Introduction

Heart failure with preserved ejection fraction (HFpEF) is a clinical syndrome in which patients have symptoms and signs of heart failure (HF) but normal or near-normal left ventricular ejection fraction (LVEF). Patients with reduced ejection fraction (HFrEF) have similar signs and symptoms of heart failure [1–3]. HFpEF accounts for more than 50 % of heart failure patients [4, 5]. The primary pathology in HFpEF is diastolic dysfunction [1]. Three conditions need to be met for the diagnosis of HFpEF, namely, (a) the presence of signs or symptoms of congestive heart failure, (b) normal or mildly abnormal systolic left ventricular function, and (c) evidence of abnormal left ventricular relaxation, filling, diastolic distensibility, and diastolic stiffness [6]. According to the consensus report by the European Society of Cardiology, evidence of abnormal left ventricular relaxation, filling, diastolic distensibility, and diastolic stiffness can be shown invasively, noninvasively by tissue Doppler as E/E’ > 15, or by the rise in biomarkers such as N-terminal pro B-type natriuretic peptide (NT-proBNP) or BNP. When the E/E’ values are between <15 and >8, confirmation by BNP levels or other echocardiographic parameters such as deceleration time (DT), ratio of early (E) to late (A) mitral valve flow velocity (E/A), left ventricular mass index (LVMI), or left atrial volume index (LAVI) are required [6]. LA strain parameters are now increasingly used in daily practice, and are known to be impaired in patients with diastolic dysfunctions. In this study, we used two-dimensional speckle-tracking echocardiography (2D-STE) to evaluate the LA function in patients suspected of having HFpEF but whose E/E’ values are in the gray zone of >8 and <15. We aimed to investigate the probable use of LA strain in diagnosing HFpEF in patients who are in this gray zone.

Patients and methods

Study population

The study comprised 83 patients who underwent echocardiography examination between January 2010 and April 2012 at the Kartal Kosuyolu Cardiovascular Education and Research Hospital and Fatih University Hospital for heart failure symptoms but who had a LVEF of more than 50 % and an E/E’ more than 8 and but less than 15, and were suspected of having HFpEF . The mean age of the patients was 64.7 ± 6.2 years, and 48.2 % were female. The diagnosis of HFpEF was made according to the consensus report by the European Society of Cardiology [6]. HFpEF was diagnosed in the following cases: when BNP was more than 200 pg/ml, LAVI was greater than 40 ml/m2, LVMI was greater than 149 g/m2 (male) or greater than 122 g/m2 (female), E/A>50 yr was less than 0.5, and DT>50 yr was more than 280 ms. Patients were divided into two groups according to the presence of HFpEF. Patients with an LVEF of < 50 %, atrial fibrillation, severe obesity, and significant valvular, pulmonary, or renal diseases were excluded. All work was done in compliance with the Declaration of Helsinki and was performed with the approval of the local ethics committee.

Two-dimensional and Doppler echocardiography

All patients underwent an echocardiographic examination in the left lateral position using the GE Vivid 7 system (GE Vingmed Ultrasound AS, Horten, Norway) with a 3.5-MHz transducer. Blood pressure and heart rate were monitored during the echocardiographic examination. The data were analyzed offline using EchoPAC (GE Vingmed Ultrasound AS). Cardiac dimensions and volumes were measured according to the American Society of Echocardiography’s Guidelines, and LVEF was calculated using the biplane Simpson method [7]. Peak velocities of early and late diastolic filling, mitral E deceleration time, and mitral E/A ratio were derived from Doppler recordings of mitral inflow. Tissue Doppler imaging was used to measure mitral annular velocities. Early diastolic velocity (E’) was measured at both the mitral septal and lateral annulus, and the mean was taken. The ratio of E/E’ was calculated by using the average E’ value. LA volume was calculated using the bi-plane area–length method as described previously; LA volume = 0.85 × ((LA area in four-chamber view)  × (LA area in two-chamber view))/LA length [8]. LA maximum volume (before mitral valve opening) was indexed to body surface area.

Speckle-tracking echocardiography

For speckle tracking analysis, images from apical four- and two-chamber views were obtained using conventional 2D gray-scale echocardiography. Three consecutive cardiac cycles were recorded while the patients held their breath, and averaged. The frame rate was adjusted between 60 and 80 frames/s. The data were analyzed offline using EchoPAC (GE Vingmed Ultrasound AS). The endocardial border was defined manually, and tracing was done by the software automatically for each view. Inadequately tracked segments were either corrected manually or excluded from the analysis. Overall, 976 segments were analyzed. A total of 20 (1.85 %) segments were excluded from the study because no analysis could be done manually and/or automatically. From apical four- and two-chamber views, longitudinal LA strain during ventricular systole (or reservoir phase; LAs-4C-res and LAs-2C-res) was obtained just before mitral valve opening; strain during late diastole (or pump phase; LAs-4C-pump and LAs-2C-pump) was obtained at the onset of the P wave on electrocardiography (Fig. 1). Global longitudinal LA strain during ventricular systole (GLAs-res) and late diastole (GLAs-pump) were calculated by averaging values obtained from all LA segments.

Fig. 1
figure 1

Left atrial Strain on 2D-STE: (a and b) demonstrate the measurement of LAs-res and LAs-pump using 2D-STE from apical four-chamber and two-chamber views respectively. Schematic diagram (c) shows left atrial strain curves

Full size image

Reproducibility

Intra- and interobserver reproducibilities were assessed for both the GLAs-res and the GLAs-pump values. For intraobserver assessment, the measurements were re-analyzed after 4 weeks. Bland–Altman analysis was performed to calculate interobserver reproducibility (mean difference, 95 % confidence interval [CI]) and intraobserver reproducibility (intraclass correlation coefficient, 95 % CI); the intraclass correlation coefficient showed good inter- and intraobserver agreement: interobserver and intraobserver agreement were assessed for GLAs-res, 1.2 (−4.0–(6.4)) and 0.89 (0.81–0.93), respectively; and for GLAs-pump, −2.0 (−5.7–(1.7)) and 0.92 (0.87–0.95), respectively.

BNP measurement

Blood samples for BNP were obtained from all patients before the echocardiographic examination. BNP was measured with the use of the immunoassay method on an ADVIA Centaur-XP device (Siemens Medical Solutions, Germany) using the kits of ADVIA Centaur BNP assay (Bayer Diagnostics, Tarrytown, N.Y.).The measurable range of the BNP assay was 2.0–5,000 pg/ml. The ADVIA Centaur BNP assay had a within-run coefficient of variation of 1.8–4.3 % and a total coefficient of variation of 2.3–4.7 % at concentrations of 29.4–1,736.0 pg/ml.

Statistical analysis

Continuous variables are expressed as mean (±SD) or median as appropriate. A p value of < 0.05 was taken as significant. The independent Student t test or the Mann–Whitney U test was used to compare parametric continuous variables. For categorical variables, the chi-squared test was used. Correlations between variables were tested by using the Pearson or Spearman correlation tests as appropriate. Stepwise multivariate logistic regression analysis was applied to identify the independent predictors of HFpEF evaluated by echocardiography or blood test. Variables with a significant p value on univariate analysis (BNP, LAVI, LVMI, DT, and GLAs-res) were included in the multivariate model. Receiver-operating characteristic (ROC) curves were plotted to determine the optimal cut-off values for GLAs-res in order to predict HFpEF and to establish the optimal cut-off points for use in clinical decision making. Statistical analyses were performed using SPSS (version 15.0 for Windows).

Results

A total of 83 patients with suspected HFpEF were included in the study. The mean age was 64.7 ± 6.2 years, and 52.2 % were female. The patients were divided into two groups two according to the diagnosis of HFpEF as described in the previous section. The clinical, echocardiographic, and demographic characteristics of the patients are shown in Table 1. Baseline demographic and hemodynamic parameters, New York Heart Association (NYHA) class, body mass index (BMI), diabetes mellitus (DM), and hypertension (HT) were similar between the two groups. Of the patients, 36 had NYHA class II and 47 had NYHA class III symptoms. The average E/E’ was 11.22 ± 1.71. In all, 37 patients had high BNP, 26 patients had LAVI > 40 ml/m2, and 15 had increased LVMI. Patients with HFpEF had higher BNP (248.7 ± 48.7 vs. 165.6 ± 26.7 pg/ml, p < 0.001), higher LVMI (132 ± 33.4 vs. 107.1 ± 17.4 g/m2, p < 0.001), higher DT (259.9 ± 44.1 vs. 231 ± 31.5 ms, p = 0.001), and increased LAVI (43.7 ± 9.4 vs. 34.1 ± 3.7 ml/m2, p < 0.001). The GLAs-res value was significantly different between the two groups (17 ± 4.1 vs. 31.9 ± 10.5 %, p < 0.001; Table 2). In univariate analysis, a good negative correlation was seen between GLAs-res and BNP (r = −0.567, p < 0.001) as well as GLAs-res and DT (r = −0.665, p < 0.001), while a moderate negative correlation was found between GLAs-res and LAVI (r = −0.458, p < 0.001) and GLAs-res and LVMI (r = −0.316, p = 0.004). Parameters found to be statistically significant predictors of HFpEF in univariate analysis – i.e., LAVI, BNP, GLAs-res, DT, and LVMI – were studied by logistic regression analysis. GLAs-res, BNP, and LAVI were found to be independent predictors of HFpEF (Table 3). In ROC analysis, the area under the curve (AUC) to predict HFpEF was 0.899 (95 % CI, 0.836–0.962, p < 0.001). A GLAs-res value of < 17.5 % predicted HFpEF with 89 % sensitivity and 55.3 % specificity (Fig. 2). The correlations between GLAs-res and BNP, DT, LAVI, and LVMI are shown in Table 4.

Fig. 2
figure 2

Receiver operating characteristic curve for GLAs-res for the prediction of HFpEF

Full size image
Table 1 Demographic and clinical characteristics of patients with and without HFpEF
Full size table
Table 2 Echocardiographic parameters of patients with and without HFpEF
Full size table
Table 3 Independent predictors of HFpEF
Full size table
Table 4 Correlation between GLAs-res and BNP, DT, LAVI, and LVMI
Full size table

Discussion

In this study, we explored LA functions by 2D-STE in patients with HFpEF. The diagnosis of HFpEF usually demands the use of a set of echocardiographic criteria and can sometimes be challenging. In our study, we found that the LA strain value can give us some idea of HFpEF before the results of BNP testing are attained. Heart failure with preserved LVEF is a clinical syndrome in which patients have symptoms and signs of heart failure, normal or near-normal LVEF, normal or near-normal LV volume, and evidence of diastolic dysfunction [1, 9, 10]. It has been reported to account for more than 50 % of all heart failure patients [4, 5]. The prevalence of HFpEF increases with age [1, 11, 12] and is reported to be more common in women than in men [13–16]. In HFpEF, dyspnea due to pulmonary congestion is frequently the earliest symptom, whereas muscle fatigue is more prominent in heart failure with reduced EF owing to the reduced cardiac output, impairment of vasodilator capacity, and abnormalities of skeletal muscle metabolism [17]. The assessment of LV diastolic function should be part of a routine examination in patients presenting with signs or symptoms of heart failure. The assessment of diastolic function and filling pressures is of great clinical importance for distinguishing this syndrome from other diseases such as pulmonary disease resulting in dyspnea, for assessing the prognosis, and for identifying underlying cardiac disease and planning the best treatment. Evidence of impaired LV relaxation, filling, diastolic distensibility, and diastolic stiffness can be acquired invasively, and is considered as providing definite evidence of HFpEF [6]. Noninvasively, blood flow Doppler and tissue Doppler assessments can be used. The ratio of E, early mitral valve flow velocity, to E’, early tissue Doppler lengthening velocity (E/E’), has been suggested as the best parameter for determining LV filling pressure [6]. When E/E’ is greater than 15, elevated LV filling pressure is established, and HFpEF can be diagnosed, whereas an E/E’ of less than 8 excludes an elevated LV filling pressure. When the E/E’ is in the borderline zone of >8 to <15, more parameters are needed to confirm the diagnosis, such as the difference between the duration of reversed pulmonary vein atrial systole flow (Ard) and the duration of mitral A wave flow (Ad; Ard-Ad > 30 ms), E/A>50 yr < 0.5 and DT>50 yr > 280 ms, LAVI > 40 ml/m2, LVMI > 122 g/m2 (female) and >149 g/m2 (male), or atrial fibrillation. We aimed to evaluate the LA strain values in patients who were in the gray zone of E/E’ >8 and <15.

Recently, LA functions have been widely studied using speckle-tracking echocardiography. Previous studies have shown that LA strain was impaired in patients with both HFpEF and HFrEF [18–22]. It has been reported that LA strain is impaired in patients with diabetes and hypertension even if the size of the left atrium is normal (LA volume indexes < 28 ml/m2) [23]. Guler et al. reported in their study that GLAs-res and GLAs-pump were closely related to LV filling pressure and their values were negatively correlated with NT-proBNP and LA volumetric parameters in patients with nonischemic cardiomyopathy [24]. In the current study, we aimed to find the LA strain values for correctly diagnosing HFpEF. We measured GLAs-res (global longitudinal LA strain during ventricular systole) and GLAs-pump (global longitudinal LA strain during late diastole) in HFpEF patients who are in the gray zone of 8 > E/E’ < 15. We found that the GLAs-res value was significantly different between the two groups (17 ± 4.1 vs. 31.9 ± 10.5 %, p < 0.001). In ROC analysis, the AUC to predict HFpEF was 0.899 (95 % CI, 0.836–0.962, p < 0.001). GLAs-res < 17.5 % predicted HFpEF with 89 % sensitivity and 55.3 % specificity (Fig. 2). This parameter may be useful for the bedside diagnosis of HFpEF when BNP is not yet available or as an additional parameter to the current criteria.

Limitations

For the evaluation of LA strain, we used the software for LV analysis, which might have influenced the echocardiographic results. Obtaining optimal images for the 2D-STE study of the left atrium was sometimes challenging. Furthermore, the study was performed with a relatively small number of patients. Studies with a larger sample size are needed to give a definitive cut-off point for the diagnosis of heart failure with preserved ejection fraction.

Conclusion

LA function as assessed by 2D-STE is impaired in patients with HFpEF. GLAs-res might be useful as an additional tool for the diagnosis of HFpEF.