Prevalence and Electronic Health Record-based Phenotype of Loss-of-function Genetic Variants in Arrhythmogenic Right Ventricular Cardiomyopathy-associated Genes

Abstract

Background

Arrhythmogenic right ventricular cardiomyopathy (ARVC) is associated with variants in desmosome genes. Secondary findings of pathogenic/likely pathogenic variants, primarily loss-of-function (LOF) variants, are recommended for clinical reporting; however, their prevalence and associated phenotype in a general clinical population are not fully characterized.

Methods

From whole exome sequencing of 61,019 individuals in the DiscovEHR cohort, we screened for putative LOF variants in PKP2, DSC2, DSG2, and DSP. We evaluated measures from prior clinical electrocardiograms (ECG) and echocardiograms, manually over-read to evaluate ARVC diagnostic criteria, and performed a phenome-wide association study (PheWAS). Finally, we estimated expected penetrance using Bayesian inference.

Results

140 individuals (0.23%; 59±18 years old at last encounter; 33% male) had an ARVC variant (G⁺). None had an existing diagnosis of ARVC in the electronic health record (EHR), nor significant differences in prior ECG or echocardiogram findings compared with matched controls without variants. Several G⁺ individuals satisfied major repolarization (n=4) and ventricular function (n=5) criteria, but this prevalence matched controls. PheWAS showed no significant associations of other heart disease diagnoses. Combining our best genetic and disease prevalence estimates yields an estimated penetrance of 6.0%.

Conclusions

The prevalence of ARVC LOF variants is approximately 1:435 in a general clinical population of predominantly European descent, but with limited EHR-based evidence of phenotypic association in our population, consistent with a low penetrance estimate. Prospective deep phenotyping and longitudinal follow-up of a large sequenced cohort is needed to determine the true clinical relevance of an incidentally identified ARVC LOF variant.

Introduction

Arrhythmogenic right ventricular cardiomyopathy (ARVC) is an inherited heart disease with an estimated prevalence of 1:1000–1:5000.¹ It is associated with sudden cardiac death, particularly in young individuals.^2-4 Most cases of ARVC (50–64%) are associated with rare genetic variants in desmosome proteins,^5,6 so genetic testing has become a useful tool in the differential diagnosis of ARVC and is an integral part of familial screening.⁷

Given that clinical symptoms are frequently absent prior to a sudden death event, there is also hope that the identification of ARVC-associated genetic variants may be used to prevent sudden cardiac death in at-risk individuals. The American College of Medical Genetics and Genomics has thus recommended reporting of incidental or secondary findings related to ARVC from clinical genomic sequencing, in part, because of this anticipated benefit.⁸ These reporting recommendations have been incorporated into clinical and, increasingly, research sequencing programs, such as eMERGE⁹ and the Geisinger MyCode Community Health Initiative (also known as the DiscovEHR cohort).^10,11

Assessing the risk associated with secondary or screening-based identification of a genetic variant requires knowledge of the population prevalence of such variants, as well as their expected penetrance (i.e., how many with a variant develop the disease). For ARVC, such data remain scarce and controversial. For example, some studies have suggested that the ARVC genotype prevalence is prohibitively high (15–20% of the general population),^12,13 but these estimates were inflated by a high rate (>99%) of missense variants, which are difficult to evaluate for pathogenicity.^14,15 In contrast, our prior work found a prevalence of 0.7% for variants reported (i.e., database-listed) to be disease-causing in the first 30,000 sequenced participants in the DiscovEHR cohort.¹⁶ This work also included the first phenotype evaluation of genotype-positive individuals identified from a broad population. In short, we reported no prior ARVC diagnoses, and no demonstrable phenotype evident from review of electronic health record (EHR) data.¹⁶ These findings suggested that penetrance in a general clinical population is low, but the study was still limited by an over-representation of missense variants of uncertain significance by current guidelines.¹⁷

These critical questions of population genomic prevalence and penetrance in ARVC thus remain open. The present study sought more definitive answers through two important improvements. First, we have doubled the size of the cohort evaluated by now including 61,019 individuals with whole exome sequencing. Second, we focused almost exclusively on putative loss-of-function (LOF) variants (both known and novel), which have very strong evidence for pathogenicity,¹⁷ with exceptions only for missense variants with a consensus of assertions for pathogenicity from multiple sources in ClinVar.¹⁸ We hypothesized that, with this more conservative ascertainment of variants, we would identify a characteristic phenotype associated with ARVC. Moreover, with a refined estimate of genetic prevalence, we also infer the penetrance for such variants in a general clinical population.

Methods

A full description of the methods used in this study is available in the supplement. Institutional review board approval was obtained for this study; all individuals previously provided informed consent to participate in the MyCode Community Health Initiative. The data that support the findings of this study will not be made available.

Results

Prevalence of LOF Variants in Desmosomal Genes in DiscovEHR

A total of 74 unique ARVC-associated variants were observed in the DiscovEHR cohort. These included variants in PKP2 (n = 24), DSP (21), DSG2 (14), and DSC2 (15). No variants in TMEM43 or JUP met criteria for inclusion (consensus of pathogenicity in ClinVar). Notably, only one of the 74 variants was a missense variant (DSG2 c.146G>A); all others were putative LOF, including frameshift (n = 24), splice site (21), stop gained (26), and start lost (2). Full variant details are presented in the Supplementary Material (Supplemental Table 1).

The identified variants were present in 140 (0.23%) individuals (G⁺); the maximum number of individuals with any given variant was 13. No individual carried more than one LOF variant across the genes of interest, although four had at least one known or expected pathogenic/likely pathogenic variant in another cardiomyopathy or arrhythmia-associated gene (HCN4, LMNA, MYBPC3, TTN, and TTR; details in the supplement). Demographic details for the G⁺ individuals, compared against the rest of the sequenced cohort, are presented in Table 1. Of particular note, the median age at last encounter was 62 years and the median body mass index was 31. No G⁺ individual had an existing diagnosis of ARVC documented in the EHR.

Table 1.

Cohort Overview

	ARVC G+	Remaining sequenced cohort*	p-value
Demographics
Individuals	140 (0.23%)	60,847
Male	46 (33%)	24,613 (40%)	0.07
EUR ancestry	129 (92%)	58,741 (97%)	0.01
Age at last encounter	62 (46–73)	62 (48–73)	0.61
Body Mass Index	31 (27–36)	30 (26–36)	0.28
Rate of investigations
ECG	115 (82%)	50557 (83%)	0.74
Echocardiogram	74 (53%)	32813 (54%)	0.80
Comorbidities
CAD	28 (20%)	14148 (23%)	0.42
Hypertension	74 (53%)	36238 (60%)	0.12
Type II DM	31 (22%)	13565 (22%)	>0.99
Heart Failure	11 (8%)	4633 (8%)	0.87
Hyperlipidemia	56 (40%)	23205 (38%)	0.66
Pulmonary Hypertension	6 (4%)	1051 (2%)	0.04

Data are presented as number (%) or median (interquartile range); EUR – European; CAD – Coronary Artery Disease; DM – Diabetes Mellitus.

^*Individuals with low quality ARVC LOF variant calls were excluded from both G⁺ and remaining cohort details.

Prevalence of LOF Variants in Desmosomal Genes from gnomAD

From analysis of the non-Finnish European cohort in the Genome Aggregation Database (gnomAD), we found 79 putative LOF variants, which were present in 112 non-Finnish European individuals (0.20%; see Supplementary Material, Supplemental Table 2).

Quantitative Measures from ECG and Echocardiography

Tables 2 and 3 show group-wise comparisons of non-specific, clinically-reported measurements from electrocardiogram (ECG) and echocardiography studies, respectively. None of these results was significantly different after adjusting for multiple comparisons.

Table 2.

Quantitative findings from ECG

	ARVC G+ (n = 140)	Variant-negative controls (n = 700)	p- value*
Number with ECG data available	115 (82%)	561 (80%)	0.64
Age at acquisition [years]	60 ± 16	61 ± 16	0.74
QRS [ms]	94 ± 21	95 ± 22	0.98
QT [ms]	391 ± 49	401 ± 43	0.03
QTc [ms]	442 ± 41	442 ± 33	0.92
PR [ms]	167 ± 32	161 ± 31	0.05
Average RR [ms]	796 ± 172	835 ± 180	0.02
PACs	9 (8%)	16 (3%)	0.02
PVCs	7 (7%)	28 (5%)	0.64
LBBB	2 (2%)	11 (2%)	>0.99
RBBB	3 (3%)	27 (5%)	0.45
Atrial Fibrillation	4 (4%)	31 (6%)	0.49
Atrial Flutter	1 (1%)	2 (<1%)	0.42
LQT	4 (4%)	11 (2%)	0.30
Pacemaker	3 (3%)	13 (2%)	0.74

^*Reported p-values were not adjusted for multiple comparisons. Data are presented as number (%) or mean ± SD. QRS: QRS duration; QT: QT interval; QTc: QT interval corrected for heart rate; PR: PR interval; RR: RR interval (inverse of heart rate); PACs: premature atrial complexes/contractions; PVCs: premature ventricular complexes/contractions; LBBB: left bundle branch block; RBBB: right bundle branch block; LQT: prolonged QT interval

Table 3.

Quantitative findings from echocardiography

		ARVC G+ (n = 140)	Variant-negative controls (n = 700)	p-value*
Number with Echo data available		74 (53%)	385 (55%)	0.63
Age at acquisition [years]		63 ± 15	62 ± 16	0.79
Ao V2 max [cm/s]	mean ± SD	146 ± 56	149 ± 58	0.68
	n (reported)	61 (76%)	332 (83%)
IVSd [cm]	mean ± SD	1.09 ± 0.21	1.07 ± 0.25	0.32
	n	59 (74%)	343 (86%)
LA dimension [cm]	mean ± SD	3.62 ± 0.64	3.70 ± 0.74	0.76
	n	56 (70%)	311 (78%)
LV EF [%]	mean ± SD	54 ± 11	56 ± 10	0.09
	n	69 (93%)	345 (90%)
LV V1 max [cm/s]	mean ± SD	102 ± 26	95.6 ± 19.3	0.15
	n	61 (76%)	329 (82%)
LVIDd [cm]	mean ± SD	4.62 ± 0.77	4.62 ± 0.76	0.99
	n	59 (74%)	344 (86%)
LVIDs [cm]	mean ± SD	3.02 ± 0.88	3.14 ± 0.86	0.36
	n	52 (65%)	310 (77%)
LVOT diam [cm]	mean ± SD	2.09 ± 0.21	2.06 ± 0.24	0.25
	n	54 (68%)	302 (75%)
LVPWd [cm]	mean ± SD	1.04 ± 0.21	1.02 ± 0.22	0.53
	n	58 (73%)	342 (85%)
MV A point [cm/s]	mean ± SD	79.3 ± 23.0	78.4 ± 23.2	0.86
	n	51 (64%)	265 (66%)
MV dec slope [cm/s2]	mean ± SD	426 ± 180	442 ± 192	0.62
	n	49 (61%)	232 (58%)
MV dec time [s]	mean ± SD	0.23 ± 0.07	0.21 ± 0.06	0.48
	n	50 (63%)	274 (68%)
MV E point [cm/s]	mean ± SD	83.9 ± 29.1	80.8 ± 23.9	0.52
	n	54 (68%)	287 (72%)
MV P1/2t max vel [cm/s]	mean ± SD	98.8 ± 35.0	97.7 ± 28.1	0.88
	n	48 (60%)	231 (58%)
PA acc slope [cm/s2	mean ± SD	615 ± 284	622 ± 253	0.73
	n	48 (60%)	264 (66%)
PA acc time [s]	mean ± SD	0.12 ± 0.04	0.12 ± 0.04	0.86
	n	50 (63%)	270 (67%)
PA V2 max [cm/s]	mean ± SD	96.1 ± 24.1	98.1 ± 22.3	0.70
	n	42 (53%)	222 (55%)
RVDd [cm]	mean ± SD	3.39 ± 0.55	3.15 ± 0.58	0.11
	n	21 (26%)	123 (31%)
Reduced RV function† (TAPSE)	N with reduced function	5 (7.4%)	26 (8.0%)	>0.99
	n (reported)	68 (85%)	324 (81%)

^*Reported p-values were not adjusted for multiple comparisons.

^†Defined as TAPSE <1.6 cm prior to July 2016, and <1.7 cm after July 2016. Data are number of individuals or mean ± SD, as indicated. Ao V2 max: Maximum velocity of distal to aortic valve flow; IVSd: Interventricular septum dimension at end-diastole; LA: Left atrium; LV V1 max: Maximum LV velocity proximal to the outflow tract; LVIDd: Left ventricular internal dimension at end-diastole; LVIDs: Left ventricular internal dimension at end-systole; LVOT diam: Left ventricular outflow tract diameter; LVPWd: Left ventricular posterior wall thickness at end-diastole; MV A point: A-point maximum velocity of mitral flow; MV dec slope: Mitral valve deceleration slope; MV dec time: Mitral valve deceleration time; MV E point: E-point maximum velocity of mitral flow; MV P1/2t max vel: Maximum velocity of mitral valve flow at pressure half time; PA acc slope: Pulmonary artery acceleration slope; PA acc time: Pulmonary artery acceleration time; PA V2 max: Maximum velocity of distal to pulmonic valve flow; RVDd: Right ventricular diastolic diameter; TAPSE: Tricuspid annular plane systolic excursion.

Evaluation of Diagnostic Task Force Criteria

The majority (82%) of G⁺ individuals had a prior 12-lead ECG available for manual review, which was consistent with the frequency in variant-negative controls (hereafter, controls; Table 4), as well as the overall DiscovEHR population (Table 1). From this review, four G⁺ individuals satisfied a major repolarization criterion for ARVC (T wave inversions (TWI) in right precordial leads, V1–V3); however, this prevalence matched that of controls (n = 4; p > 0.99). The prevalence of minor repolarization criteria (TWI in V1, V2) was also comparable between groups. Similarly, for depolarization criteria, no epsilon waves were observed (major criterion), and the prevalence of terminal activation delay (TAD; minor criterion) was comparable between groups.

Table 4.

Task force criteria findings

	ARVC G+ (n = 140)	Variant- negative controls (n = 140)	p-value
ECG
ECGs reviewed	115 (82%)	113 (81%)	0.88
Major repolarization(TWI; V1–V3)	4	4	>0.99
Minor repolarization(TWI; V1, V2)	15	17	0.71
Minor repolarization(TWI; V1–V4, with complete RBBB)	0	0	>0.99
Major depolarization (Epsilon waves; V1–V3)	0	0	>0.99
Minor depolarization (TAD; V1, V2, or V3)	0	4	0.06
Holter monitor
Number with Holter data available	16 (11%)	10 (7%)	0.30
Minor arrhythmia (Non-sustained VT, unknown axis)	2	0	0.5
Minor arrhythmia (>500 VE/24 hours)	3	1	0.62
Echocardiography
Number with echocardiogram data available	62 (44%)	57 (41%)	0.63
Major structure/function	5 (8%)	2 (4%)	0.44
Minor structure/function	0	0	>0.99

Data are number of individuals meeting each criterion; TWI: T wave inversions; RBBB: Right bundle branch block; TAD: Terminal activation delay (≥55 ms from nadir of S wave to end of QRS); VE: Ventricular ectopic beats; VT: Ventricular tachycardia

Few individuals had previously undergone Holter monitoring (n = 16 for G⁺; 10 for controls, p = 0.30; Table 4). Hence, while more G⁺ individuals than controls satisfied minor diagnostic criteria for arrhythmia, such as non-sustained ventricular tachycardia (n = 2 vs. 0) and >500 ventricular ectopies over 24 hours (n = 3 vs. 1), these findings were not statistically significant.

Of the 62 echocardiography studies reviewed in G⁺ individuals, five satisfied major criteria for ventricular dysfunction and structural alteration consistent with ARVC diagnosis. This prevalence was again similar to what was observed in the control group (2/57, p = 0.44; Table 4). Zero individuals from either group satisfied only minor structural alteration and dysfunction criteria (Table 4).

Combined, one G⁺ individual satisfied both repolarization (minor) and dysfunction (major) diagnostic criteria (of the 117 G⁺ individuals and controls who had undergone both ECG and echocardiography). Three individuals satisfied multiple diagnostic criteria from ECG (i.e., major repolarization and minor arrhythmia criteria in one instance, minor repolarization and minor depolarization in two others), but all were control subjects. Notably, none of the individuals with additional pathogenic/likely pathogenic variants outside of the desmosomal genes satisfied any ECG (n = 4 with prior history) or echocardiography criteria (n = 1 with prior history). Finally, patient-reported family history review from health records (i.e., not specific to task force family history criteria) revealed no significant differences in history of heart disease, as shown in the supplement (Supplemental Table 3).

Phenome-Wide Associations with LOF Variants in Desmosomal Genes

After removing related individuals (closer than 3^rd degree), we performed a phenome-wide association (PheWAS) comparing the remaining G⁺ individuals (n = 121) to the rest of the DiscovEHR cohort. As seen in Figure 1A, carrying an ARVC LOF variant was not associated with increased odds of any cardiovascular disease diagnoses in the EHR. Considering that variants in PKP2 are much more common than variants in other genes for genotype-positive ARVC patients (73%),⁶ we hypothesized that restricting this phenotype evaluation to individuals with variants in PKP2 (n = 40) would yield a stronger signal. However, as seen in Figure 1B, no cardiovascular disease associations were again observed in this subset.

Figure 1.

A) Phenome-wide Association (PheWAS) analysis comparing unrelated individuals in DiscovEHR 60k cohort carrying ARVC LOF variants and remaining cohort without ARVC LOF variants. No significant associations were found for any phenotypes with ARVC variant status. B) PheWAS analysis comparing subset of individuals with LOF variants in PKP2 gene against remaining cohort without ARVC LOF variants; again no significant associations with cardiovascular disease were observed. ARVC: Arrhythmogenic right ventricular cardiomyopathy; LOF: loss of function.

Estimated Penetrance of LOF Variants in Desmosomal Genes from Secondary/Screening-based Findings

Having a refined estimate of genotype prevalence from a general clinical population, in conjunction with existing estimates of disease prevalence, enables an inference of population penetrance ranges (i.e., the probability of an ARVC diagnosis in G⁺ individuals). For disease prevalence, estimates range from 1:1000¹⁹ to 1:169 000¹⁶, with 1:5000 considered the most likely.¹ For genotype prevalence, we take the estimates reported above, ranging from 0.20% (gnomAD) to 0.23% (DiscovEHR); we will assume the middle of that range (0.215%) to be most likely. Finally, we assume that 64% of patients with a diagnosis have a causal variant in one of the genes we evaluated, based on data from Groeneweg, et al (excluding patients with variants in PLN from their analysis).⁶ Based on these assumptions, the estimated penetrance is 6.0% (based on ‘most likely’ estimates) with a range from 0.2% (given lowest disease prevalence and highest genotype prevalence) to 32.0% (given highest disease prevalence and lowest genotype prevalence).

Discussion

The primary findings and novel contributions of this work are 1) a refined estimate for the population prevalence of LOF variants associated with ARVC (approximately 0.20–0.23%) based on separate assessments of gnomAD and DiscovEHR; 2) an estimate for ARVC penetrance given genomic-based ascertainment of 6.0%, with a possible range of 0.2–32%; and 3) evidence that an incidental or screening-based identification of a putative LOF desmosomal variant is not associated with a consistently detectable phenotype with respect to any particular clinical diagnosis in the EHR (PheWAS), common clinical measures from ECG or echocardiography (non-specific quantitative analysis), or even ARVC-specific task force criteria from those modalities (manual chart review). Notably, this lack of detected phenotype supports our estimated penetrance from within the possible range. These findings have considerable practical relevance for the interpretation and use of secondary findings of ARVC LOF variants from genomic sequencing and suggest that the disease likelihood for individuals ascertained in this manner is low.

Genomic Prevalence

The population prevalence of variants presumed to be causal for ARVC has been the subject of several research studies,^12,13,16 but the inclusion and over-representation of missense variants of unclear clinical significance in those analyses limited the general clinical applicability of the findings. Based on current variant interpretation standards for ARVC, few missense variants have sufficient evidence to warrant a pathogenic or likely pathogenic classification.¹⁷ In contrast, LOF variants in the desmosomal genes have very strong supportive evidence for pathogenicity,¹³ and represent the majority of variants that are likely to be reported as a secondary finding. Hence, defining the population prevalence for this specific class of variants is the most clinically relevant.

While our analysis primarily focused on a single cohort (DiscovEHR), there is evidence to suggest the findings are generalizable, at least for individuals of European ancestry. Specifically, we directly showed comparable frequency of ARVC LOF variants (0.20%) in the non-Finnish European subset of gnomAD, which was comparable to data from Walsh et al. from the ExAC database (0.24% carrier frequency for DSC2, DSG2, DSP, and PKP2).²⁰ The extent to which this prevalence differs across ethnicities is currently unknown and warrants further study.

Phenotypic Findings

Individuals identified with ARVC-associated LOF variants did not, on average, demonstrate an abnormal phenotype from evaluation of existing medical data, including 12-lead ECG (which has been shown to be the most sensitive of the task force criteria, and the most prevalent in affected individuals²¹) and echocardiogram. Hence, the focus on putative LOF variants did not result in a more consistent phenotypic signal, as we had hypothesized. This finding is consistent with our previous study,¹⁶ despite limited overlap in the variants and individuals studied because of differences in variant selection (database reporting vs. annotation-based). While these null findings could reflect limitations of EHR-based phenotype evaluation, they contrast with the strong cardiomyopathy, heart failure, and arrhythmia phenotype signals observed from a recent analysis of individuals with LOF variants in TTN, as well as an association of LMNA variants with cardiomyopathy using the same data sources.^22,23

One important consideration is that the spectrum of diseases associated with desmosomal variants (including the recently broadened definition of “arrhythmogenic cardiomyopathy”) is imperfectly understood, and variants in different genes may result in different phenotypes²⁴. We attempted to address some of these disparities through a separate PKP2 PheWAS (which is the most prominent ARVC gene), but no enrichment of signal was observed. Therefore, more research is needed to provide a systematic understanding of the genotype-phenotype relationships of desmosomal variants.

Our findings from the ECG and echocardiogram reviews are highly relevant when considering prospective clinical evaluation of individuals with these secondary genomic findings. Approximately 20% of G⁺ individuals with a previous ECG or Holter monitor satisfied some ARVC task force criteria, including four individuals satisfying a major repolarization criterion. From echocardiograms, 7% of G⁺ individuals reviewed satisfied major criteria. In an isolated clinical evaluation, these findings—particularly major criteria, in combination with the genetic variant—could prompt a diagnosis. However, in aggregate, we observed a comparable rate of these same abnormalities in our control population, which therefore cannot be specifically attributed to the ARVC variant. Hence, the prospective evaluation of ARVC G⁺ individuals identified secondarily or through genomic screening should take these findings into account with respect to diagnostic “scoring”. For example, the weight attached to the pathogenic variant (a major criterion) should be reduced or removed, requiring multiple other criteria, such as relevant family history findings, be satisfied for a definitive diagnosis. Continued clinical surveillance to identify possible disease progression, particularly for younger individuals, is still warranted however, as discussed elsewhere.¹¹

ARVC Penetrance in a General Population

Previous estimates of ARVC penetrance were based on heritability within affected families, which reported that 39–45% of relatives inheriting the familial variant were definitively diagnosed at initial evaluation,^25,26 corresponding to an incidence of 2.8 new diagnoses per 100 person years.¹¹ In contrast, our best estimate of penetrance based on inferential analysis of population-level statistics is much lower at 6.0%. While not rooted in direct observation, this estimate of low penetrance is consistent with the lack of an EHR-based phenotype observed in our cohort. The reasons for this large discrepancy between familial and population penetrance are not clear, but the mediating factors frequently cited to explain familial non-penetrance (e.g., history of endurance exercise/athletics²⁷ or secondary genetic factors²⁸) remain relevant and are also much more variable at a population level than within a single family.

An important limitation and confounding factor in this analysis is the likely survival bias associated with our study ascertainment favoring unaffected individuals. In other words, because the median age (~60 years) of our sequenced cohort is considerably older than the typical age of ARVC presentation, particularly with respect to sudden cardiac death events, the study cohort is likely to be enriched for unaffected survivors. Hence, we utilized and compared against LOF prevalence estimates from other cohorts in an effort to address this limitation. However, we note that the substantial uncertainty in disease prevalence estimates contributed more to the uncertainty in our penetrance estimate than the expected dropout due to sudden cardiac death.²⁹

Low expected penetrance has important implications for the clinical approach to prospective evaluation of secondary ARVC genomic findings. First, individuals being notified of such findings can be counseled regarding their risk of disease in light of these estimates (absent presenting symptoms or relevant personal or family history). This information may help to ameliorate the psychological stress arising from notification of genetic risk for a disease associated with sudden cardiac death. Second, while a baseline diagnostic examination should still be performed in most cases, a low (perhaps age-dependent) pre-test probability of diagnosis should be assumed in evaluating results of those tests, particularly in the absence of relevant personal or family history, to avoid over-diagnosis. As mentioned, identification of one additional major diagnostic criterion or two minor criteria (beyond the genetic variant) should not be considered sufficient to warrant a diagnosis. Additional considerations for managing these findings have recently been discussed elsewhere.¹¹

Nonetheless, this low penetrance estimate does not necessarily negate the potential value of genomic screening for LOF desmosomal variants, as the risk of sudden cardiac death for these individuals and, importantly, their siblings and children remains unknown. That is, lack of a phenotype in one individual does not preclude a significant phenotype in close relatives, who could be identified through subsequent cascade testing. Hence, these findings suggest the importance of a pragmatic approach to diagnostic testing following a genomics-based ascertainment while future work seeks to refine our capabilities for early risk stratification.

Limitations

Clinical testing, particularly echocardiograms, were not available for the entire cohort, so it is possible that some individuals could have a phenotype that could not be retrospectively assessed. Prospective patient contact and clinical follow-up testing is ongoing in this cohort to help address these gaps through an institutional genomic screening and counseling initiative.

As mentioned, there is likely a survivor bias associated with our study ascertainment favoring unaffected individuals. The relative under-representation of males in our G⁺ cohort also suggests such bias may be present. This possibility represents a critical caveat in the interpretation and application of our penetrance findings, particularly with regard to the evaluation of younger individuals at higher risk of sudden death from ARVC. Family history details collected and cascade screening performed as part of our prospective efforts will also help to shed light on the presence and significance of these potential biases and their impact on observed penetrance.

Conclusion

This work represents, to date, the most comprehensive analysis of ARVC genotype and genotype/phenotype association using a genomics-based ascertainment from a large clinical population. Despite more conservative estimates of genotype prevalence based almost exclusively on putative loss-of-function variants, a consistent phenotype signal remains elusive. These findings are supported by low penetrance estimates (6.0%) derived from inferential analysis; much lower than what has been reported within families presenting clinically with ARVC. This was a retrospective analysis without complete data available for all individuals, so prospective follow-up with deep phenotyping is still needed to confirm these findings. Yet, these results can help inform clinical decisions and patient counseling in the evaluation of individuals secondarily identified with ARVC-associated genetic variants.

Non-standard Abbreviation and Acronyms

ARVC

Arrhythmogenic right ventricular cardiomyopathy

EHR

Electronic health record

LOF

Loss-of-function

G+

Genotype-positive

TWI

T-wave inversion

TAD

Terminal activation delay

gnomAD

Genome Aggregation Database

ECG

Electrocardiogram

PheWAS

Phenome-wide association