Patients

We prospectively enrolled orthotopic HT recipients and normal control subjects from September 2013 to October 2014 at two centers. The HT patients were recruited only during routine clinic visits or after routine surveillance biopsy, not if they were hospitalized for any acute illness, including symptomatic or suspected rejection episode. Research imaging was performed as part of a routine clinically indicated echocardiographic examination during the visit, and patient follow-up was not prescribed by a specific protocol. The only exclusion criterion was heart failure, defined as LV EF < 50% as measured by the 2D biplane Simpson’s method. Normal control subjects were pediatric patients who underwent echocardiography (for indications such as heart murmur) and had structurally and functionally normal hearts. Control subjects were not age or gender matched. In all subjects, 2DE and 3DE were performed at a single setting. Four sonographers with experience in 3DE acquired all images. Clinical data, including demographics, age, height, and weight, were collected. Body surface area was calculated using the Haycock formula. Electrocardiograms obtained within 3 months of the study in each HT patient were selected from the medical records for the purpose of obtaining the QRS duration.

Seventeen HT patients who returned for clinical follow-up during the time frame of the study had a second set of 2D and 3D echocardiographic data at a median interval of 4.5 months (range, 3.6–8.0 months). Because of the potential for myocardial functional changes in the transplantation population to occur in a relatively short time frame, we elected to include these echocardiograms as de novo examinations in this investigation. Thus, 93 examinations were done in 76 subjects, consisting of 36 HT patients (mean age, 10.3 ± 6.2 years; 21 male; 53 examinations) and 40 normal control subjects (mean age, 14.6 ± 10.6 years; 10 male; 40 examinations). The institutional review boards at the University of Nebraska Medical Center and Children’s Mercy Hospitals and Clinics approved the study protocol and data transfer, and all enrolled patients and control subjects provided written informed consent.

Echocardiography

Echocardiography was performed using an Acuson SC2000 (Siemens Medical Solutions USA, Inc, Mountain View, CA) or an iE33 (Philips Medical Systems, Andover, MA). Images were optimized for gain, compression, depth, and sector width and acquired at frame rates of 50 to 80 frames/sec and 20 to 40 volumes/sec for 2DE and 3DE, respectively. For the Philips system, the X5-1 and X7-2 transducers were used for 3DE, while the S5-1 and S8-3 transducers were used for 2DE, depending on body size. For the Siemens system, the 8V3 and 4V1 transducers were used for 2DE, and the 4Z1 transducer was used for 3DE. Apical four-chamber, two-chamber, and parasternal long-axis and short-axis images were acquired during quiet respiration for 2D echocardiographic analysis. All subjects had normal right ventricular systolic function by qualitative assessment and normal estimated right ventricular systolic pressure by tricuspid regurgitation Doppler (maximal velocity < 2.5 m/sec). Three full-volume 3D echocardiographic data sets of the left ventricle were acquired from the apical four-chamber plane during quiet respiration from three consecutive cardiac cycles (single-beat captures, stitched).

Two-Dimensional and 3D Echocardiographic Analysis

From the M-mode and 2D echocardiographic acquisitions, a single observer (L.L.) measured LV EF offline using the following methods: Teichholz, bullet (5/6 area-length), and biplane Simpson. The same observer analyzed 3D echocardiographic data sets using 4D LV-Analysis 3 software (Image Arena version 4.6; TomTec Imaging Systems, Unterschleissheim, Germany) and measured LV myocardial strain, LV dyssynchrony, LV mass, and LV EF. The 3D echocardiographic full volume was postprocessed by placing points at the LV apex, mitral valve annulus, and aortic valve in a single frame. This was followed by semiautomated endocardial and epicardial tracking in the four-chamber, three-chamber, and two-chamber views in end-systole and end-diastole. Manual corrections were performed to achieve adequate tracking by the 4D LV-Analysis algorithm ( Figure 1 ). Global (longitudinal, circumferential, and radial) and regional strain, EF, mass, and SDI for the left ventricle were calculated. SDI was determined from volume-time curves of the 16 LV segments and expressed as the standard deviation of the heart rate (HR)–corrected time to reach minimal segmental systolic volume ( Figure 2 ). From the regional strain output, longitudinal, circumferential, and radial strain for the interventricular septum (septal strain) was calculated as the average of five segments constituting the interventricular septum ( Figure 3 ). To determine the reproducibility of 3D echocardiographic parameters, a blinded second observer (S.K.) repeated SDI and strain analysis in 10 randomly chosen transplantation studies, and the first observer repeated the analysis after a 3-month interval.

(A) A reconstructed 3D LV model and the results output, which includes end-diastolic volume (EDV), end-systolic volume (ESV), stroke volume (SV), EF, LV mass, global longitudinal strain (GLS), and global circumferential strain (GCS). Analyzed 3D echocardiographic data set (B) shows semiautomated endocardial and epicardial tracking for four-chamber, three-chamber and two-chamber views in end-systole and end-diastole. Green contours show endocardial tracking and blue contours show epicardial tracking.

Volume-time curves of the 16 LV segments in a normal control patient (A) and an HT patient (B) . The colored dots on each line indicate the minimal systolic volume for each segment. (A) represents a synchronous left ventricle in which the minimum volume for the various LV segments occurs at more or less the same time during the cardiac cycle, while in (B) , there is wide scatter in the timing of minimum segmental volume indicating a dyssynchronous left ventricle with high SDI.

Sixteen-segment bull’s-eye maps of 3D strain and 3D strain-time curves of the left ventricle in a normal control (A) and an HT patient (B) . Note abnormal regional strain in five septal segments in the transplantation patient (B) , demonstrated in yellow color on the bull’s-eye map. The bottom panel shows the strain curves for the same data sets. In (A) , the LV segments shorten and reach minimal length at around the same time in systole with respect to timing of aortic valve closure ( vertical line ), thus indicating normal strain. In (B) , the various LV segments shorten at varying rates and reach minimal length at varying times in systole. Note that some of the septal segments lengthen in systole and shorten after end of systole (aortic valve closure).

Statistical Analysis

Comparisons of categorical variables between groups were made using the χ ² method. Unpaired Student’s t tests were used to compare LV SDI and global and regional strain between the HT and control groups. Linear regressions were used to analyze correlations between 2D and 3D echocardiographic measurements. The EF divergence (absolute difference between 3D EF and the average of EF by 2D methods) was calculated. Correlations between EF divergence and SDI and between the various measures of strain and SDI were evaluated. For each regression, the coefficient of determination ( r ² ) was calculated. To assess for potential disproportionate influence of outliers, the regression of EF divergence and SDI was repeated without the two outliers for which SDI exceeded 20%. To assess potential group dependent difference in the correlation of EF divergence with SDI, the regression was repeated using control subjects alone and using HT patients alone. Bland-Altman limit-of-agreement analysis was used for inter- and intraobserver reliability. Intraclass correlation coefficients were calculated to test measurement variability. P values < .05 were considered statistically significant. Statistical analyses were performed using SPSS version 17.0 (IBM, Armonk, NY).