Background
Pathologic left ventricular (LV) hypertrophy is closely coupled with adverse cardiovascular events. However, normal values of LV mass determined by three-dimensional echocardiography (3DE) have not been established in a large number of healthy subjects over a wide age range. The aims of this study were to (1) validate the accuracy of 3DE for LV mass measurements against cardiac magnetic resonance (CMR), (2) establish the normal range of LV mass index in healthy subjects, and (3) investigate the effects of age, gender, and ethnic diversity on LV mass index.
Methods
In protocol 1, both transthoracic 3DE and CMR were performed on the same day in 57 patients who underwent clinically indicated CMR examinations. In protocol 2, full-volume data sets were acquired with 3DE in 390 healthy subjects. The LV endocardial and epicardial borders were semiautomatically determined at end-diastole using three-dimensional echocardiographic software. LV mass was calculated as (LV epicardial volume − LV endocardial volume) × 1.05.
Results
Excellent correlation was observed between three-dimensional echocardiographic and CMR measurements of LV mass ( r = 0.96). Bland-Altman analysis revealed bias of −4.8 g (−3.9% of the mean), with 95% limits of agreement of ±27.7 g. Normal values of LV mass indexed to body surface area were found to be 70 ± 9 g/m 2 in men and 61 ± 8 g/m 2 in women. Significant age and gender dependence, but no racial dependence, was observed for LV mass index.
Conclusions
Three-dimensional echocardiography is an accurate method for measuring LV mass. Age and gender dependence, but no ethnic dependence, of LV mass index was observed in Japanese and American populations. The reported normal reference values of 3DE-determined LV mass index according to age and gender could potentially be useful for diagnosing LV hypertrophy with excellent accuracy.
Highlights
- •
Three-dimensional echocardiography is an accurate method for measuring LV mass.
- •
Three-dimensional echocardiographic LV mass index values in healthy subjects were similar to those obtained by CMR.
- •
Age and gender dependence of LV mass index was observed.
- •
No ethnic dependence of LV mass index between Japanese and American populations was observed.
Left ventricular (LV) hypertrophy is the hallmark of an adaptive LV remodeling response, which may be associated with adverse outcomes. Cardiac magnetic resonance (CMR) is currently the most accurate and reliable imaging tool for the measurement of LV mass. However, echocardiography is the preferred modality for repeated examinations because it is more versatile, less expensive, and widely available. Although the American Society of Echocardiography recently provided reference ranges for LV mass and LV mass indexed to body surface area (BSA) using M-mode and two-dimensional (2D) echocardiography, the values were significantly larger compared with the corresponding LV mass index values obtained with CMR in normal subjects. This is likely due to the reliance of these imaging modalities on geometric assumptions. Several recent studies have reported the accuracy of LV mass measurements using transthoracic three-dimensional (3D) echocardiography (3DE). However, this modality has not gained widespread clinical use in this context, because of the tedious manual tracing of both endo- and epicardial borders that is required. Thus, normal values for 3D LV mass are not provided in the recent guidelines, because of the paucity of studies reporting normal values. Recent developments in 3D echocardiographic analysis allow the semiautomatic determination of LV mass, which could be easier to implement in daily practice.
Accordingly, the aims of this study were to (1) validate the accuracy of LV mass semiautomatically determined by 3DE against CMR reference, (2) establish the normal range of LV mass index in a large number of healthy subjects, and (3) investigate the effects of age, gender, and race on LV mass indexed to several correcting factors.
Methods
Study Subjects
In protocol 1, we prospectively enrolled 57 consecutive patients who underwent clinically indicated CMR examinations and agreed to undergo additional 3D echocardiographic examinations on the same day.
In protocol 2, a total of 410 adult healthy subjects (mean age, 43 ± 16 years; range, 20–86 years; 199 men) were retrospectively enrolled from two university hospitals (University of Chicago Medical Center and University of Occupational and Environmental Health Hospital). Normal subjects were primarily hospital employees, their relatives, and/or volunteers who were recruited through advertising. The eligibility criteria for healthy subjects included (1) no history of hypertension and normal blood pressure at the time of the echocardiographic examination; (2) no history of diabetes mellitus, hyperlipidemia, or cardiovascular disease; and (3) no history of cardiac medication use. All subjects underwent physical examinations and 2D echocardiographic studies to exclude those with wall motion and/or valvular abnormalities. The ethics committee of each of the hospitals approved the study protocol.
Three-Dimensional Echocardiography
Three-dimensional data acquisition was performed using the iE33 imaging system (Philips Medical Systems, Andover, MA) with a fully sampled matrix-array transducer (X5-1/X3-1 probe; Philips). Transthoracic 3D full-volume data sets were acquired from the apical transducer position during held end-expiration. To ensure the inclusion of the entire left ventricle within the pyramidal scan volume with relatively high volume rates, data sets were acquired using the wide-angle mode, wherein four wedge-shaped subvolumes were acquired with electrocardiographic gating during a single 5- to 7-sec breath-hold.
Three-Dimensional Speckle-Tracking Echocardiography
Three-dimensional analysis of LV volumes and mass was performed by an experienced investigator using commercial software (4D LV Analysis version 3.1.2; TomTec Imaging Systems, Unterschleissheim, Germany). After importing the 3D full-volume data sets, the apical four-chamber, two-chamber, and long-axis views and three short-axis views at end-diastole were automatically extracted from the data set. Nonforeshortened apical views were identified by initializing at end-diastole the LV apex and the center of the mitral annular plane and then connecting both sides of the mitral annulus with the largest LV long-axis dimensions. Subsequently, the 3D endocardial surface was automatically reconstructed with the papillary muscles included in the LV cavity. Manual adjustments of the endocardial surface were performed when necessary. For the determination of LV mass, the epicardial surface delineation was obtained in the end-diastolic frame. The epicardial contour was adjusted in multiple 2D short- and long-axis views extracted from the 3D data sets. LV mass was calculated as (LV epicardial volume − LV endocardial volume)×1.05, ( Figure 1 ). BSA was calculated using the Mosteller formula. Indexed parameters were calculated by dividing by BSA and height 1.7 .
CMR
CMR images were obtained using a 3-T scanner (Discovery; GE Healthcare, Little Chalfont, United Kingdom) with a phased-array cardiac coil during breath-holds gated to the electrocardiogram. Cine images were acquired in multiple short-axis views from the apex to the base of the heart, together with three long-axis views, using the steady-state free precession technique. The following general parameters were used slice thickness, 8 mm; field of view, 40 × 40 cm; scan matrix, 200 × 160; flip angle, 50°; repetition time, 3.8 msec; echo time, 1.7 msec; 20 views per segment; and 20 reconstructed phases. In every short-axis slice in which myocardium was visualized in >50% of the LV circumference, endocardial and epicardial contours were manually traced in the end-diastolic frame. LV mass was determined using the disk-area summation method (Segment, version 1.9; Medviso AB, Lund, Sweden). All tracings were performed by an investigator experienced in the interpretation of CMR images, who had no knowledge of the echocardiographic measurements.
Observer Variability
Intra- and interobserver variability of 3D LV mass measurements was assessed in 20 randomly selected subjects and reported as percentage variability, defined as the absolute difference in percentage of the mean of repeated measurements and intra-class correlation coefficients.
Statistical Analysis
Continuous data are expressed as mean ± SD or as median (interquartile range), according to data distribution. Categorical variables are presented as numbers and proportions. All statistical analyses were performed using commercially available software (JMP version 11.0; SAS Institute, Cary, NC). In protocol 1, linear regression and Bland-Altman analyses were performed to compare LV mass between the two imaging modalities. In protocol 2, differences in measurements between two groups (e.g., male vs female, American vs Japanese) were assessed using unpaired, two-tailed Student’s t tests for continuous variables and χ 2 or Fisher exact tests for categorical variables. Differences in continuous variables among more than two groups (e.g., different age groups) were assessed using one-way analysis of variance with the post hoc Tukey test. P values < .05 were considered to indicate statistical significance.
Results
Accuracy of 3D Echocardiographic Measurements against CMR
Clinical characteristics of the study population used in protocol 1 are shown in Table 1 . All data were analyzed using both 3DE and CMR. Figure 2 shows the results of the comparisons of LV geometric parameters between 3DE and CMR. Although good correlations were noted, LV volumes were significantly underestimated by 3DE compared with CMR (LV end-diastolic volume, 144 ± 52 vs 172 ± 64 mL [ P < .001, r = 0.95]; LV end-systolic volume, 82 ± 49 vs 113 ± 66 mL [ P < .001, r = 0.95]). LV ejection fraction was significantly larger with 3DE compared with CMR (46 ± 14% vs 39 ± 17%, P < .001, r = 0.90). Excellent correlation was noted between 3DE and CMR for LV mass (118 ± 39 vs 123 ± 47 g, P < .001, r = 0.96). Bland-Altman analysis revealed a bias of −4.8 g (−3.9% of the mean), with 95% limits of agreement of ±27.7 g (±22.9% of the mean).
| Variable | Value |
|---|---|
| Age (y) | 64 ± 13 (29–84) |
| Men | 29 (51%) |
| Height (cm) | 159 ± 10 (141–186) |
| Weight (kg) | 58.1 ± 14.2 (37.1–105.7) |
| BMI (kg/m 2 ) | 22.7 ± 3.8 (16.7–34.0) |
| BSA (m 2 ) | 1.58 ± 0.22 (1.26–2.21) |
| Risk factors | |
| Hypertension | 30 (53%) |
| Diabetes | 17 (30%) |
| Hypercholesterolemia | 25 (44%) |
| Chronic kidney disease | 20 (35%) |
| Clinical diagnosis | |
| Coronary artery disease | 29 (51%) |
| Hypertensive heart disease | 3 (5%) |
| Cardiomyopathy | 15 (26%) |
| Valvular heart disease | 5 (9%) |
| Others | 5 (9%) |
Normal Reference Values for 3D LV Mass
Of the 410 subjects screened, 20 (4.8 %) were excluded from the analysis because of elevated systolic blood pressure (>140 mm Hg) at the time of the echocardiographic examination ( n = 18) or poor image quality that precluded appropriate 3D analysis ( n = 2). Thus, the final group consisted of 390 normal subjects (mean age, 44 ± 15 years; 59% Japanese; 51% male). Table 2 shows the clinical and echocardiographic variables, stratified according to gender and race, respectively. Significant gender dependence was observed for all anthropometric and 3DE-derived echocardiographic parameters except age. Reference values for LV mass index were 69.9 ± 8.9 g/m 2 in men and 60.6 ± 8.1 g/m 2 in women. Japanese subjects were significantly older and had lower height, weight, body mass index, and BSA compared with American subjects, irrespective of gender. However, indexing these parameters to BSA adjusted these values to become similar between Japanese and American subjects for both genders.
| Variable | Male | Female | ||||
|---|---|---|---|---|---|---|
| Overall ( n = 199) | Japanese ( n = 121) | American ( n = 78) | Overall ( n = 191) | Japanese ( n = 109) | American ( n = 82) | |
| Age (y) | 43 ± 16 | 46 ± 17 | 40 ± 14 ∗ | 45 ± 15 | 48 ± 16 | 41 ± 13 † |
| Height (cm) | 172 ± 8 | 169 ± 7 | 177 ± 8 ‡ | 160 ± 8 ‡ | 156 ± 6 | 164 ± 7 ‡ |
| Weight (kg) | 69.6 ± 13.0 | 63.1 ± 7.8 | 79.7 ± 13.1 ‡ | 58.7 ± 13.1 ‡ | 51.7 ± 6.3 | 68.0 ± 14. 1 ‡ |
| BMI (kg/m 2 ) | 23.3 ± 3.4 | 22.0 ± 2.2 | 25.3 ± 4.0 ‡ | 22.9 ± 4.2 ∗ | 21.2 ± 2.5 | 25.2 ± 5.0 ‡ |
| BSA (m 2 ) | 1.81 ± 0.19 | 1.71 ± 0.13 | 1.97 ± 0.18 ‡ | 1.60 ± 0.19 ‡ | 1.49 ± 0.10 | 1.75 ± 0.19 ‡ |
| HR (beats/min) | 62 ± 9 | 62 ± 9 | 62 ± 9 | 65 ± 9 ∗ | 64 ± 8 | 66 ± 10 |
| SBP (mm Hg) | 124 ± 10 | 125 ± 10 | 123 ± 10 | 119 ± 11 ‡ | 120 ± 11 | 119 ± 12 |
| DBP (mm Hg) | 73 ± 8 | 73 ± 8 | 72 ± 9 | 70 ± 9 † | 70 ± 9 | 70 ± 9 |
| LVEDV (mL) | 117.6 ± 21.6 | 110.1 ± 17.7 | 129.1 ± 22.2 ‡ | 92.2 ± 16.8 ‡ | 87.1 ± 16.0 | 98.9 ± 15.5 ‡ |
| LVEDVI (mL/m 2 ) | 64.7 ± 10.8 | 64.1 ± 9.6 | 65.7 ± 12.3 | 57.6 ± 9.1 ‡ | 58.1 ± 9.0 | 56.8 ± 9.2 |
| LVESV (mL) | 45.1 ± 10.0 | 41.8 ± 8.5 | 50.2 ± 10.0 ‡ | 33.0 ± 7.8 ‡ | 30.6 ± 7.0 | 36.1 ± 7.7 ‡ |
| LVESVI (mL/m 2 ) | 24.8 ± 4.9 | 24.3 ± 4.7 | 25.5 ± 5.2 | 20.5 ± 4.2 ‡ | 20.4 ± 4.1 | 20.7 ± 4.4 |
| SV (mL) | 72.5 ± 13.2 | 68.3 ± 10.4 | 78.9 ± 14.6 ‡ | 59.1 ± 10.4 ‡ | 56.5 ± 10.2 | 62.7 ± 9.5 ‡ |
| SVI (mL/m 2 ) | 39.9 ± 6.7 | 39.7 ± 5.6 | 40.2 ± 8.2 | 37.0 ± 5.8 ‡ | 37.7 ± 5.9 | 36.0 ± 5.7 ∗ |
| LVEF (%) | 61.8 ± 3.6 | 62.2 ± 3.2 | 61.1 ± 4.0 | 64.3 ± 3.8 ‡ | 64.9 ± 3.8 | 63.5 ± 3.8 ∗ |
| LV mass (g) | 126.9 ± 19.3 | 120.3 ± 16.8 | 137.1 ± 18.6 ‡ | 97.0 ± 15.6 ‡ | 91.6 ± 14.1 | 104.2 ± 14.7 ‡ |
| LV mass index (g/m 2 ) | 69.9 ± 8.9 | 70.0 ± 8.4 | 69.7 ± 9.6 | 60.6 ± 8.1 ‡ | 61.2 ± 8.4 | 59.7 ± 7.7 |
| LV mass volume ratio (g/mL) | 1.09 ± 0.13 | 1.10 ± 0.12 | 1.07 ± 0.13 | 1.06 ± 0.14 ∗ | 1.06 ± 0.15 | 1.06 ± 0.14 |
∗ P < .05, overall male versus overall female, Japanese male versus American male, and Japanese female versus American female.
† P < .01, overall male versus overall female, Japanese male versus American male, and Japanese female versus American female.
‡ P < .001, overall male versus overall female, Japanese male versus American male, and Japanese female versus American female.
Table 3 depicts the effects of age, gender, and race on mass. Subjects were divided into three groups according to age. In each age group, no significant differences in LV mass index were observed between the Japanese and American subjects. For both Japanese and American subjects, female subjects had significantly lower LV mass index values compared with male subjects. Figure 3 shows, for both genders, linear correlations between several LV parameters and age. Advanced age showed a negative trend on LV mass index for both genders ( P = .06).
| Variable | 20–39 y | 40–59 y | 60– y | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Male | Female | Male | Female | Male | Female | |||||||
| Japanese ( n = 55) | American ( n = 46) | Japanese ( n = 36) | American ( n = 39) | Japanese ( n = 34) | American ( n = 22) | Japanese ( n = 32) | American ( n = 36) | Japanese ( n = 32) | American ( n = 10) | Japanese ( n = 41) | American ( n = 7) | |
| LVEDV (mL) | 115.7 ± 16.5 | 132.7 ± 23.3 ‡ | 94.1 ± 16.8 | 103.5 ± 17.7 ∗ | 111.4 ± 17.0 | 122.1 ± 21.1 | 86.1 ± 13.9 | 95.5 ± 11.3 † | 99.3 ± 16.2 | 128.7 ± 17.2 ‡ | 82.0 ± 15.1 | 90.5 ± 16.2 |
| LVEDVI (mL/m 2 ) | 66.0 ± 9.1 | 67.5 ± 13.0 | 62.2 ± 8.5 | 60.2 ± 10.0 | 64.6 ± 9.8 | 62.1 ± 11.0 | 57.4 ± 8.4 | 55.1 ± 6.9 | 60.5 ± 9.8 | 66.2 ± 11.1 | 55.3 ± 9.0 | 47.2 ± 7.2 ∗ |
| LVESV (mL) | 43.9 ± 8.5 | 50.8 ± 10.8 ‡ | 32.5 ± 7.5 | 37.6 ± 8.7 † | 42.3 ± 8.9 | 48.7 ± 10.1 ∗ | 29.5 ± 6.1 | 35.1 ± 6.3 ‡ | 37.8 ± 7.1 | 51.0 ± 6.4 ‡ | 30.0 ± 7.2 | 33.7 ± 8.4 |
| LVESVI (mL/m 2 ) | 25.0 ± 4.9 | 25.8 ± 5.5 | 21.5 ± 4.2 | 21.8 ± 4.8 | 24.4 ± 4.8 | 24.8 ± 5.4 | 20.0 ± 3.9 | 20.3 ± 3.8 | 23.1 ± 4.5 | 26.2 ± 4.1 ∗ | 20.2 ± 4.4 | 17.6 ± 4.2 |
| SV (mL) | 71.9 ± 9.6 | 81.9 ± 15.3 † | 61.6 ± 10.5 | 65.9 ± 10.5 | 69.1 ± 9.0 | 73.4 ± 12.9 | 56.6 ± 9.1 | 60.4 ± 7.0 | 61.5 ± 10.2 | 77.8 ± 12.3 † | 52.0 ± 9.0 | 56.8 ± 9.7 |
| SVI (mL/m 2 ) | 40.9 ± 5.2 | 41.7 ± 8.9 | 40.7 ± 5.3 | 38.3 ± 6.2 | 40.1 ± 5.8 | 37.3 ± 6.4 | 37.7 ± 5.5 | 34.8 ± 4.2 ∗ | 37.4 ± 5.9 | 40.0 ± 7.7 | 35.1 ± 5.6 | 29.6 ± 3.9 ∗ |
| LVEF (%) | 62.3 ± 3.5 | 61.8 ± 4.3 | 65.6 ± 3.4 | 63.9 ± 3.9 | 62.3 ± 3.4 | 60.2 ± 3.8 | 65.9 ± 3.6 | 63.3 ± 3.8 † | 62.0 ± 2.9 | 60.3 ± 3.1 | 63.6 ± 4.1 | 62.9 ± 4.4 |
| LV mass (g) | 124.9 ± 16.2 | 139.7 ± 19.3 ‡ | 96.5 ± 13.5 | 105.6 ± 16.8 ∗ | 120.8 ± 13.3 | 131.1 ± 15.5 ∗ | 89.5 ± 15.0 | 102.4 ± 10.6 ‡ | 111.9 ± 18.3 | 139.0 ± 20.7 † | 89.2 ± 13.2 | 106.5 ± 21.3 ∗ |
| LV mass index (g/m 2 ) | 71.2 ± 8.8 | 70.9 ± 10.0 | 64.0 ± 6.9 | 61.2 ± 8.5 ∗ | 70.0 ± 7.1 | 66.5 ± 7.1 | 59.7 ± 9.1 | 59.0 ± 6.4 | 68.0 ± 9.2 | 71.4 ± 11.7 | 60.2 ± 8.9 | 55.3 ± 8.0 |
| LV mass volume ratio (g/mL) | 1.09 ± 0.12 | 1.07 ± 0.14 | 1.04 ± 0.11 | 1.03 ± 0.14 | 1.10 ± 0.13 | 1.09 ± 0.16 | 1.05 ± 0.15 | 1.08 ± 0.13 | 1.14 ± 0.14 | 1.08 ± 0.10 | 1.11 ± 0.19 | 1.18 ± 0.17 |
∗ P < .05, Japanese male versus American male and Japanese female versus American female.
† P < .01, Japanese male versus American male and Japanese female versus American female.
‡ P < .001, Japanese male versus American male and Japanese female versus American female.
