A major stimulant of allograft rejection is recipient T cell recognition of human leukocyte antigens (HLA) in the donor kidney. Preformed donor–specific HLA antibodies (DSA) resulting in hyperacute rejection were first detected in 19691 by the complement-dependent cytotoxicity crossmatch assay (CDC-XM); widespread application of this test detected higher titer DSA and reduced hyperacute and early accelerated rejection episodes.2 Evolution to flow cytometry cross-matching (FCXM) offers improved sensitivity for low titer but nonetheless, pathogenic antibodies.3–9 Newer immunoassays (using ELISA plates or microsphere technology), where purified HLA antigens are covalently bound to a solid-phase platform, have further improved sensitivity and specificity of HLA antibody detection.10 Despite advancements in technology, newer solid-phase assays have a number of interpretive considerations that must be appreciated by clinicians in order to more appropriately apply test results to patient care.
This review for transplant clinicians first discusses the analytic and technical parameters and quantitative considerations of contemporary HLA antibody testing methods, with an emphasis on the commonly used single antigen beads (SAB) and then explores the application and utility of SAB testing pre- and post-transplant. Non-HLA antibodies, although potentially determinants of outcomes, are outside the scope of this review and will be discussed only briefly.11–14
Previous SectionNext Section
Cell-Based Assays: Detecting HLA Antibodies before Solid-Phase Assays
Panel-reactive antibody (PRA) testing, in general, estimates the percentage of potential donors to whom a recipient has HLA antibodies and approximates the risk of positive crossmatch. Early PRA assays used panels of 25–60 real donor cells selected to represent the common HLA phenotype distribution of the potential deceased donor population, which were then tested for complement-dependent cytotoxicity with recipients’ sera. Results were subject to change with different donor cells in the panel, were less sensitive for rare HLA antigens, and detected only higher titer antibodies. Although still used in conjunction with other assays, cell panels are no longer used in isolation for PRA determination. The donor–specific CDC-XM assay detects high-titer DSA required to bind complement for demonstration of antibody presence, whereas lower titer DSA may be detected by FCXM, with a positive result requiring only antibody binding and not complement activation.
All cell-based assays are subject to false-positive results caused by autoantibodies and non-HLA antibodies10,15–21 as well as false-negative results when the antibody is very low titer but still has clinical relevance.22,23 Cell-based assays are now routinely augmented by solid-phase assays, with significantly improved sensitivity and specificity. For a more detailed review of HLA techniques and interpretation, we refer to other publications.24–32
Previous SectionNext Section
Solid-phase assays, by comparison, are more sensitive for lower titer antibodies and permit more precise determination of the specific HLA antigens and alleles to which they bind. Furthermore, complimenting traditional cell–based methods with solid-phase assays offers the potential to better discriminate immunologically relevant positive XMs from false-positive results.33
To perform these assays, recipient serum is incubated with purified HLA antigens presented on a solid-phase platform (commonly microparticle beads). A fluorescent–conjugated anti-human IgG is added, which binds to and detects HLA antibody on its antigen target when the beads are analyzed on a flow cytometer or Luminex® platform. The latter platform generates a semiquantitative output for each bead of mean fluorescence intensity (MFI), which is compared with negative control MFI to determine if HLA antibody is present (Figure 1A). It has proven difficult to align MFI measurements within and between laboratories notwithstanding the increased reporting of this metric in published research.34,35 HLA antibody can also be detected in an indirect ELISA36; however, this is less sensitive than bead platforms.10 Subsequent discussion will be restricted to bead-based assays. From a technical perspective, SAB assays allow identification of HLA antibodies for all common and many rare antigens and alleles at up to 11 HLA loci.37 SAB assays are rapid (3–6 hours), with up to 100 unique antigen beads able to be tested in a single reaction chamber, and high throughput, with additional multiplexing ability to test many patients simultaneously.
MFI of single antigen bead assays has analytic limitations and cannot be used as a quantitative metric of antibody amount. (A) An ideal test should always be able to distinguish antibody binding (green fluorescent signal) from negative control (white) with a clear threshold and no overlap between the MFI distributions. (B) Decreased density of antigen (Ag) on the surface of the bead will result in MFI measurement that underestimates the amount of the antibody present. (C) In contrast, nonspecific binding to the bead can result in artificially high background and signal MFI, with overestimation of antibody. (D) Interfering substances may prevent the detection of the antibody of interest with lower MFI. (E) Epitopes shared between different beads can dilute the amount of antibody bound to any single bead, with an erroneously low MFI on the given bead of interest.
Solid-phase bead assays may be supplied as pooled antigen panels, single-patient phenotypes, or SAB. Pooled panel beads with many different class I or II HLA antigens on a bead yield a positive/negative result and are used for screening.38 Phenotype (also called ID) beads are each coated with class I or II HLA antigens of an individual patient–derived cell line39 and estimate PRA by the percentage of positive beads. SAB are each coated with a single HLA antigen37 and yield a list of discrete antibody specificities. Specificities are then compared with HLA frequencies in the donor population40–42 to determine the calculated panel-reactive antibody (cPRA), which is presently the best estimate of likelihood of a positive XM/DSA to a randomly selected donor.43,44 SAB results further enable virtual crossmatching (VXM) to identify DSA pretransplant, in turn facilitating allocation and risk assessment.45 VXM has also been used without cell-based XM in some transplant circumstances where legislation permits, or under study conditions, with acceptable (equivalent rejection rates and graft survival at a population level) results.23,46 Additional enhancements of the SAB assay, such as detecting antibodies capable of binding complement component C1q, have been developed to detect potentially more injurious antibodies.47,48
Previous SectionNext Section
Interpretive Considerations of Solid-Phase Assays
Defining a Positive Result
The numeric output of Luminex® SAB is a trimmed MFI (or channel shift in flow cytometry SAB) (Table 1). SAB analysis considers the MFI along with other factors in determining a positive result: laboratories validate their SAB assays with known negative and positive sera, establish a working MFI threshold for antibody detection, and correlate this threshold with positive FCXM results as an important laboratory–specific standardization process within a program. MFI thresholds may be modified on the basis of patient history, control values, different HLA loci, recipient HLA typing, and consideration of epitope/antigen groups (Table 1). As such, a strict MFI level above which clinically relevant antibodies are consistently identified is challenging to define.
Factors affecting MFI values in SAB HLA antibody assays
The Metric Is Not the (Whole) Message
Ideally, a standard amount of antigen would be present on the bead of interest and would be bound proportionally to antibody amount, yielding a reliable and quantitative MFI result (Figure 1A). Unfortunately, several of the assay properties result in MFI being, at best, a semiquantitative output.34
Antigen Density Is Not Equivalent between Beads
Antigen density varies between beads both within a single assay (Figures 1B and 2) and between manufacturers.49 MFI differences between beads in an individual assay may be caused by variable antibody amounts but also, differing amounts of target antigen present.50
Individual single antigen beads within a single-assay reaction can have significant differences in the density of target antigen affecting maximum MFI. A saturating amount of W6/32 antibody that binds ubiquitously to class I HLAs was mixed with single antigen beads. The unadjusted results are shown, with the fluorescence representative of the density of the target antigen on each bead. Wide variation in density is seen.
Nonspecific Serum Components May Bind to the Bead in Addition to Antibodies of Interest
Nonspecific binding of non-HLA antibodies to beads (for example, in the presence of drugs, such as intravenous Ig [IVIG], inflammation, or infection) may increase background MFI of both control beads and beads of interest (Figure 1C); isolated MFI of a target bead may be misleadingly high. Control values may provide clues to this issue but are rarely routinely provided in clinical laboratory reports or research studies.
Interference May Prevent Binding of the Antibody of Interest
High-titer antibody leading to complement activation and deposition of C1 complex on the bead,51,52 IgM antibody,53 or other serum factors54,55 may interfere with binding of the secondary detection reagent, giving false-negative results, and is known as the prozone effect (Figure 1D). Serum dilution or treatment with EDTA/dithiothreitol, osmosis, or heat has been shown to reduce this effect50,51,56 but is often variably applied (Figure 3). Interference may also be caused by drugs (e.g., IVIG)50,57,58 or nonspecific serum proteins.59 We further note that dilution does not result in a predictable decline in bead MFI (Figure 4); therefore, the MFI of a single bead in an undiluted serum cannot be assumed to quantitatively or reliably represent antibody amount.
Removal of interfering factors with serum treatment can significantly increase MFI on beads of interest. (A) Neat serum (no treatment) SAB results of a sensitized renal transplant recipient. MFI appears on the y axis. Each bar represents a single bead. The bar graph illustrates MFI measurement for each HLA allele, with self-antigens (expected to be negative) identified with purple arrows and donor antigens indicated with red arrows. The A2 beads representing donor antigens are clearly negative. An XM would be predicted to be negative, because no antibodies to donor antigens are identified; however, the FCXM was strongly positive. (B) After treatment of the serum with dithiothreitol (DTT), antibody to donor HLA-A2 is detected at MFI>20,000.
Dilution of serum in SAB assays does not result in a predictable or quantitative reduction in bead MFI. Serial dilutions were performed on dithiothreitol-treated serum tested on SAB in 10 consecutive patients with DSA. The expected change in MFI from baseline over dilution if neat serum MFI quantitatively represented antibody amount is shown with the dashed lines. MFIs of the highest bead(s) contributing to DSA in each of the sera are shown at each dilution as a percentage of the neat serum MFI (by definition, 100%). The MFIs vary widely through dilution; although antibody concentration is decreasing by a known amount through serial dilution, MFI changes are not proportional. Neat serum MFI does not, therefore, reliably represent antibody amount.
Target Epitopes May Be Shared between Beads
Groups of antigens (and their corresponding SAB) may share common or public epitopes targeted by antibodies (Figure 1E, Table 2). Antibody to a shared epitope may be diluted across these multiple beads, reducing MFI on any single bead of interest26 (Figure 5).
Epitopes targeted by HLA antibodies may be shared across multiple beads in a single assay, lowering the MFI detected on any individual bead. Multiple beads share the Bw6 epitope. In this case, the mismatched donor antigen of interest is B35, with a normalized bead MFI of 1216 (weak). Many centers would predict a negative FCXM with this donor; however, it was strongly positive. Closer examination reveals that the majority of antigens sharing the Bw6 epitope are clustered together in this MFI range. Neither a B7 nor a B5 cross–reactive epitope group explains this reactivity pattern entirely; rather, it is likely that an antibody to Bw6 is diluted across multiple beads sharing this epitope. During in vitro XM, where only the B35 target was present, the antibody can bind without epitope dilution, yielding a positive result. CREG, cross reactive group.
Examples of common Class I cross–reactive (public) epitope groups
Conformational Changes of Antigens on Beads
Artificial attachment of an antigen to a bead may result in conformational changes of the protein,60,61 potentially leading to exposure of epitopes not normally found in vivo (neoepitopes), yielding false-positive results,62–65 or concealment of immunologically relevant epitopes (cryptic epitopes), giving false-negative results.
Laboratory Assay Modifications
Many laboratories modify the manufacturer methods to enhance assay sensitivity, detect different antibody isotypes, reduce assay time, or reduce reagent costs. These modifications may also affect comparability of MFI between laboratories.
It is our laboratory experience that antibodies to Cw and DP antigens must be detected at much higher MFI than DSA at other loci before a positive XM occurs. Other laboratories may use different MFI thresholds for identifying antibodies at different HLA loci on the basis of their own validation studies.
Other Reagent and/or Lot-Specific Variables
Antigens may be represented in the SAB assay by differing numbers of unique beads; there are, for example, in a current reagent lot, three A2 beads and five DQ2 beads but only one C7 bead. Subsaturating antibody may be diluted across multiple beads in a multiplexed assay more so than if only a single bead was present, thereby lowering MFI on any single bead of interest. For higher antibody levels, maximum total bead MFI may be higher for those antigens with multiple beads. We note that there is no validated method of combining MFI across multiple beads, however summing is frequently reported in the literature.
Lot-to-lot variability in bead reactivity, bead specificities, and reagents has additionally been recognized as a confounder to inter- and intralaboratory reproducibility.35,66,67 MFI differences may also originate from differences in instrument type, maintenance, and operators, reagent differences between vendors, and variability in secondary reagent properties.
These considerations do not negate the major role that SAB assays have played in the last decade, permitting significantly more reliable identification of HLA antibody specificities and strengths. Rather we encourage caution when extrapolating research reporting MFI data to other centers, because these data are rarely characterized with sufficient detail of the above factors to appropriately generalize them. Even more importantly, an MFI result as a single metric cannot be analyzed in isolation. The patient’s history of sensitizing events, their clinical status and pathology, their own and their donor’s HLA typing, and other related test results (e.g., FCXM) in the user’s own laboratory must also be considered for comprehensive understanding. Additionally, program clinical thresholds for crossing antibodies, regional standards, and regulations for listing of unacceptable antigens as well as differences between organs may affect final recommendations.
Previous SectionNext Section
Utility of HLA Antibody Testing Pretransplant
Waitlist Testing for HLA Antibodies
Testing for HLA antibodies while waitlisted or during transplant workup serves to identify transplant candidates with potentially reduced access to acceptable donors by virtue of preexisting HLA antibodies acquired through pregnancy, transfusion, or prior transplant. The cPRA serves as an estimate of the percentage of potential donors to whom a recipient will have DSA (with corresponding risk of a positive XM). Beyond application of cPRA in quantifying transplant access for patients and care providers, those patients with the highest cPRA (and lowest access to acceptably mismatched donors) may be prioritized in allocation schema to improve equity of transplant access. Because antibody levels and specificities wax and wane over time, repeated testing (usually 3–12 times per year depending on regional regulations) is performed to obtain the most comprehensive immunologic profile. Even if transient, detected antibodies may represent potential for future immunologic memory responses if targeted to donor antigens and affect subsequent risk assessment.
Testing at the Time of Transplant
Absence of DSA, especially in high titer, has generally been considered a prerequisite for successful transplantation, and at a minimum, a negative pretransplant XM remains desirable. However, below these higher antibody levels1,68 exists a spectrum of antibody specificities and strengths with variable clinical effects, potentially affecting a substantial proportion of waitlisted individuals with broad antibody specificities.50
DSA with a Positive XM: Is Forewarning Always Enough?
A positive XM with solid-phase evidence of HLA DSA is associated with worse short and long-term allograft outcomes8,33,45; conversely, in the absence of HLA DSA, a positive XM may not correlate as strongly with outcomes.69 Identification of pretransplant DSA in living70–77 and deceased donation78 offers an opportunity to lower antibody levels through desensitization and permit transplant; however, long-term outcomes are variable: acute AMR rates of 37%–45% are reported in desensitization series compared with <5% of controls, with transplant glomerulopathy (a form of chronic AMR) occurring in up to one half of patients compared with <10% of controls.79–81
MFI of pretransplant DSA are not, however, consistently predictive of outcomes. Gloor et al.80 showed increased AMR and TG in patients with positive XM at all levels of DSA MFI, and although a clear increase in AMR was seen with total MFI >10,000, the other patients with DSA also had an increase in AMR that was not influenced by MFI. Other studies report graft survival effect at MFI>1500,82 and still others find it at 3000 MFI,83 emphasizing threshold variability. Extreme levels of MFI (>10,000) may correlate with complement–fixing C1q–positive DSA,48,84–86 but these data do not show any increase in adverse outcomes with pretransplant C1q(+)DSA over standard DSA detection methods.47,86,87
DSA with a Negative XM: Defining Risk Remains Challenging
The increased sensitivity of SAB allows DSA to be detected even with a negative XM.80,88,89 AMR rates of 20%–55% are reported under these circumstances,22,23,83,88,90,91 with AMR negatively affecting the impact of DSA on graft survival in some studies90 but not all.83
In series where DSA are associated with worse allograft survival,82,83,88,92–95 DSA MFI is an imperfect discriminator of outcomes. In contrast, other groups identified no change in rejection or graft survival in patients who were DSA positive but XM negative.96–101 Studies are generally small and retrospective, with difficulty comparing laboratory methods between centers. No DSA features (class, number, and/or MFI) beyond detection alone consistently predict outcomes,83,90 and importantly, many patients with DSA do not have adverse outcome; strategies to better risk stratify pretransplant DSA are needed.
VXM: Antibody Testing to Identify the Correct Donor
Despite historical associations of high PRA with immunologically poor outcomes,102,103 studies with these improved technologies have shown that patients with high PRA but no DSA have graft- and rejection-free survival comparable with unsensitized recipients.45,82,88,94,104 VXM compares recipient antibody specificities with donor antigens, permitting rapid and earlier identification of DSA-negative donors in deceased donor acceptable mismatch strategies45,104 and living donor paired exchange programs worldwide.105–108
Previous SectionNext Section
Utility of HLA Antibody Testing Post-Transplant
It has been a decade since the multicenter studies by Terasaki and Ozawa,109,110 in which patients with HLA antibodies detected on solid-phase platforms reported higher rates of allograft failure at 1 (6% versus 3%)109 and 2 (15.1% versus 6.8%) 110 years than those without HLA antibodies. Despite widespread ensuing interest in understanding the role of HLA antibodies and DSA in diagnosing and treating antibody-mediated complications, to date no study has clearly defined optimal timing for antibody testing, characteristics of antibodies likely to be associated with adverse outcomes, or definitive treatment strategies. Studies have been retrospective and nonrandomized, with noncomparable patient phenotypes, differing thresholds for DSA detection, variable testing time points post-transplant, and nonstandard treatment strategies. Despite this, when collectively examined, available studies do provide insights into the relationship of DSA with histopathology and impaired function, the prognostic utility of testing, and potential strategies for treatment and prevention.
Incidence and Prevalence of De Novo DSA and Timing of Outcomes
Examining numerous studies utilizing solid-phase platforms,111–124 the median onset to de novo donor–specific HLA antibodies (dnDSAs) varies from 3.8115 to 68 months.113 Cumulative prevalence at 3 years post-transplant is similarly varied from 6%116 to 38%,113 with lifetime cumulative prevalence even more disparate. Class II antibodies predominate in most115–125 but not all reports.111–114 With differences in antibody MFI cutoffs, baseline immunosuppression, and frequency of testing, definitive conclusions remain elusive.
The timing of dnDSA occurrence to onset of graft dysfunction ranges from months to years,111,113,114,116,118,123,126–130 suggesting multiple pathways of injury and potentially diverse modifying factors. A comprehensive natural history study coupled to protocol and for-cause biopsies showed the mean time to dnDSA as 4.6 years, with the most important independent predictor being immunosuppressive nonadherence (60% cumulative prevalence versus 20% in adherent patients at 10 years; odds ratio, 8.75).116 Subclinical antibody–mediated injury (peritubular capillaritis and C4d positivity) occurred in more than one half of patients with stable function, suggesting a period of latency between onset of inflammation and dysfunction that was shortened by nonadherence. dnDSA was associated with significantly reduced graft survival of 50% at 11 years post-transplantation (versus >90% without dnDSA), with similar findings in other series.109,111,116,121,124,129,131 Notably, in all studies, the relationship between dnDSA and adverse outcome is imperfect; better understanding of which DSAs are more deleterious is urgently needed.
dnDSA and Features of AMR
Late AMR is associated with chronic pathology and worse dysfunction at the time of diagnosis, mixed cellular and humoral features, and subsequently, lower treatment response rates,132–134 but it is notable that DSA MFI does not reliably further stratify those individuals who will respond to treatment from those who will do not.133 Indeed, in one study, those who responded to treatment had higher antibody level as estimated by MFI than nonresponders.135 Others that do report an association of MFI and treatment response134,136 note wide overlap in MFI ranges between response groups with no clear discriminating thresholds. Additionally, DSA MFI metrics do not correlate with the chronicity and severity of pathology of AMR.136 However, both late rejections and concomitant resistance to treatment are strongly associated with class II DSA, especially HLA-DQ, antibody specificities.119,126,133–135,137
Recent studies have indicated that post-transplant complement–fixing DSA (detected by C1q binding in an SAB assay) may confer greater risk to allografts than noncomplement-binding DSA. In a study of 1016 patients, post-transplant C1q+DSA were associated with worse pathology and 5-year graft survival (54%) than C1q−DSA. Also, those who developed C1q+DSA after transplant but had C1q−DSA before transplant had the worst outcomes.47 However, another study found C1q+DSA more likely to be of DQ specificity and also, associated with a 30% decrement in 5-year survival.84 To bind complement, antibodies must be IgG1/3 isotope and be of sufficiently high titer; closer examination of these studies and others indicates that antibody amount (estimated albeit imperfectly by MFI) is the major predictor of C1q positivity and outcomes,48,84–87,117 and the role of this new assay in addition routine SAB testing remains to be fully determined.
DSA without Dysfunction
With latency between detection of DSA and dysfunction/injury, it is not clearly known how to treat patients with dnDSA in whom no dysfunction or histologic antibody–mediated injury has occurred. One study reports progression of inflammatory pathology, despite optimization of immunosuppression and IVIG.116 Clinical trials are needed to determine the optimal expectant approaches in these patients and better identify which DSA are likely to have clinical effect.34
Non-HLA Antibodies and AMR
Studies continue to explore the role of non-HLA antibodies in AMR outcomes, although purely non-HLA antibody–mediated rejection remains rare.138 Despite early enthusiasm for the role of MICA antibody testing to predict AMR,11,139 more recent studies question this association and the benefit of routine MICA testing in addition to routine HLA antibody detection.138,140 In recent studies of angiotensin II type 1 receptor antibodies,12,141,142 a majority of recipients with an adverse outcome also had HLA antibodies, and again, the incremental value of the test must be weighed against the cost of widespread screening, with many centers using this testing on a case-by-case basis instead.
Treatment Options: Is There a Role for Monitoring Antibodies?
Several studies reported on the change in MFI levels of DSA in response to various treatments of AMR.134,135 In one, patients who received treatment with IVIG, anti-CD20, and plasmapheresis had greater reductions in MFI compared with patients receiving only IVIG; however, histologic and temporal features suggested that the latter group may have had initially worse rejection features, confounding this association. Reductions in DSA MFI correlated with favorable functional outcome at 3 months post-treatment, a finding replicated in other studies133; however, despite improved creatinine, considerable subclinical inflammation persisted, and longer-term outcomes are not known. Walsh et al.134 studied a combination of rituximab, bortezomib, and plasmapheresis in patients with early and late AMR; treatment was more successful in those with early AMR who concomitantly had less dysfunction, fewer chronic findings on pathology, less nonadherence, and lower antibody levels with fewer DQ specificities. Everly et al.135 report on four different combinations of treatments in AMR, ACR, and mixed rejection with a reduction in DSA MFI associated with improved graft survival at 5 years; however, no clinical feature other than DQ antibody specificity predicted this response, and no treatment regimen was superior. However, other studies confirm that early timing of AMR is the most important factor in predicting response to treatment.143 With a multiplicity of confounders, conclusions of treatment efficacy on the basis of antibody characteristics cannot presently be drawn.
A more crucial observation is that antibodies are rarely eradicated (and in many cases, not even significantly reduced), even in patients with clinical improvement134,135; treating AMR to a likely unattainable goal of antibody elimination bears a substantial risk for immunotoxicity.
Equally important is recognizing that the clinical improvement that accompanies reduction in DSA metrics is more easily demonstrable with routine biochemistry testing than an isolated antibody metric, with the limitations outlined earlier in this paper. The role of serial HLA antibody testing to guide treatment after diagnosis remains to be determined.
Finally, the most effective therapeutic strategy for AMR and post-transplant DSA remains unknown, particularly in the absence of dysfunction or inflammation; prudent consideration of the risks and the benefits in individual patients is recommended.
Can DSA Be Prevented?
Given generally poor responses to therapy, strategies to prevent dnDSA development must be considered. Beyond addressing nonadherence144–146 or less potent immunosuppression,112,118,120,124,125 increasing evidence that class II mismatching, especially at DQ, is a major determinant of DSA development114,124,137,145 must prompt considerations of optimizing class II matching. Early cell–mediated inflammatory pathways and infections are also associated with new HLA antibody development147,148; control of these pathways represents additional strategies to be explored (Table 3).
Clinical considerations in antibody testing and interpretation
- Copyright © 2015 by the American Society of Nephrology
In this issue, David Eckels discusses the clinical implications of solid-phase human leucocyte antigen (HLA) antibody testing for organ transplantation (Eckels, 2008).
Although the manuscript was considered too controversial according to American standards, it is an excellent basis for discussions on this important topic and, hopefully, will stimulate international collaboration aiming at solving the many issues, which are not supported by solid scientific evidence yet. Actually, many of the points raised in this discussion paper are not controversial at all and some of them even common use in many transplantation programs in Europe and particularly within Eurotransplant.
Since the introduction of very sensitive luminex-based assays by commercial companies, both HLA laboratories and clinicians are confused. They were all educated by the old dogma that donor-specific antibodies are always a contraindication for transplantation (Patel & Terasaki, 1969) and many of them immediately gave a similar value to antibodies detected by these newly developed assays. However, it is clear that this dogma is not valid anymore, and the pretransplant assessment of donor-reactive HLA-specific antibodies should rather be considered as a parameter for the risk of adverse complications after transplantation (Gebel et al., 2003). Within Eurotransplant the standard cross-match is still the original complement-dependent cytotoxicity (CDC) assay, known to be clinically relevant. Many American colleagues consider this approach as not ethical (even reviewers of our papers make such remarks) but transplant results in (highly) sensitized patients are at least as good in Europe compared to the USA (Doxiadis et al., 2005). On the other hand, complicated and very expensive desensitization programs are applied by different centres in the USA in order to transplant ‘sensitized’ patients after removal of antibodies, which are not detectable in CDC and, until recently, not even monitored in Europe. Desensitization is probably useful for a proportion but certainly not for all patients included in these programs.
A recent retrospective study in highly sensitized patients, transplanted on the basis of a negative CDC cross-match, showed that the presence of non-complement fixing donor-specific antibodies detected by luminex is associated with a (treatable) rejection in only a subpopulation of patients while many patients with donor-specific antibodies only detectable in luminex have an excellent long-term kidney graft survival (van den Berg-Loonen et al., 2008).
The challenge is, as stated by David Eckels, to discriminate clinically relevant from non-clinically relevant antibodies. In order to do so, it is essential to define the actual specificities of the antibodies present in the serum of a patient. Preferentially, one should be able to explain these specificities by a previous sensitizing event in the history of the patient. In the current era with our increased knowledge of the antibody epitopes present on the HLA antigens (Duquesnoy, 2006; El-Awar et al., 2007), this is feasible. However, the problem is that many centres do accept all specificities generated by the computer programs linked to the commercial screening assays without evaluating whether these antibody specificities make any sense, immunologically speaking.
As already stated by David Eckels, luminex-based assays from different companies may generate different antibody specificities, which is clearly an argument to look critically at the results. One of the reasons why not all antibody specificities are relevant is the fact that these assays are based on antibody binding to isolated HLA molecules, which may have a different conformation than the HLA molecules naturally expressed on the cell membrane of the donor organ.
But even the presence of well-defined donor-specific HLA antibodies is certainly not always a contraindication for transplantation. On one hand, these antibodies may lead to hyperacute or early acute humoral rejection but, on the other hand, they may be associated with no negative clinical effect (van den Berg-Loonen et al., 2008) or even an enhanced graft survival (Koka et al., 1993).
The challenge is to preassess the risk associated with the presence of donor-specific antibodies and to use this knowledge for donor selection and/or the immunosuppressive policy around or after transplantation.
I do not agree with the statement of David Eckels that we should drop the CDC test as, in contrast to antibodies detected by the other very sensitive assays, a positive CDC cross-match due to donor HLA-specific IgG antibodies is known to be clinically relevant in the majority of the cases. As long as we do not agree on the relevance of all the other assays, a standard CDC cross-match is helpful to prevent transplantation of patients with detrimental donor-specific HLA antibodies. The next necessary step is to start international collaboration to determine the clinical relevance of all the different types of antibodies.
Although several reports show the presence of HLA-C, -DQA, -DQB, -DPA and -DPB antibodies in potential transplant recipients (i.e. Duquesnoy et al., 2008), no systematic analysis of their clinical relevance has been performed and as a consequence many completely different local policies are used in patients with these types of antibodies.
Both the expression of the target molecules in the transplanted organ and the titre and immunoglobulin class (or subclass) may play a determinative role in this aspect.
The basis of our decision-making should be antibody specificity and not percentage panel reactive antibodies (%PRA) as it has been for years. The old definition of percentage PRA (just based on antibody reactivity against a panel) must disappear and percentage PRA should be based on antibody specificities in combination with the frequency of the target antigens in the donor population. Actually, such an algorithm has been introduced in Eurotransplant.
In the future, one should be able to calculate two types of PRA: first, percentage PRA on the basis of antibodies, which are a contraindication for transplantation, and second, percentage PRA on the basis of antibodies, which should be taken into consideration with respect to the choice of immunosuppressive treatment if the patient underwent transplantation despite the presence of these antibodies.
Although knowledge of antibody specificities is important, even more important is a reliable definition of acceptable mismatches, those HLA antigens towards the patient never made a (potentially detrimental) antibody response. Donor selection based on acceptable mismatches has shown to be a very efficient tool to enhance transplantation of highly sensitized patients within Eurotransplant (Claas et al., 2004). If the future definition of acceptable mismatches is based on very sensitive luminex assays, one can even consider transplanting highly sensitized patients without performing a pretransplant cross-match provided that donor and recipient are typed at a high resolution level, thereby excluding a possible effect of allele-specific HLA antibodies.
It is clear that the introduction of solid-phase HLA antibody tests has had a tremendous impact on the field of histocompatibility testing. What is currently lacking is consensus on the relevance of the different types of donor-specific antibodies for donor selection (which antibodies are a contraindication for transplantation) and for immunosuppressive treatment (which antibodies are associated with treatable early acute humoral rejection).
It is essential to use the results of antibody screening assays not only to prevent transplantation of patients but to facilitate transplantation of (highly) sensitized patients on basis of acceptable mismatches or in the presence of clinically non-detrimental donor-specific HLA antibodies.
Routinely monitoring the appearance of donor-specific antibodies after transplantation may be an additional tool to prevent loss of transplanted organs provided that adequate treatment is available (Mao et al., 2007).
However, we should not forget that the best policy is prevention of antibody formation. HLA matching, and especially matching of donor and recipient for the antibody epitopes present on the HLA molecules (Dankers et al., 2004), will prevent that many retransplant candidates will finally end up as highly sensitized patients with very extensive antibody profiles in the current solid-phase assays.
- van den Berg-Loonen EM, Billen EV, Voorter CE, van Heurn LWE, Claas FHJ, van Hooff JP, et al. Clinical relevance of pretransplant donor-directed antibodies detected by single antigen beads in highly sensitized renal transplant patients. Transplantation. 2008;85:1086.[PubMed]
- Claas FHJ, Witvliet MD, Duquesnoy RJ, Persijn GG, Doxiadis IIN. The acceptable mismatch program as a fast tool for highly sensitized patients awaiting a cadaveric kidney transplantation: short waiting time and excellent graft outcome. Transplantation. 2004;78:190.[PubMed]
- Dankers MK, Witvliet MD, Roelen DL, de Lange P, Korfage N, Persijn GG, et al. The number of amino acid triplet differences between patient and donor is predictive for the antibody reactivity against mismatched human leukocyte antigens. Transplantation. 2004;77:1236.[PubMed]
- Doxiadis IIN, Duquesnoy RJ, Claas FHJ. Extending options for highly sensitized patients to receive a suitable kidney graft. Current Opinion in Immunology. 2005;17:536.[PubMed]
- Duquesnoy RJ. A structurally based approach to determine HLA compatibility at the humoral level. Human Immunology. 2006;67:847.[PMC free article][PubMed]
- Duquesnoy RJ, Awadalla Y, Lomago J, Jelinek L, Howe J, Zern D, et al. Retransplant candidates have donor-specific antibodies that react with structurally defined HLA-DR,DQ,DP epitopes. Transplant Immunology. 2008;18:352.[PMC free article][PubMed]
- Eckels DD. Solid phase testing in the HLA laboratory: implications for organ allocation. International Journal of Immunogenetics. 2008 this issue. [PubMed]
- El-Awar N, Terasaki PI, Cai J, Deng CT, Ozawa M, Nguyen A. Epitopes of the HLA-A,B, C, DR, DQ and MICA antigens. Clinical Transplantation. 2007:175–94.[PubMed]
- Gebel HM, Bray RA, Nickerson P. Pre-transplant assessment of donor reactive, HLA-specific antibodies in renal transplantation: contraindication vs risk. American Journal of Transplantation. 2003;3:1488.[PubMed]
- Koka P, Chia D, Terasaki PI, Chan H, Chia J, Ozawa M, et al. The role of IgA anti-HLA class I antibodies in kidney transplant survival. Transplantation. 1993;56:207.[PubMed]
- Mao Q, Terasaki PI, Cai J, Briley K, Catrou P, Haisch C, et al. Extremely high association between appearance of HLA antibodies and failure of kidney grafts in a five-year longitudinal study. American Journal of Transplantation. 2007;7:864.[PubMed]
- Patel R, Terasaki PI. Significance of the positive crossmatch in kidney transplantation. The New England Journal of Medicine. 1969;280:735.[PubMed]