Regular physical activity confers benefits that are widely recognized such as improved cardiovascular risk profiles and prevention of coronary heart disease and diabetes mellitus.^1,2 Regular exercise also directly and positively affects cardiac physiology (eg, increased myocardial oxygen supply and enhanced myocardial contractility), both in the general population³ and in patients with cardiovascular disease.⁴

Editorial see p 5

Clinical Perspective on p 22

Long-term exercise induces hemodynamic changes and alters the loading conditions of the heart, with specific effects depending on the type of sport and intensity, that are most evident among athletes.⁵ Cardiac adaptations in highly trained subjects include increased left ventricular (LV) and right ventricular (RV) diameters, enlarged left atrial (LA) dimensions, and increased cardiac mass and LV wall thickness.^5,6 These changes, together with a preserved ejection fraction, have classically characterized the physiology of the “athlete's heart.”⁵

Despite the evident benefits of an active lifestyle,^1–4 numerous observational studies have raised concerns that high-level exercise training may be associated with increased cardiac arrhythmia risk and even primary cardiac arrest.⁷ Initial observations from our group and others,^8–13 later confirmed by a large epidemiological study,¹⁴ have shown that long-term endurance training may promote atrial fibrillation. Complex ventricular tachyarrhythmias can also occur in highly trained individuals¹⁵; according to recent studies^16,17 they often originate from a mildly dysfunctional RV, even after excluding RV pathologies like arrhythmogenic RV cardiomyopathy.

These findings raise the possibility that long-term endurance exercise may promote the development of certain cardiac arrhythmias. Some authors have speculated that the cardiac remodeling after sustained physical activity may create an arrhythmogenic substrate.^8,17 Although information on athletes is insufficient, arrhythmia susceptibility has been extensively related to myocardial fibrosis in other clinical contexts.¹⁸ Tissue fibrosis appears as a reparative process for damaged myocardial parenchyma and results in accumulation of fibrillary collagen deposits, which may favor reentry and consequently arrhythmogenicity.¹⁸

The present study was designed to develop a rat model of long-term, intensive endurance exercise to test whether regular intense physical training can induce cardiac structural changes, particularly fibrosis, thereby generating a substrate for cardiac arrhythmias.

Methods

Experimental Design

This study conformed to European Community (Directive 86/609/EEC), Spanish, and Canadian guidelines for the use of experimental animals and was approved by the institutional animal research ethics committees. Pathogen-free, 4-week-old male Wistar rats weighing 100 to 125 g (Charles River Laboratories, France) were housed in a controlled environment (12/12-hour light/dark cycle) and fed rodent chow and tap water ad libitum.

Animals were randomly assigned to sedentary (Sed) or intensive exercise (Ex) groups. To assess time-course changes, animals in both groups were studied at 4, 8, and 16 weeks. The exercise program was based on a previously validated protocol.¹⁹ Ex rats underwent daily running training sessions on a treadmill. The treadmill had different lanes to serve as corridors for each animal and had a grid in the back that administered a small electric shock on contact to ensure that animals ran effectively. The electric shock was of constant intensity (0.3 to 2 mA), sufficient to encourage the animal to run without being harmful. The protocol included a 2-week progressive training program, starting with a 10-minute running session at 10 cm/s and increasing gradually to steady-state 60-minute running at 60 cm/s. Thereafter, animals were trained at this level 5 days a week for 4, 8, or 16 weeks. Investigators observed the treadmill sessions daily to ensure effective running. Only rats that mastered the running training and ran spontaneously with a maximum cumulative shock time of 15 seconds per 1-hour training session were included in the study. Sedentary rats were housed and fed in the same conditions.

An additional series of rats underwent 16 weeks of training followed by discontinuation of exercise (DEx) to assess the reversibility of exercise-induced changes. DEx rats were assessed after 2, 4, or 8 weeks of exercise cessation. Sedentary rats housed and fed in the same conditions over the same period served as DEx controls.

Animals were euthanized 3 days after the end of the training program to avoid immediate responses or after 2, 4, or 8 weeks from the last running session in the DEx groups. Hearts were quickly removed; weighed; dissected into LV, RV, LA, and right atrium (RA); and frozen in liquid nitrogen at −80°C or fixed for histological studies. For details on the echocardiography, electrophysiological study, tissue imaging, and biochemical studies, see the online-only Data Supplement.

Statistical Analysis

Data are expressed as mean±SEM. Statistical analysis was generally carried out with 2-way ANOVA with general linear model procedures using a univariate approach. Compound symmetry covariance structure was used for repeated measures analysis. The sphericity test, the Mauchly criterion, was used to test for departures from the assumption of compound symmetry and was consistent with the sphericity assumption in all instances. For heart weight and hydroxyproline experiments, exercise and time point were the main effects. Morphometric, real-time polymerase chain reaction and echocardiographic results were analyzed with 2-way, repeated measures ANOVA, with exercise as 1 main effect and either cardiac chamber (morphometric and real-time polymerase chain reaction experiments) or time point (echocardiography) as the repeated measures main effect. Picrosirius red and hydroxyproline deconditioning studies were analyzed with 1-way ANOVA. In the case of a significant interaction by 2-way ANOVA or a significant difference on 1-way ANOVA, Bonferroni-corrected t tests were used to assess Sed versus Ex group differences. In the absence of interaction, P values are shown for significant differences in the main effect. Ex versus Sed immunoblots and electrophysiological testing results were compared by use of t tests for nonpaired samples. The Fisher exact test was used to compare frequency variables. SPSS version 17.0 was used for statistical analysis. Detailed specifications of statistical analysis in each figure are provided in the online-only Data Supplement. Two-tailed values of P<0.05 were considered significant.

Results

Cardiac Remodeling After Long-Term Intensive Exercise Training

Cardiac mass was significantly increased by exercise (Figure 1A). Values for individual cardiac chambers, available at 16 weeks, were significantly increased in Ex rats (Table 1). No significant changes were observed in the LV/RV mass ratio. Direct measurement of wall thickness (Figure 1B) confirmed significant increments in postmortem interventricular septum (IVS) and LV free wall (FW) thickness after 8 weeks of intensive exercise, which were maintained at 16 weeks (Figure 1C). No significant differences were observed in RV FW thickness.

• Download figure
• Download PowerPoint

To evaluate further cardiac morphological and functional adaptation to long-term intensive exercise, serial echocardiograms were performed in a subset of Ex and Sed rats. Because morphological LV hypertrophy was not observed before 8 weeks, echocardiograms were performed only at baseline and after 8 and 16 weeks of training. Ex rats developed concentric LV hypertrophy at 8 weeks, manifested by increased LV wall thicknesses and ratio of IVS to LV diameter at end diastole (for echocardiographic results, see Table 2), evolving to eccentric hypertrophy/ventricular dilatation at 16 weeks. Ex rats also showed evidence of LV diastolic dysfunction at 8 to 16 weeks (decreased S wave in pulmonary vein flow, increased LV isovolumic relaxation time corrected for R-R), along with LA enlargement. A slight but statistically significant decrease in LV systolic function was also observed in Ex rats at 16 weeks. Evidence of RV diastolic dysfunction also occurred at 8 to 16 weeks (decreased E-wave velocity, prolonged E-wave deceleration time).

Intensive Exercise Training Promotes Chamber-Specific Ultrastructural Remodeling

We next evaluated tissue fibrosis. Figure 2A shows representative photomicrographs of Picrosirius red–stained RV sections from Sed and Ex rats. Diffuse interstitial collagen deposition associated with disturbances in myocardial architecture was observed in RV FWs of Ex rats after 16 weeks of training. No differences were observed in the LV (Figure IA in the online-only Data Supplement). Morphometric quantification confirmed a gradual increase in RV FW collagen with training (Figure 2B). No differences in collagen density were observed in the IVS or LV FW. Increased collagen in the RV of Ex rats at 16 weeks was also noted on Masson trichrome–stained images (Figure 3A).

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

To independently quantify fibrous tissue content, the amount of hydroxyproline, a modified amino acid specifically found in collagen, was determined. After 16 weeks of intensive exercise, animals showed significant increases in RV hydroxyproline content (Figure 3B), with no significant differences observed in the LV (Figure I in the online-only Data Supplement).

Messenger RNA (mRNA) expression of transforming growth factor-β1 (TGF-β1), fibronectin-1, matrix metalloproteinase-2 (MMP-2), tissue inhibitor of metalloproteinase-1 (TIMP1), procollagen-I, and procollagen-III was measured in all cardiac chambers of rats in the Sed and Ex groups. After 4 and 8 weeks of exercise, no significant changes were observed (Table I in the online-only Data Supplement). The results at 16 weeks are shown in Figure 4. TGF-β1, fibronectin-1, and MMP-2 mRNA expression was significantly increased in the RA, LA, and RV of Ex rats compared with Sed rats (Figure 4A, 4B, and 4C, respectively). The only significant difference for TIMP1 mRNA expression was found in the RA (Figure 4D). Finally, mRNA expression of procollagen-I was significantly increased in the RA and RV of Ex rats (Figure 4E), whereas procollagen-III was significantly increased in both the RA and LA of Ex rats (Figure 4F).

• Download figure
• Download PowerPoint

Alterations in protein expression corresponding to mRNA changes were assessed by Western blot analysis for TGF-β1, fibronectin-1, MMP-2, TIMP1, collagen-I, and collagen-III. TGFβ-1 protein levels were significantly increased in both atria and RVs of Ex rats (Figure 5A). Fibronectin showed no significant changes (Figure 5B). MMP-2 protein expression was significantly increased in the RA and LA of Ex rats (Figure 5C), whereas TIMP1 was unchanged (Figure 5D). In parallel with results for procollagen-I mRNA expression, collagen-I protein levels were significantly greater in both the RA and RV of Ex rats (Figure 5E); however, collagen-III was unchanged (Figure 5F).

• Download figure
• Download PowerPoint

These results confirm the development of significant extracellular matrix (ECM) changes after 16 weeks of intensive endurance exercise, with fibrosis clear in the RV but not LV.

Long-Term Intensive Exercise Increases Ventricular Arrhythmia Vulnerability

We then evaluated whether ventricular remodeling by 16 weeks of exercise led to changes in electrophysiological parameters and/or arrhythmia susceptibility. In vivo electrophysiological study was performed with a customized catheter inserted into the RV apex. Ex rats showed evidence of a slight delay in ventricular conduction, manifested by longer QRS duration (Table 3). No changes were noted in repolarization on the basis of ventricular effective refractory periods. During programmed stimulation with up to 3 extrastimuli, sustained ventricular tachyarrhythmias (>10 seconds) were induced in 5 of 12 Ex rats (42%) compared with 1 of 16 Sed rats (6%; P=0.05; Figure 6).

• Download figure
• Download PowerPoint

Remodeling Reverses With Detraining

To determine whether potentially adverse ventricular remodeling is reversible after exercise cessation, we compared rats allowed to recover after discontinuation of exercise (DEx groups) with age-matched Sed controls. Although showing some regression, cardiac mass remained significantly greater in all DEx groups compared with their Sed counterparts (Figure 7A).

• Download figure
• Download PowerPoint

Because all Sed groups (Sed at 16 weeks and Sed-DEx at 2, 4, and 8 weeks) presented equivalent data (not shown) for morphometric measurements, hydroxyproline levels, histology, and mRNA analyses, results for the 16-week Sed group were used for comparisons. Wall thickness increases induced by 16 weeks of intensive exercise resolved progressively in both the IVS and LV FW (Figure 7B). In contrast, no differences in RV wall thicknesses were found among the Ex, the Sed and all the DEx groups.

We then evaluated whether the fibrotic changes induced by 16 weeks of intensive exercise were also reversed by exercise discontinuation. Histopathology studies with Masson trichrome and Picrosirius red confirmed a gradual decrease in collagen content during deconditioning (Figure 7C and 7D). Similarly, collagen quantification based on image analysis of Picrosirius red–stained tissues confirmed regression of fibrosis in the RV after deconditioning (Figure 7E). Correspondingly, hydroxyproline content in RV decreased progressively during deconditioning and became nonsignificantly different from Sed rats and significantly lower than in Ex rats at 16 weeks after exercise cessation (Figure 7F). In accordance with these results, mRNA studies showed significant reversal in exercise-enhanced profibrotic markers within 2 weeks of deconditioning (Figure 8). Together, these results suggest substantial reversibility of vigorous endurance training–induced cardiac remodeling after the cessation of exercise training.

• Download figure
• Download PowerPoint

Discussion

The present study describes cardiac remodeling in a rat model of long-term, intensive exercise training, demonstrating changes in cardiac function, fibrous tissue content, fibrotic markers, and arrhythmia susceptibility following long-term endurance training, with substantial reversibility after exercise cessation. If results are similar in humans, then our findings suggest that long-term, intensive exercise can promote chamber-specific remodeling and provide a substrate for arrhythmogenesis.

Cardiac Remodeling After Intense Endurance Exercise-Training

As previously described in other models,^19–21 we found significant LV hypertrophy at 8 weeks of training. At 16 weeks, LV dilatation was also observed, leading to eccentric hypertrophy. LV hypertrophy may have accounted for the impaired diastolic function observed after 8 and 16 weeks of exercise, which in turn was associated with LA dilatation. All these findings are consistent with the features of the athlete's heart described in humans⁵ and support the potential relevance of our training program. Of note, we also observed RV diastolic dysfunction, with a trend to RV dilatation and systolic dysfunction after 16 weeks of exercise, findings that have recently been described in high-level endurance-sport practitioners.¹⁷

Chamber-Specific Myocardial Fibrosis After Intensive Exercise Training

Perhaps our most noteworthy finding is the demonstration of myocardial fibrosis after long-term, intensive exercise training. RV fibrosis was documented by collagen quantification in histological sections and analysis of hydroxyproline content. These observations were functionally paralleled by the development of RV diastolic dysfunction, with impaired relaxation potentially related to fibrotic infiltration. Moreover, we noted an increase in mRNA and protein expression of a series of fibrotic markers in the RV and in both atria. TGF-β1 expression was increased in the RA, LA, and RV after 16 weeks of intensive exercise. TGF-β1 is a potent stimulator of collagen-producing cardiac myofibroblasts²² and leads to fibrosis development. Experimental studies have reported that both genetic ablation of TGF-β1 in mice²³ and treatment with anti–TGF-β1 antibodies²⁴ inhibit fibrosis development, indicating that TGF-β1 plays a major role in collagen turnover. We also noted enhanced expression of other major components of the ECM control system, including fibronectin-1, collagens, MMP-2, and TIMP1. Collagen-I determines the stiffness of cardiac muscle, whereas collagen-III is more distensible. Thus, the ratio of collagen-I to collagen-III can be a marker of the ECM determinants of cardiac stiffness.²⁵ We observed a significant increase in collagen-I protein expression in right-sided cardiac chambers after 16 weeks of intensive exercise, whereas collagen-III expression remained unchanged, indicating that long-term, intensive exercise could increase cardiac stiffness in these chambers via altered ECM composition, a notion that was supported by echocardiographic evidence of diastolic dysfunction. These findings were accompanied by overexpression of mRNA and protein levels of MMP-2. MMP-2 is a proteolytic enzyme; activation of MMP-2 induces disruption of ECM proteins and promotes fibrogenesis. Together, these results indicate the presence of a milieu favoring the development of myocardial interstitial fibrosis, characterized by alterations in fibroblast growth factors and ECM imbalance.

Although extensive data have been published on echocardiographic remodeling with exercise,^6,26 less information is available on cardiac histological and biochemical remodeling in endurance athletes. A recent study demonstrated increased turnover of fibrosis markers in plasma from veteran athletes,²⁷ although cardiac origin was not assessed. Data in abstract form also suggest the development of ventricular fibrosis based on magnetic resonance imaging in endurance athletes.²⁸ Our finding of cardiac fibrosis represents the first direct evidence of potentially adverse cardiac remodeling after long-term, intensive exercise.

The mechanisms by which long-term, intensive exercise may promote cardiac fibrosis are unknown. It is possible that long-term cardiac overload plays a role by promoting physiological remodeling in early phases that may eventually become maladaptive in the long term. Experimental studies support the idea that physiological cardiac remodeling and pathological cardiac remodeling involve different signaling pathways,²⁹ but recent data have demonstrated that excessive stimulation of physiological systems can result in maladaptive responses.^30,31 Whether this or other mechanisms are involved in the profibrotic cardiac remodeling observed in our model is a matter to be addressed in future studies.

Of note, tissue fibrosis in this model of long-term, intensive exercise was chamber specific (ie, the LV did not appear to be affected). There are 2 potential explanations for this finding. First, assuming that exercise increases loading conditions on all cardiac chambers,⁵ it is reasonable to suppose that greater profibrotic remodeling develops in chambers suffering greater degrees of overload. In this regard, clinical studies have described higher loading conditions on the RV than on the LV during endurance sports,³² leading to transient RV dysfunction immediately after exercise.³³ Second, it has been suggested that intrinsically thinner walls could make the atria, as well as the RV, more susceptible to remodeling.^17,18 The absence of significant fibrosis in the LV agrees with previous models of long-term exercise^20,21 and supports the normal functionality of the LV of the athlete's heart.⁵ Supporting this idea, an animal model of long-term volume overload with similarities to loading condition changes in endurance training showed regional overexpression of growth factors and increased collagen deposition in the RV but not the LV.³⁴

Intense Exercise Training and Arrhythmogenic Remodeling

Despite its benefits for overall health,^1–4 numerous clinical studies in recent years have suggested that long-term high-level exercise might be associated with an increased risk of cardiac arrhythmias, mainly atrial fibrillation and ventricular arrhythmias originating from the RV.^8–16 One important aspect of this study is that the remodeling observed after long-term, intensive exercise training could represent a potential substrate for arrhythmias. Cardiac fibrosis and associated myocardial disarray provide electric heterogeneity and promote reentrant circuits and arrhythmogenesis.¹⁸ It has been reported that atrial overexpression of TGF-β1 in transgenic mice is sufficient to generate a fibrotic substrate that supports atrial fibrillation.³⁵ Similarly, increases in procollagen and MMPs have been related to increased risks of atrial and ventricular arrhythmias.^36–38 We assessed this hypothesis by evaluating the inducibility of ventricular arrhythmias during in vivo programmed stimulation studies and noted inducible sustained ventricular tachyarrhythmias in 42% of rats subjected to intensive exercise for 16 weeks, compared with only 6% of sedentary rats. In the presence of increased QRS duration, indicating ventricular conduction slowing, and the absence of changes in electrophysiological parameters reflecting repolarization-like ventricular refractory period, these results suggest that the cardiac fibrosis observed in our model could play a role in producing the increased arrhythmia susceptibility we observed in exercise rats.

Reversibility

Clinical studies have reported regression of the morphological changes characteristic of the athlete's heart after long-term detraining.³⁹ The reversibility of arrhythmogenic remodeling is of potentially great clinical importance, because it would imply that deleterious rhythm consequences of long-term endurance training can be expected to disappear after cessation of intense physical training. We accordingly assessed whether a period of rest could allow reversion of the profibrotic changes induced by endurance training in our model. The results of our exercise discontinuation study demonstrate that, after 8 weeks of detraining, virtually all the abnormal cardiac remodeling parameters resulting from intense exercise training regressed to control levels.

More studies are needed to ascertain the mechanisms that participate in both the promotion and the reversal of the fibrotic remodeling associated with long-term exercise and detraining, respectively. In addition, follow-up clinical studies are indicated to establish whether similar remodeling changes can be demonstrated in humans and, if so, whether they are reversible.

Potential Limitations

We cannot exclude the possibility that our exercise training protocol involving conditioning shocks might have induced emotional stress in Ex rats. Maximum efforts were taken to minimize stress responses. Rats that did not adapt to treadmill exercise or received excessive shocks (>15 s/h) were excluded from the study.

It is difficult to estimate precisely how our exercise program translates into human activity. As a rough approximation, considering that the typical rat life expectancy is 2 to 2.5 years, our 18-week exercise protocol (2 weeks of progressive training plus 16 weeks of intensive exercise) would be equivalent to ≈10 years of daily exercise training in humans. According to previous studies in rodents,^40,41 the intensity of our program would correspond to ≈85% of maximum oxygen uptake, equivalent to physical activity at ≈90% of predicted maximum heart rate in humans.⁴² Therefore, our results cannot be directly extrapolated to milder or more moderate forms of exercise. In addition, we studied only remodeling reversal with complete exercise cessation, which is unlikely in high-level athletes. Whether similar recovery is achieved by simply reducing the intensity of a training program is uncertain.

Only young male rats were tested in this study. Age- and gender-related factors could significantly influence exercise-related cardiac remodeling and were not analyzed in our study. Further work in other animal models, studies of age and gender effects on exercise-induced remodeling, and follow-up analyses in human populations would be of great interest.

The functional consequences of cardiac remodeling have been specifically assessed by passive pressure-strain curves in papillary muscles or LV pressure-volume curves.⁴³ Because of limited availability of hearts and the need to obtain tissue samples for histological and biochemical studies, we decided to study both functional and morphological consequences of long-term intensive exercise by performing serial echocardiograms, thus obtaining a maximum of information while being able to use each rat as its own control.

Conclusions

This study shows that long-term intense endurance training promotes heart chamber-specific remodeling and ventricular arrhythmia susceptibility in an animal model. Cessation of endurance training was able to arrest and even reverse this pathological process. These findings, if reproduced in humans, could have potentially important implications for arrhythmia risk and its management in individuals involved in high-level athletic training and practice.

Sources of Funding

This work was supported by grants from the Sociedad Española de Cardiología, Fondo de Investigaciones Sanitarias from Instituto de Salud Carlos III ( PI050210 and PS09/02362 ), Rio Hortega ( CM06/00189 and CM08/00201 ) from the Spanish Health Ministry, Universitats i Recerca ( 2005SGR00497 ), Canadian Institutes of Health Research ( MGP-6957 ), and an ENAFRA network award from Fondation Leducq ( 07-CVD-03 ).

Disclosures

None.

Footnotes