Disease Penetrance and Phenotypic Spectrum of Desmoplakin Variant Carriers in the Population

Abstract

Background:

Desmoplakin (DSP) variants cause arrhythmogenic cardiomyopathy (ACM), a disorder characterized by myocardial fibrosis, arrhythmias, and sudden cardiac death. DSP-mediated ACM often involves left ventricular (LV) dysfunction and myocardial inflammation, yet existing diagnostic criteria may underdetect disease, underscoring the need for population-based penetrance estimates.

Objective:

To assess the prevalence and phenotypic penetrance of DSP variants in a genotyped population.

Methods:

Among 200,580 UK Biobank participants with exome sequencing, DSP variants were classified as predicted-deleterious (pDel), predicted loss-of-function (pLOF), or ClinVar 2-star pathogenic/likely pathogenic (P/LP). A subset of pDel carriers underwent ECG and CMR testing, matched to genotype-negative controls. Phenotypic penetrance was assessed using ICD-10 diagnoses, ECG, and CMR. Variant clustering within functional DSP domains was also evaluated.

Results:

Of 200,580 participants, 1407 (0.7%) carried a pDel, 168 (0.08%) a pLOF, and 44 (0.02%) a ClinVar 2* P/LP DSP variant. Myocarditis occurred in 0.28% of pDel, 1.8% of pLOF, and 4.5% of ClinVar P/LP carriers versus 0.07% of controls (p<0.05). Cardiomyopathy prevalence increased from 1.4% (pDel) to 5.4% (pLOF) and 6.8% (P/LP) versus 0.6% in controls (p<0.01). DSP carriers had more frequent lateral T-wave inversions and abnormal LV strain. Missense variants clustered within two functional DSP domains.

Conclusions:

DSP pDel variants are common but show low penetrance for myocarditis and cardiomyopathy, with risk increasing with more stringent classification. Electrocardiographic and strain abnormalities may aid early detection, supporting genotype-first approaches for DSP interpretation.

Introduction

Arrhythmogenic cardiomyopathy (ACM) is an inherited myocardial disorder, characterized by ventricular arrhythmias, progressive systolic dysfunction, and sudden cardiac death¹. ACM primarily stems from cardiac desmosomal abnormalities, leading to fibrofatty replacement of the myocardium, adverse myocardial remodeling, and fibrosis². While ACM was initially characterized by predominant right ventricular dysfunction, recent evidence suggests that ACM may also present with primary LV or biventricular involvement^{3, 4}. Recent evidence reports that genotype-specific therapeutic approaches are on the horizon which highlights the important role of genotype focused approach into the study of the natural history of ACM progression⁵.

Desmoplakin (DSP), a key desmosomal protein, is essential for cardiac force transmission and conduction by linking itself to intermediate filaments and other key proteins^{6, 7}. Pathogenic DSP variants cause a distinctive ACM subtype⁸, characterized by lateral T-wave inversions, ventricular arrhythmias, and episodic myocardial injury which contributes to progressive fibrosis with preserved systolic dysfunction until late in the disease course^{9, 10}. Unlike classic ARVC, DSP-mediated cardiomyopathy exhibits an inflammatory component, wherein episodes of myocardial injury occur in “hot phases” of inflammation, resembling myocarditis in overall disease progression^10–12. Phenotypically, such “hot phases” clinically manifest with shortness of breath, chest pain, elevation in troponin, and electrocardiographic changes; Notably, there is even new evidence of potential progression of DSP-cardiomyopathy during endurance exercise likely due to increased risk of myocardial injury; And evidence of treatment of such inflammatory hot phases has shown improved arrhythmia and heart failure outcomes in this population^13–16. Of note as DSP-cardiomyopathy can affect both ventricles and carries a high risk for heart failure and major arrhythmias, reported rates are less common in ALVC dominant subtype^{17, 18}.

Given the distinct presentation of DSP-cardiomyopathy, traditional diagnostic criteria for ACM exhibit poor sensitivity and accuracy for DSP-carriers^{19, 20}. Thus, there is a need to identify specific phenotypic markers associated with predicted deleterious (pDel) DSP variants. At the population level, timely identification of DSP carriers may offer a window of opportunity for interventions to reduce DSP-related arrhythmic outcomes. This study leverages the UK Biobank exome dataset to determine the frequency, penetrance, and phenotypic spectrum of DSP variants on a population-wide scale. We aimed to determine penetrance of pDel DSP variants and identify specific, easily accessible markers to help in early identification of DSP carriers at risk of disease progression.

Methods

Analyses were conducted in the UK Biobank (UKBB), with access provided under application number 48286. We restricted analysis to UKBB participants who underwent exome sequencing (ES) who provided written consent²¹. UKBB data are available to approved researchers upon application. We first assessed the ES cohort, with sub-cohorts defined by DSP-genotype category and completion of 12-lead ECG and cardiac magnetic resonance (CMR) imaging (Figure 1). An extensive variant and sample-level pipeline was applied to produce a high-quality dataset of DSP variants^22–25. DSP variants were stratified into three variant-call groups based on variant type and pathogenicity class: Group-1 (pDel): Predicted deleterious variant (pDel) carriers with a predicted loss of function variant or missense variant with predicted-deleterious effects. Group-2 (pLOF): Predicted loss of function variant (pLOF) carriers, including splice-site variants, frameshift insertion/deletions, or early stop-gain mutations. Group-3 (ClinVar 2* P/LP): ultra-filtered subset of variants listed in ClinVar as P/LP with at least two-star evidence level (sFigure 1). A DSP-positive sub-cohort underwent 12-lead ECG and CMR diagnostic testing and compared with a matched wildtype control, genotype-negative for cardiomyopathy and ion channelopathy P/LP variants (G-)²⁶,with full protocol available in eMethods^27–36. DSP variants were analyzed by gene and amino-acid regions for predicting pathogenicity^37–40. The research reported in this paper adhered to Helsinki Declaration as revised in 2013 guidelines. Statistical analyses are described in eMethods.

Figure 1: Study methodology and variant sub-cohorts.

This is a flowchart that describes the process of identifying the study population, including inclusion and exclusion criteria and variant sub-cohorts. Created with biorender.com

Results

Study Population

Among 502,493 UKBB participants, 200,580 underwent ES (55.1% female, 56.5y at recruitment on average), 42,078 had 12 lead ECG, 39,616 underwent CMR. Cardiomyopathy (CM) was present in 1,192 (0.6%); dilated cardiomyopathy (DCM) in 489 (0.2%); myocarditis in 137 (0.07%); and heart failure (HF) in 6,326 (3.2%). Ventricular arrhythmia (VA) was present in 1,787 (0.9%) and atrial fibrillation/flutter in 13,883 (6.9%) (sTable 1).

Clinical Disease Penetrance of DSP Carriers in the UKBB

Among the 200,580 individuals with ES, 1407 (0.70%) were identified to carry a pDel variant; 168 (0.08%) with pLOF variants, and 44 (0.02%) with ClinVar 2* P/LP variants (Figure 1). Among 1407 pDel carriers, 19 (1.4%) individuals had a clinical diagnosis of cardiomyopathy; 9 (0.6%) with DCM; 4 (0.3%) with myocarditis; and 53 (3.8%) individuals had HF. Those with pDel DSP variants had a four-fold increase in myocarditis incidence compared to the rest of the DSP-negative UKBB population (0.28% compared to 0.07%; p value <0.05). The pDel DSP cohort also exhibited a higher prevalence of myocarditis, CM, and DCM. 1.4% of individuals had a clinical diagnosis of CM, with the risk of developing cardiomyopathy being two-fold (hazards ratio (HR) 2.33, p < 0.001). Among pDel DSP carriers, 0.28% had a clinical diagnosis of myocarditis, with the risk of developing myocarditis being four-fold (HR 4.30, p<0.005) by logistic regression analysis (sFigure 2, sTable 2), and no noted impact on major adverse cardiovascular outcomes (MACE) (sTable 3). Overall, these findings show that DSP variants are common in the general population, yet lead to a low penetrant risk of developing myocarditis, cardiomyopathy, and DCM.

Penetrance Analysis by Gene-Evidence Category

A subset of 168 individuals were identified to have a pLOF variant, defined by the presence of a stop-gain, frameshift insertion or deletion, or splice-site mutation. The pLOF group had a higher prevalence of CM, DCM, myocarditis, and HF (Figure 2). Among this pLOF group, 5.4% (n = 9) had a diagnosis of CM compared to the 0.6% (n=1,183) of the DSP-negative group (p<0.001). A similarly significant enrichment was noted in pLOF DSP carriers for DCM, wherein 3.6% (n=6) of pLOF DSP carriers had a DCM diagnosis, compared to 0.2% (n=483) of the DSP-negative population (p<0.001) (sTable 4). Notably, the pLOF group also had a significantly higher myocarditis burden compared to the DSP-negative population, with 1.8% (n=3) in the pLOF group versus 0.07% (n =134) in the DSP-negative population (p<0.001). An enrichment in HF was seen, with 6.5% (n=11) in the pLOF group compared to 3.2% (n=6,315) of the DSP-negative population (p<0.05). Further, enrichment of shortness of breath was noted in pLOF carriers compared to the DSP-negative group (p<0.05). The impact of pLOF variants on CM and myocarditis was significant, with an HR of 9.9 [95% CI 5.16–19.1; p <0.001] for CM and 28.3 [CI:9.01–88.9; p <0.001] for myocarditis by Cox proportional hazards regression adjusted for sex (sTable 5). pLOF carriers were more likely to develop HF with an HR of 2.24 (p<0.01) (sTable 6).

Figure 2: Phenotypic penetrance and event-free survival of DSP-positive carriers identified by the presence of a pLOF mutation.

A. 5.4% (9/168) of the DSP-pos pLOF carriers were noted to have CM compared to 0.6% (1,183/200,412) of the DSP-neg population (p<0.001). B. 3.6% (6/168) of pLOF DSP carriers with DCM compared to 0.2% (483/200,412) of DSP-neg population (p<0.001). C. 1.8% (3/168) of pLOF carriers with myocarditis compared to 0.07% (134/200,412) of DSP-neg; (p<0.001). D. 6.5% (11/168) of the pLOF DSP carriers with HF compared to 0.2% (483/200,412) of DSP-neg population (p<0.05). E. Kaplan-Meier survival curve depicting time to CM diagnosis in pLOF DSP-pos carriers compared to DSP-neg individuals, noting pLOF carriers with significantly reduced event-free survival (p<0.001). F. Kaplan-Meier survival curve depicting time to myocarditis diagnosis in pLOF carriers compared to DSP-negative individuals, with significantly reduced event-free survival in DSP-pos pDel carriers (p<0.001).

A total of 44 individuals were identified to have ClinVar 2* P/LP variants. At this most stringent level for DSP variant filtering, we noted a significant enrichment in myocarditis and cardiomyopathy (sTable 7). A further marked increase in myocarditis was noted in the DSP-positive cohort (4.5%, n=2) when compared to the DSP-negative population (0.07%, n =135) (p<0.001). ClinVar 2* P/LP carriers thus had a 63.6-fold increased risk of developing myocarditis, with a HR of 63.6 [95% CI: 15.7–257; p<0.001] (sTable 8). Among ClinVar 2* P/LP carriers, 6.8% (n=3) had a CM diagnosis compared to 0.6% (n=1,189) of the DSP-negative group (p=0.002) (sFigure 3). with a 10-fold increased risk for developing CM [95% CI: 3.5–33.7; p< 0.01], with no significant impact on MACE outcomes noted in this smaller, ultra-filtered group (sTable 9). Overall, the penetrance of myocarditis and cardiomyopathy was noted to progressively increase by stringency of variant call method (Figure 3).

Figure 3: Progressively increasing myocarditis and cardiomyopathy enrichment in P/LP DSP variants by more stringent variant calling.

A. Myocarditis penetrance: 0.28% (4/1407) of pDel (p<0.05), 1.8% (3/168) of pLOF (p<0.001), 4.5% (2/44) of ClinVar 2* P/LP (p<0.001) with myocarditis vs 0.07% [133/199,173; 134/200,412; 135/200,536 respectively] of DSP-neg. B. Cardiomyopathy penetrance: 1.4% (19/1407) of pDel (p=0.001), 5.4% (9/168) of pLOF (p<0.001), 6.8% (3/44) of ClinVar 2* P/LP (p=0.002), with CM vs 0.6% [1173/199,173; 1,183/200,412; 1189/200,536 respectively] of DSP-neg.

Subclinical Disease Penetrance by ECG and CMR

Of the ES cohort, 488 individuals underwent both 12-Lead ECG and CMR (50.8% female; 56 years of age at recruitment). 122 pDel DSP carriers were compared against the matched genotype-negative control group (G-) comprising 366 individuals. VA was present in 5 (1.0%); atrial fibrillation in 28 (5.7%); HF in 12 (2.5%) of these individuals. Of 122 DSP carriers, 9.8% (n=12) had T-wave inversions (TWI) in the lateral-most lead, V6, compared to 4.8% (n=16) of the G- control (p <0.05) (Figure 4). No significant differences in repolarization abnormalities were noted in precordial, right-sided leads between the two groups.

Figure 4: Increased subclinical penetrance on 12-lead ECG and CMR in DSP-positive pDel carriers compared to matched genotype-negative control (G-).

A. Visual diagrammatic representation of ECG leads V1 and V6, in relation to the heart; with V1 providing the most right-sided precordial view and V6 placed laterally, as the left-most precordial lead, with T-inversion in V6 from likely pathogenic DSP carrier. B. 7.4% (9/122) of pDel DSP carriers with TWI in V1 compared to 7.1% (26/366) in the G- control group (p >0.1). 9.8% (12/122) of pDel DSP carriers in the subclinical cohort with T wave inversions in V6 compared to 4.4% (16/366) of the G- control. *p < 0.05. Created with biorender.com. C. 12.3% (15/122) of pDel DSP carriers with reduced peak global longitudinal strain vs 6.6% (24/366) in the G- (p <0.05). D. 4.9% (6/122) of pDel DSP carriers with reduced RVEDV vs 1.6% (6/366) in the G- (p <0.05). E. 11.5% (14/122) of pDel DSP carriers with reduced RV cardiac output vs 5.5% (20/366) in the G- (p <0.05). F. 8.2% (10/122) of pDel DSP carriers with reduced peak global longitudinal strain to 3.6% (13/366) in the G- (p <0.05).

CMR analysis demonstrated impaired mean radial left ventricular (LV) strain in the DSP-positive cohort compared to G-, with mean LV radial strain of 32.2 (± 5.8) in DSP-positive cohort and 33.5 (± 5.6) in G- (p<0.05). 12.3% (n=15) in the DSP-positive group were noted to have reduced peak global longitudinal strain, defined by <16%, with 12.3% compared to the 6.6% (n=24) in the G- control (p<0.05) (sTable 10). This occurred in the absence of any significant differences on T1 weighted CMR, LV ejection fraction (EF), end diastolic (EDV) and systolic volumes (ESV), stroke volume, and cardiac output between the two groups. CMR analysis of the right ventricle (RV) demonstrated an increased proportion of DSP-positive individuals with reduced RVEDV, RVESV, and RV cardiac output compared to the G- control, with 4.9% (6) versus 1.6% (6) with reduced RVEDV; 8.2% (10) versus 3.6% (13) with reduced RVESV; and 11.5% (14) versus 5.5% (20) with reduced RV cardiac output respectively (p<0.05). Subclinical phenotypic penetrance analysis thus indicates that DSP-carriers may have increased left-sided repolarization abnormalities and early signs of strain.

Variant Location and Mutational Hotspots

Given the association of DSP variants with a low penetrant cardiovascular phenotype, we next sought to determine whether the location of variant on the DSP gene or encoded amino acid (AA) sequence predicted pathogenicity or penetrance. The location of 168 pLOF DSP truncating variants was analyzed across three key gene regions categorized by degree to which truncating variants are prone to nonsense mediated decay (NMD) where phenotypically penetrant and non-penetrant pLOF DSP variants did not localize to a specific risk region (Figure 5).

Figure 5: Variant location as a predictor of pathogenicity.

A. Topology of the DSP gene by gene region designated by susceptibility to nonsense-mediated decay, displayed across major isoforms DSP I and II. B. Distinct DSP pLOF variants found in the UKBB across the 3 key gene regions. C. Key protein domains across the amino acid sequence of the desmoplakin protein. D. A plot of all missense DSP variants in the UKBB by amino acid position, with the blue lines signifying the P/LP missense variants with predicted deleterious effects, and grey lines denoting the missense variants that are benign, likely benign, and VUS’s without predicted deleterious effect. E. A plot of all missense DSP variants in the ClinVar database by amino acid position, with blue lines signifying the P/LP missense variants with predicted deleterious effects, and grey lines denoting the missense variants that are benign, likely benign, and VUS’s without predicted deleterious effect. F. A stacked bar graph noting the distribution of P/LP and non-P/LP missense variants in DSP in the UKBB by location, with a higher proportion of P/LP variants in the hotspots compared to non-P/LP variants (p<0.01). G. A stacked bar graph demonstrating a similar distribution with a higher proportion of P/LP DSP missense variants in hotspots compared to non-P/LP missense DSP variants in an independent pool of missense DSP variants from ClinVar (p<0.05).

In leveraging the previously published signal-to-noise tool, DiscoVari, to investigate the 1200 missense variants in DSP in the UKBB, noting a higher proportion of P/LP missense variants to fall in DiscoVari hotspots when indexed for AA sequence length compared to outside hotspots, with 451 P/LP variants falling in the 2 DiscoVari regions collectively spanning 887 amino acids in length compared to the 749 P/LP variants in the collective 1984 amino acid long area outside of the two DiscoVari hotspots (p<0.001). The DiscoVari regions corresponded to specific regions in key binding domains in the desmoplakin protein: PKP1/JUP/PKP2 binding domain (1–584) and the intermediate filament binding domain (1946–2871). The population-first findings of missense variant distribution in the UKBB in conjunction with DiscoVari signals and major binding domains in the desmoplakin protein were then leveraged to define key novel mutational hotspots as spanning amino-acid positions 1–784 and 1946–2871. A far higher proportion of P/LP missense variants [74.3% (n=892)] were found in these newly defined mutational hotspots compared to 50% (n=6062) of non-P/LP missense variants in the UKBB (p<0.001). The distribution P/LP and non-P/LP missense variants was further validated on an independent external variant database, ClinVar, which comprised a total of 21 P/LP DSP missense variants by ClinVar 2-star criteria and 1971 non-P/LP missense variants, where 85.7% (n=18) of ClinVar P/LP missense variants were found in the novel hotspots, compared to 62.1% (n=1225) of non-P/LP missenses. P/LP DSP missense variants in ClinVar were found to 3.7 times more likely to be in the novel hotspots compared to non-P/LP (p<0.05). The missense variant distribution in both the UKBB and ClinVar databases indicate that these mutational hotspots (1–784 and 1946–2871) hold a higher burden of P/LP missense DSP variants, indicating predictive value of location for variant pathogenicity.

Discussion

Given the distinct presentation of DSP-associated ACM from the conventional ACM phenotype, there is a pressing need to better understand DSP mediated disease, as existing risk-stratification and diagnostic criteria currently used for ACM have poor sensitivity and diagnostic accuracy for DSP-carriers⁴¹. Our study stands as the first, large-scale, population-based analysis that provides insights into the clinical and subclinical penetrance of DSP-associated pDel variants, revealing several key findings with significant clinical implications; notably this is reported in the UKBB which reports results in older individuals by design.

First, our data indicates an overall prevalence of individuals with pDel DSP variants to be 0.7% in the UKBB population which consists primarily of middle-aged adults, with pLOF and ClinVar 2* P/LP carriers being less common in the population. Second, a significant enhancement in penetrance of ACM-associated cardiac phenotypes including CM, DCM, HF, and myocarditis was observed in the pLOF DSP-positive cohort. Third, while overall cardiovascular penetrance of DSP variants was low, there was a progressively increasing enhancement with increasing stringency of variant call method.

Most notably, a markedly elevated proportion of individuals in all DSP-positive cohorts was found to have myocarditis compared to the DSP-negative population. Through increasingly stringent criteria for variant pathogenicity call, enhancement in the myocarditis signal in the DSP-positive group increased by 25-fold and by over 60-fold, respectively, compared to the DSP-negative population. A similar progressively enhancing penetrance was seen with cardiomyopathy. This suggests that variant type (pLOF vs pDel missenses) as well as ClinVar designation could offer higher predictive value in terms of risk of myocardial injury-related hot phases, or myocarditis, a known part of the DSP-mediated disease progression. More importantly, these results suggest that individuals at the population level who confer DSP variants are predisposed to DSP-CM related phenotypic manifestations well characterized in clinical cohorts. Thus, identification of DSP variants as a secondary or incidental finding should prompt clinical counseling and evaluation for not only cardiac manifestations, but also regarding past episodes of myocarditis and myocardial injury, which in turn might represent factors predisposing to disease progression^41–44. One notable detail from the results of this study is the increased penetrance of cardiomyopathy and myocarditis in ClinVar only sub cohort compared to the predicated deleterious sub cohort which implies that not all variants confer the same risk profile which can in theory help risk stratify individuals at further risk to develop disease that will be relevant to young carriers that might be interested in pursing athletic activities that may exacerbates an already elevated risk profile, this will certainly require more research before clear recommendation can be offered in clinical settings⁴⁵.

ECG findings in the 122 pDel cohort revealed a significantly increased proportion of individuals with TWI in lead V6 compared to the G- control. The well-established revised 2010 International Task Force Criteria clearly include negative T-waves in right-sided precordial leads as a specific minor criterion for diagnosing ARVC⁴⁶. T-wave abnormalities in lateral ECG leads indicative of LV involvement are common in DSP-disease cohorts⁴⁷. Yet, this often goes under-recognized given lateral-lead repolarization changes do not constitute a diagnostic or risk factor criterion for conventional ACM⁴⁸. Thus, this repolarization change in the lateral-most ECG lead in the DSP-positive group without concurrent changes in the anterior precordial leads demonstrates a left-dominant presentation of DSP-mediated ACM at a population level⁴⁹. While there is clear consensus that the genetic basis of ACM must be considered to better predict disease progression⁵⁰, lateral lead (V6) TWI, particularly in the absence of right-sided repolarization abnormalities in the DSP-positive population may offer value in being an early risk-predictor at a population-level.

CMR analysis demonstrates that DSP carriers can show early signs of myocardial injury measured by worsening mean longitudinal and radial LV strain in the absence of significant differences in LV volumes and function, with prior studies illustrating that worsening left heart strain may be an early marker of progression to systolic dysfunction^51–53. The increased proportion of DSP carriers with reduced RV volumes and cardiac output compared to G- supports that DSP-mediated disease has biventricular involvement, with no significant difference in EF and function indicating likely early disease. More recently there has been work that further explores the clinical outcomes and morphological markers of DSP-cardiomyopathy with integrated new risk scores that aim to risk stratify arrhythmic risk in these cohorts that are aim to arrive to similar landscape compared to the results presented in this work.^{54, 55}

Analyzing pDel missense and pLOF variants by location represents the first large population-wide analysis for location-related predictors for pathogenicity by gene region and amino acid sequence. Our study ultimately validates that DiscoVari-identified mutational regions, and more broadly, the PKP/JUP and intermediate filament binding domains in desmoplakin, hold enrichment of P/LP missense variants compared to non-hotspot regions, ultimately making it possible to define key mutational hotspots based on these findings in the UKBB which were validated by an additional, independent variant database. Given loss of function is the key mechanism for DSP-mediated disease, it is crucial to be able to interpret risk of pDel variants and further characterize pathogenicity risk in missense variants⁵⁶. A prior study found that variant location by gene region prone to NMD was a risk factor for ventricular arrhythmia in an enriched population of truncating variants in DSP with diagnosed ACM³⁸. However, in a population-first approach of pLOF DSP variants with an already low penetrance of phenotypic manifestations, this effect was muted. The DiscoVari-identified pathogenic hotspots and the key functional binding domains for the desmoplakin protein noted increased clustering of P/LP missense variants, demonstrating the functional relevance of variant location to predicted pathogenicity. P/LP DSP missense variant clustering was noted to follow a similar pattern in an independent missense-variant-repository, further validating missense variant location may be predictive of pathogenicity. This aligns with similar works that identifies variant location as a novel risk factor for DSP associated ARVC; The type and location of said variants was also associated with HF and arrhythmic outcomes.^{38, 57}

The current standard of risk stratification and diagnostic criteria for individuals with ACM is based on the modified international task force criteria, which holds poor sensitivity and diagnostic accuracy for DSP mediated disease⁴⁸. Phenotypic penetrance and genotype status assessed at this population-wide level gives us further insight into DSP-mediated disease and helps us better identify genotype-positive individuals who are at greater risk for disease. Our phenotype association studies demonstrate that individuals with pLOF variants are at greater risk for penetrant disease. Among missense variants, those that fall into mutational hotspots are more likely to be P/LP. Additionally, early subclinical markers in the setting of low-penetrant disease among DSP-carriers including left-sided repolarization abnormalities and reduced strain – i.e. left sided structural and electrophysiological findings may have the potential to better identify at-risk DSP carriers in the population. Taken together, our findings provide insight into DSP-mediated ACM at a large population level and highlight the need for updated risk stratification and diagnostic criteria for DSP carriers. Future endeavors with emphasis on data interoperability between different cohorts and biobanks while studying genetic-specific presentation in cardiomyopathy will be crucial moving forward.^{58, 59}

Limitations

The study results should be interpreted considering the following limitations. First, the larger clinical analysis of the 1407-individual DSP-positive pDel cohort is primarily based on ICD-10 diagnoses. While this approach is routinely used in population-based research, the absence of a specific diagnoses code for ACM, may have possibly contributed to the under-recognition of cardiac phenotype penetrance⁶⁰. Second, the UKBB does not include recruits under 40 years-of-age, which may contribute towards underestimating severe disease, as this may exclude the younger ACM population with more severe disease and sudden cardiac death^{61, 62}. Ethical considerations posed by the observational nature of the UKBB study precludes the use of contrast or stress agents, and thus limited CMR analysis from including gadolinium enhancement for myocardial tissue characterization⁶³.

Conclusions

While pDel DSP variants are relatively common in the general population, its overall clinical penetrance of cardiomyopathy and myocarditis is low. A higher proportion of individuals with DSP variants were noted to have myocarditis compared to the DSP-negative population, indicating susceptibility to a pro-inflammatory state in DSP-mediated disease. Left-sided repolarization abnormalities on ECG and worsening LV strain on CMR in the absence of overt systolic dysfunction are specific and early signs of disease involvement in the DSP-positive population that may hold value in being applied as early diagnostic markers and risk predictors of DSP-associated cardiomyopathy. Finally, the location of missense DSP variants by amino-acid sequence and mutational hotspot analysis can provide informative guidance on the pathogenicity of DSP variants.

Key Points.

What is known?

DSP variants are a well-established cause of ACM; however, traditional diagnostic criteria may underrecognize DSP-mediated cardiomyopathy and there is the need for genotype-specific guidance to better inform on early and accurate recognition of DSP-mediated disease. Prior studies have focused on large clinical cohorts, but a population-first approach to assess clinical and subclinical penetrance of DSP carriers has not been done yet.

What this study adds:

This study shows that there is progressive enrichment in clinical penetrance of DSP carriers by stringency of variant call for pathogenicity, with early left-sided phenotypic signs noted on ECG and CMR. This study also identifies mutational hotspots enriched for pathogenic missense variants, providing new insights into genotype-phenotype associations. Overall, these findings highlight the importance of a genotype-first approach for ACM, as identification of DSP variants as a secondary or incidental finding should prompt clinical counseling and longitudinal evaluation for distinct genotype-specific cardiac manifestations such as myocarditis.

Abbreviations

AA

Amino Acid

CMR

Cardiac Magnetic Resonance

CM

Cardiomyopathy

ES

Exome Sequencing

G-

Genotype-Negative Control Group

MACE

Major Adverse Cardiovascular Events

NMD

Nonsense-Mediated Decay

P/LP

Pathogenic/Likely Pathogenic

pDel

Predicted Deleterious

pLOF

Predicted Loss-of-Function

REVEL

Rare Exome Variant Ensemble Learner

TWI

T-wave Inversion

UKBB

UK Biobank

VA

Ventricular Arrhythmia

VUS

Variant of Uncertain Significance