Cardiac Evaluations

Cardiac function was retrospectively evaluated in 59 infants using an Xcelera (Philips Medical Systems, Andover, MA) database of ultrasound images. The first detailed diagnostic echocardiography was performed immediately after the infant was admitted to our unit. The images analyzed in this study were obtained later at the time of preoperative functional assessment, performed at a median age of 1 day (range, 0–12 days) for prenatally diagnosed infants and 2 days (range, 0–8 days) for postnatally diagnosed infants ( P = .30). These images were recorded in accordance with a standardized ultrasound protocol at the time of the preoperative evaluation after stabilization at our institution. Seven infants were excluded from the VVI analysis because their postnatal cardiac ultrasound was missing or of insufficient quality for VVI analysis: one from the prenatal and six from the postnatal diagnosis group ( P = .24). Aortic, pulmonary, mitral, and tricuspid valve dimensions were measured by one of the authors (H.K.M.) according to published recommendations. The severity of tricuspid and pulmonary valve insufficiency was classified as hemodynamically significant (moderate or severe) or insignificant (none, trivial, or mild), taking into account the vena contracta width and the area of color Doppler flow. All infants were divided into three categories according to the morphology of the mitral and aortic valves: mitral and aortic stenosis (MS/AS), mitral stenosis and aortic atresia (MS/AA), and mitral and aortic atresia (MA/AA). For functional analysis, cardiac ultrasound data were transferred in a standard Digital Imaging and Communications in Medicine format to the VVI analysis program (syngo USWP 3.0; Siemens Healthcare). Originally, the images were acquired at a frame rate of 30 to 95 Hz, although digital storage reduces the effective frame rate to 30 Hz. The data were then stored in a research archive with code numbers. All cardiac functional analyses were performed by one of the authors (H.K.M.), who was blinded to the clinical presentation and outcomes of the patients with HLHS.

RV fractional area change (FAC) values, global and segmental myocardial velocities and strain rates, and mechanical synchrony were analyzed from the apical four-chamber view using the VVI technique. Manual tracing of the RV subendocardial surface was performed in a single still frame in midsystole. Tracing began at the edge of the tricuspid valve annulus, extended to the apex of the ventricle without incorporation of the papillary muscle complex, and returned basally to the other edge of the tricuspid valve annulus. Velocity vectors were then automatically calculated for each frame of the cardiac cycle by the VVI algorithm and displayed for the complete loop ( Figure 1 ). The software divided the right ventricle automatically into six segments for regional and synchrony analysis. Tracings were accepted only if the subendocardial border was correctly followed throughout the whole cardiac cycle. If necessary, individual regions of the border were adjusted until the border was correctly tracked for each frame. Left ventricular end-diastolic volume was estimated from the apical four-chamber view using the Auto Left Heart program (syngo USWP 3.0). To minimize intraobserver variability, all measurements were repeated three times, and the mean value was used in the analyses. Mechanical synchrony was measured as the time to peak strain rate and velocity values from the beginning of the QRS complex in all six segments. The degree of mechanical dyssynchrony was quantified as the standard deviation of these values among six different cardiac segments. For intraobserver and interobserver variability, 10 randomly selected studies were analyzed twice by the same investigator (H.K.M.) 4 weeks later and once by another investigator (T.H.O.).

(A) Tracing of velocity vectors during systole. Segmental myocardial velocities (B) and strain rate (C) among the six segments of the right ventricle analyzed ( color coded ) are displayed as a function of time. In the tables, peak longitudinal velocity and strain rate are given for six segments of the right ventricle and the total average value.

Statistical Analysis

Clinical demographics and global cardiac functional data for the study groups were compared using t tests or Mann-Whitney U test for continuous parameters, Cochran-Mantel-Haenszel tests for noncontinuous parameters, and Fisher’s exact tests or Pearson’s χ ² tests for frequencies. Segmental cardiac analyses were evaluated statistically by repeated-measures analysis of variance using the segment as the repeating factor. In case of a significant segment–to–study group interaction, post hoc analyses were performed within segments using Bonferroni-corrected t tests between study groups. Logarithmic transformation was used to normalize skewed distributions for the t tests and repeated-measures analysis of variance models. Correlations ( R values) were calculated using Pearson’s correlation coefficient for normally distributed data and Spearman’s rank correlation coefficient for abnormally distributed data. Intraobserver and interobserver reproducibility was assessed for 10 randomly selected patients for FAC, velocities, and strain rate. Reproducibility was assessed from the corresponding repeated measures using intraclass correlation coefficients with 95% confidence intervals. For global variables, reproducibility was good (intraclass correlation coefficient > 0.83) ( Figure 2 ). A P value < .05 was used to define significance. SPSS for Windows version 19.0 (SPSS, Inc./IBM Corporation, Somers, NY) and SAS version 9.2 (SAS Institute Inc., Cary, NC) were used to perform statistical analyses.

Intraobserver and interobserver reproducibility of FAC, velocities, and strain rate represented as intraclass correlation coefficients (ICCs) and 95% confidence intervals (CIs).