Changes in left ventricular structure and function associated with renal transplantation: a systematic review and meta‐analysis

Abstract Aims This study aimed to examine if the cardiac changes associated with uraemic cardiomyopathy are reversed by renal transplantation. Methods and results MEDLINE, Embase, OpenGrey, and the Cochrane Library databases were searched from 1950 to March 2020. The primary outcome measure was left ventricular mass index. Secondary outcome measures included left ventricular dimensions and measures of diastolic and systolic function. Studies were included if they used any imaging modality both before and after successful renal transplantation. Data were analysed through meta‐analysis approaches. Weight of evidence was assessed through the Grading of Recommendations Assessment, Development and Evaluation system. Twenty‐three studies used echocardiography, and three used cardiac magnetic resonance imaging as their imaging modality. The methodological quality of the evidence was generally poor. Four studies followed up control groups, two using cardiac magnetic resonance imaging and two using echocardiography. Meta‐analysis of these studies indicated that there was no difference in left ventricular mass index between groups following transplantation {standardized mean difference −0.07 [95% confidence interval (CI) −0.41 to 0.26]; P = 0.67}. There was also no difference observed in left ventricular ejection fraction [mean difference 0.39% (95% CI −4.09% to 4.87%); P = 0.86] or left ventricular end‐diastolic volume [standardized mean difference −0.24 (95% CI −0.94 to 0.45); P = 0.49]. Inconsistent reporting of changes in diastolic dysfunction did not allow for any meaningful analysis or interpretation. Conclusions The evidence does not support the notion that uraemic cardiomyopathy is reversible by renal transplantation. However, the evidence is limited by methodological weaknesses, which should be considered when interpreting these findings.


Introduction
Over half of deaths in end-stage kidney disease (ESKD) are due to cardiovascular disease; the age-corrected relative risks are extreme, reaching over 100-fold in younger patients. 1 The majority of these deaths are not due to myocardial infarction as a result of coronary atheroma but due to heart failure and sudden cardiac death. [2][3][4] Consistent with this observation, treatments for traditional cardiovascular risk factors such as hypertension and elevated cholesterol are relatively ineffective in this population. [4][5][6] These observations can be explained by the near-universal syndrome of uraemic cardiomyopathy in patients with ESKD. 7,8 Left ventricular hypertrophy is the cardinal feature of uraemic cardiomyopathy, in addition to ventricular dilatation and both systolic and diastolic dysfunction. Histologically, myocytes are severely hypertrophied with myocardial disarray and diffuse interstitial fibrosis. 9 As renal function declines, these features become more prevalent and are present in up to 90% of those requiring renal replacement. 10 Such changes are strongly linked to cardiovascular outcomes with the presence of left ventricular hypertrophy associated with increased mortality in both transplant recipients and those requiring haemodialysis. 11,12 The gold standard for the treatment of ESKD is renal transplantation. 10 The associated improvement in glomerular filtration rate reduces cardiovascular risk below that of those on waiting lists. 13 However, cardiovascular risk still remains higher than healthy individuals of the same age and sex with transplant recipients displaying a three-fold increased risk. 14 The restoration of renal function associated with renal transplantation improves many factors thought to cause uraemic cardiomyopathy. As a result, it is generally assumed that kidney transplantation reduces left ventricular mass index (LVMI) and volumes and improves diastolic and systolic function. 15,16 This assertion is based on the reduction of LVMI reported in small echocardiographic studies. 15,16 However, echocardiography is not a reliable or reproducible method for the measurement of LVMI, especially when there are large changes in loading such as before and after haemodialysis. 17 As a result, cardiac magnetic resonance imaging (CMR) is now accepted as the gold-standard imaging modality for patients with ESKD. 7 Despite this, review articles continue to state that uraemic cardiomyopathy is reversed by renal transplantation. These articles will not cite any references, cite small, uncontrolled studies using either echocardiography or radionucleotide ventriculography-gated blood pool (multigated acquisition scan) scans, or refer to other review articles. 8,15,[18][19][20] The aim of this study was to perform the first systematic review and meta-analysis to establish if the features of uraemic cardiomyopathy are reversible following successful renal transplantation.

A Preferred Reporting Items for Systematic Reviews and
Meta-Analyses-compliant systematic review was conducted and was registered with the International Prospective Register of Systematic Reviews (PROSPERO; http://www.crd.york. ac.uk/prospero/, Reference CRD42018115359). 21 Published and unpublished articles and conference proceedings registered on or before 20 March 2020 were searched. The electronic databases used to search the published literature were MEDLINE, Embase, OpenGrey, and the Cochrane Library (clinical trials database and database of systematic reviews). All searches were limited to adult human studies. Reference lists of all pertinent review papers and eligible studies were reviewed. The search terms used are presented for the MEDLINE search in Table 1. These were modified for the specific databases searched.

Inclusion criteria
All full-text English-language articles assessing changes in LVMI, before and after successful renal transplant, using any form of imaging technique were included. Single-subject case reports, comments, letters, editorials, guidelines, or review papers were excluded. Studies were also excluded if participants received more than one organ type.

Study selection
Two reviewers (L.C.P. and J.P.L.) independently reviewed all titles and abstracts generated from the search strategy. Following this initial screening process, the full texts of eligible articles were reviewed independently by each author against the predefined eligibility criteria.

Critical appraisal
All papers were critically appraised independently by two reviewers (L.C.P. and A.R.). This appraisal was conducted using the Newcastle-Ottawa Scale. 22 A maximum score of 9 points can be awarded based on participant selection, comparability, and study outcome including follow-up.
The Grading of Recommendations Assessment, Development and Evaluation (GRADE) system was adopted to evaluate the quality of the evidence across studies for pooled analyses. 23 Outcome measures and data extraction Two reviewers (L.C.P. and J.P.L.) extracted data into a pre-constructed table. Information gathered included number of participants, age range, sex distribution, dialysis modality, immunosuppression regime, and time to follow-up after transplantation. The primary outcome measure was LVMI. Secondary outcome measures were left ventricular dimensions, measures of diastolic and systolic function. Any disagreement regarding study eligibility, data extraction, methodological quality, and GRADE assessment between reviewers was resolved through discussion until consensus was reached.

Statistical analysis
Statistical analysis was conducted using Review Manager 5.0 for Apple (Nordic Cochrane Centre, Copenhagen, Cochrane Collaboration, 2008). Statistical heterogeneity was assessed by χ 2 and I 2 . If χ 2 was greater than P = 0.10 and the I 2 statistic indicated that heterogeneity was present (>20%), a random-effects statistical model was adopted to calculate mean difference or standardized mean difference (SMD) between groups. When χ 2 and I 2 values demonstrated low heterogeneity, a fixed-effects model was adopted. 24 Where meta-analysis was not possible because of insufficient data, a narrative approach was adopted.

Search strategy results
The results of the search strategy are summarized in Figure 1. A total of 2547 potentially relevant citations were identified, with 26 being eligible for inclusion. The characteristics and outcomes of the 26 included studies are presented in Table 1.

Methodological appraisal
There were 23 studies that used echocardiography.  Of these studies, eight were retrospective echocardiographic data collected as part of routine clinical practice. 25,31,34,37,38,[45][46][47] Their methodological quality was  Table 1). One study was classified as fair, 35 and none were classified as good. Two studies recruited control groups, which were followed up, both consisting of individuals receiving haemodialysis. 27,35 Assessor blinding was only employed in two studies, 28,30 and none used a sample size calculation.
Three studies employed CMR, two were classified as good 48,49 and one as fair. 50 In each, recipients were recruited from local transplant waiting lists. No study performed a sample size calculation designed to detect change in LVMI. Prasad et al., 48 however, used a sample size calculation powered to detect changes in adiponectin levels. Assessor blinding was employed in all CMR studies. [48][49][50] In one study, the indication for initial CMR was routine clinical practice, 49 and in the remaining two, CMR was conducted for research purposes. 48,50 The length of follow-up across all the studies varied from 1 week to 5 years, with the most common follow-up time point being 12 months.

Study population
In total, 1998 renal transplant recipients were included of which 1229 were male. The pooled weighted mean age was 50 years (range 16-85 years). Fourteen studies reported type of transplant with a total of 840 live donor transplants and 377 deceased donor transplants. 25,27,29,31,[34][35][36]38,40,[42][43][44]48,50 A total of 1531 recipients were reported to be receiving renal replacement therapy in 24 studies. [25][26][27][29][30][31][32][33][35][36][37][38][39][40][41][43][44][45][46][47][48][49][50] In total, 127 control patients were followed up in four studies. 27,35,48,49 Two CMR 48,49 studies recruited both recipients and controls from local transplant waiting lists. Comparisons between the groups at baseline showed that there was no difference in age, sex, systolic blood pressure, or history of ischaemic heart disease. Two echocardiographic studies also recruited controls. De Lima et al. 27 recruited 74 unselected ESKD patients on regular haemodialysis, 17 who were subsequently transplanted. There was no significant difference between those transplanted and those who remained on dialysis in terms of age, gender, race, or duration of haemodialysis. No data regarding blood pressure or prior cardiac disease were presented. Keven et al. 35 also recruited both transplant recipients and randomly selected controls receiving haemodialysis. There were no significant differences reported between recipients and controls in terms of age, sex, and systolic blood pressure. There was also no recorded ischaemic heart disease in either group. Further two studies recruited controls who were studied at a single time point; in both cases, however, these were healthy controls. 36,50 Findings of all studies are summarized in Table 2.
Aetiology of ESKD was reported in 12 studies, 25,27,[29][30][31]33,35,39,41,47,48,50 with glomerulonephritis being the most commonly reported aetiology. Five studies excluded patients with ischaemic heart disease or congestive cardiac failure, 25,28,30,39,50 and a further two 32,41 only included patients who were asymptomatic from cardiovascular disease. Prasad et al. 48 reported that 10% of transplanted patients had undergone coronary revascularization. Vaidya et al. 46 reported that 43% of their cohort had a diagnosis of coronary artery disease, and McGregor et al. 37 indicated that 84% of participants had a dilated cardiomyopathy at baseline. The cohort reported by Hawwa et al. 47 included 26% with coronary artery disease and 31% with a prior diagnosis of heart failure.
All three CMR studies reported no significant overall change in LVMI. [48][49][50] However, the trends in mean change were conflicting. Patel et al. 49 observed an increase in LVMI in transplant recipients and a decrease in the control group of haemodialysis patients (À3.6 ± 16.7%/year vs. 2.75 ± 9.1%/year). Prasad et al. 48 reported a reduction in LVMI in both recipients and controls who remained on the waiting list (recipients À1.98 ± 5.5 g/m 2 and controls À0.36 ± 5.7 g/m 2 ; P = 0.44). The third CMR study by Hayer 2048 L.C. Pickup et al.     1 year. There was a significant reduction of LVMI in transplant recipients from baseline to follow-up. However, the magnitude of change was not significantly different from that observed in the control group (recipients À10 ± 24 g/m 2 vs. controls À5.6 ± 22 g/m 2 ; P > 0.05). De Lima et al. 27  The two CMR studies reported no significant overall change in LVMI following transplantation compared with the control group. 48,49 A meta-analysis was conducted of the four studies reporting change in LVMI in transplant recipient and controls ( Figure 2). A total of 236 participants were included in this analysis; the overall SMD was À0.07 (95% CI À0.41 to 0.26), P = 0.67, suggesting no difference between transplant and control groups. However, heterogeneity was moderate (I 2 = 38%). This was regarded as low-quality evidence using the GRADE approach due to low participant numbers and heterogeneity between the studies included. Subgroup analysis is also presented based on imaging modality. The two echocardiographic studies 27,35 [SMD À0.20 (95% CI À0.60 to 0.20); P = 0.33] and the two CMR studies 48,49 [SMD 0.07(95% CI À0.67 to 0.80)] showed no mean change in LVMI. There was no significant difference between the findings of the two imaging modalities (P = 0.53). However, heterogeneity in the echocardiographic subanalysis was low (I 2 = 0%) but substantial (I 2 = 77%) in the CMR subanalysis.
One CMR study by Hayer et al. 50 reported a significant improvement in LVEF from baseline to follow-up (ejection fraction 68 ± 9% to 73 ± 9%; P < 0.05). However, when comparing changes to control participants with ESKD, both Patel et al. 49 and Prasad et al. 48 reported no statistically significant change in LVEF. Meta-analysis of these two studies, consisting of 64 transplant recipients and 68 control participants receiving regular dialysis, showed no overall change in LVEF in transplant recipients compared with controls [mean difference 0.39% (95% CI À4.09% to 4.87%); P = 0.86] with high heterogeneity (I 2 = 62%) (Figure 3). The quality of evidence (GRADE) was rated as very low quality due inconsistency between the results, the low numbers of trials included, and overall participant numbers.

Left ventricular dimensions
The most reported measure was left ventricular internal diameter in diastole in 13 non-controlled echocardiographic studies with all but three reporting a significant reduction. 25,29,[31][32][33][34]36,37,[40][41][42][43]47 All three CMR studies   (Figure 4). The quality of evidence (GRADE) was rated as very low quality; this was again due to inconsistency between the results of the two included trials and low participant numbers.

Diastolic dysfunction
The most reported parameter of diastolic dysfunction was E/A ratio with three studies reporting statistically significant changes following transplantation. 25,27,28 One controlled study by De Lima et al. 27 reported a small reduction in E/A ratio (1.42 ± 0.6 to 1.10 ± 0.4; P < 0.05) at 1 year follow-up, whereas Deng et al. 28 reported a small increase (1.04 ± 0.57 to 1.21 ± 0.52; P = 0.001). An et al. 25 reported that recipients with moderate diastolic dysfunction (Grade 2) before transplantation showed a significant reduction in E/A ratio at 12 months (baseline 1.13 vs. 0.98; P < 0.05), whereas those with mild dysfunction (Grade 1) only exhibited a significant change at the 5 year follow-up (baseline 0.72 vs. 0.81 at 5 years; P < 0.05). 34 Mitral valve deceleration time was also reported in four studies, 28,29,31,42 with one study reporting a small significant increase 42 and one a small significant decrease. 28 Neither of these changes represented a change in the grade of diastolic function observed.

Discussion
Reversing uraemic cardiomyopathy is potentially the key to reducing cardiovascular morbidity and mortality in ESKD. Although no targeted therapy has been shown to achieve this, it is generally assumed that restoration of kidney function by kidney transplantation reverses the changes observed. At present, however, the evidence does not support this.
We have shown that the majority of uncontrolled echocardiographic studies reported significant reductions in LVMI. However, making conclusions based on these data is problematic. Echocardiography is unreliable when measuring LVMI due to inaccuracy where large volume fluctuations occur. 17 CMR is more accurate and reproducible and is accepted as the gold-standard imaging modality for patients with ESKD. 7 None of the three CMR studies included in our review found a significant change in LVMI. Furthermore, in a meta-analysis of the four available studies with control groups, renal transplantation was not associated with any reduction in LVMI, and subgroup analysis indicated that this finding was not affected by imaging modality. This analysis also clearly highlights that none of the controlled studies, regardless of imaging modality, reported significant changes in LVMI following transplantation.
A similar pattern was also observed in left ventricular function, with the majority of echocardiographic non-controlled longitudinal studies reporting significant improvements in  LVEF following transplant. This finding is also supported by the work of Wali et al. 51 where 102 transplant recipients with left ventricular dysfunction showed significant improvement at 1 year when assessed with radionuclide ventriculography. However, this study was not included in the systematic review as LVMI was not considered. Among the three CMR studies, the patterns of change observed were conflicting. Hayer et al. 50 report a significant change in recipients from baseline to follow-up, whereas both Patel et al. and Prasad et al. did not. In addition, there was also no convincing evidence that successful renal transplantation improves diastolic left ventricular function. It would, therefore, appear that the assumption that the features of uraemic cardiomyopathy are reversed by successful renal transplantation is not supported by the current published literature.
Before concluding that uraemic cardiomyopathy is irreversible, it is important to examine the quality of the evidence available. Studies were generally classified as poor with only two rated as good and two as fair using the Newcastle-Ottawa scoring system. In addition, the assessment of the evidence across studies for each comparison (GRADE) ranged from 'very low quality' to 'low quality'. The majority were opportunistic and unblinded, with little attempt to reduced risk of systematic bias. Only four studies, comprising a total of 109 transplant recipients, recruited a suitable control group, which was followed up. 27,35,48,49 A further limitation was the lack of sample size justification with no studies powered to detect a change in LVMI. Previous work, however, has indicated that to detect a change in left ventricular mass of 10 g with 90% power using 2D and M-mode echocardiography, 78 and 162 participants would be required, respectively. As a result, only five echocardiographic studies included in the review can be considered to have sufficient power to reliably detect clinically significant changes in left ventricular mass. 52 The number required to detect the same change using CMR is much smaller, with only 13 participants required indicating that all three CMR studies recruited adequate numbers of participants. 52 The fact that only three CMR studies have been conducted, with a total of 88 transplant recipients included, is a major weakness of the current evidence base.
The meta-analyses also demonstrated high heterogeneity, suggesting that the currently available studies do not reliably answer the question of whether uraemic cardiomyopathy is reversible. Some of this may be explained by many studies appearing to have an opportunistic design that is examining patients that happened to have a heart scan performed before and after transplantation with consequential bias, rather than being prospectively designed.
While there are weaknesses in the evidence base, it may also be true that uraemic cardiomyopathy is not reversible. Indeed, the presented meta-analysis looking at controlled studies, including those using the gold-standard technique of CMR, suggests that this might well be the case. Following renal transplantation, many traditional risk factors for cardiovascular disease persist and in some cases may develop de novo. 53 Hypertension, dyslipidaemia, and diabetes are all recognized complications of both steroids and calcineurin inhibitors, which are routinely administered following transplant. In addition, there is also persistence of non-traditional risk factors including uraemia, proteinuria, and chronic inflammation. 53 Transplantation cannot fully reverse these factors, which may explain the persistence of uraemic cardiomyopathy.
Our study has several strengths in that it included data from both echocardiography and CMR studies, which enabled all relevant data pertaining to the subject to be incorporated. The number of studies identified ensured that there was a significant pooled sample size on which conclusion could be based, although with 26 studies identified, the number of participants was only 1998, highlighting the fact that many studies were very small. There were, however, significant limitations. While there were an appropriate number of studies included in the systematic review, the number suitable for meta-analysis was small with only four studies eligible. There were also moderate levels of heterogeneity noted among the studies when meta-analysis was undertaken. Subsequent sensitivity analysis suggested that this was being driven by the conflicting findings of the CMR studies. Such heterogeneity can make the interpretation of any findings difficult. However, we took the view that demonstrating this variability between studies highlights the need for further work to be conducted in this area. Another limitation of the studies included is the short length of follow-up. It therefore cannot be concluded that uraemic cardiomyopathy might be reversible in the longer term with no study having more than 12 month follow-up. Furthermore, because we used the assessment of LVMI as the primary selection criteria, studies looking at other important features such as longitudinal strain and right heart changes were not systematically examined.

Conclusions
Reversing uraemic cardiomyopathy is a potential target for reducing the cardiovascular morbidity and mortality associated with chronic kidney disease. This syndrome has generally been assumed to be reversible by renal transplantation. Our review has highlighted that at present, it is unclear if this is true.
This review also highlights the need for adequately powered and controlled studies to answer this fundamental question and provides further insights into other potential strategies to reverse uraemic cardiomyopathy and improve the increased cardiovascular risk associated with ESKD. Cardiac remodelling following renal transplant