1 Introduction
In recent years, three-dimensional (3D) echocardiography and myocardial strain measurement have grown into a mature technology, which has been increasingly used as a clinical tool for assessing left ventricular (LV) and left atrial (LA) size and function [
1]. As 3D echocardiography does not rely on geometric assumption for estimating chamber volume and allows assessment of LV and LA size and function with a single data set acquisition [
2], 3DE (3D echocardiography) has overcome most of the technical limitations of 2D echocardiography. Although normal values for 3D echocardiography-derived LV and LA size and function have been proposed by the American Society of Echocardiography (ASE) and the European Association of Cardiovascular Imaging (EACVI) [
2,
3], these values were based on small and mostly white population studies, whereas recent studies have suggested noticeable ethnic difference in LV and LA size and function by 3D echocardiography [
4–
7]. Two newly published studies of the World Alliance Societies of Echocardiography (WASE) clearly demonstrated considerable interracial differences in LV and LA size and function by 3D echocardiography [
8,
9]. Thus, the ASE/EACVI-recommended normal values of LV and LA size and function by 3D echocardiography may not be applicable to nonwhite populations. An Echocardiographic Measurements in Normal Chinese Adults (EMINCA) study organized by the Echocardiography Working Group of the Chinese Society of Ultrasound in Medicine reported normal values of cardiac chamber, great arteries, and Doppler measurements by 2D echocardiography [
10,
11]. However, the normal values related to 3D echocardiography was not studied at that time largely due to the fact that this relatively new technique was not widely applied in China in 2011 when the EMINCA study was first designed [
10]. However, with the rapid progression of percutaneous intervention technology in valvar and congenital heart disease in China, 3D echocardiography has gained widened clinical applications for pre-operative diagnosis and post-operative assessment, and abundant experience has been accumulated in this imaging field. Thus, a set of normal values of 3D echocardiography across a wide range of ages in both sexes of Han Chinese population is increasingly needed. From September 2016 to February 2020, the Echocardiography Working Group of the Chinese Society of Ultrasound in Medicine designed, organized, and conducted a prospective, nationwide, and multicenter study, namely, the Echocardiographic Measurements in Normal Chinese Adults II (EMINCA II), to establish normal reference values of left and right ventricular and atrial size and function by 3D echocardiography and examine the physiologic factors that may affect these normal values. Here, the results of LV and LA size and function by 3D echocardiography were reported.
2 Methods
2.1 Study population
The EMINCA II study was a multicenter, prospective and cross-sectional study, which enrolled 1117 healthy adult Han Chinese volunteers from 28 collaborating echocardiographic laboratories of 20 provinces of China that were accredited by the Chinese Society of Ultrasound in Medicine. The healthy volunteers who met the inclusion and exclusion criteria were recruited by hospital staff members, health examination centers, and adjacent communities. Each of the 28 collaborating laboratories was asked to enroll at least six healthy volunteers in each of the following six age decade groups to ensure an even distribution of sexes and ages among the enrolled volunteers: 8–29, 30–39, 40–49, 50–59, 60–69, and > 70 years. Men and women should be evenly distributed. Considering the wide age spread, the enrolled men and women were divided into three age groups to simplify the data analysis and presentation by following the WASE study: young age: 18–40 years, middle age, 41–65 years, and old age, > 65 years [
8,
9]. The inclusion and exclusion criteria are listed in Tab.1. In addition, subjects with inadequate echocardiographic images were excluded [
8,
9]. Finally, a total of 866 qualified volunteers were enrolled in the EMINCA II study. The body surface area (BSA) was calculated by the formula of Du Bois D and Du Bois EF [
12].
2.2 Echocardiographic image acquisition
Transthoracic echocardiography was performed in all participants in accordance with a standardized acquisition protocol recommended by ASE [
13]. One or two experienced sonographers were selected from each of the participating laboratories who passed the national examination and received a Certification for Ultrasound in Medicine issued by the Chinese National Health Commission. All these sonographers were asked to attend an intensive training at the core laboratory (Qilu Hospital of Shandong University, China) to become acquainted with the study protocol and skillful with the standard 3D echocardiographic image acquisitions.
Ultrasound instruments from two vendors were used for data acquisition: iE33 and Epic7C (Philips Medical Systems, Andover, MA, USA) and Vivid E9 and Vivid E95 (GE Vingmed Ultrasound AS, Horten, Norway). Standard 3D and Doppler echocardiography were performed in all subjects to acquire data in the format of Digital Imaging and Communications in Medicine (DICOM), and all raw format images were transmitted to a core echocardiographic laboratory for image analysis. All subjects were connected to an electrocardiograph and scanned in the left lateral decubitus position to obtain the parasternal long- and short-axis views and apical four-, two-, and three-chamber views. The transducer was angulated carefully to avoid fore-shortening of the LV and LA, and instrumental functional keys were properly adjusted to obtain optimal images. Images were acquired during breath holding at the end of expiration to minimize the effect of respiratory motion on echocardiographic parameters. For Philips instruments, the frame rate was set at ≥ 24/s for 3D echocardiographic recordings. For GE instruments, the frame rate was set at ≥ 1/3 of the heart rate in a given subject for 3D echocardiographic recordings. Multi-beat modes were used for acquisition, and the number of four beats for each view was adjusted and stored on optical disks in DICOM format for off-line analyses.
2.3 Echocardiographic image analysis
Echocardiographic images were initially acquired at each participating laboratory and then transmitted to the core laboratory (Qilu Hospital of Shandong University, China) together with the demographic data and the informed consent form from the enrolled volunteers. Two experienced echocardiographers (Yingbin Wang and Yu Zhang ) in the core laboratory evaluated the image quality and assigned a group of specific number to every center and their enrolled volunteers without being aware of their demographic characteristics such as name, age, or sex. The two echocardiographers analyzed all the 3DE images independently without knowing each other’s results. The averaged results of the two echocardiographers were used for final statistical analysis. A 4-week interval was required for the same echocardiographer to re-analyze the same images and assess the intra-observer variability.
2.4 Quantification of LV by 3D echocardiography
LV volume, function, and strain by 3D echocardiography were quantified using a vendor-independent off-line workstation TomTec2.31 (TomTec Imaging Systems GmbH, Munich, Germany). The 3D full-volume LV data sets were processed by the off-line workstation, where the boundary of LV data sets was automatically recognized, and the planes of the four-, two-, and three-chamber views were automatically displayed (Fig.1). LV end-diastole and end-systole were defined as the frame with the largest and the smallest LV cavity area, respectively. Thereafter, the LV end-diastolic and end-systolic contours in each of the cut-planes were generated. Finally, an endocardial surface shell was created, wherein 3D LV end-diastolic volume (3D-LVEDV), 3D LV end-systolic volume (3D-LVESV), 3D LV stroke volume (3D-LVSV), 3D left ventricular ejection fraction (3D-LVEF), and strain were derived. The LV contours were manually edited by the echocardiographers to ensure an accurate measurement of LV parameters. The LV global longitudinal strain (GLS) and global circumferential strain (GCS) were derived from the endocardial layer on the basis of the entire LV contour length of longitudes and latitudes, respectively, as recommended by the ASE/EACVI [
14].
2.5 Quantification of LA by 3D echocardiography
LA volume and function were quantified using the vendor-independent off-line workstation TomTec2.31 in line with ASE/EACVI guidelines [
2,
15]. The 3D full-volume LA image was obtained in the apical four-chamber view, where the boundary of LA data sets was automatically recognized, and the planes of the four-, two-chamber, and short-axis views were automatically displayed (Fig.2). LA end-diastole and end-systole were defined as the frame with the largest and the smallest LA cavity area, respectively [
16]. The accuracy of endocardial border tracking was visually confirmed throughout the cardiac cycle. The region of interest was manually adjusted to ensure an accurate measurement of LA parameters. The LA emptying fraction (LAEF) was calculated by the software automatically.
2.6 Statistical analysis
Data in each subgroup were presented as the mean ± standard deviation. The lower and upper limits of the normal reference values (LNL and UNL) were defined as 2.5th and 97.5th percentile of the measured values from the corresponding subgroup, respectively.
Sexes were compared using independent sample t-tests. One-way ANOVA with LSD and Dunnet post-hoc test was used for comparison among age groups in both sexes. Interclass correlation coefficients for intra-observer and inter-observer variabilities were calculated using Bland–Altman analysis on GraphPad Prism (GraphPad Software, Inc). Sample t-test was used when comparing other 3DE studies. For strains with negative values, the absolute values were utilized in all statistical comparisons. A two-tailed P < 0.05 was considered statistically significant. SPSS version 26 (SPSS, Inc, Chicago, USA) was used for data analysis.
3 Results
3.1 Demographic data
From November 2017, to February 2020, 1117 individuals were enrolled from 28 medical centers in 20 provinces and municipalities of China, and a total of 866 subjects were qualified for the final analysis. The rest of the subjects were excluded due to cardiac arrhythmias or missed critical echocardiographic views. The number of volunteers initially and finally enrolled in each hospital is shown in Table S1, and a flowchart of volunteer selection is shown in Fig. S1.
The basic demographic characteristics of the study population are described in Tab.2. The median age was 48 years (range, 18–88 years), with 402 men and 464 women distributed evenly among different age groups.
3.2 Quality of echocardiographic data
Measurements of 3D LV volume and function were obtained in 790 subjects (91.2%), and those of 3D LA volume and 3D-LAEF were feasible in 783 subjects (90.4%). The reason for data deficiency was due to the fact that the software TomTec2.31 used for image analysis in this study was unable to recognize some raw data recorded from ultrasonic instruments manufactured by two vendors.
3.3 3D LV volume and function
The values of 3D-LVEDV and 3D-LVEDV indexed by BSA (3D-LVEDVI); 3D-LVESV and 3D-LVESV indexed by BSA (3D-LVESVI); 3D-LVSV and 3D-LVSV indexed by BSA (3D-LVSVI); and 3D-LVEF, 3D-LVGLS, and 3D-LVGCS are displayed in Tab.3 and Fig.3.
Women had smaller 3D-LVEDV, 3D-LVESV, and 3D-LVSV than men, and after normalization by BSA, the sex difference in 3D-LVEDV and 3D-LVESV remained considerable but not that in 3D-LVSV. The 3D LV volume considerablely decreased with age in men. The 3D-LVEF was higher in women than men as a whole, but this difference was not considerable in some age groups (P < 0.05 for all).
For LV strain measurements, the absolute values of 3D-LVGLS were lager in women than in men (P < 0.05 for all). No considerable differences were found among 3D-LVGCS subgroups, either by age or sex.
3.4 Comparison of 3D LV volume and function with current guidelines and WASE results
The mean values and standard deviations of 3D-LVEDVI, 3D-LVESVI, 3D-LVEF, and 3D-LVGLS measured in different sexes in this study were compared with those reported in a Japanese study [
5], which were recommended by the 2015 ASE/EACVI guidelines [
2], and those in WASE 3D study [
8]. The results were shown in Tab.4 and Fig.4.
All of the mean values from each group in the present study fell within the normal ranges defined by guidelines and the WASE study. However, some groups had over 10% or 20% of their cases falling outside of the guideline-defined or WASE-reported normal ranges (Fig.4).
The mean values of 3D-LVEDVI and 3D-LVESVI in the EMINCA II study were considerablely smaller in men and women than those in WASE 3D study [
8], whereas the mean values of 3D-LVEDVI and 3D-LVESVI in both sexes were almost the same with those in a Japanese study [
5] recommended by the 2015 ASE/EACVI guidelines [
2] (Tab.4,
P < 0.05 for all).
The mean values of 3D-LVEF for men and women in the EMINCA II study tended to be larger than the corresponding values in the 3D WASE study [
8], without statistical significance. In addition, the mean values of 3D-LVEF in women were similar to those in the guidelines [
2], whereas the mean values of 3D-LVEF in men were considerablely larger than those recommended by the guidelines [
5] (Tab.4,
P < 0.05 for all).
The mean absolute values of 3D-LVGLS for men and women were larger than the corresponding values in the 3D WASE study [
8], whereas the mean absolute values of 3D-LVGCS in women were smaller than those in the 3D WASE study [
8] (Tab.4,
P < 0.05 for all). Although 2D-LVGLS of −20% has been accepted as a cutoff value between normal and abnormal, this criterion is problematic in the setting of 3D LVGLS because the lower limit of normal range of 3D LVGLS was as low as −15% in the current study and the WASE study. Thus, 3D LVGLS of −15% as the cutoff value is more appropriate for Asian population.
3.5 3D LA volume and function
The values of 3D-LA minimum volume (3D-LAVmin), 3D-LAVmin indexed by BSA (3D-LAVminI), 3D-LA maximum volume (3D-LAVmax), 3D-LAVmax indexed by BSA (3D-LAVmaxI), 3D-LA end-diastolic volume (3D-LAEV), 3D-LAEV indexed by BSA (3D-LAEVI), and 3D-LA emptying fraction (3D-LAEF) are displayed in Tab.5 and Fig.3.
The men in each age group had larger 3D LA volumes measured at different phases of a cardiac cycle than the women. However, after indexation by BSA, most of these differences disappeared. The LA volumes were smaller in the young age group than in the middle and old age groups in men and women even after indexation by BSA. By contrast, the 3D-LAEF values were similar among different sex and age groups.
3.6 Comparison of 3D LA volume and function with WASE results
The current guidelines have no recommended normal ranges of 3D LA volume and function. Thus, the mean values of 3D-LAminI, 3D-LAmaxI, and 3D-LAEF measured with two vendor instruments in different gender and age groups were compared with the corresponding values reported in the WASE 3D-LA study [
9].
As shown in Tab.6, the mean values of 3D LA volumes measured at different phases in all age and sex groups were considerablely smaller than those reported in the WASE 3D-LA study [
9], and these differences remained even after indexation by BSA (
P < 0.05 for all).
The 3D-LAEF in the present study exhibited smaller mean values and standard deviations than those in the WASE 3D-LA study [
9]. However, these differences were not considerable in middle-aged men and old-aged women (
P < 0.05 for all).
3.7 Misclassification of abnormal 3D LV and 3D LA volume and function using the thresholds of WASE study and guidelines
The LLN and ULN of LV and LA measurements are shown in Tab.7. As demonstrated in Fig.5, more than 40% of the healthy Chinese volunteers enrolled in this study could be labeled as having abnormal 3D-LVEDVI, 3D-LVEF, and 3D-LVGLS by using the thresholds of the WASE study [
8], and 16% of these volunteers could be classified as having abnormal 3D-LVEF by using the thresholds recommended by the ASE/EACVI guidelines [
2,
5]. In addition, as depicted in Fig.6, more than 30% of the Chinese volunteers recruited in this study could be regarded as having abnormal 3D-LA volumes by using the thresholds of the WASE study, which provided only age-stratified normal value ranges [
9].
3.8 Reproducibility
Fig.7 shows a high intra-observer and inter-observer reproducibility in this study. The intra-class correlation coefficients for the intra-observer variability of LVEDV, LVESV, LVEF, LVGLS, and LAEF were 0.95, 0.96, 0.94, 0.89, and 0.98, respectively (P < 0.01 for all), and those for the inter-observer variability were 0.87, 0.89, 0.93, 0.90, and 0.91 respectively (P < 0.01 for all).
4 Discussion
The EMINCA II study was a prospective, nationwide and multicenter study that aimed to define the normal reference values of 3D-LV and 3D-LA volume and function in a large cohort of healthy Han Chinese volunteers over a wide range of ages. The present study has several important findings. First, substantial differences existed in a number of 3D-LV and 3D-LA measurements between sexes and among different age groups, a finding similar to that in a previous EMINCA study [10.11]. Second, the normal ranges of 3D-LV and 3D-LA volume and function reported in the present study differed from those recommended by the ASE/EACVI guidelines [
2] and presented by the WASE study [
8,
9]. Third, a substantial portion of the Chinese healthy volunteers could be misdiagnosed as having abnormal 3D-LV and 3D-LA volume and function by using the thresholds from the ASE/EACVI guidelines and the WASE study. Thus, establishing a set of Chinese-specific normal ranges of 3D-LV and 3D-LA volume and function, which should be age- and sex-stratified, is necessary to achieve a correct diagnosis of LV and LA remodeling and dysfunction in clinical practice.
4.1 3D-LV volume and function
Most values of 3D-LVEDV, 3D-LVESV, 3D-LVSV, 3D-LVEF, 3D-LVGLS, and 3D-LVGCS in this study varied with age and sex, and some of these differences remained considerable even after indexation by BSA. These results were consistent with findings of the EMINCA study [
10,
11] and suggested that the normal values of 3D LV volume and function should be age-and sex-stratified.
A comparison of the 3D echocardiographic measurements in this study and 2D echocardiographic measurements in the EMINCA study [
10] is shown in Tab.8. In both sexes, the values of 3D-LVEDV were similar to those of 2D-LVEDV, whereas the values of 3D-LVESV were larger than those of 2D-LVESV. This finding is in accordance with the WASE 3D LV study [
8]. These results were contradictory to previous reports that 2D-LVEDV was consistently lower than 3D-LVEDV, which correlated more closely with CMR-determined LVEDV [
17,
18]. The reason for this discordance may be due to the fact that the measurements of LVEDV in healthy volunteers are not affected by wall motion abnormality and can be accurately calculated with a biplane Simpson algorithm using 2D echocardiography. Meanwhile, inaccuracy in LV image acquisition and software processing by current 3D echocardiographic instruments may lead to erroneous volume calculation.
The values of 3D-LVEDV reported in this study were smaller than those in the NORRE study, which enrolled mostly white Europeans [
4]. Recently, the WASE study found that the measurements of 3D LV volume were smaller among Asians than in Blacks and Whites [
8]. In the present study, the mean values of 3D-LVEDVI and 3D-LVESVI in the Chinese volunteers were smaller than those in the WASE 3D LV study and similar to those in a Japanese study [
5] recommended by the ASE/EACVI guidelines [
2] (Tab.4). These results demonstrated the important effects of ethnicity on LV volume and suggested that the normal values of 3D LV volume and function should be stratified by ethnicity. However, the discrepancies in the derived values may be largely affected by inter-observer variabilities during data acquisition and analysis, because 3D LV images were acquired and measured by different investigators in EMINCA II, NORRE, WASE, and Japanese studies, and thus should be explained with caution.
In contrast to LV volume measurements, the values of 3D-LVEF in both sexes showed substantially lower mean value and smaller standard deviation than those of 2D-LVEF in the EMINCA study, leading to smaller values of lower and upper normal limits of 3D-LVEF than 2D-LVEF [
10]. The reason for this paradox is probably due to the fact that the 3D-LVEDV values were similar to 2D-LVEDV, and the 3D-LVESV values were larger than 2D-LVESV in the EMINCA study, resulting in smaller 3D-LVSV measurements. This phenomenon was also evident in the 2D NORRE study [
19] and the 3D NORRE study [
4]. However, in both sexes, the lower normal limits of 3D-LVEF were quite similar (50%–52%) in the WASE [
8], NORRE [
4], and current studies.
An important finding of the present study was that the absolute values of the mean value of 3D-LVGLS and 3D-LVGCS were similar to those in the WASE [
8] and NORRE [
4] studies (21, 22 for 3D-GLS and 30, 31 for 3D-GCS). However, the standard deviations of 3D-GLS and 3D-GCS in the present study were larger than those in the WASE [
8] and NORRE studies [
4], resulting in smaller values of lower normal limits and lager values of upper normal limits of 3D-GLS and 3D-GCS than those in the WASE [
8] and NORRE [
4] studies.
4.2 3D-LA volume and function
In the current study, most LA volume parameters varied with sex and age, and these physiologic variances cannot be completely corrected by BSA indexation. By contrast, the values of LAEF did not vary with different sex or ages, with the exception that old women had smaller 3D-LAEF than young women. These findings were consistent with the results of the NORRE LA study [
7] and the WASE 3D-LA study [
9] and indicated that normal values of 3D-LA volume should be stratified by sex and age.
The measurements of 3D-LAVmax were considerablely smaller in men and women than the 2D-LAV values in the EMINCA study [
10], as shown in Tab.8. In addition, the values of 3D-LA volumes in men and women of this study were substantially lower than those reported in the NORRE [
7] and WASE [
9] studies even after correction by BSA, suggesting the importance of ethnicity in stratifying normal values of 3D-LA volumes. However, the 3D-LAEF measurements in the current study had smaller mean values and standard deviations than those in the WASE study [
9], leading to relatively narrow normal ranges.
4.3 Misclassification of abnormal 3D LV and LA volume and function
A key finding of this study was that a large proportion of the healthy Chinese volunteers enrolled into the current study could be labeled as having abnormal 3D LV and 3D LA volume and function by using the thresholds from the guidelines and the WASE study. These results have important clinical implications because such a misclassification in healthy subjects may lead to unnecessary treatment and erroneous prediction of adverse outcomes.
4.4 Limitations
The EMINCA II study contains a few limitations. First, only Han Chinese adults were enrolled because the Han nationality accounts for over 90% of the Chinese people. Although the WASE study consisted of Han Chinese, only 131 Chinese adults were enrolled, who were too few to represent a large Han population in China. According to the recommendations by the ASE/EACVI guidelines, a large study in a diverse population is needed to establish normal reference ranges for 3D echocardiography for different ethnic groups2. Second, the 3D values in this study were measured by TomTec2.31, a vendor-independent off-line workstation, which has not been extensively built into current 3D echocardiographic instruments. Thus, the normal values reported in this study may differ from those measured with vendor-dependent software. Finally, the precise comparison between EMINCA II and NORRE, WASE, and Japanese studies was hindered by inter-observer variabilities, because 3D LV images were acquired and measured by different investigators in these studies. Thus, the discrepancies among these results should be explained with caution.
5 Conclusions
The EMINCA II study is a prospective, nationwide and multicenter study, which provided normal reference values of 3D-LV and 3D-LA echocardiographic measurements in healthy Han Chinese adults. Most values of 3D-LV and 3D-LA volume and function varied with sex and age, and a large proportion of these volunteers could be labeled as having abnormal 3D LV and 3D LA volume and function by using the thresholds from the guidelines and the WASE study. Thus, normal ranges of 3D LV and LA measurements should be stratified by sex, age, and race. In addition, a unified software program built in different 3D echocardiographic instruments is highly recommended to reduce extra time and costs required for off-line processing.
PDF
(5176KB)
