A Computer Match Program for Paired and Unconventional Kidney Exchanges


*Corresponding author: Inessa Kaplan, ikaplan@jhmi.edu


The number of renal transplants can be increased by implementing an exchange program involving donor-recipient pairs for whom the donors are each incompatible with their original patient but compatible with each other's patient. The number can be further increased if the exchanges are not limited to ABO incompatible pairs or combinations of two donor-recipient pairs. However, as the number of donor-recipient pairs willing to participate in such a program increases, there is a substantial increase in both the time taken to identify such matches and the potential for error. We have developed a computer program that accounts for ABO and HLA compatibility and is not limited to two-way exchanges. With our database of 60 patients and 83 donors, we have been able to identify 122 two-way and 1230 three-way exchanges with an average run time of 30 s.


The growing disparity between the number of patients on the national list waiting for a kidney transplant and the number of deceased donors has compelled transplant programs to seek ways to increase the number of organs available for transplantation. The development of the laparoscopic donor nephrectomy increased the attractiveness of living donation and there has been a steady increase in living donor renal transplantation with the greatest increase occurring among unrelated donors (based on OPTN data as of March 9, 2005). However, ABO and/or HLA incompatibilities can prevent transplantation from willing, healthy donors. While there are several protocols in place to overcome these incompatibilities, they increase risk to the recipient and cost of the transplant and may not be feasible or successful in all cases.

An alternative to treating an incompatibility is to circumvent it through a paired kidney exchange (PKE). The clinical aspects of such a program have been discussed elsewhere (1–5) and will not be addressed here. Initially, exchanges were geared to overcoming ABO incompatibility, as is shown in Figure 1A, and in North America involved two-way exchanges exclusively. Such exchanges are easy to identify manually. In contrast, the HLA system, with its high degree of complexity, poses a greater problem both in achieving compatibility and in identifying matched pairs. An example of an exchange to achieve HLA compatibility is shown in Figure 1B. The appropriate donor will be one who is both ABO compatible and lacks any incompatible HLA antigens, the latter of which may be numerous. Identifying suitable pairs for an exchange in these cases is tedious and subject to error. In some cases, compatibility can only be achieved when the number of donor-recipient pairs involved in the exchange is not limited to two. These types of cases can be extremely time consuming to identify. We have developed a computer program that takes into account ABO and HLA compatibility and allows for exchanges involving more than two donor-recipient pairs. We describe the program here and provide some examples to illustrate its power and the time saving that can be achieved.

Figure 1.

Examples of exchanges to overcome (A) ABO and (B) HLA incompatibilities.


The database includes phenotypic and demographic information about donors and recipients who have consented to being entered into an exchange program. This includes a unique numeric identifier, the HLA-A, -B, -Cw, -DR, -DRw(51, 52, 53) and -DQ antigens of the donor and recipient, the ABO types of the donor and recipient, unacceptable donor antigens and the donor and recipient ages. The information is imported from our existing patient database, which eliminates the possibility of transcription error and can be linked back to the database when a match is found, to obtain contact and physician information. The information can also be imported from a spreadsheet or entered manually. The data that can be included are not limited to those listed here and inclusion of other data for information purposes only might be desirable.


The search begins with an identified patient (Patient 1). The program examines all ABO compatible donors and eliminates anyone in the donor pool who has an antigen identified as unacceptable for patient 1. If no compatible donor is found, the search for that patient is ended. If one or more donors are found (donors: 2–n), the patients (2–n) for these donors are evaluated for compatibility with the donor for patient 1 (donor 1). If any of patients 2–n is a match for donor 1, a compatible two-way paired exchange is defined. If not, the program searches for compatible donors for patients 2–n and if one or more are found, the patients for those donors are evaluated for compatibility with donor 1. This process can be repeated for any number of iterations, at each step increasing the number of donor-recipient pairs involved in the potential exchange. The search can also be performed without consideration of ABO compatibility, which may be useful for patients who are so highly sensitized that it is unrealistic to anticipate finding a donor who is compatible for both ABO and HLA.


At the time of this writing, there were 60 patients and 83 potential donors in our database. The demographics are given in Table 1. The distributions of blood types and race are nearly identical between the donor and recipient groups. Forty-one (49.4%) of the donors were biologically related to their original recipient and among the unrelated donors, 19 of 42 (45.2%) were spouses or in-laws. Forty-six (75.4%) of the patients were sensitized. Of the sensitized patients, 21.7% were sensitized only to class I antigens, 26.0% were sensitized only to class II antigens and 52.3% were sensitized to both class I and class II antigens.

Table 1.  Demographics of patients and potential donors



  1. aNumbers in brackets are percentages.

GenderMale32 [52.4]36 [43.4]
Female29 [47.6]47 [56.6]
Blood typeA19 [31.1]26 [31.3]
B12 [19.7]19 [22.9]
AB3 [4.9]1 [1.2]
O27 [44.3]37 [44.6]
RaceCaucasoid51 [83.6]69 [83.2]
African American6 [9.8]8 [9.6]
Other4 [6.6]3 [3.6]
Unknown03 [3.6]

We selected one of the sensitized patients at random and asked our clinical testing supervisor and a technologist with 1 year experience to search for matches and we did a computerized search using a computer with a Pentium 3, 600 MHz processor. The patient (patient 1) was a blood type A with antibodies to A2, 28, 10, 11 and 33, DR53 and DQ1. In the first pass, six donors (donors: 2–7) compatible with patient 1 were found. However, donor 1 was incompatible with all of patients 2–7 (four were ABO incompatible and two were HLA incompatible). In the second pass, eight patients (patients 8–15) who could receive a transplant from donor 1 were identified and the donors (8–15) for each of these patients were evaluated for their compatibility with patients 2–7. Five donors were incompatible with all of patients 2–7 (of the 30 possible combinations, 25 were ABO incompatible and 5 were HLA incompatible). Two of the donors were compatible with one of the six patients (2–7) and one donor was compatible with two of the patients, identifying four possible three-way exchanges. The time taken to identify this exchange was: just under 2 h when done manually and 15 s when done by the computer, with the same results obtained by both the manual and computerized searches.

We have also examined the number of exchanges possible for each patient in the database and the results are shown in Table 2. There were a total of 112 two-way and 1230 three-way exchanges identified. There were five patients for whom no exchange was found. Of these five, two had compatible donors identified in the database, one with 15 compatible donors and the other with 30. This suggests that these two patients have a reasonable chance of achieving an exchange when new donors are added to the database or if more extensive exchanges are feasible. The time needed to perform the match runs ranged from 1 s to 1 min and 48 s with a mean time of 30.03 s per patient. The run time did not correlate with the number of exchanges identified but rather with the complexity of the patient's HLA incompatibilities.

Table 2.  Total exchanges identified
  1. aPatient with an exchange identified.

  2. bAll patients.

Number of patients with exchanges found35 (58.3)55 (91.7)
Range of exchanges per patienta1–151–108
Mean number of exchanges per patienta3.520.8
Mean number of exchanges per patientb2.019.1


Kidney exchanges provide an opportunity for living donor renal transplantation to patients who have an immunologically incompatible but otherwise suitable donor. The possibilities for transplantation are greatly enhanced when one is willing to consider unconventional exchanges such as those involving HLA incompatibility. Furthermore, we have shown using a database of moderate size, that permitting three-way exchanges increased the number of patients with a possible exchange by 57% to 91.7% of the patients in the database. In our single center program, we have transplanted 22 patients in 10 exchanges: eight two-way exchanges and two three-way exchanges. ABO incompatibility was involved in 14 patients while 8 patients had HLA incompatibilities.

Compatibility constraints can vary over time and such changes require re-examination of the possible exchanges. For example, adding a fifth unacceptable antigen for a patient with recently expanded HLA-specific antibody reduced the possible exchanges from 4 two-way and 68 three-way exchanges to 3 two-way and 34 three-way exchanges. In other cases, there are patients who have waited for a prolonged time because they are sensitized to numerous HLA antigens. Highly sensitized patients who are blood type O may have a very low probability of finding a donor who is both ABO and HLA compatible. Then, it might be desirable to investigate the possibility of an exchange that would permit one type of incompatibility. Pediatric patients are likely to need a second transplant at some time. Since access to a regraft will be inversely proportional to the extent of sensitization, one may want to minimize the potential for sensitization by minimizing the number of mismatched antigens in the first transplant. Thus, an exchange might be investigated for a pediatric patient with a potential live donor who is ABO-compatible and cross-match negative but who has multiple mismatched antigens.

Opportunities for transplantation are proportional to the numbers of donors and patients willing to be considered for an exchange. However, as the numbers of patients and donors and the complexity of the incompatibilities and matches increase, so does the amount of time needed to identify such matches. Further, performing such searches manually has a high potential for error. We have developed a computer program that performs searches rapidly and accurately. As each new donor-recipient pair is entered into the database, a search can be performed for the patient identifying any new possible exchanges that have arisen with the addition of this pair. All transplant programs considering exchanges beyond the simple pairwise exchange to achieve ABO compatibility should consider utilizing a computerized search program. With minor modifications, the program described here can be made web-based. It can be used for exchange programs that involve several participating transplant centers. This can be done by having the match program administered and run at a central location or, with proper security measures, by providing access to participating centers. Such a program could even be utilized on a national level through programs such as the Organ Procurement and Transplant Network. Our program is available, at no charge, to organ transplant programs. Interested parties should send a written request to the corresponding author.