Prevalence and Phenotypic Burden of Monogenic Arrhythmias Using Integration of Electronic Health Records with Genetics

Abstract

Background:

Inherited primary arrhythmia syndromes and arrhythmogenic cardiomyopathies can lead to sudden cardiac arrest in otherwise healthy individuals. The burden and expression of these diseases in a real-world, well-phenotyped cardiovascular population is not well understood.

Methods:

Whole exome sequencing was performed on 8,574 individuals from the CATHGEN cohort. Variants in 55 arrhythmia-related genes (associated with eight disorders) were identified and assessed for pathogenicity based on ACMG/AMP criteria. Individuals carrying pathogenic/likely pathogenic (P/LP) variants were grouped by arrhythmogenic disorder and matched 1:5 to noncarrier controls based on age, sex, and genetic ancestry. Long-term phenotypic data were annotated through deep electronic health record review.

Results:

Fifty-eight P/LP variants were found in 79 individuals in 12 genes associated with five arrhythmogenic disorders (arrhythmogenic right ventricular cardiomyopathy, Brugada syndrome, hypertrophic cardiomyopathy [HCM], LMNA-related cardiomyopathy, and long QT syndrome). The penetrance of these P/LP variants in this cardiovascular cohort was 33%, 0%, 28%, 83%, and 4%, respectively. Carriers of P/LP variants associated with arrhythmogenic disorders showed significant differences in electrocardiogram (ECG), imaging, and clinical phenotypes compared to noncarriers, but displayed no difference in survival. Carriers of novel truncating variants in FLNC, MYBPC3, and MYH7 also developed relevant arrhythmogenic cardiomyopathy phenotypes.

Conclusions:

In a real-world cardiovascular cohort, P/LP variants in arrhythmia-related genes were relatively common (1:108 prevalence) and most penetrant in LMNA. While HCM P/LP variant carriers showed significant differences in clinical outcomes compared to noncarriers, carriers of P/LP variants associated with other arrhythmogenic disorders displayed only ECG differences.

Introduction

Sudden cardiac arrest (SCA) occurs in 76.5 individuals per 100,000 population in the United States with a survival rate of 10.6%.¹ Among cases of sudden cardiac death with negative forensic autopsy, 13% are attributable to pathogenic/likely pathogenic (P/LP) variants in primary arrhythmogenic disorder and cardiomyopathy genes.² Moreover, among survivors of clinically-idiopathic SCA, 22% carry a P/LP variant in a channelopathy or cardiomyopathy gene.³ While individually rare, collectively P/LP variants associated with monogenic cardiovascular diseases (MCVDs) are common and can lead to clinical phenotypes that are underdiagnosed.⁴ Prior studies have integrated electronic health record (EHR) data with population-based genetic sequencing to understand the prevalence and consequences of variants associated with inherited primary arrhythmia syndromes^5–7 and arrhythmogenic cardiomyopathies,^8–12 but these variants have not been studied collectively using guideline-based definitions of pathogenicity. Furthermore, many of these prior studies were unable to perform deep EHR review with longitudinal follow-up for a thorough evaluation of disease expression.^13,14 Thus, we sought to characterize the overall prevalence and phenotypic burden, based on a thorough EHR evaluation, of variants in 55 arrhythmia-related genes in a diverse population of patients referred for cardiac catheterization.

Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request. All participants provided informed consent, and the study was approved by the Duke University Institutional Review Board. Detailed methods are available in the Supplemental Material file.

Results

Prevalence of Variants in Arrhythmia-related Genes

A total of 8,574 individuals in the Catheterization Genetics (CATHGEN) biorepository¹⁵ had exome sequencing passing quality control measures as previously detailed.⁴ Seven hundred seven P/LP variants, variants of uncertain significance (VUS), or novel null/truncating variants in 52 arrhythmia-related genes were identified in 948 individuals (11.1%). Of these, 58 variants were P/LP based on American College of Genetics and Genomics/Association for Molecular Pathology (ACMG/AMP) criteria evaluation, residing in 12 genes associated with five arrhythmogenic disorders (arrhythmogenic right ventricular cardiomyopathy [ARVC], Brugada syndrome [BrS], hypertrophic cardiomyopathy [HCM], LMNA-related cardiomyopathy, and long QT syndrome [LQTS]) in 79 individuals (Figure 1). This included two individuals with P/LP SCN5A variants, which have previously been associated with both BrS and LQTS; these individuals were grouped with both disorders for phenotypic and survival analyses.¹⁶ HCM had the greatest number of associated P/LP variants with 20 variants in 30 individuals; followed by LQTS with 16 variants in 24 individuals; ARVC with 15 variants in 18 individuals; LMNA-related cardiomyopathy with five variants in six individuals; and BrS with four variants in four individuals (Figure 2). In addition, 21 novel null/truncating variants in 13 arrhythmia-related genes were identified in 21 individuals. Novel null/truncating variants in genes predisposing to “other” arrhythmogenic cardiomyopathies (i.e., FLNC, CTNNA3, and ABCC9) were most common with eight variants in seven individuals; followed by BrS genes with four variants in five individuals; HCM genes with three variants in three individuals; and catecholaminergic polymorphic ventricular tachycardia (CPVT), LQTS, and ARVC genes, each with two variants in two individuals (Supplemental Figure I).

Figure 1.

Overview of Participant Identification and Variant Prevalence. Shaded boxes indicate groups that were compared for population characteristics and survival analysis. MCVD, monogenic cardiovascular disease; P/LP, pathogenic/likely pathogenic; QC, quality control; VUS, variant of uncertain significance; WES, whole exome sequencing.

Figure 2.

Distribution of Pathogenic/Likely Pathogenic Variants in Arrhythmia-related Genes in CATHGEN. Bar plot showing the number of pathogenic/likely pathogenic variants in arrhythmia-related genes, grouped by associated arrhythmogenic disorder. ARVC, arrhythmogenic right ventricular cardiomyopathy; BrS, Brugada syndrome; HCM, hypertrophic cardiomyopathy; LQTS, long QT syndrome.

An additional 629 VUS in 52 arrhythmia-related genes were identified in 869 individuals, nine of whom also harbored P/LP variants and two of whom also harbored novel null/truncating variants (Figure 1). Supplemental Figure II displays the total prevalence of P/LP variants, novel null/truncating variants, and VUS in arrhythmia-related genes.

Phenotypic Consequences of Pathogenic/Likely Pathogenic Variants

Structured and unstructured EHR-based phenotyping was conducted. Demographic and clinical characteristics at the time of CATHGEN enrollment for all P/LP variant carriers, VUS carriers, and variant-negative individuals (i.e., those not carrying P/LP variants in any MCVD gene as previously defined⁴ [Supplemental Table I]) are presented in Table 1. The median (interquartile range) length of follow-up among all variant carriers and matched noncarriers was 15 (9–25) years and 15 (7–23) years, respectively (p=0.5). Ninety-four percent of variant carriers and 95% of matched noncarriers had at least one Duke University Health System encounter documented in the EHR beyond their encounter related to CATHGEN enrollment (p=0.3). Supplemental Table II displays an individual-level list of P/LP variant carriers with their genotype and relevant electrocardiographic, imaging, and clinical phenotypes.

Table 1.

CATHGEN Population Characteristics by Arrhythmia-related Gene Variant Carrier Status

	P/LP (N=79)	VUS (N=826)	Variant-negative (N=7,348)
Demographics
Age, years*†	57.3 ± 9.8	60.2 ± 12.5	60.9 ± 12.0
Female	31 (39%)	328 (40%)	2,775 (38%)
Genetic ancestry‡
European	61 (77%)	583 (71%)	5,652 (77%)
African	18 (23%)	218 (26%)	1,563 (21%)
Admixed American	0	12 (1%)	38 (1%)
South Asian	0	2 (<1%)	35 (<1%)
East Asian	0	2 (<1%)	19 (<1%)
Unknown	0	9 (1%)	41 (1%)
Clinical characteristics
Body Mass Index, kg/m2	29 ± 6	30 ± 7	30 ± 7
Diabetes*†	15 (19%)	252 (31%)	2,065 (28%)
Hypertension	48 (61%)	564 (68%)	4,953 (67%)
Hyperlipidemia*†	34 (43%)	487 (59%)	4,384 (60%)
Smoker	35 (44%)	374 (45%)	3,521 (48%)
Coronary artery disease*†	31 (39%)	490 (59%)	4,578 (62%)
Heart failure*†	33 (42%)	224 (27%)	1,987 (27%)
Left ventricular ejection fraction, %	54 ± 15	55 ± 14	55 ± 14
Atrial fibrillation*†	22 (28%)	121 (15%)	1,169 (16%)
Indication for catheterization*†
Ischemic heart disease	36 (46%)	577 (70%)	5,141 (70%)
Valvular heart disease	1 (1%)	24 (3%)	193 (3%)
Congenital heart disease	1 (1%)	0	28 (<1%)
Other	41 (52%)	225 (27%)	1,986 (27%)

Continuous measures are described as mean ± standard deviation and are compared using the two-sided Student’s t-test. Categorical variables are described as N (%) and are compared using Fisher’s exact test. P/LP, pathogenic/likely pathogenic; VUS, variant of uncertain significance.

^*p <0.05 for comparison of P/LP to VUS carriers

^†p <0.05 for comparison of P/LP carriers to variant-negative individuals

^‡p <0.05 for comparison of VUS carriers to variant-negative individuals

Among 18 individuals carrying P/LP variants associated with ARVC, six (33%) met the 2010 Task Force criteria¹⁷ for definite ARVC based on EHR review. Among these six individuals, one had previously been diagnosed with ARVC, while the other five had not. None had previously undergone genetic testing. As expected, compared with matched variant-negative controls, variant carriers were significantly more likely to meet Task Force major electrocardiogram (ECG) criteria (p=0.04) and had significantly different Task Force diagnostic categorizations as a whole (p=2×10⁻¹⁶). However, there was no significant difference between variant carriers and noncarriers with respect to imaging phenotypes or clinical outcomes, such as nonischemic cardiomyopathy, syncope, atrial or ventricular arrhythmias, or implantable cardioverter-defibrillator (ICD) implantation. Thus, P/LP variants associated with ARVC were associated with differences in ECG characteristics, but not differences in clinical outcomes (Table 2).

Table 2.

Phenotypes in Carriers of Pathogenic/Likely Pathogenic Variants Associated with Arrhythmogenic Right Ventricular Cardiomyopathy

	Carriers (N=18)	Noncarriers (N=90)	P value
Demographics
Age, years	69.4 ± 5.7	69.4 ± 5.6	NA
Female	10 (56%)	50 (56%)
Genetic ancestry
European	9 (50%)	45 (50%)
African	9 (50%)	45 (50%)
Imaging
Individuals with TTE or CMR	16 (89%)	76 (84%)	1
TF major imaging criteria	0	0	NA
TF minor imaging criteria	0	0	NA
Electrocardiogram
Individuals with electrocardiogram	15 (83%)	84 (93%)	0.2
TF major electrocardiogram criteria	3 (20%)	3 (4%)	0.04
TF minor electrocardiogram criteria	6 (40%)	24 (29%)	0.4
Arrhythmias
Individuals with Holter monitor	4 (22%)	11 (12%)	0.3
>500 PVCs in 24 hours	2 (50%)	3 (27%)	0.6
TF major NSVT/VT criteria	0	0	NA
TF minor NSVT/VT criteria	7 (39%)	16 (18%)	0.06
Pathology
Individuals with right ventricular tissue sample	2 (11%)	6 (7%)	0.6
TF major pathology criteria	1 (50%)	0	0.3
TF minor pathology criteria	0	0	NA
Family history
TF major criteria (excluding G+)	0	0	NA
TF minor criteria	1 (6%)	0	0.2
TF diagnostic categories
None	0	82 (91%)	2×10 −16 *
Possible	7 (39%)	8 (9%)
Borderline	5 (28%)	0
Definite	6 (33%)	0
Other findings
Any right ventricular dysfunction	4 (25%)	9 (10%)	0.2
Nonischemic cardiomyopathy	4 (22%)	11 (12%)	0.3
Syncope	3 (17%)	10 (11%)	0.5
Atrial fibrillation/flutter	8 (44%)	29 (32%)	0.4
Implantable cardioverter-defibrillator	4 (22%)	12 (13%)	0.5
Obstructive coronary artery disease	8 (44%)	57 (63%)	0.2

Continuous measures are described as mean ± standard deviation. Categorical variables are described as N (%); p values are calculated using Fisher’s exact test for dichotomous variables and Wilcoxon rank-sum test for ordinal variables (indicated by *). P values in bold indicate p <0.05. CMR, cardiac magnetic resonance imaging; G+, genotype positivity; NA, not applicable; NSVT, nonsustained ventricular tachycardia; PVC, premature ventricular contraction; TF, 2010 Task Force; TTE, transthoracic echocardiogram; VT, ventricular tachycardia.

Among four individuals carrying P/LP variants associated with BrS, none had Type 1 Brugada pattern on any available ECG (ranging from 0 to 54 ECGs per person over 0 to 19 years of follow-up), and none were diagnosed with BrS clinically. Compared with matched variant-negative controls, carriers of P/LP variants had significantly longer PR intervals (p=0.009) and were significantly more likely to have first-degree atrioventricular block plus left axis deviation of the QRS (p=0.03), a finding associated with BrS.¹⁸ There was no significant difference between variant carriers and noncarriers with respect to QRS duration or clinical outcomes, including syncope, atrial or ventricular arrhythmias, or pacemaker or ICD implantation. Thus, P/LP variants in BrS genes correlated with conduction delay but not Brugada pattern on ECG or differences in clinical outcomes (Table 3).

Table 3.

Phenotypes in Carriers of Pathogenic/Likely Pathogenic Variants Associated with Brugada Syndrome

	Carriers (N=4)	Noncarriers (N=20)	P value
Demographics
Age, years	66.5 ± 15.5	66.6 ± 13.7	NA
Female	3 (75%)	15 (75%)
Genetic ancestry
European	3 (75%)	15 (75%)
African	1 (25%)	5 (25%)
Electrocardiogram
Individuals with electrocardiogram	3 (75%)	20 (100%)	0.2
Type 1 Brugada pattern	0	0	NA
Other findings
PR interval, ms	258 ± 69	162 ± 21	0.009
1st degree atrioventricular block + left axis deviation	2 (67%)	1 (5%)	0.03
QRS duration, ms	113 ± 3	97 ± 19	0.06
Syncope	2 (50%)	4 (20%)	0.3
Atrial fibrillation/flutter	2 (50%)	6 (30%)	0.6
Nonsustained/sustained ventricular tachycardia	0	3 (15%)	1
Permanent pacemaker	2 (50%)	2 (10%)	0.1
Implantable cardioverter-defibrillator	0	1 (5%)	1
Obstructive coronary artery disease	1 (25%)	12 (60%)	0.3

Continuous measures are described as mean ± standard deviation and are compared using the Wilcoxon rank-sum test. Categorical variables are described as N (%) and are compared using Fisher’s exact test. P values in bold indicate p <0.05. NA, not applicable.

Among 30 individuals carrying P/LP variants associated with HCM, eight (28%) developed unexplained left ventricular hypertrophy (LVH) ≥1.3 cm, consistent with a diagnosis of HCM.¹⁹ Adding individuals with secondary LVH ≥1.8 cm²⁰ resulted in 19 individuals (66%) meeting diagnostic criteria for HCM. All but one individual with unexplained LVH ≥1.3 cm had previously been diagnosed with HCM. None had previously undergone genetic testing. As expected, compared with matched variant-negative controls, variant carriers had significantly greater maximal left ventricular wall thickness (p=1×10⁻⁶) and left atrial diameter (p=0.005), and they more frequently had significant left ventricular outflow tract (LVOT) obstruction (p=0.0001) and systolic anterior motion of the mitral valve leaflets (SAM) (p=0.001). Furthermore, variant carriers more frequently had evidence of LVH (p=0.006), T wave inversions (p=7×10⁻⁵), and intraventricular conduction delay (IVCD) (p=0.03) on ECG compared to noncarriers. With regards to outcomes, variant carriers had more nonsustained ventricular tachycardia (NSVT) (p=0.003), syncope (p=2 ×10⁻⁵), heart failure (p=0.009), ICD implantation (p=0.02), septal reduction therapy (p=1×10⁻⁵), and heart transplantation (p=0.007) compared with variant-negative controls. There was no difference between variant carriers and noncarriers with respect to atrial fibrillation or flutter (AF/AFL), sustained ventricular tachycardia (VT), or SCA. Thus, P/LP variants associated with HCM were associated with differences in imaging, ECG characteristics, and clinical outcomes (Table 4).

Table 4.

Phenotypes in Carriers of Pathogenic/Likely Pathogenic Variants Associated with Hypertrophic Cardiomyopathy

	Carriers (N=30)	Noncarriers (N=150)	P value
Demographics
Age, years	71.2 ± 10.0	71.2 ± 9.9	NA
Female	9 (30%)	45 (30%)
Genetic ancestry
European	29 (97%)	145 (97%)
African	1 (3%)	1 (3%)
Imaging
Individuals with TTE or CMR	29 (97%)	123 (82%)	0.05
Unexplained LVH ≥1.5 cm	6 (21%)	0	3×10 −5
Unexplained LVH ≥1.3 cm	8 (28%)	0	7×10 −7
Unexplained LVH ≥1.3 cm or secondary LVH ≥1.8 cm	19 (66%)	14 (11%)	8×10 −9
Peak LVOT gradient ≥30 mmHg	7 (24%)	1 (1%)	0.0001
Systolic anterior motion of mitral valve leaflets	4 (14%)	0	0.001
Mitral regurgitation	9 (31%)	20 (16%)	0.1
Diastolic dysfunction	21 (72%)	91 (76%)	0.8
Individuals with CMR	3 (10%)	22 (15%)	0.8
Late gadolinium enhancement	3 (100%)	15 (68%)	0.5
Maximal left ventricular wall thickness, cm	1.8 ± 0.6	1.4 ± 0.3	1×10 −6
Left atrial diameter, cm	5.0 ± 0.8	4.5 ± 0.8	0.005
Left ventricular ejection fraction, %	43 ± 14	49 ± 13	0.03
Electrocardiogram
Individuals with electrocardiogram	26 (87%)	142 (95%)	0.1
Left ventricular hypertrophy	6 (23%)	7 (5%)	0.006
T wave inversions	15 (58%)	26 (18%)	7×10 −5
Left bundle branch block	4 (15%)	8 (6%)	0.09
Right bundle branch block	1 (4%)	24 (17%)	0.1
Intraventricular conduction delay	4 (15%)	5 (4%)	0.03
QRS interval, ms	106 ± 31	104 ± 25	0.8
Arrhythmias
Atrial fibrillation/flutter	17 (57%)	57 (38%)	0.07
Nonsustained ventricular tachycardia	11 (37%)	18 (12%)	0.002
Sustained ventricular tachycardia or VF	6 (20%)	17 (11%)	0.2
Sudden cardiac arrest	4 (13%)	6 (4%)	0.06
Other findings
Family history of sudden cardiac arrest	4 (13%)	1 (1%)	0.003
Syncope	13 (43%)	13 (9%)	2×10 −5
Heart failure	22 (73%)	68 (45%)	0.009
Implantable cardioverter-defibrillator	9 (30%)	17 (11%)	0.02
Septal reduction therapy	6 (20%)	0	1×10 −5
Alcohol septal ablation	2 (33%)	NA	NA
Surgical myectomy	4 (67%)
Heart transplantation	5 (17%)	4 (3%)	0.007
Individuals with histopathology	6 (20%)	2 (1%)	0.0003
Findings consistent with HCM	5 (83%)	0	0.1
Obstructive coronary artery disease	14 (47%)	100 (67%)	0.06

Continuous measures are described as mean ± standard deviation and are compared using the two-sided Student’s t-test. Categorical variables are described as N (%) and are compared using Fisher’s exact test. P values in bold indicate p <0.05. CMR, cardiac magnetic resonance imaging; HCM, hypertrophic cardiomyopathy; LVH, left ventricular hypertrophy; LVOT, left ventricular outflow tract; NA, not applicable; TTE, echocardiogram; VF, ventricular fibrillation.

Among six individuals carrying P/LP variants associated with LMNA-related cardiomyopathy, five (83%) developed nonischemic cardiomyopathy based on EHR review (all prior to enrollment in CATHGEN). None had previously undergone genetic testing, so a specific diagnosis of LMNA-related cardiomyopathy had not been made in any of these individuals. As expected, compared with matched variant-negative controls, variant carriers had significantly lower left ventricular ejection fraction (p=0.003). Variant carriers also had significantly more symptomatic bradyarrhythmia (p=0.02), AF/AFL (p=0.01), sustained VT (p=0.002), pacemaker implantation (p=0.0009), ICD implantation (p=0.02), and heart transplantation (p=0.0003) compared with variant-negative controls. There was no significant difference between variant carriers and noncarriers with respect to syncope. Thus, P/LP variants associated with LMNA-related cardiomyopathy were highly penetrant and correlated strongly with clinical outcomes (Table 5).

Table 5.

Phenotypes in Carriers of Pathogenic/Likely Pathogenic Variants Associated with LMNA-related Cardiomyopathy

	Carriers (N=6)	Noncarriers (N=30)	P value
Demographics
Age, years	59.9 ± 10.6	59.9 ± 9.8	NA
Female	3 (50%)	15 (50%)
Genetic ancestry
European	3 (50%)	15 (50%)
African	3 (50%)	15 (50%)
Imaging
Individuals with TTE or CMR	6 (100%)	25 (83%)	0.6
Left ventricular ejection fraction, %	19 ± 8	41 ± 16	0.003
Left ventricular end-diastolic diameter, cm	5.7 ± 1.0	5.4 ± 1.1	0.6
Left atrial diameter, cm	4.8 ± 1.4	4.1 ± 0.9	0.4
Nonischemic cardiomyopathy	5 (83%)	9 (30%)	0.02
Arrhythmias
Symptomatic bradyarrhythmia	3 (50%)	2 (7%)	0.02
Atrial fibrillation/flutter	4 (67%)	4 (13%)	0.01
Nonsustained ventricular tachycardia	5 (83%)	4 (13%)	0.002
Sustained ventricular tachycardia	4 (67%)	2 (7%)	0.003
Other findings
Family history of heart failure	6 (100%)	4 (13%)	0.0001
Syncope	1 (17%)	6 (20%)	1
Permanent pacemaker	6 (100%)	7 (23%)	0.0009
Implantable cardioverter-defibrillator	4 (67%)	5 (17%)	0.02
Heart transplantation	5 (83%)	2 (7%)	0.0003
Minimum eGFR, ml/min/1.73 m2	28 ± 23	47 ± 26	0.1
Obstructive coronary artery disease	1 (17%)	16 (53%)	0.2

Continuous measures are described as mean ± standard deviation and are compared using the Wilcoxon rank-sum test. Categorical variables are described as N (%) and are compared using Fisher’s exact test. P values in bold indicate p <0.05. CMR, cardiac magnetic resonance imaging; eGFR, estimated glomerular filtration rate; NA, not applicable; TTE, echocardiogram.

Among 23 individuals carrying P/LP variants associated with LQTS, one (4%) met diagnostic criteria for LQTS based on the presence of a QT interval corrected for heart rate using Bazett’s formula (QTc) ≥500 ms in repeated ECGs.²¹ None had a LQTS risk score²² ≥3.5 based on available EHR review. One individual carried a prior diagnosis of LQTS made outside of the Duke University Health System with genetic testing prior to enrollment in CATHGEN; their LQTS risk score was 3 based on their QTc and family history. As expected, compared with matched variant-negative controls, variant carriers had a significantly longer QTc (p=0.005) and significantly different LQTS risk score diagnostic probabilities as a whole (p=0.01); however, there was no significant difference between variant carriers and noncarriers with respect to syncope, atrial or ventricular arrhythmias, or ICD implantation. Thus, P/LP variants in LQTS genes correlated with prolonged QTc, but not differences in clinical outcomes (Table 6).

Table 6.

Phenotypes in Carriers of Pathogenic/Likely Pathogenic Variants Associated with Long QT Syndrome

	Carriers (N=23)	Noncarriers (N=115)	P value
Demographics
Age, years	69.2 ± 11.1	69.2 ± 10.9	NA
Female	8 (35%)	40 (35%)
Genetic ancestry
European	18 (78%)	90 (78%)
African	5 (22%)	25 (22%)
Long QT syndrome risk score electrocardiogram categories
Individuals with electrocardiogram	18 (78%)	86 (75%)	1
QTc ≥480 ms	7 (39%)	9 (10%)	0.01 *
QTc = 460–479 ms	2 (11%)	12 (14%)
QTc = 450–459 ms in males	1 (6%)	5 (6%)
QTc <450 ms in males or <460 ms in females	8 (44%)	60 (70%)
Long QT syndrome risk score electrocardiogram and clinical criteria
Syncope	2 (9%)	18 (16%)	0.5
Syncope with stress	0	0	NA
Congenital deafness	0	0	NA
Torsades de pointes	0	0	NA
Long QT syndrome risk score family history criteria
Family history of long QT syndrome	1 (4%)	0	0.2
Family history of unexplained sudden cardiac death	0	0	NA
Long QT syndrome risk score diagnostic probability
Low probability	14 (61%)	93 (81%)	0.04 *
Intermediate probability	9 (39%)	21 (18%)
High probability	0	1 (1%)
Other findings
QTc, ms	465 ± 39	439 ± 34	0.005
QTc ≥500 ms in repeated electrocardiograms	1 (4%)	0	0.2
Atrial fibrillation/flutter	7 (30%)	44 (38%)	0.6
Nonsustained/sustained ventricular tachycardia	2 (9%)	28 (24%)	0.2
Implantable cardioverter-defibrillator	2 (9%)	14 (12%)	1
Obstructive coronary artery disease	10 (43%)	76 (66%)	0.06

Continuous measures are described as mean ± standard deviation and are compared using two-sided Student’s t-test. Categorical variables are described as N (%); p values are calculated using Fisher’s exact test for dichotomous variables and Wilcoxon rank-sum test for ordinal variables (indicated by *). P values in bold indicate p <0.05. NA, not applicable; QTc, QT interval corrected for heart rate using Bazett’s formula.

The number of P/LP variant carriers who met diagnostic criteria for an arrhythmogenic disorder but had not been previously diagnosed with the relevant disorder was five (83%) for ARVC, zero for BrS, one (13%) for HCM, five (100%) for LMNA-related cardiomyopathy, one (100%) for LQTS, and 12 (60%) in total among all disorders (Supplemental Table III).

Given the nature of the study population, we assessed for differences in the rate of obstructive coronary artery disease between P/LP variant carriers and matched noncarriers within each arrhythmogenic disorder, but no significant difference was observed (Tables 2–6).

Phenotypic Consequences of Variants of Uncertain Significance

To determine potential phenotypic expression in VUS carriers, phenotypes were derived from International Classification of Diseases and Current Procedural Terminology codes (Supplemental Table IV) from EHR data and were compared between VUS carriers grouped by associated arrhythmogenic disorder and variant-negative individuals. There was no significant association between VUS carrier status and any phenotype for any arrhythmogenic disorder (Supplemental Table V).

Nine individuals carried both a P/LP variant and a VUS (predisposing to BrS, CPVT, ARVC, and HCM), and two individuals carried both a novel null/truncating variant and a VUS (predisposing to BrS and HCM). None of these 11 individuals met diagnostic criteria for the arrhythmogenic disorder relevant to the VUS.

Survival by Genotype Status

A time-to-event analysis for all-cause mortality was conducted comparing individuals carrying P/LP variants in arrhythmia-related genes, individuals with VUS in arrhythmia-related genes, and variant-negative individuals. Figure 1 shows an overview of participants identified for survival analysis. Overall, 4,076 individuals (49%) died at a mean age ± standard deviation of 72 ± 12 years. A total of 31 P/LP variant carriers (39%) died compared to 389 VUS carriers (47%) and 3,656 variant-negative individuals (50%). In survival analyses, there was no significant difference in time-to-death between each of these groups (Figure 3). Furthermore, there was no significant difference in survival by genotype status in analyses of ARVC, BrS, HCM, LMNA-related cardiomyopathy, or LQTS separately (Supplemental Figure III).

Figure 3.

Survival Analysis of Arrhythmia-related Gene Variants. Kaplan-Meier curves showing no significant difference in survival between groups stratified by carrier status of arrhythmia-related gene variants. P value was calculated using the log-rank test. P/LP, pathogenic/likely pathogenic; VUS, variant of uncertain significance.

Novel Null/Truncating Variants in Arrhythmia-related Genes

Novel variants are those that have never previously been identified. Since they shorten the protein-coding sequence of the gene, null/truncating variants are predicted to be deleterious. Twenty-one novel null/truncating variants were identified in 13 genes (predisposing to BrS, CPVT, LQTS, ARVC, HCM, and other arrhythmogenic cardiomyopathies) in 21 individuals (Supplemental Table VI).

Among four individuals heterozygous for novel null/truncating variants in FLNC (which predisposes to arrhythmogenic cardiomyopathy²³), three (75%) developed nonischemic cardiomyopathy, and two (50%) also had a history of sustained VT and underwent heart transplantation. Among two individuals heterozygous for novel truncating variants in MYH7, one (50%) developed obstructive HCM; and one individual heterozygous for a novel truncating variant in MYBPC3 developed obstructive HCM, had ventricular fibrillation, and underwent heart transplantation. None of the individuals carrying novel null/truncating variants in genes predisposing to BrS, CPVT, LQTS, or ARVC met diagnostic criteria^18,21 for their respective diseases.

Discussion

Using a large cardiovascular cohort, we have integrated deep phenotyping through EHR with genetic sequencing using ACMG/AMP variant classification to comprehensively study the prevalence and penetrance of variants in 55 arrhythmia-related genes (Table 7). To our knowledge, this analysis represents the largest number of such genes studied in this manner in a single cohort to date. This enabled us to make the following important observations: carriage of P/LP variants in arrhythmia-related genes is relatively common with a 1:108 prevalence; penetrance based on EHR review is variable, ranging from 0%-83%, and greatest in P/LP LMNA variants; HCM P/LP variants are associated with differences in clinical outcomes, while ARVC, BrS, and LQTS P/LP variants are only associated with ECG differences in carriers compared to matched noncarriers; the rate of missed diagnosis of Mendelian arrhythmogenic disorders among P/LP variant carriers is relatively high; VUS carrier status in arrhythmia-related genes is not associated with differences in EHR code-derived phenotypes; there is no survival difference between carriers of P/LP variants in arrhythmia-related genes, VUS carriers, and noncarriers; and incidentally discovered novel truncating variants in FLNC, MYBPC3, and MYH7 may lead to their associated arrhythmogenic disorders.

Table 7.

Arrhythmia-related Genes Studied

Inherited Primary Arrhythmia Syndromes
Diseases	Associated Genes
Brugada Syndrome	ABCC9*, CACNA1C*, CACNA2D1, CACNB2, GPD1L, HCN4, KCND3, KCNE3, KCNE5, KCNJ8, PKP2*, RANGRF, SCN1B, SCN2B, SCN3B, SCN5A*†, SCN10A
Catecholaminergic Polymorphic Ventricular Tachycardia	CASQ2, RYR2*†, TRDN*†
Familial Atrial Fibrillation	ABCC9*, KCNA5, KCNH2*, KCNJ2*, KCNQ1*, LMNA*, RYR2*, SCN5A*
Long QT Syndrome	AKAP9, ANK2, CACNA1C*†, CALM1, CALM2, CALM3, CAV3, KCNE1, KCNE2, KCNH2*†, KCNJ2*†, KCNJ5, KCNQ1*†, SCN4B, SCN5A*, SNTA1, TRDN*
Short QT Syndrome	CACNA1C*, KCNH2*, KCNJ2*, KCNQ1*
Arrhythmogenic Cardiomyopathies
Diseases	Associated Genes
Arrhythmogenic Right Ventricular Cardiomyopathy	DSC2, DSG2, DSP*†, JUP, PKP2*†, RYR2*, SCN5A*
Hypertrophic Cardiomyopathy	ACTC1, MYBPC3, MYH7, MYL2, MYL3, TNNI3, TNNT2, TPM1
Other Arrhythmogenic Cardiomyopathies	ABCC9*†, CTNNA3, DES, DSP*, FLNC, KCNH2*, KCNQ1*, LDB3, LMNA*†, PLN, SCN5A*, TMEM43, TRPM4

^*Genes associated with >1 disease

^†Most common disease association for genes associated with >1 disease

Our study provides the first assessment of the cardiovascular population prevalence of P/LP variants in arrhythmia-related genes across all arrhythmogenic disorders. The estimated prevalence for each arrhythmogenic disorder in CATHGEN is comparable to prior population-based assessments of genotype positivity for ARVC, BrS, HCM, LMNA-related cardiomyopathy, and LQTS in previously published cohorts^{6,9,10,12–14} after ACMG/AMP classification is applied to the variants identified in those cohorts (Supplemental Table VII).

As expected, there was marked heterogeneity in the expression of disease in individuals carrying P/LP variants. Interestingly, the penetrance of P/LP variants in arrhythmia-related genes in CATHGEN and prior population-based work^5–14 is consistently lower than that observed in registry-based cohorts involving probands and their family members (i.e., family-based penetrance) (Supplemental Table VIII).^24–28 The same reduction in population penetrance relative to family-based penetrance is seen in other monogenic diseases,^29,30 which may be due to secondary genetic factors and/or epigenetic influences that are enriched within families. Nevertheless, an understanding of penetrance in broader populations such as ours is important and provides complementary information to that assessed in family-based studies.

To assess for more subtle phenotypes in the absence of clear diagnostic criteria, we leveraged extensive, longitudinal EHR data to provide a comprehensive understanding of the phenotypic consequences of P/LP variants in arrhythmia-related genes. Not surprisingly, P/LP LMNA variants conferred the greatest phenotypic effects, consistent with the known high penetrance of LMNA-related cardiomyopathy. We are the first to report that incidentally discovered HCM P/LP variants are associated with significant LVOT obstruction and SAM, as well as evidence of LVH, T wave inversions, and IVCD on ECG in carriers compared to noncarriers in a cardiovascular population.

Consistent with prior population-based work utilizing contemporary variant classification schemes,¹⁴ LQTS P/LP variants were associated with prolonged QTc duration, which is in contrast to findings from population-based studies that did not use ACMG/AMP criteria to define P/LP variants.^6,7 Incidentally detected ARVC variants in the general population have previously been shown to have low pathogenicity,^10,11 which we have largely corroborated in our study. As in prior population-based studies of BrS variants,^5,7,14 we saw no ECGs with Type 1 Brugada pattern among individuals with BrS P/LP variants. This is unsurprising given that the majority of patients with confirmed BrS do not present with spontaneous Brugada pattern on their ECGs.³¹ We did not observe an increased rate of syncope in carriers of P/LP SCN5A variants compared to noncarriers, in contrast to the work of Glazer et al. in the eMERGE-III cohort.¹⁴ This may be attributed to their larger sample size or their use of EHR codes to phenotype P/LP variant carriers and noncarriers in contrast to our use of manual chart review.

A unique aspect of our study is the long-term follow-up of individuals enrolled in CATHGEN, which allowed for the assessment of survival differences. Despite the phenotypic differences noted above, we found no difference in survival in disease-aggregated analyses, nor by individual arrhythmogenic disorder. Possible explanations for this counterintuitive finding include inadequate sample size, particularly of LMNA variants, which carried the highest penetrance, and a greater contribution of acquired diseases to mortality in this cardiovascular cohort. Haggerty et al. did observe an increased all-cause mortality in carriers of putative loss-of-function variants in ARVC genes in the DiscovEHR cohort, but four-out-of-five observed deaths in the ARVC variant carrier group from their study were non-cardiac related.¹¹

In this cardiovascular population, incidentally, discovered novel null/truncating variants in FLNC and MYBPC3 resulted in arrhythmogenic cardiomyopathy and HCM, respectively, consistent with their effects in family-based studies.^23,28 One individual with a novel truncating variant in MYH7 developed obstructive HCM. While rare truncating variants in MYH7 have previously been identified in individuals with HCM, their case frequency is not disproportionately higher than that of the general population, leading these variants to be classified as not pathogenic.³² Segregation and functional studies are needed to establish the pathogenicity of the novel truncating variants associated with arrhythmogenic disorders discovered in the CATHGEN cohort.

There are several key limitations to our study. Because CATHGEN does not include children and is skewed toward an older population (average age of enrollment was 61 years old), we were unlikely to capture individuals with highly penetrant variants in arrhythmia-related genes due to survival bias, which confounds many population-based genetic studies.³³ Simultaneously, our analysis was biased toward more cardiovascular events since the CATHGEN cohort was recruited from individuals referred for cardiac catheterization. For similar reasons, we were also biased toward a higher prevalence of arrhythmia-related gene variants. Finally, phenotypic assessments were limited by the availability of clinical data in our EHR and length of follow-up. Regardless, our findings should be validated in additional cohorts with larger sample sizes and younger participants.

In summary, among individuals referred for cardiac catheterization, P/LP variants associated with monogenic arrhythmogenic disorders are common. The integration of EHR data and genetics improves our understanding of risk associated with variants in arrhythmia-related genes identified in a real-world cardiovascular cohort. Many questions remain regarding the clinical significance of these variants and the best approach for risk stratification of variant carriers. Future arrhythmogenic population genetics studies should seek to include younger participants, since disease manifestations can occur at a young age and SCA is more often due to inherited diseases in young patients. Large, diverse cohorts with detailed phenotypic data, as well as resources to conduct time-intensive chart extraction, are needed to draw meaningful conclusions from population genetics studies.³⁴ Finally, high-throughput tools for variant interpretation³⁵ and patient-specific, disease-relevant functional assays^36,37 to risk stratify carriers of P/LP variants in arrhythmia-related genes will enable genome-first approaches to Mendelian arrhythmogenic disorders to deliver on the heralded promise of personalized medicine for these patients.

Nonstandard Abbreviations and Acronyms

ACMG/AMP

American College of Genetics and Genomics/Association for Molecular Pathology

AF/AFL

atrial fibrillation or flutter

ARVC

arrhythmogenic right ventricular cardiomyopathy

BrS

Brugada syndrome

CATHGEN

Catheterization Genetics

CPVT

catecholaminergic polymorphic ventricular tachycardia

EHR

electronic health record

HCM

hypertrophic cardiomyopathy

ICD

implantable cardioverter-defibrillator

LQTS

long QT syndrome

LVH

left ventricular hypertrophy

LVOT

left ventricular outflow tract

MCVD

monogenic cardiovascular disease

NSVT

nonsustained ventricular tachycardia

P/LP

pathogenic/likely pathogenic

QTc

QT interval corrected for heart rate using Bazett’s formula

SAM

systolic anterior motion of the mitral valve leaflets

SCA

sudden cardiac arrest

VT

ventricular tachycardia

VUS

variant of uncertain significance