Protein and calorie restriction may improve outcomes in living kidney donors and kidney transplant recipients

Baseline characteristics

Thirty-five living kidney donors were randomized into either the PCR (n=15) or the control group (n=20) between May 2, 2014 and November 18, 2015 (Supplementary Figure 1). The difference in patient numbers between the groups is due to the additional inclusions to replace dropouts. Baseline characteristics of these 35 donors are listed in Table 1A. Donors in the PCR-group were more often female (P=0.028) and consequently had lower serum creatinine concentrations (P=0.021). Baseline characteristics of all transplant recipients (Table 1B) showed no differences between the two groups.

Compliance and perioperative outcome

All PCR-donors indicated that they had adhered to the diet. This was in line with a significant decrease of serum levels of prealbumin (PAB), retinol binding protein (RBP), valine and leucine after the PCR diet (Supplementary Figure 1). PCR-donors lost an average of 2.0 kg (range -4;0 kg) of bodyweight during the diet, while the control group gained 0.4 kg (range -4; +1 kg) (P=0.006). Perioperative outcomes did not significantly differ between the two arms in both donors and recipients (Supplementary Table 1A).

Postoperative outcome

Donors

At baseline, serum creatinine concentrations significantly differed between donors in the PCR and the control group (Figure 1A, postoperative day (POD)-pre). One day before surgery (i.e. at day 5 of dietary intervention), serum creatinine concentrations were similar (Figure 1A, POD-1). In a linear mixed-effects model analysis, PCR-donors showed significantly lower creatinine concentrations than the control group on POD2, POD3 and on postoperative month (POMo)1 (Supplementary Table 2A). Absolute CKD-EPI estimated glomerular filtration rate (eGFR) values did not significantly differ in both groups. To correct for baseline interpatient variability, the relative values of creatinine were calculated and analyzed. Corresponding to the absolute concentrations, relative creatinine concentrations were significantly improved in the PCR-donors on POD2, POD3 and POMo1 (Supplementary Table 2A). Subsequently, changes per time point were assessed. The PCR-donors showed a trend towards better absolute creatinine concentrations (Figure 1A); Relative creatinine concentrations were significantly better on POD2 (P=0.030), POD3 (P=0.042) and POMo1 (P=0.040, Figure 1B). Absolute eGFR values did not differ (Figure 1C). Serum urea levels were significantly lower in the PCR-group on POD-1 (P<0.001), suggesting effects of the PCR-diet on kidney function per se. This difference persisted after surgery on POD1 (P=0.003), POD2 (P=0.005) and diminished on POD3 (P=0.09) (Figure 1D). Cystatin C, as a marker of kidney function independent of muscle mass, showed no significant difference before or after the PCR-diet nor postoperatively; however, concentrations of Cystatin C did show on average a decreasing trend in PCR-donors on POD3 compared to controls (P=0.062, Figure 1E).

Figure 1. Kidney function of living kidney donors before and after kidney donation. (A) At start of the study (POD-pre), creatinine levels were significantly higher in the control group compared to the PCR group. Postoperatively, a trend towards lower levels of creatinine was observed in the PCR-group. (B) Taking POD-1 as cut-off value, relative creatinine clearance was significantly improved in the PCR-group at POD2, POD3 and POMo1. (C) Absolute glomerular filtration rate did not significantly differ between the groups. (D) Serum urea levels were significantly lower in the PCR group on POD1 and remained so in the first two postoperative days. (E) Serum Cystatin C concentrations showed no significant differences between the two groups except for a trend of lowering values in PCR-donors on POD3. Values are depicted as median ± interquartile range. PCR = protein and caloric dietary restriction; POD = postoperative day; POMo = postoperative month; eGFR = estimated glomerular filtration rate using the CKD-EPI formula. *=significant values.

Postoperative C-reactive protein (CRP) concentrations showed a trend towards lower levels in the PCR-donors, with a broad interquartile range (IQR) (Figure 2A). No differences in leukocyte numbers were seen (Figure 2B).

Figure 2. Systemic inflammatory markers in living kidney donors before and after kidney donation. (A) Levels of systemic inflammatory marker C-reactive protein before and after live kidney donation were not significantly different between the PCR and the control group. (B) Levels of leukocytes did not significantly differ between both groups either, and only reached high-normal levels on POD1 after surgery. Values are depicted as median ± interquartile range. PCR = protein and caloric dietary restriction; POD = postoperative day.

Recipients

Ideally, urine production in kidney transplant recipients starts directly after revascularization of the graft. In the PCR-group, urine production was delayed in 1/15 (7%) patients compared with 5/20 (25%) in the control group (P=0.135) (Table 2 and Supplementary Table 1B). On POD1, a trend towards lower incidence of partial acute tubular necrosis (ATN), as indicated on a MAG3 scan, was found in the PCR-group (Table 2). Delayed graft function (DGF) did not occur. Slow graft function (SGF) occurred significantly more often in the control group: 5/20 (25%) versus 0/15 (0%) (P=0.020). The incidence of biopsy-proven acute rejection (BPAR) was significantly higher in the control group than in the PCR-group: 8/20 (40%) versus 1/15 (7%) (P=0.013). Two recipients (both in the control group) developed severe acute rejection and underwent transplant nephrectomy on days six and 12 after transplantation, respectively. Data from these patients were censored from the day of transplant removal. The postoperative immunosuppressive drug regimen was similar in both groups (tacrolimus (Tac), mycophenolate mofetil (MMF) and prednisolone), except for two patients in the control group who received belatacept instead of Tac. These two patients experienced BPAR and were switched from belatacept to Tac. The Tac pre-dose concentrations showed a trend towards higher concentrations in the PCR-group on POD3 (P=0.065) and POD5 (P=0.094). No significant differences were seen in the duration of hospital stay nor in the incidence or severity of postoperative complications (Table 2).

Absolute and relative creatinine concentrations were significantly improved in the PCR-recipients as from POD1 until POD5 (Supplementary Table 2B). Assessment per time point showed no significant differences of absolute serum creatinine concentrations between groups throughout the postoperative follow-up (Figure 3A). Relative creatinine concentrations showed a trend towards lower concentrations in the PCR-group on POD4 (P=0.057), POD6, and POD14 (Figure 3B). Absolute eGFR values did not differ between groups (Figure 3C).

Figure 3. Kidney function of transplant recipients before and after surgery. (A) Serum creatinine showed a trend towards absolute lower levels in the PCR-group throughout the first 14 days after surgery. (B) Relative creatinine levels, calculated using POD-1 as cutoff value, showed a trend as from POD4 up to POD14. (C) Following the creatinine clearance, eGFR also showed a trend towards improvement in the PCR-group throughout the first days after surgery. Values are depicted as median ± interquartile range. PCR = protein and caloric dietary restriction; POD = postoperative day; POMo = postoperative month; eGFR = estimated glomerular filtration rate using the CKD-EPI formula.

Gender-based differences

Since gender was unevenly distributed between the groups, a gender stratified analysis was performed. In female donors, a consistent trend towards improved relative creatinine concentrations was observed for prolonged periods after POD1 (Supplementary Figure 2). In male donors, relative creatinine concentrations were significantly improved in the PCR-group at POMo1 (P=0.009) and POMo3 (P=0.022) (Supplementary Figure 3). Division based on gender of the donors showed for female PCR-kidney recipients a consistent non-significant trend towards improved relative creatinine concentrations (Supplementary Figure 2E–2G). The most pronounced PCR-related differences in the recipients were observed in kidneys of male donors from POD1 until POD7, yet did not reach significance due to large variation (Supplementary Figure 3E–3G).

Renal transcriptome analysis

Full genome expression profile analysis of the kidneys was performed using biopsies from 20 donors (Supplementary Table 3). A principal component analysis (PCA) showed that gender represented the largest discriminator (Supplementary Figure 4). Therefore, we separately analyzed male (n=3-5) and female (n=5-7) kidneys. The PCAs of both female (Figure 4A) and male (Figure 4B) kidney transcriptome showed small inter-variability on PC axis 1 and 2,

Figure 4. Principal component analysis of kidney tissue in female and male donors. (A) Unbiased principal component analysis (PCA) of all female donors showed the most variation of genes on the principal component (PC) axis 1 and some clustering of the two intervention groups is shown. (B) The PCA of the male donors was based on only three kidneys in the PCR group, and showed little variation on PC axis 2. Principal component (PC) 1 is depicted on the x-axis and PC2 is depicted on the y-axis, followed by the percentage of variance explained by each axis. Each symbol represents one sample of one donor. Samples of the same group are shown in the same color. PCR = protein and caloric dietary restriction.

respectively. Since one sample had to be excluded due to a low RNA yield, only 3 male patients in the PCR-group remained. Therefore, statistical analysis was restricted to the female data, showing 480 differentially expressed transcripts (DET) between diet groups. Pathway analysis indicated EIF2 Signaling as significantly upregulated (Table 3A). Other importantly enriched and activated pathways included the NRF2-pathway, G2/M checkpoint regulation and amino acid metabolism. Three significantly activated – MYCN, MYC and NKX2-3 – and two significantly inhibited – PRDM1 and SMARCB1 – transcription factors were found (Table 3B). The pathway analysis done in the male donors can be found in Supplementary Table 4.

Study design

This pilot study was designed as a randomized controlled trial and was approved by the Medical Ethics Committee (METC, MEC number 2012-134) of the Erasmus MC, University Medical Center Rotterdam, the Netherlands. All subjects provided written informed consent before inclusion. The trial was registered on October 12, 2012 in the Dutch trial registry under trial code 3663 (https://www.trialregister.nl/). This manuscript was prepared in accordance with the CONSORT 2010 statement [33].

Subjects

Living kidney donors were approached by the trial coordinator at the outpatient clinic of the department of Surgery of the Erasmus MC, University Medical Center Rotterdam, between May 2, 2014 and November 18, 2015 (Supplementary Figure 5). Inclusion criteria were age between 18-70 years old, BMI ≥ 19, no participation in another trial 30 days prior to first contact, and no known allergies to any of the ingredients of the used dietary intervention. Exclusion criteria were not meeting the inclusion criteria, or a surgery that would not take place in the Erasmus MC due to participation in the cross-over kidney donation program [15]. Randomization took place directly after informed consent. No statistical power calculation for sample size was performed due to the pilot design of the study.

Dietary intervention

Patients in the PCR group received a diet containing 30% fewer calories and 80% fewer protein [15]. The diet was given for five consecutive days prior to surgery. The diet was based on a synthetic liquid diet as described previously [15], and was supplemented with a limited number of low-protein and protein-free products (mainly fruits and vegetables) until the desired preset individual needs were met [34] (Supplementary Table 5). The calorie- and protein-restricted powder Scandishake® Mix shakes were kindly provided by Nutricia Advanced Medical Nutrition, The Netherlands. The control group had no dietary restrictions and continued their normal diet.

Surgical procedures

Preoperative, perioperative, and postoperative anesthetic care concerning drug administration, ventilation and fluid regimens was carried out according to our local protocols for kidney donors and kidney recipients. Donor kidneys were obtained via either a laparoscopic nephrectomy or a hand-assisted retroperitoneal nephrectomy (HARP) [35]. Kidney transplantation was performed via an open approach, and the kidney was positioned supra-inguinally on the external iliac artery and the external iliac vein. Additional informed consent was obtained from the transplant recipients to obtain biopsies of the renal cortex, which were taken at the end of the cold ischemia time (kidney off ice) using a punch biopsy with a diameter of 4 mm.

Immunosuppressive therapy

Kidney transplant recipients received the same initial immunosuppressive therapy, except for two patients in the control group who received belatacept (Bristol-Myers Squib, NYC, NY) instead of tacrolimus. The initial immunosuppressive treatment consisted of tacrolimus (Prograft®; Astellas Pharma, Leiden, The Netherlands), mycophenolate mofetil (MMF; CellCept®; Roche Pharmaceuticals, Woerden, The Netherlands) and prednisolone treatment. All patients received induction therapy with basiliximab (Simulect®, Novartis Pharma, Arnhem, the Netherlands). The doses, whole blood or plasma target concentrations, and phasing of immunosuppressive therapy have been described elsewhere in detail [36].

Outcome parameters

Living kidney donors

Before and after the dietary intervention, the following data were obtained: bodyweight, age, gender, length, and systolic blood pressure. The time point of first blood withdrawal before start of the dietary intervention, POD-pre, varied from one day up to over a year. Therefore, we used POD-1 values after the dietary intervention as cutoff levels for the calculation of relative values. Before and at various time points after surgery, serum levels of parameters shown in Supplementary Table 6 were determined. Cystatin C was measured on the Cobas 8000 system (Roche®) (Supplementary Table 6). A schematic overview of the experimental timeline of the study is shown in Supplementary Figure 6. Processing of the blood samples was done as described previously [15].

Kidney transplant recipients

Before and after the dietary intervention, the following data were obtained: bodyweight, age, gender, length, and systolic blood pressure. Before and at various time points after surgery, serum values of parameters shown in Supplementary Table 6 were determined. All patients underwent renal scintigraphy (MAG3) on POD1 to assess renal perfusion. In case of a gradual increase of activity in the parenchyma of the transplanted kidney without evidence of cortical excretion, acute tubular necrosis (ATN) was suspected and recorded. Delayed graft function (DGF) was defined as the need for dialysis in the first postoperative week [37]. Slow graft function (SGF) was defined as an estimated glomerular filtration rate (eGFR) ≤10 ml/min/1.73m² at day six after transplantation [20]. Surgical complications were graded using the Clavien-Dindo classification [38].

mRNA-sequence analysis

For gene expression analysis, kidney biopsies of 10 donors per group were used. Biopsies were put in 2 mL Eppendorf tubes (Eppendorf Group, New York, USA) containing 1 mL of RNAlater RNA Stabilization Reagent (QIAGEN Benelux B.V., Venlo, the Netherlands), and were kept at 4°C for at least 48 hours. Total RNA extraction and measurement of RNA concentration was done as described previously [12, 13]. The RNA quality was expressed as the RNA integrity number and values ranged between 6.1 and 8.2. mRNA polyA-affinity purification and subsequent Ion Proton (Thermo Fisher) next-generation sequencing was performed by the Microarray Department of the University of Amsterdam, the Netherlands. Handling, analysis and visualization of the data were performed in R (R foundation). Principal component analysis and density plots of the raw counts revealed an outlying sample that had a very low RNA concentration, which was omitted from further analysis. After filtering, 175,756 transcripts were analyzed. Differentially expressed transcripts between diet types within gender were calculated using DESeq2. False Discovery Rate correction was performed as described by Storey and Tibshirani [39, 40]. Complete raw and normalized microarray data and their MIAME compliant metadata have been deposited at the Gene expression Omnibus (GEO) database GSE103532 (https://www.ncbi.nlm.nih.gov/geo/).

Statistical analysis

Categorical data are presented as numbers and continuous variables as median ± interquartile range (IQR). The data were tested for normality using the Shapiro–Wilks test and subsequent visual assessment. Continuous data were compared using the non-parametric Mann–Whitney U test. A Bonferroni correction for multiple testing was performed on the postoperative kidney function parameters. In the living kidney donors, this correction was done separately for the short-term outcome until POD3 and the outcome until POMo3. In the transplant recipients, the cutoff for short-term outcome was set at POD7; the cutoff for outcome when kidney function has stabilized, was set on POMo6. To assess changes in eGFR and creatinine concentrations over time for both donors and recipients, lme4 [41] for R 3.3.3 was used to perform a linear mixed effects analysis of the relationship between the outcomes of interest and treatment category. Time was modelled as a factor variable as the relationship between the time passed between observation moments is not linear, the correlation with outcome is not known to be linear and we also wanted to model the preoperative measurement. In order to compute the difference, the treatment groups on each time point, the mixed effects model was fitted without an intercept. POD-1 values were not included in the models. In addition, data were adjusted for age and/or gender. The analyses were performed using Statistical Packages for Social Sciences 23.0 (IBM Inc., Chicago, IL, USA), R (R foundation), GraphPad Prism (GraphPad Software Inc., version 5.01), and Office Excel (Microsoft (Office) 2016).