Introduction

Inherited cardiomyopathies are a group of heart muscle diseases characterized by heterogeneous phenotypes, including dilated cardiomyopathy (DCM), hypertrophic cardiomyopathy (HCM), and arrhythmogenic right ventricular cardiomyopathy (ARVC), with higher morbidity and mortality [1, 2]. DCM, defined by the presence of left ventricular (LV) or biventricular dilatation and systolic dysfunction in absence of hypertension, valvular disease, or coronary artery disease [1, 2], is one of the leading causes of heart failure (HF) and sudden death (SD) and represents the most common condition requiring a heart transplant [1, 2]. HCM, characterized by increased LV wall thickness that is not solely explained by abnormal loading conditions [3, 4], is the most common cause of SD in young, and is considered an important cause of HF and embolic stroke secondary to atrial fibrillation [5, 6]. A diagnosis of ARVC is made mainly based on functional and structural abnormalities of the right ventricle (RV), fibro-fatty replacement of the myocardium, depolarization and repolarization alterations, arrhythmias with left bundle branch block (LBBB) morphology, and family history, with no single gold standard exam of diagnosis established to date [7]. ARVC represents one of the major causes of SD in the young and athletes [8, 9]; rarely, RV or biventricular dysfunction leads to HF [10]. Phenotypic overlap between different cardiomyopathies has been well described, leading to diagnostic uncertainty in some cases [11].

Over the past 20 years, substantial progress in the understanding of the genetic basis of cardiomyopathies has been made [1–3, 12, 13]. Recent family screening studies showed that autosomal dominant inheritance was the predominant pattern of transmission up to 48% of DCM cases; possibilities of X-linked, autosomal-recessive, and mitochondrial inheritance were less common [14]. To date, genes that encode sarcomere, cytoskeleton, desmosomes, nuclear envelop proteins, and ion channels have been found to be associated with DCM [15, 16], with TTN being the most frequently associated gene [17]. Current studies show that up to 60% of HCM patients exhibit an autosomal-dominant trait with mutations in genes mostly affecting sarcomere proteins [3, 18–20]. Moreover, mutations in genes that encode Z-disc proteins or intracellular calcium modulators have also been detected [3]. The classical inheritance pattern of ARVC is autosomal-dominant with variable expression and age-related penetrance [21]. To date, most of genes that have been found associated with the ARVC phenotype encode desmosomal proteins and cause disease in up to 50% of probands [12, 21]. However, non-desmosomal genes have also been implicated in ARVC development [12]. In addition to the genetic heterogeneity, a wide genetic overlap has also been noted among different cardiomyopathic phenotypes, and between these heart muscle disorders and channelopathies [22–29].

According to recent international guidelines, genetic testing is strongly recommended for each patient with inherited cardiomyopathy to identify a causative mutation and subsequently provide pre-symptomatic testing of relatives who are at risk of developing the same disease at a later stage; moreover, some disease-causing gene mutations have been found to be associated with more severe clinical features, presentation at an early age, overall poor prognosis, or increased risk of SD [30–34]. Therefore, genetic testing can have considerable implications in terms of early diagnosis, implementation of prognostic stratification algorithms, and timely therapeutic interventions [32, 35].

Sanger sequencing is a standard technique of molecular diagnostics for disorders predominantly related to a single causative gene. However, this screening method is laborious. The high degree of genetic heterogeneity, with more than 100 cardiomyopathy-related genes identified to date, can be addressed more efficiently by high-throughput sequencing, referred to as next-generation sequencing (NGS) [36], which includes targeted resequencing, whole exome sequencing, and whole genome sequencing. Targeted enrichment appears to be the method of choice in a clinical diagnostic setting, not only because it allows focusing on the genes relevant to a particular disorder, thereby preventing unsolicited findings, but also because it provides a superior quality of exon representation and coverage.

In this study, we developed a custom “pan-cardiomyopathy panel” containing 115 genes known to be associated with DCM, HCM, and ARVC as well as to channelopathies. Based on this panel, we performed molecular screening in 38 unrelated patients, 16 affected by DCM, 14 by HCM, and 8 by ARVC, to better investigate the genetic overlap between these three phenotypes and to identify new gene–phenotype associations using the Illumina MiSeq platform. Moreover, we investigated the possible role of genes linked to channelopathies in modifying the clinical cardiomyopathic phenotype.

The results of this study put a spotlight on novel cardiomyopathic gene–phenotype associations, extending the mutational spectrum underlying cardiomyopathies. Furthermore, our data suggest that some genes involved in channelopathies may act as genetic modifiers, modulating the clinical features and severity of cardiomyopathic phenotypes, as well as contributing to the variable cardiomyopathic phenotypic expression. Our findings support the “pan-cardiomyopathy panel” approach, since many patients carried multiple rare variants in several genes.

Materials and methods

Results

Variant assessment

In total, we detected 142 rare variants in 40 genes, of which 137 variants had a heterozygous status and 5 variants on the X chromosome had a hemizygous status. Thirty-three (23,2%) of these 142 rare variants were absent from both dbSNP and gnomAD (S5 Table). One hundred and twenty-four rare variants were missense variants (87%), while 5 were frameshift insertion/deletion variants, 3 in-frame insertion/deletion variants, 8 nonsense variants, and 2 involved a splicing site (S6 Table). The TTN gene had the largest number of rare variants (42 variants), followed by OBSCN (16 variants) and RYR2 (6 variants).

Considering rare variants for each cardiomyopathy type, in ARVC patients, the highest number of rare variants was found in TTN and RYR2 genes (4 variants) followed by OBSCN and PKP2 (3 variants) (S7 Table). For DCM patients, the highest number of rare variants was found in TTN gene (26 variants), followed by OBSCN (8 variants) and ANK2 (4 variants) genes (S8 Table). For HCM patients, the highest number of rare variants was found in TTN gene (12 variants), followed by OBSCN (5 variants) and MYBPC3 (4 variants) genes (S9 Table).

All patients carried at least one rare variant, while most of them (94.7%) carried more than one rare variant; on average, 3.7 rare variants per patient (range, 1–9) were detected, with DCM patients carrying the highest average number of rare variants (4.5 per patient), followed by HCM (3.6 per patient), and ARVC (2.4 per patient) (S10 Table).

Of the 142 rare variants, 28 were classified as potentially pathogenic, 9 of which already present in ClinVar and/or HGMD databases and/or supported by published data (Table 1 and S11 Table). Fifteen potentially pathogenic variants were missense mutations, 4 were frameshift insertion/deletion mutations, 2 were in-frame insertion/deletion mutations, 5 were nonsense mutations, and 2 affecting a splicing site (Table 1). We excluded from the list of the potentially pathogenic variants, all the TTN missense variants indicated as potentially pathogenic by the analysis with the bioinformatics tools because, as recently reported, there is no statistical difference in frequency between cases and controls and their potential pathogenic role can be established only by a segregation analysis in affected families [48, 49]. Regarding the TTN truncating variants, we followed the recent evidences that suggest to consider as potentially pathogenic only those variants affecting all TTN isoforms in the A-band region [50, 51]. We identified three TTN truncating variants in DCM patients. We considered as potentially pathogenic only the TTN truncating variant V16477fs, referred to the N2B isoform (NM_003319) as it maps in the TTN A-band region and was also not present in gnomAD database. The R2490fs variant, referred to the Novex-3 transcript (NM_1333379) and the L24944X variant, referred to the N2B transcript (NM_003319) were not considered potentially pathogenic as they mapped in the I and M-band regions of the TTN isoforms, respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Potentially pathogenic rare variants detected in our patients.

https://doi.org/10.1371/journal.pone.0181842.t001

In twelve patients (310DCM, 365DCM, 968DCM, 1584DCM, 1669DCM, 1717DCM, 1838DCM, 1685HCM, 1798HCM, 1665ARVC, 1708ARVC and 1830ARVC), no potentially pathogenic variant was identified, whereas 3 patients (76DCM, 1173HCM and 1776HCM) carried two potentially pathogenic variants (Table 2 and S10 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Genetic profile of each cardiomyopathic patient carrying potentially pathogenic rare variants.

https://doi.org/10.1371/journal.pone.0181842.t002

New gene–phenotype associations

Most of the rare variants identified in our study showed a well-established association with a specific cardiomyopathic phenotype. In addition, in 23 of 38 patients, we found at least one novel potential gene–phenotype association (Table 2 and S10 Table). As shown in Fig 1, considering the most represented genes, we found six new associations for AKAP9 gene, four for ANK2 gene, three for DLG1, OBSCN and TRPM4 genes and two for PSEN2 and SYNE1 genes. In particular, we focused on potentially pathogenic rare variants to examine the novel gene–phenotype associations (Table 2 and S11 Table). For the ARVC phenotype, we found two potentially pathogenic variants involving DMD and OBSCN genes, not previously associated with this phenotype. Indeed, we found one novel potentially pathogenic missense variant in DMD gene in patient identified as 1751ARVC (DMD_R1320H) and one potentially pathogenic missense variant in OBSCN gene.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. New gene–phenotype associations and number of novel gene rare variants for each cardiomyopathic phenotype.

https://doi.org/10.1371/journal.pone.0181842.g001

Regarding the HCM phenotype, we identified two potentially pathogenic variants involving DLG1 and AKAP9 genes, not previously associated with this phenotype. In DLG1 gene, a potentially pathogenic missense variant was found in patient identified as 1776HCM (DLG1_G639R); in AKAP9 gene, a novel potentially pathogenic nonsense variant was found in patient identified as 1833HCM (AKAP9_Y3870X).

With regard to DCM, the association between potentially disease-causing variants in OBSCN gene and this phenotype was recently described [54]. Our findings were in agreement with this new association, as we found two potentially pathogenic variants in OBSCN gene, one in patient identified as 682DCM (OBSCN_R6669H) and the other in patient identified as 1718DCM (OBSCN_A5660V). Moreover, for DCM phenotype, we identified one potentially pathogenic variant involving the AKAP9 gene, not previously associated with DCM, in patient identified as 76DCM (AKAP9_R3435X).

Discussion

The technology of NGS is now emerging as a powerful approach to comprehensively explore genetic mutations in a wide range of human pathologies [36, 55]. In this study, we used targeted resequencing to identify novel genetic variations associated with three inherited cardiomyopathies, namely HCM, DCM, and ARVC, which are common causes of mortality before the fifth decade of life. These inherited cardiomyopathies are genetically heterogeneous and are associated with rare mutations in a large number of genes, many of which overlap among the phenotypes. Moreover, considerable phenotypic and genetic overlaps between cardiomyopathies and arrhythmic syndromes have been documented [22–29]. Considering these overlaps, our study aimed to look for new potential gene–phenotype associations with a view to expand the mutational spectrum underlying cardiomyopathies, with possible implications for better diagnosis and clinical management. We designed a custom gene panel including 115 genes, known to be associated with the three cardiomyopathic phenotypes and/or arrhythmic syndromes. Using this panel, we performed molecular screening of a cohort of 38 unrelated patients affected by cardiomyopathies (16 DCM, 14 HCM, and 8 ARVC). Overall, our analysis identified a total of 142 rare variants in 40 genes and confirmed 11 rare variants (ACTC1_S236F, LAMP2_V310I, LMNA_Leu140-Ala146dup, LMNA_Q517X, MYBPC3_A364T, MYBPC3_c.821+1G>A, MYBPC3_E728X, MYH7_A1763T, MYH7_G701D, MYL2_E134A, and TNNT2_R102Q) which we previously identified in some of our patients by Sanger sequencing [52, 56], thereby demonstrating that our approach, that combined NGS and different in silico prediction tools, was both reliable and sensitive at identifying rare variants. Interestingly, for 75 genes in our panel, no rare variant was identified, suggesting that some gene–phenotype associations reported in the literature (on which our gene panel was based) are restricted to unique or few cases.

Our results confirm previous data indicating that cardiomyopathic patients could carry multiple specific genetic variants [33, 34, 57]. In fact, we found on average more than three rare variants per proband, with DCM patients carrying the highest average number of rare variants, followed by HCM and ARVC patients. Furthermore, since our subjects were selected on the basis of phenotype severity and family history of cardiomyopathy and/or SD, our results support the evidence that multiple mutations are likely to elicit a more severe phenotype. Indeed, previous studies suggested that patients carrying multiple mutations are more likely to develop the disease at an early age and with more severe clinical features, as poor prognoses are associated with compound or double heterozygotes [33, 34].

Titin is the largest described human protein, constituting one of the most abundant types of filaments in both cardiac and skeletal muscle and plays a central role in sarcomere organization. TTN has been recognized as the main disease-causing gene in DCM and is also involved in the pathogenesis of HCM and ARVC [17, 49, 58]. Our study confirmed the high prevalence of TTN rare variants in cardiomyopathies, most of which are missense. TTN missense variants are very common and often frequently benign and it was reported that TTN missense variants bioinformatically classified as “severe” did not associate with differences in clinical phenotypes in DCM patients, so arguing that many TTN missense variants are not disease-causing variants [48]. For these reasons, many TTN missense variants identified in our study were not classified as potentially pathogenic, despite the results of predictors’ analysis. Regarding TTN truncating variants, recent studies supported that not all these variants are equals in terms of clinical relevance. In particular, the TTN truncating variants with the highest probability of pathogenicity are those affecting TTN isoforms in the A-band region [50, 51]. In our study, we identified three TTN truncating variants in DCM patients: V16477fs (referred to NM_003319 transcript), L24944X (referred to NM_003319 transcript) and R2490fs (referred to NM_133379 transcript). According to recent evidences, we considered as potentially pathogenic, with a possible clinical relevance, only the V16477fs variant, as it maps in the A-band region and affects all TTN transcripts. However, it cannot be ruled out that multiple variants (missense and truncating) on the TTN protein may have a cumulative effect on a cardiomyopathic phenotype.

In this study, we extended the number of overlapping genes among cardiomyopathies and also between cardiomyopathies and arrhythmic syndromes, identifying novel potential gene-phenotype associations (Fig 1 and S2 Table). As reported in Fig 2, which shows the distribution of specific and overlapping genes carrying rare variants identified in this study, among the three cardiomyopathic phenotypes, we detected five genes (DMD, OBSCN, SYNE1, TNNC1, VCL) carrying rare variants in the three phenotypes, six genes (AKAP9, CACNA1C, CHRM2, DLG1, MYO6, PSEN2) carrying rare variants in HCM and DCM, and one gene (TRPM4) carrying rare variants in ARVC and DCM. As several patients carried variants in genes not previously related to their phenotype, the use of our “pan-cardiomyopathy panel” may prove beneficial to identify the novel potential gene-phenotype associations. In fact, a phenotype-specific targeted gene testing [30] might miss to identify several variants in cardiomyopathic patients. Conversely, a comprehensive gene testing could enhance the diagnostic and prognostic implications of genetic screening by not only allowing an early diagnosis of inherited cardiomyopathic forms but also by contributing to a more accurate prognostic stratification and patient management [31].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Distribution of specific and overlapping genes carrying rare variants identified in this study, among the three cardiomyopathic phenotypes.

https://doi.org/10.1371/journal.pone.0181842.g002

OBSCN variants were firstly described in HCM patients [59] and, more recently, OBSCN variants associated with DCM and Left Ventricular Noncompaction patients were reported [54, 60]. Our results are in agreement with the association between OBSCN gene variants and DCM phenotype, since we found eight different rare variants in 8 DCM patients. When our analysis was restricted to potentially pathogenic variants, the association between OBSCN variants and DCM remained, because two potentially pathogenic variants, OBSCN_R6669H and OBSCN_A5660V, were detected in patients identified as 682DCM and 1718DCM, respectively (Table 2). Furthermore, for the first time to our knowledge a possible association between OBSCN gene and the ARVC phenotype is evidenced. In fact, we found three different OBSCN rare variants in two out of eight ARVC probands. One of these variants, OBSCN_A5791P was classified as potentially pathogenic (Table 1 and S11 Table) and was detected in patient identified as 1662ARVC (Table 2). A possible pathogenic role for OBSCN variants detected in some of our DCM and ARVC patients is supported by the fact that these variants involve OBSCN domains having functional and/or structural importance and affect highly conserved amino acid regions across species. Obscurin is a member of a family of giant proteins that is expressed in striated muscle and interacts with titin, myomesin and small ankyrin-1 and has been proposed to be a structural protein linking the M-line of the sarcomere to the sarcoplasmatic reticulum [61, 62]. Different obscurin isoforms are known: obscurin A (~720 kDa); obscurin B (~870 kDa), which is similar to obscurin A but lacking the nonmodular COOH-terminal region and including two Ser/Thr kinase domains, called SK2 and SK1; and two smaller obscurin isoforms, one containing a full-length SK1 and a partial SK2 and the other containing only the SK1 domain. The expression of these obscurin isoforms differs; while giant obscurins A and B are expressed in higher amounts in skeletal muscle, the tandem and single kinase isoforms are largely present in cardiac muscle [62]. Variant OBSCN_R6669H involves the SK1 domain, and variant OBSCN_A5791P falls in a functionally important domain of obscurin, the tandem Rho-guanidine nucleotide exchange factor (RhoGEF) present in Rho/Rac/Cdc42-like GTPases, and thus these variants may affect obscurin cell signaling pathways. Variant OBSCN_A5660V is present in the src homology-3 (SH3)-conserved domain, a protein interaction domain. Therefore, OBSCN appears to be a good candidate gene for both DCM and ARVC phenotypes. Recent data supported OBSCN mutations as a significant causal factor for DCM pathogenesis, reporting that disease-related OBSCN mutations cause haploinsufficiency that accounts for the development of the DCM phenotype [54]. Moreover, it was suggested that OBSCN mutations could act alone or in concert with other mutations, contributing to cause DCM [54, 63]. Our data are in agreement with this last hypothesis, since patients carrying a potentially pathogenic variant in the OBSCN gene also carried other rare variants in different genes, such as PKP2, or TTN, suggesting a possible cumulative effect on cardiomyopathic phenotype.

For the ARVC phenotype, we also found an association with the DMD gene. Mutations in the DMD gene, encoding for dystrophin, are responsible for X-linked DCM and myopathies [64–66]. ARVC patients were found to be affected by myofibrillar myopathy [67, 68], facio-scapulo-humeral muscular dystrophy [69], and myotonic dystrophy type 1 [70]. An ARVC-like phenotype was described in desmin-related myopathy [71] and myotonic dystrophy type 1 [72]. In addition, some of the genes recently found to be associated with ARVC are also responsible for myopathies/muscular dystrophies, such as DES [73], LDB3 [74], and TTN [75]. Moreover, both phenotypic and genetic overlaps between DCM and ARVC were observed [52, 75–77].

We identified the DMD_R1320H variant in the patient identified as 1751ARVC, who was diagnosed with ARVC and associated LV systolic dysfunction at the age of 42 years. The novel variant is localized in the spectrin domain number 9 of dystrophin and is predicted to be pathogenic in silico because it is believed to affect two DMD transcripts. Mutations in dystrophin affecting the plasma membrane and causing myofiber loss could lead to the pathologic substrate favoring myocardial atrophy and fibrofatty replacement, which are the main pathogenetic mechanisms described in ARVC [78]. However, more evidences are necessary to consider DMD as a new gene associated with the ARVC phenotype.

Although hereditary cardiomyopathies and arrhythmic syndromes are considered distinct groups of genetic disorders, a considerable phenotypic overlap between these two apparently distinct cardiac disorders has been documented. In fact, the clinical cardiomyopathic picture often includes a wide range of features related to arrhythmias and conduction abnormalities; conversely, ion channel defects linked to arrhythmic syndromes may be associated with morphological changes and structural abnormalities that characterize some types of cardiomyopathy [5, 24–29, 79]. In a recent study that screened HCM probands for both sarcomere and non-sarcomere protein gene mutations, ANK2, PLN, and SCN5A gene variants were suggested to have modifying effects on HCM phenotypic expression [26]. Moreover, novel associations between three genes (KCNQ1, KCNH2, and KCNE2) associated with Long QT syndrome and HCM phenotype were proposed [26]. As the main goal of our study was to look for new gene–phenotype associations, we included in our custom panel some genes associated with inherited arrhythmic syndromes in order to support the evidence that genes implicated in channelopathies could modify the clinical characteristics and severity of cardiomyopathic phenotypes. We found novel potential gene–phenotype associations between genes typically related to arrhythmic syndromes (AKAP9, ANK2, CACNA1C, DLG1, HCN4, KCNJ8, NUP155 and TRPM4 genes) and the cardiomyopathic phenotypes discussed in our study (S2, S9 and S11 Tables). In particular, focusing on the potentially pathogenic variants, we detected new associations between AKAP9 variants and both DCM and HCM and between DLG1 variant and HCM (Table 2 and S11 Table).

A novel association between AKAP9 gene and DCM was observed in patient identified as 76DCM. This patient demonstrated an early-onset (15 years) DCM phenotype and presented with a history of palpitations, syncope on effort, and evidence of frequent multifocal premature ventricular complexes, non-sustained VT episodes, and atrioventricular conduction abnormalities recorded on Holter monitoring. He received an ICD. During follow-up, he suffered from recurrent episodes of sustained VT and VF discontinued by ICD shocks. This patient carried two potentially pathogenic variants: DSP_V508fs variants and AKAP9_R3435X (Table 2 and S11 Table). DSP_V508fs is a novel variant, predicted to produce a truncated protein and to reduce DSP expression. AKAP9_R3435X is predicted to produce a truncated protein from three different AKAP9 transcripts, thus suggesting either a dominant negative mechanism or a nonsense-mediated mRNA decay mechanism. We suppose that the DSP_V508fs variant is the primary disease-causing mutant since the DSP gene has been demonstrated to be associated with DCM (OMIM: 615821) and biventricular arrhythmogenic cardiomyopathy (OMIM: 607450) and that AKAP9 may be a modifying gene causing severe ventricular tachyarrhythmias, such as those occurring in the patient identified as 76DCM. This hypothesis is supported by the evidence that AKAP9 has been considered as a genetic modifier of congenital LQTS, increasing cardiac risk and disease severity in AKAP9 mutation carriers [80].

Another potentially pathogenic variant (AKAP9_Y3870X) is present in the patient identified as 1833HCM, who also carries the MYL2_E134A variant (S10 Table). The MYL2 variant may have a pathogenic role in 1833HCM patient, as previously described for the same phenotype [81], although it is classified as VUS in ClinVar database. The AKAP9_Y3870X variant may act as a disease severity modifier, even though the stop codon introduced by this mutation is located in the C-terminal part of the protein; so, its effect should be analyzed in functional studies and/or segregation in families.

DLG1 encodes the synapse-associated protein 97 protein (SAP97), a protein belonging to the membrane-associated guanylate kinase (MAGUK) family of proteins that are important for localization and organization of ion channels. SAP97 interacts with and modulates cardiac Kv4.3 channels that account for a large part of the transient outward potassium current (Ito) [82], but also binds and regulates the expression of NaV1.5, the main cardiac inward sodium channel [83]. Both channels are implicated in Brugada syndrome (BRGDA1, OMIM: 601144; BRGDA9, OMIM: 616399), and consequently a DLG1 variant may confer susceptibility to arrhythmias as demonstrated by a putative DLG1 pathogenic mutation in a patient affected by this syndrome [84]. We found the potentially pathogenic variant DLG1_G639R in patient identified as 1776HCM, also carrying the potentially pathogenic variant MYBPC3_A364T (Table 2, S10 and S11 Tables). The MYBPC3_A364T variant may be considered pathogenic for this patient, as MYBPC3 is one of the main genes associated with HCM. The DLG1_G639R variant involves the DLG1 SH3 domain, which mediates the assembly of specific protein complexes via binding to proline-rich peptides. It is conceivable that this mutation could affect DLG1-Kv4.3 or DLG1-NaV1.5 interactions, modulating their functions and favoring susceptibility for arrhythmias.

Our study demonstrates that a targeted resequencing approach can provide a relevant support for a genetic diagnosis in cardiomyopathic patients. However, in many patients, in particular, those affected by DCM, it was not possible to obtain a clear genetic diagnosis on the basis of our gene panel, thus suggesting a greater genetic heterogeneity in this cardiomyopathy respect to the others and/or a considerable role of modifier genes in determining this phenotype. Gene discovery by whole exome or genome sequencing would be required for those probands that are still without a molecular diagnosis. However, the application of this approach would make the interpretation of genetic screening results difficult for clinical purposes due to the large amount of sequencing data that need to be analysed.

We are aware that our study would need to be supported by evidences in favor of a pathogenic role of the variants reported here (e.g., lack of segregation analysis due to sporadic cardiomyopathy, and/or unavailability and/or small number of relatives available for genetic investigations and lack of functional studies) [85], not allowing us to assess the clinical relevance of the identified variants. Segregation analysis together with functional studies may concur to establish a more definite diagnosis, allowing the findings to be interpreted in the context of a clinical setting. However, families with multiple affected individuals are a rare occurrence and often a novel variant is present in a single individual with only few or no other clinically affected relatives. Consequently, many novel NGS variants are classified as VUS and cannot represent a conclusive test result at present, as recently highlighted [85].

On the other hand, we believe that the rigorous approaches used in this study to identify the cardiomyopathies associated variants (e.g. recruitment of patients with more severe phenotypes and family history of cardiomyopathy and/or sudden death, selection of variants using a MAF lower than the frequency of each cardiomyopathy, application of different in silico prediction tools to select the potentially pathogenic variants) reliably support our hypothesis that many of them may contribute to cardiomyopathy development and/or its progression. In particular, in agreement with results obtained by Marston et al. [54], mutations in the OBSCN gene could be considered a significant causal factor, alone or in concert with other mutations, for DCM and ARVC pathogenesis. Furthermore, variants in AKAP9 and DLG1 genes could act as genetic modifiers of arrhythmic risk and phenotype severity in cardiomyopathies. In addition, variants of unknown pathogenicity, sometimes complicating the clinical picture, may contribute to the disease development and/or may have modifier effects. It is possible to hypothesize that the genetic component underlying cardiomyopathic phenotypes and/or modifying disease phenotype severity could be explained by the cumulative functional effects of coding and non-coding DNA variants.

To conclude, this work highlights the challenges faced in the diagnosis of cardiomyopathies. NGS is a highly accurate and reproducible approach for routine molecular screening of patients with cardiomyopathies, but the identification of genetic defects in the proband is not sufficient in isolation for clinical decision-making or genetic counselling.

Finally, our data support the remarkable contribution that a targeted NGS approach could offer to genetic overlap knowledge between different cardiomyopathies, expanding previous results and ameliorating molecular diagnosis.

In the brackets, we report the number of the new potential gene-phenotype associations (number on the left) and the number of the overlapping genes (number on the right) detected in this study (see also underlined associations in S2 Table).

Supporting information

Acknowledgments

The authors are grateful to the patients for their participation in this study. We thank Mrs. Giovanna Petragallo for her technical support.