Experimental design.

This study conformed with the European Community (Directive 86/609/EEC) and Spanish guidelines for the use of experimental animals and was approved by the Institutional Animal Research Ethics Committee. We used an experimental animal model in which rats were conditioned to run on a treadmill for 5 days/wk for 16 wk, as previously described (3, 9). Fifty-five male Wistar rats (200–250 g, Charles River Laboratories) were randomly assigned to the following three groups: moderate exercise (MOD; 35 cm/s for 45 min), high-load (intense) exercise (INT; 60 cm/s for 1 h), and age-matched sedentary (SED) rats that served as controls. Estimates from previous studies in rats have suggested that these loads approximate 60% and 85% of maximum O₂ uptake for MOD and INT, respectively (17, 42). The final training load was reached after an initial 2-wk adaptation period in which the treadmill speed was progressively increased. Rats were supervised during all training sessions to ensure proper running; animals that did not adapt to the exercise routine were excluded from the study to avoid the deleterious effects of physical and psychological stress that could potentially bias our results. All animals were housed in a controlled environment (12:12-h light-dark cycle) and were provided with ad libitum access to food and water. At the end of the training protocol, in vivo RV and LV functional and structural remodeling were assessed in hemodynamic experiments and a two-dimensional echocardiogram. In trained rats, the echocardiography and hemodynamic experiments were carried out at least 12 h after the last training session.

Echocardiography.

Transthoracic echocardiographic experiments were performed at rest in the three groups. The procedure was performed under general anesthesia (2% isofluorane), and a heating pad and a phased-array probe 10S (4.5–11.5 MHz) attached to a commercially available system (Vivid q, GE Vingmed Ultrasound, Horten, Norway). The M-mode spectrum was traced in a parasternal short-axis plane at the level of the aortic valve; the RV outflow tract was also measured in this view. LV diameters at both end-diastole (LVEDD) and end-systole (LVESD) were measured at the level of the papillary muscles. The LV anterior wall (AW) and posterior wall (PW) thickness was measured at end diastole. LV ejection fraction (LVEF), fractional shortening (LVFS), and LV mass were estimated using the following previously validated formulas in rodents (26, 34):

RV end-diastolic area (RVEDA) and RV end-systolic area (RVESA) were measured in a four-chamber apical view focusing on the RV; RV fractional area change (RVFAC) was then calculated as follows (34a):

Five consecutive cardiac cycles with color-coded tissue Doppler imaging (TDI) images were recorded in a four-chamber apical view and saved for posterior offline analysis; special care was taken to maintain the Doppler velocity range (0.77–46.2 cm/s) as low as possible to avoid aliasing and an average frame rate of 250–300 s^–1. Measurements were calculated as the average of the five cardiac cycles. RV segmental deformation was evaluated by strain rate (SR) at basal and apical segments using a sample volume of 2 mm and a specific software package (EchoPAC, General Electric Healthcare, Milwaukee, WI). LV and RV diastolic function were estimated by the filling peak velocity (E) of the transmitral and transtricuspid flow, and mitral septal (e′M) and tricuspid lateral (e′T) annulus velocity during early filling derived from TDI (24). Additionally, isovolumetric relaxation time (IRT) of both ventricles was evaluated (Fig. 1). All cardiac dimensions were body weight indexed to account for differences in body weight. An LVEDD-to-RVEDA ratio was built to assess the balance between the structural remodeling of the LV and RV.

Reproducibility was assessed in 12 rats (4 for each study group). Intraobserver and interobserver intraclass correlations were 0.90 and 0.85, respectively, for RV systolic deformation at the basal segment, 0.88 and 0.82 for RV systolic deformation at the apical segment, 0.90 and 0.89 for e′T, and 0.88 and 0.85 for e′M.

Hemodynamic experiments.

In vivo RV and LV contractile remodeling were assessed in invasive hemodynamic experiments in a subgroup of rats (16/15/13 for the LV and 8/12/8 for the RV for the SED/MOD/INT groups, respectively). Briefly, anesthetized rats (inhaled 1.5–2% isoflurane) were intubated and ventilated (CWE, Ardmore, PA) with parameters recommended by the manufacturer and kept at 37.0 ± 0.3°C during the whole experiment with a homeothermic pad (Kent Scientific, Torrington, CT). RV hemodynamic parameters were first quantified. The right jugular vein was inspected through a <1-cm skin cut at the right aspect of the neck. A 1.9-Fr gauge sensor-tip pressure catheter (Scisence, London, ON, Canada) was inserted through a small incision and gently advanced into the RV. Once a smooth RV pressure curve was obtained, a 10-min stabilization period was allowed before data were recorded (PowerLab and LabChart v8.0, AD Instruments, Colorado Springs, CO). The catheter was thereafter removed, the right jugular vein was ligated, and the right carotid was exposed. The pressure-tip catheter was thereafter inserted into the right carotid artery and slowly advanced into the LV. Once a smooth LV pressure curve was obtained, a 10-min stabilization period was allowed before data were recorded. Data were later analyzed offline. For both RV and LV recordings, peak systolic and end-diastolic pressure as well as parameters assessing systolic function [maximal rate of rise of LV pressure (dP/dt_max)] and diastolic function [time constant of ventricular relaxation (τ), minimum dP/dt (dP/dt_min), and average dP/dt during isovolumetric relaxation (IRP dP/dt)] were quantified (LabChart v8.0, AD Instruments) in 50 consecutive beats, and the average was calculated for each animal.

In a subgroup of 21 rats (10 SED, 6 MOD, and 5 INT), pressure-volume loops were obtained with a conductance pressure-volume catheter and the ADVantage (ADV500) system (Transonic, Ithaca, NY). Calibration was conducted as per the manufacturer’s instructions before the procedure was initiated. After conventional hemodynamic measurements, the conductance catheter was gently introduced through the right carotid and advanced into the LV. The catheter was carefully mobilized until properly placed (i.e., a smooth sinusoidal curve with phase between 2 and 8° and magnitude between 1,400 and 2,600). After a 10-min stabilization period, simultaneous pressure and volume data were recorded. LV end-diastolic volume (LVEDV) and stroke work were quantified offline. LVEF was obtained as follows:

where LVESV is LV end-systolic volume.

Heart rate (HR) was obtained from a single-lead ECG strip, recorded during the hemodynamic experiments.

Wall stress estimation.

Wall stress at rest of both ventricles was estimated by means of hemodynamic and echocardiographic data. To estimate LV wall stress (σ), it was assimilated to a sphere and used the following previously described formula (23c):

where LVEDP is LV end-diastolic pressure, and IVS is interventricular septum.

The particular shape of the RV prevents it from reliably fitting into a sphere without major deviations. Accordingly, we modeled it as a truncated ellipsoid and estimated both longitudinal and circular RV wall stress (σ_long and σ_circ, respectively) as follows:

where RVEDP is RV end-diastolic pressure, a is LV end-diastolic length, b = LVEDD/2 + interventricular septum + RV end-diastole diameter, c = LVEDD/2, and h is RV wall thickness.

Fibrosis assessment.

RV fibrosis was assessed in histological preparations as previously described (3). Briefly, after euthanasia, hearts were embedded into paraffin. A basal ventricular section was subsequently obtained and stained with sirius red staining to identify collagen deposition. A single microphotograph, including the whole RV, was obtained (Panoramic Viewer, 3DHISTECH, Budapest, Hungary). Myocardial fibrosis was semiautomatically quantified with ImageJ (National Institutes of Health, Bethesda, MD), and results given as percentages. Perivascular and epicardial fibrosis were excluded from the analysis.

Statistical analysis.

Data were analyzed with SPSS Software for Windows (v19.0, IBM, Armonk, NY). A Gaussian distribution of all continuous variables was confirmed using a Kolmogorov-Smirnov test, and values are reported as means ± SE. Characteristics of the three groups of rats were compared by one-way independent ANOVA; if the ANOVA test showed an overall difference, then post hoc comparisons were performed with a least-significant difference test. P < 0.05 was considered for significance in all analyses.

Structural remodeling induced by exercise.

The echocardiographic evaluation at rest demonstrated that long-term endurance training promoted remarkable cardiac structural remodeling. Compared with SED rats, both MOD and INT training induced enlarged LVEDD and increased LV mass of a similar extent, resulting in a comparable degree of LV eccentric hypertrophy (Fig. 2 and Table 2). Invasive volume analyses confirmed a similar LV dilation in both trained groups (MOD and INT) compared with SED animals (Fig. 3, A and B). Conversely, we found a significant, intensity-dependent RV dilation (Fig. 2 and Table 2), which was progressive from SED to MOD (+16% RVEDA increase) to INT (+26% RVEDA increase) rats. A disproportionate RV dilation in relation to LV size (Fig. 4A) translated into a decreased LV-to-RV ratio in the INT group (Fig. 4B), thus reinforcing the concept of an unbalanced biventricular structural remodeling in intensively trained individuals.

Systolic functional remodeling induced by exercise.

Systolic and diastolic function were evaluated in the resting state through multimodal analyses. Consistent data obtained from echocardiography (Table 2), pressure-volume loop analyses (Fig. 3, C and D), and intracavitary pressure recordings (Fig. 5) demonstrated virtually unaltered LV systolic function after MOD or INT endurance training in healthy rats. Specifically, LVEF (Fig. 3C and Table 2) and LV dP/dt_max (Fig. 5D) showed similar values across groups.

In contrast, we found marked changes in RV systolic function after physical training. MOD physical activity induced a segment-specific increase in RV deformation: whereas myocardial deformation increased in the RV apical segment, there were no changes in RV basal segmental deformation (Fig. 6 and Table 2). Overall, the global RV systolic function evaluated with fractional area change was not modified (Table 2). Conversely, RV systolic function was remarkably impaired with the highest training load. Echocardiography displayed a diminished myocardial deformation both in the RV base and apex of the INT group compared with the MOD group (Fig. 6 and Table 2). The reduction in myocardial deformation led to a decreased global systolic function, as assessed with fractional area change (Table 2). A lower systolic contractile function of the RV in the INT group was further confirmed in invasive hemodynamic experiments (e.g., decreased dP/dt_max; Fig. 7).

Diastolic functional remodeling induced by exercise.

Exercise did not alter LV filling parameters at rest. Mitral E and e′M, LV IRT (Table 2) as well as τ (Fig. 5E), LV dP/dt_min (Fig. 5F), and IRP dP/dt (Fig. 5G) were unchanged in all groups.

Similarly, RV diastolic function was unaltered after MOD exercise. Nevertheless, the more intense exercise in the INT group prompted a decreased tricuspid E-wave and e′T, along with a prolonged IRT, and all of these parameters pointed to impaired RV diastolic function (Table 2). These data were supported by results from invasive hemodynamic experiments showing prolonged τ and decreased RV dP/dt_min and IRP dP/dt in INT-trained rats (Fig. 7, E–G).

RV and LV wall stress at rest.

Wall stress estimations at rest for the LV and RV are shown in Fig. 8. Consistent with physiological LV remodeling after MOD and INT training, we found no changes in LV wall stress at rest among all three groups. Conversely, longitudinal RV wall stress was significantly reduced in MOD-trained rats compared with SED rats; such a reduction was blunted in INT rats. Although it did not reach significance, a similar pattern was found for circular RV wall stress.

Myocardial fibrosis assessment in the RV.

RV fibrosis was assessed in histological preparations. Representative images are shown in Fig. 9A. Whereas MOD rats showed a similar fibrosis compared with SED rats, INT rats developed significantly increased RV myocardial fibrosis (Fig. 9B).

Exercise load determines the balance between LV and RV remodeling.

In our study, long-term endurance training of moderate intensity induced a comparable structural remodeling in both ventricles along with unchanged biventricular diastolic function and improved RV systolic function. In keeping with recent studies in athletes, moderate endurance training promoted harmonic biventricular dilation (12, 40) and normal biventricular systolic and diastolic function (8, 41). However, it is controversial whether balanced remodeling persists in highly trained individuals. Markedly disproportionate RV structural and functional changes have been shown in some athletes (23b, 31), but recent reports have yielded conflicting results and advocated for a balanced biventricular remodeling (5). Whereas small and heterogeneous cohorts and selection biases might explain some of these conflicting results, these differences suggest that large interindividual responses to exercise exist (23d, 36), likely determined by individual predisposition factors and the type and amount of exercise.

Our animal model, using individuals with a homogenous genetic background and very controlled exercise regimes, overcomes many of these potential biases. Through a multimodal approach, including echocardiographic and invasive hemodynamic assessment, our results strongly support that cardiac remodeling, after intense physical activity, diverges from that of moderate training and is characterized by a disproportionate and dysfunctional RV adaptation.

The RV is particularly sensitive to exercise-induced cardiac remodeling.

CO is determined by HR and stroke volume, which, in turn, depends on myocardial deformation and cavity size (4, 28). High-intensity endurance exercise requires marked increases in CO over periods of several hours, a requirement that can be reached through an increase in HR, cavity size, or myocardial deformation; in most cases, a combination of all them is needed. To accommodate the high demands of regular exercise, endurance training induces an enlargement of all cardiac chambers, thus enabling the heart to increase the stroke volume during exercise, at the cost of rising wall stress (10, 28). Altogether, the thin wall of the RV, its geometric shape, and the dramatic increase in intracavitary pressure during exercise beget a remarkably high RV workload and make it particularly vulnerable to the undesirable consequences of increased wall stress (23c). In our study, a high training load promoted a disproportionate RV dilatation and a decreased RV contractility, as demonstrated by impaired deformation and max dP/dt. At rest, mild decreases in RV contractility (20, 21, 23c) and a reduced myocardial flow (21) have been claimed to contribute to an enhanced functional and circulatory reserve during exercise in athletes (20). In our work, a very high training load yielded a disproportionate and unbalanced RV enlargement and reduced systolic function at rest accompanied by impaired RV diastolic function and fibrosis, suggesting that initial adaptive structural and functional changes become detrimental with certain cumulative training loads. On the other hand, modest changes in LV wall stress during exercise (23c) likely account for the lack of a deleterious LV remodeling. Whether myocardial flow is impaired after high training loads remains unknown.

The sort of exercise is an important determinant of cardiac remodeling. In predominantly strength sports (e.g., weightlifting), exercise bouts are characterized by variable increases in blood pressure, small changes in CO, and mild chamber dilation. Conversely, endurance training superimposes a cardiac volume overload that critically determines cardiac remodeling and likely contributes to the different response of the LV and RV. In an elegant experimental model of biventricular volume overload, Modesti et al. (29) found RV dysfunction and fibrosis with no significant functional or structural changes in the LV. Our results, in a transient, repetitive endurance exercise (and thus volume overload) model, are consistent with their findings. Indeed, we demonstrated RV systodiastolic dysfunction and fibrosis in the highest exercise load group, whereas LV function remained unaltered.

We documented an impaired RV diastolic function in intensively trained rats. RV diastolic dysfunction has been shown to predict fairly clinical outcomes in a clinical setting of RV pressure overload (11) and represent an early sign of RV damage in RV overload experimental models (23, 33). A previous report from our group (36) documented early signs of RV diastolic dysfunction in a subgroup of high-intensity endurance-trained athletes. However, RV diastolic function in athletes at rest is still a controversial issue (8, 38), and further research is warranted.

Overall, our data and previous work suggest that RV function is a critical regulator of cardiac performance during exercise bouts. Indeed, recent data from Heiskanen and colleagues (22) showed that exercise capacity in the general population is only determined by RV metabolism, reinforcing that cardiac performance limits are imposed by right heart characteristics.

Exercise-induced remodeling results in a segmental adaptation of the RV.

In our study, myocardial deformation analyses revealed diverging remodeling patterns of the basal and apical segments of the RV through different exercise intensities. Moderate-intensity exercise selectively enhanced apical deformation; intense endurance training, however, blunted such improvement and rendered the apex systolic function similar to that of sedentary animals. Furthermore, intense training associated with an impaired myocardial deformation in the basal segments of the RV. These results are consistent with data in humans, reporting a segmentary dysfunction of the basal segment of the RV in athletes (36, 39). Our work suggests that this change is critically governed by training load and supports an exercise load- and segment-dependent RV remodeling after long-term regular training. The causes of this segment-selective remodeling remain unexplored, but it is possible that the heterogeneous morphology of the RV, consisting of a trabeculated apical segment and a smooth and flat inlet segment with a larger capacity, may render the basal segment of the RV more vulnerable to exercise-induced volume overload bouts.

The clinical impact of a decreased RV basal segment deformation in most well-trained individuals is still unclear. In an exercise echocardiography study, La Gerche et al. (23a) demonstrated a lower RV basal myocardial deformation in athletes at rest and suggested a role as an adaptive, physiological feature that provides athletes with a larger contractility reserve during exercise. However, the results obtained through invasive hemodynamics in our animal model question this hypothesis and suggest that intensive training is associated with a true impairment in RV systolic function at rest.

Exercise load determines the sort of cardiac remodeling.

Our findings suggest that there is a threshold for training load determined by both intensity and duration beyond which cardiac adaptation might be no longer physiological but rather potentially deleterious, leading to fibrosis and becoming a potential trigger for arrhythmias. This finding further supports the recent notion that the relationship between physical activity load and cardiovascular outcomes follows a U-shaped, rather than lineal, relationship (18).

The mechanisms by which exercise turns deleterious at excessive loads remain obscure. RV wall stress acutely and linearly increases with exercise intensity (23c). It is therefore plausible that low-to-moderate exercise keeps RV wall stress within physiological values, whereas extreme exercise raises it away from a safety range. Similarly, transient systemic inflammation has been found in a dose-dependent way after extenuating exercise bouts. Extrinsic factors, such as subclinical myocarditis and performance-enhancing drugs, have been proposed to contribute to exercise-induced cardiac damage (18).

From a clinical point of view, it would be interesting to identify such a “reversal point.” In our work, we tested only two exercise loads and thereby cannot reliably infer a specific exercise threshold. Data for exercise-induced atrial fibrillation have suggested a variety of thresholds with potential clinical relevance (1, 6, 18), but these are lacking for the RV. Nevertheless, the ratio between benefits and deleterious consequences of physical activity is likely multifactorial and involves not only the duration and intensity of physical activity but also rather genetic and acquired individual adaptive mechanisms to exercise (18, 36).

Limitations.

Some limitations of our work should be acknowledged. First, translation of experimental conclusions to humans always warrants caution. It is difficult to estimate how our two different exercise protocols translate into human physical activity. As a rough approximation, if the lifespan of Wistar rats is 2 yr, our 18-wk exercise protocol (2 wk of progressive training plus 16 wk of intensive exercise) would be roughly equivalent to 10 yr of daily exercise training in humans. Regarding training load, our intense endurance training has been suggested to correspond to 85% of maximum O₂ uptake (3). We estimate that the moderate training protocol would correspond to 60% of maximum O₂ uptake (42). Finally, the sedentary group recapitulates a rather extreme form of inactivity; some levels of limited, voluntary exercise would provide a fairer approximation to the lifestyle of most individuals in the general population.

Only young male rats were tested in this study, so results might not be extrapolated to females or older ages. The effect of stress should always be considered a potential confounder. Nevertheless, maximum efforts were taken to minimize stress responses. Furthermore, we have previously assessed stress levels in intense exercise-trained rats and ruled out any significant effects (3). Due to a lower physical demand, we do not expect any significant effect of stress in moderate exercise-trained rats either.

Finally, most results were obtained under anesthesia. Most anesthetic agents yield some degree of hemodynamic perturbations that could potentially influence our results. Our choice for isoflurane was based on its unique combination of rapid and transient effect with minor hypotensive effects in the absence of exceedingly long exploration times (35).

Conclusions.

In this experimental model of long-term endurance training, exercise load critically determines a biphasic response of the RV performance. Moderate endurance training caused an adaptive RV remodeling characterized by mild RV dilatation and an increase in RV apical deformation. However, vigorous training caused a marked RV remodeling with disproportionate RV dilatation and impaired diastolic and systolic function.