Right ventricular strain rate during exercise accurately identifies male athletes with right ventricular arrhythmias

Abstract

Aims

Athletes with right ventricular (RV) arrhythmias, even in the absence of desmosomal mutations, may have subtle RV abnormalities which can be unmasked by deformation imaging. As exercise places a disproportionate stress on the right ventricle, evaluation of cardiac function and deformation during exercise might improve diagnostic performance.

Methods and results

We performed bicycle stress echocardiography in 17 apparently healthy endurance athletes (EAs), 12 non-athletic controls (NAs), and 17 athletes with RV arrhythmias without desmosomal mutations (EI-ARVCs) and compared biventricular function at rest and during low (25% of upright peak power) and moderate intensity (60%). At rest, we observed no differences in left ventricular (LV) or RV function between groups. During exercise, however, the increase in RV fractional area change (RVFAC), RV free wall strain (RV_FWSL), and strain rate (RV_FWSRL) were significantly attenuated in EI-ARVCs as compared to EAs and NAs. At moderate exercise intensity, EI-ARVCs had a lower RVFAC, RV_FWSL, and RV_FWSRL (all P < 0.01) compared to the control groups. Exercise-related increases in LV ejection fraction, strain, and strain rate were also attenuated in EI-ARVCs (P < 0.05 for interaction). Exercise but not resting parameters identified EI-ARVCs and RV_FWSRL with a cut-off value of >−2.35 at moderate exercise intensity had the greatest accuracy to detect EI-ARVCs (area under the curve 0.95).

Conclusion

Exercise deformation imaging holds promise as a non-invasive diagnostic tool to identify intrinsic RV dysfunction concealed at rest. Strain rate appears to be the most accurate parameter and should be incorporated in future, prospective studies to identify subclinical disease in an early stage.

Introduction

The differentiation between physiological cardiac remodelling due to intensive endurance exercise and maladaptive remodelling in endurance athletes (EAs) with a propensity for complex ventricular arrhythmias can pose a difficult clinical conundrum. Amongst EAs, ventricular arrhythmias often originate from the right ventricle.¹ The degree of right ventricular (RV) dysfunction at rest in athletic cohorts with the so-called exercise-induced or gene-elusive arrhythmogenic right ventricular cardiomyopathy (ARVC) is often mild despite the association with life-threatening arrhythmias.² This results in diagnostic overlap as also healthy EAs typically develop substantial RV remodelling characterized by chamber dilation and a low-normal ejection fraction.³

Cardiac imaging during exercise has been proposed to improve risk stratification amongst athletes at risk for complex RV arrhythmias.⁴ Using exercise cardiac magnetic resonance (exCMR), we demonstrated that exercise evaluation can unmask RV contractile dysfunction in athletes with RV arrhythmias despite there being no abnormalities at rest.⁵ However, widespread use of exCMR is constrained and exercise echocardiography may be an attractive alternative because of its availability and high cost-effectiveness. Recently, more advanced echocardiographic modalities based on deformation imaging have been proposed as a means to increase diagnostic power, but so far these have not been evaluated during exercise.⁶

Therefore, we sought to compare RV performance during bicycle exercise echocardiography in healthy non-athletic controls (NAs), healthy EAs, and subjects with exercise-induced ARVC (EI-ARVCs) with the aim of identifying the most accurate non-invasive echocardiographic parameter for differentiating athletes with underlying arrhythmogenic RV pathology. We hypothesized that deformation imaging during exercise would more accurately identify subjects with intrinsic RV myocardial disease as compared with standard resting and exercise measures.

Methods

Subjects

In the current study, EI-ARVC subjects that were included in a previous study were compared with an expanded cohort of EAs and NAs. Therefore, some of the data from the current cohort has been published previously, but the current study represents completely new measures of RV function during exercise and image data was analysed independently.⁵ For the recruitment and inclusion of EI-ARVCs, the following criteria were used: (i) previous or current competitive sports participation and intensive training (>6 h week); (ii) presence of RV arrhythmias (excluding idiopathic RVOT-VT) with monomorphic left bundle branch block morphology (either sustained, non-sustained for ≥3 beats at a rate ≥120 bpm, or frequent isolated premature ventricular beats ≥2000/day); (iii) at most mild structural and functional abnormalities consistent with ‘athlete’s heart’^4,7; and (iv) no evidence of inherited/genetic disease. Athletes with moderate or severe RV functional or structural abnormalities at rest were excluded, as this is not typical for ‘exercise-induced’ or ‘gene-elusive’ ARVC. Healthy EAs and NAs were recruited via local advertisements and were eligible if they had no history of cardiovascular disease and if there was no evidence of structural or functional cardiovascular pathology on electrocardiogram and echocardiogram. In addition, EAs had to perform regular endurance exercise (≥6 h/weeks), while NAs could only participate in recreational sports activity (≤3 h/week of mild-to-moderate non-competitive endurance exercise). The study conformed to the Declaration of Helsinki and was approved by the local Ethics Committee. All participants provided written informed consent.

Study design

First, stepwise cardiopulmonary exercise testing (50 W + 25 W/min protocol) was performed on an upright cycle ergometer (ER900 and Oxycon Alpha, Jaeger, Germany) until exhaustion. Through breath-by-breath analysis minute ventilation (VE), oxygen consumption (VO₂), and carbon dioxide production (VCO₂) were assessed. Additional measures included peak heart rate, peak power, and the ventilatory equivalent for carbon dioxide (VE/VCO₂) slope. In those subjects with an implantable cardioverter-defibrillator (ICD), the device was reprogrammed so that all pacing and shocks were suspended during the exercise protocol. None of the subjects was paced.

Secondly, after at least 3 h of rest to ensure full recuperation, bicycle exercise echocardiography with lateral tilt was performed on a programmable, electronically-braked semi-supine ergometer (Easystress, Ecogito Medical sprl, Liege, Belgium) both at rest and during low and moderate intensity exercise defined as 25% and 60% of peak power. Each stage lasted ∼5 min depending on patient echogenicity and respiratory interference. Beta-blockers, non-dihydropyridine calcium-channel blockers, and antiarrhythmic drugs were withheld for at least 24 h prior to the study protocol.

Stress echocardiography

Images were acquired using a Vivid E9 ultrasound system and analysed offline using EchoPAC version 113 (both GE Vingmed Ultrasound AS, Horten, Norway). First, the left ventricular (LV) outflow tract (LVOT) diameter was determined from a parasternal long-axis (PLAX) view at rest. Then, at each stage, different image sets were acquired and at least 10 cardiac cycles were stored for each set during exercise. LV volumes and ejection fraction, RV areas and fractional area change (RVFAC), and tricuspid annular plane systolic excursion (TAPSE) were analysed from a single-plane (RV focused) apical four-chamber view. Cardiac output was calculated as heart rate multiplied with stroke volume, determined by the Doppler velocity time integral method in the LVOT (apical five-chamber view). Systolic pulmonary artery pressure (PASP) was estimated from the maximal trans-tricuspid regurgitant velocity on CW Doppler without the addition of right atrial pressure. Determination of PASP was enhanced with agitated colloid contrast.⁸ Mean pulmonary artery pressure was calculated using the Chemla formula.⁹ Two-dimensional speckle tracking LV and RV free wall longitudinal strain (SL) and strain rate (SRL) were acquired and analysed from single-plane (RV focused) apical four-chamber grey-scale images (60–90 frames/s) in accordance with contemporary guidelines.^10,11 Myocardial dispersion index (MDI) was calculated as the standard deviation of the contraction duration (time-to-peak-strain) in a six-segment LV and RV model. Timing parameters were corrected for differences in R–R interval using Bazett’s formula.

Statistics

Sample size calculation was based on our observation that RV function is similar at rest in healthy athletes and athletes with VA and the assumption, consistent with the results of Vitarelli et al.,^5,12 that the increase in RV free wall strain (RV_FWSL) of EI-ARVCs would only be 25% of that of healthy EAs (ΔSL = 0.9 vs. ΔSL = 3.6). Accordingly, a sample size of n = 12 was determined with an 80% power to detect a difference between athletes with and without VA (α = 5%, 1−β = 80%, n = 12). Data were analysed using SPSS Statistics version 24 (IBM Corporation, Armonk, NY, USA) and Analyse-it version 5.10 (Analyse-it Software Ltd, Leeds, UK). Normality was ensured using the Shapiro–Wilk test and variables are presented as means (±standard deviation) or as medians (with 25% and 75% percentiles) accordingly. Categorical variables were compared using a χ2 test (or Fisher’s exact test) and continuous data by either a Kruskal–Wallis H test or a one-way analysis of variance with the Bonferroni post hoc correction. The cardiac response to exercise was assessed using mixed linear models that included subject group, exercise stage, and their interaction as fixed effects with the Bonferroni post hoc correction for multiple comparisons. To account for the repeated nature of the data, an unstructured variance–covariance matrix was included. Pressure-flow relationships (i.e. P/Q slopes) were calculated through linear regression of the individual mean pulmonary artery pressure-cardiac output points obtained during exercise. Receiver-operating characteristics (ROCs) were constructed and the area under the curve (AUC) was calculated to identify the best echocardiographic parameter for distinguishing EI-ARVCs. Diagnostic performance of different parameters was compared using the Delong–Delong–Clarke-Pearson method.¹³ Intra- and inter-observer variability of echocardiographic measures were assessed at rest and during exercise in a sample of 15 subjects (five out of each group) and evaluated through the coefficient of variation and the intra-class correlation coefficient (two-way mixed and absolute agreement quoted), analysed on different cardiac cycles within the image set. A P-value of <0.05 was considered statistically significant.

Results

Baseline characteristics and cardiopulmonary exercise test

Forty-six male subjects (12 NAs, 17 healthy EA’s, and 17 EI-ARVCs) were included in the study. The baseline characteristics and results of the cardiopulmonary exercise test are summarized in Table 1. The three groups were of similar age but athletes with RV arrhythmias had a larger body surface area and had competed in sports longer compared to both other groups. EI-ARVCs had discontinued intensive sports on average 6.3 (±5.5) years prior to the evaluation. Eight EI-ARVCs had an ICD and 15 received antiarrhythmic drug treatment, either alone or in combination. As evident from Supplementary data online, Table S1, 11 EI-ARVCs met the Task Force Criteria for a definitive diagnosis of ARVC, whilst 6 met criteria for possible ARVC. None of the subjects had clinical evidence of inherited disease nor a mutation on desmosomal gene sequencing. Subjects who met the major ARVC Task Force arrhythmia criteria were more likely to have an ICD compared to those who did not (P < 0.05). Isolated minor ARVC criteria were also present in four healthy EAs; two EAs had minor repolarization abnormalities and two EAs had a history of frequent (>500/24 h) ventricular premature beats. Healthy EAs had a significantly higher exercise capacity compared to both other groups as evident from their higher peak oxygen consumption and peak power (both P < 0.001 between groups).

Biventricular response to exercise

The results of the exercise echocardiography protocol are summarized in Table 2 and Figures 1 and 2. At rest and consistent with exercise-induced cardiac remodelling, EAs had greater LV volumes compared to NAs. RV areas of both EAs and EI-ARVCs were higher compared to controls (P ≤ 0.001 between groups). There were no major differences in either LV or RV functional parameters at rest. RV afterload measured by systolic pulmonary artery pressure was similar across the three groups.

During exercise, echocardiographic measures were obtained in the majority of subjects. At moderate intensity, LV ejection fraction and RVFAC could be determined in 44 subjects (96%) and LV and RV speckle tracking deformation parameters could be reliably obtained in 40 (87%; 16 EAs, 9 NAs, 15 EI-ARVCs) and 37 subjects (80%, 15 EAs, 8 NAs, 14 EI-ARVCs), respectively. Similar changes during exercise were observed using both 2D echocardiographic and deformational measurements. LV function augmented during exercise in all three groups (Figure 1), but the increase was blunted in EI-ARVCs (P < 0.05 for interaction for LV ejection fraction, LVSL, and LVSRL). However, at peak exercise only LVSL was lower in EI-ARVCs, whereas LV ejection fraction and LVSRL were similar. In contrast, the exercise-induced augmentation in RV function was clearly impaired (Figure 2). RV functional parameters were consistently lower in EI-ARVCs and neither RVFAC nor RV_FWSL increased during exercise (Figure 2B, P = 0.001 and Figure 2C, P = 0.038 for interaction, respectively). RV free wall longitudinal strain rate (RV_FWSRL), on the other hand, increased in EI-ARVCs to a lesser extent compared to both other groups (Figure 2D, P < 0.001 for interaction). The change in afterload during exercise did not differ between groups (Supplementary data online, Figure S1). Echocardiographic measures of LV and RV function had good to excellent reproducibility both at rest and during exercise with intra-class correlation coefficients ranging from 0.91 to 0.99 for intra-observer variability and 0.71 to 0.99 for inter-observer variability (Supplementary data online, Table S2). An example of a characteristic RV SL and RV SRL curve of an EA and an EI-ARVC is provided in Figure 3.

Timing and mechanical dispersion at rest and during exercise

When corrected for heart rate, there were no differences in the time-to-peak shortening of both ventricles at rest (Supplementary data online, Table S3). However, EI-ARVCs had a higher LV MDI and tended to have a higher RV MDI compared to NAs and EAs. During exercise, RV free wall time-to-peak-strain was significantly longer in EI-ARVCs compared to both NAs and EAs (both P < 0.01). As opposed to resting measures, there were no differences in either LV or RV dispersion during exercise.

Discrimination of subjects with arrhythmias

As evident from Table 3 and Figure 4, RV measures at rest were unable to differentiate EI-ARVCs from apparently healthy subjects (EAs and NAs combined) with the exception of RV mechanical dispersion index which had moderate discriminatory power [AUC= 0.72 (0.54–0.90)]. At moderate exercise intensity, the diagnostic performance of RV functional parameters to identify EI-ARVCs increased significantly. A RV_FWSRL >−2.35 (i.e. more positive) had a 93% sensitivity and 82% specificity for discriminating EI-ARVCs. Exercise-to-rest differences, with the exception of ΔTAPSE, had reasonable to good accuracy (Supplementary data online, Figure S2) as well. Separate ROC analysis for the comparison of EI-ARVCs with EAs and NAs revealed similar diagnostic performance and cut-offs (Supplementary data online, Table S4 and Figure S3).

RV functional impairment is proportional to disease severity

When we separately assessed athletes with a more severe phenotype as evidenced by the clinical decision to ICD, several interesting observations could be made (Supplementary data online, Figure S4). Firstly, as might be expected, subjects with an ICD clearly had more pronounced RV dysfunction. During exercise, RVFAC only increased in EI-ARVCs without an ICD (Supplementary data online, Figure S4A), while RV_FWSL did not increase in either subgroup (Supplementary data online, Figure S4B). RV_FWSRL increased in both subgroups of EI-ARVCs, but to a lesser degree compared to healthy subjects (Supplementary data online, Figure S4C). Importantly, RV_FWSRL during moderate exercise intensity [AUC 0.91 (0.78–1.00)] was the only variable that accurately differentiated EI-ARVC without an ICD from healthy subjects (Supplementary data online, Table S5 and Figure S5; P < 0.05 for comparison of AUC versus all other variables). In contrast, neither RVFAC nor RV_FWSL at rest or during exercise had sufficient discriminatory power.

Discussion

In this study, we examined the diagnostic performance of bicycle exercise echocardiography to differentiate male subjects with EI-ARVC from apparently healthy male EAs and NAs. We found that ARVC subjects had impaired RV functional reserve, which was concealed at rest but unmasked during exercise. Furthermore, exercise parameters outperformed resting measurements in identifying subjects with RV arrhythmias. RV_FWSRL at moderate exercise intensity was the most accurate discriminator of disease and the only echocardiographic parameter able to identify those subjects with a milder phenotype. Thus, evaluation of strain rate during exercise may be a promising non-invasive means of identifying subtle RV dysfunction that is not apparent at rest.

Deformation imaging in arrhythmogenic RV pathology

Cardiac deformation or strain imaging quantifies cardiac motion and enables evaluation of regional and global myocardial function on a spatial and temporal basis.¹⁴ As longitudinal shortening accounts for the majority of RV stroke volume generation, longitudinal deformation is a reliable estimator of RV function in healthy subjects.^15,16 In patients with ARVC, deformation imaging unmasks subclinical disease and even identifies subjects at risk of disease progression or ventricular arrhythmias.^17–19 Moreover, amongst asymptomatic ARVC mutation carriers nearly 50% of individuals have an abnormal deformation pattern, which is associated with an increased risk of disease progression at follow-up.^20,21 Intriguingly, these abnormalities were most pronounced in the basal RV free wall, a region where lower strain and strain rate values have also been observed in healthy athletes.²²

Previously, we have demonstrated that the ambiguity of lower RV strain values in athletes can be resolved by exercise imaging and evaluation of RV functional reserve.²³ Using a similar approach, we observed no difference in RV function between healthy athletes and NAs when measurements were performed when the greatest load is imposed on the right ventricle, i.e. during exercise. However, in EI-ARVCs exercise, was able to unmask RV dysfunction, which was concealed at rest. These results are in line with our previous report on RV contractile dysfunction occurring during exercise in EI-ARVCs as assessed by exCMR.⁵ However, the novelty of the current study is that we were able to show the same haemodynamic differences using simpler methodology. Given the widespread availability of exercise echocardiography and the implementation of deformation measurements in contemporary guidelines, these haemodynamic alterations can be detected in daily clinical practice.

Previous studies assessing RV functional reserve have predominantly looked at the relative change in peak strain from rest-to-exercise.^24,25 Vitarelli et al.¹² identified ΔSL as the most accurate parameter to identify subjects with ARVC. However, in these studies, cardiac function was already impaired in the rested state, as opposed to our cohort in which RV function was ostensibly normal at rest. An important finding in our current study is that RV_FWSRL during exercise was the most accurate discriminator of RV disease and outperformed all other measurements of RV function. As such, our results indicate that RV strain rate may be well suited to evaluate RV functional reserve and identify subclinical disease in athletes, and potentially in other populations. Conversely, mechanical dispersion appeared to be only discriminative at rest. We speculate this might relate the effects of exercise on electrical dispersion and the pathophysiological differences between gene-elusive and gene-positive disease.

Clinical implications

The complex interaction between genetic and environmental factors governing the development of RV arrhythmias in athletes, fuels a continued debate about the exercise-induced arrhythmia phenotype.²⁶ Regardless, accurate identification of intrinsic RV disease is relevant in athletes as complex VA frequently arise from the right side and vigorous exercise has a detrimental effect on the ARVC phenotype.^1,27,28 However, the differentiation between athlete’s heart and pathological RV remodelling is often difficult and requires a multimodality approach.²⁹ Our current data demonstrate that exercise imaging facilitates disease discrimination and is able to depict those EI-ARVCs with a more severe phenotype, which is important for risk stratification and disease management. As strain rate more closely corresponds to intrinsic myocardial contractility, we may hypothesize that measures of RV deformation imaging during exercise might have superior accuracy, particularly in early disease stages when only minor abnormalities are expected, but this remains to be established in future studies.³⁰

Study limitations

The modest sample size may have increased the probability of type II statistical errors while multiple comparisons increase the likelihood of type I errors. Nevertheless, this is the largest study to date to investigate EI-ARVCs with deformation imaging during exercise. Secondly, the results could have been confounded by disease heterogeneity, RV remodelling, prior athletic training and time since sports withdrawal of EI-ARVCs. However, as RV function did not differ between EAs and NAs and diagnostic performance was similar when both groups were analysed separately, a substantial influence in unlikely. Thirdly, the negative inotropic/chronotropic effect of antiarrhythmic drugs or the RV lead could have influenced the exercise response in EI-ARVCs. The pharmacological effects were minimized by ensuring a wash-out >24 h. Moreover, the residual effect is likely minimal given the greater impairment of RV function, whereas a drug effect would be biventricular. Furthermore, the absence of more than mild tricuspid regurgitation and similar apical deformation between EI-ARVCs with and without ICD (despite apical lead position) argue against a direct or indirect effect of the RV lead on the exercise response. Fourthly, the presence of pathology or an RV lead compromises a blinded interpretation of the images. We tried to minimize bias by analysing image data before accessing clinical data and by evaluating subjects in a random order. Finally, speckle-tracking deformation imaging has some inherent technical limitations. Firstly, adequate image quality was not available in every subject. However high-quality images were still acquired in the vast majority and similar patterns were observed both by different echocardiographic measures and previously using gold standard exCMR.⁵ Secondly, the limited temporal resolution of 2D strain may have led to under-sampling with an underestimation of true peak values. However, this limitation pertains to all groups and even though heart rate was higher in healthy subjects, meaningful differences were still demonstrable. Tissue Doppler imaging has superior temporal resolution but is challenging to perform during exercise due to its angle-dependency and increased noise and we have previously demonstrated good agreement between both techniques at rest and during exercise.²³

Conclusion

Exercise deformation imaging holds promise as a non-invasive diagnostic tool to identify intrinsic RV dysfunction which is not detectable at rest. Strain rate appears to be the most accurate parameter and should be incorporated in future, prospective studies to identify subclinical disease and assess outcome at an early stage.

Funding

This study was funded by a grant (Project G.0465.10N) from the Fund for Scientific Research Flanders (FWO), Brussels, Belgium.

Conflict of interest: G.C. was supported by a postdoctoral research grant from the Frans Van De Werf Fund for Clinical Cardiovascular Research, by the UZ Leuven Future Fund and by a Mathilde Horlait-Dapsens Scholarship; R.W. was supported as a postdoctoral clinical researcher by the Fund for Scientific Research Flanders (FWO); A.L.G. has received grants from the Fund for Scientific Research Flanders (FWO) and from the National Health and Medical Research Council (NHMRC) of Australia. All other authors declared no conflict of interest.

References

Author notes