Left Ventricular 3‐Dimensional Global Longitudinal Strain Predicts All‐Cause Mortality in Patients With Heart Transplant Authors: Shuangshuang Zhu, Chun Wu, Yiwei Zhang, Weihua Qiao, Nianguo Dong, Yuman Li, Mingxing Xie, Li Zhang Published: 2024 ABSTRACT BackgroundThe prognostic value of 3‐dimensional (3D)‐left ventricular global longitudinal strain (LVGLS) in recipients of heart transplant (HT) reremains unknown. This study aimed to determine whether 3D‐LVGLS was the more powerful predictor of poor outcomes in recipients of HT compared with 2‐dimensional (2D)‐LVGLS.Methods and ResultsAll consecutive adult patients who received HT and underwent at least 1 comprehensive 2D and 3D transthoracic echocardiographic examination for clinical surveillance were retrospectively enrolled. The end point was all‐cause mortality. Prognostic model performance was assessed according to the C‐statistic. The 3D‐LVGLS measurements were feasible in 294 of 342 patients with HT (86%). The median follow‐up time was 53 months, and 44 HT redied. Receiver operating characteristic curves revealed that the area under the curve for predicting all‐cause mortality was greater for 3D‐LVGLS than 2D‐LVGLS (0.77 versus 0.67, P=0.012). When HT stratified patients with HT into tertiles according to 3D‐LVGLS values, patients with lower 3D‐LVGLS had worse outcome (P<0.001). The multivariable Cox analysis showed that the model with 3D‐LVGLS (hazard ratio [HR],1.44 [95% CI,1.24–1.68]; P<0.001; C‐statistic=0.814) was better in predicting death than the model with 2D‐LVGLS (HR, 1.19 [95% CI, 1.06–1.32]; P=0.002; C‐statistic=0.772). The best cutoff value of 3D‐LVGLS for detecting all‐cause mortality was −16.1%, with a sensitivity of 63.6% and a specificity of 84.0%.ConclusionsThe 3D‐LVGLS was a powerful predictor of all‐cause mortality in patients receiving HT and provided greater prognostic value than 2D‐LVGLS. Our study highlighted the potential of evaluating 3D‐LVGLS for risk stratification in recipients of HT. Nonstandard Abbreviations and Acronyms 2‐dimensional 3‐dimensional acute rejection cardiac allograft vasculopathy left ventricular global longitudinal strain heart transplant speckle tracking echocardiography tricuspid annular plane systolic excursion Clinical PerspectiveWhat Is New? Three‐dimensional left ventricular global longitudinal strain measurement offers a feasible method for graft function evaluation.Heart transplant patients with lower 3‐dimensional left ventricular global longitudinal strain have worse outcome.Three‐dimensional left ventricular global longitudinal strain is a robust predictor of all‐cause mortality in heart transplant patients and provides greater prognostic value than 2‐dimensional left ventricular global longitudinal strain. What Are the Clinical Implications? This study might be the largest scale contemporary single center cohort to investigate the associations between 3‐dimensional left ventricular global longitudinal strain and long‐term prognosis of heart transplant patients.This study suggests that 3‐dimensional left ventricular global longitudinal strain may help clinicians to identify higher risk heart transplant patients. Clinical Perspective What Is New? Three‐dimensional left ventricular global longitudinal strain measurement offers a feasible method for graft function evaluation.Heart transplant patients with lower 3‐dimensional left ventricular global longitudinal strain have worse outcome.Three‐dimensional left ventricular global longitudinal strain is a robust predictor of all‐cause mortality in heart transplant patients and provides greater prognostic value than 2‐dimensional left ventricular global longitudinal strain. Three‐dimensional left ventricular global longitudinal strain measurement offers a feasible method for graft function evaluation. Three‐dimensional left ventricular global longitudinal strain measurement offers a feasible method for graft function evaluation. Heart transplant patients with lower 3‐dimensional left ventricular global longitudinal strain have worse outcome. Heart transplant patients with lower 3‐dimensional left ventricular global longitudinal strain have worse outcome. Three‐dimensional left ventricular global longitudinal strain is a robust predictor of all‐cause mortality in heart transplant patients and provides greater prognostic value than 2‐dimensional left ventricular global longitudinal strain. Three‐dimensional left ventricular global longitudinal strain is a robust predictor of all‐cause mortality in heart transplant patients and provides greater prognostic value than 2‐dimensional left ventricular global longitudinal strain. What Are the Clinical Implications? This study might be the largest scale contemporary single center cohort to investigate the associations between 3‐dimensional left ventricular global longitudinal strain and long‐term prognosis of heart transplant patients.This study suggests that 3‐dimensional left ventricular global longitudinal strain may help clinicians to identify higher risk heart transplant patients. This study might be the largest scale contemporary single center cohort to investigate the associations between 3‐dimensional left ventricular global longitudinal strain and long‐term prognosis of heart transplant patients. This study might be the largest scale contemporary single center cohort to investigate the associations between 3‐dimensional left ventricular global longitudinal strain and long‐term prognosis of heart transplant patients. This study suggests that 3‐dimensional left ventricular global longitudinal strain may help clinicians to identify higher risk heart transplant patients. This study suggests that 3‐dimensional left ventricular global longitudinal strain may help clinicians to identify higher risk heart transplant patients. Although survival rate of patients receiving heart transplant (HT) has improved significantly, graft failure remains the main cause threatening the long‐term prognosis. 1 , 2 Related complications after HT, such as acute rejection (AR) and cardiac allograft vasculopathy occur frequently and clinically asymptomatic. 3 , 4 These complications could adversely affect the myocardial function in various ways, ultimately leading to death and retransplantation. 5 Therefore, it is crucial to accurately assess the graft function and identify patients with HT who are at higher risk of poor outcomes. Left ventricular function in patients with HT is mainly evaluated by echocardiography during follow‐up. 3 Left ventricular ejection fraction (LVEF) is the most routinely used echocardiographic parameter, but it fails to monitor early abnormalities of graft function. 6 Recently, some studies have advocated that LV global longitudinal strain (LVGLS) derived from 2‐dimensional (2D) speckle tracking echocardiography (STE) can detect early changes in ventricular function and has been shown to have prognostic value for recipients of HT. 6 , 7 , 8 , 9 , 10 However, 2D‐LVGLS measurement is limited by the geometric assumptions, apical foreshortening, and out‐of‐plane motion of the speckles. In addition, due to the loss of support and hold provided by the pericardial sac, as well as the disproportion between the large size of the recipient's mediastinal cavity and the small size of the donor's heart, most of the transplanted hearts exhibited significant translational motion, 11 which may aggravate the “out‐of‐plane phenomenon.” Three‐dimensional (3D)‐LVGLS is a novel imaging parameter that allows an accurate assessment of LV function by overcoming the aforementioned limitations. 12 Moreover, Sun et al validated that 3D‐LVGLS showed excellent accuracy in evaluating ventricular function of transplanted hearts compared with cardiac magnetic resonance. 13 However, no study has investigated the prognostic value of 3D‐LVGLS in recipients of HT. A direct comparison between 2D‐ and 3D‐LVGLS for predicting adverse clinical outcomes in recipients of HT is scarce. Accordingly, the aims of our study were (1) to evaluate the feasibility of 3D‐LVGLS in assessing LV systolic function in a large consecutive cohort of recipients of HT; (2) to determine whether 3D‐LVGLS is the powerful predictor of adverse outcomes and to further compare the prognostic value of 3D‐LVGLS with that of 2D‐LVGLS. METHODS The data underlying this article will be shared upon reasonable request to the corresponding author. Study Population We retrospectively enrolled all consecutive adult patients who underwent HT at our center (Wuhan Union Hospital, China) between January 1, 2015, and December 31, 2019. All the recipients of HT underwent at least 1 comprehensive transthoracic echocardiographic examination during their routine follow‐up. For recipients with multiple echocardiographic examinations, the earliest one after discharge was prioritized for inclusion. Exclusion criteria were as follows: (1) combined transplantation; (2) retransplantation; (3) LVEF<50%; (4) nonsinus rhythm; (5) poor sound window quality failed to obtain 3D images; and (6) inadequate myocardial tracking by 3D‐STE. Ethics Statement This study was approved by the Ethics Committee of Wuhan Union Hospital, Tongji Medical College, Huazhong University of Science and Technology (IORG NO: 20210670) and was performed in accordance with the Declaration of Helsinki. All donor hearts were donated to the Red Cross Society of Hubei Province and were allocated by the China Organ Transplant Response System. We strictly abide by Chinese laws, which have been committed to the use of voluntary donation after brain death for HT since 1 January 2015. Furthermore, we obtained written informed consent from all participants. This study was registered in the Chinese Clinical Trial Registry (ChiCTR2100047545). Echocardiography All subjects underwent comprehensive transthoracic 2D and 3D echocardiographic examinations using commercially available systems (Philips iE 33; Philips Medical Systems, Andover, MA) with S5‐1 and X5‐1 transducer. Patients were placed in the left lateral decubitus position with simultaneous electrocardiographic recording. During at least a 5‐second end‐expiratory breath hold, standard 2D echocardiography and 3D full‐volume imaging were obtained. All echocardiographic images were recorded in a native Digital Imaging and Communications in Medicine format. Conventional echocardiographic measurements included mainly atrial and ventricular diameters, wall thickness, LV mass, mitral inflow propagation, and mitral annular relaxation velocities. Tricuspid annular plane systolic excursion, right ventricular (RV) fractional area change, and tricuspid lateral annular systolic velocity (S′) were used to evaluate RV function. LV volumes (end‐diastolic and end‐systolic volumes) and LVEF were calculated using the biplane Simpson's method in the apical 2‐ and 4‐chamber views. 14 Two investigators (S.S.Z. and C.W.) were involved in image postprocessing and analysis. All investigators were masked to the echocardiographic results and clinical information of the patients with HT. Two‐Dimensional Speckle‐Tracking Echocardiography In this study, 2D‐STE was performed by C.W. using commercially available AutoStrain software (Qlab13, Philips Healthcare, Andover, MA). The 2D‐LVGLS was calculated by averaging the values obtained in the apical 4‐chamber, 3‐chamber, and 2‐chamber views. The 2D‐LVGLS was obtained from the standard 2D grayscale image at frame rate of 50 ∼ 70 frames/s. The LV endocardial border was automatically traced at end‐diastole. Subsequently, the software automatically tracked the endocardial layer throughout the cardiac cycle and the observer could manually adjust the endocardial border if necessary. The 2D‐LVGLS was calculated as the average peak systolic longitudinal strain of all 16 LV myocardial segments. Three‐Dimensional Speckle‐Tracking Echocardiography Four heartbeat 3D full‐volume images were acquired. The frame rates of the volumetric images ranged from 19 to 23 MHz. After selecting full‐volume LV images with the highest quality, the investigator (S.S.Z) began 3D‐STE analysis using vendor‐independent 3D speckle‐tracking software (4D LV‐Analysis, version 3.1, TomTec Imaging Systems, Unterschleissheim, Germany). In the apical 4‐chamber, 3‐chamber, and 2‐chamber views, the largest apical long‐axis dimensions were set by the point of the LV apex and the center of the mitral annular line at the end‐diastolic frame. Subsequently, the software performed the contour tracking of the LV endocardium throughout the cardiac cycle. The papillary muscles were excluded from the endocardium by visual judgments. For 3D strain analysis, LV was automatically divided into 16 segments using standard segmentation schemes. The software provided averaged longitudinal strain‐time curves, and the peak LVGLS was determined. Manual adjustment was performed when tracking was not satisfactory. The patient with >3 segments that could not be analyzed was excluded from the analysis. Reproducibility Intraobserver and interobserver variabilities of 3D‐LVGLS and 2D‐LVGLS were assessed in 30 randomly selected subjects with HT. To evaluate the intraobserver variability, the first observer repeated the analysis after 8 weeks and was blinded to the results of the first measurements. To determine the interobserver variability, a second observer who was blinded to the first observer's measurements repeated the analysis. Outcomes Patients were followed until January 31, 2023. Follow‐up data were collected through hospital visit or telephone contacts by the investigator who was blinded to clinical details and echocardiographic data. The end point was all‐cause mortality. Statistical Analysis Statistical analyses were performed with SPSS version 25.0 (SPSS Inc., Chicago, IL), MedCalc version 19.0.4 (MedCalc Software, Ostend, Belgium), and R version 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria). A 2‐tailed P<0.05 was considered as statistically significant. Categorical data are presented as numbers and percentages and were compared using the chi‐square test or the Fisher exact test. Continuous variables are presented as mean±SD or median with interquartile range and were compared using t test or Wilcoxon‐Mann–Whitney U test. Correlation analysis was performed using Pearson's correlation coefficient. Receiver operating characteristic curves were used to determine the best cutoff value (maximum Youden index) for predicting all‐cause mortality. Kaplan–Meier survival curves were constructed and compared using the log‐rank test. Univariate and multivariable Cox regression analysis were used to determine the predictors of poor outcomes. To avoid overfitting and collinearity issues, 2 separate multivariable Cox proportional hazards models were constructed to determine the statistically significant factors of all‐cause mortality. Model performance was assessed according to the C‐statistic. The DeLong test is used to compare the C‐statistics between models. Hosmer–Lemeshow good of fit test was used to evaluate the model's calibration. The reproducibility of 3D‐LVGLS and 2D‐LVGLS was assessed using intraclass correlation coefficients and Bland–Altman analyses. RESULTS Feasibility of 3D‐LVGLS in the Evaluation of Graft Function in Patients With HT First, 15 patients with HT were excluded, including 6 cases of heart‐lung transplantation, 2 cases of heart‐kidney transplantation, 1 case of retransplantation, 2 cases with LVEF <50%, and 4 patients with nonsinus rhythm. Of the remaining 342 patients, 25 participants who could not obtain 3D images owing to poor sound window quality and 23 patients with inadequate myocardial tracking were excluded. Finally, 294 patients were included in this study (Figure 1). There were no significant differences in clinical characteristics between the main cohort and patients excluded due to inadequate tracking and poor acoustic window (Table S1). The 3D‐LVGLS measurements were feasible in 294 of 342 patients with HT (86%). Flow chart of study population. HT indicates heart transplant; and LVEF, left ventricular ejection fraction. Clinical Outcome The median follow‐up time was 53 (38–59) months, and death occurred in 44 patients with HT. The causes of death included graft failure (18%), AR (14%), cardiac allograft vasculopathy (21%), serious infection (14.0%), cancers (7%), aortic dissection (2%), cardiogenic shock (11%), cerebrovascular accident (2%), and sudden death of unknown cause (11%) (Table 1). Table 1Causes of Death After Heart TransplantationCause of deathNo. (%)Graft failure8 (18)Acute rejection6 (14)Cardiac allograft vasculopathy9 (21)Severe infection6 (14)Cancers3 (7)Aortic dissection1 (2)Cardiogenic shock5 (11)Cerebrovascular accident1 (2)Sudden death of unknown cause5 (11)Total44 (100) Causes of Death After Heart Transplantation Table 2 presents the clinical characteristics and laboratory findings between the death and survivor groups. Patients who died during follow‐up were significantly older, had lower body mass index and hemoglobin, and higher BNP (B‐type natriuretic peptide) levels. The prevalence of rejection episodes was higher in the death group. There were no significant differences in sex; incidence of hypertension, diabetes, and chronic kidney disease; cause of transplantation; donor characteristics; medication use; and other laboratory findings between the 2 groups. Table 2Clinical Features of Patients With HT According to All‐Cause MortalityVariableAll recipients (n=294)Dead (n=44)Alive (n=250) P valueDemographicsRecipient age, y48±1253±1147±120.008Male sex, n (%)234 (79.6)36 (81.8)198 (79.2)0.691Body mass index, kg/m2 23.2±3.721.9±2.923.4±3.70.013Cause for transplant0.617Ischemic cardiomyopathy, n (%)51 (17.3)8 (18.2)43 (17.2)Dilated cardiomyopathy, n (%)166 (56.5)22 (50.0)144 (57.6)Hypertrophic cardiomyopathy, n (%)13 (4.4)1 (2.3)12 (4.8)Restrictive cardiomyopathy, n (%)10 (3.4)3 (6.8)7 (2.8)Valvular heart disease, n (%)35 (11.9)6 (13.6)29 (11.6)Congenital heart disease, n (%)13 (4.4)2 (4.5)11 (4.4)Other diseases, n (%)6 (2.0)2 (4.5)4 (1.6)Donor age, y35±1136±1135±110.752Donor weight, kg63±1064±1263±100.886Ischemic time of donor heart, min355 (302–400)360 (319–382)152 (300–403)0.991ComorbiditiesHypertension, n (%)125 (42.5)15 (34.1)110 (44.0)0.220Diabetes, n (%)151 (51.4)28 (63.6)123 (49.2)0.077Chronic kidney disease (estimated glomerular filtration rate≤60 mL/min per 1.73 m2), n (%)89 (30.3)14 (31.8)75 (30.0)0.809History of acute rejection ≥2R or treated rejection, n (%)27 (9.2)8 (18.2)19 (7.6)0.025Immunosuppressant medicationsTacrolimus, n (%)281 (95.6)40 (90.9)241 (96.8)0.070Mycophenolate mofetil, n (%)280 (95.2)41 (93.2)239 (96.0)0.405Prednisone, n (%)267 (90.8)38 (86.4)229 (92.0)0.228Cardiac medicationsAspirin, n (%)105 (35.7)13 (29.5)92 (36.9)0.345Statin, n (%)167 (56.8)20 (45.5)147 (59.0)0.093Angiotensin‐converting enzyme inhibitor/angiotensin receptor blocker, n (%)92 (31.3)10 (22.7)82 (32.9)0.179Beta‐blocker, n (%)164 (55.8)24 (54.5)140 (56.2)0.836Calcium channel blocker, n (%)140 (47.6)15 (34.1)125 (50.2)0.071Laboratory findingsHemoglobin, g/L132±21119±23134±20<0.001Alanine aminotransferase, U/L19 (13–27)19 (13–35)18 (13–27)0.818Aspartate transaminase, U/L21 (17–26)21 (16–27)21 (17–25)0.895Fasting glucose, mmol/L5.9±1.66.1±1.95.8±1.60.253Creatinine, μmol/L99 (83–119)109 (82–129)97 (83–116)0.306Brain natriuretic peptide, pg/mL87 (54–147)150 (68–317)86 (52–135)<0.001Values are shown as mean±SD, median (interquartile range), or number (percentage). HT indicates heart transplant. Clinical Features of Patients With HT According to All‐Cause Mortality Values are shown as mean±SD, median (interquartile range), or number (percentage). HT indicates heart transplant. The echocardiographic characteristics are described in Table 3. The median time from cardiac transplantation to echocardiographic examinations was 350 (218–396) days. Compared with survivors, patients with HT who died had higher heart rate, lower RV fractional area change, tricuspid annular plane systolic excursion, S′, and LVEF. In addition, nonsurvivors displayed significantly impaired 2D‐LVGLS (− 16.7%±3.4% versus −18.4%±2.3%, P<0.001) and 3D‐LVGLS (−15.8%±2.2% versus −17.9%±2.0%, P<0.001) compared with survivors. Table 3Echocardiographic Features Between Death and Survivor GroupsVariableAll recipients (n=294)Dead (n=44)Alive (n=250) P valueHeart rate, bpm91±994±991±90.012Left atrial dimension, mm47.6±8.746.0±8.847.9±8.70.180Interventricular septal thickness at end‐diastole, mm10.0±1.810.0±1.710.0±1.80.960LV at dimension end‐diastole, mm42.4±4.342.6±4.942.4±4.30.860Posterior wall dimension at end‐diastole, mm10.3±3.010.1±1.710.3±3.20.778LV mass, g/m2 141 (117–161)141 (111–163)141 (118–161)0.933 E, cm/s83.4±20.384.5±19.283.3±20.50.711 A, cm/s48.1±11.348.9±11.148.0±11.40.671 E/A ratio1.77±0.411.76±0.381.77±0.420.911Deceleration time, ms192±47200±56191±450.870Septal e′, cm/s8.4±2.18.3±2.18.4±2.10.769Lateral e′, cm/s13.9±3.213.8±3.113.9±3.20.830Average E/e′ ratio7.8±2.58.0±2.67.8±2.50.600Right atrial dimension, mm37.6±5.337.6±6.137.7±5.20.912RV dimension, mm34.0±3.634.0±4.034.0±3.60.934RV end‐diastolic area, cm2 19.5 (15.6–23.0)22.1 (17.3–28.4)20.3 (17.6–24.6)0.851RV end‐systolic area, cm2 10.8 (8.7–12.7)10.1 (8.1–13.5)10.8 (8.9–12.6)0.248RV fractional area change, %45.2±5.643.0±6.845.9±5.30.004Tricuspid annular plane systolic excursion, mm15.1±3.613.7±3.815.4±3.50.008 S′, cm/s10.5±2.49.7±2.210.6±2.40.023LV EDV, mL68.0 (58.7–85.4)66.8 (58.6–86.7)68.2 (58.7–85.1)0.228LV ESV, mL25.2 (20.1–30.5)24.5 (20.8–32.3)25.2 (20.0–29.6)0.080LV ejection fraction, %64.0±4.362.6±4.764.2±4.20.021Speckle tracking2D‐LVGLS (%)−18.1±2.6−16.7±3.4−18.4±2.3<0.0013D‐LVGLS (%)−17.6±2.2−15.8±2.2−17.9±2.0<0.001Values are shown as mean±SD or median (interquartile range). 2D indicates 2‐dimensional; 3D, 3‐dimensional; EDA, end‐diastolic area; ESA, end‐systolic area; GLS, global longitudinal strain; LV, left ventricular; and RV, right ventricular. Echocardiographic Features Between Death and Survivor Groups Values are shown as mean±SD or median (interquartile range). 2D indicates 2‐dimensional; 3D, 3‐dimensional; EDA, end‐diastolic area; ESA, end‐systolic area; GLS, global longitudinal strain; LV, left ventricular; and RV, right ventricular. BNP, 2D‐LVGLS, and 3D‐LVGLS were entered into receiver operating characteristic analysis to estimate the probability of poor clinical outcome. The area under the curve for 3D‐LVGLS (0.77) was larger than that for 2D‐LVGLS (0.67, P=0.012) and BNP (0.63, P=0.032). The best cutoff value of 3D‐LVGLS for detecting all‐cause mortality was −16.1%,P with a sensitivity of 63.6% and a specificity of 84.0% (Figure 2). Figure 3 shows Kaplan–Meier survival curves. When separated by tertiles, mortality was highest in patients with HT with a 3D‐LVGLS ≥−16.6%, followed by 3D‐LVGLS in the range between −16.7% and −18.6%, and lowest in patients with HT with 3D‐LVGLS ≤−18.7% (P<0.001). When stratified by cutoff values, a 3D‐LVGLS lower than −16.1% showed a higher risk of death (P<0.001). Receiver operating characteristic curves of BNP, 2D‐LVGLS, and 3D‐LVGLS for all‐cause mortality. 2D indicates 2‐dimensional; 3D, 3‐dimensional; AUC, area under the curve; BNP, brain natriuretic peptide; GLS, global longitudinal strain; and LV, left ventricular. Kaplan–Meier event‐free survival curves. Survival curves for all‐cause mortality are shown (A) according to tertiles of 3D‐LVGLS (≤−18.7%, >−18.6 to −16.7%, and ≥−16.6%) and (B) according to the cutoff value of 3D‐LVGLS (−16.1%). 3D indicates 3‐dimensional; GLS, global longitudinal strain; and LV, left ventricular. A univariate Cox regression analysis showed that recipient age, body mass index, rejection episodes, hemoglobin, BNP, heart rate, RV fractional area change, tricuspid annular plane systolic excursion, S′, LVEF, 2D‐LVGLS, and 3D‐LVGLS were associated with higher incidence of death in patients with HT (Table 4). To avoid overfitting and collinearity in the multivariable Cox regression analysis, 2 separate models including a baseline model and 1 of the LV function parameters (2D‐LVGLS or 3D‐LVGLS) were established to predict all‐cause mortality. The baseline model was constructed with recipient's age, hemoglobin, BNP, and RV fractional area change. Multivariable analysis revealed that 2D‐LVGLS (hazard ratio [HR], 1.19 [95% CI, 1.06–1.32]; P=0.002) and 3D‐LVGLS (HR, 1.44 [95% CI, 1.24–1.68]; P<0.001) both were the statistically significant predictors of higher risk of death. The model with 3D‐LVGLS was better at predicting all‐cause mortality than the model with 2D‐LVGLS (C‐statistic: 0.814 versus 0.772, P=0.023). Figure S1 illustrates the calibration curves of the multivariable models. The model with 3D‐LVGLS had good calibration, based on its observed versus predicted plot and Hosmer–Lemeshow goodness‐of‐fit test (P=0.486). Table 4Univariable and Multivariable Cox Regression Analyses to Identify Statistically Significant Predictors for All‐Cause MortalityUnivariable Cox regressionMultivariable Cox regressionBaselineBaseline +2D‐LVGLSBaseline +3D‐LVGLSHR (95% CI) P valueHR (95% CI) P valueHR (95% CI) P valueHR (95% CI) P valueRecipient age, y1.04 (1.01–1.08)0.0071.03 (1.01–1.06)0.0231.03 (1.00–1.06)0.0241.03 (1.00–1.06)0.032Recipient body mass index, kg/m2 0.89 (0.82–0.97)0.009Hypertension0.67 (0.36–1.25)0.205Diabetes1.77 (0.96–3.28)0.067Chronic kidney disease1.20 (0.63–2.27)0.574History of acute rejection ≥2R or treated rejection2.31 (1.07–4.97)0.032Donor's age, y1.01 (0.98–1.03)0.606Cold ischemic time >6 h1.13 (0.63–2.05)0.678Hemoglobin, g/L0.97 (0.96–0.98)<0.0010.97 (0.96–0.98)<0.0010.97 (0.95–0.98)<0.0010.97 (0.95–0.98)<0.001B‐type natriuretic peptide, pg/mL4.60 (2.51–8.45)<0.0013.22 (1.61–6.44)0.0012.32 (1.12–4.81)0.0240.152(Log‐transformed)Heart rate, bpm1.04 (1.01–1.08)0.013LV mass, g/m2 1.00 (0.99–1.01)0.948 E/e′1.04 (0.93–1.16)0.518RV fractional area change, %0.93 (0.88–0.97)0.0030.93 (0.88–0.97)0.0020.95 (0.91–0.99)0.0440.077Tricuspid annular plane systolic excursion, mm0.87 (0.80–0.95)0.001 S′, cm/s0.85 (0.73–0.97)0.018LV ejection fraction, %0.92 (0.86–0.99)0.018‐LVGLS (%)1.25 (1.14–1.38)<0.0011.19 (1.06–1.32)0.0023D‐LVGLS (%)1.54 (1.35–1.76)<0.0011.44 (1.24–1.68)<0.001C‐statistic (95% CI)……0.744 (0.667–0.821)0.772 (0.702–0.841)0.814 (0.750–0.877)2D indicates 2‐dimensional; 3D, 3‐dimensional; GLS, global longitudinal strain; HR, hazard ratio; and LV, left ventricular. Univariable and Multivariable Cox Regression Analyses to Identify Statistically Significant Predictors for All‐Cause Mortality 2D indicates 2‐dimensional; 3D, 3‐dimensional; GLS, global longitudinal strain; HR, hazard ratio; and LV, left ventricular. Comparison Between 2D‐GLS and 3D‐GLS Values and Their Correlations A good correlation between 2D‐LVGLS and 3D‐LVGLS was detected (r=0.72, P<0.001). Bland–Altman plot analysis revealed a bias between 2D‐LVGLS and 3D‐LVGLS assessments, with a mean difference of −0.50% and 95% limits of agreement ranging from −4.09 to 3.08 (Figure 4). Comparison between 3D‐LVGLS and 2D‐LVGLS values and their correlations. A, Correlation analysis between 3D‐LVGLS and 2D‐LVGLS; B, Bland–Altman plots between 2D‐LVGLS and 3D‐LVGLS. 2D indicates 2‐dimensional; 3D, 3‐dimensional; GLS, global longitudinal strain; and LV, left ventricular. Reproducibility The intraobserver and interobserver reproducibility of 3D‐LVGLS was excellent, as reflected by the high intraclass correlation coefficients (intraobserver: 0.89; interobserver: 0.84). Additionally, Bland–Altman analysis demonstrated high intraobserver and interobserver agreement, with low bias (intraobserver: 0.04; interobserver: 0.32) and narrow limits of agreement (intraobserver: −3.67 to 3.74; interobserver: −4.30 to 4.93) (Table 5; Figure 5). Table 5Intraobserver and Interobserver ReproducibilityICC (95%CI)BiasLimits of agreementIntraobserver2D‐LVGLS (%)0.96 (0.89–0.98)0.27−1.26 to 1.793D‐LVGLS (%)0.89 (0.75–0.96)0.04−3.67 to 3.74Interobserver2D‐LVGLS (%)0.91 (0.78–0.96)0.36−2.01 to 2.743D‐LVGLS (%)0.84 (0.64–0.93)0.32−4.30 to 4.932D indicates 2‐dimensional; 3D, 3‐dimensional; ICC, intraclass correlation coefficient; LV, left ventricular; and GLS, global longitudinal strain. Intraobserver and Interobserver Reproducibility 2D indicates 2‐dimensional; 3D, 3‐dimensional; ICC, intraclass correlation coefficient; LV, left ventricular; and GLS, global longitudinal strain. Bland–Altman plots for inter‐ and intraobserver comparisons of 3D‐LVGLS and 2D‐LVGLS. A and C, Intraobserver variability. 3D‐LVGLS and 2D‐LVGLS determined by observer 1: first and second investigation in comparison. B and D, Interobserver variability. 3D‐LVGLS and 2D‐LVGLS calculated by observers 1 and 2. Bland–Altman plot showing biases and limits of agreement (mean±2SD) dashed lines indicated 95% limits of agreement. 2D indicates 2‐dimensional; 3D, 3‐dimensional; GLS, global longitudinal strain; and LV, left ventricular. DISCUSSION To the best of our knowledge, this is the first study to investigate the feasibility and prognostic value of 3D‐LVGLS derived from 3D‐STE in patients with HT. The main findings of this study can be summarized as follows: (1) 3D‐LVGLS measurement provided a feasible method to evaluate graft function during the follow‐up studies of patients with HT; and (2) 3D‐LVGLS was a robust index for predicting all‐cause mortality, offering superior predictive value to 2D‐LVGLS. These findings highlighted the potential of 3D‐LVGLS assessment for risk stratification in recipients with HT. Accurate assessment of graft function is crucial for early detection of complications and prediction of poor outcomes. Graft failure is a major cause of late mortality after HT, accounting for >15% of deaths. 15 The most common causes of graft failure are cardiac allograft vasculopathy and AR. These complications could adversely affect the subendocardium (dominated by longitudinal myocardial fibers) in various ways, such as myocardial edema, 16 , 17 , 18 perfusion abnormalities, 19 , 20 LV wall stress abnormalities, 21 , 22 and myocardial fibrosis. 23 These pathologic processes may mainly lead to longitudinal myocardial contractile dysfunction. Prior studies have demonstrated the prognostic significance of LVEF in patients with HT. 24 However, LVEF is a measurement that reflects the combined function of both longitudinal and circumferential fibers, and it does not have the ability to discriminate between longitudinal and circumferential fibers. Therefore, recent studies advocate that LVEF may not be the most optimal parameters to demonstrate graft function. 6 , 16 This finding was also confirmed in our research. Compared with survivors, nonsurvivors displayed lower LVEF. However, LVEF was still within normal range in both groups. Thus, more sophisticated assessment of myocardial performance in patients with HT is needed. The 2D‐LVGLS has been widely recognized as an important marker to detect subtle changes in myocardial function and some studies have demonstrated its prognostic value in patients with HT. 6 , 7 , 8 , 9 , 10 Sarvati et al reported that graft function evaluated by 2D‐LVGLS at 1 to 3 weeks after transplantation can serve as a reliable and powerful predictor of 1‐year mortality. 8 Clemmensen et al investigated 196 patients with HT and revealed that 2D‐LVGLS was a strong predictor of poor outcomes in long‐term follow‐up of patients with HT. 6 Kobayashi et al advocated that 2D‐LVGLS at the first annual evaluation may provide prognostic insights in predicting 5‐year outcomes following HT. 10 Our study also found that 2D‐LVGLS may provide prognostic value in recipients with HT, which is consistent with previous studies. However, 2D‐STE is hampered by 2D plane and out‐of‐plane motion of the speckles. This is particularly relevant in recipients with HT, in whom marked translational motion of the transplanted heart is present. 11 Similar to general cardiac surgery, the transplanted heart loses the support and hold provided by the pericardial sac. However, unlike general cardiac surgery, there is a disproportion between the small size of the donor's heart and the large size of the recipient's mediastinal cavity. Importantly, this translational motion of patients with HT may aggravate the occurrence of “out‐of‐plane phenomenon.” The 3D‐STE has been introduced as a novel technique that allows an accurate and comprehensive assessment of LV function. 12 Recently, several studies have reported the prognostic value of 3D‐LVGLS in clinical settings. 25 , 26 , 27 In an observation of patients with severe aortic stenosis and preserved LVEF, Nagata et al demonstrated that 3D‐LVGLS was a robust index for predicting adverse cardiac events. 25 Iwahashi et al compared the prognostic value of 3D‐LVGLS and 2D‐LVGLS in a series of 255 patients with ST‐segment–elevation myocardial infarction and found that 3D‐LVGLS was superior to 2D‐LVGLS in predicting LV remodeling and 1‐year prognosis. 26 However, their results cannot be extrapolated to recipients with HT, due to the unique translational motion of HT patients. This limitation is overcome by 3D‐STE by avoiding apical foreshortening, tracking myocardial motion independently of the imaging plane, and directly measuring the volume rather than relying on geometric assumptions. Thus, 3D‐LVGLS may be theoretically more applicable to recipients of HT than 2D‐LVGLS. Moreover, Sun et al also validated that 3D‐GLS showed excellent accuracy in evaluating ventricular function of transplanted hearts compared with cardiac magnetic resonance. 13 However, the feasibility and prognostic value of 3D‐LVGLS in recipients of HT have not been described, and whether 3D‐LVGLS is better than 2D‐LVGLS remained unknown. The present study is the first report to estimate the prognostic significance of 3D‐LVGLS in patients with HT and to directly compare its value with that of 2D‐LVGLS. In our study, both 3D‐LVGLS and 2D‐LVGLS emerged as robust predictors of all‐cause mortality among patients with HT. We provide the first evidence that although both 3D‐LVGLS and 2D‐LVGLS have prognostic value, 3D‐LVGLS is better at predicting outcomes in recipients of HT. Our study highlights that patients with HT can be effectively monitored by 3D‐LVGLS under conditions of sufficient image quality. Limitations This study had several limitations. First, this was a single‐center study, and the predictive value of 3D‐LVGLS in patients with HT requires further validation in multicenter studies. Second, 3D‐STE itself is limited by the low frame rates and suboptimal image quality, which may affect strain analysis and tracking quality. Subjects with poor echocardiographic image quality were excluded from the analysis. This potentially limits the generalizability of our findings. Finally, this study analyzed only patients in normal rhythm during imaging; thus, these results cannot be extrapolated to patients with various types of arrhythmias. Nonetheless, the present study may be the largest contemporary single‐center cohort to investigate the associations between 3D‐LVGLS and long‐term prognosis in patients with HT. CONCLUSIONS The study showed that 3D‐LVGLS measurement provides a feasible method to assess graft function in a relatively large cohort of patients with HT. More important, 3D‐LVGLS is a powerful predictor of all‐cause mortality in patients with HT and provides greater prognostic value than 2D‐LVGLS. Our study highlights that patients with HT can be effectively monitored by 3D‐LVGLS under conditions of sufficient image quality. Sources of Funding This study was supported by the National Natural Science Foundation of China (81922033, 82371991 and 82302229) and the Natural Science Foundation of Hubei Province (2024AFB086). Disclosures None. Supporting information Table S1 Figure S1