In the United Kingdom, as in many other countries, kidneys from deceased heart-beating (DHB; brain-dead) donors have for many years been allocated according to a national allocation scheme. The UK allocation scheme in place for most of the last decade (1998 [National Kidney Allocation Scheme] NKAS) was introduced in 1998 and was based on an analysis of factors affecting graft outcome reported in 1999 (1). A major aim of the scheme was to improve human leukocyte antigen (HLA) matching, and although it was successful in achieving its primary objective, inequity of access to kidney transplantation persisted and even increased under the scheme (2, 3).
In 2004, after a review of the 1998 NKAS, it was agreed that a new allocation scheme was required for kidneys from DHB donors. A Task Force (membership in Appendix) was established to define requirements for allocation and to propose an algorithm for a new scheme (2006 NKAS). The proposals were developed on the basis of a comprehensive analysis of factors influencing posttransplant outcomes (4), consideration of various equity factors, a review of the criteria for HLA matching in kidney allocation, and simulation evidence about the effectiveness of alternative schemes.
Here, we describe the objectives of the 2006 NKAS, explore how alternative approaches to HLA matching can lead to greater equity of access to transplant, and present the simulation evidence for the scheme that was implemented and its impact in the 3 years that it has now been in place.
Objectives for a New Kidney Allocation Scheme
The Task Force agreed a number of objectives for a new allocation scheme. The primary objective was to improve the equity of access to transplant among all patients regardless of geographical location, ethnicity, and rareness of HLA type and, so far as biologically possible, blood group and degree of sensitization to HLA specificities.
Other objectives were as follows:
- To reduce the number of patients having to wait an excessive time (>5 years) for transplant.
- To ensure that patient groups requiring well-matched grafts (pediatric and young adult patients and highly sensitized patients) should be allocated suitable kidneys.
- To match graft life expectancy to patient life expectancy (through approximate age matching of donor and recipient) and thus maximize the benefit of kidney transplantation.
- To extend some degree of priority to young adults beyond the threshold of 18 years.
- To avoid prolonged cold ischemia times (CITs) and a significantly increased risk of graft failure (4) that could result from increased exchange of organs between centers in a fully national allocation scheme (under the 1998 NKAS usually only one kidney from each donor was allocated nationally).
HLA and Access to Transplantation
A major disadvantage of HLA matching in kidney allocation is that patients with rare HLA types relative to the donor pool may be difficult to transplant and can have long waiting times. The UK kidney donor pool is predominantly white, whereas more than 20% of patients listed for transplantation are from minority ethnic groups (5), contributing to differences in the frequency of HLA alleles in the respective pools.
The possibility of using allocation based on matching the crossreactive groups (CREGs) of HLA antigens was explored as a means of increasing access to transplantation for those patients who are difficult to match. This approach has been used in the United States (6, 7). We demonstrated that when allocation is based on CREG mismatching, all patients became easier to match, without specifically addressing access to transplant for patients with particularly rare HLA types. There were also concerns that CREG matching could increase the risk of sensitization. Furthermore, analysis of more than 90,000 transplants from the international Collaborative Transplant Study concluded that CREG matching added no benefit to transplant outcomes over matching for HLA antigens (8).
A more focused approach to increasing access to transplantation for patients with rare HLA types was required. HLA matching in the United Kingdom has traditionally been based on broad specificities (1). A possible solution would be recipient antigens that are rare in the donor population could be matched to the nearest related specificity, where one exists, based on serological crossreactions and sequence information.
The gene frequencies of UK donors and transplant list patients were calculated by blood group and ethnicity. Serological and sequence information was considered for the 3 HLA-A, 15 HLA-B, and 4 HLA-DR specificities present in less than 2% of the donor population. Two HLA-B specificities, HLA-B13 and B37, were considered distinct, but each of the remaining specificities could be defaulted to a more frequent related specificity, generally using the serological crossreactions.
The rare and their defaulted specificities are shown in Table 1. In practice, for example, 0.2% of transplant list patients are HLA-A36, and this specificity occurs in 0.05% of UK donors. Mapping A36 to its common counterpart, HLA-A1, allows HLA-A36 patients access to HLA-A1 donors (18% of the donor pool). For an HLA-A36 positive donor, only HLA-A36 and not HLA-A1 positive patients would be considered matched by the national matching algorithm. By using this approach, 27% of all patients on the transplant list could benefit from an increased number of well-matched donors among a pool of 10,000. Forty-one percent of the patients who were most difficult to match could benefit compared with only 8% of those easiest to match. Analysis by ethnic group showed that patients from minority groups would benefit most: 33% of Asian, 59% of Black, and 52% of patients from other ethnic groups, compared with 23% of white patients.
Potential consequences of defaulting rare antigens are increased sensitization after transplant and poorer graft survival. Given earlier evidence that the impact of HLA mismatch on graft survival has diminished (4), the benefit of an increased chance of transplant and of receiving a relatively well-matched kidney outweighed the small increased risks associated with defaulting.
Allocation simulations were performed to inform development of a specific allocation algorithm and were based on the agreed, broad objectives of a new scheme, evidence from outcome analyses, and the concept of defaulting rare HLA specificities.
Simulation Data and Methods
Each simulation applied specific allocation criteria and allocated 8000 kidneys from 4000 donors, by using data for 2000 actual UK DHB kidney donors (2001–2004), each used twice. After two kidneys were allocated from each donor, two new patients were added to the waiting list of 5200 patients (based on the 2001 UK list and representing the size of the list at the time). Patients were added in a set order from a supplementary pool of 4000 new registrations, based on 2000 consecutive patients joining the transplant list (2001–2003), each used twice.
Each simulation generated 8000 transplants, equivalent to approximately 6.5 years of actual activity. The simulations were previously found to be reliable and closely reflect the real situation despite a stable number of patients on the list and the lack of accounting for deaths and removals from the list.
A simulation of the 1998 NKAS provided the basis for comparison of the effect of alternative schemes. Each scheme was evaluated on the basis of three key simulated outputs: the characteristics of patients transplanted, the expected survival rates of the cohort, and the characteristics of the patients remaining on the waiting list at the end of the simulation.
Developing a New Scheme
Two features of the 1998 NKAS scheme made it difficult to modify to improve equity of access to transplantation: first, only approximately 50% of kidneys were allocated nationally, and the rest were allocated according to local transplant center policy; second, the scheme was based on three “tiers” according to HLA mismatch, with prioritization within those tiers (1, 9). Preliminary simulations showed that a new, flexible, and fully national scheme was necessary to achieve the objectives.
Further simulations considered different rules and weightings for factors to be included: blood group match, HLA mismatch grade, recipient age, waiting time, donor-recipient age difference, local allocation, and patient HLA-B and DR homozygosity. A preferred algorithm was agreed (2006 NKAS) on the basis of its simulated effects relative to those of the 1998 scheme, after full discussion and refinements.
The simulated effects of the agreed scheme are demonstrated by factor-specific transplant rates relative to the previous scheme. The appropriateness of these transplant rates is evaluated by consideration of the characteristics of patients accumulated on the transplant list and of those new patients joining the list (Table 2) as described later.
2006 National Kidney Allocation Scheme
We describe the allocation scheme that was agreed (2006 NKAS) and the simulation evidence on which the decisions were based (Fig. 1).
It was agreed that absolute priority should be given to patients with a 000 HLA-A, -B, and -DR mismatch but that within that group, pediatric patients (<18 years) should receive absolute priority over adults. Because of difficulties in transplanting highly sensitized and HLA-DR homozygous patients, they were also prioritized based on simulation evidence. This indicated four tiers of 000 mismatched patients (A–D) and a fifth tier (E) comprising all other patients. Simulation results also suggested that well-matched pediatric patients (100, 010, and 110 HLA-A, -B, and -DR mismatches) should be included in the fourth tier (D) rather than in tier E, where long adult waiting times would too easily have disadvantaged them.
Priority Within Tiers: Points Score
Waiting Time Points
Although 000-mismatched pediatric patients (tiers A–B) are prioritized solely on waiting time, patients in the remaining three tiers are prioritized according to a points score. Excessive waiting times for transplant are undesirable both clinically and from the patient's perspective. Waiting time was, therefore, the most influential of the points scoring factors and the one against which other points scores were scaled.
Simulation results (Table 2) indicated that this emphasis on waiting time would increase transplants for long-waiting patients. Thirty-three percent of patients on the transplant list had waited more than 3 years, and such patients represented 36% of transplants in the simulation of the 1998 NKAS increasing to 55% in the agreed scheme.
HLA Mismatch and Age Points
Points for HLA mismatch and age were combined as shown in Figure 1 to ensure that, in general, younger patients receive well-matched grafts to minimize sensitization and facilitate retransplantation and that younger patients have some priority over older patients. This was achieved based on four levels of HLA mismatch identified in analyzes of posttransplant outcome (4). This showed significantly inferior outcome for level 4 mismatch (MM) transplants ([2 B+1 DR MM] or [2 DR MM]) compared with level 1 (000 MM), with intermediate levels (level 2: [0 DR+0/1 B MM]; level 3: [0 DR+2 B MM] or [1 DR+0/1 B MM]), having statistically comparable outcomes to level 1. Level 4 mismatched transplants were, therefore, not allocated through the NKAS. For the remaining levels, points scores were based on equations designed to prioritize younger patients for well-matched kidneys, whereas scores for older patients were less influenced by HLA mismatch.
Simulation results indicated that this approach would lead to a small increase in 000-mismatched transplants, no level 4 mismatched grafts and a compromise in moderately well matched grafts with the proportion of level 2 MM decreasing from 60% to 41% and level 3 increasing from 4% to 40%. This would allow more flexibility to transplant long-waiting and other priority patients.
The HLA mismatch distribution by age group for patients transplanted in the simulation of the 2006 NKAS showed that the proportion of well-matched transplants decreased with increasing patient age (Fig. 2). In particular, 97% of pediatric patients received level 1 or 2 mismatched simulated transplants compared with 41% of patients aged 60 to 69 years.
Simulations also indicated that age-mismatch points would slightly increase the proportion of kidneys transplanted into young adults, from 30% to 33%, consistent with the overall objectives.
Donor-Recipient Age Difference Points
To minimize the large differences between donor and recipient ages, a quadratic equation is used for these points which are subtracted from the total points score. This is only influential with large differences in age. Furthermore, pediatric recipients are not considered for donors older than 50 years because of inferior transplant outcomes (4).
The simulation of the 2006 NKAS showed that there were fewer transplants where the age difference between donor and recipient is more than 25 years (21%–16%).
To avoid excessive CIT, points are allocated to patients who are geographically close to the donor, based on preferential allocation within one of three defined areas of the UK (10). Analysis of CITs (4) showed that transporting a kidney to a center in another area added an average of 2 hr to the CIT.
Simulation results showed a small increase in the proportion of kidneys retained by the most local transplant center and fewer kidneys exchanged outside of the local area, to minimize the risk of prolonged cold ischemia. This may potentiate geographical inequity, but it is a necessary compromise.
HLA-B and -DR Homozygosity Points
In the 1998 NKAS, the high degree of priority for well-matched patients led to an excess of HLA-DR and, to a lesser extent, HLA-B homozygous patients on the waiting list because of the large number of well-matched patients identified for homozygous donors. To redress the imbalance, priority for HLA-DR homozygous patients was given for 000-mismatched grafts, and points were allocated for HLA-B and -DR homozygosity in tier E.
This resulted in a marked increase in access to transplant for both HLA-B and -DR homozygous patients in the simulation, reversing the increasing number of these patients on the waiting list. In particular, simulated transplants for HLA-DR homozygous patients increased from 13% to 18%.
Blood Group Points
Donor-recipient blood group identity was generally required but compatibility was permitted for high priority patients (tiers A–C). Furthermore, blood group A donor kidneys could be allocated to group AB patients and group O kidneys to group B patients in tiers D and E. However, to achieve equity between group O and B patients, points are deducted from blood group B patients when the donor is group O. This limited access to blood group O kidneys for blood group B patients reflected practice in the 1998 NKAS as demonstrated in the simulation results (Table 2).
Difficult to Match Patients
Rather than allocate points specifically for patients who are difficult to HLA match, defaulting rare HLA specificities to more common, related specificities would increase access to transplant for these patients. Simulations confirmed that the transplant rate would increase for all difficult to match patients, including those from ethnic minorities.
One of the objectives of the scheme was that graft and patient survival should not be compromised. Graft survival (time to failure or death) and patient survival of the simulated transplant cohort were estimated based on a number of donor and recipient characteristics. Estimated graft and patient survival in the simulations of the 1998 and 2006 NKAS were comparable.
Simulations demonstrated that national allocation of all kidneys with immediate effect would result in substantial changes in activity in some transplant centers. This was inevitable given inequity in waiting times across the United Kingdom and the priority for long-waiting patients in the 2006 NKAS. To achieve a smooth transition from the 1998 NKAS, it was agreed that local region patients would be given absolute priority in tier E in the first year, with further phasing in of the scheme until April 2009, when the 2006 NKAS was fully implemented.
Three Year Results of the 2006 NKAS
The scheme was rigorously reviewed after implementation, and at 6 months, certain benefits of the scheme were already apparent. The key results from the first 3 years are described and compared with the last calendar year of the 1998 NKAS (Table 3).
The biggest impact of the 2006 NKAS was the increased transplant rate for long waiting patients and the resultant effect on the transplant list. Twenty-eight percent of transplants were in patients listed for at least 5 years, compared with 10% of transplants in 2005. At the time the 2006 NKAS was introduced, 16% of patients on the list had been waiting at least 5 years, and this reduced to 8% 3 years later.
HLA mismatch grades significantly changed as a result of revised priorities within the 2006 NKAS as predicted by the simulations. In particular, the proportion of well-matched grafts varied according to patient age as intended, with younger patients receiving very well-matched transplants (Fig. 2b).
Defaulting of rare antigens to more common counterparts has affected 12% of transplants, increasing the transplant rate for difficult to match patients from 17% to 23%. This has also benefited ethnic minority patients, but this has been limited by a simultaneous increase in the number of minority patients joining the list; 23% of new patients listed compared with 17% previously.
The aim of eliminating the excess of HLA-DR homozygous patients on the transplant list has been achieved within the first 3 years. The transplant rate has increased as predicted, and these patients now make up 14.5% of the list, in line with the proportion of HLA-DR homozygous patients joining the list (14.2%).
With respect to recipient age, the desired increase in transplant rate for young adult patients has been achieved. The impact is most obvious when looking at the transplant list: 26% of the list at the start of the scheme were aged 18 to 40 years, and this has fallen to 19% after 3 years. Donor-recipient age differences were similar to those for the previous scheme, although it is noticeable that age matching is closer in practice (11% of transplants where age difference >25 years) than it is in simulations (16%), suggesting that further selection may be happening in the offering and accepting of organs.
There were fewer transplants for blood group O and B patients, but this was due to a decrease in the donor rate from these groups.
Exchange of kidneys between transplant centers has increased from 62% to 69%, but the rate of long distance exchanges has remained unchanged (26%), and thus no increase in CIT has resulted from the introduction of the scheme. In fact, median CIT has decreased by 1 hr compared with previously (P<0.0001), undoubtedly due to factors outside the allocation scheme.
There were concerns that a number of factors associated with the 2006 NKAS had the potential to adversely impact on graft and patient survival rates, although the simulation predictions, based on crude risk-adjusted models, did not suggest this would be the case. However, 1-year death-censored graft survival (93%, 95% confidence interval 91%–95%) and patient survival (97%, 95% confidence interval 94%–98%) were not significantly different compared with results for transplants in 2005 (P=0.8, P=0.2, respectively).
One of the major inequities in access to transplant was in relation to transplant center. It is still too early to be able to determine median waiting times to transplant, but as a result of the 2006 NKAS, median times on the list at specified points in time are becoming more similar for patients across the UK transplant centers. In 2005, median days listed ranged across centers from 374 to 981 days, compared with a range of 418 to 747 days in 2009. Although this is a clear improvement, because it was necessary to phase in the 2006 NKAS, geographical equity will take a number of years to achieve.
The 2006 NKAS for kidneys from DHB donors superseded the 1998 NKAS that was utility based and successful in achieving high levels of HLA matching among all transplants. However, the emphasis on HLA matching resulted in inequity of access to transplantation according to a number of factors such as frequency of HLA type, ethnic origin, and transplant center.
The objectives for the 2006 NKAS reflected a compromise between two conflicting approaches to kidney allocation: “utility” (to maximize outcomes) and “equity” (to ensure fair access to transplant). The objectives of the new scheme that related to utility included optimal HLA matching for patients for whom it is most relevant, increased priority for young adult patients, minimization of CITs and matching graft and patient life expectancies through age matching. Objectives concerning equity related to reductions in the waiting time for patients who wait longest, and, in general, greater equity of access to transplantation for all patients (regardless of ethnicity, blood group, HLA homozygosity, and geographical location).
Encompassing results from the analysis of factors influencing transplant outcome and the re-evaluation of HLA matching for allocation (which involved the defaulting of rare specificities to more common, related specificities), multiple simulations, and refinements were carried out and discussed in a series of meetings. Simulations provide objective evidence of the impacts of new and revised allocation schemes relative to one another, making it possible to see what compromises are required between competing objectives to identify a scheme that represents a suitable balance. Comparing across simulations rather than with what has happened in practice means that the somewhat simplified nature of the simulation models used here (e.g., constant waiting list size) did not detract from their value. The simulations do not provide reliable predictions, but they do allow meaningful comparisons between numerous options for a new allocation scheme. The role of simulations in reaching agreement on new organ allocation schemes is well established (11) and has been particularly well evidenced in respect of the French experience (12). An important consideration is that schemes are designed to be flexible so that evolving priorities for organ allocation can be incorporated as simply as possible. A points-based allocation algorithm such as that of the 2006 NKAS is ideal, because it offers total flexibility enabling the algorithm to be modified easily as priorities change. The importance of this flexibility has been noted by others (13).
The 2006 NKAS incorporates a more flexible approach to HLA matching than the 1998 NKAS to allow a greater degree of equity in relation to other factors such as patient location, ethnicity, and HLA homozygosity. The importance of well-matched grafts was recognized for pediatric patients, and a novel form of scoring was introduced linking HLA match and age to ensure that young patients receive appropriate grafts, whereas older patients, for whom sensitization is of less concern due to reduced need for retransplantation, are not made to wait for well-matched grafts. Results of monitoring the impact of the 2006 NKAS in its first 3 years demonstrate that the simulated effects of this approach have been seen in practice. This feature of the scheme demonstrates the flexibility and value of a points-based algorithm and is an approach to HLA matching in kidney allocation that has not been reported by any other regional or national kidney allocation scheme.
After 3 years, the scheme is already making good progress in achieving its objectives, with overall results similar to those observed in the simulations. There has been a significant benefit for patients waiting more than 5 years for transplant with no detrimental effect on one-year graft or patient survival. A number of other advantages of the scheme are also apparent with equity of access improving in many respects, including the achievement of equity of access to transplant for HLA-DR homozygous patients, but geographical inequity of access to transplant will take a number of years to fully address.
Kidney Advisory Group Allocation Task Force Membership
Chair: Dr. Susan Fuggle; members: Dr. David Ansell, Dr. Heather Maxwell, Ms. Paulette Cain, Ms. Dawn McPake, Prof. Peter Diggle, Dr. Gabriel Oniscu, Dr. Chris Dudley, Prof. Steve Powis, Dr. Phil Dyer, Mr. Andrew Ready, Dr. Rob Higgins, Mr. Keith Rigg, Mrs. Rachel Johnson, Dr. Anthony Warrens, Mrs. Helen Lewis, Mr. Chris Watson, and Dr. Phil Mason; NHSBT statisticians: Mr. John O'Neill and Mrs. Samantha Start.
HLA Subgroup Membership
Chair: Dr. Sue Fuggle; members: Dr. Martin Barnardo, Mrs. Linda Shelper, Dr. David Briggs, Dr. Craig Taylor, Prof. Phil Dyer, Dr. Robert Vaughan, Dr. Susan Martin and Mr. Chris Watson.
1.Morris PJ, Johnson RJ, Fuggle SV, et al. Analysis of factors that affect outcome of primary cadaveric renal transplantation in the UK. Lancet
1999; 354: 1147.
2.Rudge CJ, Fuggle SV, Burbidge KM. Geographic disparities in access to organ transplantation in the United Kingdom. Transplantation
2003; 76: 1395.
3.Fuggle SV, Johnson RJ, Bradley JA, et al. Impact of the 1998 UK national allocation scheme for deceased heart beating donor kidneys. Transplantation
2010; 89: 372.
4.Johnson RJ, Fuggle SV, O'Neill J, et al. Factors influencing outcome after deceased heart beating donor kidney transplantation in the UK: An evidence base for a new national kidney allocation policy. Transplantation
2010; 89: 379.
5.Rudge CJ, Johnson RJ, Fuggle SV, et al. Renal transplantation in the United Kingdom for patients from ethnic minorities. Transplantation
2007; 83: 1169.
6.McKenna RM, Takemoto SK. Improving HLA matching for kidney transplantation by use of CREGs. Lancet
2000; 355: 1842.
7.Takemoto SK, Terasaki PI, Gjertson DW, et al. Equitable allocation of HLA-compatible kidneys for local pools and for minorities. N Engl J Med
1994; 331: 760.
8.Laux G, Opelz G. Immunological relevance of CREG matching in cadaver kidney transplantation. Transplantation
2004; 78: 442.
9.Fuggle SV, Johnson RJ, Rudge CJ, et al. Human leukocyte antigen and the allocation of kidneys from cadaver donors in the United Kingdom. Transplantation
2004; 77: 618.
11.Wujciak T, Opelz G. Computer analysis of cadaver kidney allocation procedures. Transplantation
1993; 55: 516.
12.Jacquelinet C, Audry B, Golbreich C, et al. Changing kidney allocation policy in France: The value of simulation. AMIA Annu Symp Proc
2006; 2006: 374.
13.Hiesse C, Pessione F, Houssin D. The case to abandon human leukocyte antigen matching for kidney allocation: Would it be wise to throw out the baby with the bathwater? Transplantation
2004; 77: 623.