There are far more eligible kidney transplant candidates worldwide than there are organs available. Kidney exchange, also referred to as paired kidney donation, has become strongly established over the past decade as a modality for facilitating living kidney donation by donors who are not compatible with their intended recipients. Instead of donating directly to the intended recipient, candidate-donor pairs are matched together to participate in a mutually beneficial exchange of organs. This requires a coordinating effort to collect data about incompatible pairs and about other types of participants like nondirected (altruistic) donors and compatible pairs, and to select appropriate matches by a manual or algorithmic process.
In this chapter, we examine the common features and the idiosyncrasies of several successful paired exchange efforts around the world : in Korea, the Netherlands, Canada, Australia, India, the UK, the US, and other countries. Reviewing these differences may be reassuring for groups working to implement paired exchange, because for each potential snag that might have delayed or limited one program’s efforts, there are counterexamples in which that aspect of paired exchange was not an impediment.
Finding Exchange Opportunities
In every kidney exchange program, some incompatible pairs will fail to find an exchange opportunity. Because blood group O donors are most likely to be compatible with their recipients, there will be a blood group imbalance, with too few O donors for too many O recipients in any population of incompatible pairs. Populations of incompatible pairs also have larger numbers of highly sensitized candidates, who are difficult to match in an exchange. This imbalance limits exchange opportunities and leads to innovations like the inclusion of compatible pairs and nondirected donors, extending match arrangements to three or four or more pairs, and desensitization as part of the exchange program, all of which can enhance the proportion of incompatible pairs successfully transplanted.
Paired exchange may have the longest history in Korea, which established an exchange program in 1991. By 2003 exchanges comprised more than 10% of living kidney donation at one Korean center. Korea’s approach emphasized close human leukocyte antigen (HLA) matching between donors and recipients. Living donors were only accepted if they shared more than one DR antigen or two of four A/B antigens with their recipients. Of pairs deemed incompatible, 30% had HLA mismatching beyond these criteria, 65% were blood type incompatible, and only 5% had a positive crossmatch test. Compared with other programs that allow more HLA mismatch, this restriction may have directed more candidate/donor pairs into the exchange group, but may have ruled out more potential exchanges.
Legality, Anonymity, Travel, and Expenses
Legal restrictions on living organ donation have played a significant role in the evolution of kidney exchange in many countries. For a time in the US, people took care to refer to “kidney paired donation” rather than kidney exchange because of concerns about whether the National Organ Transplantation Act’s (NOTA’s) prohibition on giving valuable consideration to living donors encompassed a ban on exchanging one kidney for another. In 2007 the US enacted a new law clarifying that paired exchange had never been illegal under NOTA. Similar restrictions in Australia and Canada have also been a concern. In the UK only relatives and those with strong emotional ties to the candidate were permitted to be living organ donors, until legislation allowing kidney exchange was enacted in 2006. In India, some legal restrictions have been streamlined but paired donors must be first-degree relatives.
The usual practice in the US is to maintain strict anonymity among the exchange participants before surgery, with meetings possible after surgery if desired by all parties. In the Netherlands, patients in a pilot study expressed a strong preference for anonymity, so that country’s exchange program maintains anonymity. In Germany, anonymous paired kidney exchange is prohibited, so exchanges can only proceed after incompatible pairs meet one another. In Romania, a single-center program transplanted 56 pairs between 2001 and 2005. This group felt that maintaining anonymity among pairs involved in a paired donation was unnecessary and would be too logistically difficult, so they encouraged open interaction before and after the transplants.
In the Netherlands, a compact country that is densely populated, donors travel to the transplant center where their paired recipients are being cared for. In the US, which has a much larger geographic footprint, donor organs are routinely transported to the recipient center. US physicians became more comfortable with transporting organs for kidney paired donations after a study evaluating delayed transplants of other live donor kidneys showed that up to 8 hours of cold ischemia time did not have negative repercussions for a live donor kidney. In Canada, transporting live donor kidneys would be a logical response to far-flung paired donation matches, but regulations and the risky logistics of travel there have limited such transports. In Australia, most kidneys are transported despite the challenges of coordinating the transports and of prolonged cold ischemia time.
In the US financial arrangements are frequently a complicating factor in kidney paired donation. There is no obvious party responsible for paying the expenses incurred in donor evaluations for incompatible donors, many of whom never donate, nor for the tissue typing and administrative costs of maintaining a paired donation registry to locate suitable matches. Some have called for extending the concept of the standard acquisition charge, used in the US to recoup expenses related to acquiring deceased donor organs, to kidney paired donation.
Deciding which incompatible pairs, compatible pairs, and nondirected donors should be matched together for a kidney exchange is critically important for any program. Each donor should usually be blood group and tissue type compatible with the candidate to whom they will donate. Characterizing recipient unacceptable antigens and allowing centers to define their own criteria for unacceptability helps decrease the number of proposed matches that fail because of an unexpected positive crossmatch. Some transplant centers have experience with desensitization protocols for ABO-incompatible or HLA-incompatible transplants, or both. These centers may wish to combine desensitization with kidney exchange to obtain transplants for their hardest-to-match patients.
Beyond simply understanding which donors could successfully donate to which candidates in a kidney paired donation registry, matching participants together so that the largest number of people can achieve the greatest benefit requires sophisticated mathematics and specialized optimization algorithms. We refer to a “match” as a single two-way, three-way, or chain of transplants, as in Fig. 23.1 . For a fixed pool of participants in a paired donation registry, “matching” means selecting a nonoverlapping set of matches to proceed to transplant.
The decisions in a paired exchange registry are necessarily interconnected, because choosing one match means that other potentially useful matches involving any of those same participants will not be possible. Some seemingly effective matching heuristics neglect the interconnectedness of these decisions. For instance, it is not optimal to rank all possible matches and then choose the highest ranked match first, the next highest ranked available match second, and so on. This match-rank heuristic, which has been used in some operational registries, might transplant fewer people than would be possible with better matching algorithms.
There are only two correct approaches to find a set of exchanges with the largest number of transplants for two- and three-way exchanges among a specific list of participants: exhaustive search or integer programming (optimization). In exhaustive search, a computer generates every possible set of matches and then compares them to select the best set of matches, as in the Korean registry. Exhaustive search is only possible for small registries, because the number of different sets of matches possible grows incredibly quickly. Integer programming models, including many algorithms specialized for the paired donation problem, use mathematical techniques to locate and verify the best set of matches without having to explicitly describe all possible sets of matches.
However, recent evidence suggests that in actual registries more matches can be made by matching with very high frequency than by improving the match algorithm. The advantage of optimization is only relevant when registries wait for many incompatible pairs to accumulate and perform matching infrequently. Ashlagi et al. find that frequent matching is the best strategy for matching, when pairs might depart after being asked to wait too long.
By far the most significant factor in increasing the fraction of incompatible pairs matched is increasing the number of participants in a registry. All candidates benefit from increasing pool size, but the benefit is particularly pronounced for sensitized candidates. For instance, a nationwide registry in the US would create a sixfold increase in kidney exchange opportunities for sensitized candidates.
The advantage of large-scale programs over small ones has implications for the organization of registries around the world. Canada, the UK, and the Netherlands, each with population in the tens of millions, have established national kidney exchange registries. In South Korea and Turkey, kidney exchange is conducted within single-center programs. One single-center program in the US, by relying on compatible pairs, has grown its exchange program to be more than one-third of its live donor kidney volume. In the US there are many models competing: there are single-center programs, several multicenter consortia, and an effort by the United Network for Organ Sharing (UNOS) to establish a national kidney exchange registry integrated with the national deceased donor allocation system. The various registries in the US serve different but overlapping populations, because many incompatible pairs are registered for more than one of these. Unfortunately, fewer incompatible pairs and many fewer sensitized candidates will find matches if participants are divided into smaller pools than if the participants were in one larger pool.