan syndrome.
The male proband of family AJ (AJ1) was diagnosed with nonobstructive HCM, with a left ventricular wall thickness of 26 mm. He also had a prolonged QT interval of 479 ms, and left anterior fascicular block. His son (AJ2) was diagnosed with HCM at 2 months of age with an interventricular septal wall thickness of 14 mm, and intraventricular conduction delay. The proband’s mother had HCM, and his brother with HCM died awaiting heart transplantation. We found a likely pathogenic CACNA1C variant (c.1553G>A, p.Arg518His) in 2 available affected members of family AJ. Arg518His and Arg518Cys variants in CACNA1C segregate in multiple members of 3 families with a similar combined phenotype of HCM, long QT syndrome, and sudden cardiac death (20). Whole-cell patch clamp studies revealed Arg518His results in loss of current density and inactivation in combination with increased window and late channel current (20).
Finally, in 2 members of family BKJ, we found a likely pathogenic synonymous variant in MYBPC3 (c.2274C>T, Gly758Gly), which is predicted to create a novel splice donor sequence (Figure 1A). Analysis of RNA isolated from venous blood of 3 available affected members of family BKJ revealed exon 23 was truncated by 36 nucleotides, leading to an in-frame deletion of 12 amino acids (Gly758–Ile769) (Figure 1B).
We found an additional 25 variants of uncertain clinical significance (VUS) in autosomal dominant genes in 20 families (43%) of cohort 1. In family BIJ, we found a VUS in MYH7 (c.3523C>T, p.Arg1175Trp) that was filtered out of prior exome sequencing because it was located in a 486 base pair (bp) sequence of MYH7 that has an identical copy in the homologous MYH6 (Online Figure 1). The exome sequencing reads are 100 nucleotides long and map equally well to the identical MYH6 and MYH7 regions, thus they have a high probability of inaccurate mapping (mapping quality score = 0) (Online Figure 2A). Most genotyping software programs filter variants in sequence reads with a mapping quality of zero. By contrast, the WGS reads are 150 nucleotides long, extend into unique sequences flanking the MYH7 duplication, and aligned accurately; thus the variant was genotyped (Online Figure 2B). We found 2 VUS of unknown chromosomal phase in the ACADVL gene, associated with autosomal recessive very-long-chain acyl-CoA dehydrogenase deficiency, in 1 family (Online Table 5A). No rare protein coding variants were found in 23 families (50%) of cohort 1.
WGS data can detect genomic deletions, duplications, and inversions. In 184 candidate cardiac hypertrophy genes, we detected and Sanger confirmed a novel deletion encompassing exon 5 of TK2 in an affected member of family BKJ (Online Figure 3); however, the deletion was absent from additional affected family members. We also detected and Sanger verified a 7,807-bp duplication in 2 families (Online Figure 4), which is also found in multiple individuals of the 1000 Genomes Consortium Phase 3 data (Database of Genomic Variants ID: esv3643818). Both rearrangements were classified as benign.
Intronic splice-gain variants
Intronic regions outside of canonical splice signal sequences may harbor variants that create splice signal motifs and may lead to inclusion of intronic sequence in the mature messenger RNA. We found 3 intronic MYBPC3 variants in 4 families, which activate splicing of novel pseudoexons (Table 2). In 3 affected members of family SW, a variant deep within intron 12 of MYBPC3 (c.1090+453C>T) creates a splice donor sequence, which leads to inclusion of 77 bp of intron 12 in the MYBPC3 mRNA (Figures 1C and 1D). The insertion causes a frameshift after Thr363, followed by 5 amino acids in the new reading frame, and a premature stop codon. In the proband of family PM, we found a novel c.1091-575A>C variant in intron 12 of MYBPC3, which creates a new splice acceptor site and leads to inclusion of 85 bp of intron 12 sequence between exons 12 and 13 (Figures 1E and 1F). This causes a frameshift after Thr363, followed by 53 amino acids in the new reading frame, and a premature stop codon. A novel c.1244-52G>A variant in intron 13 of MYBPC3 creates a splice acceptor site in 2 affected members of the AIB family, and the proband of family DA (Figure 1G). Analysis of MYBPC3 messenger RNA revealed inclusion of 50 bp of MYBPC3 intron 13 between exons 13 and 14 (Figure 1H), leading to a frameshift after Gly407, followed by 30 amino acids in the new reading frame, and a premature stop codon.
We also found a deep intronic MYBPC3 variant (c.1928-569G>T) in 3 members of the ID family that created a potential splice acceptor AG sequence with a maximum entropy score that was below the median value of MYBPC3 canonical splice acceptors. Analysis of RNA extracted from venous blood and myectomy tissue of the ID family showed canonical splicing of exons 20 and 21 only, and the variant was classified as benign.
Mitochondrial genome variants
WGS includes sequencing of the mitochondrial genome, which may harbor variants causing mitochondrial cardiomyopathy. We found and Sanger validated an m.4300A>G variant in a conserved region of the mitochondrial isoleucine transfer ribonucleic acid gene in the proband of family OO (Figure 2). WGS and Sanger sequencing show that the variant is homoplasmic in venous blood of the proband; a woman with mild asymmetric septal hypertrophy, maximal left ventricular wall thickness of 13 mm. She was diagnosed with HCM, which progressed to a dilated phase, and she required heart transplantation at age 43 years. Her mother also has HCM, and there is a family history of sudden cardiac death in a maternal uncle. She has 2 children, 14 and 12 years of age, a nonidentical twin brother, and a sister, all of whom show no evidence of cardiac disease. The m.4300A>G variant was previously described in at least 3 families with maternally inherited HCM, mild asymmetric left ventricular hypertrophy, left ventricular dilation, heart transplantation, sudden cardiac death, and incomplete penetrance (21,22). Histochemical analysis of 2 families showed cytochrome c oxidase–negative cells in left ventricular tissue (22). The m.4300A>G variant is absent in over 37,500 mitochondrial genome sequences, and we classified the variant as pathogenic. A further 4 rare mitochondrial genome VUS were found in 4 families (Online Table 5C).
De novo variant analysis
Finding a nucleotide variant that arises anew (i.e., de novo) in a child with a genetic disorder, and is absent in unaffected parents, has been an effective approach to identify pathogenic variants and new disease gene associations (23). We searched for de novo variants in GM1 because we had also sequenced both parents. She had severe HCM with a maximal interventricular wall thickness of 42 mm and died suddenly at 36 years of age. Both parents are clinically unaffected, and there is no family history of disease. WGS did not detect any rare inherited or de novo variants in 184 candidate cardiac hypertrophy genes. When searching all protein coding genes, we found and Sanger confirmed a novel de novo VUS; NM_004824.3:c.1476+5G>A in the extended splice region of CDYL, which is predicted to disrupt splicing of exon 6 (Figure 3).
Genetic findings in Cohort 2
We found 5 pathogenic or likely pathogenic variants in 5 (42%) probands of cohort 2 (Table 2). A well-described single nucleotide duplication of MYBPC3 (c.2373dupG, p.Trp792ValfsTer41) found in family ALG segregated to her affected sister, and an MYH7 variant (c.2221G>T, p.Gly741Trp) found in the proband of family BNP segregated to his affected brother and mother. A mouse model provides supportive evidence that the Arg92Trp variant in TNNT2 impacts on protein function (24). In a further 5 probands (42%), we found 8 VUS in autosomal dominant genes; 1 proband had 2 rare VUS of unknown chromosomal phase in the NDUFS2 gene, associated with autosomal recessive Complex 1 respiratory chain deficiency (Online Table 5A). No rare variants were found in the remaining probands, and no genomic rearrangements of 184 candidate cardiac hypertrophy genes, deep intronic variants, or mitochondrial variants, were found in cohort 2.
In this study, we explore the utility of WGS in patients with gene-elusive HCM and identified a clinically relevant variant in 9 of 46 families (20%) (Central Illustration). In 3 families, a likely pathogenic variant was found in genes not included in previous testing, which led to reclassification of disease, and in 1 family, prompted referral for clinical evaluation of Noonan syndrome. In 1 family, a synonymous variant was found that creates a new splice donor site. In 5 families, WGS found a pathogenic variant in noncoding regions, including 4 families with deep intronic variants that activate splicing of new exons, and 1 family with a mitochondrial genome variant. As a first-line genetic test, WGS identified a genetic cause in 5 of 12 families (42%) who had never undergone prior genetic testing, in keeping with the diagnostic yield of gene panel and exome sequencing–based approaches for HCM. This study highlights the improved genetic testing yield of WGS for HCM and, in particular, how extending genetic screening to intronic regions identified a pathogenic variant in 4 of 46 gene-elusive families (9%).
WGS expands the scope of genetic testing for HCM
Genetic testing typically targets protein-coding sequences; however, deep intronic regions may also harbor a smaller proportion of pathogenic variants that alter RNA splicing. Analysis of RNA directly demonstrates the consequences of splice-disrupting variants, as recently shown for a range of primary myopathies and muscular dystrophies (25). When RNA from a disease-relevant tissue is not available, an alternative approach is to sequence intronic regions with WGS and evaluate rare variants for the potential to create new splice sites using in silico tools. With this latter approach, we found 3 MYBPC3 intronic variants predicted to create novel splice sites. Analysis of RNA isolated from fresh venous blood, or induced pluripotent stem cell–derived cardiomyocytes, of the patients confirmed novel pseudo-exon inclusion and exon extension. One additional MYBPC3 intronic variant had no observable effect on splicing, emphasizing the requirement of RNA analysis to confirm the effects of in silico–predicted splice sites. Although clinical genetic testing centers do not yet routinely analyze RNA, this only requires an additional step of reverse transcription when compared with analysis of DNA. Furthermore, clinical genetic testing centers often require a fresh blood sample for confirming genetic testing outcomes, which provides a possible source of RNA. Sharing the results of RNA studies in public databases, such as ClinVar, is important because this information is useful for the clinical interpretation of variants by genetic testing centers who may not have access to RNA.
MYBPC3 is a major disease gene of HCM, acting predominantly through loss of function mutations that lead to haploinsufficiency. Recently, a combination of in silico analyses and in vitro assays provided evidence that a number of MYBPC3 VUS alter RNA splicing (26). Our current work directly demonstrates that deep intronic variants alter RNA splicing in 9% of gene-elusive families and the overall contribution of MYBPC3 variants to HCM is likely to be greater than previously appreciated. Genetic testing of HCM should include the intronic regions of MYBPC3, particularly when there is a family history of disease and with no strong candidates from prior genetic testing.
WGS found a common pathogenic mitochondrial genome mutation. The mitochondrial genome was sequenced to an average read depth of over 4,000 with WGS, which allows detection of variants even at low levels of heteroplasmy. We have previously shown that the mitochondrial genome can be partially reconstructed from off-target exome sequencing reads (27); however, the sequencing read depth is low. Thus, screening deep intronic regions and high depth sequencing of the mitochondrial genome can be regarded as advantages of WGS over exome and targeted gene panel sequencing approaches.
WGS data enable the detection of large genomic rearrangements. Although changes in exome sequencing read depth can indicate multiexon deletions and duplications, this approach is not well suited to detect single-exon deletions and cannot locate breakpoints that fall outside of sequenced regions. Using WGS, we detected a genomic deletion and duplication, and revealed the boundaries of the rearrangements. This information guided the design of oligonucleotide sequences to amplify across the breakpoint junctions, providing a precise genetic test for the variants. Because the deletion was not inherited with disease, and the duplication was too common to directly cause HCM, both variants were classified as benign. Our work is in keeping with previous studies that demonstrate a very minor role for large gene deletions and duplications in HCM (28,29). However, large genomic rearrangements are an important class of pathogenic variation in other cardiac disorders, such as arrhythmic cardiomyopathy (30) and the long QT syndrome (31).
Genetic findings unrelated to WGS
We made new genetic diagnoses that were not directly related to the WGS approach. Three families had pathogenic variants in genes that were not included in previous gene panel testing; however, these variants would have been detected using extended cardiomyopathy gene panel designs that include RASopathy genes and CACNA1C. A synonymous MYBPC3 variant found on exome sequencing was not initially considered to cause HCM. Following WGS and our focus on candidate splicing variants, we noted that the synonymous variant creates a possible splice site, which we confirmed with RNA studies. We identified a VUS in a coding exon of MYH7 that was filtered out of prior exome sequencing because it had a low mapping quality score due to a sequence duplication. In contrast to shorter exome sequencing reads, the longer WGS reads extended into unique sequences flanking the duplication allowing the variant to be called, highlighting an advantage of longer sequencing read lengths.
De novo variant analysis
WGS of a severely affected patient and both unaffected parents enabled a search for de novo variants across all protein coding genes, and we found a single potential splice-disrupting variant in CDYL. This gene encodes chromodomain Y-like protein, which directly interacts with the histone methyltransferase, EZH2 (enhancer of zeste 2 polycomb repressive complex 2 subunit), and maintains repressive histone methylation marks during DNA replication (32). Loss of EZH2-directed histone methylation in cardiac progenitor cells of mice leads to cardiac hypertrophy and fibrosis (33).
CDYL is intolerant of loss of function variants, with only 3 such variants found in over 138,000 individuals of the gnomAD Consortium data. We did not have access to RNA from the proband to verify an effect on splicing as she is deceased, and neither parent harbors the variant; however, multiple in silico tools predict a loss of splice site recognition. Our result raises the intriguing question of whether loss-of-function variants in CDYL cause severe HCM.
WGS as a first-line genetic test
As a first-line genetic test, WGS identified a pathogenic variant in 42% of patients who had never received prior genetic testing, which is in keeping with the diagnostic yield of other sequencing-based approaches for HCM. All of these clinically relevant variants were located in coding regions of established HCM genes and would have been found with exome sequencing or gene panel sequencing. Recently, WGS was shown to identify all but 1 known mutation in 20 patients with HCM (34). The single missed variant was an 18-bp duplication that was sequenced, but filtered, with the initial threshold settings used. Collectively, this suggests WGS can largely detect variants identified using gene panel and exome sequencing approaches. WGS data are larger and take longer to analyze than exome sequencing, which poses some challenges for incorporating WGS into clinical care. However, WGS has the potential to identify variants of all size, and from regions of the nuclear and mitochondrial genome that no other technology can identify cost-effectively.
Genetic testing recommendations for HCM
We recommend that the most cost-effective first-line genetic test for all HCM is gene panel or exome sequencing–based analysis of genes established to cause HCM, or diseases that can be misdiagnosed as HCM. If this initial genetic test does not reveal a causal variant, we recommend no further testing for people with late-onset HCM and a mild phenotype. However, if there is no family history of disease, but a severe clinical presentation, a search for de novo variants with WGS of a family trio may be considered. For gene-elusive patients with a family history of disease, WGS-based analysis of intronic regions and the mitochondrial genome may reveal a pathogenic variant in an additional 9% of this challenging subgroup of HCM.
For patients in whom no causal variant was identified in 184 candidate cardiac hypertrophy genes, we did not extend analysis to additional genes because there is no evidence for a role in cardiac hypertrophy, and any found variants would have a speculative role in disease. Future investigations of these families will require sequencing of additional affected family members, where available, to allow filtering of shared variants, and combining the WGS data in a larger cohort of gene-elusive patients to increase statistical power to find novel gene associations.
In this study, we demonstrate how WGS can detect genetic variants not identified with sequencing of protein-coding exons only; thus WGS improves the yield of genetic testing for HCM. In particular, we find an additional 9% of gene-elusive HCM patients have pathogenic variants in deep intronic regions of MYBPC3 that result in a splice-gain. We also demonstrate that WGS as a first-line genetic test identifies pathogenic variants in 42% of patients tested. The favorable outcomes of WGS will translate to more accurate diagnosis of HCM and diseases that can be misdiagnosed as HCM, and therefore facilitate more targeted therapies with the ultimate goal to improve clinical management.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: Genetic testing of patients with familial HCM should include protein coding and noncoding regions of genes that have established associations with the disease as well as genes that cause diseases that mimic HCM and variants predicted to cause RNA splicing errors.
TRANSLATIONAL OUTLOOK: Further studies are needed to quantify the RNA isoforms resulting from splicing errors in patients with HCM and to clarify how these correlate with clinical features of the disease.
The authors acknowledge the Sydney Informatics Hub and the University of Sydney’s high performance computing cluster Artemis for providing the high-performance computing resources that have contributed to the research results reported within this paper.
This research is supported by a grant from NSW Health Genomics Collaborative Grants Program. Dr. Ingles is a recipient of a Heart Foundation of Australia Future Leader Fellowship (#100833). Dr. Semsarian is the recipient of a National Health and Medical Research Council (NHMRC) Practitioner Fellowship (#1059156). Dr. Dinger is an employee of Genome.One, a commercial genetic testing company. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- base pair
- hypertrophic cardiomyopathy
- variant of uncertain significance
- whole genome sequencing
- Received December 20, 2017.
- Revision received March 18, 2018.
- Accepted April 24, 2018.
- 2018 American College of Cardiology Foundation
- Maron B.J.,
- Ommen S.R.,
- Semsarian C.,
- Spirito P.,
- Olivotto I.,
- Maron M.S.
- Semsarian C.,
- Ingles J.,
- Maron M.S.,
- Maron B.J.
- Maron B.J.,
- Maron M.S.,
- Semsarian C.
- Semsarian C.,
- CSANZ Genetics Council Writing Group
- McKenna A.,
- Hanna M.,
- Banks E.,
- et al.
- Whiffin N.,
- Minikel E.,
- Walsh R.,
- et al.
- Abyzov A.,
- Urban A.E.,
- Snyder M.,
- Gerstein M.
- Ross S.B.,
- Fraser S.T.,
- Bagnall R.D.,
- Semsarian C.
- Lott M.T.,
- Leipzig J.N.,
- Derbeneva O.,
- et al.
- Boczek N.J.,
- Ye D.,
- Jin F.,
- et al.
- Bannwarth S.,
- Procaccio V.,
- Lebre A.S.,
- et al.
- Taylor R.W.,
- Giordano C.,
- Davidson M.M.,
- et al.
- Ertz-Berger B.R.,
- He H.,
- Dowell C.,
- et al.
- Cummings B.B.,
- Marshall J.L.,
- Tukiainen T.,
- et al.
- Ito K.,
- Patel P.N.,
- Gorham J.M.,
- et al.
- Ceyhan-Birsoy O.,
- Pugh T.J.,
- Bowser M.J.,
- et al.
- Pilichou K.,
- Lazzarini E.,
- Rigato I.,
- et al.
- Liu Y.,
- Liu S.,
- Yuan S.,
- et al.
- Cirino A.L.,
- Lakdawala N.K.,
- McDonough B.,
- et al.