Original Article
National Allocation of Cadaveric Kidneys by HLA Matching — Projected Effect on Outcome and Costs
N Engl J Med 1991; 324:1032-1036April 11, 1991
- Abstract
- Abstract
Background.
Although receiving a cadaveric kidney matched at the HLA-A, B, and DR loci enhances graft survival in cyclosporine-treated patients, the value of a national system of kidney allocation based on HLA matching, with the attendant increased likelihood of better matching, is still questioned. Some fear that the costs of a national system are unjustified when only a small fraction of donors would exactly match any of the 16,000 potential recipients anyway. We estimated the effect on graft survival of the use of HLA matching for all allocations of cadaveric kidneys in the United States.
Methods.
The graft-survival rates in five mutually exclusive groups of transplants with increasing numbers of HLA mismatches were estimated by partitioning the data for 22,190 first-time recipients of cadaveric kidneys. Overall graft survival was projected as a weighted average with use of the percentages of transplants in the hierarchical groups in recipient waiting pools of various sizes. We compared the benefits and costs of HLA matching in a national system with those of introducing cyclosporine, which was projected to enhance graft survival by 7 percentage points at 10 years.
Results.
Sharing kidneys nationally on the basis of hierarchical HLA matching was estimated to enhance graft survival by an additional 5 percentage points at 10 years. The anticipated five-year cost of national allocation of kidneys by HLA matching for 7000 recipients, including consideration of the costs of graft removal and dialysis after transplant rejection, would be $6.5 million less than the cost of using cyclosporine alone.
Conclusions.
The use of an HLA allocation system will not add to the cost of renal transplantation, but it will improve the long-term results to the same extent as cyclosporine. We propose the initiation of a national kidneysharing system based on hierarchical levels of HLA matches. (N Engl J Med 1991; 324:1032–6.)
Media in This Article
Figure 1
Actuarial Five-Year Graft-Survival Curves (Solid Lines) and Projections (Dashed Line) for 365 First-Time Recipients of Cadaver Kidneys Matched for Six HLA Antigens and 21,621 Recipients of Transplants with at Least One HLA-Antigen Mismatch.Figure 2
Projections of the Rate of Graft Survival over a 10-Year Period for Specific Categories of Renal Transplants.Article Activity
- Article
A RECENT study showed that kidneys available to more than 100 centers had an extremely high rate of graft survival if the donors and recipients were matched for two antigens at the HLA-A locus, two at the HLA-B locus, and two at the HLA-DR locus (a six-antigen match).1 Through the mutual cooperation of all kidney-transplantation centers in the United States, more than 600 kidneys matched for six antigens have been transplanted during the past three years through a program administered by the United Network for Organ Sharing. As a result of this successful trial and mounting evidence that HLA matching is correlated with the outcome of transplantation of cadaveric kidneys,2 3 4 5 6 we now ask, Should all cadaveric kidneys be allocated on the basis of HLA matching? Experience with small pools of recipients has shown that only with a large national pool can sufficient numbers of well-matched transplants be provided.7 The HLA system is so polymorphic that only a small fraction of cadaveric kidneys would match any of the 16,000 potential recipients exactly.
The important concern today is whether or not the benefit of sharing the available kidneys nationally on the basis of HLA matching would be large enough to justify the necessary effort and expense. Pfaff et al. reported that organ sharing based on HLA matching results in improved overall early graft survival, especially in recipients with high levels of HLA antibodies reactive to cells from randomly selected donors.5 However, Held et al. stated that the use of HLA matching as the basis of national kidney sharing would result in a five-year improvement in graft survival of only 2 to 3 percentage points.8 Since these projections were not based on the use of hierarchical HLA matching, we present such calculations based on data from the UCLA Kidney Transplant Registry. Hierarchical HLA matching was defined as a protocol by which cadaveric kidneys are shared among centers on the basis of a prioritized sequence of explicitly stated degrees of HLA compatibility between the transplant donor and the recipient.
Methods
Life-table analyses were done with data reported to the UCLA Kidney Transplant Registry for 22,190 first-time recipients of cadaveric kidneys who underwent transplantation from 1984 through 1989. Only patients with missing HLA data were excluded. Patients who died were considered to have had graft failure. These data typify transplantation during the cyclosporine era, since over 90 percent of these patients received some form of cyclosporine treatment. As a base line for the comparisons, we used four data sets containing 5751 recipients who underwent transplantation without cyclosporine from 1980 through 1983, 365 recipients of transplants matched for six HLA antigens, 511 recipients of transplants from parental donors,10 and 1359 recipients of transplants from HLA-identical siblings. A simple hierarchical protocol involving five mutually exclusive groups of transplants with increasing numbers of HLA mismatches was chosen for study: those with no HLA-A, B, or DR mismatches; those with no HLA-B or DR mismatches but one or more HLA-A mismatches; those with no HLA-A or B mismatches but one or more HLA-DR mismatches; those with no HLA-DR mismatches but one or more HLA-B mismatches; and those with other types of mismatches. Several categories are collections of incrementally mismatched grades. For example, the "other" category is composed of transplants with no HLA-A mismatches and one mismatch each at the HLA-B and DR loci, transplants with one mismatch at all three loci, and transplants with two mismatches at each locus. The hierarchy was constructed by collapsing selected incremental grades because some categories had few transplants and because graft survival was relatively constant across subcategories. In addition, these groupings correspond to priority matching categories for which the percentages of transplants for recipient waiting pools of various sizes are known; for example, the study by Mickey et al.7 contained several hierarchical protocols, all of which, when similarly analyzed, yielded results nearly identical to those reported below. Survival probabilities at one year (S[1]) and half-lives beyond one year, assuming constant risk for specific categories of first-time recipients of cadaveric renal transplants, were estimated with cases from all higher-order categories excluded (Table 1Table 1
Graft-Survival Rates and Graft Half-Lives for Specific Categories of First-Time Recipients of Cadaveric Kidneys and Current and Estimated Percentages of Transplants Organized According to the Hierarchical HLA-Matching Protocol.*). On the basis of data from previous studies,10 11 12 graft survival (S) beyond one year was projected for each group by the formula S(y) = S(1) · (1/2)[(y—1)/HL], where y is the number of years after transplantation and HL is the half-life. Also shown in Table 1 are the percentages of transplants with the various degrees of HLA compatibility to be used as weights in calculating overall survival. For a pool of 10,000 patients awaiting transplantation (i.e., an effective pool size of 3000, with an allowance made for ABO blood-group identity and patient availability), weights were taken directly from Mickey et al.7 The percentages of the first four hierarchical groups for patient pools of 20,000 and 50,000 were obtained from smoothed estimates (cubic splines) of the data from the study by Mickey et al.7 The relative frequency of the "other" fraction was assigned in such a way that the total equaled 1. Tests of the significance of differences between groups were done by comparing differences in proportions with the appropriate standard error and assuming asymptotic normality. All reported P values are two-tailed. Finally, a cost analysis was performed that assumed a discount rate of 8 percent in present-value calculations, according to financial functions in the Lotus 123, release 2.2, computer program.Results
The one-year rates of graft survival and graft half-lives in the various transplant categories are shown in Table 1. Only two-year survival results based on 50 transplantations were available for the grafts matched for six antigens that were transplanted after 1987.9 Figure 1Figure 1
Actuarial Five-Year Graft-Survival Curves (Solid Lines) and Projections (Dashed Line) for 365 First-Time Recipients of Cadaver Kidneys Matched for Six HLA Antigens and 21,621 Recipients of Transplants with at Least One HLA-Antigen Mismatch. shows the graft-survival curves (solid lines) for 365 patients whose transplants were matched for six antigens and the 21,621 Registry recipients whose transplants had at least one HLA mismatch (i.e., 22,190 total transplants minus the 569 with no HLA-A, B, or DR mismatches). Also projected in Figure 1 is the five-year survival (dashed line) for transplants matched for six antigens, assuming a constant risk after one year. The results for the transplants with a six-antigen match were significantly (P<0.0001) better from three months to two years, with one- and two-year survival rates of 86.6 percent and 82.1 percent as compared with 76.4 percent and 69.4 percent, respectively, for the recipients whose transplants had mismatches. The long-term result based on projections from the two-year survival point predicts a 22-percentage-point difference in the survival rate at five years.For all other grafts in Table 1, the estimates were based on data points for at least five years. Graft survival and half-lives for first-time recipients of cadaveric kidneys are shown together with the percentages of the transplants with the various degrees of HLA compatibility for the current experience of the Registry and pools of 10,000, 20,000, and 50,000 recipients. These values were used to compute the total effect, with the use of cyclosporine, of random matching and of allocating grafts according to the hierarchical HLA-matching protocol. With better matches, the one-year graft-survival rates are superior to those for transplants with poor mismatch grades, which have a 75 percent rate of graft survival after one year. In addition, the half-lives tend to increase with better matches, although it appears that both major histocompatibility complex Class I antigens need to be matched to produce benefit3 — that is, the half-lives exceeded 9 years in the group with no HLA-A, B, or DR mismatches or with no HLA-A or B mismatches, as compared with 6.9 years in the group with no HLA-B or DR mismatches or with no HLA-DR mismatches. The 1-year rate of graft survival for the transplants matched for six antigens was 86.6 percent, and the half-life was 13.0 years.
The rate of graft survival over a 10-year period is plotted in Figure 2. During the period from 1980 to 1983, before cyclosporine was used, the 1-year graft survival rate was 62 percent, and the 10-year rate was 25 percent. The introduction of cyclosporine therapy resulted in an increase (P<0.0001) in graft survival to 77 percent at one year. However, the advantage of cyclosporine treatment diminished over time, so that after 10 years the difference in survival was only 7 percentage points. This disparity results from the fact that the rate of transplant loss was nearly the same in the cyclosporine and the precyclosporine eras (half-life, 7.2 vs. 6.9 years; P not significant).
When kidneys were allocated to the centers according to a hierarchical HLA-matching system with a patient pool of 20,000, the additional benefit at one year was 5 percentage points (P<0.0001 for the difference in overall graft survival between current percentages of transplants with the various degrees of HLA compatibility and the percentages in a patient pool of 20,000). Since the rate of long-term loss was lower in this group than in randomly matched patients receiving cyclosporine, the projected advantage at 10 years is still 5 percentage points. This means that instead of the current 32 percent graft survival at 10 years, the rate would be 37 percent. Transplants from donors with one different HLA haplotype (parents) have a 53 percent survival at 10 years.
The cadaveric kidneys matched for six antigens had the same rate of loss as transplants from parental donors. Transplants from HLA-identical siblings had the highest survival rate at 1 year and 10 years (94 percent and 68 percent, respectively).
The five-year costs of 7000 transplantations performed without cyclosporine, with cyclosporine, and with the hierarchical HLA-matching protocol are shown in Table 2Table 2
Present-Value Costs for the First Five Years after Transplantation for 7000 First-Time Recipients of Cadaver Kidneys.*, along with the assumptions made for the various component costs. In most instances, the cost figures, taken from recent analyses,13 14 15 are the lowest estimates. Even assuming no inflation, the additional cost of cyclosporine was $52 million over the five-year period. If a hierarchical HLA-matching protocol was added, savings of $6.5 million could be realized. A negative cost would actually result from the greater long-term graft survival and a minimal one-time required expenditure of $1,000 to cover the difference between local and long-distance telephone charges, air-freight charges, and the time involved in long-distance transport arrangements. HLA typing is currently performed in all potential recipients and cadaveric kidneys; therefore, typing does not create additional costs. The discount rate of 8 percent was chosen as an estimate of a reasonably safe, medium-term investment; varying this rate (0 to 40 percent) would change the amount of savings but not the conclusion that savings do occur with sharing.Discussion
From the trials involving over 600 transplants, the cadaveric kidneys matched for six antigens provided a superior outcome despite transplantation at multiple centers. This evidence, plus the previous studies showing the effectiveness of HLA matching,2 3 4 5 6 leaves no doubt that HLA matching will benefit patients fortunate enough to receive a matched transplant. There are two important questions: Can a sufficient number of patients receive well-matched kidneys to affect graft survival significantly, and how costly will it be?
To put the anticipated magnitude of improvement in perspective, we compared the effect of hierarchical HLA matching with the effect of cyclosporine, universally accepted as having a tremendously positive influence on transplantation. The use of cyclosporine resulted in an impressive 15-percentage-point improvement in one-year graft survival. When examined from a 10-year perspective, however, the calculated increase in graft survival was only 7 percentage points (Fig. 2). The long-term effectiveness decreases because the rate of transplant loss with cyclosporine after one year is the same as the rate of loss without cyclosporine (the half-lives are equal, as shown in Table 1). On the other hand, the calculated rate of graft loss with hierarchical HLA matching is lower than the rate of loss with cyclosporine. Thus, the 5-percentage-point improvement in graft survival at one year remains constant for 10 years. At the end of 10 years, therefore, the effect of hierarchical HLA matching nearly equals the magnitude of that produced by the introduction of cyclosporine.
The costs associated with matching and sharing kidneys (e.g., airline freight charges and long-distance telephone calls) were purposely set high in our analysis, whereas costs connected with graft failure (e.g., nephrectomy and dialysis) were set low and did not include factors associated with failure such as hospitalization for episodes of rejection. Despite the fact that our cost projections were biased against sharing, we still found savings when kidneys were allocated according to hierarchical matching.
Some contend that HLA matching would discriminate against blacks. To correct for inequities that might develop in an allocation system based entirely on HLA matching, Starzl et al. have suggested a point system that takes other factors into consideration.16 Patients forced to wait for long periods, presumably because they are difficult to match, receive an allowance in the form of points allocated for waiting time. When kidneys were allocated according to a point system rather than a center-driven system in a local two-year trial, transplantation in patients with longer waiting times and those with high levels of HLA antibodies was performed more frequently through the point system than the center-driven system.17 Therefore, a change to a national system will not suddenly decrease the number of black recipients undergoing transplantation; rather, it may increase it. Increased organ donations from the black community should be a major objective for the long term, since it is the only way that black recipients can obtain well-matched transplants and avoid the consequences of rejection, which include hospitalization, increased immunosuppression, and possible graft loss followed by a return to dialysis.15
Whether or not survival of the graft in every patient improves with hierarchical matching remains in question. If matching were the sole criterion for organ allocation, patients should expect to receive kidneys matched as well as or better than those they would receive under the present system, given an infinite waiting period. Under a practical system that assigns points for factors such as waiting time, some patients would expect to receive kidneys that were less well matched even though the aggregate survival would improve.
In conclusion, an allocation system based on hierarchical HLA matching would (by our calculations) result in higher overall survival of renal grafts without increasing costs. As a possible smooth transition from the established system to a hierarchical matching protocol, we propose that at a minimum, cadaveric kidneys with no HLA-antigen mismatches or only one be shared nationally, as are grafts matched for six HLA antigens.
Supported in part by a grant (02375) from the National Institute of Diabetes and Digestive and Kidney Diseases.
We are indebted to Ms. Lynne Gjertson, OP.A., for her assistance in cost analysis and to Ms. Beverly Looman and Ms. Jane Schoenberg for their assistance in the preparation of the manuscript.
Source Information
From the UCLA Tissue Typing Laboratory, 950 Veteran Ave., Los Angeles, CA 90024, where reprint requests should be addressed to Dr. Terasaki.
- References
References
1
Takemoto S, Terasaki PI. A report on 341 six-antigen matched transplants. In: Terasaki PI. ed. Clinical transplants 1989. Los Angeles: UCLA Tissue Typing Laboratory, 1989:463–70.
2
Opelz
Full Text | Web of Science | Medline3
Gjertson DW. Short- and long-term effects of HLA matching. In: Terasaki PI, ed. Clinical transplants 1989. Los Angeles: UCLA Tissue Typing Laboratory, 1989:353–60.
4
Persijn GG, D'Amaro J, De Lange P, et al. The effect of mismatching and sharing of HLA-A, -B, and -DR antigens on kidney graft survival in Eurotransplant 1982 to 1988. In: Terasaki PI, ed. Clinical transplants 1988. Los Angeles: UCLA Tissue Typing Laboratory, 1988:237–48.
5
Pfaff
Web of Science | Medline6
Gilks
CrossRef | Web of Science | Medline7
Mickey
CrossRef | Web of Science | Medline8
Held PJ, Kahan B, Hunsicker LG, et al. The impact of HLA mismatches on kidney graft survival. Presented at the 16th Annual Scientific Meeting of the American Society of Transplant Surgeons, Chicago, May 30—June 1, 1990. abstract.
9
Takemoto
Web of Science | Medline10
Cook DJ. Long-term survival of kidney allografts. In: Terasaki PI, ed. Clinical transplants 1987. Los Angeles: UCLA Tissue Typing Laboratory, 1987:277–85.
11
Takiff
CrossRef | Web of Science | Medline12
Terasaki
Web of Science | Medline13
National Institutes of Health. The national kidney and urologic diseases advisory board 1990 long-range plan: window on the 21st century, Bethesda, Md.: National Institutes of Health, 1990. (NIH publication no. 90–583.)
14
Evans
Web of Science | Medline15
Manninen
Web of Science | Medline16
Starzl
CrossRef | Web of Science | Medline17
McClelland J, Terasaki PI, Berne T, et al. Evaluation of the UNOS point system for kidney recipient selection. In: Terasaki PI, ed. Clinical transplants 1989. Los Angeles: UCLA Tissue Typing Laboratory, 1989:471–6.
- Citing Articles (19)
Citing Articles
1
Cathi L Murphey, Thomas G Forsthuber. (2008) Trends in HLA antibody screening and identification and their role in transplantation. Expert Review of Clinical Immunology 4:3, 391-399
CrossRef2
Selda Emre Aydingoz, Steven K. Takemoto, Brett W. Pinsky, Paolo R. Salvalaggio, Krista L. Lentine, Lisa Willoughby, Beverly Hoover, Thomas A. Burroughs, Mark A. Schnitzler, Ralph Graff. (2007) The impact of human leukocyte antigen matching on transplant complications and immunosuppression dosage. Human Immunology 68:6, 491-499
CrossRef3
S. M. Gore. 2005. Transplantation. .
CrossRef4
Cynthia Davis, Gurch Randhawa. (2004) ???Don???t Know Enough about It!???: Awareness and Attitudes Toward Organ Donation and Transplantation among the Black Caribbean and Black African Population in Lambeth, Southwark, and Lewisham, United Kingdom. Transplantation 78:3, 420-425
CrossRef5
Barbara A. Bresnahan, Christopher P. Johnson, Matthew J. McIntosh, Donald Stablein, Sundaram Hariharan. (2002) A Comparison Between Recipients Receiving Matched Kidney and Those Receiving Mismatched Kidney from the Same Cadaver Donor. American Journal of Transplantation 2:4, 366-372
CrossRef6
Dave L Roelen, Inge Stobbe, Neil T Young, Simone P.M.J van Bree, Ilias I.N Doxiadis, M Oudshoorn, Peter J Morris, Kathryn J Wood, Frans H.J Claas. (2001) Permissible and immunogenic HLA-A mismatches: cytotoxic T-cell precursor frequencies reflect graft survival data. Human Immunology 62:7, 661-667
CrossRef7
Stefanos A. Zenios, Glenn M. Chertow, Lawrence M. Wein. (2000) Dynamic Allocation of Kidneys to Candidates on the Transplant Waiting List. Operations Research 48:4, 549-569
CrossRef8
Inge Stobbe, Ellen van der Meer-Prins, Jacqueline MA Smits, Ilias IN Doxiadis, Frans HJ Claas. (1999) In vitro reactivity of allospecific cytotoxic T lymphocytes does not explain the taboo phenomenon. Transplant Immunology 7:4, 215-220
CrossRef9
Schnitzler, Mark A., Hollenbeak, Christopher S., Cohen, David S., Woodward, Robert S., Lowell, Jeffrey A., Singer, Gary G., Tesi, Raymond J., Howard, Todd K., Mohanakumar, T., Brennan, Daniel C., . (1999) The Economic Implications of HLA Matching in Cadaveric Renal Transplantation. New England Journal of Medicine 341:19, 1440-1446
Full Text10
Helderman, J. Harold, Goral, Simin, . (1999) The Allocation of Cadaveric Kidneys. New England Journal of Medicine 341:19, 1468-1469
Full Text11
Masaru Hirata, Paul I. Terasaki, Yong W. Cho. (1996) CYTOMEGALOVIRUS ANTIBODY STATUS AND RENAL TRANSPLANTATION: 1987-1994. Transplantation 62:1, 34-37
CrossRef12
(1995) HLA Matching in Renal Transplantation. New England Journal of Medicine 332:11, 752-753
Full Text13
Takemoto, SteveTerasaki, Paul I.Gjertson, David W.Cecka, J. Michael. (1994) Equitable Allocation of HLA-Compatible Kidneys for Local Pools and for Minorities. New England Journal of Medicine 331:12, 760-764
Full Text14
Suthanthiran, ManikkamStrom, Terry B.. (1994) Renal Transplantation. New England Journal of Medicine 331:6, 365-376
Full Text15
Thomas E. Starzl, Anthony J. Demetris, Massimo Trucco, Noriko Murase, Camillo Ricordi, Suzanne Ildstad, Hector Ramos, Satoru Todo, Andreas Tzakis, John J. Fung, Michael Nalesnik, Adriana Zeevi, William A. Rudert, Mirjana Kocova. (1993) Cell migration and chimerism after whole-organ transplantation: The basis of graft acceptance. Hepatology 17:6, 1127-1152
CrossRef16
P.I. TERASAKI, S. TAKEMOTO, M.S. PARK, B. CLARK. (1992) Landsteiner Award. HLA epitope matching. Transfusion 32:8, 775-786
CrossRef17
Takemoto, Steve, Terasaki, Paul I., Cecka, J. Michael, Cho, Yong W., Gjertson, David W., . (1992) Survival of Nationally Shared, HLA-Matched Kidney Transplants from Cadaveric Donors. New England Journal of Medicine 327:12, 834-839
Full Text18
T.E Starzl, N Murase, S Ildstad, C Ricordi, A.J Demetris, M Trucco. (1992) Cell migration, chimerism, and graft acceptance. The Lancet 339:8809, 1579-1582
CrossRef19
Richard A. Sherman. (1991) Briefly Noted. Seminars in Dialysis 4:4, 292-292
CrossRef
