Data on myocardial deformation during the internationally widely used second-trimester screening are scarce and confusing. Reference values of time to peak strain are missing. The aims of this study were to assess reference values derived from two-dimensional speckle-tracking echocardiography for global and regional longitudinal right ventricular (RV) and left ventricular (LV) strain, strain rate, and time to peak global strain and to determine the influence of heart rate and gender on these strain parameters.
Seventy-five healthy fetuses were enrolled during second-trimester ultrasound (20–24 weeks). Clips with high frame rates (mean, 132 frames/sec) and two-dimensional (B-mode) grayscale images of apical or basal four-chamber views of both ventricles were used for offline analyses.
There were no statistically significant differences in global strain and strain rate between both ventricles ( P = .679 and P = .734, respectively) or among the RV, septal, and LV free walls. Regional measurements, modeled also as an interaction of wall and segment (basal mid and apical), showed only a small, statistically significant difference between the basal RV and LV free walls. Strain and strain rate values were independent of heart rate. The mean time to peak LV global strain adjusted for heart rate was statistically significantly shorter than the RV value ( P < .0001]). Strain, strain rate, and time to peak global strain were not found to be associated with gender.
The establishment of second-trimester two-dimensional speckle-tracking echocardiographic reference values for global and regional strain, strain rate, and time to peak global strain in a healthy fetal cohort is a mandatory prerequisite for its use in evaluating (pathologic) changes in both ventricular functions during pregnancy.
Quantification of ventricular myocardial function in fetuses with congenital heart disease remains a challenge. Ultrasonography is still the imaging modality of choice for the second-trimester pregnancy evaluation because of its relatively low cost, real-time capability, safety, and operator comfort and experience. Traditional methods such as the one-dimensional M-mode technique (e.g., dimensions and fractional shortening) and two-dimensional (2D) imaging (e.g., ejection fraction), E/A ratio, and myocardial performance index are more often used to evaluate global left ventricular (LV) function. However, these methods are not always applicable to fetal hearts with complex congenital malformations. Moreover, the current indices provide little information about regional alterations in ventricular myocardial contraction during a normal pregnancy.
Two-dimensional speckle-tracking echocardiography (2DSTE) is one of the newer diagnostic, commercially available imaging modalities that allow the assessment of myocardial deformation. Lagrangian strain (%) is defined as the instantaneous lengthening (positive value) or shortening (negative value) related to the initial muscle length, commonly determined by the end of diastole. Strain rate is by definition the temporal derivative of strain (1/sec) and is also equivalent to the shortening velocity per fiber length. Speckle tracking–derived deformation measurements assessed using different ultrasound machines and software packages are not always comparable.
The establishment of normal strain and strain rate values in a healthy fetal cohort is a mandatory prerequisite for its use in evaluating (pathologic) changes in the function of both ventricles during pregnancy. However, the data in this area are scarce and confusing. Strain and strain rate data differ enormously in magnitude and even in their ranges. Published data on both ventricles were not systematically assessed; for example, global right ventricular (RV) and LV strain have been assessed with, but also without, tracking of the interventricular septum (IVS), using different ultrasound machines and software packages (e.g., Velocity Vector Imaging [VVI], Siemens Healthcare, Erlangen, Germany ; Automated Function Imaging, GE Healthcare, Waukesha, WI ) and often using low frame rates. High frame rates have been recently found to improve offline VVI assessment of fetal myocardial function; LV Lagrangian strain was significantly higher when assessed with a high frame rate. Finally, the published correlations of global and regional strain and strain rate data with advancing pregnancy are based on cross-sectional data (e.g., gestational age range, 13–40 weeks). The numbers of subjects during routine second-trimester echocardiography are rather limited.
Detailed knowledge of the normal timing of peak LV systolic deformation and its degree of synchronicity in healthy children and young adults has been recently established. In children with congenital heart disease (e.g., with valvar aortic stenosis), time to peak global systolic strain increased significantly with the severity of the stenosis and decreased shortly after balloon dilatation of the valve. Time to peak global (LV and RV) strain and its degree of synchronicity in healthy fetuses have not yet been studied.
There is a need to use age-specific reference values, assessed using the same ultrasound and software package, for the adequate interpretation of measurements on 2DSTE during abnormal pregnancy. For these reasons, we aimed to (1) assess reference values for global and regional LV and RV myocardial strain, strain rate, and time to peak global strain in the longitudinal direction in a healthy fetal cohort during the internationally widely used second-trimester screening and (2) determine the influence of heart rate, frame rate, and gender on these strain parameters.
Mothers who were referred to the outpatient clinics at three major prenatal diagnostic centers in Israel for routine second-trimester fetal evaluation between August 1, 2010, and May 1, 2011, were invited to enroll in the study. This routine echocardiographic examination was done according to the national obstetric guidelines between 20 and 24 weeks of gestation. Fetuses with structural (congenital) heart defects or failure, abnormal cardiac rhythm, and intrauterine growth retardation were excluded. Other exclusion criteria consisted of monochorionic twins, maternal hypertension, chronic maternal illness, and recent acute illness or poor echocardiographic image quality. Informed consent was obtained from each participant. This study was approved by the institutional committee on human research.
Three physicians with extensive experience in performing fetal echocardiography from tertiary obstetrics (Z.W. in Haifa, S.H. in Tiberias) and pediatric cardiology centers (A.L. in Haifa) participated in the study. A complete transthoracic fetal echocardiographic examination was performed according to the recommendations of the American Society of Echocardiography. Image acquisition was done using a strict protocol. For each of the fetuses, a minimum of three clips of 2D images of three to eight cardiac cycles from the apical or basal four-chamber view of each ventricle were stored in raw data format and provided for offline 2D speckle-tracking echocardiographic analyses. All echocardiographic recordings were made using a Vivid I digital ultrasound scanner (GE Vingmed Ultrasound AS, Horten, Norway) equipped with a 5-MHz linear transducer (available at all three centers). Special attention was taken to achieve a high frame rate: B-mode image depth was reduced and sector width was narrowed to achieve high frame rates (aiming at ≥90 frames/sec). Data were stored at the same frame rate as the acquisition frame rate. Fetal electrocardiograms were not recorded.
Only one experienced pediatric cardiologist (L.K.), who did not take part in the prenatal screenings and was blinded to the studies, performed the 2D speckle-tracking echocardiographic analyses offline. The analyses were done on one workstation equipped with EchoPAC software (GE Vingmed Ultrasound AS).
Analysis of 2D Images
Clips with 2D (B-mode) grayscale images of apical or basal four-chamber views with the RV free wall, the IVS, and the LV free wall were chosen. The original frame rate was displayed on the image. Heart rate could not be displayed because of the inability to record fetal electrophysiologic signals. To identify the beginning and end of one cardiac cycle, the mitral valve was followed in time. End-diastole was defined by the complete closure of the atrioventricular (AV) valve. The time cursor was placed just before the AV (mitral) valve closure (“first beat”) and again just after a consequent second event of AV valve closure (“second beat”). Heart rate could then be calculated from the beat-to-beat time intervals and automatically recorded for each analysis.
Measurements of global and regional longitudinal myocardial strain, strain rate, and time to peak global longitudinal strain were taken in the following way (see also Figure 1 ).
Images of the left ventricle and the right ventricle were analyzed either simultaneously (from the same clip) or separately (from different clips) during the same examination. The reader was required to mark the endocardial border of each ventricle separately, starting from the basal septum through the apex until the basal free wall of the left and right ventricles. The endocardial borders of each ventricle were automatically detected and could be adjusted manually, if necessary. The points with the lowest strain (first and second AV valve closures, respectively) were then chosen as the initial muscle length (and marked as zero in the strain curve). The myocardium shortened in the systolic period and lengthened in the diastolic period and returned to the original length at the end of the diastole period. The LV and RV myocardium were each automatically divided into six equally sized segments (two basal, two mid, and two apical segments per ventricle). The left ventricle included the septum and LV free wall, while the right ventricle included the septum and RV free wall, thus providing identical regional segmentation for both ventricles (readers blinded to the results). Global longitudinal strain, strain rate, and time to peak global longitudinal strain of each ventricle as well as the regional (segmental) strain and strain rate data were automatically recorded ( Figure 1 B). For each subject, all analyses were repeated using three different heart cycles.
Data were analyzed using SAS version 9.1 (SAS Institute Inc., Cary, NC). Statistical significance was defined as a P value ≤ .05.
Study variables are summarized as mean ± SD. Correlations were assessed using Pearson’s correlation coefficients; 90% reference ranges (5th and 95th percentiles) and their respective 95% confidence limits are calculated for global strain, strain rate and time to peak.
Repeated-measures analysis-of-variance models were fitted to the strain, strain rate, and time-to-peak-strain data. Global strain, strain rate, and time to peak strain were modeled as a function of ventricle (left or right). Frame rate and heart rate were modeled in a similar manner. Regional measurements were modeled as a function of wall (LV free wall, RV free wall, septum) and segment (apical, middle, basal) with an interaction term for wall by segment. Model-estimated means were obtained from the models and are presented together with 95% confidence intervals; pairwise comparisons of the model-estimated means are presented if discussed.
To assess intrareader reliability, random samples of 12 cases were read on two separate occasions by the same reader. To assess interreader reliability, random samples of 12 cases were read on two separate occasions by two different readers. Interclass correlation coefficients (ICCs) were calculated from models as described above for global strain, strain rate, and time to peak and used as indices of intrareader and interreader reliability and agreement. The ICCs were interpreted in a similar manner to correlation coefficients: we considered an ICC > 0.80 as excellent, 0.60 ≤ ICC ≤ 0.80 as good, 0.40 ≤ ICC ≤ 0.60 as moderate, and ICC < 0.40 as poor.
A total of 78 mothers at 20 to 24 weeks of pregnancy gave informed consent for this second-trimester echocardiographic study. Eighty-two fetuses were evaluated for inclusion in the study. Of those subjects, seven fetuses (8.5%) were subsequently excluded during the second-trimester study because of fetal movement and incomplete echocardiographic clips (no apical or basal four-chamber view). Finally, 75 fetuses were enrolled in the second-trimester echocardiographic studies (including four bichorionic twins). Indications for echocardiography included positive family history for congenital heart disease ( n = 28), routine second-trimester screening ( n = 25), and ruling out (suspected) cardiac abnormalities ( n = 22). All video clips were recorded with a median frame rate of 132 frames/sec (range, 86–171 frames/sec). Three cardiac cycles were chosen for the offline analysis of each ventricle. Tracking was feasible in all segments of the apical or basal four-chamber view.
Global Longitudinal Myocardial Strain, Strain Rate, and Time to Peak Global Strain
Global longitudinal myocardial strain, strain rate, and time to peak global strain were assessed for the left ventricle (226 trackings from the basal septum through the LV apex to the basal LV free wall) and the right ventricle (224 trackings from the basal septum through the RV apex to the basal RV free wall), separately.
The mean frame rates of the left ventricle and the right ventricle were not found to be statistically significantly different ( F [1, 73] = 0.02, P = .879): mean, 127.70 ± 16.77 versus 127.29 ± 17.08 frames/sec; median, 132 frames/sec (range, 86–171 frames/sec) versus 132 frames/sec (range, 86–171 frames/sec). The mean calculated fetal heart rates of the left and right ventricles were not statistically different ( F [1, 73] = 1.89, P = .174): mean, 150.88 ± 11.07 versus 148.41 ± 11.47 beats/min; median, 151 beats/min (range, 127–187 beats/min) versus 148 beats/min (range, 110–186 beats/min). There was no significant correlation between global strain and strain rate results and heart rate. However, a statistically significant negative correlation was found between time to peak global strain and heart rate (in both ventricles), meaning that a longer time to peak global strain corresponded to a lower heart rate.
Reference values and their respective 95% confidence limits for global longitudinal myocardial strain, strain rate, and time to peak global strain are presented in Table 1 . There were no statistically significant differences in global strain and strain rate between both ventricles ( F [1, 73] = 0.17, P = .6792, and F [1, 73] = 0.12, P = .7342, respectively). On the other hand, the mean time to peak LV global strain adjusted for heart rate was statistically significantly shorter than that of the right ventricle ( F [1, 73] = 17.95, P < .0001). The model estimated means with respective 95% confidence intervals for strain, strain rate, and time to peak global strain in the longitudinal direction are presented in Figure 2 .
|Measurement||Left ventricle||Right ventricle|
|Strain (%)||−18.57 to −33.70||−18.83 to −31.67|
|(−31.1 to −34.3) (−16.0 to −20.5)||(−30.7 to −33.4) (−17.0 to −20.7)|
|Strain rate (1/sec)||−1.87 to −4.07||−1.87 to −4.13|
|(−3.9 to −6.5) (−1.8 to −2.1)||(−3.8 to −4.9) (−1.7 to −2.1)|
|Time to peak global strain (sec)||163.67 − 233.67||174.33 − 258.33|
|(142.3−175.0) (225.3−243.0)||(163.3−190.3) (247.7−280.7)|
Regional Myocardial Strain and Strain Rate
We first assessed whether there were differences between strain and strain rate values in the septum when measured through the right versus the left side. We found no statistically significant differences in strain and strain rate (strain: F [1, 73] = 1.21, P = .2755; strain rate: F [1, 73] = 0.59, P = .4464), and therefore, the data measured from the septum were pooled together and modeled as one. Overall, 900 cardiac segments of healthy fetuses were analyzed (226 from the LV free wall, 450 from the IVS, and 224 from the RV free wall).
There were no statistically significant differences in strain and strain rate among the RV free wall, LV free wall, and septum ( F [2, 148] = 0.74, P = .4770, and F [2, 147] = 0.42, P = .6595, respectively). The 90% reference ranges (5th and 95th percentiles) and their respective 95% confidence limits of these three walls are presented in Table 2 .
|Measurement||LV free wall||IVS||RV free wall|
|Strain (%)||−18.73 to −33.38||−19.54 to −32.52||−18.34 to −32.44|
|(−15.6 to −19.1) (−32.2 to −39.2)||(−18.4 to −20.0) (−31.9 to −34.6)||(−17.6 to −20.1) (−32.0 to −33.5)|
|Strain rate (1/sec)||−2.15 to −5.19||−2.15 to −4.57||−2.13 to −4.55|
|(−2.0 to −2.3) (−4.7 to −5.8)||(−2.1 to −2.4) (−4.3 to −5.0)||(−1.9 to −2.3) (−4.5 to −4.9)|
The wall-by-segment interaction term was not statistically significant for strain rate ( F [4, 296] = 0.45, P = .7724) but was for strain ( F [4, 296] = 4.64, P = .0012). We therefore explored this further with pairwise tests of the model-estimated means. Figure 3 shows the model-estimated mean values with respective 95% confidence intervals (of the three walls as well as three segments per wall). We found statistically significant differences only between the apical and middle segments ( P = .0035) and the apical and basal segments ( P < .0001) mean strain values of the LV free wall, with no significant differences in other segments of the LV free wall or the RV free wall or IVS. Additionally, the basal segment was found to be significantly different between the RV and LV free walls ( P = .0207). No other pairwise comparisons were found to be statistically significant.
The ICCs for intrareader and interreader variance are presented in Table 3 . Most ICCs were interpreted as good or excellent.
|Variable||Global ICC||Segmental ICC|
|Right ventricle||Left ventricle||RV free wall||LV free wall||IVS|
|Time to peak global strain||0.54||0.90||0.32||0.57|