Initially, 86 KTX patients transplanted in childhood were enrolled between December 2016 and October 2018 in this cross-sectional study. The post-transplant care of the patients is performed by the 1st Department of Pediatrics of Semmelweis University, prospectively referring them to the Heart and Vascular Center of Semmelweis University for an echocardiographic examination. During this visit blood samples were also taken and detailed medical history and demographic data were obtained. We reviewed the etiology of the KTX patients’ primary kidney disease, their history of dialysis, graft source and medication from electronic medical records. Glomerular filtration rate (eGFR) was calculated according to the Schwartz formula or the CKiD U25 formula when analysing patients up to 25 years of age.

Inclusion criteria consisted of having a functioning allograft (no need for dialysis), and no history of cardiac disease. Exclusion criteria were poor echocardiography image quality (n = 9) and an established diagnosis of significant cardiac disease (hypertrophic cardiomyopathy; n = 1, hemodynamically significant aortic regurgitation; n = 2). The final study population consisted of 74 KTX patients [age: 20 (14–26) years, under 18 years: n = 40 patients] transplanted in childhood (age at KTX: 10 ± 4 years). An age- and gender-matched healthy population served as controls (CTR) [n =74, age 20 (13–23) years].

Conventional echocardiographic exams were performed using a GE Vivid E95 (equipped with a 4V-D matrix-array transducer; GE Healthcare, Horten, Norway) or a Phillips Epiq 7G (equipped with an X5-1 matrix-array transducer; Phillips Medical Systems, Best, Netherlands) ultrasound system. A standard acquisition protocol consisting of loops from parasternal, apical, and subxiphoid views was used according to current guidelines (19). LV wall thicknesses and diameters were evaluated in parasternal long-axis view. Relative wall thickness (RWT) was calculated as end-diastolic posterior wall thickness multiplied by 2 and divided by LV end-diastolic internal diameter. Pulsed wave Doppler interrogation at the level of the mitral valve coaptation was obtained to determine early (E) and late diastolic (A) peak LV inflow velocities and their ratio (E/A) and deceleration time (DT). Systolic (s’), early (e′), and late diastolic (a′) velocities at the mitral lateral and medial annuli and the tricuspid free wall annulus were measured using pulsed-wave tissue Doppler imaging. LV filling pressure was estimated by dividing the transmitral E wave with the tissue Doppler imaging-derived averaged lateral and medial annular e’. RV basal diameter was measured as the maximal transverse dimension in the basal third of the RV inflow from the RV-focused apical four-chamber view in the end-diastolic frame. M-mode-derived tricuspid annular plane systolic excursion (TAPSE) was calculated as the maximum longitudinal displacement of the tricuspid annulus. Atrial volumes were estimated using the Simpson method. The left atrial volume index (LAVi) and right atrial volume index (RAVi) were normalized to the body surface area (BSA). Peak pulmonary artery systolic pressure (PASP) was calculated from the tricuspid regurgitant jet signal velocity and right atrial pressure, which was estimated using inferior vena cava diameter and collapsibility (REF).

Beyond the routine echocardiographic protocol, ECG-gated full-volume 3D data sets reconstructed from four or six cardiac cycles optimized for the right or left ventricle were obtained for offline analysis. Image quality was verified at the bedside to avoid “stitching” and “dropout” artefacts of the 3D data. Further measurements were performed on a separate workstation using dedicated software (4D RV-Function 2 and 4D LV-Analysis 3; TomTec Imaging, Unterschleissheim, Germany). The software detects the endocardial surface of the left- and right ventricles, and following manual correction, it traces its motion throughout the cardiac cycle. We determined the end-diastolic volume index (EDVi), end-systolic volume index (ESVi), and stroke volume index (SVi) normalized to BSA, and to characterize global LV and RV functions, EFs were also assessed. By tracing the end-diastolic epicardial contour we measured LV mass index (LVMi) normalized to BSA. The dedicated 3D LV analysis software also enables 3D speckle-tracking analysis, therefore, LVGLS and LVGCS were also calculated. The software automatically measures LV twist and torsion using basal and apical short axis deformation, and systolic dyssynchrony index (SDI) as the standard deviation of the time takes to reach minimum systolic volume for each LV segment.

Our custom 3D RV deformation analysis was previously described in detail (17, 18). All analyses were performed on a standard computer. The input to our software (ReVISION method; Argus Cognitive, Lebanon, New Hampshire) is a series of 3D meshes exported from TomTec 4D RV-Function 2, taken at even time instants during the cardiac cycle.

The analysis begins with the reorientation of the 3D RV mesh. A local coordinate system is defined for each mesh series, where the basis vectors correspond to the longitudinal, radial and anteroposterior directions.

Second, the wall motions of the 3D RV model are decomposed in a vertex-based manner (e.g., for quantifying the magnitude of longitudinal motion, only the movement of the vertices along the vertical axis is taken into account). This decomposition step transforms the original mesh series into several series of meshes, each corresponding to a decomposition type (i.e., which movement directions are enabled and which ones are not).

In the final step, by the decomposition of the model's motion along the three anatomically relevant axes, we can measure the volume change of the right ventricle attributable to the specific direction separately. To assess GCS, we create 15 equidistant circumferential contours on the first mesh of each series and calculate their positions at later time instants. For GLS, 45 longitudes are generated by connecting the apex, the predefined vertices of the base, and vertices in equidistant latitude to the latter. The pulmonary and tricuspid annular planes are excluded from this part of the analysis. Our software calculates 3D RV GCS and GLS by averaging the length changes of each circumferential contour or longitude referenced to their end-diastolic length. Global area strain (GAS) is also calculated by the relative change of the endocardial surface between end-diastole and end-systole. The software also allows the measurement of segmental strain values, therefore, longitudinal, circumferential and area strains of the septum and the free wall were also calculated.

We performed statistical analysis using STATISTICA version 13.4 (TIBCO Software Inc, Palo Alto, CA, United States). The normal distribution of our variables was verified using the Shapiro–Wilk test. Data are presented as mean ± standard deviation (SD), median (interquartile ranges), or percentage, as appropriate. Groups were compared with the unpaired Student's t-test or Mann–Whitney U test for continuous variables and the chi-square or Fisher's exact test for categorical variables according to normality. Correlations between variables were evaluated by Pearson's rank correlation test or non-parametric Spearman correlation. p-values < 0.05 were considered to be statistically significant.

Intra- and interobserver variability of the most relevant parameters were also assessed. The operator of the first measurements (BL) and a second expert reader (ZL), both blinded to the study groups, repeated the measurements in a randomly chosen subset of 5–5 subjects from each group. We calculated Lin's concordance correlation coefficient and coefficient of variation.

The baseline statistics of the KTX and CTR groups are presented in Table 1.

There was no significant difference regarding the age, gender, weight, BSA and heart rate of the two groups. However, the controls were significantly taller than the transplant patients. Furthermore, the KTX group presented with significantly higher systolic blood pressure, while the diastolic blood pressure did not differ.

In the KTX group, 86% (n = 64) received dialysis prior to transplant, 41% of whom (n = 26) was treated with hemodialysis. Preemptive transplant occurred in 14% (n = 10) of the patients. Their mean age at KTX was 10 ± 4 years, and 112 ± 82 months passed since the transplantation.

The underlying primary kidney disease were as follows: kidney agenesis or dysplasia (n = 14), polycystic kidney disease (n = 9), obstructive uropathy (n = 4), focal segmental glomerulosclerosis (n = 14), nephronophthisis (n = 8), chronic glomerulonephritis (n = 7), pyelo- or interstitial nephritis (n = 5), congenital disorders (n = 6), miscellaneous (n = 5) and unknown etiology (n = 2).

The additional clinical characteristics of the transplant patients are shown in Supplementary Table S1.

The conventional 2D echocardiographic parameters of the patients are summarized in Table 2.

LV end-diastolic diameter (LVIDd) did not differ between the transplant and control groups, while LV end-systolic diameter (LVIDs) was significantly lower in the transplant patients. However, both the interventricular septal thickness (IVSd) and the posterior wall thickness (PWd) were shown to be higher in the KTX group and the RWT was also higher in them.

Both the transmitral E wave and A waves were significantly increased in the KTX patients, while the E/A ratio did not differ, and neither did the DT. The mitral lateral s’ and e’ were significantly lower in the transplanted group, on the other hand, the mitral lateral a’ did not show any difference. Accordingly, E/e’ was significantly higher in the KTX group.

Neither the RV basal diameter, the TAPSE nor the PASP showed any difference between the transplanted and the control patients.

Regarding atrial dimensions, LAVi and RAVi did not differ between the two groups.

The 3D and LV speckle tracking data are shown in Table 3.

Both the LVEDVi and LVESVi were significantly increased in the transplant group (Figure 1). However, neither the LVSVi, nor the LVEF showed any statistically significant difference. Nevertheless, the LVMi was significantly increased in the KTX patients.

LVGLS was decreased in the transplant patients, while the LVGCS did not differ (Figures 2, 3). SDI was significantly higher in the KTX patients, and so was the LV twist, however, the LV torsion did not show any difference.

RVEDVi and the RVESVi were both increased in the transplant group (Figure 1), although the RVSVi did not differ, whereas the RVEF was significantly decreased in the KTX group.

The data of the global and segmental 3D RV deformation measures are presented in Table 4.

3D RVGLS was significantly reduced in the KTX patients, while the RVGCS did not differ between the groups (Figures 2, 3). The RVGAS was lower in the transplanted patients.

The septal longitudinal strain (SLS) did not show any difference, however, the septal circumferential (SCS) and area (SAS) strains were significantly reduced in the KTX group. The free wall circumferential (FWCS), longitudinal (FWLS) and area strains (FWAS) did not differ between the groups.

As shown in Table 5, the length of dialysis did not correlate with either 3D LVEF, 3D LVGLS, 3D LVGCS, 3D RVEF, or 3D RVGLS. However, it showed significant, albeit weak to moderate correlations with 3D RVGCS and various segmental 3D RV deformation measures, including SCS, SLS, SAS, FWCS, FWLS and FWAS.

A handful of meaningful laboratory parameters also showed correlation with the 3D echocardiographic parameters. These findings are presented in Supplementary Table S2. The serum calcium levels showed significant relationship with 3D LV twist and torsion, while the serum phosphorus levels correlated significantly with 3D LVEDVi, LVESVi, LVSVi, LVMi and 3D RVSVi, also, the CaxP product correlated well with 3D LVEDVi and LVSVi. Furthermore, the serum magnesium levels showed direct relationship with 3D LVEDVi, LVESVi, LVSVi and LVMi, while also correlating with 3D RVEDVi and RVSVi. The creatinine levels showed significant, albeit weak, correlation with 3D LVESVi, LVMi and LVEF. On the other hand, GFR did not show any relationship with the 3D LV or RV morphological or functional parameters. The urinary protein to creatinine ratio correlated weakly, but statistically significantly with 3D LVESVi, LVMi, LVEF, LVGCS, RVGLS and RVGCS. We found no significant difference in terms of LV or RV morphological and functional measures between the hypertensive and non-hypertensive patients, as presented in Supplementary Table S3. Nevertheless, we are likely not powered to detect a difference by dividing the KTX group to subgroups. When comparing patients who receive steroids as part of the immunosuppressive regime, and patients who do not, our results showed that people on steroids had significantly larger 3D LVEDVi, LVESVi, LVMi, RVEDVi and RVESVi, however, there was no difference in the functional parameters, as shown in Supplementary Table S4.

The intra- and interreader variability analysis of our key 3D parameters demonstrated good agreement in the assessment of these measures (Supplementary Table S5).

Our study has a number of limitations that should be acknowledged for adequate interpretation. Firstly, this is a single-centre, retrospective study with a limited number of cases—further multicentre expansion of the population would strengthen our findings. Secondly, the cross-sectional study design did not allow the assessment of the prognostic power of the 3D parameters. The ReVISION Method is currently not a commercially available software, however, it was previously validated against gold-standard cardiac magnetic resonance imaging. Despite discussing the possible role of FGF-23 in the ventricular remodeling, we did not measure its serum levels in our population. Lastly, cystatin C was not measured in our cohort, therefore, we calculated GFR using the form of CKiD U25 formula which does not implement this laboratory measure.

Using three-dimensional speckle tracking echocardiography, we found distinct morphological and functional changes in both ventricles in pediatric kidney transplant patients. With the ReVISION Method we also measured the changes of right ventricular motion components, which were previously undescribed in this population. However, the pathophysiological background and clinical relevance of our RV strain findings are still unclear and require further research. Nevertheless, our data support the use of a comprehensive echocardiographic protocol applying advanced techniques in the care of kidney transplant patients.