Exercise does not influence development of phenotype in PLN p.(Arg14del) cardiomyopathy

The study population was recruited from 3 Dutch university medical centres (Amsterdam, Groningen and Utrecht). Subjects who were a carrier of PLN p.(Arg14del) and ≥ 18 years of age were invited by a letter to participate in a detailed interview to document exercise history or could self-apply via the website of the PLN Patient Foundation (www.​plnheartdiseasef​oundation.​org). The study protocol was exempted from approval by the Medical Ethics Committee of the Amsterdam University Medical Centres because the Act of Medical Research involving Human Subjects did not apply (W17_076 #17.094).

Genotype was derived from sequencing of PLN in a diagnostic setting [2]. Clinical and demographic data were drawn from the ACM registry [16].

Clinical presentation was defined as a first visit for PLN p.(Arg14del) related complaints or family screening. We used 3 clinical endpoints: a clinical diagnosis of ARVC or DCM, sustained VA and hospitalisation for HF [17]. Sustained VA was defined as ventricular tachycardia of > 30 s or < 30 s when terminated electrically (including appropriate implantable cardioverter-defibrillator (ICD) therapy) or pharmacologically; ventricular fibrillation (VF); and/or resuscitated or aborted cardiac arrest, as described previously [7]. When an individual reached one of these clinical endpoints, he/she was defined as having a cardiac phenotype and considered penetrant for disease. ARVC diagnosis was assigned according to the 2010 Task Force Criteria [18]. DCM diagnosis was based on decreased left ventricular ejection fraction (LVEF) and increased left ventricular volume (LVEDD > 112% of predicted corrected for age and body surface area, and a reduced LV function [EF < 45% or fractional shortening < 35% on echo] or cardiac MR with LVEF < 45% and left ventricular end-diastolic volume > 2 SD above reference value) [17, 19]. Risk factors for malignant ventricular arrhythmia or HF, previously established in PLN p.(Arg14del), were also collected: LVEF < 45%, (non‑)sustained ventricular tachycardia (VT), > 500 PVC count/24 h, number of leads showing negative T leads and presence of low-voltage ECG [6, 7].

Exercise history was obtained by structured telephone interviews. Participants were prompted to list regular exercise done since the age of 10 years for leisure including sports, for work, and for transportation (by bicycle). Details of collection of intensity and duration of each activity have previously been described [9]. All responses were transcribed on pre-prepared data collection sheets.

For each participant, type and intensity of the activities were expressed in metabolic equivalent (MET) using the Ainsworth compendium [20] (FvL). This MET-score reflects the energy expenditure in comparison to a resting state [21]. By multiplying the frequency and duration of each activity with the MET-values, an estimation of annual MET-hours was obtained. We calculated exercise dose (MET-hours/year) preceding and following clinical presentation.

Differences in survival between individuals when split by median or quartiles of annual MET-hours of activity were evaluated with the log-rank test. Cox proportional hazard models with a Firth correction for rare outcomes were used to test the association of exercise prior to clinical presentation and outcomes. Known risk factors for PLN p.(Arg14del) carriers were selected as covariates [6, 7]. These covariates at clinical presentation, were controlled for using different models based on the hypothetical relations between variables. Model 1 adjusts only for age and sex, whereas model 2 consists of all covariates. Missing values were accounted for by multiple imputation (R-package MICE; 20 imputation sets with each 10 iterations). Results were pooled over the imputation sets using Rubin’s rules. Restricted cubic spline analyses were performed to further assess the association between mean MET-hours/year until presentation as a continuous variable and risk of outcomes. A two-sided p-value of < 0.05 was considered significant. Statistical analyses were performed using SPSS (version 25.0; IBM Corp., Armonk, NY, USA), and RStudio (version 1.1.383; RStudio, Boston, MA, USA).

Out of 590 adults from the PLN Registry (April 2017), 346 (58.6%) agreed to be interviewed, of whom 75 were self-referred via the PLN Foundation’s website. Of them, 299 (86.4%) were interviewed, of whom 207 (69.2%) had an adequate amount of clinical information available to determine phenotype and clinical outcomes with a median follow-up duration of 5.9 years.

Tab. 1 summarises the clinical characteristics. A minority were male (39.1%) and most presented because of family history (65.7%). At last follow-up 18% had experienced a sustained VA and more than 9% had been hospitalised for HF. Nine participants experienced both a sustained VA and hospitalisation for HF at last follow-up.

Exercise history before and after clinical presentation are shown in Tab. 2. As can be appreciated, this was not a particularly athletic cohort. Prior to presentation, the study population participated in activities with a median of 1661 MET-hours/year, which included leisure activities, such as sports, with a median of 490 MET-hours/year (9.4 MET-hours/week). This equals walking for pleasure for 2.5 h per week. The Dutch Health Council advises adults to exercise at a moderate intensity for half an hour 5 days per week at a minimum [22], which is met by half of the Dutch population [23].

Figure 1 shows median MET-hours/year before presentation in individuals with and without: (a) a sustained VA, (b) HF hospitalisation, (c) any cardiac phenotype during follow-up (including sustained VA, HF, ARVC or DCM diagnosis), (d) ARVC diagnosis, (e) DCM diagnosis and (f) any cardiomyopathy diagnosis at last follow-up. For all outcomes, there was no significant difference in amount of exercise (MET-hours/year) prior to presentation. Tab. 1 shows that there were no differences in clinical outcome when comparing individuals participating in the 50% higher mean MET-hours/year with those in the lower 50% mean MET-hours/year until presentation. This includes a similar frequency of sustained VA (18.3% vs 18.4%; p = 0.974) and hospitalisation for HF at last follow-up (9.6% vs 8.7%; p = 0.827). There were also no differences in clinical outcome characteristics when comparing individuals from the 25% highest mean MET-hours/year with those in the 25% lowest mean MET-hours/year until presentation (data not shown). Figure S1 in the Electronic Supplementary Material shows similar results when looking only at the median MET-hours/year spent on leisure activities, including sports.

Thirty-eight individuals (18.4%) had experienced a sustained VA at last follow-up, including 11.6% (n = 24) with a sustained VT/VF and 6.8% (n = 14) with appropriate ICD therapy. We also asked participants on the circumstances of their first arrhythmia. Almost half (n = 12/25; 48%) were physically active (e.g. bicycling, gardening).

We then looked at survival free from clinical outcomes during follow-up. Nine patients were excluded due to insufficient duration of follow-up. Tab. 3 shows the baseline characteristics of this subpopulation, and Table S1 in the Electronic Supplementary Material lists the univariate and multivariate analyses of survival free from VA and HF hospitalisation. There was no difference between the active and inactive groups in survival free from new sustained VA or HF hospitalisation during follow-up (Fig. 2a, panels 1 and 2) or from diagnosis of ARVC or DCM (baseline or follow-up) (Fig. 2a, panels 3 and 4). There was also no difference between the quartiles of activity in survival free from diagnosis of ARVC or DCM (baseline or follow-up) or new sustained VA or hospitalisation for HF during follow-up (see Figures S2 and S3 in Electronic Supplementary Material).

Next, we used Cox regression with a Firth correction for rare outcomes to assess the independent effect of exercise (MET-hours/year) controlling for clinical risk factors on survival free from sustained VA or HF hospitalisation (see Table S1 in Electronic Supplementary Material). Model 1 showed that when controlling for sex and age of presentation, there was no association with the amount of exercise (MET-hours/year) and either sustained VA or HF hospitalisation. Model 2, which includes clinical covariates, likewise showed that sustained VA and HF hospitalisation were not associated with the amount of exercise. Restricted cubic splines showed non-linearity, with model 1 showing p = 0.242 for HF hospitalisation and p = 0.177 for VA and model 2 showing p = 0.308 for HF hospitalisation and p = 0.391 for VA (see Figure S4 in Electronic Supplementary Material). The relationship between mean MET-hours/year until presentation and all outcomes showed in all curves a bell-shaped relationship with the top right after the median of 1660 mean MET-hours/year until presentation.

Figure 2b and S3 (in Electronic Supplementary Material) show the association between exercise and life-time penetrance when split by median MET-hours/year and quartiles of MET-hours/year respectively. These figures highlight that the amount of exercise prior to presentation had no effect on the penetrance of an arrhythmic phenotype, hospitalisation for HF or ARVC/DCM diagnosis.

Most participants (64.7%) decreased their annual exercise duration and intensity after first presentation (Tab. 2). This was significant for both MET-hours and for total hours spent on any activity (p < 0.001). More than half (n = 131; 63.2%) of the study population decreased their MET-hours after presentation, and most also decreased their time spent on being active (n = 127; 61.4%). In contrast, 63 individuals (30.4%) increased their level of activity in MET-hours/year and 68 (32.9%) individuals increased their time spent on activity.

Exercise restriction is advised for ARVC patients and those with a (likely) pathogenic desmosomal gene variant and this advice is generally also given to PLN p.(Arg14del) carriers with ARVC. To assess the independent effect of decreasing exercise after presentation controlling for the same risk factors as previously, we used Cox regression. The univariable and multivariable analyses (see Table S2 in Electronic Supplementary Material) show similar results as presented in supplementary Tab. 1. After controlling for covariates, both model 1 and 2 showed no association with decreasing amount of exercise after presentation and either sustained VA or HF.

The relationship between athlete status and the ARVC-phenotype has been studied extensively. Currently, both definite ARVC patients and those with a (likely) pathogenic variant in a desmosomal gene, are discouraged from competitive or frequent high-intensity endurance exercise [8].

An association between exercise and the risk of VA, HF and diagnosis of ARVC in desmosomal variant carriers has been established by multiple studies [9, 24]. In addition, for TMEM43 p.(Ser358Leu) related ARVC, ≥ 9.0 MET-hours/day in the year prior to ICD implantation was associated with an 9.1-fold increased hazard of first appropriate ICD discharge [25]. Moreover, even in gene-elusive ARVC patients exercise has been shown to have a negative effect by modifying cardiac structure and promoting arrhythmias [26]. However, exercise in Dsp haploinsufficient mice partially restored dysregulated gene expression without changing cardiac function, showing that the detrimental effect of exercise is not that straightforward [27].

For a yet unaffected, asymptomatic, PLN p.(Arg14del) carrier, data from the current study strongly suggest there is no reason to limit exercise if this is of limited-moderate intensity (i.e. 6 METs or less) [8]. This contrasts with desmosomal gene carriers and might be explained by the fact that this PLN cohort was not particularly active or that different pathophysiological mechanisms eventually lead to disease. For unaffected asymptomatic PLN p.(Arg14del) carriers that want to exercise more intensively, and thereby fulfil the definition of an athlete (≥ 18 MET-hours/week), the uncertainty of an effect of more intensive exercise has to be discussed and participation should be a shared decision.

For an affected individual, the data also suggest that mild-moderate exercise as reflected by the number of MET-hours does not negatively influence disease course. However, it has been observed that incident malignant arrhythmias in PLN p.(Arg14del) are often associated with exercise. Previously, it was shown that 74% of malignant VA in PLN p.(Arg14del) variant carriers occurred during or shortly after exercise, independent of the intensity [7]. Additionally, out of 19 individuals of whom the circumstances of SCD were known, 13 (68%) were exercising. Similarly in our cohort, those who experienced a VA and could recall their activity at their first arrhythmia, almost half (n = 12/25; 48%) said they were active. Thus, for the patient already displaying risk factors related to the development of an arrhythmia (decreased LVEF, a prior sustained or non-sustained VT, low-voltage ECG, the number of leads with negative T waves or PVC count), and for those fulfilling 2010 Task Force Criteria for ARVC [18], it is reasonable to discuss exercise restrictions and make a shared decision on the intensity of participation [8]. In short, it appears that exercise can be an arrhythmic trigger in PLN cardiomyopathy patients but is likely not associated with disease development or progression.