Patients with coronary heart disease, dilated cardiomyopathy and idiopathic ventricular tachycardia share overlapping patterns of pathogenic variation in cardiac risk genes

Introduction

A total of 17.5 million people died from cardiovascular diseases in 2012 and of these, 7.4 million deaths were due to coronary heart disease (CHD), making it the number one cause of cardiac deaths, according to the WHO (World Health Organization, 2018). More than 75% of the cardiovascular disease (CVD) deaths occurred in low- and middle-income countries, with high rates in Kazakhstan, despite numerous state-run health programs. The age-standardized mortality rate of cardiovascular disorders in Kazakhstan is amongst the highest in the world. According to the latest WHO data published in 2018, there were 47,651 coronary heart disease deaths in Kazakhstan, amounting to 34.56% of total deaths. The age-adjusted death rate is 306.02 per 100,000 population, which puts Kazakhstan as eighth in the world, and approximately 9.6-fold higher than Japan (2018: 31.55 per 100,000 of population ranks Japan #181 in the world) and 4.05-fold higher than Austria (a death rate of 75.74 per 100,000 population ranks Austria #143 in the world) (World Health Organization, 2018).

Kazakhstan is located in the middle of Central Asia on the ancient Great Silk roads. Its vast territory covers 2,724,900 km² and it is the world’s 9th largest country. Kazakhstan has a population of more than 17 million people (2016), with 131 ethnicities, including Kazakh (63% of the population), Russian, Ukrainian, German, Uzbek, Tatar, and Uyghur as the most predominant groups. Historically, the ethnic Kazakhs were nomadic and migration led to the admixing of western and eastern tribes. Kazakhstan is ethnically and culturally diverse, which is due, in part, to the forced migration of settlers and mass deportations of ethnic groups, starting in the 19th century until the first third of the 20th century (Lee & Akilzhanova, 2015).

Genetic studies in Kazakhstan are challenging because of the genetic heterogenicity introduced by many ethnicities. However, data from diverse, heterogeneous populations exposed to the same environmental conditions and similar lifestyles yield important information on natural genetic plasticity. Such data are fundamental to genetic epidemiology and are critical to dissect natural polymorphisms from pathogenic alterations (Akilzhanova et al., 2014). Genome-wide data and linkage disequilibrium patterns are unavailable for Central Asian populations and are not represented in publicly available databases.

Cardiovascular disease encompasses a range of conditions from diseases of the vasculature, myocardial infarction, and congenital heart disease, most of which are heritable. Enormous effort has been invested in understanding the genes and specific DNA sequence variants responsible for this heritability (Abraham et al., 2014; Al-Hassnan et al., 2016).

Dilated cardiomyopathy (DCM) accounts for 30–40% of all heart failure cases and is a leading cause of heart transplantation (Haas et al., 2015). An autosomal dominant inheritance pattern of transmission and some autosomal recessive or X-linked recessive familial cases have been reported as indicated by the familial aggregation of DCM (30–50% of all DCM cases) (Al-Hassnan et al., 2016; Haas et al., 2015). Mutations in more than 30 disease genes have been linked as causative mechanisms of DCM (Haas et al., 2015; Meder et al., 2011).

In contrast, CHD, which is typically a result of coronary artery disease (CAD), is a complex disease driven by interactions between genetic factors and environmental stimuli and stressors. The genetic basis of CAD/CHD has been addressed by multiple genome-wide association studies (GWAS) enrolling thousands of individuals (Lieb & Vasan, 2013; Consortium, 2013; CARDIoGRAM Consortium, 2010; Coronary Artery Disease, C4D). Genomic risk scores (GRS) have been developed using millions of datapoints (SNPs) and plasma markers. These can contribute to our understanding of critical mechanisms to provide the maximum benefit to the individual, especially before the early stages of pathogenesis. Classical CHD markers defined by the Framingham risk score (FRS), including age, cholesterol, smoking status, blood-pressure, and diabetes status are not predictive in a timely way (Akilzhanova et al., 2014). Genetic risk loci have been reported for cell proliferation genes, inflammation and immunity related genes, cholesterol and lipid biogenesis genes, among others. However, only a relatively minor risk could unambiguously be attributed to the wealth of common genetic variants in CHD heritability. Family-based analyses revealed different heritability estimates for distinct sub-phenotypes of CHD (Fischer et al., 2005). There is a remarkable consistency in genetic association findings across cohorts (with varying phenotype definitions), underscoring that different manifestations of CHD may have a common genetic architecture (Lieb & Vasan, 2013; Kitsios et al., 2011). Rare variants with a larger impact and/or more common variants with a smaller biological impact were proposed to cause the observed missing heritability (Haas et al., 2015; Schunkert, Erdmann & Samani, 2010; Manolio et al., 2009).

Ventricular arrhythmias in patients with structural heart disease are responsible for the majority of sudden cardiac deaths (SCD). CAD, previous myocardial infarction is the most common heart disease in which sustained ventricular tachycardia (VT) occurs and reentry is the predominant mechanism. Other cardiac conditions, such as idiopathic DCM, Chagas disease, sarcoidosis, arrhythmogenic cardiomyopathies, and repaired congenital heart disease may also present with VT in follow-up (Hadid, 2015).

Recurrent ventricular tachycardia (VT) is an important cause of increased morbidity and mortality in patients with non-ischemic DCM. DCM differs from postinfarction ischemic cardiomyopathy by comprising multiple different etiologies with variable disease progression and prognosis. There is a need for an individualized approach to risk stratification and treatment based on genetic information.

Idiopathic ventricular tachycardia is defined as VT that occurs in patients without structural heart disease, metabolic abnormalities, or long QT syndrome. 10% of all patients referred for evaluation of VT show no obvious structural heart disease. Idiopathic VT is characterized by a structurally normal heart and QRS morphology consistent with site of origin from typical locations of idiopathic ventricular arrhythmias (in particular, the ventricular outflow region). An absence of structural heart disease is usually suggested if an electrocardiogram (ECG) (except in Brugada syndrome and long QT syndrome [LQTS]), echocardiogram, and coronary arteriogram are collectively normal (Srivathsan et al., 2005). However, magnetic resonance imaging (MRI) may identify structural abnormalities even if all other test results are normal. Idiopathic VT comprises multiple discrete subtypes that are differentiated by their mechanism, QRS morphology, site of origin, the response to pharmacologic agents, and evidence of catecholamine dependence.

They include right ventricular (RV) monomorphic extrasystoles, RV outflow tract (RVOT) VT, left ventricular (LV) outflow tract (LVOT) VT, idiopathic LV tachycardia (ILVT), and idiopathic propranolol-sensitive (automatic) VT (IPVT). Idiopathic VT from the RVOT and LV are monomorphic and generally not familial. Catecholaminergic polymorphic VT (CPVT), Brugada syndrome, and LQTS are inherited ion channelopathies (Srivathsan et al., 2005).

Polymorphic VT may cause syncope and sudden death in Brugada syndrome. Patients with idiopathic VT monomorphic forms have a better prognosis than do patients with polymorphic VT and structural heart disease. Prognosis for patients with VT secondary to ion channelopathies is variable (Srivathsan et al., 2005).

Ninety-two patients diagnosed with ventricular tachycardia (VT) with either coronary heart disease (CHD), dilated cardiomyopathy (DCM) or idiopathic ventricular tachycardia (iVT), were enrolled in a study to evaluate the genetic profile and variants in known cardiac risk genes by targeted next generation sequencing (NGS). We hypothesized that shared morphological and phenotypical features among these subgroups might originate from an overlapping molecular basis. In addition, we assumed that the spacious genepool of the population of Kazakhstan that is fueled by more than 100 different ethnic groups deems this study cohort a challenging but valuable source for interpreting disease-associated genetic variations. Our results provide an important contribution to the understanding of human genetic diversity.

Materials & Methods

Results

Ninety-two unrelated patients with ventricular tachycardia (VT) and either coronary heart disease (CHD), dilated cardiomyopathy (DCM) or idiopathic ventricular tachycardia (iVT) were prospectively enrolled for genetic analyses to evaluate their genetic profile and variation in known cardiac risk genes (Bekbossynova et al., 2018; Akilzhanova et al., 2019).

The clinical characteristics of the patients in the three clinical subgroups are summarized in Table 1 and listed in detail in Table S1. The predominant NYHA functional classes were II-III for CHD, III-IV for DCM, and I-II for the iVT subgroup. LVEF, LA, LV ESD, and LV EDD significantly differed between the iVT and the two other subgroups (Table 2). The proportion of familial cases was nearly the same in all three subgroups, ranging from 25–29.7%.

Genetic variants

Targeted enrichment and sequencing, variant calling and stepwise filtering of the annotated variants identified a total of 307 unique variants in 74 genes totaling up in 456 variants for the overall study group. The frequency and pathogenicity of variants associated with the different arrhythmogenic syndromes within the different subgroups studied is shown in Table S2. Variants included: one in/del variant, four splice-site variants, and 451 single-nucleotide variants (SNV) within the coding exonic regions. Seven (0.15%) of the SNVs were unique stop-gain variants, three of which were residing in the TTN gene. 168 HGMD mutations (61 unique) were observed in 37 genes. 33% of the HGMD mutations were predicted to have an increased pathogenic potential (class I and II variants); 49% were classified class III variants and 18% of the observed HGMD variants were classified as benign by all prediction algorithms (Fig. 1). In contrast to the HGMD variants, a higher proportion (40%) of the novel and rare variants (≤0.5% in ESP6500 or 1000G db) was predicted to be pathogenicity classes I and II (Fig. 1).

Variants with the highest pathogenicity score (class I variants) made up approximately 8.7% of the CHD VT subgroup, and 6.3% and 18.9% in the DCM VT and the iVT groups, respectively (Fig. 2). The prevalence of class II variants was moderately lower (30.4%) in the CHD VT group for the DCM VT and the iVT groups (31.4% and 43.2%, respectively).

Statistical testing of and/or rare variants between the different clinical subgroups (testing was done including and omitting titin variants) indicated no difference in the average number of HGMD variants or rare variants alone between the three subgroups (Table S3, Fig. S1A, Fig. S1B).

Classification of 307 variants according to ACMG guidelines showed that 9 (2.9%) variants were classified as pathogenic, 9 (2.9%) were likely pathogenic, 98 (31.9%) had uncertain significance, 73 (23.8%) were likely benign and 118 (38.4%) were benign (Fig. 3, Table S2). ACMG pathogenic and likely pathogenic variants were observed in classes III and IV, and contrary benign and likely benign variants were observed in class I and class II variants. Most variants were variants of uncertain significance.

Distribution of the functional effects of detected mutations

We evaluated whether there might be a differential distribution of variants between the disease groups in relation to their functional context. We grouped the genes into seven categories using information from GeneCards®and The Human Gene Database https://www.genecards.org/ (cell membrane, cytoskeleton, sarcomere, metabolism, intercalated disc, ion flux, and nucleus) based on their molecular function and/or subcellular association. Distribution of class I–IV variants according to their molecular function/association are shown in Table 4. There was a moderate, but statistically insignificant, underrepresentation of variants in the cell membrane (4.3%) genes and intercalated disc (4.3%) genes in CHD VT patients compared to the other groups (DCM VT: 9.4 and 12.5; iVT: 10.8% and 21.6%, respectively) when we included only class I and II variants in the analysis (Fig. 4 and Fig. S3 and Table S5). Variants in the metabolism-associated genes were moderately overrepresented in the CHD VT subgroup (13%) compared to the DCM VT subgroup (3.1%) alone.

Table 4:

Distribution of class I–IV variants according to their molecular function/association.

Class I (highest pathogenic potential) variants were predicted disease causing by at least seven of the tools, class II (intermediate pathogenic potential) variants were predicted disease causing by 4–6 of the tools, class III (low pathogenic potential) were predicted disease causing by 1–3 prediction tools, and class IV (benign) was predicted disease causing by none of the tools (0).

	Class I	Class II	Class III	Class IV
Cell membrane	0 (0%)	7 (41.2%)	8 (47.1%)	2 (11.8%)
Cytoskeleton	5 (9.8%)	12 (23.5%)	22 (43.1%)	12 (23.5%)
Sarcomere	14 (11.7%)	34 (18.3%)	62 (51.7%)	10 (8.3%)
Metabolism	4 (23.5%)	2 (11.8%)	10 (58.8%)	1 (5.9%)
Intercalated disc	6 (20.0%)	5 (16.7%)	15 (50.0%)	4 (13.3%)
Ion flux	12 (30.8%)	7 (17.9%)	18 (46.2%)	2 (5.1%)
Nucleus	7 (21.2%)	3 (9.1%)	21 (63.6%)	2 (6.1%)

DOI: 10.7717/peerj.10711/table-4

Based on their relative frequencies HGMD mutations in LAMA2 (34.3%), MYBPC3 (31.2%), MYH6 (18.7%), KCNQ1 (15.6%), GAA (15.6%) and DSG2 (12.5%) were predominant in the DCM VT subgroup (Fig. 5). The mutation and variant distribution of the CHD VT subgroup strongly overlapped with the patterns for the other subgroups (Table S2, Table S6, and Fig. S4). PRKAG2 mutations p.G100S (n = 3, 13%, CM136115) and novel p.H222Q variant were observed in four CHD VT patients (Table S2). Statistical testing suggested a trend towards an increased frequency of PRKAG2 variants in the CHD subgroup (CHD VT: 13.04% vs. DCM VT: 3.1% and iVT: 2.7%; p-value 0.053) (Fig. S4). There was a prevalence of iVT mutations in genes encoding ion flux.

There were 9 pathogenic variants of ACMG, including W746C in GAA in patients with CHD, R218Q KCNJ2 and R5338X TTN in patients with iVT, and F244L MYH7, Q353X LMNA, L17465X TTN, W21011X TTN, c.2334+1G >A DSG2, c.477+1G >A KCNQ1 in patients with DCM (Table S2). Sanger sequencing confirmed some of the observed genetic variants. (Fig. S5).

Discussion

We evaluated the contribution of molecular genetic variants in genes associated with cardiac disorders in Kazakhstani population. We identified a significant proportion of possible pathogenic variants using molecular genetic screening with a targeted next-generation sequencing (NGS) panel (Bekbossynova et al., 2018; Akilzhanova et al., 2019). We obtained data on the distribution of genetic variants, the number of mutations, and the mutational burden of patients with ventricular tachycardia of various etiology.

NGS technologies have emerged as an efficient alternative to Sanger-sequencing, providing the analytical characteristics for the comprehensive exploration of genetic mechanisms (Frese, Katus & Meder, 2013; Roberts et al., 2013). Furthermore, it is believed that NGS will be increasingly important in the studies of monogenic and complex diseases, such as common cardiovascular diseases (CAD, cardiomyopathies and others) in which one or more variants in a single gene, or multiple variants in different genes, are involved (Haas et al., 2015; Meder et al., 2011; Frese, Katus & Meder, 2013; Roberts et al., 2013). The ability of NGS to generate high-throughput qualitative and quantitative sequence information has enabled investigations that were previously technically infeasible or cost prohibitive (Schrijver et al., 2012).

There are some disadvantages to the use of NGS, including the incomplete representation and coverage of exons, which poses the risk of limiting sensitivity and the inability to detect clinically significant mutations. Targeted enrichment of certain genes followed by the use of NGS for high-throughput genetic testing of genes for heart disorders is now becoming feasible and technically proven. This has been evidenced by the almost complete coverage and high accuracy of the approach, offering greater sequencing depth with reduced costs and data burden (Meder et al., 2011; Schrijver et al., 2012).

We sequenced three groups of patients with CHD VT, DCM VT and VT of unknown etiology (idiopathic) in our study to identify genetic variants that were associated with the three cardiovascular phenotypes and to evaluate the level of genetic variation in cardiac risk genes in these distinct subgroups. We designed and optimized a custom target-enrichment assay of 96 genes associated with cardiac disorders using Haloplex technology (Agilent Technologies, Santa Clara, USA) (Bekbossynova et al., 2018; Akilzhanova et al., 2019).

Targeted enrichment and sequencing and stepwise filtering of the annotated variants identified a total of 307 unique variants in 74 genes totaling up in 456 variants for the overall study group. The filtering step is crucial for bioinformatics analysis to reduce the number of probable and potentially pathogenic variants such as the exclusion of common variants present in the Single Nucleotide Polymorphism database (dbSNP). Filtering is based on assumptions about the attributes of the disease-causing variant(s), including the effect of the variant on the protein, the presumed absence of the variant in the dbSNP database, or the frequency cutoffs based on minor allele frequency from the 1000 Genomes Project (Schrijver et al., 2012).

The DCM and other common congenital heart disorders affect approximately 1–4 people per 10,000 population. We selected a MAF cut-off of 0.5% for rare variants in order to balance the rate of false positives (non-pathogenic variants) at the cost of losing variants with moderate/intermediate pathogenicity (true positives) that provoke increased susceptibility for complex diseases like CHD. With the introduction of multigene panels, exome sequencing, and whole-genome sequencing, the numbers of variants identified per person has increased progressively, and disease mutation databases contain potentially benign variants that were previously classified as disease causing. The interpretation of genetic variants is complex.

We used a combined score of 10 prediction tools to determine the pathogenic potential of each variant (HGMD listed variants and novel or rare variants). Class I (highest pathogenic potential) variants were predicted as being disease causing by at least 7 of the tools; class II (intermediate pathogenic potential) variants were predicted as being disease causing by 4–6 of the tools; class III (low pathogenic potential) were predicted as being disease causing by 1–3 prediction tools; and class IV (benign) was predicted as being disease causing by none of the tools (0).

As expected, the MAF was the lowest for the variants with the most severe pathogenicity (class I) (0.000279), followed by intermediate pathogenicity variants (class II) with an MAF of 0.00267 being nearly a 10-fold higher than that for class I. The average MAF for the variants with the lowest pathogenicity (class III variants with low pathogenic potential) was relatively common at 0.00372.

The pathogenicity of sequence variants was classified using an automatic variant classifier according to the ACMG guidelines. The classifications were: ‘pathogenic’, ‘likely pathogenic’, ‘likely benign’, ‘benign’, or ‘uncertain significance’. Information about HGMD listed variants and the classification of pathogenicity found in the ClinVar, Varsome and final verdict according to the ACMG/AMP were summarized in Table S2.

We classified 307 variants according to ACMG guidelines, which showed that 9 (2.9%) variants were pathogenic, 9 (2.9%) were likely pathogenic, 98 (31.9%) had uncertain significance, 73 (23.8%) were likely benign and 118 (38.4%) were benign (Fig. 3). ACMG pathogenic and likely pathogenic variants were observed in classes III and IV, and contrary benign and likely benign variants were observed in class I and class II variants. Most variants had uncertain significance.

Our results show that complex methods are required to make a final interpretation of sequenced variants.

Patients of all subgroups were clinically diagnosed according to common international classification criteria. Despite the many common genetic risk variants identified for CHD in GWAS, they only account for a small percentage of the expected heritability (Roberts et al., 2013; Qian & Bodmer, 2012). The predisposition to CHD is estimated to be approximately 50% genetic, although the 36 variants identified by CARDIoGRAM and the follow-up CARDIoGRAMplusC4D project only accounted for about 10% of the heritability. Rare risk variants with minor allele frequencies ≤1–5%, complex gene-gene interactions (epistasis), and undiscovered common variants are thought to cause this discrepancy. We sequenced 23 individuals with CHD VT and observed a spectrum of genetic variation that quantitatively (frequency of genetic variants) and qualitatively (molecular function) strongly overlapped with DCM VT and iVT. 43.5% (10/23) of CHD VT patients carried a class I variant, whereas only 5% (3/60) of the control cohort (KCG), 31.3% (13/32) of the DCM VT and 54.1% (20/37) of iVT patients carried a class I variant. If class II variants were added, 69.5% (16/23) of CHD VT, 75% (24/32) of DCM VT and 81.1% (30/37) of iVT patients carried on average two variants of high to intermediate pathogenicity. High- to intermediate pathogenicity variants in LAMA2, MYBPC3, MYH6, KCNQ1, GAA, and DSG2 predominated in CHD VT patients at similar frequencies as those observed for DCM VT and iVT patients. This similarity points to a common molecular disease-association. Independent of multiple tested disease-associated mutations (HGMD) and rare or newly identified variations with increased pathogenic potential, there was no statistically significant difference in the frequency of genetic variation between the three subgroups. Our results confirmed that DCM and iVT patients frequently carry multiple mutations or variants with high pathogenic potential, which has been shown in previous research. The high frequency of rare genetic variants with increased pathogenic potential (class I and II) in CHD patients was unexpected. The commonly referenced concepts of cardiomyopathies as monogenic disorders has been challenged (Haas et al., 2015; Qian & Bodmer, 2012; Lopes et al., 2013), indicating complex interactions of genes and the significance of rare variants. Distinct clinical phenotypes, including LQT, Brugada syndrome and HCM, revealed that multiple mutations, and rare potential pathogenic and functional variants in affected individuals could synergistically or additively alter penetrance, age-of-onset, or disease progression (Meder et al., 2011; Kapa et al., 2009; Stattin et al., 2012; Allegue et al., 2015).

TTN, GAA, LAMA2 and MYBPC3 harbored the most variants in the three subgroups which confirm the high impact of these genes in complex pathogenesis of cardiomyopathies and VT demonstrated in previous studies (Herman et al., 2012; Golbus et al., 2012; Bit-Avragim et al., 2001; Carboni et al., 2011).

Classification of the variants according to their cellular function showed that sarcomere function, ion-flux, nuclear function, and metabolism were predominantly affected by variants with the highest pathogenic potential (class I variants) in CHD VT patients. A similar pattern was observed for class I variants in DCM VT and iVT patients. In addition, iVT patients carried variations potentially affecting the cytoskeleton and intercalated disc. Pooling class I-III variants yields metabolism-associated variants in >60% of CHD VT patients, which is second behind variants in sarcomere genes. On a gene basis, PRKAG2 mutations were overrepresented in the CHD sub-group versus DCM and iVT, (p-values = 0.053 and 0.054, respectively). The mean age of the four heterozygous mutation carriers was 67.7 years (+/- 5.1y). PRKAG2 encodes the γ2 regulatory subunit of the AMP-activated protein kinase AMPK and mutation-associated defects account for a cardiac syndrome triad consisting of familial ventricular preexcitation (Gollob et al., 2001), conduction system disease, and cardiac hypertrophy mimicking (HCM) (Gollob et al., 2002), with a significant proportion of those progressing to DCM. HCM-associated PRKAG2 mutations are generally not associated with myocyte and myofibrillar disarray, which are the pathognomonic features of HCM, but with pronounced vacuole formation within myocytes due to excessive glycogen accumulation (Aggarwal et al., 2015; Arad et al., 2002). This may be explained by the central regulatory function of AMPK during acute low-energy states in which ATP-consuming pathways are shut off, like glycogen, cholesterol and fatty acid synthesis and the ATP-producing pathways are enhanced, such as fatty acid oxidation and glucose uptake. The potential functional role of AMPK in atherosclerosis has recently been shown by the protective effect of melatonin on the cardiovascular system, since flow shear stress-induced apoptosis in bone marrow mesenchymal stem cells could be reversed via the activation of AMPK (Yang et al., 2016).

Conclusions

We showed that in patients with VT, secondary to coronary artery disease, DCM, and idiopathic etiology, multiple rare mutations and clinically significant sequence variants in classic cardiac risk genes associated with cardiac channelopathies and cardiomyopathies were found in a similar pattern and at a comparable frequency. CHD VT patients were found to carry rare genetic variants with an increased pathogenic potential at a comparable frequency as DCM VT and iVT patients. These variants were found in genes related to sarcomere function, nuclear function, ion flux, and metabolism. Our study size was limited but this pilot study suggests that monogenic diseases can serve as an insightful model for complex disorders. Patients with coronary heart disease, dilated cardiomyopathy, and idiopathic ventricular tachycardia share overlapping patterns of pathogenic variation in cardiac risk genes. A greater in-depth statistical analysis like sub-grouping of participants according to other features like ethnicity, severity, or anamnesis (familial vs. sporadic, etc.) was not possible at this stage due to the group size limitation. Additional studies including more patients with and without ventricular tachycardia will be needed to generate a deeper insight into genotype-phenotype correlation of CHD and cardiomyopathies and between idiopathic ventricular tachycardia types.

Supplemental Information