Exercise Training Improves Diastolic Function in Heart Failure Patients

It is well established that clinical presentations are similar in heart failure patients with reduced and preserved ejection fraction. Many studies have shown that exercise tolerance is reduced to the same extent in heart failure with reduced ejection fraction (HFREF) as it is in heart failure with preserved ejection fraction (HFPEF) (^20,24). Moreover, mounting evidence indicates that left ventricular diastolic function is impaired in patients with heart failure, independent of ejection fraction (^39,40), and that contractile abnormalities can be present in patients with HFPEF despite preserved left ventricular systolic function (^11,37). On the other hand, several differences are evident in left ventricular functional and structural characteristics in patients with HFREF and HFPEF. Increased left ventricular stiffness in tandem with concentric remodeling and increased left ventricular mass are common features in HFPEF. This contrasts with left ventricular dilation and resting contractile dysfunction, which are more common in HFREF (¹¹). The unequal remodeling characteristics suggest that outcomes can be distinct despite similar treatment because the positive outcomes in clinical trials with HFREF have contrasted with the neutral results recently reported in HFPEF (²⁸).

It is well known that exercise training improves exercise capacity and quality of life in patients with heart failure (²⁶). Ample evidence supported in recent meta-analyses indicates that exercise training can also improve left ventricular function, decrease dilatation, and reduce hospital admissions in patients with HFREF (^9,15), but little is known about patients with HFPEF (¹⁸). Indeed, it has never been shown whether exercise training can improve diastolic dysfunction in patients with HFPEF. Given the association of mitral filling patterns with functional class and prognosis in patients with heart failure (^12,29), most studies hitherto have used mitral inflow velocities and time intervals to measure the effect of exercise training on diastolic dysfunction (^4,18). However, it is unclear whether mitral inflow velocities have a uniform response to exercise training because patients with different levels of dysfunction show distinct filling patterns (²²). For example, early (E) mitral flow has a prolonged deceleration time (DT), and its contribution relative to atrial (A) mitral flow to ventricular filling is reduced in patients with mild diastolic dysfunction, whereas E mitral flow becomes predominant with rapid deceleration in patients with more severe diastolic dysfunction (²²). These unique filling patterns suggest that exercise training may require unique responses to improve diastolic dysfunction in patients with different levels of diastolic dysfunction. Given the hypothesis that left ventricular functional and structural differences can lead to different responses to treatment and exercise in patients with reduced and preserved ejection fraction, this study aimed to investigate the effects of exercise training on exercise tolerance and left ventricular function and structure in heart failure patients with preserved, mild, and moderate to severe reduction of ejection fraction.

METHODS

Patient recruitment.

Patients admitted to Bnai Zion Medical Center, Haifa, Israel, were considered eligible to participate in the study if they presented signs or symptoms of heart failure. Exclusion criteria included uncontrolled hypertension, unstable angina pectoris, abnormal hemodynamic response, uncontrolled cardiac arrhythmias, ischemic ECG changes during stage 1 of the exercise tolerance test (modified Bruce protocol), uncontrolled metabolic disease (e.g., uncontrolled diabetes and thyroid disease), and medical conditions that limit participation in exercise (e.g., peripheral arterial occlusive disease and musculoskeletal disorder). The hospital ethics committee approved the study protocol, and written informed consent was obtained from every patient.

Study design.

Subjects were randomly assigned to exercise training plus usual care or usual care alone (control group) in a randomization ratio of 2:1. Twice as many patients were assigned to the exercise training group to compensate for the attrition and lower adherence in patients with heart failure who participate in cardiac rehabilitation programs (³). Patients who completed less than 80% of exercise sessions were excluded from the analysis. Usual care consisted of regular appointments with a cardiologist and optimized medication. Patients in the control group did not receive instructions or any form of exercise training. Patients in both groups were assessed at baseline and 6 months after randomization (follow-up).

Subjects were classified into three patient groups on the basis of baseline left ventricular ejection fraction (LVEF), according to the recommendations of the American Society of Echocardiography (²¹): preserved (>55%), mild (from 45% to 54%), and moderate to severe (<45%) reduction of LVEF.

Exercise testing.

Each subject performed a symptom-limited graded exercise treadmill test according to a standard modified Bruce protocol. To examine exercise tolerance, we recorded treadmill exercise time and estimated METs according to the guidelines provided by the American College of Sports Medicine (¹):

where fractional grade is expressed in decimal form and speed is expressed in meters per minute (1 mph is equal to 26.8 m·min⁻¹).

The percentage of change in exercise tolerance (METs) was calculated as follows: (follow-up METs − baseline METs) / baseline METs × 100. A clinically meaningful improvement in exercise tolerance was judged to a 10% increase from baseline to follow-up METs, as previously described (¹⁸).

Blood pressure (by auscultation) and a 12-lead ECG were recorded at rest, throughout exercise, and at regular intervals during recovery until HR and ECG had returned to baseline (recovery data not used for this study). Maximal HR was determined as the highest HR achieved during the last 30 s of exercise testing. Treadmill tests were terminated according to the guidelines recommended by the American College of Sports Medicine (¹). The same cardiologist, who was blinded to the treatment group, supervised all treadmill tests.

Echocardiographic evaluation.

All subjects underwent a complete resting echocardiographic examination using a Siemens ACUSON Sequoia machine. Three consecutive cardiac cycles were analyzed and averaged for each patient. LVEF was measured using the modified biplane Simpson method from the apical four- and two-chamber views, whereas left ventricular end-diastolic diameter (EDD) and left ventricular end-systolic diameter (ESD) were measured at M-mode in the parasternal long-axis view. Transmitral inflow velocities were assessed by pulsed-wave Doppler, with the sample volume placed between the mitral leaflet tips in the apical four-chamber view. The peak E and late (A) transmitral velocities and the E-wave DT of E filling velocity were assessed, and the E/A ratio was calculated for the evaluation of diastolic function. One cardiologist performed all echocardiographic evaluations and was blinded to the treatment group.

Exercise training.

Patients trained three times a week for 6 months. In the first month, each exercise session consisted of 10 min of warm-up exercises, 15 min of aerobic exercise, and 10 min of cooldown with stretching exercises. The aerobic exercise consisted of interval training performed on a treadmill or bicycle ergometer—five sets of 3-min exercise at 70%–75% of maximal HR interspersed with 1-min active recovery at 45%–55% of maximal HR. In the following 5 months, aerobic exercise duration was progressively increased up to 35 min by increasing the number and duration of exercise sets (7 × 5 min), while maintaining the duration of active recovery. HR was continuously monitored by ECG with each minute rhythm strip hard copy recorded. Exercise intensity was adjusted progressively throughout the study to ensure that all exercise training sessions were performed within the established HR.

Statistical analysis.

Normal data distribution was confirmed by the D’Agostino–Pearson omnibus test and Shapiro–Wilk normality test. Data are presented as mean ± SD. Continuous variables with nonnormal distribution were log-transformed before statistical testing, and means were transformed back for presentation. To examine the effect of treatments on exercise tolerance, cardiac function, and left ventricular dimensions, a two-way mixed-model ANOVA was used to compare results between treatment groups over time (treatment × time). When significant interactions were observed, t-tests were applied to determine the location of differences within each treatment relative to baseline as well as between treatments at baseline and follow-up. The association between treatments and the proportion of patients who improved exercise tolerance above the clinical threshold (10%) was examined through contingency tables with the χ² statistic. To account for the inflation of type I error that may occur with multiple comparisons, statistical significance was corrected with the Holm–Bonferroni method where appropriate. Mean differences were adjusted for potential confounders where appropriate. P < 0.05 was considered indicative of statistical significance.

The estimation of sample size was based on previous research (¹⁵), which has demonstrated that exercise training induces a moderate effect on left ventricular function, whereas the control group is expected to have no change. This analysis revealed that on the basis of the randomization ratio of 2:1, in each patient group, 19 patients in the exercise training group and 9 in the control group would be required to detect a time × treatment interaction with a moderate effect size (F = 0.25) in left ventricular function with 80% of probability, at an α level of 0.05. The total number of patients was inflated to account for patients who would not be able to complete the program or would be unable to perform the follow-up evaluations (dropouts) (³).

RESULTS

A total of 103 patients were eligible and agreed to participate in this study. Among patients from the exercise training group (n = 67), one was unable to attend the exercise program, and one did not perform the follow-up echocardiographic evaluation. Of the patients who received usual medical care and constituted the control group (n = 36), one was unable to attend both echocardiographic evaluations, whereas two patients were unable to attend the echocardiographic follow-up evaluation. Thus, a total of 98 patients completed the study and were included in the analysis. The ratio of patients assigned to exercise training and usual care alone was similar in all patient groups, that is, 22 versus 12 patients in the moderate to severe group, 23 versus 10 patients in the mild group, and 20 versus 11 patients in the preserved ejection fraction group. No adverse events were registered during exercise testing or during aerobic interval training. Our patients tolerated the exercise training protocol and did not have any problems in maintaining their target HR range throughout the exercise sessions.

Baseline characteristics.

Baseline characteristics are shown in Table 1. There was a greater proportion of women and a lower prevalence of prior myocardial infarction in patients with preserved LVEF. On the other hand, there were more patients treated with angiotensin-converting enzyme inhibitors and spironolactone in the moderate to severe group compared with the other patient groups. No significant differences were found between exercise training and usual care alone in any patient group with respect to demographic, cardiovascular risk factor profile, and medication regimen characteristics.

Left ventricle function.

LVEF increased after exercise training in all patient groups (Fig. 1), whereas it remained unchanged with usual care alone (treatment × time interaction in preserved: F = 6.33, P = 0.02; mild: F = 15.53, P < 0.01; moderate to severe: F = 28.78, P < 0.001).

Diastolic function improved after exercise training but not after usual care alone in patients with preserved (treatment × time interaction: F = 4.73, P = 0.02) and mild systolic dysfunction (treatment × time interaction: F = 3.97, P = 0.03). The E/A ratio increased after exercise training in both patient groups, whereas it remained unchanged with usual care alone (Fig. 2A). Furthermore, the E-wave DT decreased after exercise training in both patient groups, whereas it remained unaltered after usual care alone (Fig. 2B).

When the data were pooled, diastolic function was similar between the exercise training and usual-care-alone groups in patients with moderate to severe systolic dysfunction (Fig. 2). However, among these patients, there were different degrees of diastolic dysfunction, which led us to conduct separate analyses in patients that presented a short DT (DT < 160 ms) (²⁵), indicating severe diastolic dysfunction (restrictive filling), and patients that presented prolonged DT (DT > 160 ms), indicating mild diastolic dysfunction (impaired relaxation). In patients with severe diastolic dysfunction, exercise training increased the short DT toward normal (from 129.8 ± 6.1 to 166.6 ± 11.8 ms, P < 0.001) and promoted a significant decrease in the E/A ratio (from 1.58 ± 0.11 to 1.24 ± 0.22, P = 0.02), whereas in patients with mild diastolic dysfunction, exercise training decreased the prolonged DT (from 243.5 ± 37.0 to 221.5 ± 19.0 ms, P < 0.001) and enhanced the E/A ratio toward normal (from 0.79 ± 0.11 to 0.90 ± 0.08, P < 0.001). No such changes were observed in the usual-care-alone group.

Left ventricle dimensions.

Left ventricular dimensions were altered by exercise training in the mild and moderate to severe patient groups (F = 3.75, P = 0.035 and F = 9.39, P < 0.01, Fig. 3) but not in patients with preserved systolic function. In patients with moderate to severe systolic dysfunction, left ventricular dimensions decreased after exercise training, whereas this was not observed in the control group. In contrast, in patients with mild systolic dysfunction, exercise training promoted a decrease in ESD (P < 0.001) but not in EDD, whereas usual care alone had no effect on left ventricular diameters.

Exercise tolerance.

Baseline and follow-up exercise tolerance values are shown in Figure 4. Exercise tolerance increased in all exercise groups, whereas it remained unchanged in all usual-care-alone groups (treatment × time interaction: preserved: F = 4.81, P = 0.04; mild: F = 6.17, P = 0.02; moderate to severe: F = 4.07, P = 0.05). Moreover, the number of patients with a 10% improvement in exercise tolerance, conventionally used as clinically relevant, was higher in all exercise patient groups when compared with the usual-care-alone groups (moderate to severe: 62% vs 22%, χ² = 3.96, P = 0.04; mild: 48% vs 11%, χ² = 3.61, P = 0.05; preserved: 45% vs 0%, χ² = 7.51, P < 0.01).

DISCUSSION

The main findings of the present study were that exercise training improved exercise tolerance and cardiac function in patients with moderate to severe, mild, and preserved LVEF, with all those receiving usual medical care alone remaining unaltered.

It is well known that exercise training can improve systolic function in patients with heart failure and reduced ejection fraction (HFREF) (¹⁵). In this study, we showed that left ventricular performance could also improve in heart failure patients with near-normal and normal LVEF (HFPEF). These results are intriguing because left ventricular function and its driving mechanisms differ in HFREF and HFPEF. The preserved LVEF indicates that left ventricular performance is well maintained, but recent studies have shown that myocardial contractile dysfunction can be present in patients with HFPEF (¹¹). For example, recent evidence demonstrated that systolic mitral annular velocity and left ventricular longitudinal shortening are reduced in patients with HFPEF compared with controls (^7,34). It is speculated that the reasons for the depressed contractile performance in these patients are concentric remodeling and ventricular stiffening, resulting in reduced myocardial contractility and systolic reserve (^6,39). Indeed, although systolic function is arguably not as impaired in HFPEF as in HFREF, recent studies have shown that mild limitations in resting contractile function can become quite limitative during physical exertion (^5,24). Exercise training did not change concentric remodeling in our patients with HFPEF, suggesting that it might improve contractile function by reducing ventricular stiffening. Numerous factors determine myocardial stiffness, including the expression of different isoforms and phosphorylation status of the cytoskeletal protein titin, phosphorylation of sarcomeric proteins, and the amount, distribution, and architecture of fibrillar collagen in the extracellular matrix (^6,22). Evidence from experimental studies in animals indicates that exercise training can reduce collagen volume fraction after myocardial infarction (³⁶), but the consequences of exercise training on the expression and phosphorylation of myofilament and cytoskeletal proteins remain unclear. An alternative candidate mechanism is the restoration of β-adrenergic intracellular signaling transduction after exercise training (¹⁰). This could improve contractile dysfunction related to receptor downregulation and abnormal calcium metabolism on one hand (³²) and reduce cardiomyocyte stiffness through the phosphorylation of cytoskeletal (titin) or sarcomeric proteins on the other (³³). It should be noted that changes in left ventricular loading conditions and chamber geometry have a significant influence on LVEF. However, left ventricular end-diastolic dimensions did not change with exercise training in our patients with mild systolic dysfunction and HFPEF, supporting the notion that left ventricular contractile function may have improved in these patients.

Another main finding of this investigation was that mitral inflow velocities and filling patterns improved in all exercise training groups, remaining unchanged after usual care alone. This is an important finding because mitral inflow velocities and filling patterns are associated with functional class and prognosis in heart failure patients (¹²). In contrast to normal patients, the E-wave velocity during left ventricular relaxation is reduced in association with a prolonged DT in patients with mild systolic dysfunction (impaired relaxation) (²²). When diastolic pressures increase such that atrial contraction cannot increase left ventricular filling, the E-wave becomes predominant again in tandem with a rapid DT, indicating severe diastolic dysfunction (²²). Severe diastolic dysfunction is associated with a poor prognosis, especially if the restrictive filling pattern persists after treatment (²⁹). In this study, exercise training improved diastolic dysfunction in patients with impaired relaxation as well as advanced diastolic dysfunction in many patients with heart failure and moderate to severe systolic dysfunction. The E-wave velocity contribution to left ventricular filling increased after exercise training in patients with impaired relaxation, whereas it decreased toward normal in patients with a restrictive filling pattern. Moreover, DT returned to values toward normal in both groups after exercise training, independent of its initial value. These results are consistent with many previous studies in HFREF (^4,35) but are in contrast with those reported in two recent studies (^18,31), in which diastolic function did not improve with exercise training in patients with near-normal and normal LVEF. This discrepancy is likely to be due to differences in sample size, inclusion criteria, definition of normal ejection fraction, and exercise training protocol. Another important element to retain from these data is that exercise training seems to have an extensive mechanistic action because the causes of diastolic dysfunction may not be the same in patients with HFREF and HFPEF. It has been proposed that reduced elastic recoil due to impaired contractile function and left ventricular dilatation decrease intraventricular pressure gradient in patients with HFREF (¹¹). Our data indicate that both elements improve in these patients with exercise training. In contrast, the end-diastolic pressure–volume relation shows an inappropriate upward and leftward shift in patients with HFPEF, indicating increased filling pressures associated with left ventricular stiffness (²). Our results indicate that end-diastolic dimensions remain unaltered with exercise training in patients with HFPEF, leaving it unclear whether it has any effect on left ventricular passive stiffness. On the other hand, left ventricular relaxation is impaired in both subsets of heart failure (³⁸), indicating abnormalities in calcium handling or phosphorylation of sarcomeric proteins (²²). There is indeed evidence from experiments in animals that exercise training normalizes calcium handling in tandem with the expression of SERCA2a and phosphorylation status of phospholamban (¹⁷) and that it restores the phosphorylation of sarcomeric proteins, in particular myosin light chain 2 (¹⁰). It is important to consider that mitral flow parameters are affected by left ventricular loading conditions. Numerous factors can mediate alterations in loading conditions induced by exercise training. An obvious and intuitive suggestion might be neurohormonal activation because it is elevated in all patients with heart failure with increased sympathetic activation and circulating catecholamines (²⁰). Indeed, exercise training improves baroreflex sensitivity and decreases the activation of the sympathetic nervous system in patients with HFREF (³⁰). This decreases afterload and reduces peripheral vascular resistance, all of which help to improve cardiac function and decrease remodeling.

Exercise intolerance, manifested by premature fatigue and dyspnea during physical exertion, is the prime manifestation of heart failure. It is associated with reduced quality of life and increased mortality (¹⁶). It is well known that exercise training improves exercise tolerance in patients with moderate to severe systolic dysfunction (²⁶). In the present study, most patients in all exercise training groups improved exercise tolerance above the limit that is considered clinically relevant, which was not observed with usual care alone. However, our data extended the previous observations (^18,26) by showing that exercise tolerance improved more pronouncedly in patients with moderate to severe systolic dysfunction compared with the other groups. These observations suggest that exercise training should be considered as a fundamental nonpharmacological treatment in heart failure, eventually with greater benefits in patients with worse prognosis. Nevertheless, this requires further investigation.

Despite these findings, the literature thus far cannot elucidate the mechanisms by which exercise training improves exercise tolerance in different subsets of heart failure. It is complex to discern the importance of cardiac function to exercise tolerance in different subsets of heart failure because left ventricular function and structure differ in patients with reduced and preserved ejection fraction. Nonetheless, left ventricular function improved in all exercise training groups, in parallel with exercise tolerance, and remained unchanged with usual care alone, suggesting that exercise training improves exercise tolerance in concert with cardiac function. Numerous studies have postulated that exercise tolerance is weakly associated with left ventricular function (¹⁹), but a limitation of most exercise studies, as well as ours, is that mechanistic measurements have been made only at rest. Our data indicate that exercise training improves left ventricular filling dynamics in all subsets of heart failure. It is proposed that increased left ventricular pressures associated with a small preload reserve lead to exercise intolerance, resulting from failure to increase stroke volume and increased pulmonary venous pressure (^23,27). By improving left ventricular filling dynamics, exercise training could decrease pulmonary vascular resistance and increase stroke volume during exercise. This could, in turn, attenuate the appearance of dyspnea and improve exercise tolerance.

Besides these mechanisms, studies in patients with dilated and ischemic cardiomyopathies reported that exercise training improves endothelial dysfunction and decreases vascular resistance (^13,14), resulting in improved ventricular–vascular responses during exercise and in improved exercise capacity (^13,14). Although the exercise mechanistic role remains to be established, mounting evidence indicates that patients with normal or near-normal systolic dysfunction also have abnormal vasomotor and ventricular–vascular coupling responses during exercise (⁵) and could benefit from exercise training. Moreover, the importance of functional, metabolic, and histological changes in skeletal muscle to improving exercise tolerance in patients with HFREF has been widely recognized (⁸). However, this remains to be elucidated in patients with mild systolic dysfunction and preserved ejection fraction.

Limitations.

A limitation of this study is that we have used exercise grade and speed on a treadmill test as the indicator of improved functional capacity, instead of a direct measure of exercise capacity such as oxygen consumption. A few studies have shown that a learning effect may occur with exercise duration, and indeed, our subjects did not perform a familiarization exercise test to overcome this issue. However, the designated assessor was blinded to the treatment groups, and some of the control groups showed slight declines in exercise tolerance, indicating that the differences between groups reflect true changes in exercise capacity and not a learning effect.

It can be argued that the sample size is small in all groups, in particular, those pertaining to controls. However, estimates from previous literature indicate that our sample sizes are large enough to detect differences in exercise tolerance and left ventricular function induced by exercise training and control groups and avoid type I errors with 80% certainty.

Patient management, exercise prescription, and implications for future research.

Exercise intolerance is the cardinal manifestation of heart failure, and it is common to all heart failure patients. The present findings showed that exercise training improves exercise tolerance and left ventricular function in heart failure patients independent of their baseline condition. Thus, it is recommended that exercise training be included in the treatment of all heart failure patients. The question that remains open is which mechanisms are used by exercise training to improve left ventricular function and exercise tolerance in patients with HFREF and HFPEF. Indeed, the effects of exercise training on the cellular and molecular mechanisms of left ventricular stiffness and active relaxation and their association with systolic and diastolic reserves represent promising avenues of research in the future. Future studies should also investigate whether exercise training improves endothelial dysfunction, ventricular–vascular coupling responses to exercise, and skeletal muscle characteristics in patients with HFPEF.

Our results also indicate that moderate-intensity interval training offers a sufficiently strong stimulus to improve exercise tolerance and cardiac function. High-intensity exercise training has been shown to have a superior cardiovascular effect in patients with HFREF (³⁵). However, it remains to be determined if high-intensity exercise training interferes with the safety and compliance of heart failure patients compared with moderate exercise training programs.

Summary.

The present study showed that exercise training could improve exercise tolerance and left ventricular function in heart failure patients, independent of their baseline systolic dysfunction. These results are relevant because they indicate that exercise training can improve the course of heart failure independent of the degree of baseline left ventricular dysfunction.

This study was supported by grant SFRH/BD/33122/2007 from the Fundação para a Ciência e a Tecnologia, Portugal.

None of the authors have professional relationships with companies or manufacturers who will benefit from the results of the present study.

The authors received no funding for research from the National Institutes of Health, Wellcome Trust, or the Howard Hughes Medical Institute.

The authors declare no conflict of interest.

The results of the present study do not constitute endorsement by the American College of Sports Medicine.