Clinical implication of disturbed left atrial phasic functions in the heterogeneous population associated with hypertension or atrial fibrillation

Patients affected by hypertension and/or PAF, consecutively referred to our hospital from May 2018 to December 2018, were prospectively recruited for this study. Hypertension was diagnosed as systolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg on three or more occasions of different day, or as the current use of antihypertensive drugs in the presence of a documented history of hypertension according to the guidelines [4]. PAF was defined as AF, episodes confirmed by at least one electrocardiography (ECG) within a year, that terminates spontaneously or with intervention within 7 days of onset according to the guidelines [5]. Lone AF [5‐9] referred to AF develops in a subset of younger persons (age ≤ 60 years) without clinical or echocardiographic evidence of cardiopulmonary disease (including hypertension, coronary heart disease) or any known risk factors for AF (including diabetes mellitus, obesity, increased alcohol consumption, sleep apnea et al). Only the ones with hypertension as a possible causative factor for AF and the lone AF individuals were included in the study. Patients with other known causes of AF such as coronary heart disease, heart failure, prior heart surgery, structural heart disease, thyroid dysfunction or renal failure were excluded from the study. Other exclusion criteria were LV ejection fraction (EF) < 50%, non-sinus rhythm during examination, moderate or severe valvular disease, hypertrophic cardiomyopathy, any other arrhythmia, history of ablation procedure, pacemaker implantation or cardiopulmonary surgery, chronic obstructive pulmonary disease and inadequate echocardiographic images. A total of 77 patients met the selection criteria during the period of enrollment. Within the study population, 25 patients had isolated hypertension but not PAF, 24 had lone AF, and 28 hypertensive patients with new-onset PAF, namely, those patients who have been first documented PAF episodes after known a history of hypertension. The control group consisted of 28 healthy individuals who came to our hospital for health check-up, without history of hypertension, PAF or other cardiovascular or systemic disease and with normal findings on clinical examination, ECG, and echocardiography.

The study protocol was approved by the local ethics committee, and informed consent was obtained from all patients before participation.

All subjects, with a synchronous ECG connected, underwent a complete and standard transthoracic echocardiography (TTE) using a Philips iE33 ultrasound machine equipped with a S5-1 probe (Philips Medical Systems, Eindhoven, The Netherlands). All measurements were given as the average values of 3 consecutive cardiac cycles.

All measurements and evaluation were performed according to the guidelines of American society of echocardiography [10]. The LVEF was measured using the modified Simpson’s biplane method. Transmitral peak early diastolic filling velocity E was recorded by pulsed-wave Doppler at the tips of the mitral valve leaflets in an apical four-chamber view. Tissue Doppler imaging was applied in the pulsed-wave Doppler mode to record the mitral annulus peak early diastolic velocities (e’) at the septal and lateral positions. E/e’ was calculated as E divided by the average of the septal and lateral e’ velocities. LA volumes were calculated from the apical four- and two-chamber views of the LA using the biplane method of the discs. The maximum LA volume was indexed by dividing the body surface area to acquire the LA volume index (LAVI).

Two-dimensional (2D) apical four-chamber, two-chamber and long-axis views acquired with at least 60 frames per second were digitized during five consecutive cardiac cycles in cine-loop format.

Views were imported to the 2D speckle-tracking workstation, TomTec-Image Arena 4.0 (2D Cardiac Performance Analysis; TomTec Imaging System, Munich, Germany). Each view was analyzed according to the following steps: using P wave onset of the ECG as the reference point and selecting 3 cardiac cycles. Next, tracing the LA endocardial surface manually by a point-and-click approach. An epicardial surface tracing was automatically generated by the system, creating a region of interest which could be manually adjusted to cover the full thickness of the LA wall. Before processing, a cine loop preview feature visually confirmed that the internal line followed the LA endocardium throughout the 3 cardiac cycles. The software divided the LA wall into 6 segments in each view, and automatically generated longitudinal strain and SR curves of 6 segments and the average curve of 6 segments (Fig. 1). The following indexes were measured from LA strain and SR curves (Fig. 2): (1) peak atrial longitudinal strain and SR during atrial booster pump phase (PALS_pump, PALSR_pump), reservoir phase (PALS_res, PALSR_res), and conduit phase (PALS_cond, PALSR_cond) from average strain and SR curves of 6 segments in the apical four-chamber and two-chamber views, respectively; (2) time to PALS (TpS) and time to PALSR (TpSR), with the referent point at the onset of P-wave, of LA 15 segments in apical four-chamber, two-chamber and long-axis (including 3 segments of LA posterior wall and excluding 3 segments of aortic wall) views during each phase (TpS_pump, TpS_res, TpSR_pump, TpSR_res and TpSR_cond). The LA dyssynchrony index in each phase was defined as the standard deviation of 15 TpS or 15 TpSR parameters in this phase which was corrected by the RR interval duration (TpS_pump-SD%, TpS_res-SD%, TpSR_pump-SD%, TpSR_res-SD% and TpSR_cond-SD%). All PALS and PALSR indexes were obtained by averaging values measured in the apical four-chamber and two-chamber views. Calculated the proportion of reservoir deformation (PALS_res) contributed to conduit phase (PALS_cond/PALS_res ratio) and pump phase (PALS_pump/PALS_res ratio), as well as the ratio of the pump phasic component and the conduit phasic component (PALS_pump/PALS_cond ratio and PALSR_pump/PALSR_cond ratio). The ratio indices were used as indexed parameters of conduit and pump function for further assessment of phasic functions. Both three strain ratios (pump/res, pump/cond, cond/res) and one SR ratio (pump/cond) were calculated for comparing two components of LA emptying function (early passive and later active) among the four study groups.

Inter- and intra-observer variability of LA strain and strain ratio indexes measurements were assessed using Bland-Altman plots (Fig. 3) with data from 20 patients, 5 patients randomly selected from each group respectively, examined twice by one observer twice who was blinded to the results of the previous measurements and by a second observer who was blinded to the values obtained by the first observer, respectively. Overall, small differences were observed for all left atrial strain and strain ratio indexes measurements because most of the differences were within the range of 95% limits of agreement, which suggesting good repeatability and reproducibility in LA strain and strain ratio indexes.

Continuous data were presented as the mean ± SD and dichotomous data as a percentage. Between-group comparisons of continuous variables were performed using one-way analysis of variance (ANOVA), followed by the Bonferroni post hoc test to adjust for multiple comparisons, when normality and homogeneity of variance assumptions are satisfied; otherwise the equivalent non-parametric tests were used when Kolmogorov-Smirnov was in favor of absence of normal distribution or the Levene’s test was in favor of absence of homogeneity of variance. Categorical variables were compared using chi-square tests or Fisher’s exact tests as appropriate. Logistic regression analysis was performed to identify independent differentiators in the heterogeneous population. Significant variables selected in univariate analysis were entered into the multivariate logistic regression analysis. Receiver operating characteristic (ROC) curve was constructed to determine the optimal cut-off value which combine the higher value of specificity plus sensitivity.

Main clinical and conventional echocardiographic characteristics of the four study groups were shown in Table 1. No significant difference in age, sex and heart rate (HR) was observed among the four study groups. Indexed LV mass (LVMI), LAVI and E/e’ ratio were increased in hypertensive patients than normotensive patients (all P < 0.05), but no significant differences was found between hypertensive patients with PAF and those without PAF. Hypertensive patients with PAF had longer period of known hypertension as compared with those without PAF.

The comparisons of phasic LA mechanical functional parameters among the four study groups were shown in Tables 2 and Fig. 4.

Logistic regression analyses were performed to differentiate lone AF from healthy subjects (Table 3) and to identify differentiators for occurrence of AF in hypertensive patients (Table 4). All variables significantly associated with occurrence of AF in univariate analysis were involved in multivariate analysis. PALSR_cond (odds ratio (OR) 0.006, 95% confidence interval (CI) 0.000-0.581, P = 0.028) and TpS_pump-SD% (OR 2.294, 95% CI 1.228-4.285, P = 0.009) were found to have independent ability to differentiate lone AF from healthy subjects in our study (Table 3). Fig. 5a shows the ROC curves constructed using the two differentiators and their combination for determining their abilities of differential diagnosis. The optimal cutoff value of PALSR_cond was recommended as 1.475 s^-1 with sensitivity of 85% and specificity of 71% (area under the ROC curve (AUC) = 0.825, P < 0.001), and the optimal cutoff value of TpS_pump-SD% was recommended as 3.25% with sensitivity of 60% and specificity of 71% (AUC = 0.707, P = 0.015). The combination of the two differentiators, PALSR_cond ≤ 1.475 s^-1 and TpS_pump-SD% ≥ 3.25%, yielded sensitivity of 85%, specificity of 71% and increased AUC = 0.845 (P < 0.001) than individual parameters. While in hypertensive patients (Table 4), PALS_pump (OR 0.620, 95% CI 0.457-0.843, P = 0.002) was found to be an independent differentiator for occurrence of AF or not. ROC analysis identified the optimal cutoff value of PALS_pump as 14.2% with sensitivity of 81% and specificity of 84% (AUC = 0.838, P < 0.001) for distinguishing the hypertensive patients with AF from those without AF (Fig. 5b).

Logistic regression analysis was also performed to indicate the key differentiator between lone AF group and hypertension with AF group (Table 5). Multivariate analysis revealed LAVI (OR 1.077, 95% CI 1.021-1.136, P = 0.006) was an independent characteristic for reflecting different LA remodeling in two types of AF patients. The optimal cut-off value of LAVI was 29.3 mL/m² (sensitivity, 60%; specificity, 75%; AUC = 0.695, P = 0.019) which determined by ROC curve (Fig. 5c).

LA function is known to be divided into three parts: reservoir, conduit, and booster pump function. The three components are mutual interdependence and can be redistributed to compensate for each other in order to maintain cardiac output in early stage of some pathophysiological conditions. In our study, the depression in PALS_res, PALS_cond and PALSR_cond shown in hypertensive patients suggested that LA conduit function was the most severely impaired, followed by reservoir function. The relations of LV longitudinal strain and LA strain were described in previous studies [11‐13] The impairment of LA reservoir and conduit strain were likely reflecting impaired LV longitudinal strain in hypertensive subjects. Hypertension caused an increase in LV wall stress, leading to myocyte hypertrophy and myocardial fibrosis, which resulted in impaired myocardial relaxation and increased LV diastolic stiffness, thereby further induce elevated LV diastolic filling pressure. The impairment of LA reservoir and conduit function better reflects the atrial response to increased ventricular filling pressures. Although LAVI was significantly larger in hypertensive patients than in controls, no significant differences were observed in PALS_pump, PALSR_pump and TpS_pump-SD% between them. This finding suggested that LA booster pump function was still preserved in early stage of hypertension in absence of AF. It is worthwhile to note that our results indicated that hypertensive patients showed a compensatory increasing contribution of booster pump phase to emptying function, whereas in normotensive individuals the LA emptying blood into LV occurred predominantly in atrial conduit phase. Although the absolute value of strain, SR and dyssynchrous in pump phase didn’t show significant differences between the patients with isolated hypertension and controls, an increased proportion of atrial active contractile strain was observed which compensated for the reduction in the proportion of conduit deformation to emptying function. As shown in our study, the indexed parameters of pump function, including PALS_pump/PALS_res ratio, PALS_pump/PALS_cond ratio and PALSR_pump/PALSR_cond ratio, were significantly higher, while the indexed parameters of conduit function, namely, PALS_cond /PALS_res ratio, was significantly lower in patients with isolated hypertension than other three groups as a result of increased active contractility of atrial myocardium [14]. In the early filling of the LV, the LA acts as a conduit, passively emptying during LV relaxation, which is strongly influenced by LV compliance [15]. Hypertension is associated with changes in both LV compliance and relaxation, which influences the balance between early and late filling. After early filling and diastasis, the fall in passive volume shift from LA into LV, forces the atrium to both increase its stroke volume during contraction and recruit more contractile force to work against the increased ventricular pressure and to ensure efficient pumping [14, 16]. Augmented pump function is one of the mechanisms compensating for decreased early filling in patients with reduced LV compliance. An increase in LA active contractility is considered to be caused by the increase of LA volume——Frank-Starling’s law [17]. As a result of optimal use of the Frank-Starling mechanism of the atrial muscle, atrial shortening is augmented with chamber dilation until extreme dilation no longer provokes the Frank-Starling response [18].

Our study demonstrated that PALS_pump and PALSR_pump were lower in patients with both hypertension and PAF than in patients with isolated hypertension. Atrial myocardium being predisposed to increased load and wall stress for a long time, which are developing atrial myocardial fibrosis and dispersion in atrial electrical conductivity. The cumulative long-term effects of high blood pressure may be the underlying precipitating factor and the crucial point for AF. During AF, LA reservoir and conduit functions are impaired and the booster pump function is absent. In sinus rhythm, the atrial myocardial properties could hardly return to normal instantly. Our results also revealed the impairment on LA mechanical function in lone AF patients even with normal LA size, which manifested as not only reduced atrial myocardial deformation properties but also uncoordinated motion during each phase. This result was consistent with previous studies concerned with atrial impairment in AF [19]. Moreover, our findings extend prior work, confirming that PALSR_pump was further reduced in patients with both hypertension and PAF when compared to the lone AF patients. Extreme dilation and LA muscle fibrosis may account for the depressed LA booster pump function in hypertensive patients with PAF [20]. In addition, our results showed LA conduit function was not significantly different between the hypertensive patients with and without PAF, which are consistent with Barbier’ study [21]. However, Cui et al [22] suggested the occurrence of PAF in hypertensive patients is associated with enhanced LA reservoir and conduit function, while our results indicated that PALSR_res was reduced in hypertensive patients with PAF when compared to those hypertensive patients without PAF. Possible reasons for the contradictory results may be various patient clinical characteristics, for example, the hypertension duration, the clinical stages of AF, or the frequency and total number of PAF episodes.

Importance of LA phasic function evaluation is increasingly recognized for its incremental value in terms of prognosis and risk stratification in various disease states [1, 23‐28]. We studied the clinical relevance and prognostic utility of disturbed LA phasic function in heterogeneous population associated with hypertension or AF. Differences in LA phasic function between different groups remained significant ability of differential diagnosis after adjustment for other confounders. These differences were also useful for the stratification of LA mechanical performance. Hypertension and AF are both important risk factors for stroke, heart failure and overall mortality [29‐31]. LA enlargement and dysfunction induced by hypertension was found as independent determinants of new-onset AF [32, 33]. Because of high prevalence of hypertension, it appears to be the most common risk factor for AF and be responsible for more AF than any other risk factor [34, 35]. AF was associated with increased cardiovascular events in hypertensive patients, while hypertension increased the risk of stroke and cardiovascular mortality in patients with AF [36, 37]. Given hypertension is a modifiable risk factor for AF [38], we attempted to determine the threshold of phasic LA strain that distinguishes between isolated hypertension and hypertension with AF, which would be helpful to guide and monitor the progress of treatment of hypertension for preventing AF or reducing the risk of AF. In the current study, we demonstrated that in hypertensive patients, reduced PALS_pump was an independent differentiator for complications of AF or not after multivariate analysis. This result implied that LA mechanical deformation during pump phase was associated with higher risk for developing AF and more important than other factors in the development of AF. Therefore, we proposed atrial strain measurements as a stratification method in hypertension patients to select those at risk of imminent AF development because occurrence of AF was associated with impairment of LA myocardial properties. Remodeling of LA mechanical function in lone AF patients can be detected by 2D STE before structural remodeling. Decreased PALSR_cond combined with increased TpS_pump-SD% can distinguish lone AF patients with normal LA size from healthy subjects, which suggesting cardiologist to guide patients who referred to hospital for episodic palpitation to have a 2D STE evaluation of phasic LA mechanical function. On the other hand, LAVI was proved to be the independent characteristic for reflecting different LA remodeling in two types of AF, which suggesting the occurrence and developing mechanism of AF associated with hypertension, i.e., irreversible LA enlargement with a histological change causing electrical changes in the atrium and then resulting in AF with LA dyssynchrony even during sinus rhythm.

There were several limitations of the present study. First, as dedicated software for LA strain analysis has not yet been released, we used the software for LV analysis to study LA strain. Second, the control group enrolled in our present study consisted of those healthy individuals who came to our hospital for health check-up, without history of cardiovascular or systemic disease and with normal findings on clinical examination, conventional ECG and echocardiography. However, we didn’t make more aggressive efforts, such as 24-hour Holter ECG, to identify if these healthy controls were asymptomatic PAF.