In postnatal life, patients with single ventricle (SV) with morphologic right ventricles have a worse prognosis than those with morphologic left ventricles. The aim of this study was to test the hypotheses that (1) the SV in fetuses with SV has decreased longitudinal strain compared with fetuses with normal cardiac anatomy, and (2) fetuses with SV right ventricular (RV) morphology have decreased strain compared with those with SV left ventricular (LV) morphology.
Fetal echocardiograms with SV RV and SV LV morphology were retrospectively compiled. Postprocessing analysis of the dominant ventricle was done using syngo Velocity Vector Imaging version 2.0. Peak global longitudinal strain (GLS) and global longitudinal strain rate (GL SR) were generated. Both the right and left ventricles were analyzed in fetuses with normal cardiac anatomy for comparison.
Fifty-four fetuses with SV (18 with LV morphology and 36 with RV morphology) were included in the study and compared with 54 controls matched for gestational age. Global longitudinal strain and GL SR were compared between fetuses with SV and normal fetuses and among SV subsets. When all four categories were compared (normal left ventricle, normal right ventricle, SV left ventricle, and SV right ventricle), there was no difference in GLS ( P = .49) or in GL SR ( P = .32) between any of the categories.
Comparable GLS and GL SR values between fetal SV of LV or RV morphology, as well as normal fetal left and right ventricles, reflect in utero preservation of systolic function of the SV heart.
The diagnosis and management of patients with congenital heart disease often begins in fetal life. Fetal assessment often extends beyond an anatomic diagnosis to include an analysis of fetal heart function. Comprehensive evaluation of ventricular function is especially critical in fetuses with single ventricle (SV). A combination of altered preload conditions on the SV, the absence of a second ventricle contributing to ventricular function, and underlying abnormal myocardial fiber orientation may provide the substrate for eventual ventricular failure in patients with SV.
In postnatal life, patients with SV of right ventricular (RV) morphology experience a worse prognosis than those with SV of left ventricular (LV) morphology. A recent report from the Pediatric Heart Network examining 476 Fontan survivors demonstrated that SV RV morphology was associated with worse cardiac function. In the past, evaluation of fetal heart function has been performed with heart failure indices, which, although very useful, remain indirect indicators of function and may not reflect mild abnormalities in function. Therefore, a direct assessment of fetal cardiac function that identifies early and prehydropic changes is needed. Two-dimensional (2D) speckle-tracking has evolved as a new clinical and research tool to quantify regional and global myocardial function. Summary values of strain, global longitudinal strain (GLS) and global longitudinal strain rate (GL SR), have been proposed as clinically useful measures of ventricular systolic function. Studies of systolic strain in fetuses with normal cardiac anatomy have been performed and demonstrated to be feasible. In postnatal life, patients with SV have been shown to have decreased strain compared with the same ventricle in normal patients for both the right ventricle and the left ventricle.
The question central to this investigation is, Do these changes in function, however subtle, begin in fetal life? We hypothesized that in fetal life, (1) differences in GLS and GL SR exist between the normal ventricle and the SV, and (2) the GLS and GL SR of the SV right ventricle are less than those of the SV left ventricle.
All fetal echocardiograms obtained between November 1, 2007, and March 30, 2012, at Lucile Packard Children’s Hospital with diagnoses of SV were retrospectively reviewed from the Siemens syngoDynamics database workstation (Siemens Medical Solutions USA, Inc., syngoDynamics Solutions, Ann Arbor, MI). Exclusion criteria for the study group included the absence of a severely hypoplastic ventricle, moderate or greater atrioventricular valve regurgitation, arrhythmia, multiple gestation pregnancy, and poor image quality. Moderate or greater atrioventricular regurgitation was qualitatively determined on the basis of the vena contracta width, color jet extension into the atrium, and atrial size.
Control studies were selected from the database by reviewing fetal echocardiograms from patients with normal cardiac anatomy during the same time period. Indications for fetal echocardiography in this control group included family history of congenital heart disease or referral for suspected cardiac abnormality. Fetuses referred for maternal antibodies, chromosomal abnormalities, maternal diabetes, multiple gestation pregnancy, increased nuchal thickness, active arrhythmias, hydrops, or any cardiac abnormalities were excluded. Control patients were age matched by selecting for gestational age within 1 week of patients with SV. This retrospective study was approved by the institutional review board at Stanford University.
Echocardiographic Image Acquisition
Studies were obtained using the Siemens Sequoia C512 revision 12.0 (Siemens Medical Solutions USA, Inc., Mountain View, CA) or the Phillips iE33 (Philips Medical Systems, Bothell, WA). Images were acquired as 2-sec clips, per the echocardiography lab protocol. All images were stored digitally and analyzed offline. The four-chamber view, with clear delineation of the atrioventricular valve attachments and a well-defined apex, was required for inclusion in the study. Images with the clearest endocardial border were selected for further analysis. Ventricular morphology was determined by review of recorded images by a study investigator. Right ventricular traits included the presence of coarse muscular trabeculations running parallel to the axis of the ventricle, a moderator band, and an atrioventricular valve with attachments to both the free wall and the septum. Left ventricular traits included an inferior portion with fine trabeculations following oblique angles, as well as an atrioventricular valve with attachments to only the free wall. Variations in anatomic abnormalities in SV preclude the identification of ventricular morphology using classic ventricular traits, such as the level of atrioventricular valve, the number of leaflets of the atrioventricular valve, the presence of inlet and outlet portions of the right ventricle, and septoparietal trabeculations constituting the conal component of the right ventricle.
For 2D strain analysis, the ventricular endocardial border in the four-chamber view was traced. One cardiac cycle was defined manually, with end-diastole defined by atrioventricular valve closure on the 2D image and verified by superimposed M-mode tracings and the velocity curve zero crossover. Endocardial tracing of the SV or normal ventricle was manually performed, with an automated 42-point analysis. Tracing began on the lateral ventricular wall at the atrioventricular valve attachment, extending along the length of the ventricle to the medial attachment of the valve on the interventricular septum. In the case of double-inlet left ventricle, the tracing encompassed the attachment of the atrioventricular valves at either basal edge of the ventricle. Papillary muscles were not included in the tracings.
Two-dimensional GLS and GL SR were obtained from the displacement velocity along the tracked contour of endocardial tracings. The strain curve was displayed using syngo Velocity Vector Imaging version 2.0 (Siemens Medical Solutions USA, Inc.), which provides angle-independent 2D velocity, strain, and strain rate using speckle-tracking. This analysis is independent of ultrasound machine vendor. The strain curves were displayed individually for the six ventricular segments. In studies with ventricular septal defects, the corresponding segment of the interventricular septum was removed from the global calculation. The peak of each instantaneous strain and strain rate curve was averaged and displayed as a separate, average peak value ( Figure 1 ). The average peak longitudinal strain was recorded as GLS and average peak longitudinal strain rate as GL SR.
Two readers (U.T.T., T.A.T.) performed strain analysis on the SV fetuses. Within the Velocity Vector Imaging software program, all previous analyses are saved with tracking results displayed. When the endocardial tracking was deemed suboptimal, the second reader performed a repeat analysis, and the measurement judged to have superior tracking was selected. To avoid bias, a separate reader (H.Y.S.) blinded to the strain measurement for the age-matched subject with SV performed strain analysis on the control patients.
In a subset of 18 patients (16% of the total cohort), GLS and GL SR were independently measured by a third reader (T.A.T.) blinded to the results of the first measurement to assess interobserver variability. The measurements were separated in time by ≥30 days. The SV population was chosen for interobserver assessment because this population is expected to have greater inherent variability in measurement and would provide a conservative estimate of agreement.
Comparisons of GLS and GL SR among the following groups were performed using analysis of variance: normal left ventricle, normal right ventricle, SV left ventricle, and SV right ventricle. The effect of gestational age on GLS and GL SR was analyzed using linear regression. Interobserver variability was assessed by calculating absolute mean percentage difference between readers and the intraclass correlation coefficient. All data analyses were performed using Stata version 11 (StataCorp LP, College Station, TX).
Sixty-one SV studies were initially reviewed, and of these, 54 echocardiographic studies with SV were analyzed: 18 with LV morphology and 36 with RV morphology ( Table 1 ). The remaining seven studies were eliminated for poor image quality or inadequate views, yielding a feasibility rate of 88% in the study group. A frame rate of 30 frames/sec was used, the standard compression performed by digital storage of Digital Imaging and Communications in Medicine clips by the syngoDynamics workstation. Gestational age ranged from 15 to 38 weeks, with a mean of 28.2 ± 6.3 weeks. The control group was composed of 54 age-matched controls, with a mean gestational age of 28.1 ± 6.3 weeks.
|Single left ventricle||Single right ventricle|
|Lesion||n||Mean gestational age (weeks)||Lesion||n||Mean gestational age (weeks)|
|Tricuspid atresia||9||26.3||Hypoplastic left heart syndrome||33||28.8|
|Pulmonary atresia with intact ventricular septum||6||27.8||Unbalanced atrioventricular canal (right dominant)||2||24.4|
|Double-inlet left ventricle||2||28.0||Double-outlet right ventricle with mitral atresia||1||20.6|
|Unbalanced atrioventricular canal (left dominant)||1||33.0|
There was no significant difference in either GLS or GL SR between any comparison of normal left ventricle, normal right ventricle, SV left ventricle, and SV right ventricle ( Figure 2 ). The data for these measures are summarized in Table 2 .
|Ventricle type||n||GLS (%) ∗||GL SR (sec −1 ) †|
|Normal left ventricle||54||−15.6 ± 3.3||−1.5 ± 0.4|
|Single left ventricle||18||−17.1 ± 5.0||−1.6 ± 0.5|
|Normal right ventricle||54||−16.5 ± 3.9||−1.6 ± 0.4|
|Single right ventricle||36||−16.2 ± 4.8||−1.4 ± 0.5|
Interobserver variability analysis revealed that the absolute mean percentage difference between readers was high, 23% for GLS and 21% for GL SR, and the intraclass correlation coefficient was low for both, 0.39 for GLS and 0.50 for GL SR.
The effect of gestational age on LV GLS, RV GLS, LV GL SR, and RV GL SR was evaluated for all fetuses. Gestational age had a significant but weak relationship in all comparisons ( Figure 3 ).