Despite the increase in the number of deceased organ donors in recent years, live kidney donation still remains a major source of kidney transplantations in Australia. Not infrequently, a willing and healthy living donor is deemed incompatible owing to blood group incompatibility or unacceptable donor-specific antibodies (DSAs). Kidney paired donation (KPD) provides a solution to this dilemma by pairing two or more incompatible pairs together to facilitate the exchange of live-donor kidneys between the willing donors (1–4). The success of KPD depends on several factors, such as the number of incompatible pairs in the database, the proportion of blood group–incompatible versus human leukocyte antigen (HLA)-incompatible pairs, the level of sensitization of enrolled patients, and the rules for allocation of KPD donors to recipients. The Netherlands (1, 5) and the United Kingdom (UK) (3) have well-established national KPD programs, whereas in the United States, several regional (6) or single-center (2, 7) programs have established successful programs in the past few years.
In Australia a national KPD program was established in August 2010. The implementation of the Australian Paired Kidney Exchange (AKX) program required that all participating transplant centers agree on donor and recipient medical suitability criteria and the variables and rules used for allocation. We previously reported on the allocation rules agreed for the AKX program (4), which were based on simulations using different algorithms and data from a previous single-center KPD experience (8). Our proposed approach was to ignore any HLA matching rules and to use matching based on acceptable mismatches by excluding donors from matching to recipients with DSAs greater than 2000 mean fluorescence intensity (MFI) for any class 1 or 2 antibodies by Luminex single-antigen bead (SAB). This conservative approach to antibody assignment was chosen to avoid the breakdown of identified chains due to positive crossmatches (9) and to reduce the risk of inferior long-term outcome in the presence of strong DSAs in patients with negative crossmatch at transplantation (10). Depending on the level of sensitization of patients enlisted in a KPD program, our approach could potentially disadvantage sensitized patients by excluding crossmatch-compatible donors. We therefore undertook the task to review the activity of the first 12 months of the AKX program.
A total of 53 donor-recipient pairs were included in the 4 match procedures performed between August 2010 and August 2011. There were 24 couples who participated in 1, 14 in 2, 8 in 3, and 7 in all 4 rounds. Match run conditions are summarized in Table 1. Twenty pairs were included in the first, 25 in the second, 26 in the third, and 33 in the fourth match run. In match runs 2 and 3, an additional altruistic donor each contributed to the donor pool.
Type of Immunological Incompatibility in Enlisted Pairs
Only 10% of recipients were included because of ABO blood group incompatibility without HLA antibody against their coregistered donor. Most recipients (90%) were included because of HLA sensitization; 40% were HLA and ABO incompatible with their coregistered donor, and 50% were HLA incompatible (but ABO compatible) with their coregistered donor. Table 2 summarizes the blood type distribution of donors and recipients; 30% of the donors were blood group O, and there were twice as many blood group O recipients. In the 53 registered recipients, the median calculated panel-reactive antibody (cPRA) for loci -A, -B, -DR, and -DQ was 81% (mean, 71 ± 32). cPRAs of greater than 75% and greater than 90% were found in 58% and 42% of the registered recipients, respectively.
The matching algorithm does not consider any HLA matching rules, and allocation is only based on acceptable mismatches by excluding donors from matching to recipients with DSAs greater than 2000 MFI for any class 1 or 2 antibodies against any of the donor HLA loci (4). The computer program selects between competing match offers on the basis of prespecified ranking rules, such as favoring 3-way over 2-way exchanges, to maximize the number of patients receiving a transplant, and then favoring patients with low versus high match probability, which primarily gives an advantage to a recipient with high versus low PRA. In match run 2, and despite the inclusion of an altruistic donor, no suitable matches were identified using the 2000-MFI cutoff for unacceptable mismatches, and a subsequent run was performed using the 8000-MFI cutoff (Table 1). From match run 3, we agreed with referring centers that ABO-incompatible matching could be considered as an option for selected patients with low blood group–specific titers. After limiting possible exchanges to 2-way and 3-way chains, without limitation of chain length with altruistic donors and eliminating multiple matches, 31 KPD and 2 cadaveric list recipients were matched through the AKX program. Three recipients were matched with multiple donors in subsequent match runs. The match probability by cPRA in the 27 crossmatch-negative matched recipients compared with the 53 registered patients is shown in Figure 1.
Calculated Panel-Reactive Antibody in Listed Versus Matched Recipients
There was an inverse relationship between cPRA and match probability. Of the 37 matched recipients, the matched rate was 86% if the cPRA was less than 25% but was 43% if the cPRA was 90% to 100% (Fig. 1). However, a matching donor in a chain combination could still be identified in a significant proportion of highly sensitized patients. Of the 53 listed recipients, 58% had a cPRA greater than 75%, and in 42% of them the cPRA was greater than 90%. In patients who underwent transplantation, a cPRA greater than 75% was present in 45%, and a cPRA greater than 90% was present in 35% of the recipients (Fig. 2).
Complement-Dependent Cytotoxic Crossmatches
In total, 37 complement-dependent cytotoxic (CDC) crossmatches were performed between matched donors and recipients. All the KPD donors and the altruistic donor re-called for CDC crossmatch testing in the first 2 rounds were also crossmatched against all recipients on the trays. Thus, an additional 112 T-cell crossmatches (TCXMs) and B-cell crossmatches (BCXMs) results were available for analysis, in which 60% of recipients were tested against the donors despite the presence of single or multiple DSAs. The aggregated results of all crossmatches are reported in Table 3.
Crossmatch Results of Matched Recipients
All 27 recipients who were matched using the 2000-MFI cutoff for unacceptable mismatches had a negative CDC TCXM and BCXM. In match run 2, a less stringent exclusion of unacceptable mismatches setting the cutoff at 8000 MFI resulted in 10 patients finding a possible donor. In 7 patients, 1 or more DSAs at strengths ranging from 2254 to 6753 MFI were present; in 4 patients with a low-level single DSA (2254–3373 MFI), both TCXM and BCXM were negative. However, 3 patients with 1 or more DSAs at a strength greater than 6000 MFI had a positive BCXM, but negative TCXM, against their matched donor and were not accepted for transplantation resulting in the breakdown of chains.
Three patients were matched to an ABO-incompatible donor because an initial blood group titer indicated that they were suitable for ABO-incompatible transplantation. However, in 2 patients, additional testing against donor red blood cells revealed an excessively high blood group antibody titer, and it was decided not to progress to transplantation. Thus, after computer matching and CDC crossmatch testing, 27 recipients were booked for live-donor kidney transplantation.
Panel Crossmatch Results
There were 21 positive TCXM with positive BCXM and 32 with positive BCXM and negative TCXM results. All patients with positive TCXM had 1 or more class 1 DSA at greater than 8000 MFI. In two patients, the initial testing suggested a weak or moderate DSA, but the strength of the DSA was found to be greater than 8000 MFI when the Luminex assay was repeated in the immunoglobulin (Ig)M-depleted serum sample. A positive BCXM was found in 47% of patients with DSAs at greater than 8000 MFI and in 38% with DSAs at 2000 to 8000 MFI (Fig. 3A). DSAs against C-locus were responsible for 15% of positive TCXMs, and DSA against DQB1 accounted for 50% of positive BCXMs (Fig. 3B).
Kidney Paired Donation Transplants and Outcomes
In the first year of the AKX program, 20 recipients have undergone successful kidney transplantation. One 3-way chain and two 2-way chains did not progress to transplantation. One offer was refused by the transplant surgeon because of concerns in relation to the vascular anatomy of the donor kidney, 1 because of a donor withdrawing consent, and 1 because of an acute illness of a matched recipient. The breakdown of these chains resulted in 7 patients not undergoing transplantation. Of the patients who underwent transplantation, 18 were KPD recipients, and 2 were transplant wait-list recipients who received the kidney from the last KPD donor in an altruistic donor chain. This represents 34% transplant rate of all registered KPD recipients, 49% of all matched recipients, and 74% of crossmatch-negative matches.
The mean follow-up period was 256 ± 131 days, and 95% of the patients remained rejection-free throughout this period. Fourteen patients underwent a total of 18 kidney biopsies (3 per cause, 15 protocol), and 6 patients did not have a kidney biopsy either per cause or protocol. One patient with delayed graft function not requiring dialysis demonstrated mild acute tubular necrosis on day 3 and showed BK virus nephropathy at 3 months. One recipient with a cPRA of 99% who received a kidney from an ABO-incompatible donor showed abundant C4d deposition and possible antibody-mediated rejection with mild peritubular capillaritis. Records on renal allograft function available to date revealed that serum creatinine levels were 108 ± 27 μmol/L at discharge, 94 ± 22 μmol/L at 3 months (n=20), 96 ± 29 μmol/L at 6 months (n=15), and 88 ± 11 μmol/L at 1 year (n=5).
The first-year experience of the Australian KPD program demonstrates that by using a stringent virtual crossmatch approach, it is possible to achieve match rates of greater than 50% even with a small donor-recipient pool where most recipients are highly sensitized. Thus, using a conservative approach to antibody assignment to avoid positive crossmatches does not seem to significantly disadvantage sensitized recipients.
These results are of particular importance in view of the relatively small number of unsensitized ABO-incompatible recipients enrolled in the program, reducing the ability to easily match donors to multiple recipients. In many instances, the barrier of blood group incompatibility can be easily overcome without substantial changes in the immunosuppressive regime (11, 12), and this has become standard practice in many transplant centers in Australia, where the costs for apheresis for ABO-incompatible transplants are not capped. This has contributed to the small number of ABO-incompatible patients without HLA antibodies being listed in AKX.
The options for patients with high-level HLA sensitization are more limited, and there are still concerns with regard to expense, increased morbidity, and inferior long-term outcomes associated with desensitization techniques (10, 13, 14), although they have been shown to provide a significant survival benefit for HLA-incompatible patients, as compared with waiting for a compatible organ (15). If directed donation is not possible because DSAs are not amenable to desensitization and conventional HLA matching in the cadaveric program results in exceedingly long waiting time, the only hope for highly sensitized recipients could be KPD. With the use of our virtual crossmatch approach to KPD allocation (4), sensitized patients, with the exception of those with a cPRA greater than 90%, do not seem to be overtly disadvantaged in favor of the unsensitized patients. Moreover, that 35% of recipients who underwent transplantation had a cPRA greater than 90% is a remarkable outcome. To perform transplantation on a larger number of these more highly sensitized patients, a less conservative approach to antibody assignment may be required. A restrictive policy to reduce the increased risk for antibody-mediated rejection and the increased long-term graft dysfunction will result in such transplantations being avoided in patients who have no other options. Match run 2 demonstrates that a higher threshold for exclusion from matching in the presence of DSAs of up to 8000 MFI resulted in 3 recipients with negative TCXM and BCXM despite DSAs greater than 2000 MFI undergoing transplantation without evidence of antibody-mediated rejection on protocol biopsies after up to 12 months of follow-up. On the other hand, the only instances of positive BCXM in matched recipients were found with this level of antibody assignment, leading to breakdown of chains. Moreover, the results of the crossmatch testing performed against the panel of unmatched recipients demonstrate that all positive TCXM and 85% of positive BCXM were found in the presence of DSAs at greater than 2000 MFI, which is in line with previously reported data (9). It is worth noting that a significant proportion of recipients had HLA-C or DQ antibodies. This underlines the importance of using SAB assay for better definition of the exact specificities in view of the higher rates of acute rejection episodes especially antibody-mediated acute rejections and graft losses for immunologic reasons among recipients with pretransplant DSAs against HLA-C, -DP, and -DQ (14, 16, 17). We do not routinely perform flow crossmatches because CDC crossmatching and SAB assay are well-established techniques for transplantation organ assignment, and flow crossmatching is a very sensitive technique that could unnecessarily disadvantage highly sensitized recipients with a negative CDC crossmatch and low-level DSAs. However, in some of the patients with DSAs 2000 to 4000 MFI, flow crossmatches were performed and found to be positive and would therefore suggest that our approach is also a good predictor of flow crossmatch results.
The threshold of 2000 MFI to exclude donors from matching to recipients with a DSA seems to be an excellent predictor of negative CDC crossmatch for class 1 antigens. This prediction is dependent on the accuracy of HLA antibody testing. Initially, 1 sample was found to be crossmatch positive despite absence of DSAs greater than 2000 MFI. One possible explanation for this result is the presence of blocking IgM antibodies (18), and, indeed, after IgM depletion, the index serum demonstrated a significantly higher strength of the DSAs. This observation highlights the need to have exactly agreed laboratory standards for correct antibody assignment. An alternative explanation would be the prozone effect (19), where steric hindrance from a high-level HLA antibody could compete for binding by anti-IgG detection antibody on the SAB. However, although we do not routinely perform testing to unmask prozone in our laboratories, this problem seems to be uncommon, because we rarely encounter positive crossmatches without apparent DSA.
An alternative consideration to allow highly sensitized recipients the chance of live-donor kidney transplantation is to leave the decision about whether and how to proceed to the discretion of the transplant center, when low levels of DSA are amenable to early therapeutic intervention. A review of all crossmatch data available indicates that approximately 20% of donor-recipient combinations that were excluded by virtual crossmatching because of DSAs 2000 to 8000 MFI had a negative TCXM and BCXM, and in most instances, there was only 1 or 2 DSAs at a strength of greater than 4000 MFI. The risk of this approach is that identified matches have a greater risk of breakdown, because the initial acceptance by the transplant center may later be withdrawn. The result will be that the other immunologically suitable pairs that were in the same chain will miss out on the potential organ exchange. Because donors have to be re-called to provide fresh samples for crossmatches, the psychological and emotional implications of informing them of a potential match but to be told later that the match had been undone should not be discounted. In our program, there were 7 pairs (two 2-way and one 3-way chains) who were scheduled for exchange surgeries and who did not progress to transplantation. Anecdotal reports by clinicians caring for these patients would indicate significant emotional distress to those pairs who did not progress to transplantation after the initial promise was not fulfilled.
When compared with available data from the first-year activity of other programs in the Netherlands (1), the UK (3), and Texas (7), the transplant rates of the AKX program (34% of enrolled pairs) seem to be excellent. Compared with the cumulative number of enrolled recipients, the transplant rate in the first year was 40% in the Dutch program (1), 9% in the UK (3), and 14% in the Austin (7). It is worth noting that these programs differ significantly in the proportion of recipients included because of pure ABO incompatibility with their coregistered recipients without the presence of significant HLA sensitizations. Recipients who were ABO incompatible with their intended donor were 56% in the Dutch program (1), 44% in the UK (3), and 36% in the Austin (7), compared with only 10% in Australia. On the other hand, in contrast to our program, neither the Dutch nor the UK program included altruistic donors in their first year. The inclusion of altruistic donors is an effective approach to facilitate matching in a KPD program (20). The inclusion of 2 altruistic donors in AKX resulted in a 4-way chain (3 KPD recipients and 1 wait-list recipient) and a 3-way chain (2 KPD recipients and 1 wait-list recipient). If the altruistic donors were removed, only 2 KPD recipients could have been matched in a 2-way chain. Thus, the total number of transplants purely by KPD matching would have been 15, resulting in a transplant rate of 28%.
Early outcomes of the 20 recipients who underwent transplantation were excellent; all patients left the hospital within the center’s usual length of stay without any surgical complications or early rejection episodes, with only 1 case of delayed graft function not requiring dialysis, unrelated to the relatively short cold ischemia of 6 hours of the shipped kidney. At this stage, there are not sufficient data to comment on long-term allograft outcomes. However, the observation that with the currently available follow-up data, a biopsy-proven rejection was found in only 1 recipient, who had a cPRA of 99% and received a kidney from an ABO-incompatible donor, of the 20 patients who underwent transplantation indicates that avoiding moderate and strong DSAs against all HLA loci is a valid strategy to minimize the risk of rejection. The major stumbling block in the AKX program to date has been the failure to progress to transplantation surgery in 25% of crossmatch-negative matched pairs for nonimmunological reasons. Strategies to minimize the rate of nonconversion of matches to transplantations have been devised.
In conclusion, high transplant rates in a KPD program using virtual crossmatching can be achieved even with a small pool consisting largely of highly sensitized recipients, indicating that KPD is a valid and effective solution for patients with immunologically incompatible donors even in the context of highly sensitized recipients.
MATERIALS and methods
The AKX program was established in August 2010 as a new national initiative to enhance live kidney donation. Within the first 12 months, 78 incompatible donor-recipient pairs had been referred to the program. Allocation procedures to match compatible combinations were scheduled every 3 months. Only pairs who satisfied all agreed medical criteria for donor and recipient registration, molecular HLA typing, and HLA alloantibody testing were included in match run procedures. Thus, 53 donor-recipient pairs were included in any of the 4 match procedures performed between October 2010 and August 2011. Crossmatches between the new donors and recipients were performed in the histocompatibility laboratory of the matched donors’ state.
Human Leukocyte Antigen Typing and Detection of Anti–Human Leukocyte Antigen Antibodies
Donor molecular HLA typing and recipient HLA antibody testing for the loci -A, -B, -C, -DRB1, -DPB1, -DQB1, and -DRB3/4/5 were performed at the 5 state histocompatibility laboratories as previously reported (4). Donor HLA typing was performed using direct DNA sequencing. Recipients were tested for HLA class 1 and 2 directed IgG antibody in their sera using SAB Luminex technology (Luminex, One lambda Inc., Conga Park, CA); the MFI signal for each SAB was normalized against the negative control to correct for nonspecific binding.
Calculated Panel-Reactive Antibody
The cPRA of enrolled recipients was based on HLA-A, -B, -DR, and -DQ antigen frequencies derived from the phenotypes of 200 representative participants in the Busselton Health Study, which were found to have similar frequencies for HLA-A, -B, and -DR of donors in the National Organ Matching System (NOMS) registry and the United Network for Organ Sharing (data not shown). The correlation between the cPRA derived using phenotypes from our cohort and the cPRA obtained using the OPTN calculator was R2=0.981. We compared the cPRA distribution for patients who were registered, were matched, and underwent transplantation.
Computer Program Matching
Allocation of suitable live-donor matches was performed using a software module developed by the NOMS as previously described (4). The software takes advantage of advances in HLA technology to enable virtual crossmatching. HLA alleles of each locus and HLA antibody specificities were entered in the computer program at the 4-digit level. Briefly, the algorithm does not consider any HLA matching rules for allocation, and matching is only based on acceptable mismatches by excluding donors from matching to recipients with DSAs greater than 2000 MFI for any class 1 or 2 antibodies against any of the donor HLA loci -A, -B, -C, -DRB1, -DPB1, -DQB1, -DQA1, and -DRB3/4/5 (4). By default, the program considers exclusively ABO-compatible matching, but the option of ABO-incompatible matching is available in selected cases.
Complement-Dependent Cytotoxic Crossmatches of Matched Pairs
After suitable matches in a chain were identified and the peak and historical antibody record was reviewed, TCXM and BCXM between a matched donor and recipient were performed in the histocompatibility laboratory of the matched donor’s state using fresh donor cells in duplicate. If all links within a chain had negative CDC crossmatches, donation procedures in the centers were arranged within 3 months of allocation and performed simultaneously. Sera from all recipients were distributed 1 week before match runs 1 and 2, but from match run 3 onward, only sera from matched recipients were sent to the laboratory of the matched donor. Donors re-called for CDC crossmatch in runs 1 and 2 were crossmatched against all recipients on the trays to test the accuracy of the virtual crossmatch in predicting a negative crossmatch in relation to SAB antibody strength.
The authors thank the collaborators of all participating transplant centers, in particular, those involved in exchange transplant surgeries at Westmead Hospital, Royal Prince Alfred Hospital, John Hunter Hospital, Prince of Wales Hospital, Royal North Shore Hospital, and The Children’s Hospital at Westmead, in New South Wales; Monash Medical Centre and Royal Melbourne Hospital, in Victoria; and Sir Charles Gairdner Hospital and Royal Perth Hospital, in Western Australia.
The authors also thank Mrs Jenni Wright and the NOMS team for their support throughout the program and Dr Scott Campbell, chair of the National Renal Transplant Advisory Committee, for his advice and support.
Finally, the authors thank the patients and their donors for agreeing to participate in this pioneering study.
1. de Klerk M, Keizer KM, Claas FH, et al.. The Dutch national living donor kidney exchange program. Am J Transplant
2005; 5: 2302.
2. Montgomery RA, Zachary AA, Ratner LE, et al.. Clinical results from transplanting incompatible live kidney donor/recipient pairs using kidney paired donation
2005; 294: 1655.
3. Johnson RJ, Allen JE, Fuggle SV, et al.. Early experience of paired living kidney donation in the United kingdom. Transplantation
2008; 86: 1672.
4. Ferrari P, Fidler S, Wright J, et al.. Virtual crossmatch approach to maximize matching
in paired kidney donation. Am J Transplant
2011; 11: 272.
5. de Klerk M, Witvliet MD, Haase-Kromwijk BJ, et al.. Hurdles, barriers, and successes of a national living donor kidney exchange program. Transplantation
2008; 86: 1749.
6. Hanto RL, Reitsma W, Delmonico FL. The development of a successful multiregional kidney paired donation
2008; 86: 1744.
7. Bingaman AW, Wright FH, Murphey CL. Kidney paired donation
in live-donor kidney transplantation. N Engl J Med
2010; 363: 1091.
8. Ferrari P, Woodroffe C, Christiansen FT. Paired kidney donations to expand the living donor pool: the Western Australian experience. Med J Aust
2009; 190: 700.
9. Morris GP, Phelan DL, Jendrisak MD, et al.. Virtual crossmatch by identification of donor-specific anti-human leukocyte antigen antibodies by solid-phase immunoassay: a 30-month analysis in living donor kidney transplantation. Hum Immunol
2010; 71: 268.
10. Lefaucheur C, Loupy A, Hill GS, et al.. Preexisting donor-specific HLA antibodies predict outcome in kidney transplantation. J Am Soc Nephrol
2010; 21: 1398.
11. Flint SM, Walker RG, Hogan C, et al.. Successful ABO-incompatible kidney transplantation with antibody removal and standard immunosuppression. Am J Transplantat
2011; 11: 1016.
12. Montgomery JR, Berger JC, Warren DS, et al.. Outcomes of ABO-incompatible kidney transplantation in the United States. Transplantation
2012; 93: 603.
13. Gloor J, Stegall MD. Sensitized renal transplant recipients: current protocols and future directions. Nat Rev Nephrol
2010; 6: 297.
14. Ntokou IS, Iniotaki AG, Kontou EN, et al.. Long-term follow up for anti-HLA donor specific antibodies postrenal transplantation: high immunogenicity of HLA class II graft molecules. Transpl Int
2011; 24: 1084.
15. Montgomery RA, Lonze BE, King KE, et al.. Desensitization in HLA-incompatible kidney recipients and survival. N Engl J Med
2011; 365: 318.
16. Gilbert M, Paul S, Perrat G, et al.. Impact of pretransplant human leukocyte antigen–C and –DP antibodies on kidney graft outcome. Transpl Proc
2011; 43: 3412.
17. Tran TH, Dohler B, Heinold A, et al.. Deleterious impact of mismatching for human leukocyte antigen–C in presensitized recipients of kidney transplants. Transplantation
2011; 92: 419.
18. Kosmoliaptsis V, Bradley JA, Peacock S, et al.. Detection of immunoglobulin G human leukocyte antigen–specific alloantibodies in renal transplant patients using single-antigen-beads is compromised by the presence of immunoglobulin M human leukocyte antigen–specific alloantibodies. Transplantation
2009; 87: 813.
19. Schnaidt M, Weinstock C, Jurisic M, et al.. HLA antibody specification using single-antigen beads—a technical solution for the prozone effect. Transplantation
2011; 92: 510.
20. Roodnat JI, Zuidema W, van de Wetering J, et al.. Altruistic donor triggered domino-paired kidney donation for unsuccessful couples from the kidney-exchange program. Am J Transplant
2010; 10: 821.