OUP user menu

The role of exercise echocardiography in the diagnostics of heart failure with normal left ventricular ejection fraction

Jaroslav Meluzín, Jan Sitar, Jan Křístek, Robert Prosecký, Martin Pešl, Helena Podroužková, Vladimír Soška, Roman Panovský, Ladislav Dušek
DOI: http://dx.doi.org/10.1093/ejechocard/jer082 591-602 First published online: 17 June 2011

Abstract

Aims Few data are available on the exercise-induced abnormalities of myocardial function in patients with exertional dyspnoea and normal left ventricular ejection fraction (LV EF). The main aims of this study were to determine the prevalence of isolated exercise-induced heart failure with normal ejection fraction (HFNEF) and to assess whether disturbances in LV or right ventricular longitudinal systolic function are associated with the diagnosis of HFNEF.

Methods and results Eighty-four patients with exertional dyspnoea and normal LV EF and 14 healthy controls underwent spirometry, NT-proBNP plasma analysis, and exercise echocardiography. Doppler LV inflow and tissue mitral and tricuspid annular velocities were analysed at rest and immediately after the termination of exercise. Of the 30 patients with the evidence of HFNEF, 6 (20%) patients had only isolated exercise-induced HFNEF. When compared with the remaining patients, those with HFNEF had a significantly lower resting and exercise peak mitral annular systolic velocity (Sa) and the mitral annular velocity during atrial contraction, lower exercise peak mitral annular velocity at early diastole, and lower exercise peak systolic velocity of tricuspid annular motion. The multivariate logistic regression analysis including both parameters standardly defining HFNEF and the new Doppler variables potentially associated with the diagnosis of HFNEF revealed that NT-proBNP, LV mass index, left atrial volume index, and Sa significantly and independently predict the diagnosis of HFNEF.

Conclusion A significant proportion of patients require exercise to diagnose HFNEF. Sa appears to be a significant independent predictor of HFNEF, which may increase the diagnostic value of models utilizing the variables recommended by the European Society of Cardiology guidelines.

  • Exercise echocardiography
  • Diastolic heart failure

Introduction

Many patients encountered in everyday cardiological clinical practice suffer from exertional dyspnoea of unexplained aetiology. Following exclusion of non-cardiac causes of exertional dyspnoea, cardiologists frequently face the situation that resting non-invasive cardiac examinations including echocardiography, magnetic resonance imaging, and other demonstrate normal heart morphology and function. In such a situation, exercise may be helpful in identifying the reason for symptoms. The most likely explanation for exertional dyspnoea is exercise-induced systolic, diastolic, or combined myocardial dysfunction frequently resulting in heart failure symptomatology. Although exercise-induced systolic dysfunction due to myocardial ischaemia is commonly diagnosed by several non-invasive methods, the non-invasive approach to diagnose exercise-induced diastolic heart failure (i.e. heart failure with normal left ventricular ejection fraction, HFNEF) is not very well clinically validated and utilized. At present, echocardiography represents the most widely utilized tool for the assessment of diastolic function and for the non-invasive diagnosis of HFNEF.1 The basic step in the diagnostics of HFNEF is the evidence of an increase in left ventricular filling pressure (LVFP) as indicated by an increase in the ratio of early LV inflow to early diastolic mitral annular velocity. At present, the diagnostics of HFNEF relies on resting examinations.1 However, Burgess et al.2 demonstrated in an invasive part of their study that approximately one-quarter of the patients manifested an elevated LVFP only during exercise. It means that exercise echocardiography focusing on the evaluation of diastolic function may be the basic step for the diagnosis of HFNEF manifested only during exercise. Recently, several authors suggested that the pathophysiology of HFNEF is a complex process involving not only worsening of relaxation and an increase in myocardial stiffness accounting for the increase in LVFP,38 but also abnormalities in longitudinal systolic function and apical rotation.36,9,10 In addition, HFNEF is frequently associated with pulmonary hypertension,11,12 which is known to worsen right ventricular (RV) systolic and diastolic function. Indeed, the prevalence of resting RV systolic dysfunction in patients with HFNEF was found to range from 33 to 50%.13 However, few data are available on the exercise-induced disturbances of various components of myocardial function in patients with HFNEF and on the relationship of longitudinal systolic functional abnormalities to exercise tolerance in patients with exertional dyspnoea having the normal LV EF. Thus, the purpose of this study was (i) to assess the relationship of LV and RV systolic and diastolic longitudinal function to the exercise tolerance in patients with unexplained exertional dyspnoea and preserved LV EF, (ii) to determine the prevalence of exercise-induced increase in LVFP indicative of exercise-induced HFNEF, and (iii) to determine whether the parameters of LV or RV longitudinal systolic function can contribute to the diagnostics of HFNEF.

Methods

Study population

Patients with chronic exertional dyspnoea were selected from a cohort of patients who were referred to the 1st Department of Internal Medicine or to the Radiology Department at St Anna Hospital from October 2008 to December 2010 for elective invasive or CT coronary angiography to exclude coronary artery disease (CAD). They suffered from no or atypical chest discomfort, but the limiting symptom was exertional dyspnoea. Patients with the absence of coronary atherosclerosis or with only a mild insignificant coronary artery stenosis (luminal diameter narrowing <40%), with no history of myocardial infarction, no valvular or congenital heart disease, cardiomyopathy, liver, renal, or lung disease were screened. The control group consisted of 10 patients with atypical chest discomfort but without exertional dyspnoea, in whom invasive (n = 9) or CT (n = 1) coronary angiography excluded CAD, and of four asymptomatic healthy volunteers. None of the controls suffered from any cardiovascular disease or a disease predisposing to diastolic dysfunction (hypertension, diabetes mellitus, hyperlipoproteinaemia). Initially, all patients and controls underwent laboratory examination, spirometry, and baseline echocardiography at rest. Only those clinically stable, without renal, liver, and respiratory insufficiency or anaemia entered this study. Echocardiography had to prove normal LV EF (≥50%) and to exclude any wall motion abnormality, previously unrecognized valvular or congenital heart disease, or cardiomyopathy. All patients and controls gave their written consent to the investigations. The study complies with the Declaration of Helsinki and was approved by the Ethics Committee at St Anna Hospital.

Invasive catheter and CT coronary angiographies

Invasive catheter and multidetector 64-slice CT coronary angiographies were performed using standard techniques on ALLURA XPER FD 10 (Philips, Best, The Netherlands) and GE Light Speed VCT (GE, Milwaukee, WI, USA), respectively. The coronary lesions were analysed in multiple projections and expressed as percent of luminal diameter narrowing.

Laboratory examination and spirometry

N-terminal pro-B-type natriuretic peptide (NT-proBNP) plasma levels were measured at rest before exercise by an immunoassay method with the Elecsys 2010 analyzer (Roche, Basel, Switzerland). The maximal intra-assay variation was 4.2%. To exclude anaemia, renal or liver insufficiency, the complete blood count, serum creatinine, alanine aminotransferase, aspartate aminotransferase, bilirubin, and alkaline phosphatase were assayed.

The parameters evaluated on spirometry included forced expiratory volume in 1 s (FEV1), forced vital capacity (FVC), and their ratio (FEV1%). Forced expiratory volume in 1 s and FVC were expressed as percentages of predicted normal values. Values <70% were considered abnormal and indicative of lung disease.

Echocardiography at rest and immediately after exercise

Echocardiographic examinations were performed using Vivid 7 (the first 39 subjects) and Vivid E9 (GE, Milwaukee, WI, USA) with M3S and M5S transducers, respectively. Gray scale 2D images from apical four- and two-chamber views and parasternal short-axis views at the base, at the level of papillary muscles, and at the apex were recorded. Transmitral, aortic, and pulmonary venous flows were recorded using pulsed Doppler echocardiography. The sample volumes with a fixed length of 5.0 or 5.1 (Vivid 7) mm were placed between the tips of the mitral and aortic leaflets in the apical four- and five-chamber views, and in the right paraseptal pulmonary vein in the apical four-chamber view. In patients with tricuspid regurgitation, the transtricuspid systolic pressure gradient was determined by continuous Doppler and the right atrial pressure was visually estimated from the filling of jugular veins. Guided by colour-coded 2D echocardiography, pulsed-wave Doppler tissue imaging (DTI) of mitral and tricuspid annular motion was recorded at three annular sites adjacent to the septum, lateral LV wall, and free RV wall in the apical four-chamber view. The narrow angle sector (30–40°) was used to obtain DTI of individual walls at the high frame rates >150 frames/s. Tissue Doppler sample volumes of 5.9 (Vivid 7) or 6.0 mm were used. All Doppler recordings were done during shallow respiration or end-expiratory apnea. Three to ten consecutive cardiac cycles were digitally stored for offline analysis using EchoPAC PC versions 6.1.0. and 108.1.5 (GE Medical Systems, Horten, Norway).

Immediately after resting echocardiography, symptom-limited exercise was performed by bicycle ergometry (Kettler X7, Siemens, Germany) in a sitting position. The initial workload of 25 W was increased by 25 W every 2 min until the limited symptoms appeared (dyspnoea, leg, or general fatigue). Immediately after exercise, in the patient's position corresponding to the pre-exercise examination at rest, the following images were obtained: transmitral pulsed Doppler filling flow and Doppler tissue recordings of septal and lateral mitral annular motion and of tricuspid annular motion. The succession of post-exercise images was kept constant and their acquisition did not exceed the period of 90 s after the end of exercise. The duration of the exercise was measured and the maximal exercise tolerance was expressed as the number of metabolic equivalents (METs) according to the recommendation of the American College of Sports Medicine.14 For the METs calculation, the maximal workload exerted for at least 1 min was used. Severely limited exercise tolerance was defined by the presence of METs <6.

Echocardiographic data

Echocardiographic measurements were performed according to the recommendations of the American Society of Echocardiography and the European Association of Echocardiography.15 Left ventricular mass was estimated according to the Devereux formula.16 Left ventricular and left atrial volumes were calculated with the method of discs using apical four- and two-chamber views. Left ventricular ejection fraction was determined as (EDV − ESV)/EDV, where EDV, end-diastolic volume and ESV, end-systolic volume. End-diastolic volume, ESV, LV mass, and left atrial volume were indexed to body surface area. From conventional pulsed-wave Doppler recordings of mitral valve inflow, the following parameters were assessed: peak early (E) and late (A) diastolic transmitral flow velocities, early filling deceleration time (DT), and A-wave duration (Ad). From pulmonary venous flow recording, the duration of atrial reversal flow (ARd) was determined and the difference between ARd and Ad (ARd – Ad) was calculated. Pulmonary artery systolic pressure (PASP) was calculated from tricuspid regurgitant flow velocity (V): PASP = 4V2+ 5 mmHg at a normal jugular vein filling. From DTI, peak systolic (Sa-tric), early (Ea-tric), and late (Aa-tric) tricuspid annular velocities were measured. To determine peak systolic (Sa), early (Ea), and late (Aa) mitral annular velocities, the values obtained at septal and lateral sides of the mitral annulus were averaged. E, A, DT, and tissue Doppler velocities were analysed both at rest and immediately after the end of exercise (exe). The data are presented as a mean of 3–6 consecutive heart cycles in patients with sinus rhythm or as a mean of 4–10 heart cycles in patients with atrial fibrillation. All echocardiographic measurements were performed by one experienced observer (J.M.). For the determination of intra- and inter-observer variabilities, the resting and exercise Doppler parameters of 10 randomly selected patients were reassessed 2 months after the initial analysis by 2 observers (J.M. and H.P.). The data are presented as the mean absolute difference of repeated measurements and expressed as percentage of the mean of two absolute measurements.

Definition of heart failure with normal left ventricular ejection fraction

In our study, the presence of HFNEF was defined according to the algorithm published by experts of the Heart Failure and Echocardiography Associations of the European Society of Cardiology (ESC),1 which relies on functional parameters analysed at rest. The diagnosis of HFNEF required the following criteria to be fulfilled: (i) signs or symptoms of heart failure, (ii) normal or mildly abnormal systolic LV function defined by LV EF >50% and LV end-diastolic volume index (EDVI) < 97 mL/m2, and (iii) presence of diastolic LV dysfunction. Non-invasive evidence of diastolic LV dysfunction relied on the conventional and tissue Doppler echocardiography and plasma levels of NT-proBNP assessed under resting conditions. The presence of diastolic dysfunction was defined by any of the following three findings: (i) E/Ea >15, (ii) by the combination of borderline E/Ea (15 > E/Ea > 8) and the presence of at least one of the following pathologies: plasma NT-proBNP > 220 pg/mL, E/A < 0.5, and DT > 280 ms in patients over 50 years of age, Ard – Ad > 30 ms, left atrial volume index (LAVI) >40 mL/m2, LV mass index >122 g/m2 in females or >149 g/m2 in males, or the presence of atrial fibrillation, and (iii) by the combination of NT-proBNP > 220 pg/mL and the presence of at least one of the following pathologies: E/Ea >8, E/A <0.5, and DT >280 ms in patients over 50 years of age, Ard – Ad > 30 ms, LAVI >40 mL/m2, LV mass index >122 g/m2 in females or >149 g/m2 in males, or the presence of atrial fibrillation. Exercise-induced HFNEF was defined by the presence of exercise-induced dyspnoea and an increase in E/Ea >15 or E/Ea-sept (Ea measured at the septal annular corner) >13.2 Based on these definitions, patients with exertional dyspnoea were divided into patients without the evidence of HFNEF (group AB) and those with the evidence of HFNEF (group AA) determined under the resting conditions and/or during exercise.

Statistical analysis

Standard measures of summary statistics were used to describe the primary data: relative and absolute frequencies, arithmetic mean supplied with standard error of the mean. A one-way ANOVA model was used to compare the experimental groups in continuous variables and a standard t-test was applied for detailed mutual comparisons of the groups. An ML-χ2 test was applied to compare the experimental variants in categorical variables. All examined continuous parameters revealed normal sample distribution with homogeneous variance in compared groups (Shapiro-Wilk's test, Levene's test), except for NT-proBNP with asymmetric distribution of log-normal type. The NT-proBNP values were therefore logarithmically transformed prior to any parametric analysis. The diagnostic power of potential predictors was assessed on the basis of receiver operating characteristics (ROC) curves. The ROC analysis was performed using a ROC web calculator (Eng, 2006) for curve fitting, SPSS 17.02 (SPSS, Inc., 2009) for the area under the curve (AUC) computation and testing, and MedCalc 11.1.0.0 (MedCalc Software 1993–2009) for computation of sensitivity and specificity. The computation was based on binormal assumption. Both univariate and multivariate logistic regression strategies were applied to quantify the association of examined predictive factors and binary coded risk endpoints (severely limited exercise tolerance, HFNEF). Odds ratios (ORs) with 95% confidence limits (CIs) were estimated and tested in a Wald χ2 test. The parameters with the potential risk power (providing at least P < 0.10 in univariate logistic regression) were then examined for mutual correlation and the interaction terms were coded and tested for significantly correlated pairs of variables. If used in the models, effective cut-off values of continuous variables were optimized on the basis of ROC analysis. Construction of multivariate models for the presence of HFNEF as dependent variable adopted external knowledge, because some of the independent variables served also as factors conditionally defining HFNEF. Therefore, we used the set of defining parameters as initial entry to the model and then examined the added value of the other examined parameters over this baseline. The second approach employed values of potential predictors before and after physical exercise and examined their predictive power. In all the strategies used, the final set of potential predictive factors and interaction terms was subjected to objective stepwise selection algorithm in multivariate logistic regression (driven by the maximum likelihood ratio test). In all tests, P–values <0.05 were considered statistically significant.

Results

Ninety-eight patients with exertional dyspnoea and 16 controls were initially screened. Of them, nine patients with dyspnoea were excluded because of the evidence of lung disease on spirometry, one patient for either a significant tricuspid regurgitation, or aortic regurgitation, anaemia, poor quality of echocardiography, and LV EF <50%, respectively. Of the controls screened, one patient was excluded because of LV hypertrophy and one for frequent atrial extrasystoles. Thus, the final cohort analysed consisted of 84 patients with a history of exertional dyspnoea (group A) and of 14 healthy controls (group B). All group A patients developed dyspnoea during exertion, mostly combined with a general or leg fatigue and were found to have a decrease in exercise tolerance (METs <8). Based on the result of bicycle ergometry, group A patients were divided into those with a severely limited exercise tolerance (METS <6, group Al, n = 51) and those with a mildly-to-moderately limited exercise tolerance (METs ≥6, group A2, n = 33). The baseline characteristics of group A1, A2, and B patients are demonstrated in Table 1. Of patients with exertional dyspnoea, those with a severely limited exercise tolerance were most frequently older women with the elevation of NT-proBNP. Seventy-eight patients were in sinus rhythm, the remaining six patients were found to have atrial fibrillation. Pulmonary artery hypertension (pulmonary artery systolic pressure ≥35 mmHg) was found in four patients, all of whom had the evidence of HFNEF.

View this table:
Table 1

Baseline clinical, ergometric, and echocardiographic data in patients with exertional dyspnoea and healthy controls

Group A (n= 84)Group A1 (n= 51)Group A2 (n= 33)Group B (n= 14)P-value
Demographics
 Age (years)65.1 (1.0)67.6 (1.2)a61.3 (1.6)b61.8 (1.7)b0.002
 Males (%)25 (29.8)8 (15.7)a17 (51.5)b7 (50.0)b0.001
 Hypertension (%)74 (88.1)45 (88.2)29 (87.9)0.961
 Diabetes (%)24 (28.6)18 (35.3)6 (18.2)0.089
 Hyperlipidaemia (%)45 (53.6)30 (58.8)15 (45.5)0.230
 Obesity (%)38 (45.2)26 (51.0)12 (36.4)3 (21.4)0.095
 BMI (kg/m2)29.5 (0.5)30.4 (0.7)a28.0 (0.6)b26.9 (0.9)b0.010
Medication
 Diuretics (%)42 (50.0)28 (54.9)14 (42.4)0.264
 Beta-blockers (%)50 (59.5)33 (64.7)17 (51.5)0.229
 Hypolipidaemics (%)60 (71.4)40 (78.4)20 (60.6)0.987
 ACE-I/AT II (%)51 (60.7)31 (60.8)20 (60.6)0.077
Blood analysis
 NT-proBNP (ng/L)146.9167.7a119.7b78.5c0.024
Ergometry
 Exercise duration (s)461.0 (15.0)394.6 (15.9)a563.6 (18.2)b644.0 (37.3)c<0.001
 Exercise tolerance (METS)5.7 (0.1)5.0 (0.1)a6.7 (0.1)b7.5 (0.2)c<0.001
Echocardiography
 EDVI (mL/m2)44.5 (1.1)44.1 (1.5)45.1 (1.5)44.2 (2.1)0.888
 ESVI (mL/m2)15.9 (0.5)16.1 (0.7)15.5 (0.7)14.7 (1.0)0.612
 LV EF (%)64.6 (0.6)63.8 (0.8)65.7 (1.1)66.8 (1.2)0.137
 LV mass index (g/m2)91.0 (2.6)93.8 (3.7)86.5 (3.2)84.1 (4.6)0.222
 LAVI (mL/m2)29.6 (1.0)29.9 (1.3)29.2 (1.5)28.3 (1.6)0.822
 Ard – Ad (ms)−12.0 (2.7)−10.5 (3.5)−14.6 (4.3)−14.7 (10.8)0.765
  • ACE-I, angiotensin-converting enzyme inhibitor; Ad, duration of late transmitral filling wave; Ard, duration of atrial reversal flow; AT II, angiotensin II receptor blocker; BMI, body mass index; EDVI, end-diastolic volume index; EF, ejection fraction; ESVI, end-systolic volume index; group A, all patients with dyspnoea; group A1, patients with dyspnoea and severely limited exercise tolerance (METs <6); group A2, patients with dyspnoea and mildly-to-moderately limited exercise tolerance (METs ≥6); group B, healthy controls; LAVI, left atrial volume index; LV, left ventricular; METs, metabolic equivalents; NT-proBNP, N-terminal pro-B-type natriuretic peptide. Data are presented as mean ± SEM or number (%). NT-proBNP is described by geometric mean due to highly skewed distribution.

  • P-value: statistical test comparing groups A1, A2, and B (ML-χ2 test for categories, one-way ANOVA test for continuous variables).

  • a–cMarks of mutual statistically significant differences between the compared variants A1, A2, B (ANOVA post hoc test; P< 0.05).

Haemodynamic and Doppler results at rest and immediately after the end of exercise

Tables 2 and 3 show haemodynamic and Doppler data at rest and immediately after the end of exercise. In all patients, the angle between the Doppler ultrasound beam and the direction of blood flow or annular corner motion was <25 degrees. At rest, group A1 and A2 patients differed only in Sa and Ea, both being significantly decreased in patients with METs <6. At peak exercise, group A1 patients exhibited lower heart rate and systolic blood pressure. Of the Doppler variables, Sa-exe, Ea-exe, and Aa-tric-exe were lower, whereas the E-exe/Ea-exe was higher in group A1 when compared with group A2. Complete or nearly complete fusion of early and late Doppler filling or tissue velocities, the presence of atrial fibrillation, or the worse quality of exercise echocardiography did not allow to assess resting and exercise A, DT, Aa, Sa-tric, Ea-tric, and Aa-tric in 6, 4, 6, 3, 3, 11, and in 6, 17, 6, 3, 3, 12 subjects, respectively.

View this table:
Table 2

Haemodynamic and Doppler results at rest

Group A (n = 84)Group A1 (n= 51)Group A2 (n = 33)Group B (n= 14)P-value
Haemodynamics
 Heart rate (bpm)74.5 (1.3)73.9 (1.6)75.5 (2.3)78.7 (3.1)0.416
 Systolic BP (mmHg)134.1 (1.9)136.0 (2.7)a131.2 (2.5)a,b123.2 (3.1)b0.041
 Diastolic BP (mmHg)84.4 (1.0)84.4 (1.4)84.4 (1.5)78.6 (1.9)0.095
LV Doppler results
 E (cm/s)75.2 (1.9)74.2 (2.3)76.8 (3.1)69.9 (2.0)0.408
 A (cm/s)81.5 (1.6)82.1 (2.0)a80.6 (2.9)a69.4 (2.6)b0.014
 E/A0.92 (0.03)0.90 (0.03)0.95 (0.05)1.01 (0.03)0.222
 DT (ms)185.0 (4.1)185.5 (5.6)a184.1 (5.7)a159.3 (4.4)b0.042
 Sa (cm/s)8.2 (0.2)7.8 (0.2)a8.8 (0.3)b9.0 (0.3)b0.003
 Ea (cm/s)8.3 (0.2)8.0 (0.3)a8.8 (0.3)b9.8 (0.3)c0.009
 Aa (cm/s)10.3 (0.2)10.0 (0.3)10.7 (0.4)10.8 (0.4)0.278
 Ea/Aa0.82 (0.03)0.80 (0.03)0.85 (0.05)0.92 (0.05)0.171
 E/Ea9.3 (0.3)9.5 (0.3)a8.9 (0.4)a7.3 (0.4)b0.005
RV Doppler results
 Sa-tric (cm/s)13.5 (0.3)13.4 (0.5)13.8 (0.4)15.2 (0.6)0.157
 Ea-tric (cm/s)12.3 (0.4)11.9 (0.5)12.8 (0.7)14.0 (0.9)0.147
 Aa-tric (cm/s)15.7 (0.4)15.7 (0.6)15.7 (0.6)16.1 (1.1)0.925
 Ea-tric/Aa-tric0.8 (0.0)0.8 (0.0)0.8 (0.0)0.9 (0.0)0.167
  • A, peak late transmitral flow velocity; Aa, peak late diastolic annular velocity; BP, blood pressure; DT, deceleration time of E-wave; E, peak early transmitral flow velocity; Ea, peak early diastolic annular velocity; RV, right ventricular; Sa, peak systolic annular velocity; tric, measured at the tricuspid annulus adjacent to the free RV wall. Other abbreviations as in Table 1. Data are presented as mean ± SEM.

  • P-value: statistical test comparing groups A1, A2, and B (ML-χ2 test for categories, one-way ANOVA test for continuous variables).

  • a–cMarks of mutual statistically significant differences between the compared variants A1, A2, B (ANOVA post hoc test, P< 0.05).

View this table:
Table 3

Haemodynamic and Doppler results immediately after the end of exercise

Group A (n= 84)Group A1 (n= 51)Group A2 (n= 33)Group B (n = 14)P-value
Haemodynamics
 Heart rate-exe (bpm)120.0 (2.0)113.8 (2.4)a129.5 (2.9)b132.9 (3.4)b<0.001
 Systolic BP-exe (mmHg)180.1 (2.9)174.8 (3.8)a188.0 (4.1)b185.0 (4.7)a,b0.049
 Diastolic BP-exe (mmHg)91.8 (1.3)90.9 (1.7)93.2 (1.8)87.1 (1.6)0.233
LV Doppler results
 E-exe (cm/s)99.1 (2.3)100.3 (3.2)97.4 (3.0)89.7 (3.9)0.218
 A-exe (cm/s)98.7 (2.1)97.6 (2.6)100.3 (3.6)91.0 (3.2)0.275
 E-exe/A-exe1.01 (0.03)1.03 (0.04)0.98 (0.05)0.99 (0.04)0.632
 DT-exe (ms)144.0 (3.5)143.7 (4.5)144.6 (5.3)140.2 (4.1)0.899
 Sa-exe (cm/s)11.0 (0.3)10.1 (0.3)a12.5 (0.4)b13.6 (0.6)b<0.001
 Ea-exe (cm/s)11.1 (0.2)10.5 (0.3)a12.0 (0.3)b12.4 (0.4)b0.001
 Aa-exe (cm/s)11.9 (0.3)11.3 (0.4)a12.9 (0.5)a,b13.8 (0.4)b0.002
 Ea-exe/Aa-exe0.97 (0.03)0.97 (0.04)0.96 (0.04)0.91 (0.03)0.686
 E-exe/Ea-exe9.1 (0.3)9.7 (0.4)a8.2 (0.3)b7.3 (0.4)b0.001
RV Doppler results
 Sa-tric-exe (cm/s)17.4 (0.3)17.0 (0.4)a18.1 (0.4)a,b19.5 (0.5)b0.021
 Ea-tric-exe (cm/s)14.3 (0.4)14.0 (0.4)14.6 (0.8)15.4 (0.5)0.359
 Aa-tric-exe (cm/s)19.0 (0.5)18.1 (0.5)a20.5 (0.9)b19.3 (0.8)b0.044
 Ea-tric-exe/Aa-tric-exe0.76 (0.03)0.79 (0.03)0.70 (0.04)0.81 (0.03)0.131
  • Exe, exercise. Other abbreviations as in Table 2. Data are presented as mean ± SEM.

  • P-value: statistical test comparing groups A1, A2, and B (ML-χ2 test for categories, one-way ANOVA test for continuous variables).

  • a–cMarks of mutual statistically significant differences between the compared variants A1, A2, B (ANOVA post hoc test, P< 0.05).

Resting and exercise echocardiography and the diagnostics of heart failure with normal left ventricular ejection fraction

Of the 84 patients with exertional dyspnoea, 24 (29%) had the evidence of HFNEF based on the analysis of data at rest (echocardiography, plasma NT-proBNP, electrocardiography). Of them, resting echocardiography alone was able to reveal HFNEF in 13 (54%) patients (2 patients had E/Ea >15; 11 patients had E/Ea between 8 and 15 in a combination with either left atrial dilation or with LV hypertrophy or Ard – Ad >30 ms). In the remaining 11 patients (46%), the diagnosis of HFNEF had to be made by a combination of parameters derived from various methods (echocardiography, electrocardiography, and plasma NT-proBNP). Of the 24 patients with HFNEF determined under the resting conditions, 11 patients had the exercise-induced E/Ea >15 or E/Ea-sept >13 suggestive of elevated LVFP also during exercise. Six patients had the isolated, only exercise-induced HFNEF (exercise-induced E/Ea >15 or E/Ea-sept >13 with no evidence of HFNEF under the resting conditions). Thus, of all 30 patients with HFNEF, the combined resting echocardiography, electrocardiography, and NT-proBNP analysis allowed to diagnose 24 (80%) of the cases of HFNEF, whereas the remaining 6 (20%) patients relied exclusively on the exercise Doppler echocardiography.

Longitudinal left ventricular and right ventricular systolic function in patients with heart failure with normal left ventricular ejection fraction

Tables 4 and 5 show the baseline clinical, haemodynamic, and echocardiographic results in patients with exertional dyspnoea with and without HFNEF and in healthy controls. When compared with patients with dyspnoea but not with the evidence of HFNEF (group AB), patients with HFNEF (group AA) were older and were more frequently on beta-blockers. They had higher plasma NT-proBNP, lower exercise systolic blood pressure, and worse exercise tolerance. Of the echocardiographic parameters, group AA patients were found to have higher ESVI, LAVI, LV mass index, E/Ea, E-exe/Ea-exe and lower LV EF. These findings are expected and logical as, by definition, the pathology in these parameters is involved in the diagnostics of HFNEF according to the ESC guidelines. More interestingly, when compared with group AB patients, those with HFNEF had significantly lower Sa, Sa-exe, Ea-exe, Aa, Aa-exe, and Sa-tric-exe. When comparing Sa, Ea, Aa, E/Ea, and Sa-tric changes associated with the exercise (▵Sa = Sa-exe – Sa at rest, etc) between groups AA and AB, the differences in ▵Sa (2.2 ± 0.3 cm/s vs. 3.2 ± 0.2 cm/s, P = 0.001), ▵Ea (1.9 ± 0.4 cm/s vs. 3.2 ± 0.3 cm/s, P = 0.045), and in ▵(E/Ea) (0.6 ± 0.5 vs. −0.5 ± 0.2, P = 0.03) were significant.

View this table:
Table 4

Baseline clinical, haemodynamic, and NT-proBNP results in patients with exertional dyspnoea with or without the evidence of heart failure with normal left ventricular ejection fraction and in healthy controls

Group AA (n= 30)Group AB (n= 54)Group B (n= 14)P-value
Demographics
 Age (years)68.7 (1.7)a63.2 (1.2)b61.8 (1.7)b0.008
 Males (%)10 (33.3)15 (27.8)7 (50.0)0.301
 Hypertension (%)25 (83.3)49 (90.7)0.315
 Diabetes (%)8 (26.7)16 (29.6)0.773
 Hyperlipidaemia (%)17 (56.7)28 (51.9)0.672
 Obesity (%)17 (56.7)21 (38.9)3 (21.4)0.071
 BMI (kg/m2)30.3 (0.9)29.0 (0.6)26.9 (0.9)0.081
Medication
 Diuretics (%)18 (60.0)24 (44.4)0.172
 Beta-blockers (%)23 (76.7)a27 (50.0)b0.014
 Hypolipidaemics (%)21 (70.0)39 (72.2)0.829
 ACE-I/AT II (%)19 (63.3)32 (59.3)0.714
Blood analysis
 NT-proBNP (ng/L)326.8a94.2b78.5b<0.001
Haemodynamics at rest
 Heart rate (bpm)72.7 (2.1)75.5 (1.7)78.7 (3.1)0.289
 Systolic BP (mmHg)137.2 (3.7)a132.4 (2.1)a,b123.2 (3.1)b0.042
 Diastolic BP (mmHg)85.0 (2.0)84.1 (1.2)78.6 (1.9)0.086
Haemodynamics at exercise
 Heart rate-exe (bpm)113.3 (3.6)a123.7 (2.3)a,b132.9 (3.4)b0.002
 Systolic BP-exe (mmHg)170.0 (4.4)a185.8 (3.6)b185.0 (4.7)b0.017
 Diastolic BP-exe (mmHg)90.3 (2.2)92.6 (1.6)87.1 (1.6)0.234
  • Abbreviations as in Tables 13. Group AA, patients with exertional dyspnoea and the evidence of heart failure with normal left ventricular ejection fraction; group AB, patients with exertional dyspnoea without the evidence of heart failure with normal left ventricular ejection fraction; group B, healthy controls. Data are presented as mean ± SEM or number (%).

  • P-value: statistical test comparing groups AA, AB, and B (ML-χ2 test for categories, one-way ANOVA test for continuous variables).

  • a–cMarks of mutual statistically significant differences between the compared variants AA, AB, B (ANOVA post hoc test, P< 0.05).

View this table:
Table 5

Exercise tolerance and echocardiographic results in patients with exertional dyspnoea with or without the evidence of heart failure with normal left ventricular ejection fraction and in healthy controls

Group AA (n= 30)Group AB (n= 54)Group B (n= 14)P-value
Ergometry
 Exercise duration (s)421.7 (28.7)a482.8 (16.5)a644.0 (37.3)b<0.001
 Exercise tolerance (METs)5.3 (0.2)a5.9 (0.1)b7.5 (0.2)c<0.001
 No. of patients with METs <6 (%)25 (83.3)a26 (48.1)b0 (0.0)c<0.001
LV Doppler results
 Sa (cm/s)7.3 (0.2)a8.7 (0.2)b9.0 (0.3)b<0.001
 Sa-exe (cm/s)9.4 (0.3)a11.9 (0.3)b13.6 (0.6)c<0.001
 Ea (cm/s)7.5 (0.4)a8.8 (0.3)a,b9.8 (0.3)b0.001
 Ea-exe (cm/s)9.7 (0.4)a11.9 (0.3)b12.4 (0.4)b<0.001
 Aa (cm/s)9.2 (0.4)a10.7 (0.3)b10.8 (0.4)b0.007
 Aa-exe (cm/s)10.7 (0.6)a12.5 (0.3)b13.8 (0.4)b0.001
 E/Ea10.8 (0.4)a8.5 (0.3)b7.3 (0.4)b<0.001
 E-exe/Ea-exe11.3 (0.5)a8.0 (0.2)b7.3 (0.4)b<0.001
RV Doppler results
 Sa-tric (cm/s)13.0 (0.7)a13.8 (0.4)a,b15.2 (0.6)b0.017
 Sa-tric-exe (cm/s)16.5 (0.6)a18.0 (0.4)b19.5 (0.5)b0.006
 Ea-tric (cm/s)12.1 (0.8)12.3 (0.5)14.0 (0.9)0.271
 Ea-tric-exe (cm/s)14.7 (0.7)14.0 (0.4)15.4 (0.5)0.354
 Aa-tric (cm/s)15.3 (0.9)15.9 (0.5)16.1 (1.1)0.784
 Aa-tric-exe (cm/s)18.3 (0.7)19.3 (0.6)19.3 (0.8)0.586
Conventional echocardiography
 EDVI (mL/m2)46.7 (2.2)43.2 (1.1)44.2 (2.1)0.281
 ESVI (mL/m2)17.6 (1.1)a14.9 (0.5)b14.7 (1.0)b0.027
 LV EF (%)62.6 (1.0)a65.7 (0.8)b66.8 (1.2)b0.011
 LV mass index (g/m2)106.8 (5.1)a82.1 (2.1)b84.1 (4.6)b<0.001
 LAVI (mL/m2)35.1 (1.9)a26.6 (0.8)b28.3 (1.6)b<0.001
 Ard – Ad (ms)−7.9 (4.9)−14.1 (3.2)−14.7 (10.8)0.607
  • Abbreviations as in Tables 14. Data are presented as mean ± SEM. P-value: statistical test comparing groups AA, AB, and B (ML-χ2 test for categories, one-way ANOVA test for continuous variables).

  • a–cMarks of mutual statistically significant differences between the compared variants AA, AB, B (ANOVA post hoc test, P< 0.05).

Intra- and inter-observer variability

Mean absolute differences of intra-observer repeated measurements (J.M.) for Sa, Ea, Aa, E, and A were 0.15 cm/s (1.8%), 0.27 cm/s (2.8%), 0.13 cm/s (1.2%), 1.2 cm/s (1.6%), and 2.7 cm/s (3.4%), respectively. Mean absolute differences of inter-observer measurements for Sa, Ea, Aa, E, and A were 0.47 cm/s (5.7%), 0.43 cm/s (4.5%), 0.33 cm/s (3.1%), 2.4 cm/s (3.2%), and 3.6 cm/s (4.4%), respectively. Mean absolute differences of intra-observer repeated measurements (J.M.) for Sa-exe, Ea-exe, Aa-exe, E-exe, and A-exe were 0.41 cm/s (3.4%), 0.17 cm/s (1.4%), 0.30 cm/s (2.2%), 2.5 cm/s (2.6%), and 1.4 cm/s (1.4%), respectively. Mean absolute differences of inter-observer measurements for Sa-exe, Ea-exe, Aa-exe, E-exe, and A-exe were 0.39 cm/s (3.2%), 0.34 cm/s (2.7%), 0.20 cm/s (1.5%), 3.6 cm/s (3.8%), and 2.00 cm/s (2.0%), respectively.

Predictors of a severely limited exercise tolerance in patients with normal left ventricular ejection fraction

Of the pre-exercise resting clinical and echocardiographic variables (Tables 1 and 2), the univariate analysis identified the following potential predictors of a severely limited exercise tolerance (METs <6): age, gender, BMI, NT-proBNP, Sa, and Ea. Two alternative models demonstrating the association of potential risk parameters with severely limited exercise tolerance resulted from multivariate analysis (Table 6), both including age and therefore age-adjusted. Sa was proved as a significant predictor of the endpoint, even more functional in the continuous scale (as primary variable) than as coded according to the cutoff point. When coded in risk values Sa < 8.4 cm/s, it significantly interacted with the values Ea < 8 cm/s. The Sa and Ea optimal cutoff points were derived from the ROC analysis. Sa < 8.4 cm/s predicted METs <6 with a sensitivity of 68.6% and a specificity of 60.6% (AUC 0.67; 95% CI 0.56, 0.80; P = 0.009), Ea < 8 cm/s predicted METs <6 with a sensitivity of 56.9% and a specificity of 66.7% (AUC 0.62; 95% CI 0.50, 0.74; P = 0.043).

View this table:
Table 6

Association of potential risk factors with severely limited exercise tolerance (METs <6) as the risk endpoint in a multivariate logistic regression modela

Parameters included in modelCoefficient (SE; P level)Multivariate adjusted OR (95% CI)
Model including Sa as a continuous variable
 Parameters included in model
  Null model (intercept only)−2.003 (2.441)
  BMI (continuous)0.176 (0.069; P= 0.010)1.19 (1.04; 1.37)
  Age >65 years1.250 (0.538; P= 0.020)3.48 (1.19; 10.17)
  Sa (continuous)−0.394 (0.192; P= 0.039)0.67 (0.46; 0.96)
 Overall model parameters
  −2*log (likelihood) – null model/final model112.6/91.3
  χ2(df = 3); P-value21.2; P< 0.001
 Cross-validationb
  Correctly classified patients with METs <6/METs ≥680.4%/63.6%
Model including Sa as a binary coded variable
  Null model (intercept only)−6.109 (2.137)
  BMI (continuous)0.191 (0.070; P= 0.007)1.21 (1.06; 1.39)
  Age >65 years1.265 (0.539; P= 0.019)3.54 (1.21; 10.37)
  Sa < 8.4 cm/s and Ea < 8.0 cm/s1.251 (0.608; P = 0.040)3.49 (1.04; 11.72)
 Overall model parameters
  −2*log (likelihood) – null model/final model112.6/91.4
  χ2(df = 3); P-value21.1; P< 0.001
 Cross-validationb
  Correctly classified patients with METs <6/METs ≥684.3%/63.6%
  • aMultivariate logistic regression model based on stepwise procedures driven only by statistical measures (log-likelihood function); dependent variable Y, probability of METs <6.

  • bCross-validation performed retrospectively on calibration data set (n= 84).

  • Abbreviations as in Tables 1 and 2.

Potential contribution of longitudinal left ventricular and right ventricular systolic function to the diagnostics of heart failure with normal left ventricular ejection fraction

Of the parameters listed in Tables 4 and 5, the following parameters were proved to be the potential predictors of HFNEF in univariate analysis: age, NT-proBNP, systolic blood pressure, heart rate, Sa, Ea/Aa, E/Ea, Sa-tric-exe, ESVI, LV EF, LV mass index, and LAVI. Then, we constructed a model based on parameters defining HFNEF—a multivariate model selected the following multivariate adjusted predictors: NT-proBNP > 200 pg/mL, LV mass index > 100 g/m2, E/Ea > 8, and LAVI > 35 mL/m2 (Table 7). When the set of parameters defining HFNEF was enriched by the other significant potential predictors (Sa, Sa-tric, Aa, Ea), only Sa was selected as an independent predictor, which even slightly increased the predictive value of the models (Table 7). The other parameters were excluded due to mutual correlations and redundancy. Models including both values before and immediately after exercise were not possible due to a very strong pairwise correlation of these values. Although the parameters measured after the exercise revealed higher predictive values in ROC analysis in comparison with the values obtained before exercise (Sa, Ea, Aa, Sa-tric), the multivariate model based on exercise values did not reveal any significantly improved predictive value (data not presented). The ROC analysis of potential predictors of HFNEF including both resting and exercise parameters revealed the following best six predictors: E-exe/Ea-exe (AUC 0.88; 95% CI 0.79, 0.96; P < 0.001; optimal cutoff >9.5, sensitivity 86.2%, specificity 83.3%), NT-proBNP (AUC 0.83; 95% CI 0.73, 0.93; P < 0.001; optimal cutoff >200 pg/mL, sensitivity 76.7%, specificity 87.0%), Sa-exe (AUC 0.83; 95% CI 0.74, 0.93; P < 0.001; optimal cutoff <10 cm/s, sensitivity 83.3%, specificity 83.3%), Ea-exe (AUC 0.79; 95% CI 0.68, 0.90; P < 0.001; optimal cutoff <10 cm/s, sensitivity 60.0%, specificity 87.0%), E/Ea (AUC 0.79; 95% CI 0.69, 0.90; P < 0.001; optimal cutoff >8.0, sensitivity 82.8%, specificity 70.4%), Sa (AUC 0.79; 95% CI 0.69, 0.89; P < 0.001; optimal cutoff <8.0 cm/s, sensitivity 73.3%, specificity 72.2%). When analysing the change of Sa from rest to exercise (▵Sa), ▵Sa < 2.5 cm/s predicted HFNEF with a sensitivity of 63.3% and a specificity of 79.6% (AUC 0.72; 95% CI 0.60, 0.84, P = 0.001).

View this table:
Table 7

Association of potential risk factors with heart failure with normal left ventricular ejection fraction as the risk endpoint in a multivariate logistic regression modela

Parameters included in modelCoefficient (SE; P level)Multivariate adjusted OR (95% CI)
Model based on HFNEF defining parameters
 Null model (intercept only)−5.151 (1.165)
 NT-proBNP >200 ng/L4.046 (0.485; P< 0.001)57.19 (22.09; 147.91)
 LV mass index >100 g/m22.771 (0.982; P= 0.004)15.98 (2.26; 112.86)
 E/Ea > 82.379 (0.963; P= 0.013)10.80 (1.59; 73.39)
 LAVI >35 mL/m22.349 (0.969; P= 0.015)10.47 (1.52; 72.16)
Overall model parameters
 −2*log (likelihood) – null model/final model109.5/45.7
 χ2(df = 4); P-value63.8; P< 0.001
Cross-validationb
 Correctly classified patients with/without HFNEF93.3%/92.6%
Model based on HFNEF defining parameters and the other examined parameters
 Null model (intercept only)−4.446 (2.824)
 NT-proBNP >200 ng/L3.595 (0.674; P< 0.001)36.43 (9.72; 136.46)
 LV mass index >100 g/m23.096 (1.062; P= 0.004)22.12 (2.67; 183.03)
 LAVI >35 mL/m22.959 (1.053; P= 0.005)19.28 (2.36; 156.89)
 Sa (continuous)−1.008 (0.381; P= 0.009)0.36 (0.17; 0.78)
Overall model parameters
 −2*log (likelihood) – null model/final model109.5/42.9
 χ2(df = 4); P-value66.5; P< 0.001
Cross-validationb
 Correctly classified patients with/without HFNEF93.3%/98.2%
Model based on HFNEF defining parameters and Sa as binary coded parameter
 Null model (intercept only)−4.974 (1.129)
 NT-proBNP >200 ng/L4.032 (0.521; P< 0.001)56.32 (20.28; 156.32)
 LV mass index >100 g/m23.383 (1.113; P= 0.002)29.46 (3.21; 270.32)
 Sa < 8 cm/s and E/Ea > 82.753 (0.919; P= 0.003)15.71 (2.52; 97.93)
 LAVI >35 mL/m22.431 (0.985; P= 0.014)11.38 (1.60; 80.93)
Overall model parameters
 −2*log (likelihood) – null model/final model109.5/41.7
 χ2 (df = 4); P-value67.6; P< 0.001
Cross-validationb
 Correctly classified patients with/without HFNEF94.4%/92.6%
  • aMultivariate logistic regression model based on stepwise procedures driven only by statistical measures (log-likelihood function); dependent variable Y, probability of METs <6.

  • bCross-validation performed retrospectively on calibration data set (n= 84).

  • Abbreviations as in Tables 1 and 2.

Discussion

Our study brings several important findings. We demonstrated that even the combined non-invasive approach to the diagnostics of HFNEF utilizing echocardiography, electrocardiography, and NT-proBNP, recommended by the current ESC guidelines,1 does not reveal the significant proportion of patients with the evidence of exercise-induced elevation of LVFP. In our study, 7% of patients (6 of 84) with unexplained exertional dyspnoea required exercise to diagnose HFNEF. In addition, we demonstrated a significant independent association of LV longitudinal systolic dysfunction with the severely limited exercise tolerance and with the diagnosis of HFNEF. Sa < 8.4 cm/s along with Ea < 8.0 cm/s as the interaction term independently predicted the severely limited exercise tolerance (<6 METs). The Sa value as a continuous variable or Sa < 8.0 cm/s in interaction with E/Ea > 8 was more closely associated with the diagnosis of HFNEF than several echocardiographic parameters currently involved in the HFNEF diagnostics, according to the ESC recommendations.

Diagnostics of heart failure with normal left ventricular ejection fraction and the role of exercise

At present, HFNEF accounts for ∼30–50% of all heart failure patients1719 and is the reason for unexplained dyspnoea in many patients with preserved LV EF.12,20 The diagnosis of HFNEF relies on the assessment of patient's history, the finding of non-dilated left ventricle with a normal EF, and the evidence of diastolic dysfunction under resting conditions.1 The pivotal step in the determination of diastolic dysfunction is the evidence of an increase in LVFP.1 However, in some patients with a normal LV EF, LVFP at rest is normal and rises only under conditions of cardiovascular stress including exercise. In such patients the resting estimation of LVFP gives only incomplete information and stress-induced changes in LVFP must be analysed in order to diagnose the stress-induced HFNEF. Borlaug et al.12 performed diagnostic exercise catheterization in 55 patients with exertional dyspnoea, LV EF >50%, normal brain natriuretic peptide assay, and normal resting haemodynamics. Thirty-two (58%) patients were found to have exercise-induced HFNEF as indicated by an elevation of PCWP ≥25 mmHg during exercise. In a study of Burgess et al.2 combining exercise echocardiography and invasive LV diastolic pressure measurements, approximately one-quarter of the patients manifested an elevated LV diastolic pressure only during exercise. In the largest study published thus far, Holland et al.21 analysed rest and exercise E/Ea ratios reflecting LVFP in 493 patients with normal LV EF (>50%) referred for exercise echocardiography. Seventy-five patients had raised E/Ea with exercise, of whom 41 had normal E/Ea at rest suggesting an isolated, only exercise-induced elevation of LVFP (8% from the total 493 exercised patients). However, non-invasive studies performed thus far included unselected cohorts of patients with frequent exercise-induced wall motion abnormalities due to ischaemia that may have affected the behaviour of the exercise E/Ea. In addition, these studies relied on the echocardiographic E/Ea ratio omitting other variables contributing to the diagnosis of HFNEF. In our study, in order to establish the diagnosis of HFNEF, we used the complex non-invasive approach recommended by the ESC.1 It relies not only on the E/Ea ratio, but also on the other echocardiographic parameters, on the presence of atrial fibrillation, and on the plasma levels of NT-proBNP. We demonstrated the utility of exercise echocardiography even in the application of ESC guidelines for the diagnostics of HFNEF. Of the 30 patients with HFNEF, the combined resting echocardiography, electrocardiography, and NT-proBNP analysis allowed to diagnose 24 (80%) of the cases of HFNEF, whereas the remaining 6 (20%) patients relied exclusively on the exercise Doppler echocardiography. Thus, omitting the exercise, HFNEF can be undiagnosed in a significant proportion of patients with exertional dyspnoea.

Longitudinal left ventricular and right ventricular systolic function in patients with the diagnosis of heart failure with normal left ventricular ejection fraction

In patients with documented or suspected HFNEF, diastolic dysfunction characterized by altered relaxation and/or an increase in LVFP was repeatedly found to be associated with longitudinal LV systolic dysfunction represented by a decrease in the peak mitral annular systolic velocity3,5,6,9,10 or a reduction in longitudinal systolic strain and strain rate.10,22,23 The association of longitudinal systolic and diastolic dysfunction results from their close coupling demonstrated in several studies.3,2326 Increased longitudinal contraction is able to improve early diastolic LV filling through the effect of elastic recoil. On the other hand, more complete relaxation contributes to more vigorous contraction through the Frank–Starling mechanism. Both systolic and diastolic longitudinal functions are energy-dependent processes vulnerable to subendocardial ischaemia due to microcirculation disturbances typically localized in the subendocardium. The subendocardium is also primarily involved in the fibrotic process accompanying the development of hypertensive and diabetic cardiomyopathy or other diseases frequently associated with diastolic dysfunction.27,28 As the subendocardium contains longitudinally oriented myocardial fibres, the linkage of diastolic and systolic longitudinal dysfunction in early stages of heart failure is obvious. García et al.6 demonstrated that the peak systolic mitral annular velocity (Sa) belongs to the most useful parameters for identifying HFNEF, having even a higher predictive value than the peak early diastolic mitral annular velocity (Ea). Our results correspond to such a finding. The multivariate logistic regression model including the current HFNEF defining parameters and other variables potentially predicting HFNEF revealed the significant potential of Sa in predicting HFNEF, when used as either continuous variable or in association with E/Ea (combination of Sa < 8.0 cm/s and E/Ea > 8.0). Recently, Tan et al.4 showed that systolic and diastolic abnormalities found in HFNEF patients at rest are exaggerated during exercise and that some abnormalities may only occur under exercise. As demonstrated in Table 5, we also observed exaggerated differences in echocardiographic functional parameters (Sa-exe, Ea-exe, E-exe/Ea-exe) between groups AA and AB under the exercise. However, in a multivariate logistic regression model, exercise Doppler parameters did not provide any improvement of the predictive power reached for the diagnosis of HFNEF when compared with the corresponding values obtained at rest. Interestingly, Sa-exe was significantly lower in group AB when compared with controls, suggesting the contribution of myocardial dysfunction to exercise dyspnoea in these patients, even if they did not fulfill the current criteria for the diagnosis of HFNEF. Thus, exercise can likely reveal subtle functional abnormalities in patients with exertional dyspnoea without the evidence of HFNEF, probably representing a very early stage of heart failure.

In HFNEF patients, LV dysfunction is frequently associated with pulmonary hypertension,11,12 which may worsen RV function. Indeed, both systolic and diastolic RV dysfunction were frequently found in HFNEF patients.13 In early stages of HFNEF, pulmonary hypertension may only occur during the exercise.12 This fact is likely to explain our results of a significantly lower Sa-tric-exe in patients with HFNEF when compared with the remaining patients (Table 5). Thus, HFNEF is commonly associated with a complex of various functional abnormalities of a different clinical meaning and importance.

Components of myocardial function associated with a low exercise tolerance in patients with normal left ventricular ejection fraction

Many reports studying the relationship of various echocardiographic parameters to exercise capacity or tolerance included patients with a significant CAD,2931 myocardial ischaemia,2,29 resting wall motion abnormalities,29,30 or decreased LV EF.30,31 In these studies, A, Sa, Ea, and mainly E/Ea correlated best with the exercise capacity.2931 However, the results of these studies cannot be extrapolated to patients with HFNEF as mainly myocardial ischaemia or wall motion abnormalities may have accounted for significant correlations of many echocardiographic parameters mentioned with the exercise tolerance. The data on the relationship of various echocardiographic parameters to exercise tolerance in patients with non-ischaemic exertional dyspnoea having normal LV EF are rare. Ha et al.32 studied 45 such patients with diastolic stress echocardiography and demonstrated the relationship of E/Ea both at rest and during exercise to the exercise duration. In a subset of 113 patients with no ischaemia analysed in the study of Burgess et al.,2 age, limiting dyspnoea, Ea, and E/Ea both at rest and post-exercise were among the univariate predictors of reduced exercise capacity <8 METs. In 56 patients with HFNEF, Tan et al.4 found the following exercise echocardiographic parameters to significantly correlate with the peak oxygen consumption: Sa, Ea, E/Ea, apical rotation, mitral flow propagation velocity, and LV untwist in early diastole. Recently, Borlaug et al.33 identified a complex of disturbances contributing to reduced exercise tolerance in subjects with HFNEF including endothelial dysfunction, blunted exercise-induced increases in chronotropy, contractility, and vasodilation resulting in impaired dynamic ventricular-arterial coupling responses during the exercise. In our study, body mass index, age over 65 years, and Sa as a continuous variable or Sa < 8.4 cm/s along with Ea < 8 cm/s were found to be independent predictors of a severely limited exercise tolerance <6 METs. Thus, low exercise tolerance in patients with unexplained exertional dyspnoea, normal LV EF, and the absence of any congenital, valvular, or CAD is likely of a multifactorial aetiology including age, body mass index, LV diastolic and longitudinal systolic dysfunction, endothelial dysfunction, stiff arterial system, and likely also chronotropic incompetence.

Study limitations

Our stress data were not obtained during exercise because of the suboptimal quality of echocardiographic recordings made during exercise on bicycle ergometry. Instead, we used post-exercise data obtained within 90 s after the termination of the exercise. At that time, fusion of early and late diastolic Doppler filling and annular velocities may inhibit the complete exercise data analysis in some patients. The recruitment of our patients, who had been referred for coronary angiography to exclude CAD and frequently suffered from numerous risk factors for diastolic dysfunction, is likely to have affected the prevalence of HFNEF in our patients with exertional dyspnoea. The prevalence of isolated, only exercise-induced HFNEF must be viewed cautiously due to a relatively small number of patients included.

Conflict of interest: none declared.

Funding

The study was supported by a grant of the Ministry of Education of the Czech Republic (MSM, No. 0021622402) and by the European Regional Development Fund – Project FNUSA-ICRC (No. CZ.1.05/1.1.00/02.0123).

References

View Abstract