High prevalence of genetic variants previously associated with Brugada syndrome in new exome data

Authors

  • B Risgaard,

    Corresponding author
    1. Danish National Research Foundation Centre for Cardiac Arrhythmia (DARC), University of Copenhagen, Copenhagen, Denmark
    2. Laboratory of Molecular Cardiology, Department of Cardiology, The Heart Centre
    3. Department of Cardiology, The Heart Centre, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark
    • Corresponding author: Bjarke Risgaard, MD, Department of Cardiology, Section 2142, Copenhagen University Hospital, Rigshospitalet, Blegdamsvej 9, DK-2100 Copenhagen, Denmark.

      Tel.: +45 35 45 65 01;

      fax:+45 35 45 65 00;

      e-mail: bjarkerisgaard@gmail.com

    Search for more papers by this author
  • R Jabbari,

    1. Danish National Research Foundation Centre for Cardiac Arrhythmia (DARC), University of Copenhagen, Copenhagen, Denmark
    2. Laboratory of Molecular Cardiology, Department of Cardiology, The Heart Centre
    3. Department of Cardiology, The Heart Centre, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark
    Search for more papers by this author
  • L Refsgaard,

    1. Danish National Research Foundation Centre for Cardiac Arrhythmia (DARC), University of Copenhagen, Copenhagen, Denmark
    2. Laboratory of Molecular Cardiology, Department of Cardiology, The Heart Centre
    Search for more papers by this author
  • AG Holst,

    1. Danish National Research Foundation Centre for Cardiac Arrhythmia (DARC), University of Copenhagen, Copenhagen, Denmark
    2. Laboratory of Molecular Cardiology, Department of Cardiology, The Heart Centre
    3. Department of Cardiology, The Heart Centre, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark
    Search for more papers by this author
  • S Haunsø,

    1. Danish National Research Foundation Centre for Cardiac Arrhythmia (DARC), University of Copenhagen, Copenhagen, Denmark
    2. Laboratory of Molecular Cardiology, Department of Cardiology, The Heart Centre
    3. Department of Cardiology, The Heart Centre, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark
    4. Department of Medicine and Surgery, Faculty of Health Science, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • A Sadjadieh,

    1. Department of Cardiology, Copenhagen University Hospital of Bispebjerg, Copenhagen, Denmark
    Search for more papers by this author
  • BG Winkel,

    1. Danish National Research Foundation Centre for Cardiac Arrhythmia (DARC), University of Copenhagen, Copenhagen, Denmark
    2. Laboratory of Molecular Cardiology, Department of Cardiology, The Heart Centre
    3. Department of Cardiology, The Heart Centre, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark
    Search for more papers by this author
  • MS Olesen,

    1. Danish National Research Foundation Centre for Cardiac Arrhythmia (DARC), University of Copenhagen, Copenhagen, Denmark
    2. Laboratory of Molecular Cardiology, Department of Cardiology, The Heart Centre
    Search for more papers by this author
  • J Tfelt-Hansen

    1. Danish National Research Foundation Centre for Cardiac Arrhythmia (DARC), University of Copenhagen, Copenhagen, Denmark
    2. Laboratory of Molecular Cardiology, Department of Cardiology, The Heart Centre
    3. Department of Cardiology, The Heart Centre, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark
    Search for more papers by this author

  • The authors declare no conflict of interest.

Abstract

More than 300 variants in 12 genes have been associated with Brugada syndrome (BrS) which has a prevalence ranging between 1:2000 and 1:100,000. Until recently, there has been little knowledge regarding the distribution of genetic variations in the general population. This problem was partly solved, when exome data from the NHLI GO Exome Sequencing Project (ESP) was published. In this study, we aimed to report the prevalence of previously BrS-associated variants in the ESP population. We performed a search in ESP for variants previously associated with BrS. In addition, four variants in ESP were genotyped in a second Danish control population (n = 536) with available electrocardiograms. In ESP, we identified 38 of 355 (10%) variants, distributed on 272 heterozygote carriers and two homozygote carriers. The genes investigated were on average screened in 6258 individuals. This corresponds to a surprisingly high genotype prevalence of 1:23 (274:6258). Genotyping the four common ESP-derived variants CACNA2D1 S709N, SCN5A F2004L, CACNB2 S143F, and CACNB2 T450I in the Danish controls, we found a genotype prevalence comparable with that found in ESP. We suggest that exome data are used in research, as an additive tool to predict the pathogenicity of variants in patients suspected for BrS.

Today hundreds of variants in 12 genes have been associated with Brugada syndrome (BrS), which is considered a rare and inherited Mendelian disorder with a prevalence ranging between 1:2000 and 1:100,000 [1-5]. BrS is a primary arrhythmic syndrome, characterized by ST-segment elevations in the right precordial leads (V1–V3), with an increased risk of sudden cardiac death, due to malignant ventricular arrhythmias, in the absence of a structural heart disease [6]. Since BrS was first described in 1992 [7], it has been associated with several genetic variations affecting the cardiac sodium current, the potassium transient outward current, and the calcium current [8-10]. However, in the clinical setting SCN5A is the only recommended gene for targeted screening in patients with BrS, but only 15–30% of all cases are caused by loss of function in this gene [11]. SCN5A encodes the alpha subunit (Nav1.5) of the cardiac sodium channel complex, and was the first gene to be associated with BrS [8, 12]. SCN5A has also been associated with several other arrhythmogenic diseases such as lone atrial fibrillation and long QT syndrome (LQTS) [13-15].

In the last years, several other genes have been associated with BrS. Untill now, the function of the sodium channel complex is also known to be modulated by RANGRF, SCN1B, SCN3B and GPD1L genes which have all been associated with BrS phenotypes [16]. Other currents important for the action potential have also been associated with BrS, including the inward transient potassium current and the inward calcium current. As such, in the setting of BrS, the inward transient potassium current has been shown to be affected by mutations in the KCNJ8, KCND3, KCNH2 and KCNE3 genes [16]. Finally, changes in the inward calcium current has been shown to be caused by mutations in the CACNA1C, CACNA2D1, and CACNB2 genes [9].

Until recently, there has been little knowledge regarding the distribution of genetic variations in the general population. Potentially it could be a problem, when rare variants are associated with a rare disease, as some of these variants most likely just have a modifying effect [16]. In June 2011 this problem was partly solved, when exome data from the NHLBI GO Exome Sequencing Project (ESP) was published [17].

The aim of this study was to report the prevalence of variants present in the ESP population, previously associated with BrS. In addition, we aimed to establish if multiple in silico tools could distinguish between variants represented in ESP (ESP-derived) and variants not represented in ESP (non-ESP-derived) adding further information to the pathogenicity of these variants.

Materials and methods

In ESP, next-generation sequencing was carried out for all protein-coding regions in 6500 persons from different population studies [17]. Clinical data on the ESP populations were not available even on request. None of these studies specifically included patients with channelopathies or other heart diseases, and at least two studies excluded such patients [18].

In order to find all genes and variants previously associated with BrS, a search in the Human Gene Mutation database (HGMD) was conducted on 1 June 2012, for ‘Brugada syndrome’ [19]. In this way 279 variants in 11 genes were identified. Furthermore, a literature search in the PubMed database was carried out. The following search query was used: {(Brugada) or [Brugada syndrome] or [‘Brugada syndrome’ (Mesh)]} and {[Genetic*] or [‘Genetics’ (Mesh)]} and (mutation). We found 390 articles that matched the search term, and these were systematically examined. In this way, 74 SCN5A variants reported by Kapplinger et al. [11], one SCN5A variant reported by Marangoni et al. [20], as well as KCNJ8 S244L reported by Barajas-Martínez et al. [21] were additionally included. In total, 355 variants in 12 genes, previously associated with BrS, were included in this study (Table 1).

Table 1. Variants previously associated with Brugada syndrome and present in the ESP population
GeneVariantAmino acidEuropean Americans genotypeAfrican Americans genotypeAll genotype
Minor /minorMinor/majorMajor/majorMinor /minorMinor/majorMajor/majorMinor /minorMinor/majorMajor/major
  1. ESP, NHLBI GO Exome Sequencing Project.

CACNA1Cc.1468G>Ap.G490R054170002010056180
c.5510G>Ap.C1837Y014169012014026183
c.5639G>Ap.R1880Q064158011994076152
c.6040G>Ap.V2014I044092011916056008
CACNA2D1c.2126G>Ap.S709N14542490122011466450
c.2751A>Tp.q917H044288002199046487
CACNB2c.428C>Tp.S143F01242880022030126491
c.1195C>Tp.L399F044296012202056498
c.1349C>Tp.T450I01842820122020196484
c.1614C>Ap.D538E034297012202046499
GPD1Lc.839C>Tp.A280V014299002203016502
KCND3c.1798G>Cp.G600R014299002203016502
KCNH2c.2617G>Ap.G873S004300042199046499
SCN1Bbc.641G>Ap.R214Q02122160212390233455
SCN3Bc.29T>Cp.L10P004299012201016500
KCNJ8c.1265C>Tp.S422L01942810122020206483
SCN5Ac.103G>Ap.G35S014131001957016089
c.481G>Ap.E161K014247002139016386
c.647C>Tp.s216L01142050120440126249
c.659C>Tp.T220I044200002016046216
c.694G>Ap.V232I004163031935036098
c.1127G>Ap.R376H014200002046016246
c.1577G>Ap.R526H014191012026026217
c.1844G>Ap.G615E054198002063056261
c.1855C>Tp.L619F024204002056026260
c.1981C>Tp.R661W004298012198016496
c.2150C>Tp.P717L004264012152016416
c.3718G>Cp.E1240Q014299002203016502
.3727G>Ap.D1243N014299002203016502
c.3878T>Cp.F1293S044202002046046248
c.3922C>Tp.L1308F00423901720880176327
c.3956G>Tp.G1319V004237012099016336
c.4573G>Ap.V1525M014184002013016197
c.5494C>Gp.Q1832E004248032133036381
c.5770G>Ap.A1924T014204002112016316
c.5851G>Tp.V1951L05419201920600246252
c.5903T>Gp.I1968S004189012063016252
c.6010T>Cp.F2004L12441590220121266171

The ESP exome data was searched 1 July 2012, for these missense and nonsense variants. Due to lack of data regarding variations positioned in introns and UTR regions in ESP, these were not included in our search. In addition, the literature was searched for data on functional studies and familial co-segregation on variants represented in ESP. Familial co-segregation was defined as at least two genotype positive family members having the same phenotype.

To test if the ESP population harboured an overrepresentation of variants associated with BrS, four common variants in our own healthy control population (n = 536) of Northern European origin were genotyped, as described previously [22]. This control population had no history of arrhythmias or other cardiac diseases. electrocardiograms (ECGs) were available for the entire control population [23, 24]. In those patients harbouring a variant, ECGs were examined by two physicians independently to exclude ST-segments elevations in the right precordial leads (BrS type 1 pattern) as well as incomplete right bundle branch block.

Phylogenetic- and/or physicochemical-based phenotype prediction analyses

In this study, four in silico prediction tools (Grantham values, Polyphen, sift, and Conservation across species) were used to predict if missense variants associated with BrS were ‘benign’ or ‘pathogenic’. The detailed use of these prediction tools has been described in detail by Giudicessi et al. in patients with the LQTS [25].

Grantham physicochemical values were calculated using the Grantham amino acid difference matrix. In this study, values above 100 were considered radical (pathogenic), and values below 100 were considered conservative (benign) [25, 26]. Using Polyphen predictions (version 2.2.2) [27], each variant were labelled ‘probably damaging’, ‘possibly damaging’ or ‘benign’. ‘Probably damaging’ and ‘possibly damaging’ were considered ‘damaging’ (pathogenic). sift predictions (version 4.05) were calculated [28], and variants were classified as ‘tolerated’ (benign) or ‘damaging’ (pathogenic). Finally, the degree of Conservation Across Species was obtained from HGMD [19], and classified as occurring at a position with no substitutions (conserved or pathogenic) or ≥1 substitution (not conserved or benign). Using each in silico tool, we calculated the percentage of variants predicted to be ‘pathogenic’. Any observed difference was tested with the χ2 test for categorical data and a significance level of p < 0.05 was used.

Results

In the ESP population, 38 of 355 (10%) variants previously associated with BrS were identified(Tables 1 and 2). All variants in ESP were missense. When available, functional characterization and familial co-segregation data was recorded (Table 3).

Table 2. Genes and variants associated with Brugada syndrome
GenesNumber of variants% of variants in ESP
  1. ESP; NHLBI GO Exome Sequencing Project.

RANGRF10
KCNE310
SCN5A3257
SCN1B333
CACNA2D1450
KCND3250
KCNH2250
CACNA1C757
CACNB2757
GPD1L1100
SCN3B1100
KCNJ81100
Table 3. Functional data and family co-segregation for genes and variants in the ESP population
GeneAmino acidGenotype frequencyFunctional dataFamily co-segregation
  1. ESP, NHLBI GO Exome Sequencing Project.

CACNA1Cp.G490R5Loss of functionNo
p.C1837Y2No data availableYes
p.R1880Q7No data availableNo data available
p.V2014I5Loss of functionNo
CACNA2D1p.S709N47No data availableNo data available
p.q917H4No data availableNo data available
CACNB2p.S143F12No data availableNo
p.L399F5No data availableYes
p.T450I19No data availableYes
p.D538E4No data availableNo data available
GPD1Lp.A280V1Loss of functionYes
KCND3p.G600R1Gain of functionNo data available
KCNH2p.G873S4Gain of functionNo data available
SCN1Bp.R214Q23Loss of functionNo data available
SCN3Bp.L10P1loss of functionNo
KCNJ8p.S422L20Loss of functionNo data available
SCN5Ap.G35S1No data availableNo
p.E161K1No data availableNo data available
p.s216L12Loss of functionNo
p.T220I4No data availableNo data available
p.V232I3No data availableNo data available
p.R376H1Loss of functionYes
p.R526H2No data availableNo data available
p.G615E5No data availableNo data available
p.L619F2No data availableNo data available
p.R661W1No data availableNo data available
p.P717L1No data availableNo data available
p.E1240Q1No data availableNo data available
p.D1243N1No data availableNo data available
p.F1293S4No data availableNo data available
p.L1308F17No data availableNo data available
p.G1319V1Loss of functionNo data available
p.V1525M1No data availableNo data available
p.Q1832E3No data availableNo data available
p.A1924T1Gain of functionNo data available
p.V1951L24No data availableNo data available
p.I1968S1Loss of functionNo data available
p.F2004L27Loss of functionNo

In ESP there were 272 heterozygote carriers and two homozygote carriers of one of these variants. The genes investigated were screened on average in 6258 individuals, corresponding to a genotype prevalence of 1:23 (274:6258; Table 3). ESP harboured 22 of 303 (7%) variants in SCN5A. In other words, 93% of the variants in SCN5A were not found among nearly 6500 individuals.

When genotyping the four common ESP-derived variants CACNA2D1 S709N, SCN5A F2004L, CACNB2 S143F, and CACNB2 T450I in a healthy Danish control population, 18 heterozygote carriers were found corresponding to a genotype prevalence of 1:30 (18:536). That is comparable with the genotype prevalence of these variants in ESP (1:30 vs 1:61). The mean age of these Danish patients were 63 years (range: 56–71) and none of them had any clinical manifestations of BrS. None of the examined ECGs had ST-segment elevations or incomplete right bundle branch block.

Prediction analyses of ESP-derived vs non-ESP-derived variants

Using the Grantham chemical values, 24% (9/38) of ESP-derived variants were predicted to be pathogenic compared with 32% (74/232) of the non-ESP-derived variants (p = 0.9). The ESP-derived variants were predicted pathogenic in 50% (19/38) of the cases, using Polyphen, compared with 85% (197/232) of the non-ESP-derived variants (p < 0.0001). Finally, using both sift and Conservation Across Species we found that 42% (16/38) and 47% (18/38) of the ESP-derived variants were predicted pathogenic compared with 81% (189/232) and 83% (192/232) of the non-ESP-derived variants, respectively (p < 0.0001).

The frequencies of the ESP and non-ESP-derived variants were also calculated when ≥3 tools predicted pathogenicity. In this synergistic use of the prediction analyses, 47% (15/38) of the ESP-derived and 75% (174/232) of the non-ESP-derived variants were predicted pathogenic (p < 0.0001; Fig. 1). Analyses were performed on all missense mutations found in ESP, but the prediction analyses could not be conducted on 219 SCN5A variants and on one SCN1B variant, as deletions and stop codons were introduced.

Figure 1.

Percentage of variants predicted to be pathogenic, using different predictions analyses on variants present and not present in NHLBI GO Exome Sequencing Project (ESP).

Discussion

In this study, 355 variants associated with BrS were identified. Searching the ESP population, 38 variants distributed on 272 heterozygote carriers and two homozygote carriers were found which corresponds to a genotype prevalence of 1:23 (Table 1). The prevalence of BrS in the general population has been estimated to range between 1:2000 and 1:100,000.

The ESP represents the general population in this study, but we cannot exclude the possibility that there might be an overrepresentation of BrS in ESP, although the prevalence is much higher than expected. To test this, we genotyped four common variants found in ESP in our own healthy control population with available ECGs [23, 24]. It is noteworthy although, that none of these controls had a flecainide or ajmaline test performed. Hence, we cannot exclude that some patients might be asymptomatic carriers although the mean age were >60 years and thus, well above the average age of diagnostic of BrS. In this population, the prevalence of these variants was comparable with that found in ESP (1:30 vs 1:61; Table 4). This is pointing towards that ESP does not harbour an overrepresentation of these variants.

Table 4. The prevalence of four variants in ESP and in the Danish control population
GenesAmino acidDanish control populationESP Population
Found in number of patientsPrevalenceMean age (year)Found I number of patientsPrevalence
  1. ESP, NHLBI GO Exome Sequencing Project.

CACNA2D1p.S709N61:89471:137
CACNB2p.S143F31:178121:540
CACNB2p.T450I61:89191:314
SCN5Ap.F2004L31:178271:237
Overall181:30 (18:536)631051:61 (105:6399)

The ability to distinguish rare pathogenic variants from similarly rare yet non-pathogenic variants has been a challenge for several years. With the increasingly widespread use of genetic testing we might see that more and more rare variants will be established in the disease causation of certain rare diseases [25]. In this regard, large scale exome population data may help permit discrimination between low-frequency genetic variants and disease causing variants. This is especially applicable in low prevalence diseases such as BrS, where it is expected to find only very few cases in the general population. Furthermore, large-scale exome data on the general population might also be needed per ethnic group or geographical region because of the occurrence of such rare variants and their frequency might be region-dependent.

Besides using exome data, in silico tools have been developed to assess the phylogenetic- and/or physicochemical properties of amino acid changes altered by rare single nucleotide variants. Thus, these tools are thought to increase the likelihood of predicting the ‘true’ pathogenicity of certain variants. Recently, Giudicessi et al. published results supporting the potential clinical utility of the synergistic use of these tools. The classification of rare variants, in patients with the LQTS, was enhanced when ≥3 tools predicted pathogenicity [25]. In this study, using ≥3 of these in silico tools it was found that 47% and 75% of the ESP-derived and non-ESP-derived variants were predicted pathogenic, respectively, p < 0.0001. This was the case even though several non-ESP-derived variants in SCN5A were not predicted because stop codons were introduced.

The ESP most likely represents the general population, and even there was clinical data regarding the phenotypic presentation on the cohort; it is not likely that there would be significantly different conclusions considering the reduced penetrance and variable expressivity seen in BrS. Even though the presence of variants in the ESP population might raise questions about disease causation, it does not definitively exclude the possibility of pathogenicity. It is conceivable that some variants present in ESP could in fact be disease causing for instance, the finding of GPD1L A280V that was initially identified by London et al. in 2007. They studied a large family consisting of several members diagnosed with BrS [29]. In this study, it was convincingly demonstrated that the variant A280V affects the intracellular localization of GPD1L as well as it decreases the surface expression of SCN5A and thereby probably decreases the inward sodium current [29]. In other words, even though this variant was present in ESP in one individual, it might still play the pivotal role in the pathogenicity.

Screening for inherited cardiac diseases in family members has become an important tool in the family cascade screening. In a recently published HRS/EHRS consensus document screening is recommended in family members and appropriate relatives following the identification of a BrS-causative mutation in an index case [12]. Furthermore, it is recommended that SCN5A targeted testing can be useful in cases where a cardiologist suspects BrS following a clinical examination. This is only recommended or possible because of a low prevalence (2%) of variants of uncertain significance in controls with a signal to noise ratio of 1 in 10 [12].

The ESP only harboured 22 of 303 (7%) variants in SCN5A previously associated with BrS. That is, 93% of the variants in SCN5A, associated with BrS, were not found among nearly 6500 controls confirming the important role and usefulness of this gene in cascade screening in families. Nevertheless, it is important to keep in mind that the absence of variants in ESP in itself, do not establish disease causation, but certainly strengthen the possibility of one.

Lack of properly sized control populations is a problem when dealing with rare genetic variations in the context of rare monogenetic diseases. Without this we might misdiagnose family members undergoing genetic testing, and this may have great consequences for the treatment, clinical advice and follow-up. We suggest that exome data, like ESP, are used in research and in the everyday clinical practice as a tool alongside with other known prediction tools, to get a better understanding of the pathogenicity of the variants associated with BrS or other rare inherited diseases.

In conclusion, 38 of 355 (10%) variants, associated with BrS, were identified in the ESP population distributed on 272 heterozygote carriers and two homozygote carriers corresponding to a genotype prevalence of 1:23. Importantly, 93% of the variants in SCN5A were not represented among nearly 6500 individuals confirming the pivotal role of this gene in cascade screening in families.

Genotyping four common ESP-derived variants in a healthy Danish control population, 18 heterozygote carriers were identified corresponding to a comparable genotype prevalence with that found in ESP. These results suggests that exome data be used together with other known prediction tools, to better understand the pathogenicity of variants associated with BrS or other rare inherited diseases.

Acknowledgement

The authors would like to thank the NHLBI GO Exome Sequencing Project and its ongoing studies which produced and provided exome variant calls for comparison: the Lung GO Sequencing Project (HL-102923), the WHI Sequencing Project (HL-102924), the Broad GO Sequencing Project (HL-102925), the Seattle GO Sequencing Project (HL-102926) and the Heart GO Sequencing Project (HL-103010).

The work was supported by The Danish Heart Foundation (12-04-R91-A3790-22689), The Danish National Research Foundation Centre for Cardiac Arrhythmia (DARC), The John and Birthe Meyer Foundation, The Research Foundation at the Heart Centre, Rigshospitalet, The foundation of Edith and Henrik Henriksens mindelegat (50892) and The A.P. Møller foundation for the Advancement of Medical Science.

Ancillary