The Prognostic Value of Left Atrial Reservoir Functional Indices Measured by Three-Dimensional Speckle-Tracking Echocardiography for Major Cardiovascular Events

Abstract

Background: Left atrial (LA) volume and left ventricular longitudinal strain (LVLS) have significant prognostic values for major cardiovascular events (MACEs). Prognostic values of LA reservoir functional indices measured by 3-dimensional (3D) speckle-tracking echocardiography (STE) were evaluated.

Methods and Results: A total of 264 patients, who underwent 2-dimensional (2D) echocardiography and 3DSTE for various underlying heart diseases, were followed up to record MACE. After a mean follow up of 547±435 days, 30 patients developed MACE: 7 cardiac deaths, 6 strokes, 1 non-fatal myocardial infarction, and 22 admissions for heart failure (5 of these had cardiac death after discharge, whereas 1 sustained stroke after discharge). Receiver operating characteristic curve analysis was performed to determine the optimal cut-off levels of 4 LA functional indices: LA emptying fraction (LAEmpF), LA longitudinal strain (LALS), LA circumferential strain (LACS), and LA area change ratio (LAAC), using 3DSTE. Among these factors, 2DLVLS, 3DLAEmpF, and 3DLALS demonstrated a higher hazard ratio (>5.0) than other variables. The 3DLAEmpF and 3DLALS had a higher average treatment effect (ATE) and ATE on the treated (ATT), respectively, than the other indices after propensity score matching. Addition of 3DLAEmpF to the base model using clinical variables and LV ejection fraction or 2DLVLS demonstrated higher prognostic power.

Conclusions: LAEmpF calculated using 3DSTE possessed additive prognostic values for the prediction of MACE.

Left atrial (LA) functional indices have been reported as powerful predictors of major cardiovascular events (MACEs).^1–6 In particular, LA longitudinal strain (LALS), as determined by two-dimensional (2D) speckle-tracking echocardiography (STE), and the LA emptying fraction (LAEmpF) emerged as indices that represent left ventricular (LV) diastolic function and defects in LA myocardial tissue.^7–19 Three-dimensional (3D) STE is the method of choice for the simultaneous analysis of LA volumes and function due to its availability, shorter measurement time, and reproducibility.^19–25 Although LV strains calculated by 3DSTE showed prognostic significance in patients with limited diseases,^26,27 no studies have compared the LA reservoir indices measured by 3DSTE.

This study aimed to determine whether the LA reservoir functional indices LAEmpF; LA emptying fraction, LALS; peak global LA longitudinal strain, LACS; peak global LA circumferential strain; and peak global area change ratio, LAAC, predict MACE better than other indices calculated by standard 2-dimensional echocardiography (2DE), 2DSTE, and LA volume indices calculated by 3DSTE. We also sought to identify which reservoir indices have better prognostic power.

Methods

Study Design

Participants and Clinical Measures A total of 367 randomly selected patients were retrospectively analyzed in the study. All individuals underwent 2DE and 3DSTE examinations for heart disease between 4 January 2013 and 28 February 2016. Of note, the first 2 patients who visited the echocardiography laboratory either on Tuesday or Thursday morning were selected. Exclusion criteria were as follows: (1) atrial fibrillation or history of pulmonary vein isolation (n=64); (2) frequent premature atrial or ventricular contractions (n=2), (3) moderate to severe mitral or aortic regurgitation (n=9); and (4) mitral stenosis and mitral valve replacement (n=5). Furthermore, 16 patients were excluded because of poor quality images. Seven patients were lost during follow up. Among the 367 initially enrolled patients, data from 264 (mean age 65±16 years, 60% male patients) were included in the analysis. Additionally, 60 healthy subjects (aged 23–48 years) were recruited as normal controls from hospital employees and fellows in training that met the following inclusion criteria: age >22 years, no history of cardiovascular or lung diseases, no symptoms 1 month before recording, absence of cardiovascular risk factors (systemic hypertension [HT], smoking, diabetes, and dyslipidemia) assessed at an annual health check in the hospital; no cardioactive or vasoactive treatment, and abnormal results on electrocardiography and physical examination. Exclusion criteria were as follows: status as a professional athlete, pregnancy, body mass index >30 kg/m², no data from patient’s annual health check, and poor 2D echocardiographic image quality. All patients and healthy volunteers provided written informed consent. The local institutional research committee approved the study (F20141130).

Clinical Measures The following clinical characteristics were evaluated using the patients’ medical records: HT, diabetes mellitus, dyslipidemia, past and present smoking habit, chronic kidney disease (CKD: eGFR <45 mL/min/1.73 m²), and history of coronary artery disease (CAD). Clinical records were reviewed to determine the presence of MACE (cardiac death, non-fatal myocardial infarction, stroke, and admission for decompensated heart failure).

Echocardiographic Acquisition and Analysis

All images were obtained using a commercial ultrasound machine (Artida 4D; Toshiba Medical Systems, Nasu, Tochigi, Japan). M mode, 2D, and Doppler measurements were obtained following the American Society of Echocardiography guidelines.^28,29 Tissue Doppler-derived peak-systolic, early, and late-diastolic velocities of the septal mitral annulus were recorded. The LV ejection fraction (LVEF) was calculated using the Simpson biplane method and measured from apical 4- and 2-chamber views. The LA maximal and minimal volumes (LAVmax and LAVmin) were calculated from 2 orthogonal 2DE images via the modified Simpson’s method.²⁹ The maximal LA volume index (LAVImax) and minimal LA volume index (LAVImin) were calculated by dividing LAVmax and LAVmin by the body surface area, respectively.

LAEmpF was calculated as follows:

LAEmpF = (LAVmax − LAVmin) / LAVmax

To measure the 2D LV longitudinal strain (LVLS), endocardial borders were traced on the end-systolic frame using three apical views (4-, 2-, and 3-chamber views). The end-systole was defined by the QRS complex or the smallest LV volume during the cardiac cycle. The software tracks speckles along the endocardial border and myocardium throughout the cardiac cycle. The peak LVLS was computed automatically and generated regional data from 6 segments and an average value for each view. Peak global longitudinal strain was determined as 2DLVLS. The 3D imaging for speckle tracking analysis was performed from an apical position using a commercially available scanner (Artida 4D; Canon Medical Systems, Nasu, Tochigi, Japan) with a fully sampled matrix-array transducer (PST-25SX). Images of the LA were recorded and consisted of 4 wedge-shaped sub-volumes. These were acquired over four consecutive cardiac cycles during a single held breath. Width and depth were held constant to better control the images’ temporal and spatial resolution sector. This resulted in a mean temporal resolution of 32±2.7 volumes per second. The 3DSTE was performed using an offline analysis system (Ultra Extend).^20,21,25 Specifically, the 3DSTE analysis of LA volume required examiners to consecutively set several markers on 2 orthogonal apical views (i.e., from the septal to the lateral edges and from the posterior to the anterior edges of the mitral valve ring). The LA endocardial border was automatically detected using 3D-tracking software (Canon Medical Systems). The examiner subsequently performed a manual adjustment. The software automatically determined the global volume through the entire cardiac cycle. Continuous global volume values were calculated and represented in graphic images. The LAVmax, LAVmin, LAEmpF, LALS, LACS, and LAAC were automatically calculated (Figure 1). The 3D image acquisition took approximately 30 s including setup, whereas the analysis time took approximately 1–2 min (30–60 s for tracing and 30 s for calculations).

Figure 1.

A representative case of 3DSTE analysis for left atrial reservoir functional indices LA images using 3DSTE in a 33-year-old male with HT. The LA endocardium was semi-automatically determined. The volume rate was 28.6/s. White dotted line: global time–volume curves; White line: time–global strain curves. Yellow line: time–radial strain curve that was not assessed in the study. 3DSTE, 3-dimensional speckle-tracking echocardiography; LA, left atrial; HT, hypertension; LS, longitudinal strain; CS, circumferential strain; AC, area change ratio; LAVImax, maximal LA volume index; LAVImin, minimal LA volume index; LAEmpF, LA emptying fraction.

Reproducibility Analysis

Inter- and intra-observer reliability of the 3DLA indices were assessed in 40 randomly selected patients. Intra-observer measures were obtained ~1 week apart in random order. To assess the inter-observer reliability (test–retest reliability), a complete dataset was acquired by a second examiner within 10 min of the first analysis. Each dataset was subsequently analyzed by trained echocardiographers who were blinded to the results of the other examination.

Comparison Between 2DLALS and 3DLALS

A comparison between 2DLALS and 3DLALS was performed in 40 patients with high-quality 2DE images. Patients excluded from the comparison study were those with >2 inadequate tracking segments in the 2D speckle-tracking analysis.

Statistical Analysis

The 2DE and 3DSTA-derived LA volumes were indexed by body surface area. Data were analyzed using commercial software JMP, version 14.0 (SAS Institute Inc., Cary, NC, USA), IBM SPSS version 22 (IBM Corp, Armonk, NY, USA), and R version 3.1.0 (R Foundation for Statistical Computing, Vienna, Austria).

Continuous data are expressed as the mean±SD. Categorical data were presented as numbers or percentages. Differences between continuous variables were assessed using the unpaired Student’s t-test. Additionally, differences between categorical variables were assessed with the chi-squared (χ²) or Fisher’s exact tests. Statistical significance was determined by P<0.05. Relationships between the 2 parameters were evaluated by linear regression analysis.

Data reproducibility was assessed using the intra-class correlation (ICC) with a variance component procedure (restricted maximum likelihood method of estimation). In this analysis, the observer and participant were entered as random effects. Biases and limits of agreement were also calculated using the Bland-Altman plot, to verify the agreement between the 2 datasets of indices.

The cut-off values were defined as the values yielding the maximum area under the receiver operating characteristic curve (ROC), denoted by area under the curve (AUC). The latter was calculated using the sensitivity and specificity of each index. The Cox proportional hazard model was used for the MACE multivariate analysis. Eight models were constructed to avoid problems linked to multi-collinearity (LA volume and functional indices). Propensity score matching was estimated to reduce the potential selection bias and to balance differences between the 2 groups. Patients were matched using a greedy method with 1 : 1 pairs. Variables other than the LA volume and functional indices that contributed to the propensity score were as follows: age, gender, HT, CAD, CKD, and LVEF <40%. The average treatment effect (ATE) and the average treatment effect on the treated (ATT) of each LA index and 2DLVLS were presented to compare prognostic power. Kaplan-Meier survival analysis was performed to plot MACE. The log-rank test was used to analyze the survival curves. Finally, the AUC was calculated for the risk models to identify patients with MACE during the follow up and compare the predictive power of the models using LA indices.

Results

Normal Values

The LA volume and reservoir indices calculated by 3DSTE in 60 normal healthy subjects are shown in Supplementary Table 1. No significant differences were observed between males and females for all indices. There were larger variations in peak LACS compared with those in LALS among normal subjects.

Intra- and Inter-Observer Variability of 3DSTE Indices

We observed outstanding intra- and inter-observer (test–retest) reliability of LA reservoir functional measures. All ICCs were >0.9 (Supplementary Table 2). For inter-observer reliability, the limits of agreement were 7.0%, 7.3%, 7.6%, and 8.4% for LAEmpF, LALS, LACS, and LAAC, respectively.

Correlation Between 2DLALS and 3DLALS

The linear regression analysis of 2DLALS and 3DLALS in 40 patients (R²=0.739, P<0.0001) is shown in Supplementary Figure 1A. The 3DSTE measurements tended to be lower than 2DSTE measurements.

Clinical Characteristics of Study Patients and Number of MACE During Follow up

Baseline clinical characteristics and echocardiographic findings from 264 study patients are shown in Table 1. “Others” included patients with HT, diabetes, premature contractions, and non-significant mild valvular diseases with or without symptoms (palpitations, chest pain, and dyspnea on exertion).

Table 1. Clinical Characteristics and Echocardiographic Findings

Clinical characteristics
Number	264
Age (years)	65±16 (20–93)
Male	159 (60%)
Underlying heart diseases
CAD	54 (20%)
Hypertensive heart disease	17 (6%)
Dilatated cardiomyopathy	22 (8%)
Hypertrophic cardiomyopathy	23 (9%)
Secondary cardiomyopathy	12 (5%)
Others	136 (52%)
2D Echocardiography
LVDD (mm)	50.0±7.9
LVDS (mm)	35.5±11.8
2DLAVImax (mL/m2)	41±15
2DLAVImin (mL/m2)	27±13
LAEmpF (%)	36±16
LVEF (%)	55±16
LVLS (%)	9.5±4.0
E (cm/s)	71±22
A (cm/s)	66±37
E/A	1.3±0.8
Deceleration time (ms)	193±44
e’ average (cm/s)	7.1±2.6
E/e’ average	11.5±5.6
TRPG (mmHg)	27±8
3D Echocardiography
3DLAVImax (mL/m2)	45±15*
3DLAVImin (mL/m2)	29±15*
LAEmpF (%)	37±14*
LALS (%)	16.3±7.3
LACS (%)	17.8±10.3
LAAC (%)	36.8±19

Data are presented as mean±SD, unless otherwise stated. *P<0.001 vs. 2D. 2D, 2 dimensional; 3D, 3 dimensional; A, peak mitral flow velocity at atrial contraction; CAD, coronary artery disease; E, peak mitral flow velocity of the early rapid filling; e’, early diastolic velocity of the mitral annulus; E/e’, E/average of lateral and medial early diastolic velocity of the mitral annulus; LAAC, left atrial area change ratio; LACS, left atrial circumferential strain; LAEmpF, left atrial emptying fraction; LALS, left atrial longitudinal strain; LAVI_max, maximal left atrial volume index; LAVI_min, minimal left atrial volume index; LVDD, left ventricular diastolic dimension; LVDS, left ventricular systolic dimension; LVEF, left ventricular ejection fraction; LVLS, left ventricular longitudinal strain; TRPG, peak pressure gradient of tricuspid regurgitation.

During the mean follow-up period of 547±435 days (range: 1–1,618 days), 30 patients experienced MACE: 2 cardiac deaths, 1 non-fatal myocardial infarction, 5 strokes, and 22 hospitalizations for heart failure. Among the patients hospitalized for heart failure, 5 patients experienced cardiac death and 1 patient sustained a stroke as the second event after discharge).

Interrelationship Between LA Volume Indices, LA Reservoir Functional Indices, and 2D LV Systolic and Diastolic Indices

All 4 3DLA functional indices had a higher correlation coefficient with LAVImin than with LAVImax (Table 2). LAEmpF correlated best with LAVImin out of the 4 indices. LALS had the highest correlation with peak 2DLVLS and E/e′ compared with other 3DLA indices. LAEmpF demonstrated a very high quadratic correlation with LAAC (R²=0.872, P<0.0001; Supplementary Figure 1B).

Table 2. Correlation Coefficient Among 3D Left Atrial Reservoir Indices and Peak 2D Systolic and Diastolic Functional Indices

	3DLALS	3DLACS	3DLAAC	LAEmpF	Peak 2DLVLS	2DLVEF	E/A	E/e’
3DLALS	1	0.598	0.81	0.76	0.515	0.488	0.042	0.372
3DLACS	0.598	1	0.938	0.876	0.323	0.291	0.067	0.284
3DLAAC	0.81	0.938	1	0.922	0.416	0.381	0.042	0.339
LAEmpF	0.76	0.876	0.922	1	0.423	0.344	0.109	0.358
Peak 2DLVLS	0.515	0.323	0.416	0.423	1	0.665	0.066	0.414
2DLVEF	0.488	0.291	0.381	0.344	0.665	1	0.106	0.365
E/A	0.042	0.067	0.042	0.109	0.066	0.106	1	0.08
E/e’	0.372	0.284	0.339	0.358	0.414	0.365	0.08	1

2DLVEF, 2D left ventricular ejection fraction; 2DLVLS, 2D left ventricular longitudinal strain; 3DLAAC, 3D left atrial area change ratio; 3DLACS, 3D left atrial circumferential strain; 3DLALS, 3D left atrial longitudinal strain. Other abbreviations as in Table 1.

Determining the Cut-Off Criteria for MACE and Univariate Analysis

Cut-off criteria were determined using the ROC as maximal AUC values for MACE. Cut-off values for each index were as follows: 2DLAVImax >43.7 mL/m² (AUC=0.686; P=0.0022); 2DLVLS <6.04% (AUC=0.704, P=0.0008); 3DLAVImax >40.8 mL/m² (AUC=0.717, P<0.0001); 3DLAVImin >27.4 mL/m² (AUC=0.769, P<0.0001); 3DLAEmpF <33.0% (AUC=0.783, P<0.0001); 3DLAAC <25.4% (AUC=0.755, P<0.0001); 3DLACS <15.1% (AUC=0.702, P=0.0007); and 3DLALS <10.6% (AUC=0.791, P<0.0001).

Determining the Best Predictive Factor for MACE in LA Functional Indices

A univariate and multivariate Cox proportional hazards model was used after adjusting for age, sex, HT, CAD, CKD, DM, and LVEF <40% to determine factors that were independently associated with MACE (Table 3 and Table 4). Among the 8 indices, 2DLVLS, 3DLAEmpF, and 3DLALS had a hazard ratio (HR) >5.0, and 3DLAEmpF had the highest HR (6.59).

Table 3. Univariate Analysis for MACE

	Cut-off	MACE+ (n=30)	MACE− (n=234)	HR (95% CI)	P value
Age (years)		74±11	64±16	1.06 (1.02–1.09)	0.002
Male		16 (53%)	143 (61%)	1.56 (0.75–3.19)	0.231
DM		12 (40%)	59 (25%)	2.05 (0.96–4.22)	0.055
Hypertension		24 (80%)	144 (62%)	2.34 (1.02–6.32)	0.063
Smoking		17 (57%)	81 (35%)	2.27 (1.10–4.76)	0.027
Dyslipidemia		16 (53%)	100 (43%)	1.39 (0.72–3.11)	0.28
eGFR <45 (mL/min/1.73 m2)		14 (47%)	41 (18%)	4.18 (2.00–8.63)	<0.001
CAD		13 (43%)	41 (18%)	3.61 (1.71–7.42)	<0.001
LVEF (%)		42±15	54±15	1.05 (1.02–1.06)	<0.001
LVEF	<40%	11 (37%)	46 (20%)	2.25 (1.03–4.67)	0.033
2DLVLS	<6.04%	16 (57%)	41 (19%)	6.01 (2.83–13.10)	<0.001
2DLAVImax	>43.7 mL/m2	19 (63%)	71 (30%)	3.23 (1.56–7.04)	0.002
3DLAVImax	>40.8 mL/m2	25 (83%)	111 (47%)	4.20 (1.75–12.47)	0.003
3DLAVImin	>27.4 mL/m2	24 (80%)	90 (38%)	5.49 (2.38–14.89)	<0.001
3DLAEmpF	<33.0%	25 (83%)	79 (34%)	8.48 (3.52–25.17)	<0.001
3DLAAC	<25.4%	21 (70%)	63 (27%)	5.24 (2.47–12.05)	<0.001
3DLACS	<15.1%	24 (80%)	95 (41%)	4.47 (1.95–12.01)	0.001
3DLALS	<10.6%	21 (70%)	46 (20%)	7.44 (3.51–17.15)	<0.001

CI, confidence interval; DM, diabetes mellitus; eGFR, estimated glomerular filtration rate; HR, hazard ratio; MACE, major adverse cardiovascular events. Other abbreviations as in Table 1.

Table 4. Multivariate Analysis for Prediction of MACE

	Model 1		Model 2		Model 3		Model 4		Model 5		Model 6		Model 7		Model 8
	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value
CAD	2.71 (1.20– 5.97)	0.014	1.78 (0.76– 4.11)	0.178	2.44 (1.09– 5.37)	0.028	2.50 (1.11– 5.52)	0.024	2.31 (1.04– 5.03)	0.036	2.22 (1.01– 4.87)	0.047	2.51 (1.12– 5.52)	0.023	1.99 (0.90– 4.37)	0.084
eGFR <45 (mL/ min/1.73 m2)	3.21 (1.50– 6.84)	0.002	2.31 (1.01– 5.20)	0.044	3.02 (1.39– 6.50)	0.005	2.59 (1.18– 5.62)	0.016	2.72 (1.26– 5.81)	0.009	2.89 (1.33– 6.17)	0.006	3.02 (1.40– 6.47)	0.004	2.36 (1.09– 5.10)	0.027
LVEF <40%	0.94 (0.39– 2.17)	0.879	0.51 (0.20– 1.28)	0.152	1.06 (0.45– 2.39)	0.889	0.92 (0.39– 2.08)	0.845	0.76 (0.33– 1.70)	0.509	0.80 (0.34– 1.80)	0.588	0.97 (.42– 2.19)	0.948	0.68 (0.29– 1.55)	0.364
2DLAVImax >43.7 mL/m2	2.31 (1.07– 5.22)	0.037
2DLVLS <6.04%			5.37 (2.06– 13.73)	<0.001
3DLAVImax >40.8 mL/m2					2.92 (1.16– 8.88)	0.035
3DLAVImin >27.4 mL/m2							3.90 (1.59– 11.00)	0.005
3DLAEmpF <33.0%									6.59 (2.60– 20.18)	<0.001
3DLAAC <25.4%											3.80 (1.65– 9.30)	0.002
3DLACS <15.1%													3.31 (1.39– 9.15)	0.011
3DLALS <10.6%															5.57 (2.32– 14.06)	<0.001

Abbreviations as in Tables 1,3.

Following propensity score matching, ATE and ATT were calculated for each LA index and 2DLVLS. The 3DLAEmpF and 3DLALS had a higher ATE (0.15 and 0.24, respectively) and ATT (0.21 and 0.20, respectively) compared with the other indices (Table 5).

Table 5. Average Treatment Effect (ATE) and Average Treatment Effect on the Treated (ATT) for Each LA Reservoir Index

	2DLAVImax	2DLVLS	3DLAVImax	3DLAVImin	3DLAEmpF	3DLALS	3DLAAC	3DLACS
	>43.7 mL/m2	<6.04%	>40.8 mL/m2	>27.4 mL/m2	<33.0%	<10.6%	<25.4%	<15.1%
ATE	0.0089	0.17	0.095	0.12	0.15	0.24	0.13	0.096
ATT	0.14	0.14	0.065	0.15	0.21	0.20	0.13	0.12

Abbreviations as in Tables 1,3.

Comparison of the Prognostic Power of Models Using LA Reservoir Functional Indices

We first performed a logistic regression analysis that included the following: age, sex, HT, CAD, DM, CKD, and LVEF <40% and calculated AUC for the base model. Subsequently, each LA reservoir index was added to the base model and the AUC was evaluated. Among the 8 models, the model including 3DLAEmpF or 3DLALS better predicted future MACE than the base model (P=0.0361 and 0.01, Table 6A). Furthermore, these models showed significantly higher AUC than the base model added with 2DLAVImax (P=0.0202, Table 6B). When the base model was made with 2DLVLS instead of LVEF <40%, the addition of 3DLAEmpF to the base model produced a significantly higher AUC (P=0.02, Table 6C), but 3DLVLS did not show a significant difference.

Table 6. AUC for the Base Model and the 8 Models Added With Each LA Index or 2DLVLS

(A) BASE MODEL (CAD+CKD+LVEF)	AUC	P value
BASE MODEL	0.7295
BASE MODEL+2DLAVImax	0.7649	0.3
BASE MODEL+2DLVLS	0.7701	0.098
BASE MODEL+3DLAVImax	0.7697	0.2969
BASE MODEL+3DLAVImin	0.7891	0.1379
BASE MODEL+3DLAEmpF	0.8186	0.0361
BASE MODEL+3DLAAC	0.7834	0.0819
BASE MODEL+3DLACS	0.7828	0.1076
BASE MODEL+3DLALS	0.8222	0.0109
(B) BASE MODEL (CAD+CKD+LVEF)	AUC	P value
BASE MODEL+2DLAVImax	0.7649
BASE MODEL+2DLVLS	0.7701	0.8561
BASE MODEL+3DLAVImax	0.7697	0.8863
BASE MODEL+3DLAVImin	0.7891	0.451
BASE MODEL+3DLAEmpF	0.8186	0.1387
BASE MODEL+3DLAAC	0.7834	0.4634
BASE MODEL+3DLACS	0.7828	0.5979
BASE MODEL+3DLALS	0.8222	0.0069
(C) BASE MODEL (CAD+CKD+2DLVLS)	AUC	P value
BASE MODEL	0.7567
BASE MODEL+2DLAVImax	0.7841	0.2638
BASE MODEL+3DLAVImax	0.8108	0.0675
BASE MODEL+3DLAVImin	0.8128	0.0732
BASE MODEL+3DLAEmpF	0.8372	0.0202
BASE MODEL+3DLAAC	0.7858	0.2312
BASE MODEL+3DLACS	0.8042	0.0885
BASE MODEL+3DLALS	0.8123	0.1062

AUC, area under the curve; CKD, chronic kidney disease. Other abbreviations as in Table 1.

Survival Analysis

Following propensity score matching, study patients were divided into unaffected and affected groups based on the cut-off values. Seven indices (i.e., 2DLAVImax, 2DLVLS, 3DLAVImax, 3DLAVImin, 3DLAEmpF, 3DLALS, and 3DLAAC) showed significant differences in survival curves between groups. Figure 2 outlines the Kaplan-Meier survival curves using the cutoff criteria of 3DLAEmpF and 3DLALS. Supplementary Figure 2 shows all 8 survival curves before and after propensity score matching.

Figure 2.

Kaplan-Meier survival curves in 2 groups divided by cut-off values determined by 3DLALS and 3DLAEmpF after propensity score matching. 3DLALS, 3-dimensional left atrial longitudinal strain; 3DLAEmpF, 3-dimensional left atrial emptying fraction.

Discussion

The present study showed that 3DLAEmpF and 3DLALS as measured by 3DSTE had additive clinical utility in MACE prediction. Following propensity score matching, 3DLAEmpF and 3DLALS had higher ATE and ATT, and both indices showed significantly different survival curves. A previous report has described the superiority of 3D-derived LA indices over 2DE-derived ones as predictors for MACEs.³⁰ We also reported the superiority of 3DSTE-derived LA volumes and LAEmpF to 2DE as prognostic indicators for MACE.²⁵ This report noted the greater prognostic power of 3DLAEmpF compared with that of 3DLAVImax and 3DLAVImin. It also compared the prognostic values of multiple LA reservoir functional indices calculated by 3DSTE. Likewise, 3DSTE-derived LAEmpF and LALS had good predictive power.

Peak LALS, LACS, LAAC, and LAEmpF represent LA reservoir function. LA reservoir function is influenced by multiple factors, including LV longitudinal contraction, LV filling pressure, and LA myocardial compliance.^{3,5–7,11,14,17,19,31}

In these 4 indices, the 3DLAEmpF showed a highly quadratic correlation with LAAC. This suggests that these indices were almost synonymous. LACS is considered easily influenced by organs surrounding the LA, which prevent LACS to represent the LA compliance. LALS is also a good prognostic indicator. Unfortunately, the significant additive power of 3DLALS disappeared when the base model was structured using 2DLVLS instead of LVEF. This was the second important finding. LALS is more influenced by LVLS than other LA reservoir indices.^9,14,18,32 This study showed a moderate but higher correlation between LALS and LVLS compared with other LA indices. LALS may lose some independent power after replacing LVEF with LVLS.^9,18 The independent value of LALS for predicting MACE could change on the basis of the difference in cardiac disease and clinical characteristics of the study patients. If the disease produces more myocardial injury and/or serious tissue damage in the atrium than the ventricle, the independence of LALS as a prognostic factor for MACE would increase. Atrial sarcoidosis,^33,34 ATTR amyloidosis,^35,36 atrial cardiomyopathy^35,37 with long-lasting atrial fibrillation,³⁸ or atrial ischemia³⁹ are examples of such diseases.

Compared with other 3D STE-derived LA indices, LAEmpF is a balanced index that represents LV diastolic function, LV longitudinal systolic function, and LA compliance, which results in good prognostic value. In this study, stroke and hospitalization for heart failure were observed more frequently than non-fatal myocardial infarction or cardiac death as an adverse event. This may be the reason for more accurate prediction of MACE by 3DLAEmpF than by other indices.

Some studies demonstrated the inferiority of LAEmpF to LALS as a prognostic indicator using 2DE and 2DSTE.^9,16 However, 3D echocardiography (3DE) produces more accurate and reproducible LA volume measurements than 2DE.^30,31,40,41 The prognostic values of LAEmpF using 3DE should be evaluated in the context of various specific cardiovascular diseases.

This study also demonstrated that using 3DSTE produces feasible and reproducible results for measuring LA reservoir indices, similarly to LA volume indices.²⁵ The required time for image acquisition and analysis was short enough for clinical use. Importantly, the simultaneous analysis of volume and functional indices reduces the time and work required for multi-image acquisition, as well as minimizes inter- and intra-observer variability. However, the absolute values of volume parameters often change based on the modality used, although there is a certain degree of correlation. LAEmpF may be a more reproducible and robust parameter when measured by different modalities.

Our results demonstrate a good correlation between LALS measured with 2D and 3DSTE, although the 3DSTE measurements tended to be lower than that obtained by 2DSTE. However, we only used high-quality 2DE images for comparison. Acquiring adequate 2D images for 2DSTE was often difficult by plane through phenomenon. Also, noises from the one frame of one segment influence 2DSTE images more than 3DSTE images. Correlation might be lower with moderate-quality 2D images.

Considering the lower internal pressure in the LA than in the LV and the multiple organs attached to the LA (pulmonary veins, right atrium, aorta, and lungs), we expected different values from the 2 plains by 2DE for 3D structures. Especially when LA are enlarged, obtaining a correct long axis by 2DE may be difficult.

Study Limitations

The present study has several limitations. First, the indices for active pump function in the LA were not assessed because of the difficulty in determining the timing of atrial contraction in patients with tachycardia and/or low voltage P wave. Because a normalized LALS is not always associated with a normalized LA volume,⁴² evaluating the booster pump function using a more elegant method may be promising.^14,39,43 Second, this was a retrospective study. Future studies should focus on patients with each specific cardiovascular disease. Third, this was a single-center study that only enrolled Japanese patients. A multicenter study with multiracial patients will establish the prognostic value of LA reservoir functional indices using 3DSTE. Finally, we did not evaluate the prognostic significance of LV strain calculated by 3DSTE.

A major limitation of LV analysis using the same software was the difficulty in obtaining the data of the apical area. Inclusion of the full LV in the pyramidal angle necessitates a highly skilled operator. Reproducibility of numerical data was not always good even among experienced sonographers. Sufficient LV data could be obtained only in 40% of the randomly selected patients. Improvements in the 3DSTE technique for the measurement of LV and RV strain (e.g., wide angle and better time resolution) may yield different results.

However, primary and secondary atrial cardiomyopathy is commonly encountered in the aging society.^35,37 Once initiated, LA dysfunction tends to progress rapidly owing to the weaker construction and thinner wall of the LA than the LV. LA functional indices may continue to be significant prognostic indicators in older patients.

Conclusions

LAEmpF and LALS calculated by 3DSTE better predicted future MACEs than other indices calculated by 3DSTE and 2DLVLS. Additionally, LALS was more influenced by the LVLS compared with other reservoir functional indices. Future prospective studies are needed to assess the prognostic values of LA functional indices using 3DSTE in each disease.

Acknowledgments

The authors acknowledge the assistance with raw data recordings from sonographers, Ms. Watanabe A, Ms. Watanabe M, Ms. Kitai H, and Ms. Ishida M.

Disclosures

The authors declare that there are no conflicts of interest.

Funding

None.

IRB Information

The ethics committee of the Showa University Fujigaoka Hospital approved the present study (F2014113). The procedures of this study were in accordance with the Declaration of Helsinki.

Data Availability

All data generated or analyzed during this study are included in this published article and the supplementary information files. The deidentified participant data will be shared on a reasonable request basis after permission from the local ethics committee. Please directly contact the corresponding author to request data sharing.

Supplementary Files

Please find supplementary file(s);

http://dx.doi.org/10.1253/circj.CJ-20-0617

References

• 1.   Rosca M, Lancellotti P, Popescu BA, Pierard LA. Left atrial function: Pathophysiology, echocardiographic assessment, and clinical applications. Heart 2011; 97: 1982–1989.
• 2.   Gupta S, Matulevicius SA, Ayers CR, Berry JD, Patel PC, Markham DW, et al. Left atrial structure and function and clinical outcomes in the general population. Eur Heart J 2013; 34: 278–285.
• 3.   Hoit BD. Left atrial size and function. J Am Coll Cardiol 2014; 6: 493–505.
• 4.   Caselli S, Canali E, Foschi ML, Santini D, Di Angelantomio E, Pandian NG, et al. Long-term prognostic significance of three-dimensional echocardiographic parameters of the left ventricle and left atrium. Eur J Echocardiogr 2010; 11: 250–256.
• 5.   Teo SG, Yang H, Chai P, Yeo TC. Impact of left ventricular diastolic dysfunction on left atrial volume and function: A volumetric analysis. Eur J Echocardiogr 2010; 11: 38–43.
• 6.   Stefanadis C, Dernellis J, Toutouzas P. A clinical appraisal of left atrial function. Eur Heart J 2001; 22: 22–36.
• 7.   Murata M, Iwanaga S, Tamura Y, Kondo M, Kouyama K, Murata M, et al. A real-time three-dimensional echocardiographic quantitative analysis of left atrial function in left ventricular diastolic dysfunction. Am J Cardiol 2008; 102: 1097–1102.
• 8.   Russo C, Jin Z, Homma S, Rundek T, Elkind M, Sacco RL, et al. Left atrial minimum volume and reservoir function as correlates of left ventricular diastolic function: Impact of left ventricular systolic function. Heart 2012; 98: 813–820.
• 9.   Ersbøll M, Andersen MJ, Valeur N, Mogensen UM, Waziri H, Møller JE, et al. The prognostic value of left atrial peak reservoir strain in acute myocardial infarction is dependent on left ventricular longitudinal function and left atrial size. Circ Cardiovasc Imaging 2013; 6: 26–33.
• 10.   Pagola J, Gonzalez-Alujas T, Flores A, Muchada M, Rodriguez-Luna D, Sero L, et al. Left atrial strain is a surrogate marker for detection of atrial fibrillation in strokes. Stroke 2014; 45: e164–e166.
• 11.   Donal E, Behagel A, Feneon D. Value of left atrial strain: A highly promising field of investigation. Eur Heart J Cardiovasc Imaging 2015; 16: 356–357.
• 12.   Galli E, Fournet M, Chabanne C, Lelong B, Leguerrier A, Flecher E, et al. Prognostic value of left atrial reservoir function in patients with severe aortic stenosis: A 2D speckle-tracking echocardiographic study. Eur Heart J Cardiovasc Imaging 2016; 17: 533–541.
• 13.   Sarvari SI, Haugaa KH, Stokke TM, Ansari HZ, Leren IS, Hegbom F, et al. Strain echocardiographic assessment of left atrial function predicts recurrence of atrial fibrillation. Eur Heart J Cardiovasc Imaging 2016; 17: 660–667.
• 14.   Donal E, Galli E, Schnell F. Left atrial strain: A must or a plus for routine clinical practice? Circ Cardiovasc Imaging 2017; 10: e007023.
• 15.   Santos AB, Roca GQ, Claggett B, Sweitzer NK, Shah SJ, Anand IS, et al. Prognostic relevance of left atrial dysfunction in heart failure with preserved ejection fraction. Circ Heart Fail 2016; 9: e002763.
• 16.   Freed BH, Daruwalla V, Cheng JY, Aguilar FG, Beussink L, Choi A, et al. Prognostic utility and clinical significance of cardiac mechanics in heart failure with preserved ejection fraction: Importance of left atrial strain. Circ Cardiovasc Imaging, doi:10.1161/CIRCIMAGING.115.003754.
• 17.   Singh A, Addetia K, Maffessanti F, Mor-Avi V, Lang RM. Left atrial strain for categorization of LV diastolic dysfunction. JACC Cardiovasc Imaging 2017; 10: 735–743.
• 18.   Modin D, Biering-Sørensen SR, Møgelvang R, Alhakak AS, Jensen JS, Biering-Sørensen T. Prognostic value of left atrial strain in predicting cardiovascular morbidity and mortality in the general population. Eur Heart J Cardiovascular Imaging 2019; 20: 804–815.
• 19.   Thomas L, Marwick TH, Popescu BA, Donal E, Badano LP. Left atrial structure and function, and left ventricular diastolic function: LACC State-of-the-Art review. J Am Coll Cardiol 2019; 73: 1961–1977.
• 20.   Ebato M, Maezawa H, Tanabe A, Yamaya S, Suzuki H, Takeyama Y. Left atrial area strain measured by three dimensional speckle tracking echocardiography represents global and regional left atrial function at once. J Am Coll Cardiol 2012; 59: E1269.
• 21.   Mochizuki A, Yuda S, Oi Y, Kawamukai M, Nishida J, Kouzu H, et al. Assessment of left atrial deformation and synchrony by three-dimensional speckle tracking echocardiography: Comparative studies in healthy subjects and patients with atrial fibrillation. J Am Soc Echocardiogr 2013; 26: 165–174.
• 22.   Kleijn SA, Aly MF, Terwee CB, van Rossum AC, Kamp O. Three-dimensional speckle tracking echocardiography for automatic assessment of global and regional left ventricular function based on area strain. J Am Soc Echocardiogr 2011; 24: 314–321.
• 23.   Prez de Isla L, Feltes G, Moreno J, Martinez W, Saltijeral A, de Agustin JA, et al. Quantification of left atrial volumes using three-dimensional wall motion tracking echocardiographic technology: Comparison with cardiac magnetic resonance. Eur Heart J Cardiovascular Imaging 2014; 15: 793–799.
• 24.   Yamano M, Yamano T, Iwamura Y, Nakamura T, Shiraishi H, Shirayama T, et al. Impact of left ventricular diastolic property on left atrial function from simultaneous left atrial and ventricular three-dimensional echocardiographic volume measurement. Am J Cardiol 2017; 119: 1687–1693.
• 25.   Tsujiuchi M, Yamauchi T, Ebato M, Maezawa H, Nogi A, Ikeda N, et al. Prognostic value of left atrial size and functional indices measured by 3-dimensional speckle-tracking analysis. Circ J 2019; 83: 801–808.
• 26.   Yu HK, Yu W, Cheuk DK, Wong SJ, Chan GC, Cheung YF. New three-dimensional speckle-tracking echocardiography identifies global impairment of left ventricular mechanics with a high sensitivity in childhood cancer survivors. J Am Soc Echocardiogr 2013; 26: 846–852.
• 27.   Xu Y, Shi J, Zhao R, Zhang C, He Y, Lin J, et al. Two and three-dimensional myocardial strain imaging in the interrogation of sex differences in cardiac mechanics of long-term survivors of childhood cancers. Int J Cardiovasc Imaging 2019; 35: 771–779.
• 28.   Picard MH, Adams D, Bierig SM, Dent JM, Douglas PS, Gillam LD, et al. American Society of Echocardiography recommendations for quality echocardiography laboratory operations. J Am Soc Echocardiogr 2011; 24: 1–10.
• 29.   Lang RM, Badano LP, Mor-Avi V, Afilalo J, Armstrong A, Ernande L, et al. Recommendations for cardiac chamber quantification by echocardiography in adults: An update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. J Am Soc Echocardiogr 2015; 28: 1–39.e14.
• 30.   Wu VC, Takeuchi M, Kuwai H, Iwatani M, Nagata Y, Otani K, et al. Prognostic value of LA volumes assessed by transthoracic 3D echocardiography: Comparison with 2D echocardiography. JACC Cardiovasc Imaging 2013; 6: 1025–1035.
• 31.   Hidemaro S, Masanori K, Ryuhei T, Takashi Y, Koji O, Mitsunobu T, et al. Relationships among pulmonary capillary wedge pressure, dry weight and natriuretic peptide in patients undergoing hemodialysis: A three-dimensional speckle tracking echocardiography study. J Echocardiogr 2020; 18: 160–168.
• 32.   Solomon SD, Biering-Sørensen T. LA strain when ejection fraction is preserved: A new measure of diastolic function? JACC Cardiovasc Imaging 2017; 10: 744–746.
• 33.   Fleming HA. Sarcoidosis of the heart. Am J Med 1978; 64: 915–916.
• 34.   Tigen K, Sunbul M, Karaahmet T, Tasar O, Dundar C, Yalcinsoy M, et al. Early detection of bi-ventricular and atrial mechanical dysfunction using two-dimensional speckle tracking echocardiography in patients with sarcoidosis. Lung 2015; 193: 669–675.
• 35.   Nochioka K, Quarta CC, Claggett B, Roca GQ, Rapezzi C, Falk RH, et al. Left atrial structure and function in cardiac amyloidosis. Eur Heart J Cardiovasc Imaging 2017; 18: 1128–1137.
• 36.   Henein MY, Suhr OB, Arvidsson S, Pilebro B, Westermark P, Hörnsten R, et al. Reduced left atrial myocardial deformation irrespective of cavity size: A potential cause for atrial arrhythmia in hereditary transthyretin amyloidosis. Amyloid 2018; 25: 46–53.
• 37.   van den Berg MP, Mulder BA, Klaassen SHC, Maass AH, van Veldhuisen DJ, van der Meer P, et al. Heart failure with preserved ejection fraction, atrial fibrillation, and the role of senile amyloidosis. Eur Heart J 2019; 40: 1287–1293.
• 38.   Goldberger JJ, Arora R, Green D, Greenland P, Lee DC, Lloyd-Jones DM, et al. Evaluating the atrial myopathy underlying atrial fibrillation: Identifying the arrhythmogenic and thrombogenic substrate. Circulation 2015; 132: 278–291.
• 39.   Bauer F, Jones M, Qin JX, Castro P, Asada J, Sitges M, et al. Quantitative analysis of left atrial function during left ventricular ischemia with and without left atrial ischemia: A real-time 3-dimensional echocardiographic study. J Am Soc Echocardiogr 2005; 18: 795–801.
• 40.   Mor-Avi V, Yodwut C, Jenkins C, Kühl H, Nesser HJ, Marwick TH, et al. Real-time 3D echocardio-graphic quantification of left atrial volume: Multicenter study for validation with magnetic resonance imaging. JACC Cardiovasc Imaging 2012; 5: 769–777.
• 41.   Suh IW, Song JM, Lee EY, Kang SH, Kim MJ, Kim JJ, et al. Left atrial volume measured by real- time 3-dimensional echocardiography predicts clinical outcomes in patients with severe left ventricular dysfunction and in sinus rhythm. J Am Soc Echocardiogr 2008; 21: 439–445.
• 42.   Huynh QL, Kalam K, Iannaccone A, Negishi K, Thomas L, Marwick TH. Functional and anatomic response of left atrium to change in estimate left ventricular filling pressure. J Am Soc Echocardiogr 2015; 28: 1428–1433.
• 43.   Kawasaki M, Tanaka R, Ono K, Minatoguchi S, Watanabe T, Iwama M. A novel ultrasound predictor of pulmonary capillary wedge pressure assessed by the combination of left atrial volume and function: A speckle tracking echocardiography study. J Cardiol 2015; 66: 253–262.