Right ventricular free wall longitudinal strain as determined by speckle tracking echocardiography is a noninvasive predictor of acute cellular rejection in adult heart transplantation

Acute cellular rejection (ACR) is a major complication after heart transplantation. Endomyocardial biopsy (EMB) remains the gold standard for its diagnosis, but with concerning complications. We evaluated the usefulness of speckle tracking echocardiography (STE) and the biomarkers for detecting ACR after heart transplantation. We prospectively studied 60 transplant patients with normal left and right ventricular systolic function who underwent EMB for surveillance sixth months after transplantation. Sixty age- and sex-matched healthy individuals constituted the control group. Conventional echocardiographic parameters left ventricular global longitudinal, radial and circumferential strain (LV-GLS, LV-GRS and LV-GCS, respectively), left ventricular systolic twist (LV-twist) and right ventricular free wall longitudinal strain (RV-FWLS) were analyzed just before procedure. We also measured biomarkers at the same moment. was the only independent predictor of ACR (odds ratio = 0.57; 95% CI = 0.40–0.81; p = 0.02). RV-FWLS < 18.0% had an accuracy of 85% to predict ACR.


Introduction
Heart transplantation is the choice treatment for select patients with end-stage heart failure (1, 2). Although signi cant advances in immunosuppressive therapy have been bene cial in decreasing cardiac allograft rejection, graft failure remains one of the major associated complications (3,4). For this reason, adequate monitoring of heart transplant patients to diagnose and initiate speci c therapy for transplant rejection in a timely manner is important, albeit challenging.
Endomyocardial biopsy (EMB) is the widely accepted gold standard for the diagnosis of acute cellular rejection (ACR) (5). However, it is invasive and is associated with complications in 0.2 to 5.5% of cases; these complications include tricuspid regurgitation, cardiac perforation and cardiac tamponade (6)(7)(8)(9).
Additional limitations of EMB include subjectivity of the pathological analysis and also signi cant interobserver variability, which compromises reliability and reproducibility (10)(11)(12).
Therefore, there has been much effort to develop noninvasive and accurate methods that can reduce the need for EMB, including biomarker detection, imaging techniques and genetic tests (13,14). The most used biomarkers in patients with heart transplantation are troponin and BNP, however there is a considerable heterogeneity among studies about the time dosage and the predict value for detect ACR (15,16).

Speckle tracking echocardiography (STE) is a newly incorporated technique in clinical practice based on
identi cation and frame-by-frame tracking of natural acoustic myocardial markers that arise from the interaction of ultrasound waves and small myocardial elements. These speckle patterns are unique in each segment, and changes in their positions enable the determination of multiple parameters involved in cardiac mechanics.
In the present study, we aimed to assess the value of STE derived strain measurements and biomarkers for the noninvasive detection of ACR after heart transplantation. We also evaluated the ventricular dynamics of heart transplant patients in comparison with control individuals using STE.

Patients
From January 2014 to November 2018, we prospectively studied heart transplant patients who underwent EMB in the 6 th month after orthotopic heart transplantation for the diagnosis of ACR for surveillance. All the included patients were asymptomatic and did not present any hemodynamic compromise at the time of enrollment.
The exclusion criteria were as follows: age < 18 years, arrhythmia, left ventricular ejection fraction < 0.55, right ventricular dysfunction of any degree (FAC < 35%), vascular graft disease, humoral rejection, two or more previous cellular rejection episodes, chronic kidney disease, Chagas disease reactivation, limited echocardiographic window for STE analysis and inconclusive EMB analysis.
All patients underwent the same protocol according to prede ned steps. First, enrolled patients underwent complete echocardiographic analysis for assessment of left and right ventricular function. For those with normal systolic function, images were acquired for mechanical analysis by STE, and blood was withdrawn for biomarker tests. The patients then underwent EMB and, according to the results, were divided into the following groups: 1) without signi cant ACR (grades 0 and 1R) and 2) with signi cant ACR (grades 2R and 3R).
Individuals without a clinical history of any disease known to interfere with myocardial physiology or structure and who were matched by sex and age with the transplant patients constituted the control group. This group underwent STE for analysis of ventricular mechanics and was compared with the groups of transplant patients with and without signi cant ACR. The study protocol was approved by our ethical committee, and all patients gave written informed consent to participate.
Endomyocardial biopsy EMB was performed through the internal jugular or femoral vein under radioscopy. A minimum of 3 distinct ventricular myocardium fragments were collected (each consisting of at least 50% myocardium) and sent for anatomopathological analysis. A sample was considered su cient when at least 3 myocardial fragments were obtained for analysis by optical microscopy after xation in 10% formalin and staining of the laminae with hematoxylin and eosin. Two experienced cardiac pathologists blinded to the echocardiographic results analyzed all biopsies. The grade of rejection was based upon the recommendations of the International Society for Heart and Lung Transplantation (17). The results of EMB were described as grade 0 (without rejection), grade 1R (mild rejection, low grade), grade 2R (moderate rejection, intermediate grade) or grade 3R (severe rejection, high grade) (17). In our study, grades 2R and 3R were considered to indicate signi cant ACR.

Echocardiography
On the day of EMB, just before the procedure, patients underwent echocardiographic examination on a commercially available machine equipped with an MS5 probe (GE Vivid 9, GE Healthcare, Milwaukee, Wisconsin, USA). Image acquisition and assessment were performed according to the recommendations of the American Society of Echocardiography (18). Left ventricular ejection fraction was obtained by Simpson's rule throughout apical 4-and 2-chamber views, and left ventricular mass was calculated using the equation proposed by Devereux et al. (19), indexed by body surface area to derive the left ventricular mass index. Right ventricular parameters were analyzed in the apical 4-chamber view focused on the right ventricle. Right ventricular systolic function was assessed with the conventional parameters recommended for routine clinical practice: tricuspid annular plane systolic excursion, systolic excursion velocity, and fractional area change, which were obtained with M-mode, pulsed tissue Doppler and twodimensional echocardiography, respectively. Diastolic function was evaluated based on mitral in ow E/A pattern, E/A ratio, E velocity deceleration time, annular tissue Doppler curves (e'/a'), and E/e' ratio.
To assess cardiac mechanics, 3 consecutive cardiac cycles were recorded. Left ventricular short-axis and apical views were acquired using two-dimensional grayscale second-harmonic imaging at a frame rate of 50-80 frames per second. Left ventricular short-axis views at the basal, mid (papillary muscles) and apex levels were acquired to analyze circumferential and radial strain, while left ventricular apical 4-, 2-and 3chamber views were used to assess left ventricular global longitudinal strain (LV-GLS) and an apical 4chamber view was used to analyze right ventricular free wall longitudinal strain (RV-FWLS).
Speckle tracking echocardiography analysis was performed o ine using dedicated software (EchoPAC, version BT11, GE Healthcare). All echocardiographic measurements were performed by one specialist blinded to the clinical data. End-systole was determined by pulsed-wave Doppler at the time of aortic valve closure. After the ventricular endocardial border was manually traced, the epicardial borders were automatically de ned to create regions of interest according to the ventricular segmentation; if necessary, adjustments were made by the operator. In particular, care was taken not to include the myocardial trabeculae and the pericardium. Following this step, myocardial speckles were automatically tracked by dedicated software and, in the case of suboptimal tracking, further manual adjustments were allowed, resulting in strain curves that were exported to a spreadsheet. Longitudinal and circumferential peak strains were de ned as the greatest negative de ection, while radial strain was de ned as the greatest positive value, before aortic valve closure on the strain curve. Global results were obtained by averaging the values obtained for all segments on each plane: accordingly, the LV-GLS was obtained from the mean of 18 segments acquired in the apical 4-, 2-, and 3-chamber views, while circumferential and radial strain were measured by averaging the values from all the segments of the basal, mid and apex levels. Left ventricular twist is the wringing motion of heart around its long axis. It was calculated as the net absolute difference between apical and basal rotations (LVtwist = ROTapical -ROTbasal). By widely assumed convention, apical rotation had positive values and basal, negative (20). RV-FWLS was obtained by averaging the values of the 3 right ventricular free wall segments: basal, medial, and apical. Care was taken to obtain the best visualization of the right ventricle to enable accurate delineation of its endocardial border. Irregular cardiac cycles or those containing premature ventricular beats were excluded.
The intraobserver reproducibility of strain measurements was assessed in a subsample of 30 randomly selected patients 3 months after the initial evaluation; the observer was blinded to the previous results. Interobserver variability was assessed in the same subsample by a second blinded experienced echocardiographer.

Biomarkers
Biomarker analysis was performed before EMB. For this, a 20 mL blood sample was collected from a peripheral vein to determine the plasma levels of cardiac troponin I and B-type natriuretic peptide (BNP).
Cardiac troponin I levels were quanti ed with high sensitivity 3-step sandwich immunoassay using direct chemiluminescent technology and constant amounts of 2 monoclonal antibodies. An auxiliary reagent was included to reduce nonspeci c binding using an ADVIA Centaur TnI-Ultra commercial kit (Siemens Healthcare Diagnostics, Tarrytown, New York, USA). The level of detection was 0.006 ng/mL (levels <0.006 are reported as 0.005 ng/mL). The normal range of cardiac troponin I was considered <0.04 ng/mL. Plasma concentrations of BNP were determined with a 2-step sandwich immunoassay using direct chemiluminescent technology and constant amounts of 2 monoclonal antibodies using an Advia Centaur commercial kit (Siemens Healthcare, Malvern, Pennsylvania, USA). The level of detection was 2 pg/mL. Levels <2 are reported as 1 pg/mL.

Statistical analysis
Categorical variables were compared using Pearson chi-square tests, Fisher exact tests, or likelihood ratio tests. Continuous variables were compared using analysis of variance and Tukey's test (normal distribution) or Kruskal-Wallis and Dunn's tests. The results are expressed as the means with standard deviations or as the medians with interquartile ranges. Linear correlations were tested using the Spearman rank method. The sensitivity, speci city, accuracy, and positive and negative predictive values of each test for predicting ACR were calculated.
Associated risk factors of ACR (p<0.05) were analyzed by were introduced in a logistic regression model with a forward stepwise approach for multivariate analysis, while receiver operating characteristic (ROC) curve analysis was conducted to determine the accuracy and an optimal cut-point value (the optimal cutpoint was assessed by jointly maximizing sensitivity and speci city).
The interobserver and intraobserver reproducibility of LV-GLS, LV-Twist and RV-FWLS were assessed using intraclass correlation coe cients and 95% con dence intervals (CIs) in one-way random and twoway mixed models.
All analyses were performed using SPSS version 17 (SPSS Inc., Chicago, Illinois, USA). A p value < 0.05 was considered statistically signi cant.

Results
Eighty-nine patients were initially included in the study. Among these patients, 29 were excluded because of the following characteristics: 2 due to cardiac arrhythmia, 4 due to left ventricular systolic dysfunction, 4 due to right ventricular systolic dysfunction, 8 due to previous ACR episodes, 4 due to a limited echocardiographic acoustic window for STE analysis, 03 due to inconclusive results of EMB, 2 due to humoral rejection and 2 due to Chagas disease recurrence. A total of 60 heart transplant patients and 60 control individuals constituted the nal study population ( Figure 1).
The baseline characteristics of the heart transplant patients included in this study are described in Table  1. The clinical characteristics of the groups with and without signi cant ACR detected by EMB, as well as those of the control group individuals, are described in Table 2.

Speckle tracking echocardiography
The variables obtained by echocardiography are shown in Table 3. There were no differences between the group with signi cant ACR and the group without signi cant ACR regarding these parameters. However, heart transplant patients had a greater septum and posterior wall thickness, larger left atrium diameter, higher E/E', and higher relative thickness and left ventricular mass index values than control individuals. In addition, heart transplant patients had lower right ventricular fractional area change, systolic velocity of the tricuspid annulus, E' velocity, A' velocity and tricuspid annular plane systolic excursion than control individuals.
The LV-GLS, LV-GCS, LV-GRS, RV-FWLS and LV-twist values are shown in Table 4. The absolute values of these variables were signi cantly lower in the heart transplant patients without rejection than in the control individuals, except LV-twist. In the group with signi cant ACR, the LV-GLS, LV-twist and RV-FWLS were signi cantly lower (in absolute value) than that in the group without signi cant ACR (12.5% ± 2.9 vs 14.8% ± 2.3, p = 0.002 ; 13.9° ± 4.8 vs 17.1° ± 3.0, p = 0.048; 21.4%± 3.2 vs 16.6% ± 2.9, p<0.001; respectively), as shown in Figure 2. Table S1 (see the Supplementary Appendix) shows sensitivity, speci city, NPV, PPV and accuracy for each parameter related to the diagnosis of ACR degree ≥2R.

Predictors of acute cellular rejection
A multiple logistic regression model was constructed to determine the predictive factors of ACR. In univariate analysis, the predictors of ACR were cardiac troponin I levels, RV-FWLS, LV-GLS and LV-twist.  Figure 3 shows the ROC curve of RV-FWLS.

Discussion
The main nding of the present study was that RV-FWLS determined through STE is a noninvasive independent predictor of signi cant ACR in heart transplant patients with preserved LV and RV systolic functions. A cutoff value of < 18.0% had 85.0% accuracy for detecting signi cant ACR. In this population, the very strong negative predictive value of RV-FWLS may help to rule out ACR. In addition, we observed that modular values of LV-GLS and LV-twist were lower, and cardiac ultrasensitive troponin I levels was higher in transplant patients with signi cant ACR compared with patients without signi cant ACR.
In accordance with previous reports in the literature (21), we con rmed in our population that heart transplant patients had a characteristic cardiac geometric remodeling featured by greater septum and posterior wall thickness, larger left atrium diameter, and greater left ventricular mass index values than matched non transplanted controls. Heart transplant patients also showed lower values for conventional right ventricular systolic function parameters, such as fractional area change, tricuspid annulus systolic velocity and tricuspid annular plane systolic excursion. Moreover, regarding left ventricular diastolic function, heart transplant patients had lower tissue Doppler velocities and higher E/e´ ratios, suggesting impaired relaxation and increased left ventricular lling pressures.
We demonstrated that heart transplant patients without rejection present unique ventricular dynamics, characterized by lower LV-GLS, LV-GCS, LV-GRS and RV-FWLS, in comparison with control individuals. We have con rmed the data recently published by Ingvarsson et al. (21), which showed that echocardiographic measurements from 124 heart transplant patients were different from reference values, unless LV-CGS. Unfortunately, the investigators did not analyze LV-Twist. In our study, we used a non-transplanted control group matched by age and sex to con rm these results. Multiple mechanisms may explain the different echocardiographic ndings in heart transplant patients. The pathophysiology involves the consequences of surgical trauma, such as ischemic injury and the release of in ammatory mediators, in addition to previous pulmonary hypertension compromising right ventricular dynamics and risks associated with rejections, cardiac biopsies and immunosuppressive medications. We assume that preserved LV-twist might be responsible for maintaining cardiac function in these transplant patients without rejection.
ACR is a signi cant and frequent complication of heart transplantation. In the rst year, it is the most common cause of mortality. Currently, EMB is the clinical gold standard in screening for graft rejection after heart transplantation and is actually the only tool for the diagnosis and classi cation of allograft rejection (5). Much effort has been made to improve the consistency, reliability and reproducibility of the histopathological evaluation of EMB. However, several issues make EMB assessment more di cult and less reproducible than it should be. Critical issues include the subjective and challenging pathological interpretation of EMBs and the risks associated with the procedure (22). Considering these limitations, noninvasive techniques to detect cardiac rejection have been evaluated, such as analysis of biomarkers (cardiac troponin I and BNP) and the use of various imaging modalities, including echocardiography, computed tomography, magnetic resonance imaging, positron emission tomography and intragraft gene expression pro ling (14,23).
Previous studies have shown that myocardial strain has higher sensitivity than conventional echocardiography and therefore may be an important tool to detect early subclinical cardiac dysfunction (24). Although myocardial strain imaging has been reported to have potential for the detection of graft dysfunction in the early stage (25,26), its diagnostic value has not been widely recognized yet. Our study showed that the LV-GLS was lower in transplant patients with signi cant ACR compared with patients without signi cant ACR, as reported in a recent meta-analysis by Elkaryoni et al (27). When ACR occurs, myocardial deformation is impaired due to in ammatory cellular in ltration and myocardial edema and can be re ected by myocardial strain parameters. However, the diagnostic value of other strain parameters by 2D STE on ACR detection is still unknown.
Mingo-Santos et al (28). demonstrated a predictive role of STE parameters in the diagnosis of ACR (RV-FWLS and LV-GLS, with threshold values of < 17% and < 15.5%, respectively). That study classi ed biopsies into 3 groups (0, 1R, and ≥ 2R) and was not able to detect any difference in measurements between the 1R and ≥ 2R groups. However, our study divided biopsy results into only two groups according to the grade of rejection: biopsies without signi cant rejection (0 and 1R) and biopsies with signi cant rejection (≥ 2R). This division was based upon the clinical meaning of the rejection grade, since cases with grades of 0 or 1R do not require immediate intervention via adjustment of immunosuppressive medications, whereas this adjustment is necessary in patients presenting with 2R and 3R rejection. In agreement with the study from Mingo-Santos et al. (28), we con rmed the predictive value of RV-FWLS for the diagnosis of ACR, highlighting the high speci city of this parameter. RV-FWLS values > 18.0% rule out cardiac rejection with a speci city of 88.4%. Our cutoff value was slightly higher than that reported by Mingo-Santos et al. (28). We speculate that it might be due to different echocardiographic equipment or even the characteristics of the studied populations.
To the best of our knowledge, this is the rst study to analyze right ventricular strain using STE in patients with normal right ventricular systolic function (fractional area change > 0.34) to diagnose clinically signi cant ACR. Extra care was given in our study to restrict patient selection to a xed period of time (6 months post heart transplantation) to minimize the possible in uence of the time since heart transplantation on right ventricular strain, as recommended by the current guidelines (29). Additionally, this xed period of selection minimized the possible bias of pretransplantation ischemic injury, which can manifest in up to the sixth month post heart transplantation. The rst six months is a period of adaptation during which many patients can still present some degree of right ventricular systolic dysfunction. The two studies that have analyzed right ventricular strain using STE to diagnose ACR did not exclude patients with fractional area change below the lower limit of normality, included individuals as early as 10 days post heart transplantation and one of them did not allocate heart rejection events according to their clinical signi cance (28,30).
As we show in our results, LV-twist values were signi cantly lower in the group with signi cant ACR than that in the group without signi cant ACR (13.9° ± 4.79 vs 17.1° ± 3.02, p < 0.048). In parallel to our results, the unique previous study that applied 2D-STE-derived LV-twist measurement to detect rejection in heart transplanted patients have demonstrated that LV-twist decreased more in group with ACR than in group without ACR (9.6 ± 2.7 vs 12.2 ± 2.3) degrees, p < 0.0001) (31). We postulated that twist precede deterioration in LVEF, suggesting early myocardial involvement in cardiac rejection. With the advancement of technology that has made this technique more available and increasingly feasible, this parameter of cardiac mechanics has been increasingly studied in other pathological situations, and can be applied in this type of patient (32,33).
Our study failed to nd any association between diastolic markers and rejection. The results found in the literature are highly con icting and could not be reproduced by our data. This can be explained by a limitation of diastolic dysfunction parameters due to their dependence on heart rate (which is generally elevated in transplant patients, with fusion of E and A waves), loading conditions and donor age (34)(35)(36)(37)(38).
As acute rejection promotes cardiomyocyte necrosis and results in compromised cardiac mechanics, cardiac troponin and BNP have been evaluated as potential diagnostic tools for ACR (39,40). There have been controversial results on these biomarkers in the eld of heart transplantation (41,42). Our study used an ultrasensitive assay for cardiac troponin I that detects 10 to 100 times lower levels than standard assays. Troponin was measured before biopsy so that this procedure did not interfere with its serum levels. Troponin I levels were signi cantly elevated in patients with signi cant ACR than in those without it. However, in multivariate analysis, troponin I was not an independent predictor of ACR. Serum BNP levels were not different between groups; this nding can be explained by the suggestion of some studies that BNP remains altered in most patients for up to one year after heart transplantation (43,44). Our results are in accordance with those of Bader et al. (45), who observed that BNP levels did not predict rejection at any time point after heart transplantation and suggested that BNP is not clinically useful for the detection of acute cellular rejection.

Limitations
The limitations of this study should be addressed. First, this was a single-center study with a small number of patients and a limited number of rejection episodes graded equal to or above 2R (17 out of 60 samples). Despite its extensive validation, STE is still an evolving technique, and improvements such as better tracking accuracy are still needed. Additionally, STE accuracy is highly dependent on image quality.
Suboptimal resolution can produce a negative impact on the nal results. Nevertheless, despite these limitations, we were able to successfully perform speckle-tracking analysis of both left and right ventricular longitudinal strain in 95% of the patients. The reproducibility of the parameters was good and was similar to that reported in other studies. Finally, our results must be independently validated in a prospective external cohort, preferably in multicenter studies, before they can be used in clinical practice.

Conclusions
Heart transplant patients have altered left ventricular dynamics compared with control individuals. STE is a valuable technique for the noninvasive detection of signi cant ACR in patients with heart transplant, and RV-FWLS (cutoff point < 18.0%) is an independent predictor of ACR with good accuracy and high negative predictive value.

Perspectives
With the increasing number of heart transplantation in the world, new tools for detecting ACR is highly desirable. In our study we demonstrated that the analysis of RV-FWLS is able to predict ACR. It seems to be an accurate and noninvasive technique for analysis of heart transplant patients. All authors listed on the title page have read the manuscript, attest to the validity and legitimacy of the data and its interpretation, and agree to its submission to BMC Cardiovascular Disorders.

Consent for publication
Not applicable Availability of data and materials All data generated or analysed during this study are included in this published article. If you have questions or additional information, the datasets used and/or analysed during the current study are available from the corresponding author upon reasonable request.

Competing interests
The authors declare that they have no competing interests.  The values are presented as the mean (standard deviation) or number (percentage). CTRCD = Cancer therapeutic-related cardiac dysfunction; ACEI = angiotensin converting enzyme inhibitor; ARB = aldosterone receptor blocker CCB = calcium channel blocker    The values are presented as the median (interquartile range). *Mann-Whitney BNP = brain natriuretic peptide; ACR = acute cellular rejection.   Study ow diagram. EMB = endomyocardial biopsy; ACR = acute cellular rejection.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SupplementaryAppendixACR.docx