Study population
Twenty-two kidney transplant candidates with ESRD who successfully underwent CFVR assessment at the Queen Elizabeth Hospital, Birmingham (QEHB), United Kingdom between March 2019 and March 2020 were included in this analysis. These individuals were research participants in the Chronic Renal Impairment in Birmingham Coronary Flow Reserve (CRIB-FLOW) study or the Prospective Study of the Effects of Renal Transplantation on Uraemic Cardiomyopathy using Magnetic Resonance Imaging (RETRACT) echocardiogram sub-study, both of which examined CFVR in patients with ESRD.
Participants were > 18 years old, considered suitable for kidney transplantation by the renal transplant team at QEHB, had estimated glomerular filtration rate (eGFR) < 15 ml/min/1.73 m2 and were pre-dialysis or on peritoneal dialysis (PD). Exclusion criteria were: pregnancy, haemodialysis (HD), diabetes mellitus, uncontrolled hypertension, known ischaemic heart disease, moderate/severe valvular heart disease and contraindication to adenosine or sulphur hexafluoride contrast agent (SonoVue, Bracco, Milan, Italy).
Blood pressure
Office blood pressure (BP) was measured using an automated BP monitor (BpTRU, VSM Medtech, Coquitlam, BC, Canada), which takes 6 BP readings over 6 min. After exclusion of the first reading, an average of the remaining 5 readings was used to represent office BP.
Transthoracic echocardiography (TTE)
All subjects underwent comprehensive two-dimensional echocardiography by a British Society of Echocardiography accredited physician (AR). Studies were performed on a Philips iE33 machine (Philips, Eindhoven, Netherlands) using a S5-1 transducer for TTE and myocardial contrast echocardiogram (MCE) studies and a S8-3 transducer for CFVR measurements. Echocardiograms were stored under an anonymous code and analysed offline by a single investigator (AR) using commercially available software (IntelliSpace Cardiovascular, Philips, Eindhoven, Netherlands).
Left ventricular mass was estimated using the Cube formula and indexed for body surface area [11]. The Simpson’s biplane method was used to measure left ventricular volumes and ejection fraction [11]. Diastolic function was quantified using multiple parameters [12]. Global longitudinal strain (GLS) was assessed in the 3 standard apical views using speckle tracking.
Doppler coronary flow velocity reserve
Doppler CFVR assessment was performed as previously described [13]. The left anterior descending artery (LAD) was identified on colour Doppler in the anterior inter-ventricular sulcus. Pulse wave Doppler signals of LAD flow were recorded to measure coronary flow velocity (CFV) at rest and at hyperaemia. SonoVue was used, if necessary, to identify LAD flow and to improve the spectral Doppler trace. Hyperaemia was induced by an infusion of adenosine at a rate of 140micrograms/kg/min for 3 min. Subjects were advised to abstain from caffeine for 24 h prior to adenosine administration. CFVR was calculated as hyperaemic CFV/rest CFV. For each variable in the CFVR calculation, the highest values of 3 cardiac cycles were averaged.
Myocardial contrast echocardiography
Myocardial contrast echocardiography was performed as previously described [13]. Images were taken in the 3 apical views using low-power continuous MCE at a mechanical index (MI) of 0.1. Sonovue was continually infused using an oscillating infusion pump that maintains microbubbles in suspension (Vueject, Bracco, Milan, Italy). The infusion rate was started at 70-100ml/hr but adjusted to ensure sufficient myocardial opacification without excessive contrast attenuation. Triggered high MI (1.0) flash echocardiography was performed at end-systole, where the myocardium is at its thickest, to destroy microbubbles in the myocardium and to observe replenishment. The sequence was initially performed at rest and then repeated after adenosine vasodilator stress as above. The absence of regional wall motion abnormalities or sub-endocardial perfusion defects on vasodilator MCE was deemed sufficient to exclude flow limiting CAD.
Laboratory analysis
N-terminal pro-brain natriuretic peptide (NTpro-BNP) was assayed using the Alere point of care assay (Alere, Massachusetts, USA). High sensitivity C-reactive peptide was assayed using the Architect MULTIGENT CRP Vario assay (Abbott, Illinois, USA). The remaining laboratory parameters were assayed using standardised automated methods. The fluorescence responses of 16-analytes of inflammation, atrial stretch, cardiac fibrosis, kidney injury and LVH were obtained using Human Magnetic Luminex® Asssays (R&D Systems, Minneapolis, MN, USA) and the Bio-RAD Bio-Plex™ 200 system for analysis. Concentrations were calculated using the Bio-Plex Software Manager™ (version 6.1) generated standard curves and a 5PL logistic curve fitting technique as per the manufacturer’s instructions.
Statistical analysis
Statistical analysis was performed using SPSS version 26 (SPSS Inc, Chicago, Illinois). The Shapiro–Wilk test was used to assess data normality. Continuous variables are expressed as mean ± standard deviation for parametric data or median (interquartile range—IQR) for non-parametric data. Unpaired group comparisons for continuous data were made using the unpaired t-test or the Mann-Whitney U test. Unpaired categorical data were compared using Fisher’s exact test. Correlation was assessed using the Pearson correlation coefficient. Univariable and multivariable linear regression models were performed with CFVR as the dependent variable. Factors known to influence CFVR (age, systolic BP, left ventricular mass index) as well as markers of anaemia (haemoglobin, iron), bone mineral disease [calcium, phosphate, parathyroid hormone (PTH)] and inflammation (high sensitivity C-reactive peptide, tumour necrosis factor-α, interleukin-6, interleukin-8, interleukin-10) were included as independent variables in regression models. Binary logistic regression was also performed, with CFVR < 2 as the dependent variable, and the parameters listed above as independent variables. Parameters that were significant in univariable analysis were entered into multivariable regression models. A variance inflation factor > 5 was taken to represent collinearity. Statistical tests were 2-tailed, and a p value < 0.05 was considered statistically significant.