Abstract
ABSTRACT. Donor-recipient age matching has been proposed as a means of improving overall outcomes in deceased donor renal transplantation. It was hypothesized that donor-recipient age matching would improve patient survival time in younger recipients while not adversely affecting patient survivals in older recipients because they seldom outlive their grafts. By use of data from United Network of Organ Sharing Standard Transplant and Analysis and Research Files 50,320 patients were identified who underwent a first deceased donor renal transplantation between January 1, 1990, and December 31, 1997. Adjusted patient survival and death-with-graft function patient survival were analyzed from the date of transplantation. Patient survival was affected by donor age for all recipient age groups, including recipients older than 55 yr. The effect of donor age on patient survival is greater than that seen with HLA matching. The effect of donor age on patient survival persisted even when censoring recipients in whom grafts failed before death, suggesting that both longevity and quality of graft function are important in patient survival. Donor-recipient age matching is occurring to a limited degree in this population. Donor-recipient age matching would improve survival in younger recipients but would adversely affect survival in older patients by reducing the availability of younger donor kidneys for this group. The issue of donor-recipient age matching needs to be debated among the public and transplantation community so that a logical and just system can be developed.
Patient mortality after deceased donor renal transplantation is lower than in patients on dialysis waiting for a transplant (1); however, limited organ availability is resulting in increased waiting times for patients seeking renal transplants from deceased donors (2). The most common cause of graft failure is death of the patient with a functioning graft (3). The risk of dying with a functioning graft increases with increasing recipient age (4). The United Network of Organ Sharing (UNOS) organ allocation system uses HLA matching as a means to improve the outcome of transplantation but gives no consideration to donor-recipient age matching (5). As the waiting list grows and waiting times for deceased donor renal transplantation increase, pressure is growing on transplantation programs to use more expanded-criteria donors (ECD). Because of the donor shortage, maximizing utility of available organs is increasingly important. Use of older donor kidneys in young recipients leads to greater organ wastage due to early graft failure with the need for retransplantation, whereas old recipients receiving younger kidneys die with functional grafts, reducing the potential lifespan of the organ had it been given to a younger recipient. Giving young recipients older kidneys, which have little chance of functioning for the life of the patient, exposes them to the risks of sensitization, to a return to dialysis with its increased morbidity and mortality, and to more uremic exposure with its associated morbidity and mortality. On the other hand, reserving ECD kidneys for older patients may also influence patient outcomes in this population. The purpose of the study presented here is to examine the effect of donor age on patient survival after first deceased donor renal transplantation to better understand how donor-recipient age matching might affect patient outcomes.
Materials and Methods
Patients undergoing renal transplantation in the United States were identified via the Organ Procurement and Transplant Network (OPTN) database obtained from the Standard Transplant Analysis and Research File (OPTN data as of August 27, 2002). All first deceased donor renal transplantations performed between January 1, 1990, and December 31, 1997, were identified. Multiorgan transplants and patients with previous organ transplants were excluded from the analysis.
A total of 50,322 patients who met the above criteria for analysis were identified. Two patients were excluded from the study because of missing donor age data. Patients who received a first deceased donor renal transplant were divided into groups on the basis of recipient age, donor age, and the difference in donor recipient age for analysis. Death-with-graft function patient survivals were determined by censoring all patients who died without graft function at the time of death. Patient survival and death-with-graft function patient survivals were calculated from the date of transplantation.
Cox proportional-hazard modeling was performed by forward stepwise regression on SPSS software version 10.1. Adjusted 5- and 10-yr survivals were calculated at the mean of the covariates for the population. The following covariates were included in the model: transplantation year, cytomegalovirus status, kidney diagnosis, recipient age, race and gender, peak panel reactive antibody, HLA mismatch, donor age, race and gender, prior dialysis treatment, and cold ischemia time. Only covariates that would be typically available at the time of transplantation and were used to determine the allocation of the kidney were included in the model.
To determine the difference between the actual and random donor and recipient age distribution of kidneys, the total number of kidneys and the percentage of total number of kidneys were determined in each donor age group. The projected number of kidneys in each donor recipient age category given random allocation was determined by multiplying the percentage of kidneys for a given donor age group by the total number of recipients in a given recipient age group. The percentage difference from random allocation was calculated by subtracting the expected number of kidneys from the actual number of kidneys divided by the expected number of kidneys and multiplied by 100.
To determine the 10-yr survivals based on different donor-recipient age matches, the recipients were divided into three groups on the basis of recipient age (0 to 40 yr, 41 to 54 yr, and 55 yr or older). For each group, adjusted patient survivals were determined by Cox proportional-hazard modeling at the mean of the covariates for five donor age categories (0 to 17 yr, 18 to 29 yr, 30 to 41 yr, 42 to 54 yr, and 55 yr or older). For each recipient age group, the adjusted 10-yr survivals were plotted on the basis of the donor age category to determine the expected percentage survival at 10 yr.
Results
A total of 50,320 patients underwent first deceased donor renal transplantation between January 1, 1990, and December 31, 1997. Frequency distribution of donor and recipient ages was examined (Figure 1). The donor pool had a bimodal distribution with a mean age of 31 yr. A large peak at 20 yr of age and a second, lower peak at age 45 yr were seen. The range in donor age was from under 1 to 83 yr. In contrast, the recipient distribution showed unimodal distribution with a mean age of 44.4 yr and median age of 45.0 yr. The range in recipient age was from under 1 to 86 yr. The donor-recipient age difference was calculated by subtracting recipient age from donor age for each transplantation. The frequency distribution of kidney transplants based on donor-recipient age difference is shown in Figure 2. Donors were an average of 11 yr younger than recipients, with a wide range—from 78 yr younger to 59 yr older.
Figure 1. Frequency distribution of recipients and donors by age. (A) Recipients (mean age, 44 yr). (B) Donors. Donor age had a bimodal distribution, with the first peak at age 20 yr and the second peak at 45 yr; mean donor age was 33 yr.
Figure 2. Frequency distribution of transplantations by donor-recipient age difference. Donor recipient age difference = (donor age) − (recipient age). Donors averaged 11 yr younger than recipients.
Mean donor and recipient ages by year of transplantation were calculated (Table 1). Over the 7-yr observation period, mean donor age increased by just over 2 yr, and the mean recipient age increased by almost 4 yr. Table 2 and Figure 3 show the distribution of donor organs by recipient and donor age groups. As previously reported by Kasiske and Snyder (6), older recipients received fewer young kidneys, whereas children and young adult recipients received fewer older donors than would be expected on the basis of random allocation by age.
Table 1. Mean donor and recipient age and number of transplants by transplant year
Table 2. Distribution of cadaveric kidneys by donor and recipient age
Figure 3. Distribution of deceased donor kidneys by donor and recipient age. % difference from expected = ([actual no. of transplants] − [expected no. of transplants])/(expected no of transplants × 100).
Cox regression analysis of pretransplantation variables revealed that age of the recipient and a renal diagnosis of diabetes had the largest effect on patient survival. Donor age was the third most important variable with regard to patient survival (Table 3).
Table 3. Risk factors for survival in recipients of first deceased donor renal transplants: 1990–1997
Adjusted patient survivals based on HLA mismatch for the entire population are shown in Figure 4. The 10-yr survival difference between a zero to one mismatch and a six-antigen mismatch is 7% (69% versus 62%), and this was primarily the result of the much poorer patient survival seen in six antigen mismatch kidneys versus all other HLA mismatch groups. The adjusted patient survival based on donor age is shown in Figure 5.
Figure 4. Adjusted patient survival based on HLA mismatch. Survival curves were plotted at the mean of the covariates for the population.
Figure 5. Adjusted patient survival based on donor age group. Survival curves were plotted at the mean of the covariates for the population.
The 10-yr survival difference between recipients of donor kidneys age 0 to 17 and donor kidneys of 65 yr or more was 14% (71% versus 57%). To better understand the affect of graft function on patient survival, death-with-graft function patient survival was determined. Figure 6 shows the adjusted patient survivals by donor age in patients who died with functioning grafts. Donor age remained a significant factor even when excluding graft failure before death, indicating that the effect was not entirely due to the decrement in survival seen with a return to dialysis (P < 0.001, analysis not shown).
Figure 6. Adjusted patient survival with graft function based on donor age. Survival curves were plotted at the mean of the covariates for the population.
In the initial analysis of the effect of donor age on survival, donor age groups less than 42 yr had no affect on patient survival. To better define the breakpoint in survival detriment associated with advanced donor age, adjusted 5- and 10-yr patient survivals were determined in 5-yr increments of donor age between donor ages 21 and 55 yr (Figure 7). This analysis indicated that survival begins to fall between donor age 36 and 40 yr.
Figure 7. Patient survival at 5 and 10 yr based on donor age. Percentage survival was calculated from adjusted curves for each donor age group at the mean of the covariates.
Table 4 shows the 10-yr survivals of deceased donor kidney recipients on the basis of both donor and recipient age groupings. For all recipient age groups, increasing donor age was associated with poorer 10-yr survivals. The decrements in survival between the best and worst outcomes were 8%, 11%, and 13% for the recipients 0 to 40 yr old, 41 to 54 yr old, and 55 yr or older, respectively. The greatest decrement in survival occurred in the oldest recipient group in spite of the fact that 81% of patient deaths occurred with graft function, compared with 65% and 74% for recipient age groups 0 to 40 yr and 41 to 54 yr, respectively.
Table 4. Adjusted 10-yr patient survivals by donor age and recipient age
Discussion
The waiting list for deceased donor renal transplants is growing rapidly while the kidney donor pool is growing slowly, mainly due to the acceptance of older donor kidneys (6). This has put increasing pressure on transplantation programs to accept even more elderly donor kidneys for transplantation. The current UNOS allocation protocol is based on the quality of HLA matching and waiting time, with no provisions for donor-recipient age matching. Our data indicate that donor age is an important predictor of long-term patient survival after renal transplantation, and this effect was seen in all adult age groups, including patients older than 55 yr. This effect is not entirely explained by graft loss and return to dialysis with its higher mortality because donor age remained a significant factor even for recipients who died with functioning grafts. The strong effect of donor age on patient mortality in the older recipient population suggests that poorer graft function interacts with patient morbidity to hasten death.
One limitation of our data is that we cannot rule out a subtle bias on the part of transplantation programs to give more marginal kidneys to patients with more comorbidity and risk of death for their given age, which would accentuate the effect of donor age on survival independent of the effect of donor age per se.
The ethics of transplantation require that we balance medical utility and justice. HLA matching is based on data that graft survival is enhanced by better matching and is a form of maximizing utility. Because the primary goal of end-stage renal disease treatment is to maximize patient survival, patient survival—not graft survival—should be the standard by which the utility of organ allocation is measured. Our data indicate that donor age had a greater effect on patient survival than the HLA matching currently used in the UNOS algorithm. HLA matching is important in immunologically mediated graft failure, but it has no effect on the quality of graft function in the absence of rejection. In contrast, donor age of more than 40 yr affects both graft survival and quality of graft function, and both are important in patient survival after renal transplantation.
Given the demographics of donors and recipients, most recipients should be able to obtain a kidney that is younger than their chronological age, except for patients at the extremes of age, such as children or the elderly. Our data suggest that a policy of favoring younger kidneys for younger adult patients would result in better patient survival in this age group; however, this would result in fewer young kidneys available for older recipients, with worsening survivals in this group. Overall, longer graft survival in young adults may lead to fewer retransplantations, lessening the pressure on the waiting list, which has benefits for all patients by reducing waiting times.
Some selection of donors is already occurring at extremes of age (OPTN data as of August 27, 2002). Although this may make sense from a utilitarian standpoint, a codified system for allocation of organs based on age should be considered because the current system leaves this decision to the discretion of the transplantation team and is not in keeping with the principles of an open and just system of allocation. The current increase in the number of patients on the waiting list is primarily the result of the increased acceptance of older adults (7). Given an increasing incidence of end-stage renal disease in the elderly, this pressure on the donor organs will continue. If the current allocation system continues as is and more advanced age kidneys are accepted for transplantation, as suggested by our data, younger recipients will receive older kidneys as they compete for this scarce resource.
The use of ECD kidneys is one attempt to deal with the limited deceased donor pool and increasing waiting times (8). In patients who agree to accept these kidneys, waiting times for transplantation may be shortened. For some, such as patients 60 yr or older for whom long waiting times may cause de facto exclusion from the benefits of renal transplantation, this approach may make sense. Previous analysis of the use of marginal donor kidneys still showed a survival benefit to renal transplantation over dialysis in elderly patients (9). In this analysis, the definition of marginal kidney was donor age over 55 yr or a 10-yr history of diabetes or hypertension in the donor, cold ischemia time more than 36 h, or a non–heart-beating donor. The effect of donor age on patient survival appears from our data to be a continuous variable after the age of 40 yr and affects survival in all recipient age groups. It is likely at some point that the mortality risk of transplantation will exceed that of patients continuing dialysis if more and more marginal kidneys are used. Also, the experience with marginal kidneys comes from data acquired before the concept of ECD kidneys was introduced. These were kidneys that were accepted for transplantation and may not reflect the outcomes with a more liberal acceptance policy. Recently, 39% of potential ECD kidneys were discarded because they were deemed unacceptable for transplantation (9). The effect on mortality of transplanting previously discarded kidneys is unknown.
The implementation of ECD depends largely on regional waiting times. In regions with short waiting times, recipients may be better off waiting for standard-criteria kidneys because of their superior outcomes. Only in areas with long waiting times would ECD kidneys be beneficial for those patients who are not likely to survive or remain healthy enough to receive a standard-criteria kidney. Although the program was developed to increase the donor pool by using more marginal kidneys that were previously discarded, its indirect, and possibly more important, effect may be providing a means of age matching, because recipients of these kidneys require prior informed consent. Given the poorer graft and patient survivals with ECD kidneys, most young potential recipients may be counseled to opt out of the program, whereas older recipients would likely be counseled to consent to the program. The net effect of this will be to increase the number of older recipients receiving older kidneys.
Traditionally, graft survival has been the primary end point by which medical utility of kidney allocation was judged. Although longevity of graft function is important in patient survival, it is clear from our data that the quality of graft function is also important in patient survival. Because donor age affects both quality and longevity of graft function, its effect on patient survival is greater than that seen with HLA matching. Long-term patient survival should be the primary end point by which we judge the efficacy of our allocation system. As the scarcity of donors relative to recipients increases, it becomes increasingly important to wisely allocate this resource. Limited donor-recipient age matching is already occurring without a clear system, and donor age has a significant effect on recipient survival in all ages. Unfortunately, there is no win-win approach when dealing with organ allocation and donor age. Systems that benefit survival in one age group will affect others negatively. The issue of donor-recipient age matching needs to be debated among the public and transplantation community so that a logical and just system can be developed.
Acknowledgments
The authors of this article have no commercial conflicts of interest. Statistical analyses of data were performed by Dr. Douglas S. Keith with the assistance of Dr. Angelo deMattos and Dr. Jonathan Prather.
- © 2004 American Society of Nephrology