Neonatal congenital tricuspid valve (TV) dysplasia and/or displacement (Ebstein's malformation) with severe tricuspid regurgitation (TR) is a challenging condition in which outcomes are frequently poor. Little is known about left ventricular (LV) function during the perinatal period in patients with congenital TV disease. The objective of this study was to evaluate LV function in fetuses with congenital TV anomalies associated with significant TR.
Serial fetal echocardiograms in 16 fetuses with congenital TV dysplasia and/or displacement (five neonatal survivors and 11 fetal or neonatal deaths) were reviewed. LV stroke volume, LV end-diastolic volume (LVEDV), LV end-diastolic dimension (LVIDd), the LV eccentricity index, thoracic and cardiac areas and the cardiothoracic area ratio (CTAR), the right atrium area index, and LV longitudinal strains were compared according to gestational age and clinical outcome.
The gestational age-adjusted LVEDV (Z-score) was lower in late gestation (−1.2 ± 1.2 at last examination ≥ 28 weeks) than earlier in gestation (0.3 ± 1.5 at last examination < 28 weeks) and LV output was lower than reported late-gestation normal values. LV short-axis dimension correlated with LV volume and CTAR. LV mid-septal strain was lower than the normal average of fetal mid-septal strain and correlated with the LV eccentricity index. Among these parameters, only the LV eccentricity index differed between survivors and non-survivors.
Neonatal congenital tricuspid valve (TV) dysplasia and/or displacement (Ebstein's malformation) with severe tricuspid regurgitation (TR) is frequently a challenging condition with a poor prognosis. We and others have previously observed a high mortality rate among patients diagnosed with severe TR in utero1–3. However, several groups have reported encouraging results among neonates undergoing palliation or repair of severe neonatal Ebstein's malformation4–7. Right ventricular (RV) volume overload and atrialization can cause paradoxical movement of the interventricular septum (IVS) and may compress the left ventricle (LV), potentially compromising LV cardiac output, and some of the more promising surgical strategies for treating this condition aim to facilitate functional recovery of the LV through RV volume reduction8. Little is known about LV function during the perinatal period in patients with congenital TV disease. The objective of this study was to evaluate LV function in fetuses with congenital TV anomalies associated with significant TR.
Fetuses diagnosed with Ebstein's malformation or congenital TV dysplasia who were followed at our center from 2004 to 2010 were reviewed. Fetuses were excluded if the pregnancy was terminated.
Measurements and assessments were performed in a retrospective blinded fashion by a single examiner, using DICOM images which were automatically stored at a frame rate of 30 Hz. For fetuses undergoing ultrasound examination more frequently than monthly, only examinations with relatively good-quality images were selected, with the intent of analyzing studies at roughly monthly intervals. Gray-scale and Doppler echocardiographic measurements included distance from the level of the mitral valve (MV) annulus to the LV apex (LV length) in the four-chamber view at end-diastole, LV end-diastolic dimension at the mid-papillary level in the short-axis view (LVIDd; D1), LV end-diastolic dimension in a plane parallel to the IVS and perpendicular to the short-axis LV dimension (D2), LV and RV width just below the MV annulus at end diastole in the four-chamber view; LV diastolic and systolic volumes using the single plane area-length method in the four-chamber view both at end-diastole (LVEDV) and when the LV area was at a visual minimum (LVESV), aortic valve diameter (AoD) from hinge to hinge in the LV long-axis view during systole, and the flow velocity integral of the pulsed Doppler wave at the aortic valve (VTI Ao) using planimetry of the area beneath the Doppler spectrum. For these measurements, end-diastole was defined as MV closure in the four-chamber view, and as the frame immediately preceding inward movement of the LV posterior wall in the short-axis view. In recording flow velocity spectra, the Doppler cursor was positioned as parallel as possible to the axis of blood flow. Other sonographic measurements included the transverse thoracic cross-sectional area at the level of the four-chamber view, the cross-sectional area of the heart in the four-chamber view; the cross-sectional area of the right atrium (RA) and atrialized RV, the combined cross-sectional area of the functional RV, left atrium (LA) and LV, the presence of fetal hydrops, determined by the presence of ascites, effusions and/or body wall edema, and the presence of antegrade flow across the pulmonary valve confirmed by color Doppler.
For fetuses with a well visualized four-chamber view, LV longitudinal strain was measured using Syngo Velocity Vector Imaging (VVI) software (Siemens Medical Solutions, Mountain View, CA, USA). Images recorded at a frame rate < 30 Hz were excluded from strain analysis. The subendocardial surface of the LV was traced and VVI software automatically calculated strain using the feature-tracking algorithm. Feature-tracking accuracy was visually confirmed, and tracing was corrected until consistent endocardial tracking was obtained. Only one cardiac cycle was used for analysis. Longitudinal negative peak strains were calculated and strains for all segments were averaged to express a global strain.
Parameters analyzed included LV stroke volume, which was calculated as the product of the cross-sectional area of the aorta (3.14 × (AoD/2)2) and VTI Ao and was compared with published data9, the LV ejection fraction (LVEF), calculated as ((LVEDV − LVESV)/LVEDV), the LV eccentricity index, defined as D2/D1, the cardiothoracic area ratio (CTAR), calculated as cardiac area/thoracic area, and the RA area index, calculated as RA area divided by the combined area of functional RV, LA and LV. Strains were compared with normal data from our center based on images stored at 30 Hz. As septal and free wall functional parameters, mid-septal strain and mid-free-wall regional strain were used. The SAS score10 was also calculated. The scoring system for predicting outcome proposed by Andrews et al.10 includes five parameters (CTAR, RA area index, pulmonary valve flow, ductal flow, RV-to-LV width ratio). In this scoring system, each parameter was scored from 0 to 2 (total score, 0 to 10) and fetuses with a score > 5 are predicted to have a poor outcome.
Z-scores of anatomical measurements were calculated on the basis of data from normal fetus studies at our center. All parameters were compared to clinical outcome (death before hospital discharge vs. neonatal survival). Changes in parameters across gestational age were also assessed.
Continuous variables were compared between groups using paired t-test analysis, and categorical variables were compared using the chi-square test. To assess relationships between two continuous parameters, linear regression was performed. Data are presented as mean ± SD. Significance was set at P < 0.05. For intra- and interobserver variability in LV functional parameters, Bland–Altman analysis was conducted. Data from a subset of 10 randomly selected patients were analyzed by two examiners who were blinded to each others' measurements.
The study was conducted with approval from the Committee for Clinical Investigation at Children's Hospital Boston.
Sixteen fetuses were included in the study (Table 1). Between one and 13 serial examinations were performed in each fetus (4.3 ± 2.8). From these, 48 examinations were selected for analysis. Overall, average gestational age at the time of the echocardiogram was 29.3 ± 5.2 weeks; the average gestational age in the first study was 25.4 ± 4.9 weeks, and that in the last study was 34.0 ± 2.5 weeks. Among the 48 examinations selected for analysis, four-chamber views were available for all and short-axis views were available for 34 (71%). For strain analysis, four-chamber images recorded at a frame rate of ≥ 30 Hz were available for 40 out of 48 (83%). Among these 40 examinations, successful analysis was achieved in 21 (53%). Aortic Doppler flow results were available for 38 studies.
Table 1. Clinical profiles of the 16 fetuses studied
Of the 16 fetuses, one died in utero, 10 died during the neonatal period and five survived. Among the 10 neonates who died, four underwent surgery (systemic-to-pulmonary arterial shunt in one, tricuspid valvuloplasty in two, Starnes procedure in one) and six did not. Three of these six neonates died in the delivery room and the others died within a few days after birth and did not undergo surgical intervention despite intensive medical management aimed at stabilization to undergo surgical intervention. Two of the patients who survived the neonatal period underwent a Cone operation later. There were no deaths beyond the newborn period. On average, the SAS score was 3.2 ± 2.7 at the first prenatal echocardiogram performed before 28 weeks of gestation and worsened to 5.7 ± 1.6 at 28 weeks of gestation or later.
Four fetuses, three of which died during the neonatal period, were treated with transplacental medication (digoxin in three and indomethacin in one to promote ductal closure, aiming to reduce pulmonary regurgitation and promote antegrade flow through the RV). Among the nine fetuses with hydrops, one survived beyond the neonatal period and eight did not (P = 0.15). Pulmonary arterial forward flow was recognized prenatally in six fetuses, three of which survived beyond the neonatal period and three of which did not. In four of these fetuses, there was a predominant right-to-left shunting through the ductus arteriosus (two neonatal survivors), but this was no longer present at the final prenatal study in any of them.
Gestational changes in LV anatomy and function
LV stroke volume increased across gestational age. However, in late gestation when no fetus had substantial antegrade RV output and all had left-to-right shunting through the ductus arteriosus, LV stroke volume was lower than reported normal values for combined ventricular (RV + LV) output (CVO) in all but one fetus. In four fetuses that survived beyond the neonatal period, LV stroke volume was larger than the published normal LV output9, while in seven of the 11 fetuses that died before 1 month postpartum (including intrauterine fetal demise), LV stroke volume remained below published normal values (P = 0.11) (Figure 1a). The LVEDV Z-score was lower in late gestation (−1.2 ± 1.2 at ≥ 28 weeks) than earlier in gestation (0.3 ± 1.5 at < 28 weeks) and, on the last measurement, was < 0 in all but two fetuses. The LVEDV Z-score was < − 1 in six of the 11 fetuses that did not survive beyond the neonatal period and > − 1 in four of the six that did (P = 0.59) (Figure 1b). No consistent changes were observed in LVEF, which remained > 40% in most fetuses (Figure 1c). Due to limited availability of images and low feasibility of analysis, gestational changes in global and segmental longitudinal strain were unclear. However, mid-septal longitudinal strains were lower than normal average (14.1 ± 4.1) except for one point of evaluation, as was global strain, with three exceptions (Figure 2).
Relationship between LV anatomy and function and outcome
Both LVIDd and LV length Z-scores correlated modestly with the LVEDV Z-score (Figures 3a and 3b). The LVIDd Z-score correlated negatively with CTAR and positively with LV mid-septal strain (Figures 3c and 3d).
A modest correlation was observed also between the LV eccentricity index and longitudinal mid-septal strain (Figure 4). Figure 5 depicts relationships between global and mid-septal strains and between global and mid-free-wall strains. In fetuses in which global strain was > 15, free wall strain was higher than the normal average mid-free-wall strain (15.9 ± 3.8), and mid-septal strain remained lower than the normal average (14.0 ± 4.0) with one exception, regardless of global strain.
Average values for these anatomical and functional parameters at the first (25.3 ± 4.9 weeks) and final (34.0 ± 2.5 weeks) examinations are summarized according to outcome group in Table 2. At the first examination, no differences were observed with respect to eventual outcome. At the last prenatal examination, there were significant differences in the LV eccentricity index between survivors and patients who died before hospital discharge. No significant differences in these parameters were observed between fetuses with and without hydrops, at either first or last examination.
Table 2. Anatomical and functional parameters at first and last prenatal echocardiograms
Survival to discharge
Data are given as mean ± SD. CTAR, cardiothoracic area ratio; GA, gestational age; LV, left ventricular; LVEDV, left ventricular end-diastolic volume; LVEF, left ventricular ejection fraction; LVIDd, left ventricular end-diastolic dimension; RA, right atrium.
26.6 ± 5.0
22.7 ± 4.0
33.5 ± 2.9
35.2 ± 0.9
0.1 ± 1.7
0.6 ± 1.4
− 1.6 ± 1.1
− 0.2 ± 1.0
LV length Z-score
0.4 ± 1.2
− 0.3 ± 2.1
0.0 ± 1.6
− 0.1 ± 0.6
0.5 ± 1.0
0.3 ± 1.3
− 1.2 ± 1.9
− 1.1 ± 1.2
LV eccentricity index
1.4 ± 0.2
1.5 ± 0.1
1.8 ± 0.3
1.3 ± 0.3
58 ± 10
49 ± 8
53 ± 7
47 ± 14
LV stroke volume (mL)
0.4 ± 0.0
0.9 ± 0.5
3.2 ± 1.3
5.2 ± 2.8
RA area index
0.73 ± 0.55
0.83 ± 0.49
0.93 ± 0.30
0.97 ± 0.19
0.42 ± 0.23
0.41 ± 0.08
0.63 ± 0.12
0.62 ± 0.12
Validation of measurements
Intraobserver variation for strain value was − 0.5 ± 3.3 and interobserver variation was 1.0 ± 2.9. Intra- and interobserver variations in LV end-diastolic volume measurements were 9 ± 9% and 9 ± 16%, respectively, and in LV end-systolic volume were 17 ± 17% and 8 ± 27%, respectively.
As previously reported1–3, we found that severe TR diagnosed during the prenatal period was associated with very poor outcome, with more than half of the patients dying either prenatally or during the neonatal period. The SAS score at the last prenatal ultrasound exam was > 5 on average, indicative of a poor prognosis according to Andrews et al.10. Gestational age-adjusted LV volume in late-gestation fetuses with severe Ebstein's malformation or TV dysplasia was smaller than in mid-gestation fetuses. In general, the small LV volumes were related to relative diminution in the short-axis dimension rather than the long-axis, consistent with geometric changes in the septum related to RV volume overload and atrialization. This logic is supported by the correlation between CTAR and LV short-axis dimension, which suggests that progressive right heart dilation contributes to LV compression in the septum-to-free wall direction and consequent reduction in LV dimension and volume in late gestation.
LV systolic function also appeared to be impaired, based on low septal strain and relatively low global strain. More pronounced abnormalities of septal strain were associated with more prominent distortion of LV geometry. However, due to free wall compensation for septal dysfunction, LVEF remained > 40% in most cases.
Septal dysfunction due to leftward shift of the septum has been reported in adult patients with pulmonary hypertension and in an empirical animal model of LV unloading11, 12. In addition, leftward shift of the septum was reported to cause more transversely oriented oblique fibers and result in reduced LV longitudinal function13. These reports are in line with our finding of correlation between the LV eccentricity index and LV mid-septal strain. Longitudinal function of the LV septum was reduced, possibly due to septal deformation, as suggested by the higher eccentricity index. Moreover, a dilated right heart resulting in a smaller LV short-axis dimension and a higher eccentricity index might also contribute to septal longitudinal systolic dysfunction.
In a study of postoperative adults, Louie et al. reported that RV volume overload was responsible for a decrease in LV septal-to-free wall systolic fractional shortening but not for base-to-apex systolic shortening along the LV long axis14. In that study, all patients were studied after surgical intervention and without contribution of pericardial restraint. In contrast, in our cohort, a dilated right heart could compress and constrain the LV. Thus, the differences between those findings and ours may be related both to differences in LV volume restriction and in methodology (i.e. fractional shortening vs. strain).
Late in gestation, left-to-right shunting through the ductus arteriosus was observed in all cases. Because the LV is providing essentially all cardiac output in such circumstances, it might be expected that LV stroke volume would rise to the normal level of CVO in order to compensate for the non-functioning RV in late gestation. However, in late gestation in most fetuses in this study, LV output remained below the average normal CVO, and among fetuses that did not survive beyond the neonatal period, LV output was generally less than even normal LV stroke volume. This finding indicates that these fetuses were in a low cardiac output condition. Lower LV stroke volume observed during late gestation was partly due to smaller LV volume, as LVEF was unchanged. However, LV septal dysfunction also might play a part in late-gestation low cardiac output in this population.
Considering the relationship between CTAR and LV dimension, and the resultant reduction in LV volume, it is possible that the dilated right heart causes compression of the LV in an anteroposterior direction. The consequent reduction in LV short-axis dimension could restrain LV volume and contribute to septal dysfunction, which might in turn lead to low LV output. RV exclusion in neonates with severe Ebstein's malformation was reported to improve surgical survival4, 6, 15. Moreover, Takagaki et al. reported LV functional recovery after surgery in adults with Ebstein's malformation8.
Given our observations, it was not surprising that patients who survived beyond the neonatal period had a lower eccentricity index, since the LV supplies systemic circulation regardless of RV function and surgical management. Although medical and surgical interventions may have led to changes in LV volume and function, seven out of 11 fetuses that died before discharge did not undergo surgery and three of these seven died just after delivery, it is possible that prenatal LV anatomy and possibly related functional abnormalities are important.
Outcome could not be predicted from early-gestation fetal echocardiograms in this study, which is probably because the adverse physiologic consequences of severe TR in utero are likely to be progressive over the course of gestation2, 3. However, our findings suggested that the LV eccentricity index could be a possible predictor of fetal outcome in later gestation. Although Pavlova et al. reported that fetuses with Ebstein's malformation that developed hydrops had a smaller diameter of the fossa ovalis, and suggested that reduced left-to-right interatrial flow may affect potential LV volume after birth, LV anatomical parameters in this study were not predictive of fetal hydrops16.
There are several important limitations to this study. In addition to the relatively small number of subjects, this was a retrospective analysis of fetal echocardiograms that were not obtained with the intention of performing further analysis, especially for speckle tracking echocardiography. As a result, technical features were not standardized, and the frame rate was not optimal for VVI analysis17. There were a number of factors disadvantageous for tracking, such as thinning or atrialization of the septum, out-of-frame motion of the LV tethered by a dilated RA and RV, and the low frame rate used in many of the echocardiograms. Thus, the frequency of successful tracking was extremely low compared to previous studies using the same software18–21. Due to the small size of the fetal heart, there are spatial limitations to tissue tracking, and, in addition to issues related to low frame rate, the limited spatial discrimination in the small fetal heart could contribute to underestimation of strain18. The method of volume measurement in this study was based on the normal elliptical shape of the LV, which did not necessarily reflect the shape of the LV in this fetal population with RV dilation and hypertension and septal deformation. Furthermore, intra- and interobserver variations of volumetric measurements were relatively large, especially for end-systolic volume.
LV function and anatomy are abnormal in fetuses with congenital TV anomalies associated with severe TR and may be important factors in the outcome of these patients. This is a preliminary study, and investigation is warranted to characterize further the LV functional parameters and their role in outcome in this complex group of patients.