Cardiovascular Effects of Unilateral Nephrectomy in Living Kidney Donors at 5 Years

Supplemental Digital Content is available in the text.


135
Chronic kidney disease (CKD) is an independent risk factor for cardiovascular morbidity and 136 mortality. 1 There is an inverse association between estimated glomerular filtration rate 137 (eGFR) and cardiovascular risk. 1 This risk remains elevated even after adjusting for 138 comorbidities such as hypertension and diabetes. 1, 2 Whilst the threshold eGFR at which 139 cardiovascular risk rises is debatable many studies have found that risk increases significantly however, reported an increase in cardiovascular mortality with a hazard ratio of 1.40 153 compared to healthy controls raising concern about the long term safety of kidney donation. 11 154 Furthermore, the Chronic Renal Impairment in Birmingham (CRIB)-DONOR study 155 (NCT01028703) highlighted potentially important short term adverse changes in 156 cardiovascular structure and function. 12 Compared to controls, donors at 12 months after to follow up the same cohort at 5 years to examine the medium term effects of kidney 160 donation on cardiovascular structure, function and hemodynamics. Late gadolinium enhancement (LGE) was defined based on definitions previously 218 described. 17 Quantification was made using full width half max methodology. 18 For 219 assessment of T1, T2 and ECV, the myocardium at the mid ventricular slice was segmented 220 into American Heart Association (AHA) segments and global values were calculated as an 221 average of the valid segments see Figure S1. 19     Continuous variables were assessed graphically using histograms to determine normality. Non-268 parametric data was log10 transformed and assessed graphically. For continuous data, within-group change from baseline to 12 months and baseline to 5 years was analyzed using paired 270 samples t tests. Between-group difference was analyzed using independent samples t tests to 271 compare within-group change at 5 years between groups and generate the p value for the 272 primary end point. Non-parametric data was analyzed in a logged format then anti-logged and 273 displayed as multipliers. Categorical data are displayed as counts and percentages, between-274 group changes are displayed as relative risks and 95% confidence intervals and analyses were 275 performed using MedCalc for Windows, version 19.4 (MedCalc Software, Ostend, Belgium)..

276
Interactions between each variable and donor/control status were determined by general linear 277 models. Multivariable model analysis was carried out using linear regression and incorporating 278 any significant interactions. An interval censored cox regression was used for analysis of the 279 combined blood pressure endpoint using the icenReg package in R. T1, T2 and ECV, which 280 were measured at 5 years only, were analyzed using independent samples t tests.

281
Reproducibility was assessed using intraclass correlation coefficients.

295
Records from all 124 subjects who took part in the original study were reviewed. Of these, 1 296 had died of bronchial carcinoma and 3 were not contactable; 120 were approached; 50 kidney 297 donors and 45 healthy controls agreed to participate, see Figure 1. There were no significant 298 differences in baseline demographics between those who attended follow up at 5 years and 299 those that were lost to follow up other than a cardiovascular family history, see Table S1. 300 One kidney donor and one healthy control declined a CMR study. Nine subjects did not 301 undergo a baseline CMR study, therefore, there were 42 kidney donors and 42 controls with 302 paired sets of end-point data (baseline and 5 year CMR data). Three subjects had 303 contraindications to 3T CMR and had 1.5T scans using the same protocol.

305
Data are presented in Table 1. One control subject was diagnosed with diabetes and one with 306 ischaemic heart disease. There was an increase from baseline in the prevalence of self-307 reported hypertension in kidney donors (4% to 16%) with little change in controls (7% to 308 9%). At 5 years the proportion of donors and controls on anti-hypertensive medication was 309 not different between groups.

320
Effects on LV mass, volumes, geometry and function 321 At 5 years, change in LV mass in kidney donors were not different to healthy controls +0.40g 322 (95% CI -4.68, 5.49), p=0.876, see Table 2 and Figure 2Aa. 323 There was no significant difference in the changes in LV or left atrial (LA) volumes indexed 324 for body surface area, LV geometry, GLS or GCS at 5 years, see Table 2.  Table 2. LGE at the right ventricular insertion points was seen in 4 living kidney donors (% 331 of LV mass, 0.87± 0.15%) and in one control. There was no left ventricular myocardial 332 LGE.

334
There were no between-group difference in office blood pressure or heart rate at 5 years, see 335 Table 3. Compared to baseline, office systolic blood pressure fell in both groups at 5 years.

336
Ambulatory and central blood pressures however increased in both groups over time but were 337 not significantly different between-groups at 5 years. The proportion of subjects with a 338 diagnosis of hypertension (on ambulatory blood pressure monitoring criteria) showed no 339 significant differences, see Table 3. A further sub analysis using a composite end-point of clinically significant increases in blood pressure also showed no significant differences 341 between the two groups. The hazard ratio (HR) for hypertension using the combined outcome 342 in donors relative to controls was increased but not significant, [hazard ratioHR 1.38 (95% 343 confidence interval 0.74, 2.60), p=0.313], see Supplemental methodology 4. Carotid IMT at 344 5 years was greater in donors vs. controls but had not increased significantly from previous 345 values. 346 At 12 months, there was an increase from baseline in PWV in kidney donors, which was not 347 seen in controls. From 12 months to 5 years, PWV increased in both groups and by 5 years 348 the between-group difference was not significantly different, see Figure 2Bb. A similar 349 pattern was observed in AI75 in which there was a small increase in kidney donors at 12 350 months compared to a fall in healthy controls. AI75 at 5 years was not significantly different 351 between kidney donors and controls. Aortic distensibility in the proximal ascending and 352 descending aorta decreased in both groups over time with no between-group difference.

354
Biochemical data are given in Table S3 There was an increase in high sensitivity C -reactive 355 protein, high sensitivity troponin T and vitamin D over time in both donors and controls. At 356 12 months, the prevalence of detectable troponin T was greater in donors than controls; at 5 357 years the prevalence had increased in both groups reducing the between-group difference. 12

358
Serum urate and FGF23 were higher in donors than controls at 5 years.  Table S4. There was no significant influence of sex or LV mass at baseline on change in LV 363 mass at 5 years. Change in ambulatory systolic blood pressure however, was significantly associated with change in LV mass. None of the other variables were significant when 365 included in a multivariable model with change in day systolic blood pressure.

366
Reproducibility for primary end-point 367 There was high reproducibility for LV mass assessment. The interclass correlation 368 coefficients (95% confidence intervals) for inter-study, intra-observer and inter-observer  Table S5).  In the first CRIB-DONOR study, there was a significant increase in LV mass in 396 kidney donors compared to healthy controls at 12 months. 12 These results were confirmed by 397 a later small uncontrolled study of 23 kidney donors. 31 Our latest results suggest that these 398 changes resolve over time. The reasons for these fluctuations are unclear. Effects due to 399 random chance cannot be excluded but there may have been influences on LV mass at 12 400 months due to circulating and haemodynamic factors which we either did not measure or 401 were unable to detect. A contributing factor to the reduction in between group differences at 5 402 years may have been the reduction over time in the differences in eGFR. In donors, while 12 403 month iGFR was reduced by about 30ml/min/1.73m 2 , by 5 years there was a mean increase 404 from this nadir of 2ml/min/1.73m 2 . In contrast, eGFR in healthy controls declined by about 1 405 ml/min/1.73m 2 per year. In the first CRIB-DONOR study, we found a significant association has been reported at only 11%, consistent with this we found no difference in hemoglobin at 415 12 months in our cohort. 36 We did not however measure erythropoietin, which has been 416 associated with LV hypertrophy. 37

417
Change in LV mass was chosen as the primary outcome for this study because of the well-418 recognised association of LV hypertrophy with adverse clinical outcomes and the graded 419 relationship between LV mass and prognosis. 38 In the Framingham study, LV mass was 420 second only to age in its ability to predict cardiovascular morbidity and mortality. 39 We 421 acknowledge that a causative relationship cannot be assumed and that a meta-analysis has 422 questioned the validity of using LV mass as a surrogate for total mortality in CKD, this study 423 however, included patients on dialysis and many of the studies were of inadequate size and 424 duration and measured LV mass by echocardiography which has limitations in CKD subjects. The increase in self-reported hypertension in the living kidney donor group at 5 years was not 427 consistent with the use of anti-hypertensives or associated with a significant increase in mean 428 office or ambulatory blood pressures compared to the control group. Likewise, we found no 429 significant difference in hypertension prevalence according to European Society of 430 Hypertension ambulatory blood pressure monitoring criteria or combined endpoint analysis. 26 It is likely that the apparent finding of increased rates of hypertension in donors was a result 432 of surveillance bias. 41 This phenomenon has been seen repeatedly in living kidney donor 433 studies. 41 Our study was not powered to detect small effects on blood pressure and as the 434 ambulatory blood pressure values in donors at 5 years were numerically slightly higher than 435 those in controls we suggest that longer and larger studies of ambulatory blood pressure in The major strength of this study is that it was a blinded end-point analysis from a prospective 455 longitudinal study of a donor cohort with an appropriately healthy control group allowing 456 assessment of serial change. We experienced a high return rate for a longitudinal study with 457 79% from the original cohort.

458
Limitations include potential selection bias due to attrition as a result of the longitudinal 459 design. Whilst attempts were made to minimise changes in techniques and methodology, 460 upgrades to our imaging system meant that the magnetic resonance scanner used at 5 years 461 was 3T rather than 1.5T. Signal-to-noise ratio and artefact increases with increasing field 462 strength and can potentially effect scan quality, however the field strength itself is not 463 deemed to have a significant influence on mass and volume quantification. 45 Our cohort 464 were predominantly Caucasian and therefore cannot be generalizable to all kidney donors. It 465 has previously been established that risk is highly likely to be race and age dependent. 41

466
Finally, we recognise that due to the large number of variables analysed, some significant 467 differences are likely to occur by chance and that our sample size limited our ability to detect 468 small changes in secondary end points.

689
There is no evidence of detrimental changes to cardiovascular structure/function, arterial 690 hemodynamics, blood pressure or cardiac biomarkers in living kidney donors at 5 years. year change for participants with paired data sets.

713
A; Left ventricular mass (g) and B; Adjusted pulse wave velocity (m/s).
ACEi; Angiotensin converting enzyme inhibitor, BMI; Body mass index. IHD; Ischaemic heart disease. N; Number of subjects. NSAIDS; Nonsteroidal anti-inflammatories. SD; Standard deviation. TIA; Transient ischaemic event *One subject started smoking during follow up period for a total of 3.5 years and then gave up. Data are displayed as mean ± SD at baseline for the whole cohort. Mean (95% CI) are displayed for within-group change and between-group difference.
*Within-group change and 95% CI were determined using paired samples t tests. Results are displayed as the mean change in values (95% CI) between baseline and 12 months and baseline and 5 years for each group. † Between-group difference and 95% CI were determined using independent samples t tests for comparison of within-group change at 5 years between groups. Results are displayed as the mean difference in values (95% CI) between groups for within-group change. The p value for LVM is from an independent samples t test. ‡ Data at baseline is that at 5 year follow up only. Between-group difference and 95% CI were determined using independent samples t tests for comparisons of mean values at 5 years between groups.  Data are displayed as mean ± SD or geometric mean [95% CI] at baseline for the whole cohort. Mean (95% CI) are displayed for within-group change and between-group difference.
*Within-group change and 95% CI are from paired analyses. † Between-group difference and 95% CI are from unpaired analyses. § Non-parametric data was log10 transformed prior to analysis. Values for within-group change and between group differences are displayed as antilogged values with (95% CI). These values are multipliers. For example, for AI 75 the within group change at 12 months is 1.17 in living kidney donors. The 12 month result on average is ×1.17 the baseline value but the 95% confidence interval indicates that the multiplier could be anywhere between 1.04 and 1.31. The multiplier for between group differences is 0.97 meaning the within-group 5 yr. change in donors is ×0.97 that of the change seen in controls. Therefore the between group difference is a ratio of the donor multiplier to control multiplier.
|| For categorical data the baseline prevalence is presented and the within group change is the incidence at 12 months and 5 years in those who did not have the condition at baseline. Prevalence and incidence are given as counts (percentage). The between-group difference is the relative risk (incidence for donors relative to controls) and 95% confidence intervals. ¶ The definition of hypertension was in accordance with the European Society of Hypertension guidance and in addition included those who had commenced on antihypertensive medication. 26 PWV has been adjusted for MAP and HR as recommended by the AHA to standardise vascular research. 23 AHA; American Heart Association. ABPM; Ambulatory blood pressure monitoring. AI75; Augmentation index corrected for a heart rate of 75. AdjPWV; Adjusted pulse wave velocity. BMI; Body mass index. CIT; Carotid intima-media thickness. CI; Confidence interval. DBP; Diastolic blood pressure. HR; Heart rate. IMT; Carotid intima-media thickness MAP; Mean arterial pressure. PWV; Pulse wave velocity. SBP; Systolic blood pressure. SD; Standard deviation