Background
Right ventricular (RV) failure is a major cause of morbidity and mortality in patients with hypoplastic left heart syndrome (HLHS), but the longitudinal course of RV volumes through staged palliation (SP) has not been previously investigated. The aim of this study was to evaluate RV volume and function longitudinally through SP of HLHS using real-time three-dimensional echocardiography.
Methods
A total of 18 subjects with HLHS were prospectively studied at four time points from diagnosis through stage 2 (SP2). Real-time three-dimensional echocardiographic full-volume data sets were acquired in high–frame rate mode with electrocardiographic gating. Volumetric and functional analyses were performed using a semiautomatic contour detection algorithm. Eighteen age-matched and sex-matched normal infants (aged 0–6 months) were studied at comparable time points as controls.
Results
Presurgical examinations (pre–stage 1 [SP1]; n = 18) were performed at a mean age of 4 days, post-SP1 examinations ( n = 17) at a mean age of 20 days, pre-SP2 examinations ( n = 14) at a mean age of 4.6 months, and post-SP2 examinations ( n = 14) at a mean age of 5.5 months, constituting a total of 63 examinations. The mean values of RV end-diastolic volume indexed to body surface area (EDVi) at the four time points were 87 ± 30, 104 ± 39, 112 ± 34, and 102 ± 35 mL/m 2 , respectively. There was an increase in EDVi ( P = .024) from pre-SP1 to post-SP1 but no significant change between post-SP1 and pre-SP2. The decrease in EDVi after SP2 did not reach statistical significance. Mean RV ejection fractions (EFs) were 50 ± 5%, 45 ± 5%, 46 ± 5%, and 38 ± 4%, respectively. There was a trend toward decreasing EF throughout SP, with statistically significant decreases from pre-SP1 to post-SP1 ( P = .003) and from pre-SP2 to post-SP2 ( P < .001). In normal infants, the mean RV EDVi was 50 ± 10 mL/m 2 (approximately half that of patients with HLHS), and the mean EF was 51 ± 3%. There was good interobserver agreement for EDVi, end-systolic volume indexed to body surface area, and EF.
Conclusions
Real-time three-dimensional echocardiography is a reproducible means for evaluating RV volumes and EFs in patients with HLHS. Indexed RV diastolic volume remains stable to slightly increased, and RV EF deteriorates as the first two stages of surgical palliation are accomplished. The findings of this study highlight the adverse physiology of HLHS, which deteriorates even among early survivors despite SP.
Despite advances in surgical techniques and perioperative management, hypoplastic left heart syndrome (HLHS) remains one of the most high-risk congenital heart diseases (CHDs), with only 65% of children born with HLHS reaching the age of 5 years. One of the primary causes of morbidity and mortality among this population is right ventricular (RV) failure after surgical staged palliation (SP). RV dysfunction before SP has been shown to be associated with decreased intermediate-term and overall survival in HLHS. An important contributor to RV failure in HLHS is progressive tricuspid valve regurgitation leading to RV dilation and changes in RV geometry. Serial examinations of RV performance over the course of SP may provide valuable information to guide the medical and surgical management of patients with HLHS. Current knowledge of RV volume and function in HLHS are derived mostly from cross-sectional studies, but the longitudinal course of RV performance through SP has not been previously studied.
The current strategy for surgical management of HLHS incorporates three surgical palliative stages. The first stage (SP1; Norwood) consists of reconstruction of the hypoplastic aorta to augment its size and establish its origin from the pulmonary root, coupled with construction of a source of pulmonary blood flow either as a conduit from the right ventricle or as a systemic–to–pulmonary arterial shunt. The atrial septum is excised to ensure decompression of the left atrium. The result is a circulation with total mixing of saturated and desaturated blood within the right heart, a volume-loaded right ventricle delivering both systemic and pulmonary cardiac output, and a right ventricle afterloaded to the level of the aortic blood pressure. Stage 2 (SP2; Glenn) involves takedown of the SP1 source of pulmonary blood flow when pulmonary vascular resistance permits adequate prograde pulmonary blood flow without ventricular support. A superior vena cava–to–pulmonary artery connection then supplies pulmonary blood flow, and the RV volume overload is thus decreased or eliminated. In stage 3 (Fontan), separation of the systemic venous and pulmonary venous blood is accomplished when the patient has achieved chronic accommodation to SP2 and is large enough to receive a direct connection from the inferior vena cava to the pulmonary arteries adequate for the long term. There remains considerable attrition at and between each stage, with most attrition usually between SP1 and SP2.
Assessments of RV volume and function in patients with HLHS are difficult. The nongeometric shape of the right ventricle hampers assessments from conventional two-dimensional (2D) echocardiographic measures. Conventional measures of ventricular shortening are not applicable to the single right ventricle, because of the fundamental differences in geometry. Changes in RV volume load and geometry between stages render applications of 2D techniques even more challenging. Cardiac magnetic resonance (CMR) imaging is the current reference standard for the quantification of RV volumes and ejection fraction (EF), but there are several limitations for performing CMR imaging in routine clinical practice for neonates and infants. Besides the issues of high cost, low availability, and relative contraindications, performing CMR imaging in these fragile infants is technically difficult and necessitates the administration of general anesthesia. Real-time three-dimensional (3D) echocardiography (RT3DE) is well suited for volumetric and functional assessment of the right ventricle in this setting. There have been significant improvements in both 3D transducer technology (matrix array) and volumetric quantification techniques in recent years. Both transthoracic and transesophageal RT3DE have been validated for the evaluation of RV volumes and function. RT3DE has been shown to provide a fast and reproducible assessment of RV volumes and EF with fair to good accuracy compared with CMR reference data in CHD. Segmental analysis of the right ventricle in patients with CHD is feasible using RT3DE. Moreover, RT3DE has been shown to have good interobserver and intraobserver test-retest reproducibility for the assessment of RV volumes and EF in patients with CHD and healthy controls, which makes it a valuable technique for serial follow-up. Because RV failure is an important cause of morbidity and mortality after SP, an accurate and reproducible technique for serial RV assessment would facilitate the early identification of RV dysfunction and improve risk stratification in HLHS. The aim of our study was to prospectively evaluate RV volumes and function in patients with HLHS using RT3DE through SP.
Methods
This was a prospective longitudinal observational study of patients with HLHS at a tertiary pediatric referral center. The institutional review board at the University of Nebraska Medical Center and the Children’s Hospital and Medical Center approved the study protocol. Informed, written consent was obtained from the legal guardians of all patients who met the inclusion criteria.
Patients
Patients were prospectively enrolled between February 2009 and September 2011. The two anatomic subtypes of HLHS included were aortic atresia/mitral atresia and aortic atresia/mitral stenosis. Variants of HLHS such as unbalanced atrioventricular septal defect, double-outlet right ventricle with mitral atresia and arch hypoplasia, and critical aortic stenosis were excluded. Patients were also excluded if their gestational age was <36 weeks, their birth weights were <2.5 kg, or extracardiac malformations were present. Thirteen of these patients were diagnosed by prenatal echocardiography. Patients diagnosed prenatally were delivered at a high-risk perinatal center, started on prostaglandin E1 infusion, mechanically ventilated when necessary for apnea, and transported for tertiary-level care. All patients underwent SP1 palliation with a right ventricle–to–pulmonary artery conduit (Sano shunt) as the provision for pulmonary blood flow, per our institutional practice. Elective bidirectional cavopulmonary anastomosis (SP2) was performed usually between 4 and 6 months of age on the basis of the treating cardiologist’s clinical analysis, including level of urgency for single ventricular unloading, trends in arterial oxygenation, and somatic growth.
Transthoracic research echocardiography was performed at four time points: the neonatal period as soon as the diagnosis was made in our center (pre-SP1), after the first palliative surgery (post-SP1), before the second palliative surgery (pre-SP2), and after the second palliative surgery (post-SP2). Post-SP1 and post-SP2 echocardiographic studies were performed ≥5 days after each surgery, with the patient breathing spontaneously and without any inotropic support. Seventeen neonates were receiving prostaglandin E1 infusion at the time of the pre-SP1 examination. The pre-SP1 examinations were performed during mechanical ventilation in the six neonates who had to be electively ventilated because of apnea from prostaglandin E1 infusion (before transport to our center) at the time of the study. The ventilated infants were not pharmacologically paralyzed to suppress respiratory effort. Eighteen age-matched and sex-matched infants (aged 0–6 months) with structurally normal hearts were studied at comparable time points to serve as controls.
Three-Dimensional Echocardiography
The iE33 ultrasound system and matrix X7-2 transthoracic transducer (Philips Medical Systems, Andover, MA) were used for the acquisition of real-time 3D echocardiographic research images. Large-sector (full-volume), narrow-sector (Live 3D; Philips Medical Systems), and focused wide-sector (Live 3D zoom) data sets were acquired using the high–frame rate mode (53 ± 7 volumes/sec for the full-volume data sets) by a single experienced sonographer (A.P.) in all patients. The full-volume mode used electrocardiographic gating to merge pyramidal scans obtained over seven consecutive heartbeats. The data sets were obtained from the apical or slightly medial four-chamber, parasternal long-axis, and subxiphoid views in grayscale and with color Doppler. To minimize stitch artifacts, the acquisitions were performed during quiet breathing or with the patient mechanically ventilated, and with minimum transducer movement during acquisition. Four full-volume data sets of the entire right ventricle were obtained in each study. The raw full-volume data sets were transferred from the ultrasound system via CD-ROM or flash drive to a dedicated research workstation for offline analysis using 4D RV Function version 1.2 (Image Arena platform; TomTec Imaging Systems, Unterschleissheim, Germany).
One good-quality full-volume data set in each study (without stitch artifacts) was selected for analysis. The software performed RV border detection over one cardiac cycle using a physics-based modeling algorithm. The algorithm consisted of the construction of an initial RV “cast” and semiautomatic RV contour detection with manual correction options, the details of which have been described elsewhere. Briefly, this four-dimensional RV cast computation method was accomplished in the following steps: (1) Three landmarks were defined for construction of the initial RV cast, and a cut plane that passed through the landmarks was calculated to derive a four-chamber view for RV endocardial border tracing. (2) On the basis of the known 3D coordinates of the four-chamber view, a short-axis view was computed (perpendicular to the four-chamber view), and endocardial border was traced in this view. (3) A coronal view (orthogonal to the short-axis view) was computed on the basis of the tricuspid valve landmark, and the previous two tracings, which was then optimized by the user. (4) A second outer contour of the coronal view was traced between the tricuspid and pulmonary (neoaortic valve) annuli. (5) On the basis of the contours of the three previously traced planes (four-chamber, short-axis, and coronal), the program applied a 3D surface with specific RV properties to create the initial 3D RV cast, which was stored as a triangulated mesh. (6) Finally, using the initial cast surface as a guide, a semiautomatic contour detection algorithm with manual correction options was applied to complete the computation ( Figure 1 ). The contour tracing and cast computation were considered adequate if identification of the anatomic landmarks from the automatically displayed RV volume by the software and manual tracing of the RV endocardial border at end-systole and end-diastole in the three planes were feasible. The full volume 3D echocardiographic data sets were analyzed using this software by a single observer (B.A.G.). Twenty measurements at the four time points were repeated in randomly chosen patients by a second observer (P.G.) to assess interobserver variability.
Statistical Analysis
Continuous data are expressed as mean ± SD and categorical data as proportions. Total counts and percentages are reported for categorical variables. Paired t tests were applied to compare ventricular volumes and EF between palliative stages. The level of statistical significance was set at P < .05. Interobserver variability was tested using the method of Bland and Altman and by calculating the mean percentage error. Mean percentage error was derived as the absolute difference between the two sets of observations, divided by the mean of the observations: [| X 1 − X 2 |/mean( X 1 , X 2 )] × 100. Analysis was performed using Minitab version 16.1 (Minitab Inc., State College, PA).
Results
Between February 2009 and September 2011, 18 patients were enrolled for the study. Patient demographics are shown in Table 1 . Pre-SP1 examinations ( n = 18) were performed at a mean age of 4 days (range, 2–11 days), post-SP1 examinations ( n = 17) at a mean age of 20 days (range, 15–28 days), pre-SP2 examinations ( n = 14) at a mean age of 4.6 months (range, 4.0–5.2 months), and post-SP2 examinations ( n = 14) at a mean age of 5.5 months (range, 5.1–6.0 months), constituting a total of 63 studies. The pre-SP1 echocardiographic studies in six neonates were performed during mechanical ventilation. Fourteen patients completed all four serial examinations, and four patients died. Of the patient deaths, three were during the period between SP1 and SP2, and one preceded SP1.
| Variable | Pre-SP1 | Post-SP1 | Pre-SP2 | Post-SP2 |
|---|---|---|---|---|
| Weight (kg) | 3.3 ± 0.6 (1.95–4.0) | 3.35 ± 0.83 (2–5.6) | 6 ± 0.9 (4.7–7.6) | 6.7 ± 1.1 (5.0–8.5) |
| Height (cm) | 51 ± 3.5 (43–56) | 51 ± 3.2 (45–58) | 62 ± 3.5 (56–69) | 66 ± 4.6 (55–71) |
| BSA (m 2 ) | 0.20 ± 0.02 (0.15–0.23) | 0.21 ± 0.03 (0.15–0.28) | 0.31 ± 0.03 (0.26–0.37) | 0.33 ± 0.04 (0.28–0.39) |
| SpO 2 (%) | 86 ± 10 (64–96) | 78 ± 7.8 (58–87) | 70 ± 8.5 (53–82) | 80.1 ± 3.5 (76–86) |
| FiO 2 (%) | 0.21 | 0.23 ± 0.05 (0.21–0.35) | 0.21 | 0.24 ± 0.07 (0.21–0.4) |
It was feasible to draw contours for RV cast computation in all 63 studies. The mean values of RV end-diastolic volume (EDV) indexed to body surface area (EDVi) at the four time points were 87 ± 30, 104 ± 39, 112 ± 34, and 102 ± 35 mL/m 2 , respectively. There was an increase in EDVi ( P = .024) from pre-SP1 to post-SP1 but no significant change between post-SP1 and pre-SP2 ( P = .302), as shown in Figure 2 . The decrease in EDVi after SP2 did not reach statistical significance ( P = .392). There was a statistically significant increase in the mean RV end-systolic volume (ESV) indexed to body surface area (ESVi) noted through SP ( Figure 3 ). The mean RV EFs at the four time points were 50 ± 5%, 45 ± 5%, 46 ± 5%, and 38 ± 4%, respectively. There was a trend toward decreasing EF throughout SP, as shown in Figure 4 , with a statistically significant decrease from pre-SP1 to post-SP1 ( P = .003) and from pre-SP2 to post-SP2 ( P < .001). Statistical power analysis of the paired t test was performed for comparison of mean EDV, ESV, and EF to examine the ability to detect differences at the levels of confidence specified (sample size = 14, power = 95%, α = 0.05; standard deviations of the differences for EDV, ESV, and EF of 18 ml/m 2 , 14 ml/m 2 , and 4.25%, respectively). These results demonstrate confidence with the ability to accurately detect a change of about 20% in EDV, 25% in ESV, and 10% in EF on the basis of this sample size. There were too few deaths in this series for meaningful statistical analysis. However, at the pre-SP1 examination, two of the four patients who ultimately died had EDVi values > 2 standard deviations above the mean, and one of these also had an EF > 2 standard deviations below the mean.
