|
|
| |||||||||||||||||||||||||||||||
Background The benefits of HLA-A, B, and DR matching of cadaveric kidney grafts and recipients remain controversial when viewed from the perspective of social equity and graft survival.
Methods We estimated graft survival using proportional-hazards techniques, adjusting for patient and donor characteristics, for a series of 30,564 Medicare patients receiving a first cadaveric kidney transplant between 1984 and 1990. The effects of minimal achievable HLA mismatches and maximal matching on graft survival were estimated by simulated allocation of a sample of organs to a sample of 20,000 candidates for transplantation.
Results The adjusted one-year graft survival was 84.3 percent for grafts with no mismatches and 77.0 percent for grafts with four mismatches. National rationing of donor organs to achieve minimal mismatching and maximal matching could potentially decrease the average number of HLA mismatches from 3.6 to 1.2, with a corresponding increase in the number of matches. As a consequence, projected five-year graft survival could potentially increase from 58.5 percent to 62.9 percent. This would be associated with a decrease in the proportion of kidneys allocated to black recipients from 22.2 to 15.0 percent.
Conclusions Under ideal circumstances, a policy of maximal matching of cadaveric renal transplants would increase five-year graft survival by a comparatively small 4.4 percentage points, but the actual benefit is likely to be smaller.
This study examined the experience of 30,564 Medicare-covered kidney-transplant recipients from the United States Renal Data System to estimate the effects of HLA mismatches on graft survival (both short and longer term). Using these survival data and calculations of maximal achievable matching based on a representative national waiting list of 20,000 patients, we estimated the impact of such a policy on systemwide graft survival.
Methods
All data were derived from the United States Renal Data System, which includes data on 93 percent of all renal transplantations performed in the United States11. The accuracy of these data has been assessed and reported11.
We included in the analysis all 30,564 cadaveric transplantations performed from 1984 through 1990 in recipients 10 to 60 years of age who had first undergone dialysis after 1980. Of a total of 42,329 cadaveric transplantations performed during this period, 3076 were excluded because the recipients did not meet the age criterion, 3768 were excluded because they were not first transplantations, and 6566 were excluded because the recipients had begun dialysis before 1980. There was some overlap in these subgroups.
All identified donor antigens at the HLA-A, B, and DR loci that were known to be absent in the recipient were considered to represent mismatches. No attempt was made to account for differences in split antigens12. "Blank" antigens, where one rather than two antigens were identified at the donor HLA-A, B, or DR locus, were assumed to represent homozygosity and were not counted as mismatches. Multivariate (Cox proportional hazards and Bailey-Makeham) and univariate (Kaplan-Meier) methods were used to estimate the association of mismatches and cadaveric graft survival13,14,15,16. The assumption of the proportionality of hazards was verified by visual inspection of the negative log of the Kaplan-Meier estimates of graft survival. The Bailey-Makeham and Cox models produced nearly identical estimates of graft survival based on the number of mismatches, so only the latter is reported here. Covariates, listed in Table 1, were selected on the basis of theory and statistical significance. Survival estimates were adjusted to reflect the average characteristics of the population undergoing transplantation from 1984 to 1990. However, when proposals for matching are presented, the estimates have been adjusted to the HLA-matching characteristics of the 1989 cohort of patients undergoing transplantation.
|
|
A random sample of 6000 cadaveric organs transplanted in 1989, sorted according to the date of transplantation, was assigned to a simulated waiting list of 20,000 patients randomly selected from the group of patients who received a cadaveric graft from 1986 to 1990. The waiting list was kept at a constant 20,000 patients with replacements randomly drawn from a separate list. Donor organs were sequentially allocated only to recipients matched for ABO blood type with the following hierarchy: six HLA-A, B, and DR matches; no HLA mismatches, but fewer than six matches; one HLA mismatch; and so on up to a total of six HLA mismatches. In cases in which more than one patient had the same number of mismatches, the ties were broken by the selection of the recipient with the highest number of matches. In cases in which more than one patient had the same number of mismatches and matches, the recipient was selected randomly. Since patients who had never undergone transplantation were not included in this list of candidates, it is likely that our model overstates the extent of matches achievable between donors and potential recipients. We evaluated the systemwide effect of achieving the maximal numbers of matches possible nationally by comparing the adjusted five-year rate of graft survival for patients undergoing transplantation from 1984 to 1990 (adjusted to reflect the characteristics of the 1989 cohort) with the survival rate estimated to result from reducing the number of mismatches and increasing the organ-preservation time. Actual five-year graft survival was computed for each level of mismatch after adjustment for the average number of matches and preservation time reported for the period from 1984 to 1990. For maximal matching, the estimates were based on the following:
Hmismatches = 1 - [S1984-1990]M x
M +
where Hmismatches is the estimated failure fraction at five years with maximal matching, S1984-1990 is the average survival at five years in the period studied (1984 to 1990), estimated in a proportional-hazards model with control for center effect; M is the proportional-hazards coefficient for the effect of HLA matches on graft failure, estimated with stratification for center effect; pt is the proportional-hazards coefficient for the effect of preservation time on graft failure, estimated with stratification for center effect; M is the change in average matches resulting from maximal matching; and pt is the change in average preservation time resulting from maximal matching.
The overall five-year estimate is the weighted mean survival of the grafts for all mismatches. (Weights were the percentages of patients at each level of mismatch.) The weights used for the period from 1984 to 1990 were the distribution of transplants according to the number of mismatches during this period; the weights for maximal-matching results were the percentages of mismatched grafts calculated by the matching simulation.
A close comparison of graft failure according to the number of mismatches (Table 2) indicates that the failure rates at five years predicted with the maximal-matching method are slightly lower for each level of mismatch than the actual results for the period from 1984 to 1990. There are several competing reasons for these differences. With maximal matching, the average total organ-preservation time was assumed to increase by 2.5 hours, from a mean of 24.4 hours (1984 to 1990) to 26.9 hours, which was the actual average preservation time for nationally shared organs from 1989 to 1990. (Over the entire period from 1984 to 1990, the difference in preservation time between nationally shared and locally used organs was 4.0 hours.) In addition, the average number of HLA matches would increase (except for a small decrease in the number of matches involving zero mismatches) under maximal matching, which would improve predicted graft survival.
|
The distribution of covariates among the 30,564 patients who underwent transplantation and met the criteria for this study is shown in Table 1. The predominance of male and white recipients (relative to their representation among patients who required dialysis) and the preponderance of white donors have been reported previously18. Both the multivariate and univariate estimates, shown in Figure 1, indicate that grafts with fewer HLA mismatches tend to survive longer. The survival of grafts with no mismatches is substantially better than that of grafts with one mismatch, whereas the survival of grafts with one to six mismatches is more homogeneous.
Figure 2 presents two alternative estimates of graft survival according to the number of HLA mismatches: estimates of the relative risk are reported separately for each level of mismatch and as a continuous variable for grafts with one to six mismatches. Given the arbitrary selection of grafts with four HLA mismatches as the reference group, the linear estimate is the more important, although for the grafts with one to six mismatches the results of the two methods are similar. There was a 4 to 6 percent reduction in the rate ratio for each unit decrease in the number of mismatches (P<0.001). However, the rate of graft failure for a graft with no mismatches was one third lower than that for a graft with four mismatches (factor, 0.65; P<0.001). In absolute terms, the adjusted five-year rate of survival (Figure 1) was 65 percent for grafts with no mismatches, 55 percent for grafts with four mismatches, and 52 percent for grafts with six mismatches. With the exception of the estimate for grafts with five mismatches, all estimates were significantly different from the reference group of grafts with four mismatches (P values ranged from <0.001 to 0.029).
|
|
In this study, 2.6 percent of all transplantations involved no mismatches between the donor and recipient; the rate was 1.7 percent for the period from 1984 to 1987 and 3.7 percent for the period from 1988 to 1990. (A policy of mandatory sharing of kidneys matched for six antigens was adopted in October 1987.) With the maximal-matching method, 19.6 percent of all transplantations could theoretically involve no mismatches, representing a fivefold increase over the percentage calculated for 1988 to 1990. The new method could also reduce the average number of mismatches from 3.6 to 1.2, with 97 percent of donor organs being allocated to recipients with two or fewer mismatches.
The five-year rates of graft failure from 1984 to 1990 are shown in Table 2 (weighted average, 41.5 percent). The rate of graft failure theoretically possible with the method of maximal matching is 37.1 percent. Thus, the projected improvement in five-year graft survival would be 4.4 percentage points.
Rationing of organs with the maximal-matching method would shift recipients' demographic characteristics toward those of the donor pool. As shown in Table 3, recipients selected by the maximal-matching method were similar in age and sex to patients who received kidneys in 1989. However, the new method would increase the already high likelihood that white candidates, rather than black candidates or candidates of other races or ethnic groups, would receive cadaveric organs. Given that the donor pool is composed of more whites than nonwhites and that blacks are more likely to have unidentified antigens, it is not surprising that the method would have such an effect.
|
Our results indicate that HLA matching has a statistically significant and clinically important impact on short- and longer-term graft survival, even in the post-cyclosporine era. However, the impact of HLA mismatches is not linear over the entire range of zero to six mismatches, in that progressive increases in the number of mismatches from one to six have only a small effect on survival as compared with the large benefits afforded by the use of a graft with no mismatches.
Even before the United Network for Organ Sharing mandated national sharing of organs matched for six antigens in October 1987, there was considerable debate about whether the potentially detrimental effect of increased organ-preservation time might cancel out any benefit of matching. Several recent studies had discounted the possibility of a negative effect of the increased organ-preservation times necessary for national organ sharing19,20. Many of these studies, however, involved much smaller samples than those used in our analysis. Furthermore, the claim that survival of well-matched kidneys with longer preservation times is better than that of poorer-matched kidneys with shorter preservation times, although true, is of limited importance when viewed from a systemwide standpoint, as discussed below for the maximal-matching method, which if implemented might involve the shipment of organs of all levels of matching. This would be in contrast to the current policy of shipping only organs matched for six HLA antigens.
Some recent large studies have found that graft failure is most strongly associated with long (>24 hours) organ-preservation times21,22,23. In contrast, we found that organ-preservation time has linear effects on both short- and longer-term graft survival. The 8 percent higher risk of graft failure associated with an increase of 12 hours in the organ-preservation time (relative risk, 1.08; P<0.001) (Figure 3) translates to a difference of 2.4 percentage points in the absolute five-year rate of graft survival (56.1 percent vs. 58.5 percent for procedures performed from 1984 to 1990).
Although it is difficult to predict the effect of maximal matching on organ-preservation times, the estimated increase of 2.5 hours is probably low (see the Methods section). If an increase of more than 2.5 hours is required for national matching of all organs, the benefits from maximal matching will be reduced from the maximal estimate of 4.4 percentage points. These results clearly suggest the benefits of shortening organ-storage times. To achieve this, tissue typing, for example, might be performed before an organ is retrieved.
According to our simulation, a mismatch of no HLA-A, B, or DR antigens could be achieved with cadaveric grafts in 19.6 percent of cases, as compared with the rate of 3.7 percent actually observed since 1987, the year that organs matched for six antigens were required to be shared24. It should be stressed that these estimates are the theoretical limits of what could be achieved with an idealized maximal-matching system.
Evaluation of the systemwide effect of maximal matching on graft survival provides results that are much less dramatic than those obtained for grafts with no mismatches. Although the benefit of a graft with no mismatches is unequivocal, the systemwide increase in average graft survival is at most 4.4 percentage points five years after transplantation. Furthermore, in an analysis of the data from this project for the period from 1988 to 1990, only 50 percent of the grafts matched for six antigens under the maximal-matching system were actually transplanted into the identified recipient (40 percent into patients undergoing a first transplantation) (data not shown). The systemwide benefit of 4.4 percentage points reported in this analysis would be reduced to 2.2 percentage points if the past experience of the placement of kidneys is a guide to how maximal matching might be implemented. Since further splits of HLA loci are being identified, we would expect that the probability of finding no antigen mismatches will be reduced. This negative effect may be counterbalanced by a better outcome of more closely matched donor-recipient pairs.
Given that the donor pool is primarily white, the maximal-matching proposal would, not surprisingly, allocate fewer organs to nonwhite recipients. In 1989, 26.2 percent of transplant recipients were nonwhite; 22.2 percent were black. Under the maximal-matching system, the percentage of nonwhite patients receiving transplants would decrease by one third to 17.7 percent, and the percentage of blacks receiving kidneys would decrease to 15.0 percent. Although this finding is not surprising given the predominance of white donors, it would further impede the access of blacks and other nonwhites to transplantation22.
In summary, the improvement in five-year graft survival resulting from a maximal-matching program is likely to be in the range of 2 percentage points. Whether this benefit is worth the economic and social costs, including the rationing of fewer donor organs to black recipients, is an open question. The reported adverse effects of ischemic injury, which were found to be continuous and linear over the entire range of observed times, suggest potential advantages to shortening organ-storage times.
Supported by a contract (NO1-DK-8-2234) with the National Institute of Diabetes and Digestive and Kidney Diseases, a grant (17-C-90255/5-01) from the Health Care Financing Administration, and the Urban Institute, Washington, D.C.
Source Information
From the Department of Internal Medicine, School of Medicine (P.J.H., F.K.P.), and the Departments of Health Services Management and Policy (P.J.H.), Epidemiology (F.K.P.), and Biostatistics (R.A.W.), School of Public Health, University of Michigan, Ann Arbor; the Division of Immunology and Organ Transplantation, School of Medicine, University of Texas, Houston (B.D.K.); the Department of Internal Medicine, University of Iowa Hospital and Clinics, University of Iowa, Iowa City (L.G.H.); the Health Policy Center, Urban Institute, Washington, D.C. (D.L., J.R.G.); the National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, Md. (L.Y.C.A.); the Woodrow Wilson School, Princeton University, Princeton, N.J. (D.S.G.); and the Department of Preventive Medicine and Biometrics Division, Uniformed Services University School of Medicine, Bethesda, Md. (H.K.). A preliminary presentation of this paper was delivered at the annual meeting of the American Society of Transplant Surgeons, Chicago, June 1, 1990.
Address reprint requests to Dr. Held at the University of Michigan, 315 W. Huron St., Suite 240, Ann Arbor, MI 48103.
References
| |||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
