Outcomes of Kidneys from Donors After Cardiac Death: Implications for Allocation and Preservation

Authors


* Corresponding author: Robert A. Montgomery, rmonty@jhmi.edu

Abstract

Although donation after cardiac death (DCD) kidneys have a high incidence of delayed graft function (DGF) and have been considered marginal, no tool for stratifying risk of graft loss nor a specific policy governing their allocation exist. We compared outcomes of 2562 DCD, 62 800 standard criteria donor (SCD) and 12 812 expanded criteria donor (ECD) transplants reported between 1993 and 2005, and evaluated factors associated with risk of graft loss and DGF in DCD kidneys. Donor age was the only criterion used in the definition of ECD kidneys that independently predicted graft loss among DCD kidneys. Kidneys from DCD donors <50 had similar long-term graft survival to those from SCD (RR 1.1, p = NS). While DGF was higher among DCD compared to SCD and ECD, limiting cold ischemia (CIT) to <12 h decreased the rate of DGF 15% among DCD <50 kidneys. These findings suggest that DCD <50 kidneys function like SCD kidneys and should not be viewed as marginal or ECD, and further, limiting CIT <12 h markedly reduces DGF.

Introduction

During the last decade the kidney transplant waiting list has increased more than 260%, yet the number of deceased donor kidney transplants performed has increased by only a modest 16% (1). As the gap between the demand for kidney transplants and the supply of organs widens, expansion of the donor pool has become increasingly compelling. Methods for expanding the number of deceased donor kidneys available have ranged from educating the public about the importance of organ donation (2) to increasing the utilization of marginal donor organs (3).

In 2002 Port and colleagues established a set of characteristics that defined the subgroup of deceased donor referred to as expanded criteria donors (ECD). Kidneys from these donors had a 70% or greater increased risk of graft loss compared to kidneys from standard criteria donors (SCD) (4). Specifically, ECD donors were defined as being older than 60 years or aged 50–59 with two of the following three characteristics: cerebrovascular event as cause of death, history of hypertension, and terminal creatinine of 1.5 mg/dL or higher. In an attempt to decrease allocation time and thus cold ischemic time (CIT), a waiting list flag was implemented to identify patients willing to consider ECD kidneys (5).

At the time ECD criteria were identified, >98% of deceased donor kidneys came from donors after brain death (DBD), with the remaining <2% from donors after cardiac death (DCD). Over the last decade, the utilization of DCD kidneys has increased to 8%, yet factors associated with outcomes for DCD kidneys have not been established. As such, the current national deceased donor allocation policy does not provide guidelines specific for DCD kidneys and the allocation of these kidneys is left to the purview of the organ procurement organization (OPO). DCD transplants are known to have higher rates of delayed graft function (DGF) which in other settings have been associated with poorer long-term results (6–8). In some OPOs, DCD organs are allocated using the ECD list suggesting that operationally these kidneys are considered equivalent to ECD kidneys. In other OPOs, DCD organs are offered to every patient on the deceased donor list but because the prevailing feeling is that these kidneys are marginal, many patients may refuse them adding significantly to the time to allocation.

Considering the increased utilization of DCD organs, a better understanding of the factors that predict outcome would seem imperative, perhaps resulting in a more consistent and informed allocation policy. Over the last decade as experience with DCD transplantation has grown, many single center studies have reported equivalent long-term graft survival and equivalent rates of primary non-function (PNF) between recipients of DCD kidneys and SCD kidneys (10,12–18). Further, several retrospective analyses of the United Network for Organ Sharing (UNOS) database, including reports by Rudich et al. and more recently Doshi and Hunsicker, have compared short and long term outcomes among recipients of DBD and DCD kidneys, and have demonstrated that DCD kidneys have equivalent patient and graft survival rates at 5 years compared to DBD (SCD and ECD) kidneys (19,20).

Forty-four percent of DCD kidneys are currently preserved with pulsatile perfusion (PP), a higher percentage than for either ECD (18.4%) or SCD (11.5%) kidneys. This may be explained by the fact that DCD kidneys have the highest rate of DGF. Among other DBD subgroups, poorer long-term allograft outcomes have been associated with DGF. Thus, a trend toward a decreased risk for DGF among PP preserved deceased donor kidneys has been the justification for pumping such a high percentage of DCD kidneys (reviewed in 6).

Despite intense and renewed focus on the utilization of DCD kidneys, no national study has identified a subset of DCD kidneys with graft survival comparable to SCD kidneys or identified ways to minimize the incidence of DGF. In this study we explored the factors that define risk for graft loss among recipients of DCD kidneys, and evaluated characteristics that distinguish DCD kidneys from ECD kidneys and contribute to the rate of DGF among DCD kidneys. We found that DCD kidneys from donors less than 50 years old have equivalent graft survival to SCD kidneys, and that limiting CIT to less than 12 h is the most effective method for reducing DGF.

Methods

Study design and population

We carried out a secondary data analysis of a prospective cohort study in adult primary deceased donor renal transplant recipients. The study population included patients who are available for analysis in the UNOS Standard Transplant Analysis and Research (STAR) Files. We initially evaluated 170,668 recipients who underwent renal transplantation between January 1993 and December 2005. We then excluded: (1) pediatric recipients (<18 years old) (n = 9434), (2) adult recipients who underwent multi-organ transplantation (n = 2238), (3) adult live donor renal transplant recipients (n = 58,404) and (4) deceased donor recipients with a history of a previous renal transplant (n = 14,122). Enbloc and dual kidney transplants were not excluded (DCD n = 58; DBD n = 2380).

The recipients were stratified according to the subgroup of donor kidney received: ECD, SCD and DCD. ECD was defined as having been DBD with donor age greater than 60 years or age 50–59 years and two of the following three characteristics: donor hypertension, donor creatinine greater than or equal to 1.5 mg/dL, or donor cause of death due to intracranial hemorrhage (n = 12 812). SCD were defined as DBD not meeting ECD criteria (n = 62, 800). DCD donor subgroup was identified based on the UNOS designation of donation after cardiac death, specifically the recovery of organs from a donor whose heart has irreversibly stopped beating (n = 2562). Approximately, 4.7% of information on donor kidney subtype received was missing (n = 7970). Exploratory data analysis demonstrated no differences between donor and recipient pairs with donor kidney subtype information and those without. The data point was thus assumed to be missing at random, and donor–recipient pairs missing this information were excluded.

Death-censored graft survival (DCGS) was used for all analyses. No information on the time of graft loss was missing. DCGS times for patients that died with a functioning graft were defined as the day of death, and these observations were censored. In the overall analysis comparing DCD kidney survival to SCD and ECD kidney survival, no graft losses were excluded based on cause of graft loss.

Almost 40% of information regarding cause of graft loss was missing (n = 66 699). Because more than 10% of the information on cause of graft loss was missing, missing information regarding cause of graft loss was imputed using five iterations of mlogit prediction matching (21). Imputing cause of graft loss was necessary to more accurately define DGF as the need for dialysis in the first post-transplant week not secondary to PNF, hyperacute rejection, graft thrombosis, or surgical complication. Less than 1% of first week post-transplant dialysis information was missing (n = 327), and recipients missing this information were excluded prior to imputation. The imputed information was only used for comparing survival of kidneys with DGF to those without DGF within donor subgroups and comparing survival of kidneys with DGF across donor subgroups. For the analyses comparing outcomes between kidneys that developed DGF and kidneys that did not develop DGF, graft losses due to primary non-function (PNF), hyperacute rejection, thrombosis and surgical complications were excluded. Because the definition of DGF excludes those causes of graft loss, all graft losses not due to DGF, including those due to PNF, hyperacute rejection, thrombosis, and surgical complications, would have by default been analysed in the no DGF subgroup. This would have artificially lowered the graft survival rate for the no DGF group, and have made DGF seem like a predictor of improved graft survival.

Statistical analyses

Unadjusted DCGS was estimated using Kaplan–Meier methodology, and the survival functions between donor subgroups, by preservation method, and based on the development of DGF were compared with the log-rank test. To determine DCD donor factors predictive of long-term DCGS and the impact of preservation method on long-term DCGS, we developed adjusted Cox proportional hazards models specific to the DCD donor subgroup, based upon preservation method and specific to recipients of DCD kidneys who went on to develop DGF.

Identification of which recipient and donor characteristics that were important in the regression models occurred in two phases. In the first phase, we identified all UNOS donor and recipient covariates that were found to be statistically significant (p < 0.05) predictors of the outcome measure on unadjusted univariate analysis and included these variables in an initial multivariate model. Exploratory data analysis was then used to determine the most functional form of each covariate. In the second phase, a parsimonious model was developed using stepwise testing, forward and backward selection, of nested models looking for a reduction in the Akaike Information Criterion (AIC). Utilizing the AIC allowed for the identification of the fewest number of variables necessary to explain a given outcome within the data set and thus prevented the development of an over fitted model or a model that is so complex it is no longer generalizable. Further, minimization of the number of variables necessary to explain a given outcome maximized the number of donor–recipient pairs available for evaluation. This became particularly important when analysing outcomes in the DCD subgroup, as fewer than 3000 DCD kidney transplants were performed and 388 censoring events occurred. Specifically, covariates were excluded from the parsimonious model: (1) if their removal resulted in a reduction in the AIC and (2) if the group of covariates eligible for elimination did not result in a statistically significant finding on likelihood ratio testing (p < 0.05). As a result only covariates that did not explain the given outcome measure within the context of this data set were excluded from the parsimonious model. The final phase involved testing the sensitivity with which the parsimonious model predicted the given outcome measure. The sensitivity analysis was performed by comparing the results of the regression analysis generated from the full multivariate model to the results of the regression analysis generated from the final parsimonious model.

All tests were two-sided with statistical significance set at the α= 0.05 level. All analyses were performed using STATA 9.1 for Linux (Stata Corp, College Station, TX).

Results

A total of 78,174 adult deceased donor renal transplant recipients were identified who met criteria for inclusion in this analysis (Table 1). These recipients were stratified according to donor subgroup: DBD (included SCD (n = 62,800) and ECD (n = 12,812)) and DCD (n = 2562).

Table 1.  Comparison of donor and recipient characteristics by donor subgroup, donation after cardiac death (DCD) and donation after brain death (DBD). The DBD subgroup consists of both standard criteria donors (SCD) and expanded criteria donors (ECD)
CharacteristicsDCD (n = 2562)DBD (n = 75,612)p-value
  1. 1p > 0.05.

  2. 2Age in years.

  3. 3Mean peak panel reactive antibody.

Recipient
 Mean age25049<0.001
 Gender
   Male59.9%60.8%NS1
   Female40.1%39.2% 
 Ethnicity
   White51.2%51.7%<0.001
   Black33.0%29.4% 
   Hispanic 8.7%11.8% 
   Asian 5.2% 4.7% 
   Other 1.9% 3.4% 
 Peak PRA310.8 11.3 NS1
Donor
 Mean age23737NS1
 Gender
   Male67.0%58.7%<0.001
   Female33.0%41.4% 
 Ethnicity
   White85.8%74.3%<0.001
   Black 7.4%11.3% 
   Hispanic 4.9%11.3% 
   Asian 1.2% 1.9% 
   Other 0.7% 1.2% 

Graft survival

Unadjusted univariate analyses were performed on all available UNOS variables to identify recipient, donor, and/or graft characteristics predictive of graft loss in all DCD kidneys, DCD kidneys from donors <50 years, and DCD kidneys from donors ≥50 years (subset listed in Table 2). Overall, recipient age, ethnicity, blood type, peak panel reactive antibody (PRA), body mass index (BMI), diagnosis and donor age, ethnicity, blood type, blood urea nitrogen (BUN), cause of death and hypertension were most predictive of allograft outcome. In addition, among DCD kidneys from donors <50, CIT > 24 h and pulsatile perfusion preservation were associated with worse allograft survival.

Table 2.  Unadjusted Cox proportional hazards univariate analyses of multiple recipient (A), donor (B) and graft (C) characteristics. Characteristics that were statistically significant predictors of graft loss were used in multivariate models
 All DCDDCD < 50DCD > 50
Hazard ratioCIp-ValueHazard ratioCIp-ValueHazard ratioCIp-Value
(A) Recipient characteristics
 Age (18–39)1.47[1.18, 1.82]<0.0011.64[1.26, 2.12]<0.0010.99[0.98, 1.01]0.2
 Gender1.04[0.85, 1.28]0.71.06[0.83, 1.36]0.61.03[0.73, 1.46]0.9
 Race
   White0.87[0.71, 1.06]0.20.81[0.63, 1.03]0.080.96[0.68, 1.35]0.8
   Black1.29[1.05, 1.58]0.021.32[1.01, 1.72]0.051.24[0.87, 1.76]0.2
   Hispanic0.79[0.54, 1.16]0.20.96[0.61, 1.51]0.90.84[0.37, 1.91]0.7
   Asian1.05[0.66, 1.67]0.81.57[0.93, 2.65]0.090.45[0.14, 1.41]0.2
 Blood type
   A0.88[0.72, 1.09]0.20.89[0.69, 1.15]0.40.87[0.61, 1.27]0.5
   B1.41[1.07, 1.84]0.011.53[1.09, 2.14]0.011.1[0.69, 1.77]0.7
   O0.98[0.80, 1.2]0.80.92[0.72, 1.18]0.51.1[0.79, 1.57]0.5
   AB0.84[0.51, 1.39]0.50.95[0.54, 1.67]0.90.69[0.22, 2.16]0.5
 HLA mismatch
   A1.15[0.99, 1.32]0.071.15[0.97, 1.37]0.11.06[0.82, 1.38]0.6
   B1.11[0.97, 1.28]0.11.1[0.92, 1.31]0.31.11[0.86, 1.47]0.4
   DR1.02[0.89, 1.17]0.80.99[0.84, 1.18]0.91.05[0.83, 1.33]0.7
   Six Ag1.29[0.97, 1.72]0.091.26[0.87, 1.83]0.3 
   Zero Ag0.77[0.5, 1.17]0.20.76[0.46, 1.27]0.30.81[0.38, 1.73]0.6
 Peak PRA (50–80)1.73[1.08, 2.78]0.021.7[1.17, 2.47]0.0050.99[0.99, 1.01]0.6
 Wait > 950 days1.15[0.86, 1.52]0.31.44[0.96, 2.16]0.081.62[1.12, 2.36]0.01
 Hospitalized1.57[0.74, 3.32]0.21.19[0.4, 3.2]0.74.26[1.35, 13.5]0.01
 Dialysis
   None0.93[0.75, 1.16]0.50.86[0.65, 1.14]0.30.99[0.69,1.45]0.9
   Peritoneal0.89[0.64, 1.23]0.50.91[0.62, 1.34]0.60.95[0.51, 1.76]0.9
   Hemodialysis1.12[0.91, 1.38]0.31.2[0.9, 1.6]0.21.02[0.71, 1.46]0.9
 Diabetes1.07[0.85, 1.35]0.61.04[0.67, 1.61]0.91.01[0.67, 1.51]0.9
 BMI >301.28[1.01, 1.64]0.041.32[1.0, 1.74]0.051.67[1.13, 2.45]0.01
 Diagnosis
   FSGS1.52[1.05, 2.19]0.031.62[1.06, 2.49]0.031.39[0.68, 2.84]0.4
   PKD0.55[0.37, 0.82]0.0030.38[0.31, 0.68]<0.0010.82[0.48, 1.41]0.5
(B) Donor characteristics
 Age
   < 50 years0.6[0.49, 0.74]<0.0010.6[0.49, 0.74]<0.0010.6[0.49, 0.74]<0.001
   ≥50 years1.7[1.35, 2.06]<0.0011.7[1.35, 2.06]<0.0011.7[1.35, 2.06]<0.001
 Gender1.17[0.95, 1.45]0.10.99[0.76, 1.3]0.91.4[0.99, 1.98]0.05
 Race
   White1.19[0.88, 1.62]0.31.09[0.77, 1.54]0.61.22[0.62, 2.41]0.6
   Black1.04[0.72, 1.52]0.821.14[0.76, 1.72]0.51.28[0.47, 3.48]0.6
   Hispanic0.42[0.22, 0.81]0.010.52[0.26, 1.06]0.070.18[0.03, 1.28]0.09
   Asian1.74[0.77, 3.89]0.21.66[0.53, 5.2]0.41.52[0.48, 4.8]0.5
 Blood type
   A0.93[0.75, 1.14]0.50.97[0.89, 1.06]0.80.88[0.61, 1.27]0.5
   B1.4[1.06, 1.85]0.021.51[1.07, 2.14]0.021.12[0.7, 1.8]0.6
   O0.94[0.77, 1.14]0.50.86[0.68, 1.1]0.21.09[0.77, 1.54]0.6
   AB0.87[0.48, 1.58]0.70.99[0.51, 1.93]0.90.68[0.17, 2.7]0.6
 BUN >151.28[1.05, 1.56]0.021.21[0.94, 1.55]0.11.16[0.82, 1.66]0.4
 Creatinine >1.51.03[0.77, 1.39]0.81.22[0.87, 1.72]0.30.67[0.37, 1.22]0.2
 Cardiac Arrest1.11[0.77, 1.59]0.61.12[0.72, 1.73]0.61.28[0.66, 2.47]0.5
 Inotropes required0.99[0.81, 1.22]0.91.1[0.86, 1.43]0.40.81[0.56, 1.16]0.2
 Cause of death: CVA1.55[1.24, 1.93]<0.0011.5[1.12, 2.02]0.0071.22[0.85, 1.73]0.3
 BMI >351.61[1.15, 2.24]0.0051.49[0.97, 2.29]0.071.84[1.08, 3.13]0.02
 Diabetes1.39[0.91, 2.11]0.11.42[0.7, 2.87]0.30.99[0.58, 1.7]0.9
 Hypertension1.74[1.37, 2.22]<0.0011.67[1.15, 2.42]0.0071.36[0.95, 1.93]0.09
 Smoker0.96[0.78, 1.19]0.70.93[0.71, 1.22]0.60.78[0.55, 1.1]0.2
 Infection1.14[0.9, 1.44]0.31.07[0.79, 1.44]0.671.21[0.82, 1.79]0.3
 Prior malignancy1.16[0.64, 2.11]0.61.14[0.5, 2.4]0.71.17[0.43, 3.19]0.8
(C) Graft characteristics
 Pump1.21[0.99, 1.48]0.061.35[1.06, 1.73]0.020.91[0.64, 1.28]0.6
 CIT >24 hrs1.23[0.99, 1.52]0.061.3 [1.0, 1.69] 0.050.97[0.62, 1.51]0.9
 WIT >30 min1.04[0.78, 1.4] 0.8 1.05[0.78, 1.4] 0.8 1.21[0.65, 2.23]0.5
 Uncontrolled1.21[0.85, 1.73]0.3 1.21[0.85, 1.73]0.3 1.26[0.51, 3.14]0.6
 Transplant Year0.97[0.94, 1.01]0.1 0.96[0.92, 1.0] 0.060.99[0.94, 1.06]0.8
 Sharing
   Regional1.28[0.99, 1.66]0.061.26[0.92, 1.74]0.21.52[0.74, 3.1] 0.2
   National1.38[0.91, 2.08]0.1 1.34[0.8, 2.2]  0.31.34[0.86, 2.09]0.2

Adjusted Cox proportional hazards analysis demonstrated that recipients of DCD kidneys had a 21% increased risk for graft loss (adjusted hazard ratio (AHR) 1.21, p = 0.001) as compared with recipients of SCD kidneys (Table 3A). When compared with recipients of ECD kidneys, DCD recipients had a 33% decreased risk for graft loss (AHR 0.67, p < 0.001) (Table 3B). By unadjusted Kaplan–Meier survival analysis, these differences translated to a 2% reduction in 5-year DCGS comparing recipients of DCD and SCD kidneys (77.9 and 79.9%, respectively, p < 0.001), and an 11% increase in 5-year DCGS comparing recipients of DCD and ECD kidneys (77.9 and 66.9%, respectively, p < 0.001) (Figure 1).

Table 3.  Results from adjusted regression models estimating the impact of donor subgroup on risk for graft loss. Donation after cardiac death (DCD) kidneys perform slightly inferior to standard criteria donor (SCD) kidneys (A), and perform superior to expanded criteria donor (ECD) kidneys (B)
 Hazard ratioCIp-Value
(A) Donor characteristics
 Donor type
   SCDReference 
   DCD1.26[1.11, 1.43]<0.001
 BMI >351.09[1.0, 1.2]   0.05 
 Blood type
   AReference 
   B1.17[0.94, 1.46] 0.1  
   O0.96[0.82, 1.11] 0.6  
   AB1.01[0.82, 1.23] 0.9  
 BUN >151.06[1.01, 1.11] 0.02 
 Cause of death: CVA1.25[1.2, 1.31] <0.001
 Hypertension1.27[1.19, 1.35]<0.001
Recipient characteristics
 Age (18–39)1.52[1.45, 1.59]<0.001
 BMI >301.18[1.12, 1.24]<0.001
 Blood type
   AReference 
   B0.87[0.7, 1.08]  0.2  
   O1.06[0.91, 1.23] 0.5  
   AB1.03[0.87, 1.23] 0.7  
 Race
   WhiteReference 
   Black1.7 [1.62, 1.78]<0.001
 Diagnosis
   FSGS1  [0.92, 1.09] 0.9  
   PKD0.69[0.63, 0.75]<0.001
 PRA (50–80)1.19[1.06, 1.33] 0
Graft characteristics
   CIT >24 hours1.13[1.08, 1.18]<0.001
(B) Donor characteristics
 Donor type
   ECDReference  
   DCD0.74[0.63, 0.86]<0.001
 BMI >350.93[0.81, 1.06] 0.3  
 Blood Type   
   AReference  
   B0.81[0.52, 1.23] 0.3  
   O0.8 [0.59, 1.08] 0.1  
   AB0.94[0.63, 1.4]  0.8  
 BUN >151.12[1.04, 1.21]0  
 Cause of death: CVA1.14[1.0, 1.29]  0.04 
 Hypertension1.12[1.03, 1.21] 0.01
Recipient characteristics
   Age (18–39)1.24[1.12, 1.36]<0.001
   BMI >301.19[1.09, 1.3] <0.001
 Blood type
   AReference  
   B1.31[0.86, 1.99] 0.2  
   O1.27[0.94, 1.71] 0.1  
   AB0.96[0.7, 1.33]  0.8  
 Race
   WhiteReference  
   Black1.41[1.3, 1.53] <0.001
 Diagnosis
   FSGS1.02[0.87, 1.21] 0.8  
   PKD0.78[0.68, 0.9] 0  
 PRA (50–80)1.19[0.97, 1.45] 0.09 
Graft characteristics
   CIT >24 hours1.12[1.04, 1.21]0  
Figure 1.

Kaplan–Meier (KM) death-censored graft survival (DCGS) curves by donor subgroup: standard criteria donor (SCD), donation after cardiac death (DCD), and expanded criteria donor (ECD). Based upon 5-year DCGS, it is clear that not all DCD kidneys are marginal.

As expected, each ECD criterion previously identified as being significant accurately predicted increased risk for graft loss among the DBD donor population in our analysis (p < 0.001) (Table 4). However, the only ECD criterion predictive of increased risk for graft loss in the DCD donor population was age (>60 years: unadjusted hazard ratio (HR) 1.78, p = 0.001; 50–59 years: HR 1.48, p < 0.001) (Table 4). Unlike the DBD recipient population, increased risk for graft loss in DCD recipients aged 50–59 occurred regardless of donor hypertension, terminal creatinine, or cause of death.

Table 4.  Results from Cox proportional regression analysis of each ECD criterion by donor subgroup, donation after cardiac death (DCD) and donation after brain death (DBD). As previously cited in the literature ECD criteria accurately predict worse outcomes for recipients of DBD kidneys. However, the only ECD criterion predictive of increased graft loss among recipients of DCD kidneys is donor age
 Expanded criteria donor characteristics
≥60 years50–59 yearsSCr > 1.52Hypertension3COD ICH4
  1. 1p > 0.05. 2Serum creatinine in mg/dL in a donor aged 50 to 59 years. 3History of hypertension in a donor aged 50 to 59 years. 4Cause of death, intracranial hemorrhage in a donor aged 50 to 59 years.

DCD (p-value)1.78 (0.001) 1.48 (0.001) NS1NS1NS1
DBD (p-value)1.90 (<0.001)1.37 (<0.001)1.14 (0.001)1.31 (<0.001)1.21 (<0.001)

When stratified by age, DCD kidneys from donors older than 50 years have a 80% increased risk for graft loss and greater than a 15% reduction in 5 year DCGS compared to DCD kidneys from donors younger than 50 years (AHR 1.8, p < 0.001; 65.9 and 81.6%, respectively, p < 0.001) (Table 5 and Figure 2). The risk for graft loss and DCGS was equivalent between recipients of either a SCD kidney or a DCD kidney from a donor younger than 50 years (AHR 1.1, p > 0.05; 79.9 and 81.6%, p > 0.05) (Table 4 and Figure 3). Recipients of DCD kidneys from donors older than 50 years and recipients of ECD kidneys had similar risks for graft loss and 5 year DCGS, and compared to recipients of SCD kidneys their outcomes were significantly worse (DCD ≥ 50 years: AHR 1.77, p < 0.001 and 65.9%, p < 0.001; ECD: AHR 1.8, p < 0.001 and 66.9%, p < 0.001) (Table 6 and Figure 3).

Table 5.  Results from the adjusted regression models estimating the impact of donation after cardiac death (DCD) donor age on risk for graft loss. Recipients of DCD kidneys from donors older than 50 years have significantly higher risk for graft loss compared to recipients of DCD kidneys from donors younger than 50 years
 Hazard ratioCIp-Value
Donor
 Age 
   <50 yearsreference 
   >50 years1.8 [1.39, 2.22]<0.001
   BMI >351.58[1.12, 2.24] 0.009
 Gender
   Malereference 
   Female1.06[0.84, 1.33] 0.6  
Recipient
 BMI >301.34[1.06, 1.71] 0.02 
 Hospitalized1.44[0.59, 3.49] 0.4  
 Wait >950 days1.14[0.89, 1.46] 0.3  
Figure 2.

Kaplan–Meier (KM) death-censored graft survival (DCGS) curves for recipients of donation after cardiac death (DCD) kidneys by donor age. Kidneys from DCD donors greater than 50 years old perform inferior to kidneys from DCD donors younger than 50 years old (p < 0.001).

Figure 3.

Kaplan–Meier (KM) death-censored graft survival (DCGS) curves for recipients of standard criteria donor (SCD) kidneys, donation after cardiac death (DCD) kidneys from donors younger than 50 years, DCD kidneys from donors older than 50 years and expanded criteria donor kidneys (ECD). With regard to 5-year DCGS, SCD kidneys and DCD kidneys from donors younger than 50 years have equivalent outcomes, and ECD kidneys and DCD kidneys from donors older than 50 years have equivalent outcomes.

Table 6.  Results from the adjusted regression models and Kaplan–Meier survival analysis estimating the impact of donor subgroup on risk for graft loss and 5-year death-censored graft survival (DCGS). Donation after cardiac death (DCD) kidneys from donors younger than 50 years perform as well as kidneys from standard criteria donors (SCD). DCD kidneys from donors older than 50 years perform similar to kidneys from expanded criteria donors (ECD)
 Risk of graft lossGraft survival1
Hazard ratiop-Value5-yearp-Value
  1. 1Death-censored graft survival.

  2. 2p > 0.05.

SCDreference 79.9 
DCD <50 years1.1 NS281.6NS2
DCD ≥50 years1.77<0.00165.9<0.001
ECD1.8 <0.00166.9<0.001

Graft function

Compared to SCD kidneys, DCD kidneys had a higher incidence of PNF (0.7 and 1.6%, respectively, p < 0.001), whereas compared to ECD kidneys, DCD kidneys had a lower incidence of PNF (1.82 and 1.6%, respectively, p = 0.1). DCD kidneys from donors younger than 50 years had a slightly higher, but not statistically significant, incidence of PNF compared to DCD kidneys from donors older than 50 years (1.71 and 1.33%, respectively, p = 0.1).

With regard to DGF, recipients of DCD kidneys were found to have a higher incidence of DGF than DBD kidneys (DCD 38.7%, SCD 19.5%, ECD 30%, p < 0.001). Again, DCD kidneys from donors younger than 50 years were associated with a lower incidence of DGF (36% compared to 45.9% for DCD kidneys from donors older than 50 years, p < 0.001). Recipients of deceased donor kidneys that experienced DGF did have inferior 5-year DCGS (72.5% vs. 82.9% for those that did not develop DGF, p < 0.001). Interestingly among the subgroup that developed DGF, DCD kidneys from donors <50 tolerated DGF better than either SCD or ECD kidneys. In fact, DCD kidneys from donors <50 with DGF were associated with a 23% decrease in risk for graft loss compared to SCD kidneys with DGF and a 52% decrease compared to ECD kidneys with DGF (AHR 0.77, p = 0.03 and AHR 0.48, p < 0.001, respectively). Furthermore, recipients of DCD kidneys from donors younger than 50 years that developed DGF had 5 year DCGS of 81.1% compared to a 75% 5 year DCGS among recipients of SCD kidneys that developed DGF (p < 0.001). A similar trend was seen with DCGS for recipients of DCD kidneys from donors older than 50 years with DGF compared to recipients of ECD kidneys with DGF (67.4 and 63.3%, respectively, p < 0.01).

CIT was strongly correlated with the development of DGF. Specifically, the incidence of DGF among DCD kidneys was reduced 15% when CIT was less than 12 h compared to CIT greater than 12 h (p < 0.001). In addition, with CIT greater than 12 h there is an incremental increase in DGF among DCD kidneys up to 63.1% for those with CIT greater than 48 h. Despite the fact that more than 80% of DCD kidneys were procured locally, greater than 83% of DCD kidneys were found to have a CIT longer than 12 h.

Nearly half of all DCD kidneys (43.8%; n = 1122) were preserved using pulsatile perfusion (PP) preservation. Compared to cold storage (CS), PP preservation resulted in a 3.8% reduction in the incidence of DGF (CS: 42.3% vs. PP: 38.5%, p < 0.001). PP substantially reduced the incidence of DGF among DCD kidneys from donors older than 50 (CS: 52.8% vs. PP: 44%, p < 0.001), however, PP was not associated with improved graft survival compared to CS (p > 0.05). Among DCD kidneys from donors younger than 50, PP was associated with a 2.4% decrease in incidence of DGF, but PP was associated with a 35% increased risk for graft loss (HR 1.35, p = 0.02).

Furthermore, in adjusted models that accounted for factors other than the preservation method that might influence the outcomes of either group, DCD kidneys from donors younger than 50 preserved with PP were also found to have significantly lower graft survival and a 1.38 increased risk for graft loss compared to CS kidneys (AHR 1.38, p = 0.02) (Table 7A). Interestingly within this same donor subgroup, the 34.5% of patients receiving a PP-preserved kidney who experienced DGF had even worse graft survival when compared to the 37.1% of patients receiving CS kidneys who experienced DGF (5-year DCGS: 75% vs. 85%; AHR 1.95, p < 0.04) (Table 7B and C). Among DCD kidneys from donors older than 50 years, no statistically significant differences were seen in graft survival based on preservation method (Table 8).

Table 7.  Unadjusted 5-year death-censored graft survival (DCGS) and adjusted risk for graft loss among cold storage versus pulsatile perfused DCD kidneys from donors younger than 50 years. Pulsatile perfused DCD kidneys from donors younger than 50 years had worse 5-year DCGS and increased risk for graft loss compared to young DCD kidneys that were cold preserved, particularly among the subset of kidneys that go on to develop delayed graft function
(A) Multivariate model: DCD < 50 years
 Hazard ratioCIp-Value
Graft
 Pump1.38[1.05, 1.81]0.02 
 CIT > 241.3 [0.99, 1.72]0.06 
Donor
 BMI > 402.43[1.32, 4.46]0    
 Blood type B0.33[0.09, 1.28]0.1  
 Cause of death: CVA1.36[0.97, 1.92]0.08 
 Hypertension1.79[1.16, 2.76]0.01 
Recipient
 Age (18–39)1.46[1.09, 1.96]0.01 
 Blood Type B4.74[1.24, 18.1]0.02 
 Race (black)1.19[0.88, 1.61]0.3  
 PRA (20–80)1.47[0.96, 2.24]0.08 
Diagnosis 
FSGS1.55[0.96, 2.5] 0.07 
 PKD0.5 [0.27, 0.93]0.03 
(B) Multivariate model: DCD < 50 years & DGF
 Hazard ratioCIp-Value
 
Graft
 Pump1.95[1.06, 3.57]0.03 
 CIT > 241.06[0.59, 1.88]0.9  
Donor
 BMI > 350.66[0.23, 1.88]0.4  
 Cause of death: CVA1.73[0.86, 3.46]0.1  
 Hypertension2.63[1.1, 6.33] 0.03 
Recipient
 Age (18–39)2.04[1.15, 3.61]0.01 
 Race (black)1.1 [0.6, 2.0]  0.8  
 PRA (20–80)1.08[0.11, 7.57]0.9  
 Diagnosis
   FSGS1.27[0.29, 5.54]0.7  
   PKD0.81[0.29, 2.27]0.9  
(C) 5-year death-censored graft survival
 Cold storagePulsatile perfusionLog-rank
 
All84.6%82.3%0.05
DGF84.7%75.3%<0.001 
Table 8.  Unadjusted 5-year death-censored graft survival (DCGS) and adjusted risk for graft loss among cold storage versus pulsatile perfused DCD kidneys from donors older than 50 years. Pulsatile perfusion is not associated with improved outcomes in DCD kidneys from donors older than 50 years
(A) Multivariate model: DCD ≥ 50 years
 Hazard ratioCIp-Value
Graft
 Pump 1.29[0.63, 2.67]0.5  
 CIT > 24 hrs 1.03[0.49, 2.14]0.9  
Donor
 BMI > 35 1.74[0.61, 4.97]0.3  
 Gender0.8[0.38, 1.66]0.5  
Recipient
 BMI > 30 0.69[0.29, 1.63]0.4  
 Hospitalized72.3 [5.87, 891]0.001
 PRA 0.99[0.97, 1.01]0.2  
 Wait > 950 days 1.99[0.96, 4.16]0.07 
(B) Multivariate model: DCD ≥ 50 years with DGF
 Hazard ratioCIp-Value
 
Graft
 Pump 0.88[0.39, 1.73] 0.6  
 CIT > 24hrs 1.13[0.58, 2.21] 0.7  
Donor
 BMI > 35 2.06[0.74, 5.8]  0.2  
Recipient
 BMI > 30 0.73[0.28, 1.88] 0.5  
 Hospitalized46.2 [3.81, 560.4]0    
 PRA 0.67[0.08, 5.47] 0.7  
 Wait > 950 days 1.91[0.91, 3.99] 0.09 
(C) 5-year death-censored graft survival
 Cold storagePulsatile perfusionLog-rank
 
All65.6%71.6%0.9
DGF68.4%65.9%0.3

Discussion

Kidney transplantation is associated with a 68% reduction in long-term mortality compared to patients that remain on dialysis (22,23). This improvement in long-term patient survival has garnered tremendous support for kidney transplantation and resulted in a 260% increase in the deceased donor kidney transplant waiting list over the last decade (1). Unfortunately the current supply of deceased donor kidneys cannot meet the increasing demand for donor kidneys, and more than 6% of patients on the deceased donor waiting list die each year awaiting transplantation (24). The expansion of the donor pool to include kidneys from ECD donors has increased the donor pool modestly (approximately 10%) but at some cost in terms of outcomes (1,4,25). Multiple studies have demonstrated that nationwide optimization of DCD procurement could increase the donor pool between 11 and 350%, resulting in 1338–133, 700 more kidney transplants each year (9–11). Despite this fact, many in the transplant community have been slow to embrace DCD kidneys due to concerns about increased DGF rates and inferior long-term results (6). In many OPOs DCD kidneys are allocated only to recipients on the ECD list.

Identification of donor characteristics predictive of long-term graft survival among recipients of DCD kidneys is paramount to more widespread acceptance and utilization of this valuable donor source. We demonstrated that the only criterion in the current ECD definition (4) that significantly predicted graft survival in DCD kidneys was donor age. Specifically, DCD kidneys from donors younger than 50 years old performed as well as SCD kidneys with regard to long-term graft survival, and DCD kidneys from donors older than 50 years had graft survival rates comparable to ECD kidneys at 5 years. These findings support previous single center reports demonstrating equivalent long-term outcomes between DCD and SCD kidneys (13).

DCD donor kidneys are associated with a 20% increased incidence of DGF. However, as previously reported (20,25), we found that the increased incidence of DGF did not translate to worse overall long-term graft survival for DCD kidneys. Among those kidneys that went on to develop DGF, DCD kidneys tolerated DGF better than DBD kidneys with 23–52% decrease in risk for graft loss. Further, we found that when CIT was limited to less than 12 h, the rate of DGF in DCD kidneys (25.2%) approached that of SCD kidneys (19.5%).

Interestingly, while our analysis confirmed the ability of PP preservation to reduce the incidence of DGF, this effect was minimal among DCD kidneys from donors younger than 50. Perhaps even more importantly, PP was associated with decreased DCGS, particularly for recipients that experienced DGF. These findings suggest that at least among DCD kidneys from donors younger than 50 PP is not associated with clear benefit, whereas reducing CIT to less than 12 h is associated with lower rates of DGF and improved long-term outcomes. Since PP is often associated with prolongation of CIT, use of PP for DCD organs, especially when the donor is <50 years, may need to be reconsidered.

The fact that DGF does not appear to affect long-term survival among DCD kidneys remains counter-intuitive. It has been previously shown in both small single center studies and in rat transplantation models that the long-term functional consequences of DGF may be related more to the injury of brain death than ischemia reperfusion (27).

DCD kidneys are still considered by many in the transplant community to be marginal organs and this may in fact be contributing to the increased rates of DGF. Carter et al. found that kidneys labeled as marginal were often turned down by multiple centers prior to placement, resulting in prolonged CIT and increased rates of DGF (28). Additionally, the study demonstrated that proper allocation of marginal kidneys could result in decreased CIT and DGF rates (28). Our data further emphasize the importance of proper allocation of DCD kidneys given the finding of a 15% reduction in the rate of DGF among DCD kidneys with less than 12 h of CIT. Interestingly, despite greater than 80% of DCD kidneys being allocated locally, more than 85% of DCD kidneys were found to have CIT greater than 12 h, suggesting the need for both systems and practice changes. The ability of PP to reduce the incidence of DGF among DCD kidneys appears limited, and with the finding of an association between increased risk for graft loss and PP preserved DCD kidneys, the utilization of PP preservation among DCD kidneys may need to be further investigated.

We must acknowledge several limitations of our study. Our study is retrospective and is subject to the many confounders inherent in any large database analysis, such as residual and unmeasured confounding and missing data. The process of imputing missing data, for example, introduces potential bias by assuming that values are missing at random. In addition, we attempted to account for selection bias by developing multivariate models that controlled for multiple confounding factors and thus allowed us to compare outcomes by preservation method among donor–recipient pairs that were equivalent in other ways. However, the UNOS database does not contain variables on surgeon rationale for PP over CS or the type of PP machine used, and as a result, our study is an analysis of results from an already selected process based on clinical intuition and clinical decision making. Furthermore, while we did demonstrate an association between PP and increased risk for graft loss, we also acknowledge that small single-center reports have demonstrated improved outcomes with PP preservation (reviewed in 29). Thus, it may be prudent to identify centers with good PP preservation outcomes, establish best practices, and formally evaluate the utility of PP versus CS preservation in a randomized controlled clinical trial.

Our findings suggest that DCD kidneys from donors younger than 50 should be allocated using the standard deceased donor waiting list, whereas the ECD list should be used for DCD kidneys from donors older than 50. This may have the effect of reducing time to allocation and thus CIT. The findings of this study further suggest that expansion of DCD utilization, especially from donors <50 years, could be accomplished without a reduction in allograft function or survival. Perhaps with a more precise definition of marginal kidneys and with the recognition that DGF can be minimized in DCD kidneys, more surgeons will be comfortable transplanting DCD kidneys, and as a result more OPO's will begin to recover kidneys from DCD donors ultimately resulting in much needed expansion of the deceased donor pool.

Acknowledgment

Dr. Segev is supported by the American Society of Transplantation Clinical Science Faculty Development Grant.

Ancillary