SEARCH

SEARCH BY CITATION

Keywords:

  • Expanded criteria donor;
  • kidney transplantation;
  • marginal donor;
  • pre-implantation biopsy;
  • score

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. References
  8. Supporting Information

The predictive value of pre-implantation biopsies versus clinical scores has not been studied extensively in marginal donors. Pre-implantation biopsies were performed in 313 kidneys from donors that were ≥ 50 years of age (training set, n = 191; validation set, n = 122). The value of the donor clinical parameters and histological results in predicting 1-year estimated glomerular filtration rate (eGFR) <25 mL/min/1.73 m2 was retrospectively evaluated. In multivariate analysis, the only clinical parameters associated with low eGFR were donor hypertension and a serum creatinine level ≥150 μmol/L before organ recovery. Clinical scores (Nyberg and Pessione) were not significantly associated with graft function. Regarding histological parameters, univariate analysis showed that glomerulosclerosis (GS) (p = 0.02), arteriolar hyalinosis (p = 0.03) and the Pirani (p = 0.02) and chronic allograft damage index (CADI) (p = 0.04) histological scores were associated with low eGFR. The highest performance in predicting low eGFR was achieved using a composite score that included donor serum creatinine (≥150 μmol/L or <150 μmol/L), donor hypertension and GS (≥10% or <10%). The validation set confirmed the critical importance of taking into account biopsy and clinical parameters during marginal donor evaluation. In conclusion, clinical scores are weak predictors of graft outcomes with marginal donors. Instead, a simple and convenient composite score strongly predicts graft function and survival and may facilitate optimal allocation of marginal donors.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. References
  8. Supporting Information

The continuing shortage of organ donors has led to increasing use of kidneys from ‘marginal’ deceased donors (1). The majority of such donors are considered marginal due to their age and the presence of glomerulosclerosis (GS) and renal arteriosclerosis (2). Although a significant proportion of marginal donors can act as an acceptable alternative to optimal donors, there is considerable heterogeneity in graft function and survival rates. This may be a consequence of histopathological differences between aging kidneys. For example, an autopsy study of subjects without evidence of renal disease or hypertension showed that, beyond the age of 50 years, there is wide variation in the percentage of sclerotic glomeruli (3). Accordingly, many centers discard kidneys with extensive GS, even though it has been shown that GS should not be regarded as the only criterion when assessing donor quality (4,5). Indeed, kidneys allocated from marginal donors with major comorbidities and risk factors may be discarded unless a pre-implantation biopsy reveals the absence of major histological damage.

There is clearly a need to refine the clinical and histopathological criteria for graft selection in marginal donors in order to identify kidneys that are at the highest risk of graft dysfunction or failure. Clinical scoring systems based on information available at the time of donor nephrectomy have previously been proposed to assess donor risk factors and guide appropriate transplantation strategies (6, 7). Nyberg et al. developed and validated a scoring system that takes into account donor age, cause of death, history of hypertension or diabetes, creatinine clearance, cold ischemia time and the presence of renal artery plaque (6). The resulting donor score correlated with early renal function and delayed graft function. Data from the French registry showed that vascular brain death, donor serum creatinine levels and donor hypertension or diabetes are predictors of graft loss (7). To help clinicians predict graft dysfunction or graft loss, these scoring systems have been validated across the whole population of deceased donors. Using this methodological approach, however, donor age carries a relatively high weight such that, when applied to a limited population of elderly donors, the predictive performance of these scoring systems could be restricted.

Another approach to identifying high-risk kidneys is to use a pre-implantation biopsy in order to characterize potential kidney grafts (8), help predict the graft outcome (9,10), and provide a reference point for analysis of subsequent biopsies (8,11). The percentage of sclerotic glomeruli (9,12) and the degree of tubulointerstitial (12,13) and chronic vascular lesions (10,14–19) prior to transplant are all associated with a worse graft outcome. These histological changes have been integrated into a histological scoring system that has proven useful in determining the allocation of kidneys from donors that are over 60 years of age (20). Importantly, however, accurate interpretation of a donor biopsy should not be allowed to increase the cold ischemia time, which is well established as a predictor of long-term graft survival in marginal donors. Additionally, the studies that formed the basis for this scoring system varied in terms of patient selection, indication for pre-implantation biopsy and the definition of graft outcomes.

Despite the growing use of marginal donors, the relationship between donor clinical scores, pre-implantation biopsy results and graft outcomes has not been extensively studied in this population. Indeed, in many centers, pre-implantation biopsies are not performed routinely and donor clinical scores alone are used as the basis for decision-making with regard to accepting or discarding kidneys. We have, therefore, evaluated the predictive performance of clinical scores and histological scores in a population of marginal donors, and assessed the use of a simple composite algorithm to stratify the risk of graft dysfunction or failure. We focused the analysis on donors aged over 50 because this restricted population will include a high proportion of patients reaching the expanded criteria donor (ECD) definition, or at least with risk factors associated with poor graft outcome, as initially reported by Port et al. (21). Accordingly, 61% of our study population met the ECD definition.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. References
  8. Supporting Information

Study population

We retrospectively studied the records of 313 recipients of a single kidney transplant at two centers (training set, Necker Hospital, from January 1998 to July 2006, n = 191; Validation set, Necker Hospital, August 2006 to April 2007 and Saint-Louis Hospital, from September 2004 to April 2007, n = 122), in whom the kidney donor was deceased and aged ≥50 years, and in whom the graft was adequately biopsied in the operating room at time zero (according to the Banff'97 criteria) (22). The value of pre-implantation biopsies versus clinical scores in predicting graft performance was studied in the training set, and a scoring system was developed from this cohort. The scoring system was then validated in the validation set. All patients included in the study were required to have clinical follow-up data for a minimum of 1-year posttransplant. Patients who died with a functioning graft during the first year posttransplantation were excluded from the analysis.

Data collection

The following data were collected: donor characteristics (age, sex, cause of death, serum creatinine levels prior to organ recovery, history of hypertension or diabetes), cold ischemia time, recipient characteristics at the time of transplant (age, sex, number of previous kidney transplants, time on dialysis, current level of panel reactive antibodies [PRA], number of human leukocyte antigen [HLA] mismatches), and graft status at 1-year posttransplant (estimated glomerular filtration rate [eGFR] according to the Modification of Diet in Renal Disease [MDRD] formula and outcome of transplantation, including death). Low eGFR was defined as <25 mL/min/1.73 m2 at 1-year posttransplant, and included graft loss during year 1.

Two clinical donor risk scores, the Nyberg score (6) and the Pessione score (7), were calculated from these data. The Pessione score was calculated by adding one point for each of the following risk factors: age > 60 year, hypertension, diabetes, serum creatinine > 150 μmol/L and cerebrovascular cause of death.

Renal allograft histopathology

Pre-implantation biopsies were performed by the transplant surgeon using a 16G Tru-Cut® needle. The biopsy was subsequently fixed in alcohol, formol and acetic acid, embedded in paraffin, and then processed for light microscopy (staining for hematoxylin-eosin, periodic acid-Schiff, Masson-Trichrome and silver methenamine). Histopathologic studies were based on the Banff'97 scoring system (22). The number of glomeruli, number and percentage of sclerotic glomeruli and histological scores according to the Banff classification were used to calculate two histological risk scores: the ‘Pirani score’, previously used by Remuzzi et al. to define the allocation of marginal kidneys (20), and the histologic chronic allograft damage index (CADI score) (23). The Pirani score was calculated as follows: changes in each evaluated component of the kidney tissue (vessels, glomeruli, tubules and connective tissue) received a score ranging from 0 (if no changes were observed) to 3 (if marked changes were present). The sum of these scores was defined as the Pirani score, which could range from 0 to 12. The CADI score was based on the individual component scores for interstitial inflammation, interstitial fibrosis, glomerular sclerosis, glomerular mesangial matrix increase, vascular intimal proliferation and tubular atrophy, each individual parameter being scored from 0 to 3 as described (23). Supplemental Table S1 shows the cutoffs used by the different histologic scoring systems. All biopsies were performed concomitantly with transplantation such that histopathological findings could not influence the allocation of grafts, and this potential selection bias was eliminated.

Immunosuppression

The patients from the training set usually received 2 g of mycophenolate mofetil (CellCept®, Roche Pharmaceuticals, Basel) pre-operatively, followed by mycophenolate mofetil 1.5 g b.i.d. from day 1 to day 45 postoperatively, and then 1 g b.i.d., and prednisolone (500 mg pre-operatively, 125 mg on day 1, then 20 mg/day for 15 days, progressively tapered to 10 mg/day at day 30). The patients also received 4 mg/kg/day of cyclosporine (Neoral®, Novartis Pharma AG, Basel) in divided doses starting on the seventh day after transplantation, with a target whole-blood cyclosporine concentration of 600–800 ng/mL 2 h after cyclosporine intake during the first 3 months. A few highly sensitized patients received tacrolimus (Prograf®, Astellas, Tokyo) instead of cyclosporine, which was started at the day of transplantation, and adjusted to achieve trough levels of 8–12 ng/mL. In addition, all patients received induction therapy with antithymocyte globulin (rabbit ATG, Thymoglobulin®, Genzyme, Lyon, France) or basiliximab (Simulect®, Novartis Pharma AG,Basel, Switzerland). Acute rejection episodes were treated with bolus steroids (500 mg/day) for 3 days, followed by a rapid steroid taper.

Statistical analysis

All statistical analyses were performed using SAS software package version 8.1 (SAS Institute, Inc., Cary, NC). The results are expressed as numerical values and percentages for categorical variables and as mean ± SD for continuous variables, unless otherwise stated.

The impact of clinical and histological parameters on the graft outcome was evaluated using an eGFR < 25 mL/min/1.73 m2 at 1-year posttransplant, including the graft loss during year 1, as the end-point. A univariate analysis was performed using the χ2-test to compare the percentages. Univariate logistic regression was used to determine the odds ratios (OR) and confidence intervals. Multivariate analysis was performed after adjustment for recipient and transplant variables including recipient age, prior kidney transplantation, gender matching, HLA-A, B and DR mismatching (0–6), time on dialysis and the cold ischemia time. Multivariate logistic regression was used to estimate adjusted OR and confidence intervals. Receiver operating characteristic (ROC) curves were used to compare the predictive value of the clinical parameters, GS, Pirani score and a composite score on eGFR at 1-year posttransplant.

For graft survival analysis, we considered the return to dialysis as the end-point, and reported death-censored graft survival. Death-censored graft survival was estimated using Kaplan–Meier survival curves. The impact of clinical and histological parameters on graft survival was analyzed using the log-rank test. A two-sided p-value of 0.05 was considered to be statistically significant.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. References
  8. Supporting Information

Population characteristics

The training set consisted of 191 recipients in whom the graft was adequately biopsied in the operating room at time zero (according to Banff'97 criteria) (22). The mean follow-up time after transplantation was 33 ± 22 months, with all patients having a minimal follow-up time of 1 year.

The recipient and donor characteristics are summarized in Table 1. The mean recipient age was 50.6 ± 10.4 years (range: 24–71). Most of the recipients received their first allograft and were not sensitized. Delayed graft function occurred in 42.4% of the cases. Thirty-seven recipients (19%) had one episode of acute rejection after a mean follow-up of 33 ± 22 months. The mean donor age was 60.1 ± 7.4 years (range: 50–80 years); 92 donors (48.2%) were ≥60 years of age. Brain death was caused by cerebrovascular accident (CVA) in over half the donors (55.0%). The cold ischemia time was >24 h in 73 cases (38.2%). A history of hypertension was reported in 59 (31%) donors, and 29 (15.2%) had diabetes.

Table 1.  Recipient and donor characteristics (Training Set)
Parametern = 191
  1. HLA, human leucocyte antigen; PRA, panel reactive antibodies.

  2. aAt time of transplant.

  3. bPrior to organ recovery.

Recipient characteristicsa
 Age (mean ± SD, years) 50.6 ± 10.4
 Male/female ratio128/63
 HLA A + B mismatch (mean ± SD) 1.9 ± 1.0
 HLA DR mismatch (mean ± SD) 0.7 ± 0.7
 First/second/third kidney transplant162/26/3
 Peak PRA level
 <30%89.7%
 30–79%7.9%
 ≥80%2.4%
 Delayed graft function (n, %)81 (42.4%)
Donor characteristics
 Deceased donor (n, %)191 (100.0%)
 Age (mean ± SD, years)60.1 ± 7.4
 Male/female ratio110/81
 Death of cerebrovascular origin (n, %)105 (55.0%)
 History of hypertension (n, %)59 (30.8%)
 History of diabetes (n, %)29 (15.2%)
 Serum creatinine (mean ± SD, μmol/L)b101 ± 51
 Cold ischemia time (mean ± SD, hours)22.7 ± 7.2

Association between clinical characteristics and graft function

The mean eGFR of the training set was 47.5 ± 19.4 mL/min and 43 ± 16 mL/min at 3 months and 1-year posttransplant, respectively. Twenty-seven patients (14.1%) had eGFR < 25 mL/min at 1 year posttransplant. The univariate analysis revealed that a donor history of hypertension and serum creatinine prior to organ recovery were both significantly associated with low eGFR at 1-year posttransplant (p = 0.04 and p = 0.007, respectively) (Table 2). After adjustment for recipient and graft characteristics, these two parameters remained significant, and were independent predictors of low eGFR at 1-year posttransplant.

Table 2.  Univariate and multivariate analysis of the association between donor-related clinical parameters and low eGFR at 1-year posttransplant
Donor parametereGFR at 1 yearp Univariate analysisp Multivariate analysisa
≥25 mL/min (n = 164)<25 mL/min (n = 27)
  1. eGFR, estimated glomerular filtration rate by the MDRD formula; CVA, cerebrovascular accident.

  2. aThe multivariate logistic regression analysis included all the parameters significantly associated with eGFR at 1-year posttransplantation by univariate analysis or conventionally associated with graft outcome, and was adjusted for recipient characteristics (age, sex, number of HLA mismatches, previous kidney transplantation, cold ischemia time and time on dialysis).

  3. bThe Nyberg score is calculated by using the following risk factors: donor age, history of hypertension, creatinine clearance of the donor, HLA mismatch, cause of death.

  4. cThe Pessione score is calculated by adding 1 point for each following risk factors: donor age ≥60 years, history of hypertension, diabetes, serum creatinine ≥ 150 μmol/L, death by CVA.

Male donor (%)54.974.00.060.48
Cause of death  0.630.75
 CVA (%)54.359.3  
 Non-CVA (%)45.740.7  
Serum creatinine
 Mean ± SD (μmol/L) 97.1 ± 42.0125.1 ± 83.0 0.0070.02
 ≥150 μmol/L (%)12.225.90.06
Donor age
 Mean ± SD (year)60.2 ± 7.559.0 ± 7.00.490.72
 ≥60 years (%)48.844.40.68
Hypertension (%)28.048.10.040.03
Diabetes (%)15.214.80.9 0.48
Nyberg scoreb24.8 ± 4.325.6 ± 5.00.4 
Pessione scorec1.58 ± 1.11.92 ± 1.30.13

Neither clinical donor risk score (Nyberg or Pessione) showed an association with a low eGFR at 1 year (Table 2).

Association between histopathological variables and graft function

The mean number of glomeruli measured in the pre-implantation biopsy was 16.3 ± 7.3 in the training set; the mean proportion of sclerotic glomeruli was 10.1 ± 12.2%. Seventy-seven biopsies (40.3%) showed GS ≥ 10%, and 37 (19.4%) had GS ≥ 20%. Interstitial fibrosis/tubular atrophy (IF/TA) lesions were observed on 50 biopsies (26.2%), of which 39 were grade I, eight were grade II and three were grade III (Table 3).

Table 3.  Univariate and multivariate analysis of the association between histological parameters and low eGFR at 1-year posttransplant
Histological parametersneGFR at 1 yearp Univariate analysispa Multivariate analysis
≥25 mL/min (n = 164)<25 mL/min (n = 27)
  1. aThe multivariate logistic regression analysis was adjusted for recipient characteristics (age, sex, number of HLA mismatches, prior kidney transplantation, cold ischemia time and time on dialysis).

Sclerotic glomeruli (%)
 06938.4%22.2%0.020.01
 1–94526.2% 7.4%  
 10–194018.9%33.3%  
 ≥203716.5%37.0%  
Arteriolar hyalinosis
 06236.0%11.1%0.030.07
 17537.8%48.1%  
 24623.2%29.6%  
 3 8 3.0%11.1%  
Chronic vascular changes
 05930.5%33.3%0.830.66
 19148.8%40.7%  
 23718.9%22.2%  
 3 4 1.8% 3.7%  
Tubular atrophy
 09553.7%25.9%0.060.17
 18440.9%63.0%  
 2 9 4.3% 7.4%  
 3 3 1.2% 3.7%  
Interstitial fibrosis
 012365.9%55.6%0.320.40
 15528.7%29.6%  
 2 9 3.7%11.1%  
 3 4 1.8% 3.7%  
Pirani score
 ≤313071.3%48.1% 0.056-
 4–65626.2%48.1%  
 >7 5 2.4% 3.7%  

Patients with a low eGFR at 1 year had a higher percentage of GS than those with eGFR > 25 mL/min/1.73 m2 (15.1 ± 13.2% vs. 9.3 ± 11.8%, p = 0.02), a higher Pirani score (3.6 ± 2.3 vs. 2.6 ± 1.9, p = 0.02), and a higher CADI score (4.0 ± 2.6 vs. 2.8 ± 2.0, p = 0.04). After adjustment for recipient and graft variables, a multivariate logistic regression analysis including all the Banff classification parameters showed that GS was the only significant, independent histological predictor of low eGFR at 1 year (Table 3).

The predictive value of the CADI, Pirani and Banff scores in terms of predicting a low eGFR at 1-year posttransplant were compared by multivariate analysis ajdusted for recipient characteristics (Table 4). CADI, Pirani and Banff scores significantly and independently predicted a low eGFR. The area under the curves (AUCs) of these three scoring systems were compared to clarify which one is the best histological scoring system for the evaluation of donor biopsies. Whereas the AUCs of the CADI score and of the Banff parameters were identical, the AUC of the Pirani score was significantly higher than the AUC of the CADI score (p = 0.005) and the AUC of the Banff parameters (p = 0.009).

Table 4.  Multivariate logistic regression models evaluating CADI, Pirani and Banff scores for the prediction of low eGFR at 1-year posttransplant
 ORpAUC p
  1. AUC, area under the ROC (receiver operating characteristics analysis) curve; OR: adjusted odds ratio.

CADI score1.30.0090.76CADI versus Pirani0.005
Pirani score2.7 0.00030.79Banff versus Pirani0.009
Banff1.30.0130.76CADI versus Banff0.55 

Composite scoring system

The predictive effect of a combination of clinical (serum creatinine before organ recovery and donor history of hypertension) and histological parameters (percentage of GS and Pirani score) was assessed using multivariate analysis (Table 5). Models 1 and 2, respectively, evaluated the additional predictive value of GS and the Pirani score in combination with the two clinical parameters. These models showed that the two histopathological parameters each independently predicted low eGFR at 1 year, and also increased the predictive value when added to the two clinical parameters. Model 3 included both GS and the Pirani score, and showed that the Pirani score did not increase the prediction of the model when added to GS. Of all the associations tested (clinical only, histopathological only, or composite scoring), the composite scoring system that included donor serum creatinine levels (<150 μmol/L or ≥150 μmol/L), donor hypertension and GS (≥10% or <10%) showed the highest predictive value for low eGFR at 1 year posttransplant.

Table 5.  Multivariate logistic regression models evaluating the effect of adding histological parameters to donor clinical factors for the prediction of low eGFR at 1-year posttransplant
 Model 1Model 2Model 3
ORpORpORp
  1. AUC, area under the ROC (receiver operating characteristics analysis) curve; OR: adjusted odds ratio.

Serum creatinine ≥ 150 μmol/L3.40.042.70.093.00.07
History of hypertension2.40.082.40.072.40.09
Sclerotic glomeruli ≥ 10%5.3 0.001  4.3 0.009
Pirani score  1.30.03 1.140.28
AUC 0.84  0.79  0.84 

Predictive value of the composite scoring system for graft function

The composite score based on these three factors (donor serum creatinine, donor hypertension and GS) was evaluated in terms of its predictive value for eGFR at 1 year in the training set (Figure 1 and Table 6). The ROC analysis showed that the calculated AUC of the four evaluated parameters (composite score, Pirani score, clinical parameters and GS) were significantly different from each other (global test, p = 0.007). Whereas the AUC of the composite score was only numerically higher than the AUC of the percentage of GS alone, it was significantly higher than the AUC of the Pirani score (0.84 vs. 0.79, p = 0.001) and the AUC of the clinical parameters (0.84 vs. 0.78, p = 0.009) (Figure 1).

image

Figure 1. Receiver operating characteristics (ROC) curves for clinical, histopathological and composite scoring systems as predictors of low eGFR at 1-year posttransplant. The composite score included history of hypertension, donor serum creatinine before organ recovery and percentage of glomerulosclerosis. The clinical risk score included only history of hypertension and donor serum creatinine before organ recovery. Sensitivity is the proportion of true positive results; 1-specificity is the proportion of false positive results. A value of 0.5 is no better than that expected by chance (the null hypothesis). A value of 1.0 reflects a perfect indicator (global test, p = 0.007; composite score vs. glomerulosclerosis, p = NS; composite score vs. Pirani score, p = 0.001; composite score vs. clinical parameters, p = 0.009).

Download figure to PowerPoint

Table 6.  Univariate and multivariate analysis of a composite score including clinical parameters (CP, donor serum creatinine ≥ 150 μmol/L and/or donor history of hypertension) and the percentage of glomerulosclerosis (GS) to predict the risk of low eGFR at 1-year posttransplant
Composite scorenUnivariate analysisMultivariate analysis
Low eGFR at 1 year (%)pORp
  1. AUC, area under the ROC (receiver operating characteristics analysis) curve; OR, adjusted odds ratio.

CP = 0 and GS < 10%77 5.20.00031.00.0003
CP = 0 and GS ≥ 10%4012.5 5.2 
CP ≥ 1 and GS < 10%3713.5 5.5 
CP ≥ 1 and GS ≥ 10%3735.1 27.5  
AUC    0.84  

Finally, multivariate logistic regression showed that, using this composite score, the adjusted OR for the prediction of low eGFR at 1 year ranged from 1.0 if none of the three factors (i.e. donor serum creatinine ≥ 150 μmol/L, donor hypertension, or GS ≥ 10%) were present, to 5.2 if GS > 10% in the absence of the two clinical parameters, and to 27.5 if all three factors were present (p = 0.0003) (Table 6).

Predictive value of the composite scoring system for graft survival

Eight patients of the training set died during the follow-up. The patient survival rate was 98.9% and 96.1% at 1 and 3 years, respectively. Across the total study population, the death-censored graft survival was 94.2% at 1 year, 91.4% at 2 years and 87.5% at 5 years posttransplant. Among the recipients with a donor serum creatinine ≥ 150 μmol/L before organ recovery, the death-censored graft survival was numerically lower than for those with serum creatinine < 150 μmol/L (p = 0.09, log-rank test) (Figure 2). Whereas donor history of hypertension did not affect the death-censored graft survival, GS was highly associated with graft survival (p = 0.0006) (Figure 2).

image

Figure 2. Association of (A) donor serum creatinine before organ recovery (B) donor history of hypertension and (C) glomerulosclerosis with death-censored graft survival (Kaplan–Meier estimates).

Download figure to PowerPoint

The association of the Pirani score and the composite score with graft survival was also evaluated (Figure 3). Both the Pirani score (p = 0.0001) and the composite score (p = 0.0059) were highly associated with the death-censored graft survival (Figure 3).

image

Figure 3. Association of (A) Pirani score and (B) the composite score with death-censored graft survival (Kaplan–Meier estimates). GS+, ≥10% glomerulosclerosis. CP+, presence of either donor history of hypertension or donor serum creatinine ≥ 150 μmol/L before organ recovery.

Download figure to PowerPoint

Validation

The scoring system was tested on data from a validation set of 122 patients. The recipient and donor characteristics of the training and validation sets were similar. The mean recipient age was 51.5 ± 10.4 years (range: 27–70). The mean donor age was 61.2 ± 8.1 years (range: 50–81 years); 58 donors (47.5%) were ≥60 years old. Brain death was caused by CVA in 75 (61%) patients. A history of hypertension was reported in 41 (34%) donors, and 10 (8%) had diabetes.

The mean eGFR of the validation set was 45 ± 19 mL/min at 1-year posttransplant. Twenty-five patients (20.5%) had eGFR <25 mL/min at 1 year. The donor history of hypertension and serum creatinine prior to organ recovery were, respectively, significantly (34.1% vs. 13.7%, p = 0.01) and numerically (37.5% vs. 19.8%) associated with a low eGFR at 1 year (Table 7). Glomerulosclerosis was also highly associated with a risk of low eGFR at 1-year posttransplant (43.4% vs. 2.9%, p < 0.0001), thus confirming the results obtained in the training set. The composite scoring system was highly associated with graft outcome (p < 0.0001, Kruskal Wallis test), and highlighted the impact of GS on graft outcome. The clinical parameters (donor serum creatinine ≥ 150 μmol/L and/or donor history of hypertension) increased from 31% to 56%, the risk of low eGFR at 1 year in patients with a glomerulosclerosis ≥ 10%.

Table 7.  One-year graft outcome in the validation set of 122 patients
ParametersLow eGFR at 1 year (%)p
  1. CP, donor serum creatinine ≥ 150 μmol/L and/or donor history of hypertension.

Clinical parameters
 History of hypertension (n = 28)34.1% versus 13.7%0.01
 Serum creatinine ≥ 150 μmol/L (n = 8)37.5% versus 19.8%NS
 CP+ (hypertension and/or serum creatinine ≥ 150) (n = 46)28.3% versus 15.8%0.1
 Sclerotic glomeruli ≥10% (n = 53)43.4% versus 2.9%<0.0001
Composite score
 CP = 0 and GS < 10% (n = 49)2/49 (4%)<0.0001
 CP = 0 and GS ≥ 10% (n = 26)8/26 (31%) 
 CP ≥ 1 and GS < 10% (n = 20)0/20 (0%) 
 CP ≥ 1 and GS ≥ 10% (n = 27)15/27 (56%) 

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. References
  8. Supporting Information

Our results, which provide the first comparison of clinical and histological scoring systems in a population of marginal donors, suggest three main conclusions regarding graft function and survival in recipients of a kidney allograft from a marginal donor. First, widely-used clinical scores perform weakly in terms of predicting graft outcome in this population. Second, histological findings at the time of transplantation can predict low eGFR at 1-year posttransplant. Third, a simple composite score (including clinical and histological parameters) showed the best predictive value for graft function and outcome. This scoring system is based on only two clinical variables and one simple histopathological parameter (which is readily available at the time of organ procurement), and is suitable for routine use in the clinic. Together, these three parameters provided a better prediction of 1-year renal function than clinical variables alone, and were also highly predictive of death-censored graft survival. Together, these data highlight the importance of pre-implantation biopsies when using transplants from marginal donors.

It remains difficult to define a poor graft outcome. Previous studies showed that posttransplant renal function in the first year predicts long-term kidney transplant survival. We defined a primary endpoint of 25 mL/min of eGFR at 1-year posttransplant, because this value corresponds to creatinine levels that have been reported to predict a 5-year graft survival of less than 40%. Considering this result to be unacceptable, we choose this endpoint to define the poor graft outcome. This predicted poor graft survival is even more critical when considering that the increasing shortage of available deceased-donor kidneys, and the increase of the mean deceased-donor age, led to the proposal of using older kidneys on young recipients. It was noteworthy that, in our study, even if the recipient age ranged from 24 to 71 years, the mean recipient age was 50.6 years, and 33 (17%) recipients were younger than 40 years of age.

It is noteworthy that neither of the two previously validated clinical scores used in our study were predictive of graft function. Pessione et al. studied allograft risk factors in 7209 deceased-donor kidney transplant recipients and showed that the cerebrovascular cause of death, a donor history of hypertension, and elevated serum creatinine levels before organ recovery were significant, independent donor risk factors for graft survival; donor age was a statistically significant, but not independent, risk factor (7). In the study of Pessione et al., only 7.6% of the grafts were from donors aged over 60 years. Nyberg et al. developed a scoring system based on seven clinical and biological donor variables in a population of 90 deceased-donor transplant recipients (6). In their analysis, donor age was associated with creatinine clearance at 1 month posttransplant, and was then introduced into their scoring system. No data are provided by Nyberg et al. about donor age, but the majority of donors appeared to be below 50 years of age. Interestingly, data on whether donor age is an independent risk factor for recipient renal function remain conflicting (24–29). The design of our study markedly reduces the effect of donor age, since all donors were over 50 years of age, which may explain why the clinical scores were not predictive of graft outcome in our population.

As previously reported, we found that a donor history of hypertension was associated with a worse graft outcome (7, 29–31). The presence of hypertension is one of the most relevant factors to consider in the evaluation of the donor, as it underlies the crucial importance of evaluating arterial disease in elderly potential donors (7). The level of renal function of the donor has also been associated with graft outcome (7, 30, 32). For example, Carter et al. have shown that a combination of calculated creatinine clearance <80 mL/min and a long-standing history of hypertension in an older donor reduced graft survival to 53.6% at 3 years (30), while Pessione and colleagues have reported that impaired renal function (donor serum creatinine > 150 μmol/L prior to organ recovery) correlates strongly with poor early graft survival (7).

Our study, which only included single kidney transplantations, confirms the effectiveness of the histopathological scoring system developed by Remuzzi et al. (20) in predicting renal function and allograft survival. This histological scoring system, first described by Pirani and Salinas-Madrigal (33) and slightly modified by Karpinski et al. (10), quantifies the severity of histological changes (GS, interstitial fibrosis, tubular atrophy and vascular disease). Remuzzi et al. have shown that when kidney grafts, obtained from donors >60 years of age, were allocated for single or dual transplantation on the basis of this scoring system, graft survival was similar to that of single grafts from younger donors, and was substantially better than for single grafts from donors >60 years of age that were selected and allocated on the basis of standard clinical criteria (20). According to the Remuzzi scoring system, 56 grafts from our training set would have been used for dual transplantation (score 4–6), and five kidneys would not have been considered for transplantation (score ≥ 7). Because all biopsies were performed concomitantly with transplantation in the present study, the histopathological findings did not influence the allocation of grafts. The mean eGFR was 45.3 ± 15.2 mL/min, 38.4 ± 17.8 mL/min and 29.5 ± 16.9 mL/min at 1-year posttransplant in recipients of a graft with a score of 0–3, 4–6 and ≥7, respectively (p = 0.01, Kruskal–Wallis test). Moreover, as shown in Figure 3A, the death-censored graft survival was highly associated with different groups defined by Remuzzi, thus confirming, through a different approach, the high value of this scoring system in predicting graft survival.

Since interobserver variability in grading lesions is a well-recognized phenomenon, we focused our histological analysis on GS. This is easy to quantify in a biopsy section, even by a nonspecialist histopathologist, and also allows rapid assessment of pre-implantation biopsies to facilitate decision-making on organ use and allocation. Previous studies have demonstrated that tubulointerstitial and vascular lesions are associated with a poor functional graft outcome (10,12,14,16), but these parameters did not reach statistical significance in our study. Nevertheless, when unequivocally identified, moderate to severe interstitial fibrosis and arteriosclerosis are also important parameters that should enter into the donor evaluation process. Consistent with previous reports (9,12), we found that GS is significantly and strongly associated with graft outcome. In an analysis of United Network for Organ Sharing (UNOS) data, the percentage of GS was closely correlated with graft survival, delayed graft function and primary nonfunction (p < 0.001) (34). Interestingly, in a recently published registry analysis, Sung et al. found no correlation between GS and graft failure in ECD kidney recipients (35). However, as acknowledged by the authors, an important limitation to registry analyses of biopsy data is the potential for selection biases; the transplanted kidneys with greater degrees of GS being more carefully selected. Accordingly, only 17% of transplanted ECD kidneys with a pre-implantation biopsy showed greater than 10% GS in this registry analysis. This important selection bias was eliminated in our study, in which histopathological findings did not influence the allocation of grafts. The highly variable results obtained after transplantation from a marginal donor may be partly due to differences in the percentage of sclerotic glomeruli, which varies considerably between older donors. For example, one study has reported the 95% confidence limits for the percentage of sclerotic glomeruli to be 1.5–23.0% (mean 9.5%) in 75-year-old subjects without evidence of renal disease or hypertension (3). Consistent with this, although the percentage of sclerotic glomeruli was higher in donors >60 years than those aged 50–60 years in our study (12.3 ± 12.6% vs. 8.1 ± 11.5%, p = 0.02), regression analysis failed to show any significant association between donor age and sclerotic glomeruli (r2= 0.015, p = NS). Indeed, 9/20 donors aged >70 years had only ≤10% sclerotic glomeruli. Clearly, a decision to reject these donors on the basis of age alone would have been inappropriate. Nevertheless, GS alone should not be used as the only criterion for kidney selection (4,36).

Consistent with the marginal nature of the donor organs, the recipient graft function was suboptimal, and 42% of patients experienced a delayed graft function, even though most of them had a delayed introduction and low target whole-blood concentrations of their calcineurin inhibitor. Using the Kidney Disease: Improving Global Outcomes (KDIGO) classification (37), 64% and 20% of the recipients, respectively, had renal function grade of 3 T (30– 59 mL/min/1.73 m2) and 4 T (15–29 mL/min/1.73 m2) at 1 year of age. However, death-censored graft survival rates (94.2% at 1 year and 87.5% at 5 years) suggest a reasonable outcome for many patients. Our composite score, based on the donor history of hypertension, serum creatinine ≥ 150 μmol/L before organ recovery, and glomerulosclerosis ≥ 10%, indicates an adjusted OR of 27.5 for low eGFR at 1 year in cases where all three risk factors are present, which is an argument that could be used to decline kidneys. It is interesting, though, that the death-censored graft survival at 5 years reached 60% in the poor-prognosis group (Figure 3), suggesting that the composite score should be used to quantify the risk associated with a given donor rather than exclude potential donors. Allocation of poor-prognosis kidneys should be an individual decision that involves, for a given patient, the unique opportunity of transplantation, expected improvements in quality and quantity of life versus hemodialysis, and the patient's particular risk factors.

In conclusion, clinical scoring systems that are widely used to predict graft outcomes from marginal donors perform poorly compared to histological results from pre-implantation biopsies. We propose a simple and convenient composite scoring system that is strongly predictive of poor graft outcome following transplantation from a marginal donor, and which may allow an optimized allocation of marginal kidneys. Use of this scoring system may improve informed consent by permitting a more accurate discussion of the likelihood of success with the patient prior to transplantation of a graft form an elderly donor. A prospective validation is now required before applying this clinico-histological scoring system in clinical practice.

Conflict of Interest Statement. The authors have no conflicts of interest related to this study. This study received no external funding.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. References
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. References
  8. Supporting Information

Table S1: Cutoff values used by the different allograft biopsy scoring systems.

Please note: Blackwell Publishing are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
AJT_2394_sm_TableS1.doc22KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.