Unacceptable human leucocyte antigens: how to navigate between increased immunological risk and waiting time?

The continuous improvement of techniques used to detect antibodies directed against human leucocyte antigens (HLAs) during the last decade has important implications for our understanding of transplantation immunology and alloimmunization and consequently the allocation of kidneys for transplantation. For decades, the complement-dependent cytotoxicity (CDC) assay was the main test used to detect anti-HLA antibodies. Cytotoxic antibodies cause hyperacute rejection and their detection by a positive crossmatch using donor cells has been an absolute contraindication against transplantation [1]. The later development of more sensitive solid-phase techniques for the detection of anti-HLA antibodies showed that patients can also develop anti-HLA antibodies that do not necessarily cause a positive CDC crossmatch and acute rejection [2–4]. Indeed, retrospective analyses of historic pretransplant sera have shown that ∼10–20% of patients with a negative CDC crossmatch were unknowingly transplanted across a positive solid-phase donor-specific antibody (DSA) [5, 6]. In a meta-analysis of retrospective studies, transplantation with preformed DSA was associated with a 2- to 4-fold increase in the risk of acute antibody-mediated rejection (AMR) and a 70% increase in graft loss [7]. Although there is a large consensus for the exclusion of HLAs targeted by cytotoxic antibodies, no clear-cut and validated guidelines are available for the exclusion of HLAs targeted by solid-phase anti-HLA antibodies alone [3]. While some solid-phase antibodies are pathogenic, this is obviously not the case for all of them. Indeed, a significant proportion of patients with solid-phase DSA develop neither AMR nor accelerated graft loss [3]. Two characteristics of anti-HLA antibodies have been associated with an increased risk of AMR and graft loss. First, solid-phase antibodies detected at higher titres appear to be more pathogenic [5, 8]. Second, antibodies able to bind complement C1q [9] or C3d [10] have been shown to be more pathogenic than those unable to do so. These two markers might be linked, as a higher titre of DSA is also associated with increased complement binding [11, 12]. Selection of unacceptable HLAs based on antibody titre is hampered by the fact that mean fluorescence intensity (MFI) of single antigen bead assays is only a crude marker of antibody titre. In addition, MFI can vary depending on the antigen density on beads, interference of the secondary detecting antibody by serum factors and conformational changes of antigen on beads [13]. Although there is no definite consensus on MFI cut-offs to determine unacceptable antigens [3], many centres consider that transplantation across a DSA with an MFI >2000–3000 should not be undertaken [5, 14]. Moreover, if a solid-phase anti-HLA antibody is detected against an HLA present on a previously rejected graft, this HLA mismatch (MM) should be excluded, irrespective of its MFI [15]. Clinicians are also confronted with complex choices in selecting unacceptable HLAs based on solid-phase antibody screening. Whereas cytotoxic antibodies are detected by a CDC assay that is either positive or negative, results of single antigen bead solid-phase assays are expressed on a continuous scale. Lowering the MFI thresholds for taking a DSA into account can lead to a large spectrum of unacceptable HLA MMs that will severely restrict the available donor pool and prolong waiting time [13]. Therefore, exclusion of MMs that put the patient at increased risk of rejection and graft loss have to be balanced against leaving the patient longer on dialysis. The annual mortality on the Eurotransplant (ET) waiting list in 2015 was close to 6% [16], an important competing risk to consider when selecting unacceptable HLAs for a particular patient. To maintain equal chances in the access to transplantation, allocation systems should be adapted to compensate for the reduced probability of finding a suitable organ in case the donor pool is restricted by unacceptable HLAs. ET has two mechanisms that aim to improve kidney allocation for patients with HLA antibodies. First is the acceptable MM programme, which is highly efficient but is restricted to highly immunized patients with predominantly cytotoxic antibodies. As a result, <3% of deceased donor kidneys are allocated through this programme [17]. Second is the MM probability (MMP) parameter, which is part of the standard Eurotransplant Kidney Allocation System (ETKAS), adds points to compensate for a reduced probability to receive a kidney offer. It is based on an algorithm taking into account the HLA allelic frequency, ABO blood group and the percentage of panel reactive antibodies (PRA) [18]. Currently, very limited evidence exists on the ability of MMP points to compensate for the effect of unacceptable HLAs on waiting time. This question has now been addressed in a study from Ziemann published in this issue of NDT [19]. The authors conducted a retrospective cohort study in a series of large German transplant centres and validated their predictive model in a second German cohort. This article investigates the impact of unacceptable antigens, quantified by virtual PRA (vPRA), on the chance of being subsequently transplanted. The vPRA reflects the proportion of donors that cannot be accepted because of either ‘repeated mismatched’ HLAs present in a previously rejected graft or unacceptable HLAs targeted by CDC and solid-phase antibodies. As this latter parameter is not yet clearly defined, it falls within a ‘grey zone’ where each centre has different criteria. This report shows that the choice of unacceptable antigens had a significant effect on waiting time despite attempted compensation by MMP in the ETKAS allocation system. In a multivariate model, a 1% increase in vPRA prolonged waiting time by 1.3 weeks, resulting in a prolongation of waiting time by ≥1 year with vPRAs >50%. Why the MMP parameter did not adequately compensate for the increased waiting time that HLA-sensitized patients have to undergo remains hypothetical. How can we explain these findings? First, the MMP corrective factor might not properly reflect HLA immunization. A first limitation of this factor is that in patients with rare HLA alleles, MMP reaches its maximum (100 points) irrespective of the patients’ immunization. Second, the data of Ziemann suggests that MMP might also poorly reflect the true impact of vPRA on the donor pool. Indeed, MMP is calculated based on the percentage of PRA in the most recent trimestral screening and not by the HLAs indicated as unacceptable (vPRA). This implies the potential for significant underestimation of the true restriction of the donor pool by the full scope of unacceptable antigens. The importance of this problem is suggested in the article by a median PRA in the most recent serum of 0% in patients with a baseline vPRA of 51–84% and 26% in those with a baseline vPRA of 85–100%. Importantly, when the difference between vPRA and the most recent screening PRA was added to the predictor variables of the multivariate model, the effect of vPRA on the additional waiting time disappeared, suggesting that compensation based on the true restriction of the donor pool has the potential to completely prevent the excess waiting time in patients with unacceptable antigens. Hopefully, in ET, the vPRA value will replace the percent PRA value in the MM probability calculation at the end of 2017 [Dr Jan De Boer (ET), personal communication]. The article reports another important finding with strong implications for patients on the waiting list. The Eurotransplant Senior Program (ESP) allocates at the local and regional level kidneys from donors ≥65 years of age to the longest-waiting blood group–identical recipient ≥65 years of age, without regard to HLA matching. In Germany, patients participating in the ESP are excluded from ETKAS allocation [18]. The article clearly documents that inclusion in the ESP with a vPRA >50% increases waiting time to an extent that makes it unlikely for older patients to receive a graft during their lifetime. This merely reflects the fact that the probability of finding a matched kidney is low for sensitized patients in the setting of a severely restricted local and regional donor pool. The data provide a strong argument against the restriction of older recipients with a high number of unacceptable antigens to exclusive ESP allocation in Germany. In Belgium, older recipients also remain in the classical ET allocation system, and in The Netherlands, immunized patients are excluded from ESP. It is clearly necessary that German patients in this category are also granted access to a large donor pool providing a reasonable chance to receive an organ. Finally, the article is also a wake-up call for clinicians to become aware of the implications unacceptable HLAs have on transplantability. Unacceptable HLA specificities are, in general, entered by transplant coordinators or the HLA lab into the ET computer system, which automatically calculates the vPRA. This information, and the degree of restriction of the donor pool it entails, is not necessarily known by the transplant physicians. We feel there is an urgent need to have dedicated meetings between transplant physicians, transplant coordinators and histocompatibility specialists to discuss these issues and share the decision process of excluding HLA specificities from the donor pool. The authors are grateful to Dr Jan De Boer of the Eurotransplant medical staff for helpful information on the current algorithm used for the calculation of mismatch probability. None declared. (See related article by Ziemann et al. Unacceptable human leucocyte antigens for organ offers in the era of organ shortage: influence on waiting time before kidney transplantation.