The study population included 111 patients from three tertiary-care centers in Switzerland with a definite (n = 80), borderline (n = 19), and possible (n = 12, all first degree family members of patients with confirmed ARVC/D) diagnosis of ARVC/D according to the 2010 TFC, who had an ECG recorded between February 1987 and March 2013 [7]. This baseline ECG was the first available ECG. Clinical information regarding demographics and symptoms were obtained from hospital records at the time of the baseline ECG. Coronary ischemia was ruled out in the presence of precordial TWI. If patients with precordial TWI presented with typical angina, or were >35 years old and presented with ventricular tachycardia or arrhythmogenic syncope, coronary angiography was performed to rule out significant coronary artery disease. In all other patients non-invasive tests such as cardiac MRI, myocardial perfusion scintigraphy, stress TTE or treadmill stress testing were performed to rule out coronary ischemia. This study was approved by the Ethics Committee of the Canton of Zurich and the Ethics Committee of the Canton of Bern, Switzerland, and has been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. Patients gave written informed consent for prospective inclusion in this study.

ECG and transthoracic echocardiography (TTE) definitions and criteria used in this study are presented in Table 1.

ECGs were recorded at rest (25 mm/s, 10 mm/mV amplitude) with standard lead positions, digitized with a high-resolution scanner, and analyzed with a digital caliper (Iconico; screen caliper version 4.0; http://​www.​iconico.​com)[15]. To increase the accuracy of measurements, all ECGs were enlarged four times. ECG intervals were measured in two consecutive sinus beats in each lead; the mean value of the two beats was used. If the difference between the two beats was >10 ms, then the mean of three beats was taken. Each ECG was independently analyzed by two blinded experienced readers. Differences in the interpretation of the ECG parameters were adjudicated by a third cardiologist, and a final conclusion was made by consensus. ECG variables were measured as previously described [13].

Conventional M-mode, 2-D, and color Doppler echocardiography was performed in 111 patients at the time of the baseline ECG by experienced cardiologists according to guidelines [16, 17]. Left ventricular (LV) involvement was diagnosed when regional wall motion abnormalities or a reduced ejection fraction (EF <50% by Simpson’s biplane method) were present, and other causes were excluded [18]. Some ECG data from 90 patients and TTE data from 70 patients used in this study has been previously reported by our group [11, 13]. While we focused on predictors of adverse outcome in these previous studies, the current study investigates ECG changes during long-term follow-up, and their associations with TTE findings. Thus, data on follow-up ECGs have not been previously reported by our group.

Continuous variables are presented as mean ± SD or median (with interquartile ranges, IQR), and were compared using a two-sided t-test or Mann–Whitney U-test, as appropriate. Categorical variables are reported as frequency (percentage) and compared between groups by Fisher’s exact test or χ² –test of independence, as appropriate. Logistic univariate regression analysis was used to correlate between TWI in inferior leads as the response, and number of precordial leads with TWI as the explanatory variable. For comparison of ECG data between baseline and follow-up, we used the Wilcoxon signed-rank test for continuous variables, and Mc Nemar’s test for categorical variables. A two-sided p-value of ≤0.05 was considered significant. Statistical analysis was performed using R programming language (R Development Core Team, 2009) and GraphPad Prism 5 (GraphPad Software Inc., La Jolla, CA, USA).

Patient baseline characteristics are summarized in Table 2. A baseline ECG was available in 111 patients. TTE at the time of the baseline ECG was performed in all 111 patients. Ninety-eight patients were index patients, and 13 were family members. Sixty-four (58%) patients had an implantable cardioverter-defibrillator. ECG did not display abnormal findings in 14 patients (13%). Epsilon-waves were most frequently observed in V1 (14%). Epsilon-like waves in any lateral precordial lead (V4, V5, or V6) were present in a minority of patients (5%), with one patient displaying these potentials in V2 through V6. Complete right bundle branch block (RBBB) was observed in 14 (13%), incomplete RBBB in 3 (3%), complete left bundle branch block in 2 (2%), left anterior fascicular block in 7 (6%), and unspecific QRS notching in 29 (26%) patients. On echocardiography, regional wall motion abnormalities were most commonly observed in the RV subtricuspid region (n = 66; 60%).

A follow-up ECG was available in 77 patients. In the remaining 34 patients, a follow-up ECG was not obtained due to short follow-up periods (n = 14) or loss of follow-up/heart transplantation/death (n = 20). The median time between baseline and follow-up ECG was 4 years (IQR 1.9-9.2 , range 0.3–21.8 , mean 5.6 years). The follow-up ECG was considered normal in 9 patients (12%). On average, ECG findings were progressive during follow-up, although individual disease course showed some variations. This progression was statistically significant for major depolarization criteria, which was particularly attributed to an increase in patients with epsilon waves in V2 at follow-up, and an increase in the maximum QRS duration (Table 3). More patients had TWI in the precordial leads at follow-up. Out of 8 patients without any precordial TWI at baseline, 4 (50%) later displayed precordial TWI. An illustrative diagram of electrocardiographic de- and repolarization abnormalities at baseline and their changes during follow-up is shown in Figure 1. Three patients with minor repolarization criteria at baseline did not have any repolarization criteria at follow-up. Four patients with major repolarization criteria at baseline had minor repolarization criteria in the absence of complete RBBB at follow-up (Figure 2), whereas two patients with major repolarization criteria displayed minor repolarization criteria in the presence of complete RBBB at follow-up. Repolarization criteria were completely absent in three patients, who had major repolarization criteria at baseline (Figure 3).

From the patients without epsilon waves at baseline, 19% had these abnormalities at follow-up. In all patients displaying epsilon waves at baseline these potentials also remained at follow-up. In one patient, follow-up ECG revealed a dynamic Brugada-like ECG pattern (Figure 4). Epsilon-like waves in V4, V5 or V6 were present in a higher proportion of patients at follow-up compared to baseline (9%). Median terminal activation duration (TAD) at follow-up was similar compared to baseline. An increase in the number of precordial leads with TWI was particularly attributed to new TWI in leads V4 and V5 at follow-up compared to baseline (49% vs. 36% for V4, and 38% vs. 26% for V5, respectively). At follow-up, complete RBBB was observed in 12 (16%), incomplete RBBB in 2 (3%), complete left bundle branch block in 2 (3%), left anterior fascicular block in 5 (6%), and unspecific QRS notching in 23 (30%) patients. The maximum corrected QT interval did not significantly differ between baseline and follow-up.

TTE data was available in all 77 patients at the time of the follow-up ECG. Echocardiographic data also indicated disease progression, and correlated with the ECG findings. Accordingly, LV involvement was present in 20 patients at follow-up (26% vs. 14% at baseline, p = 0.04). At follow-up, 15 (19%) patients remained with only one RV region involved, whereas 48 patients (62%) had ≥2 RV regions involved (p = 0.14 compared to baseline).

TWI in inferior leads were significantly associated with major repolarization criteria (p = 0.02), extensive RV involvement (p = 0.04), the presence of TWI in either lateral precordial lead (V4, V5 or V6; p = 0.05), and the presence of definite ARVC/D (p = 0.05), whereas the presence of echocardiographic LV involvement (p = 0.11) did not reach statistical significance. Regional wall motion abnormalities of the subtricuspid region were not significantly associated with TWI in inferior leads (p = 0.16). Univariate logistic regression revealed a highly significant incremental association between the absolute number of precordial leads with TWI and TWI in inferior leads (odds ratio 1.45, 95% confidence interval 1.16 - 1.83, p = 0.001).

Except for one study [19], four previous studies have suggested that ECG changes are generally progressive in patients with ARVC/D. Yet, the extent and speed of disease progression shows some individual variability that may be based on genetic and environmental factors [9‐11, 20‐22].

Recent data suggests that electrocardiographic diagnostic features of ARVC/D such as epsilon waves and TWI may be dynamic and transient [8], which may potentially impede appropriate diagnosis of ARVC/D at certain time points during the disease course, particularly at early stages.

This study focused on patients with less advanced disease in comparison to previous studies analyzing data from patients with more advanced forms of ARVC/D, and underscores that disease progression can be visualized by consecutive ECG tracings, although the disease course shows some individual variations. Importantly, the percentage of patients with epsilon waves and average QRS duration significantly increased over time, and the increase in the number of precordial leads with TWI showed a tendency towards significance. With regard to TWI, the percentage of patients with precordial TWI beyond V3 and TWI in inferior leads increased, and correlated well with more extensive echocardiographic RV involvement and new onset LV involvement at follow-up. ARVC/D has been recently recognized as a biventricular disease with early LV involvement [23]. With respect to this, repolarization abnormalities beyond V1-V3, but also inferior lead TWI may serve as early ECG markers of biventricular involvement and widespread disease. The fact that depolarization abnormalities appear to better reflect disease progression compared to TWI is not completely clear. It is possible that repolarization abnormalities are not only caused by structural myocardial changes, but may also constitute a physiological variant [14], may vary depending on exercise status of a given patient [24], be caused by prolonged bouts of paroxysmal tachyarrhythmia or frequent ventricular premature contractions, i.e. T wave memory [25], and subtle differences in positioning the precordial electrodes, i.e. ECG reproducibility [26]. Anatomic predisposition, particularly in women, may affect lead positioning, and thus alter the presence and distribution of leads with TWI (Figure 3). Of note, in this study all four patients with major repolarization criteria at baseline, who presented with minor repolarization criteria in the absence of complete RBBB at follow-up, were women [8]. We were also able to document a dynamic Brugada-like ECG pattern in leads V1/V2 in one patient, as previously reported, although the criteria for Brugada Type I ECG were not fulfilled in the absence of J-point elevation ≥2 mm and TWI in V2 [27, 28]. Interestingly, this patient was one out of 13, in whom TWI in inferior leads were documented in the absence of major and minor repolarization criteria (34% of patients with TWI in inferior leads). In contrast to the study by Quarta et al. [8], dynamic ECG changes in our study primarily originated from changes in repolarization rather than depolarization abnormalities. Furthermore, as opposed to that previous study, all three patients with major repolarization criteria at baseline and absence of any repolarization criteria at follow-up had advanced forms of ARVC/D, suggesting that dynamic repolarization abnormalities may not only affect early forms of the disease.

Based on our data, depolarization abnormalities seem to be far less prone to reversal, because they probably better reflect myocardial infiltration by fibro-fatty tissue. This interpretation is further supported by our data on QRS duration, which also significantly increased during follow-up. As the proportion of patients with antiarrhythmic medication were not significantly different at baseline and follow-up, and TAD did not change over time, this increase can be attributed to a progression in depolarization abnormalities also affecting early depolarization such as bundle branch block and QRS notching. As such, the percentage of patients with RBBB and QRS notching increased during follow-up.

The presence of TWI in lateral leads in patients with ARVC/D is well established, and a noninvasive indicator of biventricular involvement and severe RV dilation [29‐32]. However, data about TWI in inferior leads in ARVC/D is scarce and their role not completely understood. Whereas most studies support the concept that TWI in inferior leads are a marker of progressive disease with extensive RV involvement and LV involvement [31, 32], other studies suggest that such ECG changes could also be associated with isolated wall motion abnormalities of the subtricuspid region. Steriotis et al. have recently reported a considerable prevalence of TWI in inferior leads in ARVC/D, particularly in the presence of severe RV dilation, but did not report on the prevalence of LV involvement in their cohort. Our study confirms these previous findings with more than one third of patients displaying TWI in inferior leads at baseline – far more than depolarization abnormalities in these leads. We further add to the concept that inferior lead TWI in ARVC/D are associated with extensive disease as reflected by a significant association with definite disease status, extensive RV involvement, and TWI in lateral precordial leads, and a tendency towards significance for LV involvement. On the contrary, we did not observe a significant association between inferior lead TWI and isolated regional wall motion abnormalities of the subtricuspid region. TWI in inferior leads could even constitute markers for widespread biventricular disease, especially in the presence of major repolarization abnormalities implying that TWI in inferior leads may have a rather non-benign character. Based on our findings, we believe that ECGs obtained once yearly can serve as an informative, cheap and readily available monitoring tool. In case of progressive ECG findings, further more expensive diagnostic tests such as echocardiography, cardiac MRI or Holter monitoring can be performed, and follow-up strategies can be refined. In the absence of ECG changes and a clinical deterioration during follow-up, in our routine clinical practice we perform TTE every 2 years, and cardiac MRI every 2–5 years to monitor disease progression. Yet, as an isolated diagnostic test ECG may not drive therapy per se, and ECG findings have to be incorporated in the whole clinical picture, especially when arrhythmic episodes are lacking. Accordingly, as a sole diagnostic method, ECG is not sensitive and specific enough to establish the diagnosis of ARVC/D. Therefore, other diagnostic tests including imaging (e.g. cardiac MRI with stress testing), genetic testing, family history and endomyocardial biopsy are necessary and required by the TFC. Since TWI in the precordial leads are not very specific for ARVC/D, e.g. are commonly seen in patients suffering from coronary artery disease immediately after a coronary ischemic episode, and can persist for several weeks thereafter, it is important to exclude coronary ischemia by diagnostic tests such as coronary angiography or cardiac MRI.

Due to the observational and partially retrospective nature of this study, echocardiographic data was mainly obtained from previous reports. The duration of individual follow-up and lack of a follow-up ECG in some patients is a limitation that may have led to an underestimation of ECG changes. Variable ECG filtering may have affected the detection of epsilon waves, but not the detection of TWI, which was the main reversible ECG feature found in our cohort. Due to the retrospective study design our ECGs did not include right-sided chest leads (V1R, V2R, V3R, V4R), which could be included in future prospective studies investigating the diagnostic and prognostic utility of ECG in ARVC/D. Baseline and follow-up cardiac MRI were only available in a minority of patients, and thus associations between ECG and MRI findings could not be addressed in this study.