Article Text

PDF

Original article
Independent association of left atrial function with exercise capacity in patients with preserved ejection fraction
  1. Kenya Kusunose,
  2. Hirohiko Motoki,
  3. Zoran B Popovic,
  4. James D Thomas,
  5. Allan L Klein,
  6. Thomas H Marwick
  1. Heart and Vascular Institute, Cleveland Clinic, Cleveland, Ohio, USA
  1. Correspondence to Dr Thomas H Marwick, Heart and Vascular Institute, Cleveland Clinic J1-5, 9500 Euclid Avenue, Cleveland, OH 44195, USA; marwict{at}ccf.org

Abstract

Objective Left atrial (LA) function is an important contributor to cardiac function in a variety of disease states and especially with exercise. The purpose of this study was to assess the role of LA function in the left ventricular (LV) filling pressure (E/e') response to exercise and exercise tolerance in patients with preserved LV ejection fraction (LVEF).

Methods LA contractile function (LA strain peak negative), conduit function (LA strain peak positive), and reservoir function (LA strain total) were measured from speckle tracking in 486 patients with preserved LVEF and negative exercise echocardiography for ischaemia. Maximal exercise tolerance was expressed as the estimated metabolic equivalents (METs). Per cent predicted METs was calculated in accordance with described nomograms. Features associated with exercise capacity were sought in a multiple linear regression, and compared using standardised β.

Results Patients with increased stress E/e' had significantly lower LA strain profiles and larger LA volume than the remainder. LA strain total was strongly associated with exercise capacity in multivariate analysis (ß=0.21, p<0.001). E/e' rest was also associated with exercise capacity (ß=−0.11, p=0.001). Other independent correlates of exercise capacity were age (ß=−0.36, p<0.001), male gender (ß=0.34, p<0.001) and body mass index (ß=−0.23, p<0.001). The best predictor of per cent predicted METs was total LA strain in multivariate analysis.

Conclusions The association of reduced LA function with impaired exercise capacity was similar to that of elevated LVE/e', emphasising the role of ventriculo-atrial coupling to cardiac dysfunction with preserved LVEF.

  • Echocardiography
  • exercise capacity
  • atrial function
  • cardiac function
  • myocardium
  • myocardial function
  • diastolic function
  • ventricular function
  • EBM
  • doppler
  • echocardiography (three-dimensional)
  • doppler ultrasound
  • contrast echocardiography
  • tissue doppler
  • stress echocardiography

Statistics from Altmetric.com

Left atrial (LA) volume is a sensitive surrogate marker of the severity and duration of left ventricular (LV) diastolic dysfunction and is an independent predictor of exercise capacity with normal LV systolic function.1 ,2 Recent work has shown that direct measurement of LA function using the speckle tracking method is related to diastolic dysfunction.3 ,4 However, there is limited information regarding the role of LA function on exercise capacity, and the incremental value of LA function in addition to standard measurements is poorly understood.5

Increased LV diastolic pressure with exertion has been identified as playing an important role in the genesis of exertional dyspnoea,6 and allows dyspnoea to be linked with LV filling pressure (E/e') response to stress.7 Increased post-exercise LV pressures are associated with decreased exercise capacity and outcome.6 ,8 The role of LA function in this response is undefined. The aim of this study was to assess the role of LA function in the LVE/e' response to exercise and exercise tolerance in patients with preserved LV ejection fraction (EF) without CAD.

Methods

Study population

Between January and April 2010, 672 patients had an exercise echocardiogram at the Cleveland Clinic for clinical indications, mainly for the exclusion of CAD, and the investigation of chest pain or dyspnoea. The study group of 517 patients was defined after exclusion of patients who were in atrial fibrillation/flutter at the time of exercise, EF <45%, moderate or severe valvular disease, echocardiographic evidence of exercise-induced myocardial ischaemia, or had known coronary artery disease. Thirty-one subjects (6%) were excluded due to poor image quality, therefore, 486 subjects were included for final analysis. Clinical variables and body mass indices were recorded at the time of the exercise echocardiogram. A history of diabetes mellitus, hypertension, dyslipidemia and smoking were abstracted from the medical record and entered into a prospectively maintained database. The protocol was approved by the Cleveland Clinic Institutional Review Board.

Echocardiography

Transthoracic echocardiography was performed in the left-lateral decubitus position by experienced sonographers before and after symptom-limited exercise using a commercial ultrasound machine (iE33, Philips Medical Systems, Bothell, Washington, USA). Standard imaging windows were used, including parasternal long and short axes, as well as apical two- and four-chamber views. Echocardiographic images were digitally recorded and downloaded to an imaging server for off-line analysis. End-diastolic LV dimension, end-systolic LV dimension, interventricular septal thickness and LV posterior wall thickness were measured from the M-mode echocardiogram. The LV mass index was estimated from the formula of Devereux et al9 The LVEF was calculated using the Simpson's biplane method. All measurements were performed and averaged over three cardiac cycles.

The following indices of LA volume were calculated according to a previous study.3 ,10 Total LA volume is a composite of three distinct phases: the maximum volume, the minimum volume and the pre-contraction volume. Total LA stroke volume was calculated as maximum LA volume-minimum LA volume. We described the LA volume index as the maximum LA volume divided by body surface area. The LA expansion index was calculated as (total LA stroke volume/minimum LA volume) ×100. The following measurements were taken from pulmonary vein velocities obtained at the right upper pulmonary vein: peak S-wave inflow velocity during ventricular systole, peak D-wave inflow velocity during the early phase of ventricular diastole, atrial reversal velocity (A-wave) during the atrial contraction and the corresponding S/D ratio. The early transmitral flow velocity profile (E) was acquired in the apical four-chamber view using pulsed-wave Doppler at the level of the mitral valve tips during diastole. The early mitral annular tissue velocity (e') was also measured in the apical four-chamber view with the sample volume positioned at both the septal and lateral mitral annulus and e' being the average of these two values. All measurements were made off-line by an investigator blinded to all clinical and demographic information. Resting diastolic function was graded as normal, mild dysfunction (Grade I), moderate dysfunction (Grade II) and severe dysfunction (Grade III) using the ASE recommendations.11

Exercise echocardiography

All patients underwent symptom-limited exercise treadmill testing using standard treadmill protocols with 12-lead ECG monitoring. At a usual interval of 1–2%min after exercise (ie, following acquisition of the wall motion study post-exercise, sometimes with an additional delay to allow for de-merging of E- and A-waves), mitral inflow and annulus Doppler data were again obtained. Elevated LVE/e' was defined by the presence of E/e' >13.12 The duration of the exercise was measured and the maximal exercise tolerance was expressed as the estimated metabolic equivalents (METs). Predicted exercise capacity was calculated in accordance with described nomograms (predicted METs; in males =18 (0.15×age), in females =14.7 (0.13×age)),13 and per cent predicted METs (%PMETs) was described as the difference between actual and predicted METs divided by predicted METs. Reduced exercise capacity was identified by PMETs <85%, which has been related to death from any cause as well as cardiac causes.13 ,14

LA function

LA strain measurements were obtained from a high frame rate (60–80%frame/sec), acoustic-captured grey-scale-recorded images in the apical four- and two-chamber views at rest. Strain was analysed off-line (Syngo VVI, Siemens Medical Solutions, Mountain View, California, USA). After manual definition of the LA endocardial border, this was automatically tracked throughout the cardiac cycle. The software divided the LA endocardium into six segments, resembling the approach used when studying LV strain. Segments in which inadequate traces were obtained were excluded from further analysis, and the remaining segments were averaged for each view. Strain was defined as instantaneous lengthening or shortening. We used the onset of the P-wave as the reference point for the calculation of LA strain because this point most relevantly represents the LA cavity just before its contraction.3 ,15 ,16 The use of the P-wave as the reference point enabled the recognition of peak positive strain, which corresponded to LA conduit function, the peak negative strain, which corresponded to LA contractile function, and the sum of these values (LA strain total), which corresponded to LA reservoir function (figure 1).

Figure 1

Left atrial strain curve. Negative peak strain, which corresponded to LA contractile function; positive peak strain, which corresponded to LA conduit function; and the sum of these values, total strain, which corresponded to LA reservoir function, were obtained.

Statistical analysis

Data are reported as mean±SD. Comparisons between two groups were done with a t-test (or Wilcoxon signed rank tests for non-normal data). We checked for collinearity between the independent variables and exercise capacity by using Pearson's correlation coefficients. The diagnostic content of several echocardiographic parameters for separating stress E/e' and %PMETs was determined by comparison of the area under the curve (AUC) of receiver-operating characteristic curves (ROC) (MedCalc Software 12.0.4.0, Mariakerke, Belgium). Linear regression analysis was used to evaluate the associations between several variables and E/e' stress, METs and %PMETs. We found no evidence for collinearity problems in our model without the model including age×gender (variance inflation factor values <2). We performed analysis stratified by any variables for interaction. For a sample size that was about 460, multiple linear regression was sufficiently powered for inclusion of 11 predictors.17 Interobserver and intraobserver variability for LA strain total were examined. Measurements were performed in a group of 15 randomly selected subjects by one observer then repeated on two separate days by two investigators who were unaware of the other's measurements and of the study time-point. Reproducibility was expressed as the mean per cent error (absolute difference divided by the average of the two observations). Statistical analysis was performed using a standard statistical software package (SPSS software V.18.0, SPSS Inc.), and statistical significance was defined by p<0.05.

Results

Patient characteristics

Among the 486 patients included, 306 (63%) were male, the mean age was 56±13%years, and 85% were white. The baseline clinical and echocardiographic characteristics are outlined in table 1. They show an expected prevalence of risk factors, preserved LV function and normal LV mass. Normal diastolic function was present in 238 patients (49%), mild diastolic dysfunction in 172 (36%), and moderate/severe diastolic dysfunction in 76 (15%).

Table 1

Clinical characteristics

Correlates of exercise capacity

Men had a greater exercise capacity than women, and the magnitude of this difference decreased with age (p=0.007) (figure 2). An additional multivariate model with age×gender (multiple r2=0.60) (table 2, model 1) showed that METs also correlated with rest E/e' and total LA strain. Total LA strain and rest E/e' had no interaction with any variables after adjustment by age×gender (all p>0.05).

Figure 2

Association of gender with exercise capacity by age. Men had a greater exercise capacity than women, the magnitude of this difference decreased with age (p=0.007). These lines are the 10th, 50th and 90th percentile line plotted from LA strain total and E/e' rest in males and females. LA strain total spectrum was wider than the E/e' rest spectrum.

Table 2

Multivariate analysis of association of clinical and echocardiographic variables with exercise tolerance (METs) including age and gender interaction (model 1), METs (model 2), predicted METs (%PMETs, model 3), and raised filling pressure (E/e' stress, model 4)

There were no differences in LVEF, LV mass index, LA volume index and pulmonary vein flow profiles between patients with preserved and reduced exercise capacity (%PMETs <85%) (table 1). Patients in the reduced exercise capacity group had significantly lower LA strain profiles than patients in the preserved exercise capacity group. Patients with reduced exercise capacity had significantly higher rest and stress E/e' than the remainder. METs also correlated with age, gender, BMI, heart rate, smoking, hypertension, E/e' rest and LA strain total (multiple r2 =0.59) (table 2, model 2). In addition, %PMETs were correlated with age, gender, BMI, heart rate, smoking, hypertension, diabetes, E/e' rest and LA strain total (multiple r2=0.41) (table 2, model 3). To avoid collinearity, separate models were analysed for the LA peak positive strain and LA peak negative strain. The multiple r2 were modest for both models (0.39 for the LA peak positive strain and 0.38 for the LA peak negative strain). The best predictor of %PMETs was LA strain total. LA strain total <28 predicted reduced function with 61% sensitivity and 61% specificity according to ROC curve analysis (AUC=0.66). Patients with abnormal LA function (LA strain total <28) had significantly higher stress E/e' (12.7±5.2 vs 9.4±2.8, p<0.001) and E/e' rest (11.7±4.7 vs 8.2±2.5, p<0.001) than with normal LA function. The interobserver and intraobserver variability, expressed as a mean per cent error of LA strain total were 9.6±7.8% and 5.5±4.3%, respectively.

Relationship between stress E/e' and echocardiographic parameters

Patients in the stress E/e'>13 group had significantly lower LA strain profiles and larger LA volume than patients in the stress E/e' ≤13 group (table 1). There were no differences in resting LVEF and pulmonary vein flow profiles between the stress E/e' ≤13 group and the stress E/e' >13 group. There was a negative linear relationship between total LA strain and stress E/e' (r=−0.51, p<0.001). After adjustment for resting E/e', the correlation between total LA strain and stress E/e' was weak (partial r=−0.22). The stress E/e' also correlated with age, gender, diabetes mellitus, LA volume index and rest E/e' (table 2, model 4) (multiple r2=0.57). The best predictor of stress E/e' was rest E/e' and the next best was total LA strain.

A total LA strain <26.9 predicted a raised E/e' response with 72% sensitivity and 72% specificity according to ROC analysis (AUC 0.78). The AUC for the total LA strain to predict stress E/e' >13 was larger than that for the LA volume index (AUC 0.71, p<0.005) and LA peak negative strain (AUC 0.64, p<0.001). The AUC for the total strain was similar to that for the LA peak positive strain (AUC 0.76, p=0.38).

Discussion

The results of this study of a large consecutive population undergoing exercise echocardiography with preserved EF, free of valvular heart disease or exercise-induced ischaemia or known CAD history, show that exercise capacity was correlated with LA function. Resting LA function was shown to be the strongest echocardiographic correlate of exercise tolerance after accounting for confounding by age and gender. The results also confirm that increased E/e' with exercise (E/e' stress) was associated with a decrement in exercise capacity.18 In contrast, exercise capacity was not associated with variations of LV mass and EF within the range of normal values.

Exercise capacity

Exercise capacity is an important prognostic marker in most cardiovascular diseases. Widely known correlates were confirmed in this study, including older age, female gender and higher body mass index. Although women have a lower exercise capacity than men, there is a greater age-associated decline in exercise capacity in men. These results show that LA function is an important correlate of exercise capacity, independent of haemodynamic responses to stress.

In a previous study of patients without LV systolic dysfunction and significant valvular disease, LV diastolic dysfunction and elevated LVE/e' were strongly associated with exertional dyspnoea, thus limiting exercise capacity.19 LA volumes could be perceived as a marker of chronic elevation of LA pressure and, hence, indirectly of diastolic dysfunction. Indeed, increased LA volume has been identified as a predictor of cardiovascular disease, risk for stroke, atrial fibrillation and death.20 Furthermore, some studies have suggested that LA volume is an independent predictor of exercise capacity with normal LV systolic function.1 However, in this study population, there were no significant differences in LA volume index between patients with preserved and reduced exercise capacity.

LA function is reduced in proportion to the severity of hypertension, diabetes and ischaemic heart disease.21 ,22 In this study, LA reservoir function was the best predictor of decreased exercise capacity even after accounting for confounding by age and gender. The findings are concordant with previous work showing the association of reduced atrial functional reserve to handgrip in patients with heart failure and preserved EF in comparison with hypertensive controls of a comparable age.23 LA dysfunction has also been reported in patients with hypertension or diabetes with normal LA size,21 and reduced LA function is an independent predictor of mortality and provides prognostic value incremental to that of LA size in patients with non-ST-segment elevation myocardial infarction.24 Recently, LA function has also been shown to be an independent predictor of heart failure hospitalisation.25 In our population, there was a correlation between age and LA strain (R=0.22, p<0.01), consistent with previous work linking decreased LA strain with advanced age,26 possibly mediated by interstitial fibrosis.

In this study, as in previous investigations, resting LV filling pressure (E/e' rest) and increased E/e' with exercise (E/e' stress) were also associated with a decreased exercise capacity. However, we found resting LA function to be the strongest echocardiographic correlate of exercise tolerance. These results build upon a recent study,5 which showed significant correlation between atrial function (assessed by exercise a (Am)) with exercise capacity and exercise E/e (E/Em). In contrast with the case-mix of that study (50 HFNEF, 15 hypertensive and 30 healthy control subjects), in this work, nearly three-quarters of the patients had hypertension and diabetes, but only half the patients had diastolic dysfunction.

Diastolic exercise stress echocardiography

Exercise testing has been widely used to identify cardiac pathology as a cause of exertional dyspnoea, most notably cardiac ischaemia. The additional performance of mitral filling and tissue Doppler imaging of the mitral annulus at rest and post-stress has created the ‘diastolic stress test’. An exercise E/e' >13 predicts elevated LVE/e' both at rest and with exercise.27 While there are limitations to the use of E/e' as a surrogate marker of LVE/e',28 these were not pertinent to this study. The association of elevated LVE/e' with normal LVEF both at rest and with exercise has been associated with several risk factors and carries important prognostic implications. A previous study has demonstrated higher mortality rates and increased morbidity associated with more frequent hospitalisations with elevated LVE/e'.29 The results of this work are consistent with findings of previous work18 linking elevated LVE/e' with exercise with reduced exercise capacity. The cause of increased E/e' with exercise is not well characterised. The simplest explanation is that either diastolic relaxation abnormality or reduced LV compliance is responsible. However, atrial contraction may be an important means of optimising LV performance, even in the setting of normal systolic function.30

Limitations

This study included a relatively homogeneous and selected population of patients with preserved LVEF, and excluded ischaemic heart disease. In particular, we were unable to confirm or refute previous observations about ethnicity-related differences in LV function31 because the majority (85%) of patients were white. We used calculated METs, rather than oxygen consumption, as a measure of exercise tolerance. The MET is a widely accepted clinical tool for determining functional capacity and has been shown to be the most powerful predictor of long-term mortality.32 Despite the absence of reduced LVEF, we cannot exclude subclinical systolic dysfunction, which may have affected our results. Finally, although speckle strain has been widely used for LA evaluation and appears to predict cardiovascular outcomes,33 no dedicated software has been validated for the assessment of LA strain.

Conclusions

In this large population, the association of reduced LA function (especially LA reservoir function) with impaired exercise capacity was similar to that of elevated LVE/e', emphasising the role of ventriculo-atrial coupling to cardiac dysfunction with preserved LVEF. As exercise capacity is a strong independent predictor of death and cardiac events among asymptomatic patients,34 the role of LA function in predicting the clinical outcome of these patients warrants further study.

References

View Abstract

Footnotes

  • See Editorial, p 1261

  • Funding Dr Kusunose was supported by the Uehara Memorial Foundation postdoctoral fellowship and the Naito Foundation for research abroad.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval The study protocol was approved by the Cleveland Clinic Institutional Review Board.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles