The field of desensitization and incompatible transplantation has made great gains over the past decade. There are now several options and effective therapies for many patients who face antibody barriers. Kidney paired donation (KPD) and desensitization have traditionally been considered competing strategies and patients have been offered one or the other without regard for the probability of a successful outcome. It is now possible to predict which donor/recipient phenotypes will benefit from each of these modalities. KPD should be favored among patients with immunologic phenotypes that are likely to match without prolonged waiting times. However, as many as 50% of patients with incompatible donors will fail to find a match in a KPD pool and many of these patients could be desensitized to their donor. Positive crossmatch and ABO incompatible transplantation has been accomplished in selective cases without the need for heavy immunosuppression or B-cell ablative therapy. Patients who are both difficult-to-match due to broad sensitization and hard-to-desensitize because of strong donor reactivity can often be successfully transplanted through a combination of desensitization and KPD. Using these various modalities it is estimated that most patients with incompatible live donors can undergo successful renal transplantation.
As the crisis in organ availability deepens for all patients in need of kidney transplantation, those who are disadvantaged by HLA sensitization or hard-to-match blood types have been particularly disadvantaged by prolonged wait times. Based on mathematical simulations, it has been estimated that there are currently about 6000 patients in the United States on the deceased donor list who have a willing but incompatible live donor (1). Further, there are approximately 3500 new registrants each year that join the deceased donor list who would benefit from live donor transplantation if their donor blood type or HLA incompatibility could be overcome (2).
Patients who are sensitized but do not have a live donor available have been shown to have a higher transplant rates when treated with monthly doses of IVIg (3,4). There are currently three options available to patients who have an incompatible live donor: desensitization, kidney paired donation (KPD) and a combination of the two modalities. Predicting which modality is the best suited for a particular donor/recipient pair is now possible due to more sophisticated immunologic phenotyping and better understanding of the limitations of both desensitization and KPD. Figure 1 outlines a transplant modality algorithm that takes into account the clinical phenotypes that are likely to benefit from the different options and will serve as the template for this review.
There are currently two desensitization protocols for which clinical efficacy has been demonstrated: high-dose IVIg and plasmapheresis (or immunoadsorption) with low dose IVIg (PP/IVIg) (4,5). Anti-HLA antibody rebounds within days of discontinuing plasmapheresis, whereas the transplantation benefit of high-dose IVIg may continue for many months after the drug is administered. Both protocols are designed to lower DSA strength to a level that is safe for transplantation. Following preconditioning and transplantation, immunoregulartory mechanisms may promote maintenance of reduced antibody reactivity. To date there have been no randomized prospective studies comparing the clinical efficacy of these two protocols. Most centers that have begun desensitization programs have adopted the protocol that best suites the local expertise, reimbursement realities and infrastructure.
The kinetics and endpoints for pretransplant antibody removal varies in different published series depending upon techniques and assays. The literature supports that a safe margin is obtained if antibody strength is lowered below a positive CDC AHG cytotoxic crossmatch for HLA antibody and an IgG titer ≤16 by AHG for isohemagglutination. Monitoring HLA antibody strength and identity through a variety of cell-based and solid phase assays provides a continuous assessment of immunologic risk and aides in determining the optimal timing of transplantation (reviewed in Ref. 6). We believe that no single platform is adequate to fulfill all the requirements of effective monitoring. DSA must be characterized, quantified and tracked in real time and different assays are better or less well suited for each of these functions (7).
The protocols for both live and deceased donor transplants are very similar and consist of monthly infusions 2 gm/kg of IVIg until either the crossmatch is deemed safe or a total of four doses are administered. The multicenter prospective NIH sponsored IGO2 trial which compared 4 monthly doses of IVIg to placebo demonstrated a significant reduction in PRA and a higher transplant rate (39% vs. 17%) for deceased donor transplants in the IVIg arm (8). Interestingly, most of the fall in PRA occurred after the first dose of IVIg and the transplant benefit was realized after the difference in PRA between the two groups disappeared, suggesting that the IVIg effect is incompletely explained by reduction in HLA antibody. The Mayo group performed a head-to-head comparison between a single high dose of IVIg and PP/IVIg in live donor recipients and found that PP/IVIg was more effective in abrogating a positive crossmatch especially when the strength of the crossmatch was higher (9). The Cedars group added a single dose of anti-CD20 to their monthly IVIg dose regimen and achieved an 80% transplant rate among 20 patients with a 100% and 94% 1-year patient and graft survival rate, respectively (10). The reduction in PRA was similar to the IGO2 study and it remains unclear whether anti-CD20 added any efficacy to the IVIg regimen.
Plasmapheresis effectively reduces HLA antibody and isohemagglutinin in preparation for an incompatible live donor transplant (Figure 2). At least two PP/IVIg sessions are performed after transplantation and beyond that the duration of treatment is predicated by DSA levels. Our posttransplant goal has been to deplete DSA strength below that which would yield a positive flow crossmatch. There are various antibody depleting techniques in use including centrifugation, plasma membrane separation and immunoabsorption. Each has a unique side effect profile and it is not known which is best suited for this application. Low-dose IVIg (100 mg/kg) may serve to reduce the synthesis and release of endogenous antibody that occurs after plasma exchange or it may have immunomodulatory effects similar to high-dose protocols.
Splenectomy has been eliminated from most desensitization protocols except as rescue therapy for severe AMR (11,12). The use anti-CD20 remains controversial and largely unsubstantiated. Omitting anti-CD20 in our ABO incompatible patients has not resulted in an increased incidence of rejection or graft loss (13). In our protocols, anti-CD20 is used selectively for patients who have a high-risk donor/recipient phenotype (combined ABOi and +XM, high XM starting titer, multiple DSAs and multiple repeat mismatches). New compounds including Eculizumab and Bortezomib have unique mechanisms of action and are being tested in studies to prevent or reverse AMR and reduce DSA, respectively (14,15).
In our experience desensitizing 60 patients with ABO incompatible live donors the 1-year, 3-year and 5-year graft survival rates are 98%, 93% and 89%, respectively (13). Comparable graft survival rates for ABOi transplants have been achieved by groups in Japan and Sweden (Table 1A). For positive crossmatch transplants results have been less consistent and more often than not inferior to standard live donor transplants (Table 1B). The rate of AMR after desensitization for a +XM transplants is reported to be between 30% and 50% (16). Both the Mayo and Hopkins groups have performed serial antibody screening and surveillance biopsies. The rate of elimination of DSA after engraftment varies between the two groups, varies with antibody specificity (class I > class II), and is perhaps related to different induction strategies (ATG vs. anti-IL-2R antibodies) (17,18). In an important study the Mayo group demonstrated that among patients who had undergone desensitization for a +XM the rate of transplant glomerulopathy on a 1-year protocol biopsy was 22%, rising to 44% among those who sustained an early AMR (19). Longer follow-up will be needed to determine the impact of this finding on graft half-life.
Table 1. Clinical characteristics and outcomes of desensitization protocols to cross ABO (A) and HLA (B) barriers
(A) Graft survival of ABO-incompatible living donor kidney transplantation in large series from Asia, Europe and the United States
Takahashi et al. (Am J Transplant. 2004 Jul;4(7):1089–1096)
84%[1 year]; 80%[3 year]; 71%[5 year]
Ishida et al. (Am J Transplant. 2007 Apr;7(4):825–831)
94%[1year]; 90%[5 year]
Tyden et al. (Transplantation. 2007 May 15;83(9):1153–1155)
97% (17.5 (2–61) months mean follow-up [range])
Montgomery et al. (Transplantation. 2009 Apr 27;87(8):1246–1255)
98.3% (1 year); 92.9% (3 year); 88.7% (5 year)2
(B) Summary of reported graft survival in positive crossmatch transplantation
1Data shown for Group 2, representing the most recently transplanted patients.
2Death censored Graft survival using Kaplan–Meier estimation.
3Includes one heart-lung, one heart-kidney, one liver-kidney, and one heart transplant.
4Forty patients received pre-transplant IA but only 9 had a +XM prior to IA.
5Group 1 Zenapax induction; Group 2 Thymoglobulin induction.
Stegall et al. (Am J Transplant. 2006 Feb;6(2):346–351)
T AHG XM
Plasmapheresis or IVIG
Higgins et al. (Transplantation. 2007 Oct 15;84(7):876–884)
CDC; FCXM; microbead
West-Thielke et al. (Am J Transplant. 2008 Feb;8(2):348–354)
22 AA; 28 non-AA
82.6% AA; 94.1% non-AA
Higgins et al. (Lancet. 1996 Nov 2;348(9036):1208–1211)
26 months (median)
Jordan et al. (Transplantation. 2003 Aug 27;76(4):631–636)
Vo et al. (Am J Transplant. 2008 Jan;8(1):144–149)
Vo et al. (N Engl J Med. 2008 Jul 17;359(3):242–251)
Lorenz et al. (Transplantation. 2005 Mar 27;79(6):696–701)
Haririan et al. (Am J Transplant. 2009 Mar;9(3):536–542)
Vo et el. (Am J Transplant. 2006 Oct;6(10):2384–2390)
84% Group 1; 90% Group 25
Kidney Paired Donation
First proposed by Rapaport in the 1980s, KPD remained a theoretical construct for several decades while it was vetted by ethicists, legal experts and logisticians (20). KPD involves matching a potential kidney recipient who has a willing but incompatible donor to another incompatible pair. This allows the donor of each pair to give a kidney to the recipient of the other pair. The pairing results in both recipients receiving a compatible kidney. In its simplest form two donor/recipient pairs with reciprocal blood type incompatibilities (A/B and B/A) are matched (Figure 3A). However, when pairings involve matching a crossmatch incompatible pair with either blood type or crossmatch incompatible pairs, donors and recipients with blood type O can be included in KPDs greatly increasing its impact (Figure 3B). The outcomes of KPD transplants have been shown to be comparable to standard compatible live donor transplants (21).
Incompatible donor pools can be simulated using computer modeling and mathematical algorithms. When compared to the general donor/recipient population, incompatible pools contain a dramatic blood type skewing towards a greater percentage of hard-to-match O recipients (61% vs. 44%) and fewer valuable O donors (30% vs. 66%) (22). This phenomenon significantly limits the matching options for O recipients in the pool and reduces the overall match rate. Predictions from simulations about which phenotypes match in KPD pools have been validated by actual data from the Netherlands’ national KPD registry (23). Under the best conditions we have predicted that about 47% of patients would find a compatible match (1). After the first match run the remaining 53% will be enriched for hard-to-match patients who are less likely to find matches in subsequent runs as new registrants enter the pool (Figure 4). Many of these patients could be considered for desensitization.
List donation (LD) is another option that is being offered to incompatible pairs at some centers. In this modality a live donor kidney from an incompatible pair is donated to the individual at the top of the UNOS match run for the donor's blood type and in return the donor's intended recipient receives special status on the deceased donor list expediting their transplant with a deceased donor kidney (24). An exchange of a live donor kidney for a deceased donor kidney with a significantly shorter predicted half-life may be hard to justify. Unfortunately, sensitized patients among whom the donor inequity would be easiest to defend will be hard-to-match with a deceased donor kidney and a long waiting time could ensure. Further, due to the blood type skewing among incompatible pairs towards a high percentage of A donors and O recipients there will be a net exodus of blood type O kidneys from the deceased donor list, further accentuating the waiting time disparity for blood type O recipients (25). Simulations of LD have yielded two important insights: (1) LD will result in a higher percentage of transplants when the incompatible pool is small as would be the case at a single center and (2) In scenario's in which there is a large pool and an optimized KPD match the additional benefit of offering LD to the unmatched pairs is very small (Figure 4) (2).
Compatible Pairs and Nondirected Donors (NDD)
The skewing of blood types in incompatible pools can be ameliorated by the inclusion of compatible pairs or NDD. There have been instances where compatible pairs have entered into KPD pools for either altruistic reasons or to gain some benefit such as a younger donor. If large numbers of compatible pairs could be encouraged to enter into KPD pools the percentage of incompatible pairs that would find matches would increase to 75% (22). At present, however, this remains largely a theoretical consideration.
NDDs are a relatively new source of live donor kidneys for transplantation. These are individuals who desire to donate a kidney but do not have a designated recipient. There is currently no broadly accepted allocation system for NDDs and disparate allocation philosophies are being practiced at the discretion of individual transplant centers. A consensus conference on living NDD in 2002 recommended that NDD kidneys should be allocated to patients with the highest priority on the UNOS waiting list (26). This strategy results in a single transplant for each NDD. Allocating NDD to the KPD pool helps donors to more fully realize their altruism by enabling more than one transplant. NDDs not only ameliorate the blood type skewing present in KPD pools but also have the advantage of not requiring a reciprocal match.
Domino Paired Donation (DPD)
In DPD, kidneys from NDDs set off a series of simultaneous transplants terminating in a donation to the deceased donor pool (Figure 5A) (27). Transplants are performed together so the possibility of donor reneging or recipient medical ineligibility is greatly reduced. DPDs can be logistically challenging because the donor operations are started at the same time as is the case with traditional KPDs, and these often involve multiple simultaneous transplants at multiple transplant centers (28).
NEAD is a variation on DPD in which a NDD starts a chain of transplants but the last donor becomes a ‘bridge donor’ who begins another chain of transplants at a later time rather than terminating the chain by giving a kidney to the deceased donor list (Figure 5B) (29). This approach has the advantage of limiting the need for complex logistical arrangements associated with multiple simultaneous transplants. It was originally thought that NEAD would result in more transplants than DPDs because of the nonterminating nature of these chains. However, recent simulations have demonstrated that both strategies yield similar numbers of transplants (30). This is because the blood type advantage from the NDD is lost after the first transplant and NEAD will stall after several iterations due to the appearance of a donor with a difficult-to-match blood type. Since the transplants in a NEAD are not occurring simultaneously the possibility of donor reneging or medical ineligibility is real, especially if there is a long delay between when the recipient receives a kidney and bridge donor donates their kidney.
KPD versus Desensitization
KPD and desensitization have been viewed as competing strategies for overcoming blood group and HLA antibody barriers. This is a false dichotomy. Decisions about the best transplant option for a particular donor/recipient phenotype can now be made rationally using KPD and desensitization as complimentary modalities. When determining which option is best for an individual pair there are two important questions to ask: (1) how difficult will they be to match in a KPD? and (2) how difficult will they be to desensitize? The answer to these two questions will estimate the likelihood of success of either KPD or desensitization for the particular donor/recipient phenotype. This will allow the clinician to guide patients and set expectations.
The first question can be addressed using mathematical simulations to impute the probability of matching in a KPD based on the donor blood type, recipient blood type, degree of sensitization and the size of the KPD pool (1,2). Table 2 shows several examples of this concept. The most common blood type incompatibility, an A donor and an O recipient is the least likely to match in a KPD pool. Pairs with this blood type combination will need to be matched with a positive crossmatch pair in which the donor is blood type O and the reciprocal crossmatch is negative, a relatively low probability event. Thus, in a pool of 100 pairs only 13% of the pairs with this phenotype are expected to match. Increasing the pool size to 1000 only increases the match rate by 4%. Unfortunately, 30% of a KPD pool will be made up of O recipients with A donors and these patients will have prolonged waiting times and desensitization may be their best transplant option. This is in contradistinction to donor/recipient pairs with blood type combination B/A who have a match rate of 70% in a KPD pool of 100 pairs and rises to 83% when the pool increases. Patients with this phenotype should be advised that their best prospect is a KPD. When the recipient is both ABOi with their donor and is highly sensitized match rates are dismal (1%), whereas their transplant rate and outcomes with desensitization are very good (13). The table also shows that having an O donor greatly improves the probability of finding a match in a KPD. Broadly sensitized recipients benefit the most from an O donor and a large KPD pool.
Table 2. Who matches in a KPD pool?
PRA < 80 100 pairs
PRA > 80 100 pairs
PRA < 80 1000 pairs
PRA > 80 1000 pairs
This table is generated by simulation experiments based on known blood type and crossmatch distributions from the UNOS database. It shows the likelihood that different donor/recipient (D/R) phenotypes will find a match in two-way exchanges in a pool of 100 or 1000 incompatible pairs. It also shows the contribution that each phenotypic subgroup will make to a typical incompatible pool. Data like these can help the clinician to choose the best transplant option for individual D/R pairs.
O ⇒ O
A ⇒ O
B ⇒ A
Predicting who will be either difficult-to-desensitize or at risk for AMR is also possible once the immunologic profile of the donor/recipient has been determined. For positive crossmatch patients the strength of the recipient's antibody reactivity to the donor which can be estimated by crossmatch testing or semiquatitative solid phase assays is a good predictor of the length of the desensitization therapy as well as the risk of AMR after the transplant (31). In our experience the presence of multiple donor-directed antibodies, repeat mismatches, or DSA titers ≥32 are associated with a higher risk of AMR and graft loss (5,16). For ABOi transplants there are historical data from Japan that suggested that donor blood type (A1) and high starting isohemagglutinin titer (≥128) were predictive of recipient immunologic graft losses but more recent data from several groups no longer demonstrate significant differences in outcome based on these variables (13,32).
Very broadly sensitized patients with high HLA reactivity that are both difficult-to-match and difficult-to-desensitize can be transplanted by combining KPD and desensitization (Figure 1C). Since the likelihood of finding a compatible donor is low, the goal for these patients is to find a lower immunologic risk donor in the KPD pool, increasing the practicality and success of desensitization. Figure 6 shows this can be accomplished by raising the threshold for the antibody strength that defines unacceptable antigens and then searching the KPD database for genotypes that would permit a positive crossmatch with a low strength.
Technical advances in both the ability to reliably measure DSA and diagnose AMR have led to a renaissance in understanding of the role of antibody in allograft injury and longevity. It has also given hope to thousands of patients who face immunologic barriers once thought to be insurmountable. There are now several different interventions with proven efficacy that can be used to avoid or confront antibody incompatibilities. Understanding the strengths and limitations of these interventions will lead to a more rational application of therapeutic modalities that have the potential to add several thousand additional transplants each year.
Conflict of Interest
The authors of this manuscript have no conflict(s) of interest as defined by the American Journal of Transplantation.