Feature-Tracking Left Ventricular Global Longitudinal Strain Improves 6 Months After Kidney Transplantation Associated With Reverse Remodeling, Not Myocardial Tissue Characteristics

Purpose To determine whether left ventricular (LV) global longitudinal strain (GLS) measured by feature-tracking (FT) cardiac magnetic resonance (CMR) improves after kidney transplantation (KT) and to analyze associations between LV GLS, reverse remodeling, and myocardial tissue characteristics. Methods This is a prospective single-center cohort study of kidney transplant recipients who underwent two CMR examinations in a 3T scanner, including cines, tagging, T1 and T2 mapping. The baseline exam was done until 10 days after transplantation and the follow-up after 6 months. Age and sex-matched healthy controls were also studied for comparison. 2-, 3-, 4-chamber long Scan eld of view (FOV) 37 cm, repetition time (TR) 43.54 ms, echo time (TE) 1.38 ms, ip angle (FA) 50°, slice thickness 6 mm, in-plane image resolution 1.6 x 1.6 mm. Myocardial tissue tagging was performed with an ECG-gated bSSFP line tagging sequence with complementary spatial modulation of magnetization (CSPAMM). Image parameters were: FOV 32 cm, TR 48.15, TE 2.54, FA 10°, slice thickness 7 mm with a tag spacing of 7 mm. Short-axis tissue tagging was performed on 3 levels of the LV, positioned at 25%, 50% and 75% of the distance between the mitral valve annulus and the apex on a LV 4-chamber view in end-systole, and in 2- and 4-chamber long axis views. Quantitative T2 mapping was performed using a T2-prepared SSFP sequence in a mid-ventricular short axis view with the following imaging parameters: FOV 36-cm, TR 254.32, TE 1.07 ms, ip angle 35°, slice thickness 8 mm, in-plane image resolution 2.5 x 1.9 mm, acquisition in late diastole on every fourth heartbeat; T2 preparations: 0 ms, 25 ms and 55 ms. Quantitative T1 mapping was performed with a Modied Look-Locker Inversion-Recovery (MOLLI) sequence in mid-ventricular short axis view without intravenous contrast injection (Native T1). Scan parameters: FOV 36-cm, TR 316.09, TE 1.12 ms, ip angle 35°, slice thickness 8 mm, in-plane image resolution 2.1 x 1.4 mm, acquisition in late diastole on every other heartbeat, minimal inversion time 120


Introduction
Cardiovascular disease (CVD) remains the leading cause of death in patients with chronic kidney disease (CKD) and end-stage renal disease (ESRD) [1]. This increased cardiovascular risk is mainly related to changes in cardiac structure and function named uremic cardiomyopathy (UC) [2,3]. The histologic basis for UC is cardiomyocyte hypertrophy and increased interstitial myocardial brosis [4,5], that eventually cause myocardial dysfunction. Kidney transplantation (KT) is considered the most effective form of ESRD treatment and is associated with reverse remodeling, improved ventricular function and better outcomes [6], however kidney transplant recipients (KTR) are still at increased cardiovascular risk compared to general population [7,8].
Cardiac magnetic resonance (CMR) is the gold standard method to measure cardiac structure and function, with further advantage to non-invasively characterize myocardial tissue using T1 and T2 mapping [9], without intravenous gadolinium injection, which is of great interest in renal insu ciency because of the risk of nephrogenic systemic brosis (NSF) [10]. The recent development of featuretracking techniques (FT-CMR) now allow assessment of global longitudinal strain (GLS) from standard cine images without the need of other specialized pulse sequences or additional scanning time [11].
Previous CMR studies have shown subclinical features of myocardial disease characterized by reduced left ventricular (LV) GLS and increased myocardial brosis, as assessed by T1 mapping, in CKD [12] and ESRD [13,14]. Although KT was associated with reduced myocardial brosis [15], previous FT-CMR [16] and speckle-tracking echocardiography (STE) [17,18] studies found different results about the effects of KT in LV GLS, which is the most reliable and studied strain parameter. Nevertheless, the relationships between LV GLS, cardiac structure and myocardial tissue characteristics in this setting are unknown.
Accordingly, in this study we sought to evaluate whether LV GLS measured by FT-CMR improves after KT and analyze associations between LV GLS, cardiac structure (mass and volumes) and myocardial tissue characteristics (native T1 and T2). In addition, we compared CMR data of KTR at follow-up with healthy volunteers to analyze whether KT could reverse these biomarkers of UC.

Study design and participants
This is a single-center prospective cohort study in a university hospital with ESRD patients who received a kidney transplant and underwent two CMR examinations. The rst exam (baseline) was performed between the 1st and the 10 th postoperative days. The second one was performed 6 months after renal transplantation. We included consecutive patients over 18 years-old who received a kidney transplant from a living or deceased donor. We excluded patients with a contraindication to CMR (e.g., pacemaker, cochlear implant, cerebral aneurysm clip, tattooing, claustrophobia) or inability to perform breath hold. For comparison a group of age-and sex-matched healthy controls was selected from the hospital records. Healthy subjects had normal kidney function, no known chronic disease, and were not on regular medication.
This study complied with the Declaration of Helsinki, and the institutional review board of Botucatu Medical School-UNESP approved the research protocol (approval number: 972.129). All participants provided witnessed, written, informed consent. Siemens Healthineers (Erlangen, Germany) provided the use of work-in-progress # 448B (VB17A) quantitative cardiac parameter mapping (T1|T2|T2*) in this study. No person from this company had access to study data or was involved in image analysis, manuscript preparation, or any part of the study. The authors had full control of the data submitted for publication.

CMR technique and measurements
All patients underwent their examination on a 3.0-T Magnetom Verio Scanner (Siemens, Erlangen, Germany) with a phased array chest coil, according to study protocol. A cardiac cine balanced steadystate free precession (bSSFP) sequence was acquired using retrospective cardiac gating. Typically, 25 phases were acquired in 2-, 3-, and 4-chamber long axis views and a stack of short axis views. Scan parameters: eld of view (FOV) 37 cm, repetition time (TR) 43.54 ms, echo time (TE) 1.38 ms, ip angle (FA) 50°, slice thickness 6 mm, in-plane image resolution 1.6 x 1.6 mm. Myocardial tissue tagging was performed with an ECG-gated bSSFP line tagging sequence with complementary spatial modulation of magnetization (CSPAMM). Image parameters were: FOV 32 cm, TR 48.15, TE 2.54, FA 10°, slice thickness 7 mm with a tag spacing of 7 mm. Short-axis tissue tagging was performed on 3 levels of the LV, positioned at 25%, 50% and 75% of the distance between the mitral valve annulus and the apex on a LV 4chamber view in end-systole, and in 2-and 4-chamber long axis views. Quantitative T2 mapping was performed using a T2-prepared SSFP sequence in a mid-ventricular short axis view with the following imaging parameters: FOV 36-cm, TR 254.32, TE 1.07 ms, ip angle 35°, slice thickness 8 mm, in-plane image resolution 2.5 x 1.9 mm, acquisition in late diastole on every fourth heartbeat; T2 preparations: 0 ms, 25 ms and 55 ms. Quantitative T1 mapping was performed with a Modi ed Look-Locker Inversion-Recovery (MOLLI) sequence in mid-ventricular short axis view without intravenous contrast injection (Native T1). Scan parameters: FOV 36-cm, TR 316.09, TE 1.12 ms, ip angle 35°, slice thickness 8 mm, inplane image resolution 2.1 x 1.4 mm, acquisition in late diastole on every other heartbeat, minimal inversion time 120 ms; increment 80 ms. The T1 mapping scheme included 5 acquisitions after the rst inversion pulse, followed by a 3-heartbeat pause and a second inversion pulse followed by 3 acquisitions (5(3)3).

CMR analysis
The biventricular end-diastolic volume (EDV) and end-systolic volume (ESV) were measured by manual segmentation of the short axis cine images, using Argus function software (Siemens, Erlangen, Germany). The endocardial borders were traced at end-diastole and end-systole, including trabeculations and papillary muscles in the blood pool. EDV and ESV were calculated for each ventricle using the method of disc summation. Ventricular stroke volume (SV) was calculated as the difference between the EDV and ESV, and ventricular ejection fraction (EF) was (SV/EDV) × 100. LV epicardial borders were draw only at end-diastole to calculate LV mass (LVM). All volume measurements were indexed for the body surface area (BSA) and expressed in ml/m 2 .
Myocardial feature-tracking analysis was performed processing cine images using strain module of Segment Medviso software, which was previously validated in a clinical setting [19]. Circumferential and radial strains were analyzed in basal, medial and apical short axis slices by manual segmentation of the LV blood pool cavity and myocardium, while longitudinal strains were analyzed in 2-,3-and 4-chamber long axis views. Strain values were obtained for each segment and global values de ned as the mean of all segmental values. For validation, tagging strain analysis was performed using the same software to process tagged long axis views. Figure 1 displays an example of GLS feature-tracking analysis at baseline and follow-up CMR exams. T1 and T2 maps were automatically generated on the MR scanner with motion corrected images using a novel non-rigid registration algorithm [20,21]. A region of interest (ROI) was then drawn conservatively in the septal myocardium for each map, accordingly previous consensus [22].
An experienced reader (ACVM) measured ventricular volumes, mass and EF, while another experienced reader (MFB) independently performed T1, T2 and strain analysis, blinded to former results.

Analysis of reproducibility and validation of LV GLS measurements
To determine the intraobserver reproducibility of LV GLS measured by FT-CMR, 15 exams were randomly selected and the analysis repeated by the same observer about 6 months after the initial assessment.
These exams were also used to validate LV GLS measured by FT-CMR against the reference standard tagging.

Statistical analysis
Kolmogorov-Smirnov test was applied to determine appropriate parametric or nonparametric tests. Quantitative variables were expressed as mean ± standard deviation or median (interquartile range) and compared by t test or Wilcoxon signed-rank test, whereas qualitative variables were expressed by their frequencies and percentages, and compared by the chi-square test or Fisher's exact test. The relationship between changes in LV GLS and variables of interest were assessed using Pearson's correlation coe cients for continuous normally distributed variables and Spearman's correlation for categorical or non-normally distributed data. Linear regression analysis was used to evaluate the in uence of clinical variables in LV GLS changes. Intraobserver reproducibility was assessed by analyzing Bland-Altman plot.
All data were analyzed using SAS Studio 3.8 or Microsoft Excel software. A p-value of ≤ 0.05 was considered signi cant.

Participants
We consented 47 patients of whom 44 [Mean age 50 ± 11 years-old; 27 (61.4%) male] completed the two CMR exams (n = 88 CMR studies). One patient died on 11 th day after transplant surgery and two patients refuse to undergo a second CMR exam. The median time from transplant to the rst exam (baseline) was 5 days and to second exam (follow-up) was 189 days. During the study period, no patient has developed a graft loss or a cardiovascular event (acute myocardial infarction, acute coronary syndrome or arrhythmias). Control group was composed by 10 age-and sex-matched healthy controls selected from the hospital records. Table 1 describes the subject characteristics for KT (baseline and follow-up) and control groups.

CMR parameters
There were no signi cant changes in LV volumes, mass or EF. The mean native T1 decreased from 1331 ± 52 ms to 1298 ± 42 ms at 6 months (p < 0.001), but still remained higher than controls (1256 +/-33 ms, p = 0.005) [ Fig. 2]. There was no change in T2 times, suggesting that reduction in native T1 was probably related to regression of myocardial brosis, not edema. Also, there were no signi cant changes in right ventricular (RV) volumes or EF. Table 2 (Fig. 5). There were no signi cant correlations between LV GLS, T1 or T2 measurements (p > 0.05 for all). Also, we did not nd associations between changes in LV GLS with changes in serum levels of creatinine, heart rate or blood pressure (see Table 4). On univariate regression analysis none of the clinical variables examined (age, gender, body mass index (BMI), diabetes, dialysis vintage, type of donor and time after transplantation the rst exam was done) were determinants of changes in LV GLS (Table 5).

Intraobserver reproducibility and validation of LV GLS measurements
The Bland-Altman analysis for intraobserver reproducibility of LV GLS, measured by FT-CMR, showed a bias of 0.2 and a con dence interval (95% CI) of -1.3 to 1.7 (Fig. 6). There was a strong positive correlation (r = 0.84) between the LV GLS values measured by FT-CMR and tagging (Fig. 7).

Discussion
In this prospective study we demonstrated a favorable impact of KT on LV GLS measured by FT-CMR 6 months after surgery, in concordance with previous echocardiographic studies [17,18]. Although there was an improvement in LV GLS and a reduction of myocardial brosis as assessed by T1 mapping after KT, these important prognostic CMR biomarkers do not reach the normal range values characterized in controls, which could help to explain why KTR are still at increased cardiovascular risk. Yet, improvement in LV GLS was associated with reduction of left ventricular hypertrophy (LVH), but not myocardial brosis (T1) or edema (T2).
FT-CMR derived LV GLS is a powerful independent predictor of mortality in patients with ischemic or nonischemic dilated cardiomyopathy [23], reduced [24] and preserved [25] ejection fraction, incremental to common clinical and imaging risk factors. In studies using STE, LV GLS was independently associated with both all-cause and cardiovascular mortality also in CKD, ESRD and KTR [26][27][28]. To the best of our knowledge, data regarding changes in strain parameters using FT-CMR are limited to the study by Gong et al. [16] who found that LV GCS and GRS, but not GLS, improved 1 year after KT. This apparent discrepancy probably is due to the time after transplant surgery data were collected, supporting the hypothesis that GLS improves earlier in the course of recovering subclinical myocardial dysfunction. Also, besides no prior studies investigated the temporal sequence of improvement in LV GLS after KT, Enrico et al. [29] showed that LV GLS improved from baseline to 6 months after KT, but remained unchanged in the next 6 months, until 1 year after surgery. Most progressive myocardial diseases predominantly cause subendocardial dysfunction in their early stages, leading to reduction in longitudinal LV mechanics [11].
Transmural involvement results in concomitant subendocardial and subepicardial dysfunction, decreasing myocardial contractility in all directions, with impairment of LV ejection performance [30]. Explanations about why subendocardium is the most vulnerable region include it is the furthest layer from epicardial coronary ow, it undergoes greater variations in pressure and compression in both systole and diastole, and also appears to be more susceptible to early microvascular and structural changes such as brosis [31]. So, it is expected that GLS also will recover earlier than other strain parameters after effective cardiac treatment. Yet, different from us, Gong et al. [16] found signi cant improvement in LVEF 1 year after KT, and that was correlated with improvement in LV GCS and GRS. This nding could be explained because the contribution of midwall circumferential shortening has a greater impact on LV SV and LVEF than longitudinal shortening [32]. In the present study, although we found a signi cant correlation between improvements in LV GLS with others parameters of systolic function (GCS, GRS and EF), LVEF showed a slight, non-signi cant, increase in the follow-up.
Strain measures are susceptible to changes in preload and afterload [33], but in our study, we did not observe signi cant correlations between LV GLS with end-diastolic volume index (surrogate for preload) or blood pressure (surrogate for afterload), suggesting that improved LV systolic function is likely attributed to other effects of KT, including the removal of uremic toxins, reversal of LVH, and restoration of in ammation and oxidative stress state [34], rather than loading conditions. Cardiac remodeling, as assessed by LV mass and geometry, is also a strong predictor of cardiovascular and all-cause mortality in studies of asymptomatic populations [35] and in CKD [3]. In the other way, reverse remodeling after KT is associated with better outcomes in patients with cardiac dysfunction [6]. However, there are discrepancies between STE and CMR exams regarding the effects of successful KT on LVH, a common feature of UC. Similar to Patel et al.
[36] we did not observe signi cant regression in LVMi 6 months after KT, although improvements in LV GLS were associated with reductions in LVMi. Indeed, Stokke et al. [37] demonstrated that LVH can compensate impaired LV GLS for maintaining LVEF, so it is expected that improvement in LV GLS will contribute to reductions in LVH and vice versa, while LVEF remains constant.
LVH may be associated with cardiac brosis that leads to conduction disturbances and probably provides the link between UC, arrhythmia and sudden death. The use of newer T1 mapping techniques without gadolinium (native T1) has demonstrated increased myocardial T1 relaxation times indicative of diffuse interstitial brosis in ESRD [13,14] and early-stage CKD [12]. Although KT was associated with reduced septal native T1 [15], probably related to reduction in diffuse interstitial brosis, in this study we failed to prove associations between improvements in LV GLS with reductions in native T1. It has been suggested that changes in myocardial systolic and diastolic deformation are functional markers of diffuse interstitial brosis. Previous experimental study by Kramann et al. [38] reported that strain parameters not only detected LV contractile abnormalities but also correlated with the severity of interstitial myocardial brosis and hypertrophy in rat models with uremic cardiomyopathy. On the other hand, in a recent clinical study, Frojdh et al. [39] demonstrated that myocardial interstitial brosis assessed by expansion of extracellular volume fraction (ECV) and GLS were both associated with outcomes, but they correlated minimally, suggesting that diffuse myocardial interstitial expansion and contractile dysfunction may re ect different domains of myocardium disease.
Our study has some limitations. First, the baseline CMR was performed after transplantation surgery. Although performing CMR at the pre-transplant period would be more appropriate, this is not possible with deceased donors due to long waiting list times. However, all patients were clinically and hemodynamically stable at the moment of the CMR exam. Second, we did not perform histological con rmation of myocardial brosis because of the inherent risks and decreasing use of endomyocardial biopsy in daily practice, however T1 mapping is a well validated CMR technique to this end. Third, this is a single center study with small number of patients, and may have been underpowered to show associations between native T1 and GLS. Fourth, we did not measure serum biomarkers of heart failure, however, previous study has found associations between B-type natriuretic peptide (BNP) and those CMR indices in hemodialysis patients (40)]. Finally, outcomes assessment was not feasible, because of the limited follow-up duration.
Overall, our ndings reinforce the potential role of multiparametric CMR in monitoring important biomarkers of UC in ESRD patients after KT. Early detection of individuals at higher risk for CVD after KT could allow identi cation of those who might bene t from closer cardiovascular follow-up or more aggressive therapies. Future studies with increased number of patients and longer follow-up are needed to determine which of these CMR parameters will be helpful in predicting mortality or guiding therapy.

Conclusion
In this prospective study we demonstrated that LV GLS measured by FT-CMR improves 6 months after KT in association with reverse remodeling but not native T1 or T2 measurements.

Competing interests
The authors declare that they have no competing interests.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.   Data expressed as mean and standard deviation (mean ± standard deviation).
* Indicates p value < 0.05 between paired data at baseline and transplant follow-up. † Indicates p value < 0.05 between unpaired data comparing the transplant cohort at follow-up and healthy controls.