Background
There is a paucity of echocardiographic data describing right ventricular (RV) dimensions and function in the early transitional newborn period.
Methods
Fifty healthy term newborns underwent serial echocardiography at a mean of 15 ± 2 and 35 ± 2 hours of age. RV dimensions and functional indices were measured as recommended in the American Society of Echocardiography’s recent guidelines. Additional novel parameters included RV anteroinferior basal diameter, fractional area change (FAC) from the apical three-chamber view, and speckle-tracking echocardiography–derived peak longitudinal strain in the RV lateral (apical four-chamber view) and inferior (apical three-chamber view) walls. Results obtained at both time points were compared.
Results
Linear dimensions and tissue Doppler velocities were highly reproducible, while time intervals and FAC measurements were more variable. Three-chamber FAC was higher than four-chamber FAC (36 ± 5% vs. 24 ± 7%, P < .001). Lateral wall peak longitudinal strain was similar to the value for the inferior wall (22 ± 4% vs 18 ± 5%, P > .05). A small increase in RV dimensions was noted on day 2 of life (midcavity diameter, 1.71 ± 0.19 vs 1.55 ± 0.19 cm, P < .01; RV anteroinferior basal diameter, 2.24 ± 0.29 vs 2.06 ± 0.24 cm, P < .01; end-diastolic-area in the apical four-chamber view, 4.32 ± 0.64 vs 4.10 ± 0.69 cm 2 , P = .04), while no changes occurred in functional indices. RV dimensions and FAC showed moderate linear correlations with birth weight. Z scores could be computed for the majority of measured indices.
Conclusions
Using conventional and novel indices, the investigators describe a comprehensive echocardiographic protocol for neonatal RV imaging, establish reference ranges, and describe the effect of physiologic postnatal transition on RV dimensions and function. This will facilitate future investigations of RV dysfunction in neonatal cardiopulmonary disorders.
Right ventricular (RV) dysfunction is a strong predictor of outcomes in many adult and pediatric diseases. In infants, RV functional assessment is considered important for children with pulmonary hypertension and different congenital heart defects involving the right ventricle. Despite its importance, the appraisal of RV function in newborns has largely remained qualitative in routine clinical practice. The American Society of Echocardiography recently published comprehensive guidelines for assessment of the right heart in adults and children. The application of these guidelines to the newborn population is hampered by the limited availability of normative data and the lack of information on how the transitional circulation influences these measurements. The transition from fetal to postnatal life is characterized by major circulatory changes, which include a decrease in pulmonary vascular resistance (PVR), an increase in pulmonary blood flow, and closure of fetal shunts (patent ductus arteriosus [PDA] and patent foramen ovale [PFO]). Although these changes continue to occur for many weeks after birth, peak alterations take place in the first few days of life. Although research regarding the use of echocardiography to quantify neonatal RV function is growing, there remains a paucity of comprehensively acquired data, with very few studies looking at the effect of the transitional circulation during the first 48 hours of age. This is crucial because the majority of critically ill newborns present during this period. Furthermore, most RV functional parameters established for use in older children and adults are derived almost exclusively from apical four-chamber views and hence can be influenced by interventricular septal motion, which is often restrictive and variable during the first few days of life.
The primary objective of this study was to test the feasibility and reliability of a comprehensive echocardiographic protocol, including some novel indices, which were independent of septal motion, for quantifying RV dimensions and function in healthy term human neonates. This protocol was used to establish normative data and calculate Z scores for two-dimensional (2D) measurements, Doppler tissue imaging (DTI), and 2D speckle-tracking-echocardiographic parameters. Finally, we investigated the effect of early transitional changes on these measurements by comparing results between days 1 and 2 of life. We hypothesized that in healthy neonates, using a comprehensive approach, it is possible to image the neonatal right ventricle from multiple views and reliably quantify RV dimensions and function. We further hypothesized that the novel parameters established for this study would allow quantification of the size and function of the neonatal right ventricle in a manner that is independent of septal motion.
Methods
Study Design
For this prospective observational study, we recruited 50 healthy term neonates between December 2011 and September 2012 from the Mother and Baby Unit (Mount Sinai Hospital, Toronto, Ontario, Canada). The health records of all infants receiving routine postnatal care were screened, and eligible families were approached within 12 hours of birth. The institutional research ethics board approved the study, and written informed consent was obtained from all parents.
Inclusion and Exclusion Criteria
All healthy, term (gestational age, 37–42 weeks), singleton newborns with normal birth weight (BW) (10th–90th percentiles), born after uncomplicated low risk pregnancies, were eligible for recruitment. Exclusion criteria were maternal diseases (diabetes mellitus, preeclampsia, chorioamnionitis, antenatal diagnosis of placental dysfunction defined by absent or reversed end-diastolic flow in the umbilical arteries on fetal ultrasound, or the prenatal use of antidepressant medications) and any newborn disease (evidence of a perinatal depression defined for this study as umbilical cord pH < 7.0 and/or Apgar score < 5 at 5 min of age, the need for active resuscitation at birth, admission to the neonatal intensive care unit, any congenital malformation, documented episode of neonatal hypoglycemia, or clinical examination suggestive of genetic or cardiovascular abnormalities). We also intended to exclude neonates with incidental findings of congenital heart defects on echocardiography, except those with PFOs, small PDAs, or mild peripheral pulmonary artery stenosis, as these are considered “normal” during the transitional period.
Echocardiographic Methods
On the basis of American Society of Echocardiography guidelines, we developed a comprehensive RV functional protocol for neonates, including measurements derived from multiple views using 2D echocardiography, DTI, and speckle-tracking echocardiography (STE). Each infant underwent two serial evaluations, the first between 12 and 20 hours (D1) and a second between 30 and 40 hours (D2) of life, with a minimum interval of 18 hours between scans. Blood pressure was recorded using a noninvasive oscillometric method (DINAMAP Pro 100; GE Healthcare, Tampa, FL). Echocardiography was performed using a Vivid 7 ultrasound scanner with a 10-MHz transducer (GE Medical Systems, Milwaukee, WI). All 100 studies were digitally stored and analyzed offline, in a random order, using a dedicated workstation (EchoPAC version BT10; GE Medical Systems). Measurements were performed according to published guidelines.
RV Linear Dimensions
The majority of linear dimensions were measured from the RV-focused apical four-chamber (RV-4C) view ( Figure 1 A). Tricuspid valve (TV) annular diameter was defined as a straight line joining the hinge points of the anterior and septal valve leaflets. The basal diameter was measured at the basal third of the RV cavity as the maximal distance from the RV lateral wall to the septum while maintaining a parallel orientation to the TV annulus (TVA). A straight line joining the midpoint of the TVA to the RV apex constituted the RV length. The midcavity diameter was defined by a straight distance between the RV lateral wall and the septum running parallel to the TVA but passing through the mid-point of RV length. In addition, with the aim of quantifying the anteroinferior dimension of the RV cavity, we measured the RV anteroinferior basal diameter (B-PLAX) using the parasternal long-axis RV inflow view ( Figure 1 B). This was measured as the maximal diameter at the basal third of the RV cavity from its anterior to inferior walls while maintaining a parallel orientation with the TVA. All dimensions were measured in end-diastole.
Functional Measurements Obtained from an RV-4C View
Fractional Area Change (FAC)
The four-chamber RV areas at end-diastole and end-systole were calculated by tracing the endocardial borders, including the RV trabeculations within the area ( Figure 2 A). Four-chamber (FAC-4C) (expressed as a percentage) was calculated using the formula [(four-chamber RV area at end-diastole − four-chamber RV area at end-systole)/four-chamber RV area at end-diastole] × 100%.
Tricuspid annular plane systolic excursion (TAPSE) was measured using M-mode echocardiography with the line of interrogation passing through the lateral aspect of the tricuspid annulus while maintaining vertical alignment with the apex ( Figure 2 B).
TV inflow was assessed by placing a pulsed-wave (PW) sample gate of 2 mm at the tip of the TV leaflets during diastole with the Doppler beam parallel to the inflow as visualized using color Doppler echocardiography. Early (E) and late (A) TV inflow velocities and their ratio (E/A) were measured.
Myocardial Velocities
Pulsed DTI of the tricuspid annulus was obtained by placing a PW sample (gate 2 mm) just below the lateral tricuspid annulus. Peak systolic (s′), early diastolic (e′), late diastolic (a′), and peak isovolumetric contraction velocities were measured. On the PW Doppler tracings, we measured time intervals, including closing to opening of the TV (TcOt′), isovolumic relaxation time, duration of systole (S′ = beginning to end of s′ = RV ejection time [RVET′]) and diastole (D′ = beginning of e′ to end of a′) ( Figures 2C and 2D ). The e′/a′, TV E/e′, and S′/D′ ratios and the DTI-based myocardial performance index (MPI′ = [TcOt′ − RVET′]/RVET′) were then calculated.
Peak Longitudinal Strain of the RV Lateral Wall (pLS)
Grayscale images recorded at a frame rate of 80 to 100 frames/sec were analyzed offline using 2D speckle-tracking analysis software (EchoPAC; GE Medical Systems). The RV lateral wall endocardial border was manually traced at end-systole from the lateral basal attachment of the TV to the apex. The width of the region of interest was reduced to the smallest allowed by the software. The software’s automated tracking was visually inspected before accepting the results. The tracing points were manually repositioned in cases of inadequate tracking. A maximum of 10 min was allowed to obtain adequate tracking before the image was deemed nonanalyzable. Four-chamber pLS (pLS-4C) was calculated by averaging peak longitudinal strain of the basal, middle, and apical segments ( Figure 3 ).
The RV Three-Chamber View
From the apical window, an RV-focused apical three-chamber (RV-3C) view was acquired by rotating the transducer counterclockwise from the standard RV-4C view while maintaining a slight rightward tilt to keep the right ventricle in view. The probe was rotated until the left heart was completely out of view, the ascending aorta was in the center of the image, and simultaneous visualization of RV inflow, outflow, and the inferior wall was achieved ( Video 1 ; avialable at www.onlinejase.com ). Precaution was taken to avoid visualizing the anterior wall of the right ventricle. Our aim was to “capture” the maximum RV cavity while keeping these anatomic landmarks in view ( Figure 4 ). Using this view, we measured three-chamber FAC (FAC-3C) and three-chamber pLS for the inferior wall (pLS-3C) as described above. Briefly, FAC-3C (%) = [(three-chamber RV area at end-diastole − three-chamber RV area at end-systole)/three-chamber RV area at end-diastole] × 100%, where RV areas at end-diastole and end-systole were obtained by manually tracing the endocardial borders ( Figure 5 ). For pLS-3C, the inferior wall’s endocardial border was manually traced in end-systole from the lateral basal attachment of the TV to just before the pulmonary valve attachment at the infundibulum ( Video 2 ; available at www.onlinejase.com ). The values for all segments were averaged to obtain an overall pLS-3C for the inferior wall ( Figure 6 ). Similar to pLS-4C, the image was deemed nonanalyzable if adequate tracking was not achieved in 10 min. The results of RV-4C and RV-3C views were then averaged to calculate global FAC and global pLS.
Other Measurements
RVET and pulmonary artery acceleration time were measured from PW Doppler of the main pulmonary artery from the parasternal long-axis view of the RV outflow tract. Interventricular septal motion at end-systole (IVSs) was assessed by visual inspection of the interventricular septum from a 2D short-axis view acquired at the level of the mitral valve. IVSs was considered “flat” if there was complete absence of concavity toward the left ventricle. A comparison was performed between pulmonary artery acceleration time and the presence of flat IVSs, as surrogates of PVR, for the two time points. Presence or absence of PDA and PFO was also recorded. All color and DTI based measurements were averaged from three consecutive cardiac cycles.
Estimation of Reference Values and Z Scores
All echocardiographic measurements were assessed for dependency on BW. Associations with infant length or body surface area were not explored because of concerns regarding the accuracy of length measurement in the newborns. Regression models were empirically tested to optimize the goodness of fit between echocardiographic measurements and BW. Linear ( y = ax + b ), allometric ( y = ax b ), second-order polynomial ( y = ax 2 + bx + c ), and third-order polynomial ( y = ax 3 + bx 2 + cx + d ) models were tested. Selection of the most adequate model for each echocardiographic measurement was based on goodness of fit, on visual inspection of residual values, on fit diagnostic aid plots, and on fit plots of the residual values over the independent variable by linear regression and polynomial regression. Mathematical transformation of the dependent variable was considered only if the distribution of the normalized echocardiographic measurement suggested significant departure from the normal distribution.
Preliminary analysis showed that the variance of the residual values was not always homogenous across the range of BW (i.e., heteroscedasticity). However, weighted regression approaches yielded clearly overadjusted results (not shown). Therefore, weighted regressions were not further explored.
For each echocardiographic measurement, the selected regression model was used to calculate the predicted mean value according to BW. Residual values (observed measurement minus predicted mean) were calculated, and the standard deviation of the residual values was computed. Z scores were then calculated as follows:
Z score = ( observed value − predicted mean ) standard deviation of residual values .
Finally, newly computed Z scores were tested to ensure that residual association with BW (by linear regression and polynomial regression) and significant departure from the standard normal distribution (distribution histograms, box plots, normal probability plots, and Shapiro-Wilk statistic) were not present.
Statistics
Data are presented as mean ± SD or median (interquartile range), unless otherwise stated. Results from D2 scans were compared with those from D1 scans using paired Student’s t tests or Fisher’s exact tests as appropriate. The interobserver measurement reliability for all RV indices was assessed using 20 randomly selected studies. For intraobserver variability, one investigator (A.J.) performed two offline analyses 12 weeks apart to avoid recall bias, while interobserver variability was assessed by a second investigator (A.K.), who was blinded to the measurements of the first investigator. Interobserver agreement was tested using intraclass correlation, presented as intraclass correlation coefficients (ICCs) and 95% confidence interval, and Bland-Altman (BA) analysis, presented as 95% limits of agreement and coefficient of variation (COV). COV was calculated for each parameter using the formula COV (%) = (standard deviation of absolute differences between repeated measurements/arithmetic mean of all repeated measurements) × 100%. This was done to facilitate a more intuitive comparison between reproducibility of different parameters. The lower the COV, the lower the interobserver variability. We accepted P values < .05 as significant. The strength of linear dependency between major functional indices was examined by estimating Pearson or Spearman correlation coefficients as appropriate. The actual P values are shown. Multiple-comparison correction using the Bonferronis method would have required a readjustment of the significance threshold to P < .008.