Background
Despite growing interest in applying three-dimensional (3D) speckle-tracking echocardiography (STE) to measure left ventricular (LV) myocardial deformation in various diseases, normative values for 3D speckle-tracking echocardiographic parameters and the effects of demographic, hemodynamic, and technical factors on these values are unknown.
Methods
In 265 healthy volunteers (age range, 18–76; 57% women), longitudinal strain (3DLε), circumferential strain (3DCε), radial strain (3DRε), and area strain (3DAε) were measured by using vendor-specific (Vsp) 3D speckle-tracking echocardiographic equipment. LV strain was also measured by using Vsp two-dimensional (2D) and vendor-independent 3D speckle-tracking echocardiographic software packages, for comparison.
Results
Reference values (lower limit of normality) for Vsp 3D STE were −17% to −21% (−15%) for 3DLε, −17% to −20% (−14%) for 3DCε, −31% to −36% (−26%) for 3DAε, and 47% to 59% (38%) for 3DRε. Three-dimensional longitudinal strain decreased, whereas 3DCε increased, with aging ( P < .003), with different trends in men and women. Men had lower 3DLε, 3DRε, 3DAε, and 2D longitudinal strain than women ( P < .02). LV 3D strain parameters were also influenced by LV volumes and mass, image quality, and temporal resolution ( P < .02). Reference values obtained by Vsp 2D STE were −20% to −23% (−18%) for 2D longitudinal strain, −20% to −24% (−17%) for 2D circumferential strain, and 39% to 54% (28%) for 2D radial strain ( P < .001 vs Vsp 3D STE). Significantly different 3DCε and 3DRε values were obtained with vendor-independent versus Vsp 3D STE ( P < .001).
Conclusions
In healthy subjects, reference values of LV 3D strain parameters were significantly influenced by demographic, cardiac, and technical factors. Limits of normality of LV strain by Vsp 3D STE should not be used interchangeably with Vsp 2D STE or with Vin 3D STE software.
Deformation imaging by two-dimensional (2D) speckle-tracking echocardiography (STE) has emerged as a reliable quantitative tool for left ventricular (LV) myocardial function analysis, holding important diagnostic, prognostic, and management implications. Three-dimensional (3D) STE represents a further advance in myocardial deformation imaging, allowing a fast and comprehensive evaluation of all LV segments from a single 3D data set. In principle, 3D STE should be more able than 2D STE to capture the complex LV myocardial mechanics, by overcoming the issues of 2D STE related to the “out-of-plane” motion of speckles and partial information contained in a few thin slices of the left ventricle.
Although researchers are increasingly using 3D STE to study various pathologic conditions, the normal pattern of LV myocardial deformation in healthy adults, as characterized by 3D STE, remains unknown. Furthermore, a major issue of 3D speckle-tracking echocardiographic strain is the large intervendor variability of LV myocardial deformation parameters among the various commercially available software packages, which adversely affects their implementation in the clinical routine. Therefore, the definition of specific normative values for each of the available 3D speckle-tracking echocardiographic software packages becomes crucial for the reliable application and interpretation of 3D strain in research, as well as in the clinical arena. So far, no study has specifically aimed to establish normative data for LV 3D strain parameters and identify the factors that contribute to their variation in healthy subjects. Understanding the effects of demographic, hemodynamic, and technical factors on values of the various 3D strain components will help put them in the context of the particular patient and machine from which they have been obtained.
Accordingly, we have designed this prospective study to (1) identify the normative values for LV myocardial 3D strain parameters using a commercially available 3D speckle-tracking echocardiographic platform; (2) analyze the relationships of LV 3D strain parameters with demographic, hemodynamic, cardiac, and technical factors; (3) compare reference values of LV strain obtained by a vendor-specific 3D speckle-tracking echocardiographic software package with those measured by both 2D STE and a vendor-independent 3D speckle-tracking echocardiographic software package in the same population.
Methods
Study Population
Three hundred thirty-eight healthy Caucasian volunteers were prospectively screened at a single tertiary center among hospital employees, fellows in training, their relatives, and people who underwent medical visits for driving or working licenses. The study sample size was determined according to Altman, who set 200 subjects as the minimum number to enroll in a study aiming to assess reference values for biological variables. Enrollment criteria included age > 17 years, body mass index < 30 kg/m 2 , no history of cardiovascular or lung disease, no symptoms, absence of cardiovascular risk factors (i.e. systemic hypertension, smoking, diabetes, and dyslipidemia), no cardioactive or vasoactive treatment, and normal results on electrocardiography and physical examination. Exclusion criteria are shown in Figure 1 .
Blood pressure was measured in all subjects immediately before the echocardiographic examination.
The study was approved by the Ethics Committee of the University of Padua (protocol no. 2380 P, approved October 6, 2011), and written informed consent was obtained from all volunteers before screening for eligibility in the study.
Image Acquisition and Analysis
Echocardiographic data sets were obtained by using a Vivid E9 scanner (GE Vingmed Ultrasound AS, Horten, Norway) equipped with M5S and 4V probes for 2D and 3D acquisitions, respectively. The acquisition protocol included dedicated 2D data sets for measuring the LV myocardial strain parameters by 2D STE using the Q-Analysis package (EchoPAC BT12; GE Vingmed Ultrasound AS). Peak global 2D circumferential strain (2DCε) and 2D radial strain (2DRε) were measured from the parasternal short-axis view at the midpapillary muscle level, and peak global 2D longitudinal strain (2DLε) was computed from the three apical LV views. When more than two LV segments were inadequately tracked in one view, global 2DLε was no longer computed by the 2D speckle-tracking echocardiographic software, and these data sets were excluded from analysis. For consistency, the same rule was applied for 2DCε and 2DRε.
A detailed description of the 3D data set acquisition protocol and of 3D speckle-tracking echocardiographic analysis applied in this study is contained in the Supplemental Methods (available at www.onlinejase.com ). Briefly, four- or six-beat full-volume 3D LV data sets were acquired in all subjects, taking care to avoid any artifacts and to include all LV segments in the data set ( Video 1 available at www.onlinejase.com ). The image quality of 3D data sets was judged subjectively, considering the signal-to-noise ratio, the blood-tissue contrast, and the completeness of LV wall visualization, and was categorized as poor (score = 1), fair (score = 2), good (score = 3), or excellent (score = 4). Three-dimensional data sets were analyzed offline with 4D AutoLVQ (EchoPAC BT12 and BT13; GE Vingmed Ultrasound AS) by a single researcher (D.M.), with 3 years of experience with this software package. Three-dimensional LV volumes, ejection fraction, sphericity, and mass in this healthy cohort have already been reported. For 3D speckle-tracking echocardiographic analysis, the region of interest was set across the entire LV wall ( Figure 2 ). The quality of segmental tracking was validated by the software, as well as by the operator ( Supplemental Methods ; Figure 3 ). When more than three LV segments were inadequately tracked, the software did not provide the global strain values, and the data set was excluded from further analysis. The software provided four 3D strain components: longitudinal strain (3DLε), circumferential strain (3DCε), radial strain (3DRε), and area strain (3DAε) ( Figure 2 ; Videos 2–5 available at www.onlinejase.com ).
The same 3D data sets analyzed with 4D AutoLVQ software were then exported in volumetric Digital Imaging and Communications in Medicine format to a separate workstation equipped with vendor-independent, Digital Imaging and Communications in Medicine–based software for 3D speckle-tracking echocardiographic analysis (4D LV-Analysis version 3.1.3; TomTec Imaging Systems GmbH, Unterschleissheim, Germany). The same researcher (D.M.) performed the 3D speckle-tracking echocardiographic quantification, independently and blinded to the previous 2D and 3D strain measurements, to obtain four strain parameters: 3DLε, 3DCε, 3DRε, and principal tangential strain (a composite myocardial deformation parameter including both 3DLε and 3DCε measured at the endocardium).
Statistical Analysis
Normal distribution of variables and uniform distribution of subjects per age decade were assessed using the Kolmogorov-Smirnov test. Accordingly, continuous variables are summarized as mean ± SD if normally distributed and as median (interquartile range [IQR]) otherwise, whereas scalar variables are reported as percentages. Except for reporting reference values (when strain parameters other than 3DRε were considered negative), all LV strain parameters elsewhere were interpreted as absolute values, and comparisons were based on strain magnitude (with lower strain values indicating worse deformation and higher strain values better deformation, independent of mathematical sign). Similarly, lower limits of normality for all 3D speckle-tracking echocardiographic parameters refer to strain magnitude and are expressed as the 95th or, as appropriate, fifth percentile (depending on the mathematical sign of the respective strain parameter).
Differences between values in men and women were assessed using unpaired t tests for normally distributed variables and Mann-Whitney U tests otherwise. LV strain values obtained with two different software packages from the same subject were compared by using paired Mann-Whitney U tests. Pearson correlation was used to analyze the relationships between strain parameters and demographic, cardiac, and technical variables. A stepwise multivariate linear regression analysis was used to assess the association of each 3D strain parameter with the following covariates: age, gender, weight, height, blood pressure, LV volumes, LV mass, temporal resolution, and image quality of the 3D data set (having P values < .10 in individual models). Intraobserver reproducibility was analyzed in 15 random subjects by the same observer (D.M.), performing the same measurements of 3D strain parameters 2 months later. Interobserver reproducibility was analyzed in the same subjects by two independent observers (D.M. and L.P.B.). Reproducibility was assessed for all strain parameters by Bland-Altman analysis and is reported as the difference between the corresponding pairs of measurements as a percentage of their mean, in which the difference was reported as absolute percentages (strain units) and relative percentages (proportionally to the different magnitudes of 3D strain parameters).
All analyses were performed using SPSS version 21.0 (SPSS, Inc, Chicago, IL). Differences among variables were considered significant at P < .05.
Results
The final study population comprised 265 subjects ( Figure 1 ), whose characteristics are presented in Table 1 . The age of subjects ranged from 18 to 76 years, and ≥46 subjects per age decade were included in the study (mean, 53 ± 11 subjects/age decade). Women were slightly prevalent (57%). Men had higher blood pressure; larger body sizes, LV volumes, and LV mass; and lower ejection fractions than women ( Table 1 ).
| Variable | All | Women | Men | P ∗ |
|---|---|---|---|---|
| ( n = 265) | ( n = 150) | ( n = 115) | ||
| Age (y) | 45 ± 14 | 44 ± 14 | 45 ± 14 | .46 |
| Height (cm) | 170 ± 9 | 165 ± 7 | 177 ± 6 | <.001 |
| Weight (kg) | 67 ± 11 | 61 ± 8 | 76 ± 9 | <.001 |
| Body mass index (kg/m 2 ) | 23.2 ± 2.9 | 22.3 ± 2.7 | 24.3 ± 2.8 | <.001 |
| Body surface area (m 2 ) | 1.78 ± 0.18 | 1.66 ± 0.12 | 1.93 ± 0.12 | <.001 |
| Systolic blood pressure (mm Hg) | 120 ± 13 | 116 ± 13 | 126 ± 10 | <.001 |
| Diastolic blood pressure (mm Hg) | 72 ± 8 | 71 ± 8 | 75 ± 7 | <.001 |
| Heart rate (beats/min) | 66 ± 10 | 67 ± 10 | 66 ± 10 | .52 |
| Temporal resolution (volumes/sec) | 37 ± 8 | 38 ± 9 | 37 ± 8 | .34 |
| Excellent or good image quality | 77% | 79% | 74% | .42 |
| 3D LV end-diastolic volume (mL/m 2 ) | 60 ± 10 | 57 ± 9 | 63 ± 11 | <.001 |
| 3D LV end-systolic volume (mL/m 2 ) | 22 ± 5 | 20 ± 4 | 24 ± 5 | <.001 |
| 3D LV stroke volume (mL/m 2 ) | 38 ± 7 | 37 ± 6 | 39 ± 7 | .003 |
| 3D LV ejection fraction (%) | 64 ± 4 | 65 ± 4 | 62 ± 4 | <.001 |
| 3D LV mass (g/m 2 ) | 75 ± 9 | 73 ± 8 | 77 ± 10 | .001 |
The mean temporal resolution of 3D data sets was 37 ± 8 volumes/sec (range, 27–67 volumes/sec), and the mean temporal resolution of 2D data sets was 71 ± 5 frames/sec (range, 54–80 frames/sec). Image quality of 3D data sets was scored as excellent in 37%, good in 40%, fair in 19%, and poor in 4% of subjects. Forty-one percent of 3D data sets had one or two LV segments incompletely visualized throughout the cardiac cycle, with 90% of these in the apical region. The overall feasibility of global 3D speckle-tracking echocardiographic analysis with 4D AutoLVQ was 90%. The feasibility of segmental 3D speckle-tracking echocardiographic analysis with 4D AutoLVQ is presented in Figure 4 . Substantial intersegment variability in tracking quality can be observed, ranging from 100% at the midventricular level to 46% at the basal inferolateral level.
Reference Values of LV 3D Strain Parameters
Reference values and lower limits of normality for LV 3D strain parameters obtained by vendor-specific 3D speckle-tracking echocardiographic software are listed in Table 2 . Overall, men had lower magnitudes of 3DLε, 3DRε, and 3DAε than women, although this effect was not evident in all age groups. Conversely, we found no effect of gender on 3DCε measured by vendor-specific 3D speckle-tracking echocardiographic software.
| Strain parameter | All | Women | Men | P † | |||
|---|---|---|---|---|---|---|---|
| Median (Q1, Q3) | LLN | Median (Q1, Q3) | LLN | Median (Q1, Q3) | LLN | ||
| 3DLε (%) | |||||||
| All (W = 150, M = 115) | −19 (−21, −17) | −15 | −20 (−21, −18) | −15 | −18 (−19, −16) | −15 | <.001 |
| 18–29 y (W = 28, M = 20) | −20 (−21, −18) | −17 | −20 (−21, −18) | −17 | −19 (−20, −18) | −17 | .914 |
| 30–39 y (W = 27, M = 21) | −19 (−21, −18) | −16 | −20 (−22, −19) | −16 | −18 (−19, −17) | −15 | .002 |
| 40–49 y (W = 41, M = 31) | −19 (−21, −17) | −14 | −20 (−22, −17) | −14 | −18 (−19, −16) | −14 | .006 |
| 50–59 y (W = 28, M = 18) | −17 (−20, −15) | −14 | −20 (−21, −16) | −13 | −17 (−18, −16) | −14 | .002 |
| ≥60 y (W = 26, M = 25) | −17 (−19, −16) | −13 | −19 (−20, −17) | −14 | −17 (−19, −15) | −12 | .014 |
| 3DCε (%) | |||||||
| All (W = 150, M = 115) | −18 (−20, −17) | −14 | −18 (−20, −17) | −14 | −19 (−20, −16) | −14 | .978 |
| 18–29 y (W = 28, M = 20) | −18 (−20, −16) | −14 | −17 (−19, −16) | −13 | −19 (−20, −16) | −14 | .228 |
| 30–39 y (W = 27, M = 21) | −18 (−20, −16) | −14 | −18 (−19, −16) | −14 | −18 (−20, −16) | −14 | .441 |
| 40–49 y (W = 41, M = 31) | −18 (−21, −17) | −15 | −19 (−21, −17) | −15 | −18 (−21, −17) | −15 | .590 |
| 50–59 y (W = 28, M = 18) | −18 (−21, −16) | −15 | −19 (−22, −16) | −13 | −17 (−21, −15) | −15 | .286 |
| ≥60 y (W = 26, M = 25) | −19 (−22, −18) | −16 | −20 (−22, −18) | −15 | −19 (−22, −18) | −16 | .257 |
| 3DRε (%) | |||||||
| All (W = 150, M = 115) | 52 (47−59) | 38 ∗ | 54 (48, 59) | 40 ∗ | 51 (46, 58) | 38 ∗ | .021 |
| 18–29 y (W = 28, M = 20) | 53 (48, 58) | 40 ∗ | 52 (47, 57) | 40 ∗ | 55 (49, 60) | 46 ∗ | .277 |
| 30–39 y (W = 27, M = 21) | 52 (48, 57) | 36 ∗ | 53 (49, 58) | 36 ∗ | 50 (47, 57) | 40 ∗ | .273 |
| 40–49 y (W = 41, M = 31) | 52 (47, 60) | 40 ∗ | 54 (48, 62) | 42 ∗ | 50 (45, 59) | 37 ∗ | .069 |
| 50–59 y (W = 28, M = 18) | 49 (42, 59) | 37 ∗ | 54 (45, 63) | 35 ∗ | 46 (40, 52) | 37 ∗ | .040 |
| ≥60 y (W = 26, M = 25) | 53 (48, 59) | 38 ∗ | 57 (49, 60) | 39 ∗ | 51 (44, 59) | 33 ∗ | .150 |
| 3DAε (%) | |||||||
| All (W = 150, M = 115) | −33 (−36, −31) | −26 | −34 (−31, −36) | −27 | −32 (−35, −30) | −26 | .005 |
| 18–29 y (W = 28, M = 20) | −34 (−35, −31) | −28 | −33 (−35, −31) | −28 | −34 (−36, −32) | −30 | .337 |
| 30–39 y (W = 27, M = 21) | −33 (−35, −31) | −26 | −33 (−35, −31) | −26 | −32 (−35, −31) | −25 | .484 |
| 40–49 y (W = 41, M = 31) | −33 (−36, −31) | −28 | −34 (−37, −31) | −28 | −32 (−36, −31) | −26 | .027 |
| 50–59 y (W = 28, M = 18) | −32 (−36, −28) | −25 | −35 (−37, −28) | −26 | −30 (−33, −28) | −25 | .037 |
| ≥60 y (W = 26, M = 25) | −33 (−36, −31) | −24 | −35 (−36, −32) | −27 | −32 (−36, −31) | −25 | .087 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
