Article Text

Lower than expected desmosomal gene mutation prevalence in endurance athletes with complex ventricular arrhythmias of right ventricular origin
  1. A La Gerche1,3,
  2. C Robberecht2,
  3. C Kuiperi2,
  4. D Nuyens1,
  5. R Willems1,
  6. T de Ravel2,
  7. G Matthijs2,
  8. H Heidbüchel1,3
  1. 1Department of Cardiovascular Medicine, University Hospital, University of Leuven, Belgium
  2. 2Centre for Human Genetics, University Hospital, University of Leuven, Belgium
  3. 3Department of Medicine, St Vincent's Hospital, University of Melbourne, Australia
  1. Correspondence to Prof Hein Heidbüchel, Cardiology-Electrophysiology, University Hospital Gasthuisberg Herestraat 49 B-3000 Leuven, Belgium Europe; hein.heidbuchel{at}


Objective To determine the prevalence of desmosomal gene mutations in athletes with complex arrhythmias (VA) of right ventricular (RV) origin and structural RV abnormalities to evaluate whether there is sufficient genetic overlap with arrhythmogenic right ventricular cardiomyopathy (ARVC) to consider them the same or different entities.

Design Observational cohort

Setting Tertiary hospital referrals

Patients Forty-seven consecutive athletes (age 42 (11) years) with complex VA of RV morphology (excluding idiopathic right ventricular outflow tract ventricular tachycardia), who performed 14 (9) h/week of moderate to intense sport practise for 19 (9) years.

Interventions Clinical evaluation (detailed sports history, multi-modality imaging, electrophysiological study) and sequencing of five candidate desmosomal genes.

Results A clinical diagnosis of definite or suspected ARVC by task force criteria (TFC) was met in 24 (51%) and 17 (36%), respectively. ARVC classification was not related to the rate of major arrhythmic events (p=0.28). Pathogenic mutations (four novel) were identified in six athletes (12.8%), which is below published rates for familial ARVC (27–52%). Moreover, only two athletes had a suggestive family history. Severe RV dysfunction was more frequent in mutation carriers (33% vs 2%, p=0.04), but otherwise TFC features were similar to those without mutations. No mutations were found in the 20 athletes performing more than average weekly exercise, yet all met the criteria for definite or suspected ARVC.

Conclusions In this athletic cohort, we found lower than expected rates of desmosomal gene mutations, particularly among those performing the most exercise. This adds further weight to the hypothesis that an ARVC-like phenotype may be acquired through intense exercise without an identifiable genetic predisposition.

  • Exercise
  • arrhythmogenic right ventricular cardiomyopathy
  • desmosome
  • athlete
  • genetics
  • ventricular arrhythmias
  • arrhythmic right ventricular dyplasia
  • ventricular tachycardia
  • exercise, genetics
  • dyplasia

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Sports-related sudden arrhythmic death is a rare but important entity. It is conceived that exercise acts as a trigger for ventricular arrhythmias on an underlying arrhythmogenic substrate, which in itself may be acquired or inherited. In young athletes (ie, ≤35 years old), hypertrophic cardiomyopathy and coronary anomalies are the most frequently encountered underlying diseases, followed by arrhythmogenic right ventricular cardiomyopathy (ARVC) which is present in 4% to 22% of athletes with sudden death.1 2

It is well known that exertion is related to the development of arrhythmias in familial ARVC.3 4 Sport is not only the trigger for these arrhythmias but may also contribute to the underlying substrate as evidenced by the precipitous development of ARVC in plakoglobin-deficient mice that exercised vigorously.3 Also, in humans, long-term endurance exercise predicts a more severe phenotype in familial ARVC.5 Over the last decade, ARVC has been recognised to be a disease of the desmosome and predominantly of autosomal dominant inheritance. A family history is positive in at least 30%,5–7 and a mutation in one of five different desmosomal components has been characterised, accounting for 27–52% of unrelated index cases.5 6 8–11 Combining these studies, many of which report single gene analyses, an estimated mutation rate of 40% has been calculated.12 Mutations in desmoplakin (DSP), plakophilin (PKP2) and desmoglein (DSG2) are most prevalent, while mutations in plakoglobin (JUP) and desmocollin (DSC2) are rare.

We have previously described our findings in elite endurance athletes presenting with ventricular arrhythmias, of which a vast majority originated from the right ventricle (RV). However, a familial history was present in only one out of 46 cases.13 We also noticed a slight but consistent decrease in RV ejection fraction in athletes with arrhythmias compared to those without arrhythmias or sedentary controls.14 This raised the question as to whether sport had enhanced ARVC expression in patients with an underlying mutation or whether the extreme pressure and volume-load of endurance exercise could, by itself, have led to pro-arrhythmic structural RV changes.

The aim of this study was therefore to systematically analyse the five desmosomal genes for mutations in a cohort of unrelated endurance athletes with complex arrhythmias of RV origin in whom other causes of arrhythmogenic potential had been excluded. Our hypothesis was that if sport were only the promoter of ARVC expression, the proportion of genotype-positive athletes would be at least as high as in series describing familial ARVC probands (ie, 27–52%).5 6 8–11 Conversely, a lower proportion would add support to the concept that long-term RV overload during endurance exercise can lead to a similar phenotype, which could be labelled as “exercise-induced ARVC-like syndrome”.


Patient selection

The following inclusion criteria had to be fulfilled: (1) athletes were defined as those performing at least 3 h/week of sport with a moderate to intense dynamic component (class B or C)15, competitively or recreationally for >5 years; (2) qualifying RV arrhythmias excluding idiopathic right ventricular outflow tract ventricular tachycardia (RVOT-VT), required to be: recorded spontaneously or induced, have a monomorphic left bundle branch block (LBBB) morphology (a) sustained or (b) nonsustained for ≥3 beats at a rate of ≥120 beats per minute, or (c) frequent isolated premature ventricular beats (VPB; ≥2000/day) with a morphology excluding an RVOT origin and/or being associated with findings of structural RV modifications (ie, biopsy with fibro-fatty infiltration; positive late potentials; angiographic RV ejection fraction ≤50%), given the high sensitivity of these findings for underlying RV pathology.14 16 Patients with documented supraventricular arrhythmias were excluded as was any athlete with features suggestive of hypertrophic cardiomyopathy, an ion-channel disorder or coronary vessel disease (congenital or acquired). Our unit is known for its interest in athletes with arrhythmias. We estimate that approximately 25–30% of such athletes from throughout Belgium are referred to our unit for evaluation. Between 1997 and 2007, 129 athletes were evaluated (including an invasive electrophysiology study) at our institution for palpitations of unknown aetiology, (pre)syncope or sudden death. Fifty-three athletes met the study inclusion criteria of which six athletes declined participation. Therefore, the study population consisted of 47 athletes.

A validated detailed questionnaire was used to assess the cumulative number of hours of dynamic sport, performed since the age of 15 years until diagnosis.17 For each type of sport, weekly hours and total number of years for which the athlete participated were recorded. Sports were classified by the Mitchell criteria15 according to their static and dynamic components. In addition, a pedigree was constructed, asking the index about the sports level, clinical status and age/mode of death of all first- and second-degree family members.

All patients gave prior written informed consent, and the study protocol was approved by the local ethical committee.

Clinical evaluation

All athletes underwent thorough electrophysiological and functional assessments. Electrical evaluation included a 12-lead electrocardiogram (ECG) and 24 h ambulatory ECG monitoring (100%), a signal-averaged ECG (SAECG) (74%) and an electrophysiological study (100%). Morphological and functional evaluation of the RV was performed in all cases using one or more modalities—echocardiography (100%), cardiac magnetic resonance (80%) and RV bi-plane ventriculography (59%). A clinical diagnosis of ARVC was based on the McKenna Task Force Criteria (TFC) 18 of two major, one major and two minor, or four minor criteria. In accordance with previous studies,5 11 13 the diagnosis of suspected ARVC was met if ≥2 minor criteria or one major and one minor criterion were fulfilled. An RV ejection fraction (RVEF) of <50% was deemed consistent with mild RV dysfunction (minor criterion), while an RVEF<35% was classified as severe dysfunction (major criterion) regardless of imaging modality.

Mutation screening

Genomic DNA from patients was extracted from whole blood using the Chemagic DNA blood kit (Chemagen, Baesweiler, Germany). Primer pairs for all coding exons of PKP2, JUP, DSP, DSC2 and DSG2 were designed from flanking intronic sequences. Primer sequences and polymerase chain reaction (PCR) conditions are available on request. After amplification, PCR fragments were purified with the ExoSAP-IT mixture (GE Healthcare Ltd, Buckinghamshire, UK) sequenced on an ABI 3130XL Genetic Analyser (Applied Biosystems) in both directions using the BigDye Terminator cycle sequencing kit v3.1 (Applied Biosystems Inc, Foster City, CA, USA) and analysed with the SeqMan II V.4.00 8 software (DNASTAR Inc., Madison, WI, USA). Multiplex Ligation-dependent Probe Amplification (MLPA) was utilised to detect larger duplications and deletions of one or more exons.

All identified sequence variants were classified as mutations if they met the following three criteria: (1) the variant was rare: all polymorphism frequencies were tested against the HapMap database of 120 European chromosomes. Where population data was unavailable, prevalence was checked against 300 chromosomes from a Belgian control group without cardiac symptoms. (2) The variant altered a conserved amino acid. (3) The variant was determined to be of pathogenic potential on analysis with Polyphen ( All other sequence variants were classified as polymorphisms in a manner consistent with previous literature.10 11


Nominal and ordinal data frequencies were compared using χ2. In 2×2 tables, the Fisher Exact test was used given the low frequencies of some variables. Continuous variables were compared using the Student independent samples t test. A p value <0.05 was considered significant. All statistics were computed using commercial software (SPSS V.16.0).


Clinical data

The 47 athletes were aged between 23 and 66 years, and 45 were male. All athletes were unrelated, and only two had a familial history suggestive or suspicious of ARVC (death of a sibling at 29 years in one case and a father with refractory ventricular arrhythmia (VA) of RV origin in another). As detailed in table 1, most athletes were professional, and a majority practised endurance sports (cyclists 72%, distance runners 6%, triathletes 6%, and kayakers 2%) as compared with non-endurance (footballers 9%, dancers and martial arts 4%). On average, athletes performed more than five times the minimum recommended amount of exercise for prevention of cardiovascular disease19.

Table 1

Clinical characteristics of the total cohort (47 athletes)

A clinical diagnosis of definite ARVC according to TFC18 was met in 24 athletes (51%), while an additional 17 athletes (36%) met the criteria for suspected ARVC. In the six remaining athletes (13%), idiopathic RVOT-VT was excluded on the basis that VAs were of more than one morphology of which a LBBB/inferior axis pattern was not dominant.

An ECG abnormality was evident in 24 athletes (51%), six of whom were detected only on signal-averaged ECG, while structural RV abnormalities were detected in 30 athletes (64%).

No athlete died during follow-up. In 26 athletes (55%), a defibrillator (ICD) was implanted. Overall, 15 athletes (32%) had a major arrhythmic event (aborted sudden death, appropriate ICD discharge, recurrence of sustained VT) during 8 (5) years follow-up.

Genetic analysis

We found nine different heterozygous sequence variants in 10 patients (21.2%), seven novel and two previously known, of which five (in six patients, 12.8%) are manifestly pathogenic (table 2). Of these, four affected the PKP2 gene. Large deletions were found in athletes 1 and 46 and, in the latter case, this was only detectable by MLPA. Athletes 9 and 28, who were unrelated, had a previously described splice acceptor site mutation,8 while a missense mutation in athlete 39 was considered of pathogenic potential because it altered a putative phosphorylation site20. The fifth mutation was in JUP and was considered of pathogenic potential because it results in substitution of a strictly conserved amino acid (p.Ala143Thr) in a region of strong phylogenetic conservation. However, this assignment was not supported by subsequent family studies in which two first-degree relatives with the same mutation had normal phenotypes (after ECG, SAECG and echocardiographic evaluation).

Table 2

Summary of desmosomal mutations—pathogenic and unclassified variants

The pathogenicity of the remaining four missense sequence variants is uncertain and likely absent. A c.76→GA nucleotide change in PKP2 was previously described by van Tintelen et al 10 and considered non-pathogenic. For the other three missense variants, the affected amino acids are either moderately or poorly conserved, are not located in known functional domains and undergo only conservative changes. There was no family history consistent with, or even suspicious of, ARVC in any of these four subjects with unclassified variants.

Clinical features of mutation carriers versus others

Mutation prevalence was similar regardless of designation according to Task-Force Criteria—as detailed in figure 1A. Similarly, the incidence of major arrhythmic events leading to presentation, during work-up and in the follow-up period, did not differ according to TFC (figure 1B), suggesting that the inclusion or exclusion of any sub-group would not be expected to influence results.

Figure 1

(A) Prevalence of desmosomal gene mutations in total group (athletes with RV arrhythmias) and ARVC Task Force Criteria sub-groups compared with estimated prevalence from preceding literature. Mutation prevalence was similar between task force criteria sub-groups (p=0.57) and was substantially less than published literature (*references5 6 8–12). (B) Important clinical events at presentation and during follow-up according to Task Force Criteria sub-groups. Significant arrhythmic events are common but are not segregated by the extent of RV abnormalities as determined by Task Force Criteria.

Clinical features of athletes with and without desmosomal mutations was similar. Importantly, athletes without a familial history or desmosomal mutation experienced a major arrhythmic event during follow-up in 28% of cases, a proportion which is not significantly different from those with proven familial ARVC or mutations.

Electrophysiological features were similarly represented between the two groups including epsilon waves which have been regarded as relatively specific for the diagnosis of ARVC and therefore afforded a major criterion in the TFC. While spontaneous arrhythmias were similarly represented between groups, the induction of sustained VT or ventricular fibrillation (VF) on electrophysiological study favoured those with mutations, though the difference was not significant (p=0.20). On the other hand, in the cohort as a whole, those with inducible sustained VT/VF had a higher rate of major arrhythmic events during follow-up (64% vs 5%, p<0.001). As we have previously described, this may be one of the few predictive tests in this patient group.13

While morphological features were also similar in the majority of instances, severe RV dysfunction was more common among athletes with desmosomal mutations (33% vs 3%, p=0.02). Similarly, RVEF was lower. However, mild and moderate RV dysfunction was at least as common in those athletes without desmosomal mutations—examples of which are illustrated in figure 2 and as supplementary video files. Endomyocardial biopsies were performed in only 10 athletes (22%), which precludes definitive conclusions. However, fatty infiltration was not observed in any subject, whereas the finding of fibrosis (with or without evidence of myocyte degeneration) was evident in three (30%) of the biopsies—one athlete with a desmosomal mutation and two athletes with a normal genotype (table 3).

Figure 2

An example of moderate right ventricular dysfunction in a competitive cyclist without a desmosomal gene mutation. MRI demonstrating RV dilatation, mild septal flattening and moderately reduced systolic function. There is no evidence of myocardial fatty infiltration. (RVEF, right ventricular ejection fraction; LVEF, left ventricular ejection fraction).

Table 3

Clinical findings according to desmosomal genotype

The six athletes with desmosomal mutations performed significantly less exercise per week when compared with the remaining athletes (figure 3). There were no desmosomal gene mutations found among the 20 athletes who practised more than average weekly exercise, yet all had RV abnormalities amounting to categorisation as definite (12, 60%) or suspected ARVC (6, 40%).

Figure 3

Weekly exercise in athletes with and without desmosomal mutations. Weekly exercise is significantly greater in those athletes without a desmosomal gene mutation, 14.3 (8.9) vs 9.2 (2.3), p=0.005.


Prevalence of desmosomal mutations in athletes presenting with VA

Pathogenic desmosomal mutations were detected in 12.8% of our cohort of athletes presenting with complex RV arrhythmias. This is substantially less than would be expected if exercise were solely acting as a promoter of an underlying genetic predisposition. Previous studies have noted a 27–52% prevalence of mutations in the desmosomal genes of patients with a clinical diagnosis of ARVC.5 6 8–11 Moreover, we performed a comprehensive assessment of all five desmosomal genes, whereas three of the quoted studies reported high mutation prevalence in the PKP-2 gene alone. In addition, the use of Multiplex Ligation-dependent Probe Amplification identified a subject which would not have been detected in the methodology of any of the preceding studies. Lastly, we studied a population with Dutch ancestry in whom one of the highest mutation prevalence rates has been described.10 Therefore, our study represents the most comprehensive assessment to date, yet the detection of desmosomal gene mutations was low. It may be argued that the prevalence in our cohort is diluted by the inclusion of athletes without strict criteria for ARVC. However, mutation prevalence was low in all TFC groups (figure 1A) and lower than that reported in two recent studies which also evaluated suspected ARVC.11 12 Furthermore, we demonstrate that athletes remain at just as significant a risk of major arrhythmic events despite having lesser degrees of RV abnormalities and lower TFC scores (figure 1B). Finally, our underlying hypothesis is that this syndrome of RV arrhythmias in athletes is not explained by our current understanding of familial ARVC.

We previously reported a high prevalence of morphological and electrophysiological abnormalities in the RV of athletes presenting with ventricular arrhythmias.13 This raised two predominant hypotheses regarding aetiology; first, that exercise was acting as a promoter for ARVC and, second, that exercise may cause structural and electrical remodelling in its own right. Corrado et al1 reported that sudden death was attributable to ARVC more frequently in athletes than in non-athletes (relative risk 5.4, p<0.0001). This would suggest that sport practise might represent a relatively specific “stress test” for exposing underlying ARVC. Thus, in symptomatic athletes, a higher mutation prevalence than that described for non-athletic cohorts might be expected if ARVC were the underlying pathology. Conversely, the finding of a low prevalence of mutations should prompt consideration of alternative explanations. Similarly, a clinical history of familial disease may be expected in many, if not a majority, of subjects with ARVC.5–7 However, evidence of familial disease, both genetic and on history, is rare in our cohort of athletes leading to speculation that exercise may cause an acquired ARVC-like syndrome in its own right.

Classification of genotype pathogenicity

As noted by previous authors, in the absence of functional assays, the pathogenicity of desmosomal mutations is sometimes difficult to determine. Thus, the significance of some uncommon polymorphisms or “unclassified variants” remains speculative. We used the same criteria as previous studies to exclude polymorphisms.10 11 Moreover, a retrospective search for pre-symptomatic relatives, which is ongoing in our cohort, has identified first-degree relatives with ECG and/or RV abnormalities suggestive of ARVC in three of four of the pathogenic mutations involving deletions or splice site mutations but in none of the six missense mutations. We classified two of these missense mutations as pathogenic despite the absence of confirmation on family evaluation. The variable penetrance associated with ARVC may explain these findings, but the weight of evidence could support these being non-pathogenic variants, in which case the prevalence of disease-causing mutations would be even lower (8.5%), further strengthening the main conclusion of this study.

Similarities and differences between athletes with and without desmosomal gene mutations

Prognosis in our cohort (as defined by major arrhythmic events during 8 (5) years follow-up) could not be predicted by TFC classification or genotype. However, severe RV dysfunction was more common among those with a mutation. Screening for desmosomal mutations may be most applicable in this group. At the other end of the spectrum, our cohort included 17 athletes (36%) in whom imaging of the RV was normal. RVOT-VT was excluded on the basis of VPB morphology and/or additional findings of RV proarrhythmogenic remodelling. Difficulties in separating benign idiopathic RVOT-VT from ARVC have been raised before.16 21 Corrado et al16 used electro-anatomical mapping and guided biopsies to differentiate the two entities and reported abnormalities in 22% of 27 patients with “RVOT-VT”, increasing to 87% if confined to those guided by mapping. Such techniques were not routinely employed in this study except in the last five athletes in whom none had low-voltage areas.

We obtained septal biopsies in 10 athletes of whom three had evidence of fibrosis. The absence of fatty infiltration may reflect differences between exercise-induced remodelling and familial ARVC. This finding adds to speculation that sub-clinical fibrosis may be caused by intense endurance exercise and represents a pro-arrhythmic substrate, as proposed in case studies following arrhythmic arrest and sudden death in well-trained athletes.22 23

Endurance athletes and familial ARVC—convergence to a common phenotype?

Given the low prevalence of clinical and genetic evidence of familial ARVC in this cohort, we are left with the question as to why athletes might develop such changes. Our premise that intense exercise may contribute is supported by the finding that those athletes without a desmosomal mutation performed more exercise than those with a mutation. In fact, among athletes with RV arrhythmias who performed more than the mean weekly exercise (14 h), none had a desmosomal gene mutation. Current understanding of familial ARVC suggests that desmosomal alterations render them susceptible to mechanical stress. It has been demonstrated that the RV is subject to relatively greater increases in afterload during exercise when compared with that of the left ventricle (LV),24 which may be expected to translate as a greater proportional wall stress. A similar predisposition to RV over LV injury has been noted after endurance sporting events.25 26 Therefore, it stands to reason that the increased mechanical stress imposed by exercise would accelerate ARVC disease expression, and, thus, those with a desmosomal mutation would require less exercise to precipitate symptoms. As an extension on this line of reason, we propose that “extreme doses” of exercise may be sufficient to weaken the interstitial matrix of the myocardium in the absence of an established genetic risk.


Our small cohort, despite being the largest reported, means comparisons of outcomes between those with and without mutations is prone to both over-estimation and under-estimation of differences. Second, as with all genotype/phenotype associations, the significance of unclassified variants is difficult to determine with certainty. In assigning pathogenic potential, we erred on being inclusive such that our estimated prevalence may be overestimated. We did not investigate first-degree relatives of athletes who had a negative pedigree and genotype. Some additional familial cases may have been identified with this approach.


Our data indicate that, under the umbrella of the TFC-positive ARVC phenotype, there are two groups: one characterised by a high prevalence of clinical and genetic evidence of heredity and another characterised by a high volume of intense, endurance sport practise and little or no evidence of a familial predisposition. The two groups may represent a spectrum from familial ARVC to “exercise-induced right ventricular cardiomyopathy”. The cohort that we describe here may have a “mild genetic risk” (such as polymorphisms and/or unrecognised genes) which evokes an ARVC phenotype only if combined with intense endurance exercise.


Supplementary materials


  • Linked articles 195172

  • Funding The project was supported by a grant from the Jean Standaert Foundation. ALaG is supported by a research scholarship from the Australian National Health & Medical Research Council and the National Heart Foundation.

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the University Hospital, Catholic University of Leuven, Belgium.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles