Correlations between left ventricular and left atrial function assessed by speckle tracking echocardiography in patients with treated well-controlled arterial hypertension

with of in who e’ velocity <10 cm/s, average E/e’ ratio >14, and peak tricuspid regurgitation velocity >2.8 m/s. Waves E and A of the mitral inow velocity by pulsed wave Doppler from the apical four-chamber view, the E/A ratio, and the velocity waves (e’ and a’) of the mitral annulus septal and lateral regions were recorded using tissue Doppler imaging. An average value of septal and lateral mitral annulus velocities was used to estimate E/e’ ratio.


Background
Arterial hypertension (AH) is one of the most prevalent cardiovascular diseases. European guidelines present strong evidence for the use of antihypertensive treatment in moderate and severe AH, but the usage of pharmacotherapy in patients with mild AH is still debated [1]. Especially in young people, marginally elevated blood pressure (BP) is thought to have no signi cant clinical consequences. However, the results of the Systolic Blood Pressure Intervention Trial (SPRINT) [2], which forced the American cardiology societies to decrease BP thresholds for diagnosing AH [3], sparked a discussion on cardiovascular risk in mild AH.
Even in patients with well-controlled AH, without signi cant comorbidities and with normal echocardiographic indices of left ventricular (LV) systolic and diastolic dysfunction, LV and left atrial (LA) strain abnormalities may sometimes be found in speckle-tracking echocardiography (STE) [4,5,6]. In our earlier study, we revealed that in hypertensive patients with normal LV ejection fraction (EF), left ventricular global longitudinal strain (LV GLS) is reduced compared to healthy people, and its value improves after effective antihypertensive treatment [4]. Also, for left atrial (LA) function, recent studies using STE showed that in hypertensives, LA strain parameters might be impaired, despite normal LA size [7,8].
Based on these reports, we hypothesized that even in the early stage of AH, there might be some interrelated subclinical LV and LA abnormalities, detectable by STE. Therefore, we decided to investigate the relation between LA strain and LV diastolic and systolic function in a group of patients with treated, well-controlled AH.

Study group
We retrospectively analyzed the data collected in the Non-invasive Haemodynamic Assessment in Hypertension (FINE-PATH) study (ClinicalTrials.gov Identi er NCT01996085), which was conducted in the Department of Cardiology and Internal Medicine within the Military Institute of Medicine during the period 2011-2014 [9]. In brief, this trial had a prospective, randomized, controlled design (144 patients enrolled) to assess a novel approach to the treatment of AH; the study involved patients with at least a three-month history of AH de ned according to the European Society of Cardiology guidelines [1]. The exclusion criteria included secondary AH, chronic kidney disease, systolic heart failure, cardiomyopathy, signi cant arrhythmias, signi cant valvular heart disease, chronic obstructive pulmonary disease, previously diagnosed diabetes mellitus, polyneuropathy, and peripheral vascular disease. The following drug classes were used: beta-blockers, angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, calcium channel blockers, and diuretics, either alone or in combination. The study protocol was approved by the Institutional Review Board at the Military Institute of Medicine (no. 21/WIM/2011), and each patient provided written consent.
In this secondary analysis from among the whole group who attended a controlled visit after 12 months of treatment (n=121), 108 patients with acceptable ultrasound image quality were selected. In the nal analysis, 101 patients with normal LV diastolic function evaluated based on current guidelines [10] were included. Collected data included demographic characteristics, the results of a complete clinical examination, 24-h ambulatory blood pressure (ABP) monitoring (ABPM), antihypertensive treatment, and transthoracic echocardiography.
Ambulatory blood pressure monitoring Ambulatory blood pressure monitoring (Spacelabs 90207, Spacelabs, Medical Inc, Redmond, USA) was performed within 2 weeks before echocardiography. The time from 6 a.m. to 10 p.m. was considered daily activity period (daytime) with automatic blood pressure measurement in 10-minute intervals. During night rest (night-time: 10 p.m. -6 a.m.) the measurement was performed every 30 minutes. As a good blood pressure control (well-treated hypertension) was considered a mean 24-hour systolic BP<130 mmHg and diastolic BP<80 mmHg, a daytime systolic BP<135 mmHg and diastolic BP<85 mmHg, and a night-time systolic BP<120 mmHg and diastolic BP<70 mmHg Standard transthoracic echocardiography Transthoracic echocardiography was performed using a high-quality echocardiograph (Vivid 7 or E95, General Electric, United States). The examinations were analyzed off-line by an experienced echocardiographer accredited by the Section of Echocardiography of the Polish Cardiac Society, echocardiography laboratory. All LV and LA measurements were made according to the current guidelines of the European Society of Cardiology [11]. To estimate the size and function of the LA, the following standard parameters were measured: LA end-diastolic diameter, LA area, LA volume (LAV), and LA indexed volume (LAVI). LAV and LAVI were measured using a biplane algorithm from the apical four-chamber (A4C) and twochamber (A2C) views. LA enlargement was de ned as LAVI >34ml/m 2 . To assess LV function, LVEF was calculated using the biplane Simpson formula. LV mass (LVM) was calculated using the linear method according to the recommendations for cardiac chamber quanti cation by echocardiography in adults [11].
Using the parasternal longitudinal axis view), the thicknesses of the interventricular septal, the inferolateral walls, and the LV end-diastolic and end-systolic diameters were obtained. To diagnose LV hypertrophy, LVM was indexed to the body surface area (BSA) and calculated using the DuBois formula (indexed LVM-LVMI).
Diagnosis of LV diastolic dysfunction was based on the current guidelines [10], where the parameters for its identi cation and their cutoffs are as follows: LAVI >34 ml/m2, septal annular e' velocity < 7 cm/s, lateral annular e' velocity <10 cm/s, average E/e' ratio >14, and peak tricuspid regurgitation velocity >2.8 m/s. Waves E and A of the mitral in ow velocity by pulsed wave Doppler from the apical four-chamber view, the E/A ratio, and the velocity waves (e' and a') of the mitral annulus septal and lateral regions were recorded using tissue Doppler imaging. An average value of septal and lateral mitral annulus velocities was used to estimate E/e' ratio.
Speckle-tracking echocardiography (STE) Regional and global longitudinal 2D LA and LV strain was analyzed by STE using GE EchoPAC BT 12 software. LV GLS was assessed using automated imaging software. Detection of the tracked area was performed semi-automatically with two points selected at the level of the mitral annulus and the third point at the apex with the possibility of manual adjustments. The LV GLS values were averaged for all 17 LV segments: seven in the apical four-chamber view, six in the apical two-chamber view, and six in the apical three-chamber view.
Analysis of LA strain was performed off-line, obtained from a non-foreshortened apical, both A4C-and A2Cview images, using conventional 2D gray-scale echocardiography. High frame rates (60-80 frames per second) were used for analysis as recommended in the Expert Consensus Statement [12]. The analysis was performed by an experienced echocardiographer using acoustic-tracking software (EchoPAC, General Electric, USA), allowing off-line semi-automated analysis of speckle-based strain. The LA endocardial border was manually traced in both the A4C and A2C views. An epicardial border was automatically generated by the software, creating a region of interest. The LA was contoured, extrapolating across the pulmonary veins and LA appendage ori ce. Then, after eventual manual adjustment of the ROI shape, the software divided the region of interest into six segments and generated a longitudinal strain curve. To assess all LA strain values, the QRS wave onset was set as a reference point as recommended in the consensus document [13].
The obtained LA longitudinal strain (LAS) parameters were as follows [13]: positive peak strain during the reservoir phase (LASr), corresponding to the atrial reservoir function (positive value); next strain value during early diastole, corresponding to the atrial conduit function and identi ed as the LA-strain-during-conduit phase (LAScd-negative value), and the LA strain during late diastole, corresponding to active atrial contraction and identi ed as the LA-strain-during-contraction phase (LASct-negative value). LASr, LAScd, and LASct were calculated by averaging the values observed in all the LA segments (global LASr, LAScd, and LASct). When some segments were excluded due to the inability to achieve adequate tracking, LAS was calculated by averaging the values measured in the remaining segments. All measurements were obtained during sinus rhythm. As reference LA strain values, we adopted those given in a large meta-analysis carried out by Pathan et al., which included 2,542 healthy subjects [14]. Detailed measurements of LASr, LAScd, and LASct are presented in Figure 1.

Statistical analysis
Statistical analyses were conducted with Statistica 12.0 (StatSoft Inc., Tulsa, OK, USA). The distribution and normality of data were assessed visually and with the Shapiro-Wilk test. Continuous variables were presented as the mean ± standard deviation (SD), whereas categorical variables were presented as absolute and relative values (percentages). A comparison analysis was conducted for two subgroups: patients with high GLS (>the absolute value of −20%; "High GLS"; n=30) and lower GLS (≤ the absolute value of −20%; "Lower GLS"; n=71). The cut-off value was in accordance with the current recommendations [11]. The student's t-test was used for normally distributed data, whereas the Mann-Whitney U-test was used for data with non-normal distribution. A p-value of <0.05 was considered statistically signi cant.

Results
A total of 101 patients (64 men with a mean age of 45.2 years) with well-controlled blood pressure and heart rate control were selected to participate in the study. The mean values of SBP and DBP obtained from ABPM for all patients were 119.7±9.2 mm Hg and 76.7±7.5 mm Hg, respectively. Details of their demographic, clinical, and treatment data are presented in Table 1. Echocardiographic assessmentThe mean value of LAVI was 29.1 (6.8) ml/m 2 , and only 17 (16.8%) patients ful lled the criteria of LA enlargement. Myocardial hypertrophy was diagnosed in 14 (13.9%) subjects, and the mean LVMI for the whole group was 89.5 (±17.5) g/m 2 . The study population was characterized by normal LVEF (65.2 ± 3.3%). However, the mean LV GLS (−18.7 ± 2.6%) was lower than the value (−20%) which is called as a high in a healthy person according to ESC guidelines [1]. Also, analysis of LAS revealed lower values than those previously reported as normal (39% for LASr, -17% for LASct, and −23% for LAScd) [14]. In our group, these LAS parameters were 32.9%, -15.9%, and -13.9%, respectively. Detailed echocardiographic data is summarized in Table 2. Comparison of subgroups with respect to LV GLSThere were no differences between the subgroups "High GLS" and "Lower GLS" for age, sex distribution, or antihypertensive treatment. LA strain measurements were signi cantly lower in patients with lower GLS as compared to those with high GLS. Patients in the "Lower GLS" subgroup presented signi cantly reduced LASr (p = 0.0007) and LAScd (p = 0.008) as well as an increased E/e' ratio (p = 0.037). These results were supported by correlations with LV GLS for LASr, LAScd, E/e' ratio, and E' value.
GLS also proved to be correlated with LVEF. Detailed comparative data are presented in Table 3 Table 4).

Discussion
Our results revealed that even in patients with treated well-controlled AH, who have good BP control and no LV dysfunction in standard echocardiographic evaluation, the values of LV and LA strain measured by STE might be impaired. We also noted that LV GLS is the best correlated with LA strain. Surprisingly, its correlations with LASr and LAScd were even stronger than with LVEF. The interplay between LV diastolic pressures and LA function was also identi ed and con rmed in correlations of some LAS with E/A and e'.
We intended to investigate a group of young, well-treated hypertensives, mostly free from heart function impairment in standard echocardiography (LVH con rmed in only 14%). The exclusion of subjects with reduced and mid-range LVEF from the FINEPATH study along with the additional criteria for this particular analysis (LV diastolic dysfunction) provided a set of data ful lling these assumptions. Even the prevalence of myocardial hypertrophy was very low (13.9%).
However, the use of novel echocardiographic parameters revealed a high frequency of subtle abnormalities occurring in this group. More than two-thirds of patients showed LV GLS values below −20%, and LAS values were also lower than reference values. Previous studies also reported that in hypertensive patients, LV GLS is decreased despite the lack of other comorbidities [5,6]. In our earlier group of 125 patients with mild hypertension [5], we revealed impaired LV GLS values (mean −18.1%) despite normal LVEF (mean 65.3%). Moreover, we observed an association between LV GLS values and the occurrence of LV diastolic dysfunction. The incidence of LV diastolic dysfunction grew higher as the GLS value was more impaired (from 48.4% in the group of subjects with LV GLS < the absolute value of −16.3% to none in the group of patients with GLS < the absolute value of −19.9%) [5]. Similar ndings were reported in a study by Galderisi et al. [6], in which reduced LV GLS was noticed in young patients with AH, and its strong dependence on LV diastolic dysfunction and the degree of LV hypertrophy were found.
Dividing our study group depending on the LV GLS showed that patients with lower LV GLS are characterized by decreased values in parameters re ecting reservoir and conduit LA function. Only contractile LA function was comparable between the two groups of patients. Our results agree with previous reports on LAS in AH. Jarasunas et al. noticed impaired values of all LA function, including reservoir, conduit, and pump, in 63 patients with AH. Both LA conduit and reservoir function decreased with an increasing number of parameters showing LV dysfunction, while contractile function did not change [15]. It is worth mentioning that the study by Jarasunas et al. included patients with concomitant paroxysmal atrial brillation. This atrial arrhythmia may itself lead to impaired LAS [16,17,18]. It was also suggested that the LASr value proved to be a strong parameter of LA brosis in patients with AF [19].
The correlations of LA reservoir (LASr) and conduit (LAScd) strain with LV GLS values and chosen indices of LV diastolic function imply a complex interplay between LA and LV hemodynamics. It seems logical that an interplay was revealed between LA function represented by both reservoir and conduit LA strain and LV systolic function. LA reservoir strain re ects LA relaxation that lasts throughout the period of LV systole. We can treat LASr as an indicator of LA compliance. LA conduit strain spans early LV lling, so it is closely related to LV relaxation. LA contractile strain depends on LA myocardial shortening and LV lling pressures, and it is usually impaired in more advanced LA dysfunction [20]. Our results con rmed that LV longitudinal contraction is related to reservoir and conduit LA strain. We did not nd any published papers reporting such observations.
Our investigation of relations between LAS and LV diastolic function indices revealed correlations between both conduit (LAScd) and contractile (LASct) strain and e'avg, A, and E/A. The pathophysiological mechanisms of LA function can explain these correlations. Conduit LA strain re ects early LV lling; meanwhile, contractile LA strain re ects LV lling during late diastole. Therefore, it seems natural to associate LAScd and LASct with the parameters of both mitral in ow and LV myocardial velocities (especially LAScd with E and E' velocity, LASct with A and A' velocity, and both LAScd and LASct with E/A ratio) [21]. Olsen et al.'s study of patients with cryptogenic stroke revealed that those with impaired LASr < 28.2% have more reduced e'value and increased E/e' ratio compared to subjects with LASr ≥ 28.2% [22]. No previous studies have con rmed the relations we found among patients with mild hypertension and without any other comorbidities.
Other researchers have reported correlations between LAS and E/e' ratio, which is quite a good parameter for non-invasive estimation of LV lling pressure. In a study of hypertensive patients with con rmed LV diastolic dysfunction, Morris et al. [23] demonstrated that LA function assessed by STE strain was inversely related to the degree of LV diastolic dysfunction and to LV lling pressure (measured by mitral E/e' ratio). Other authors have demonstrated that global LASr provides an even better estimation of LV lling pressure [24]. However, both of these studies concerned patients with more advanced LV dysfunction and higher cardiovascular risk compared to ours.

Strengths and limitations
We realize that the methodology and role of STE in the assessment of LA deformation dynamics is still not well established. We assessed strain parameters referring to actual state-of-the-art. Moreover, the measurements were performed by an experienced echocardiographer, and their reproducibility was checked earlier on 20 randomly selected patients. Additionally, the de nition of lower LV GLS and LAS values are still unclear. The expert recommendations speak that a peak GLS in the range of -20% can be expected in a healthy person and we decided to that's why we decided to use this LV GLS cut-off point. Similarly, we tried to compare our LAS values with these obtained in a large meta-analysis.