HLA DQB1 antibodies are the commonest de novo donor-specific antibodies (DSA) detected postrenal transplantation.1-6 Their development has been shown to be associated with unfavorable allograft outcomes, namely antibody-mediated rejection (AMR), transplant glomerulopathy (TG) and allograft failure.2,5-7 It has been postulated that incorporating HLA DQ antigen matching into organ allocation schema may reduce the incidence of de novo DQ DSA and the subsequent detrimental sequelae.2,4,6,8 However, there is now increasing evidence that matching at the epitope level is superior to matching at the antigen level in predicting the development of de novo DSA.3,9 Therefore, the future of organ allocation may not only include HLA DQ antigen matching but also utilization of DQ epitope matching.3,9
To our knowledge, no previous study has systemically analyzed the impact of DQ epitope matching based on Terasaki-defined epitopes (TerEp). TerEps were established by analyzing the antibody reactivity patterns from serum eluted from known HLA antigen cell lines and determining the shared amino acid sequences on the resultant positive single antigens beads.10-13 A benefit of using TerEps over epitopes derived from computer algorithms is that we already know that these epitopes are immunogenic due to the nature by which they were defined.
In this study, we compare the immunogenicity of the different HLA antigens, individual DQB1 alleles, and DQB1 TerEps, in a large cohort of renal transplant recipients. We aim to determine if TerEp mismatches can be used to further risk stratify patients at risk of de novo DQ DSA development. Finally, we analyze the correlation between the presence of DQB1 TerEp antibodies and allograft outcomes, which has not been previously reported.9
MATERIALS AND METHODS
One thousand three patients transplanted at Imperial College Renal and Transplant Centre between 2005 and 2013 were analyzed. We included kidney alone (both deceased and living donors) and simultaneous pancreas and kidney recipients and excluded antibody-incompatible patients defined as those recipients with preformed DSA detected by luminex only (with an MFI > 500), patients with a positive flow or CDC crossmatch and ABO-incompatible patients. We also excluded patients who experienced early allograft failure in the first week posttransplant.
All patients received monoclonal antibody induction with either alemtuzumab (Campath-1H, Genzyme, UK), daclizumab (Zenax, Roche Inc, NJ), or basiliximab (Simulect, Novartis Pharma Corp, NJ). Maintenance immunosuppression in the alemtuzumab group consisted of tacrolimus (FK) monotherapy. Patients induced with an IL-2 receptor antagonist (IL-2RA) received FK with the addition of mycophenolate mofetil (MMF). All patients receive 500 mg methylprednisolone perioperatively followed by 1 week of corticosteroids only. Patients on maintenance steroids at the time of transplantation are maintained at their baseline dose posttransplant. All rejection episodes were biopsy proven and classified by Banff 2007 criteria.14 Rejection and TG were diagnosed on either surveillance or indication biopsies.
HLA Typing and Antibody Screening
HLA class I and II type was derived by 1 field, low-resolution DNA typing using polymerase chain reaction with sequence-specific primers. All recipients and donors are routinely typed at the HLA-A, HLA-B, HLA-C, HLA-DRB1/B3/B4/B5, and HLA-DQB1 loci.
Patients are routinely screened for DSA twice in the first week then at 1, 3, 6 and 12 months posttransplant and yearly intervals thereafter. Patients are also tested at times of allograft dysfunction. Exclusion of patients with our stringent definition of preformed DSA (luminex only with a MFI > 500), together with screening during the first week to detect for a memory response, aids confirmation that any DSA developed is a de novo antibody response. Screening uses LABScreen mixed beads (One Lambda, Canoga Park, CA) if the patient is nonsensitized, and they are subsequently or primarily screened using LABScreen single-antigen beads if sensitized. We include DSA to HLA-A,-B,-Cw,-DR, and -DQB1 antigens in our study. A mean fluorescence index greater than 500 by single-antigen beads on 2 separate occasions was taken as positive.
Epitope analysis was carried out on patients mismatched at a single DQB1 allele alone. Epitopes were assigned using the single antigen bead antibody patterns from the first posttransplant sample of each patient which tested positive for de novo DQ DSA. We used the serological DQB1 epitopes as defined by El-Awar et al12 to designate the epitopes in this study. Thus, the antigen specificities of the positive beads of individual patients in our cohort were compared with the defined TerEp patterns and assigned accordingly (TerEp 2001-2015, 2022-2027).11,12 A worked example together with a summary of all the studied DQB1 TerEps, the antigens on which they are carried and the corresponding amino acid sequences, can be found in the supplemental data, SDC (http://links.lww.com/TP/B467). In cases where it was not possible to determine if the antibody pattern was attributed to a single antibody or several, it was assumed that all possible TerEps for a given reactivity pattern were present. We excluded 4 patients, whose antibody pattern did not fit with any of the predefined TerEps, and which may represent unassigned epitopes.
To ensure correct HLA-DQB1 TerEp assignment, it was necessary to identify antibodies directed against HLA-DQA1. The HLA-DQA1 association tool on HLA Matchmaker was used to predict HLA-DQA1 types. Patients with positive beads arising from a suspected HLA-DQA1 antibody were discounted from the analysis. Six patients were excluded on the basis that they had de novo DQA1 HLA antibodies. Given the methods used to assign DQA1 type, it is possible that some DQA1 DSA-positive patients are still included in the cohort.
All analyses were performed using the statistical package Medcalc version 10.4.3 (Medcalc Software, Ostend, Belgium). Comparisons of means and frequencies of normally distributed variables were calculated using t tests and χ2/Fisher exact tests. The Mann-Whitney U test was used for nonparametric variables. Kaplan-Meier survival analysis was used to calculate time of event from transplant, and statistical significance was determined by log rank testing. A P value less than 0.05 was deemed statistically significant.
After a mean follow up of 4.28 ± 2.18 years posttransplant, 198 (19.74%) of 1003 patients had developed a de novo DSA. Eighty-eight (44.4%) 198 of all de novo DSA+ patients developed a DSA against a DQB1 mismatched antigen. Patient demographics are shown in Table 1.
Immunogenicity of the Different HLA Antigens
Using the conventional A, B, DR matching schema, the median HLA mismatch was significantly higher in the de novo DSA+ group at 4 (interquartile range [IQR], 3-5) mismatched HLA antigens compared with 3 (IQR, 2-4) in the de novo DSA− group (P < 0.01). The risk of patients developing a de novo DSA against a mismatched HLA antigen was significantly higher in those patients mismatched at a DQB1 antigen than any other. De novo-DQ DSA-free survival was 78.2% which was inferior to de novo-A DSA-free survival, which was 87.7% (P = 0.0005); de novo-B DSA-free survival, which was 85.0% (P < 0.0001); de novo-Cw DSA-free survival which was 91.0% (P < 0.0001); de novo-DRB1 DSA-free survival which was 90.4% (P < 0.0001); and de novo-DRB345 DSA-free survival which was 88.4% (P = 0.0003) as shown in Figure 1.
Immunogenicity of the Different DQB1 Antigens
Patients mismatched at a DQ7 allele were significantly more likely to develop a de novo DSA compared with mismatches at other DQB1 alleles. De novo DQ7 DSA-free survival was 61.0% which was significantly inferior compared with de novo DQ2 DSA-free survival, which was 88.7% (P = 0.0075); DQ4 DSA-free survival, which was 94.5% (P = 0.007); DQ5 DSA-free survival, which was 87.8% (P = 0.0019); DQ6 DSA-free survival, which was 84.3% (P = 0.0023); DQ8 DSA-free survival, which was 90.6%; and DQ9 DSA-free survival, which was 94.8% (P = 0.0057). Figure 2 shows graphically a comparison of de novo DQ DSA-free survival by DQ mismatch, except for DQ4 where the risk of de novo DQ DSA was no different than those patients who were mismatched at DQ9.
There was no difference in de novo DQ DSA prevalence in patients mismatched at 1 DQB1 allele compared with those patients mismatched at 2 DQB1 alleles. Twenty-two (14.9%) of 148 patients with 2 DQB1 allele mismatches were de novo DSA+ compared with 66/494 [13.4%] patients with a single DQB1 mismatch, P = 0.64. No patient mismatched at 2 DQB1 alleles which included a DQ7 mismatch that developed a non-DQ7 DSA alone. Of the 10 patients mismatched at 2 DQB1 alleles which included a DQ7 mismatch, 7 of them developed a DQ7 DSA alone, whereas 3 developed a DSA to DQ7 in conjunction with a DSA against the other DQB1 mismatched allele. This was not seen in other DQB1 mismatches and again suggests the immunodominance of the DQ7 antigen as shown in Table 2.
Immunogenicity and Pathogenicity of TerEps
Four hundred fifty-two patients who were mismatched at a single DQB1 antigen were used for the epitope analysis. At the time of analysis, 50 of these patients had developed a DQ DSA. The median TerEp mismatch was significantly higher in the DSA+ patients compared with the DSA− patients, as shown in Figure 3. The median TerEp mismatch in the DSA+ group was 4 (IQR, 3-6) compared with 3 (IQR, 2-4) in the DSA− group (P < 0.0001).
Categorizing the single DQB1 mismatch by the number of TerEp mismatches helped predict the risk of de novo DQB1 DSA development, as shown in Figure 4. De novo DQ DSA-free survival overall for those patients mismatched at a single DQB1 antigen was 79.4%. After separating the DQB1 mismatched patients into those mismatched at 3 DQ TerEps or less and 4 DQ TerEps or greater, de novo DQ DSA-free survival was 85.9% and 66.7%, respectively (P < 0.0001; hazard ratio [HR], 2.94; 95% confidence interval [95% CI], 1.67-5.17). In a similar analysis of AMR-free survival, a TerEp mismatch of 4 epitopes or greater also predicted AMR. Overall AMR-free survival in the DQB1 matched and mismatched patients was 98.0% and 86.5%, respectively (P = <0.0001; HR, 5.23; 95% CI, 3.04-8.99). However, separating the DQB1 mismatched patients by epitope mismatch number demonstrated that AMR-free survival with 3 DQ TerEp mismatches or less was significantly less at 89.6% compared with 80.1% in patients with 4 DQ TerEp mismatches or greater (P = 0.026; HR, 1.91; 95% CI, 1.02-3.58). The overall TG-free survival for patients with a single DQB1 mismatch was 90.1%, which was not significantly different than patients matched at DQB1, who had a TG-free survival of 92.8% (P = 0.43; HR, 1.33; 95% CI, 0.67-2.65). However, there was a trend toward a higher risk of TG in patients with 4 DQ TerEp mismatches or greater. TG-free survival was 93.4% in patients with 3 DQ TerEp mismatches or less and 82.9% with 4 DQ TerEp mismatches or greater (P = 0.11; HR, 2.00; 95% CI, 0.80-5.03) as shown in Figure 4. At the end of follow-up, there was no difference in allograft survival between the TerEp groups, with a censored allograft survival of 72.5% in the 3 DQ TerEp patients and 80.6% in the 4 or greater DQ TerEp patients (P = 0.79; HR, 1.09; 95% CI, 0.58-2.04). There was also no difference in censored allograft survival between those patients mismatched at a single DQB1 antigen versus those matched at DQB1, with an allograft survival of 76.0% and 85.6% respectively (P = 0.95; HR, 0.99; 95% CI, 0.63-1.54).
For those patients mismatched at a single DQB1 allele, we then analyzed the individual risk of de novo DQB1 DSA development by individual TerEp mismatch, the results of which are shown in Table 3. Patients mismatched at TerEp epitopes 2003, 2005, 2006, 2013, 2014, 2025, and 2027 appeared to be at higher risk of de novo DQB1 DSA development, compared with the cohort of patients matched at the respective TerEp epitopes. De novo-DQ DSA-free survival was 56.9% in those patients mismatched at TerEp 2003 compared with 81.3% in those matched at TerEp 2003 (P = 0.003; HR, 2.80; 95% CI, 0.96-8.16). In TerEp 2005 mismatched patients, de novo DQ DSA-free survival was 54.2% compared with 86.3% in patients matched at TerEp 2005, (P = 0.0002; HR, 2.60; 95% CI, 1.37-4.93). In TerEp 2006 mismatched patients, de novo DQB1 DSA-free survival was 53.5% in the mismatched group compared with 86.2% in the matched group (P < 0.0001; HR, 3.79; 95% CI, 1.96-7.32). In TerEp 2013 mismatched patients, de novo DQ DSA-free survival was 52.1% in the mismatched group compared with 86.9% in the matched group (P = 0.0022; HR, 2.39; 95% CI, 1.13-5.06). In TerEp 2014 mismatched patients, de novo DQB1 DSA-free survival was 56.2% in the mismatched group compared with 86.0% in the matched group (P < 0.0001; HR, 3.48; 95% CI, 1.83-6.61). In TerEp 2025 mismatched patients, de novo DQB1 DSA-free survival was 70.4% in the mismatched group compared with 81.3% in the matched group (P = 0.0066; HR, 2.13; 95% CI, 1.08-4.20) and finally, in TerEp 2027 mismatched patients, de novo DQB1 DSA-free survival was 73.1% in the mismatched compared with 80.7% in the matched group (P = 0.04; HR, 1.80; 95% CI, 0.93-3.48).
For the final analysis, we compared the allograft outcomes in patients who had developed de novo DSA antibodies directed against TerEp epitopes compared with those patients mismatched at a single DQB1 who did not develop the antibody being examined. No antibodies developed against TerEp epitopes 2002, 2008, 2009, 2023, and 2026. The full results of allograft survival, overall rejection-free survival, and AMR-free survival are shown in Table 4. Allograft survival was statistically inferior in those patients with antibodies directed against TerEp epitopes 2001, 2004, 2006, 2007, 2010, 2011, 2012, 2024, and 2027 compared with patients without the corresponding TerEp antibodies. Overall rejection-free survival was inferior in patients with antibodies directed against all TerEp epitopes except epitopes 2001, 2003, 2014, 2025, and 2027. Although we suspect that these would become significant with larger numbers. AMR-free survival was inferior in all patients with an antibody directed against any TerEp epitope, except for patients with antibody against TerEp 2012, of which only 2 of 8 patients with this antibody pattern developed AMR during the follow up period, giving an AMR-free survival of 68.6% in the TerEp 2012 antibody positive patients compared with 86.8% in patients without the antibody (P = 0.20; HR, 2.45; 95% CI, 0.28-21.31).
This study has validated conclusions of previous studies along with demonstrating several novel findings. We have found that a mismatch at a DQB1 allele is more likely to result in a de novo DSA than any other HLA antigen mismatch and that the DQ7 alleles appear to be the most immunogenic. Like others, we have shown that the epitope mismatch number is proportional to the risk of a corresponding alloimmune response; however, we have incorporated the use of TerEps, which is unique to this study.3 We have also shown that the immunogenicity of the different TerEps are not equal, and to our knowledge, this is the first study which has shown the adverse allograft outcomes associated with TerEp antibodies.
Given the propensity for renal allograft recipients to develop de novo DQ DSA, it has been suggested that DQ matching should be taken into consideration for organ allocation.2-4,6 However, it is known that allorecognition is epitope rather than antigen dependent and that epitopes may be shared between 2 or more HLA antigens.15 As such, numerous studies have looked at the superiority of epitope over HLA antigen matching.3,16-18 In the field of HLA antigens and transplantation, epitope analysis and relevance thereof, has largely been determined by 2 distinct but concordant methodologies. The first, HLA Matchmaker, uses a computer algorithm to estimate the difference of the polymorphic regions, consisting of specific strings of amino acids between donor and recipient HLA antigens.15,17-22 The second method developed by Terasaki and colleagues uses a comparison of amino acids in antibody reactivity patterns to identify epitopes after adsorption and elution of mouse monoclonal antibodies and allosera from recombinant HLA antigens.10-13 More recently, the same group have used the sera of transplant patients with known DQ DSA who had proven AMR to define 6 new DQ epitopes.12 As stated previously, a benefit of using TerEps in clinic studies is that we already know that these epitopes are immunogenic. However, the disadvantage of TerEps is that the list of defined epitopes is not exhaustive, and it is likely that new epitopes will be characterized over time. Neither epitope identification method determines the pathogenicity of the corresponding antibodies, which requires further investigation and clinical correlation which is analyzed in our study.
There have been studies which have shown the importance of epitope matching in preventing an alloimmune response.3,17 Epitope matching has been used in practice as part of Eurotransplant to improve the transplantation rates of sensitized patients by increasing the number of acceptable HLA antigens to include zero mismatches and also mismatches incorporating low numbers of epitope differences.23 Use of the acceptable mismatch program has led to equivocal allograft outcomes in highly sensitized patients compared with the standard “low risk” Eurotransplant cohort.23 Not only can epitope matching increase transplantation rates in sensitized patients, it could potentially reduce alloimmune response prolonging allograft survival and reduce the degree of sensitization after graft failure, making retransplantation more likely.4,9,24
Wiebe and Nickerson9 were the first to show that epitope mismatching may predict the development of class II de novo DSA more accurately than conventional matching. They also suggested that certain epitopes may be more immunogenic. In contrast to our results, they found that mismatches at TerEp epitopes 2004 and 2014 were independent predictors of DQ DSA development.9 It should be noted that our TerEp epitopes were assigned using the antibody reactivity patterns of our patients, and we were able to use the control group as a denominator in establishing immunodominant epitopes. Although we found that TerEp 2014 was the most commonly mismatched epitope, TerEp epitopes 2003, 2005, 2006, 2013, 2014, 2025, and 2027 mismatches were all associated with risk of de novo DSA development. We acknowledge that the epitopes found on a single antigen only may be overrepresented in our cohort, but this will only account for TerEp 2005. It will be important to establish the reason behind the reported differences in immunogenic epitopes, which may reflect differences in patient demographics or immunosuppression. Whatever the reason may be for the reported differences in immunogenicity, it highlights the importance of the need of collaborative work, with larger numbers to obtain more conclusive evidence of which epitopes are immunodominant. However, we do concur that epitope mismatch load is an important predictor of DSA development, and we have shown that certain epitopes appear to be less immunogenic.9,21
The major weaknesses of our study include the lack of DQA1 typing and the use of low-resolution typing alone. The appreciation of the importance of DQA1 has only relatively recently been understood, and therefore, like other centers, we have not routinely DQA1 typed our patients historically.8,25 We sought to remove patients who we believed to have DQA1 antibodies, and although we cannot guarantee within the limits of the data available that some of the remaining cohort still had DQA antibodies, of upmost importance is that these patients still have DQB1 antibodies too, which is the focus of this specific study. The use of low-resolution typing for this study was achievable given the work performed by El-Awar and colleagues, whose work previously defined the serological epitopes we incorporated. Future work will involve prospective studies incorporating the DQA type coupled with high-resolution typing, which will also enable a direct comparison between the clinical impact of epitope mismatches and antibodies determined by serological methods versus computer algorithms.
To conclude, the development of de novo DQ DSA is a potential barrier to improving long-term renal allograft survival, and given the lack of effective treatment for HLA class II de novo DSA, their prevention is vital. Consideration of DQB1 matching in organ allocation schemes may help prevent DSA development. However, this study shows that the immunogenicity of the HLA DQB1 antigens are not equal and determining mismatch at the epitope level may improve risk assessment further, and therefore incorporation of epitope matching into allocation schemes needs to be considered. This study adds to this field of work by demonstrating the clinical impact of epitope mismatch burden determined by TerEps. Furthermore, it also provides evidence of the detrimental clinical outcomes associated with the corresponding TerEp antibodies.
The authors would like to thank the contribution of the clinical transplant staff and the Histocompatibilty and Immunogenetics Laboratory, without whom this study would be possible.
1. Worthington JE, Martin S, Al-Husseini DM, et al. Posttransplantation production of donor HLA-specific antibodies as a predictor of renal transplant outcome. Transplantation
2. Willicombe M, Brookes P, Sergeant R, et al. De novo DQ donor-specific antibodies are associated with a significant risk of antibody-mediated rejection and transplant glomerulopathy. Transplantation
3. Wiebe C, Pochinco D, Blydt-Hansen TD, et al. Class II HLA epitope matching—a strategy to minimize de novo donor-specific antibody development and improve outcomes. Am J Transplant
4. Tagliamacco A, Cioni M, Comoli P, et al. DQ molecules are the principal stimulators of de novo donor-specific antibodies in nonsensitized pediatric recipients receiving a first kidney transplant. Transpl Int
5. Issa N, Cosio FG, Gloor JM, et al. Transplant glomerulopathy: risk and prognosis related to anti-human leukocyte antigen class II antibody levels. Transplantation
6. DeVos JM, Gaber AO, Knight RJ, et al. Donor-specific HLA-DQ antibodies may contribute to poor graft outcome after renal transplantation. Kidney Int
7. Willicombe M, Roufosse C, Brookes P, et al. Antibody-mediated rejection after alemtuzumab induction: incidence, risk factors, and predictors of poor outcome. Transplantation
8. Tambur AR, Leventhal JR, Zitzner JR, et al. The DQ barrier: improving organ allocation equity using HLA-DQ information. Transplantation
9. Wiebe C, Nickerson P. Strategic use of epitope matching to improve outcomes. Transplantation
10. Marrari M, Duquesnoy RJ. Correlations between Terasaki's HLA class II epitopes and HLAMatchmaker-defined eplets on HLA-DR and -DQ antigens. Tissue Antigens
11. El-Awar N, Terasaki PI, Cai J, et al. Epitopes of HLA-A, B, C, DR, DQ, DP and MICA antigens. Clin Transpl
12. El-Awar N, Nguyen A, Almeshari K, et al. HLA class II DQA and DQB epitopes: recognition of the likely binding sites of HLA-DQ alloantibodies eluted from recombinant HLA-DQ single antigen cell lines. Hum Immunol
13. El-Awar N, Lee JH, Tarsitani C, et al. HLA class I epitopes: recognition of binding sites by mAbs or eluted alloantibody confirmed with single recombinant antigens. Hum Immunol
14. Solez K, Colvin RB, Racusen LC, et al. Banff 07 classification of renal allograft pathology: updates and future directions. Am J Transplant
15. Duquesnoy RJ. Humoral alloimmunity in transplantation: relevance of HLA epitope antigenicity and immunogenicity. Front Immunol
16. Tambur AR, Rosati J, Roitberg S, et al. Epitope analysis of HLA-DQ antigens: what does the antibody see? Transplantation
17. Duquesnoy RJ. HLA epitope based matching for transplantation. Transpl Immunol
18. Duquesnoy RJ. HLA matching at the epitope level: the way to go. Clin Transpl
19. Duquesnoy RJ. The antibody response to an HLA mismatch: a model for nonself-self discrimination in relation to HLA epitope immunogenicity. Int J Immunogenet
20. Duquesnoy RJ. Clinical usefulness of HLAMatchmaker in HLA epitope matching for organ transplantation. Curr Opin Immunol
21. Duquesnoy RJ. HLAMatchmaker: a molecularly based algorithm for histocompatibility determination. I. Description of the algorithm. Hum Immunol
22. Duquesnoy RJ. Antibody-reactive epitope determination with HLAMatchmaker and its clinical applications. Tissue Antigens
23. Claas FH, Rahmel A, Doxiadis II. Enhanced kidney allocation to highly sensitized patients by the acceptable mismatch program. Transplantation
24. Kosmoliaptsis V, Gjorgjimajkoska O, Sharples LD, et al. Impact of donor mismatches at individual HLA-A, -B, -C, -DR, and -DQ loci on the development of HLA-specific antibodies in patients listed for repeat renal transplantation. Kidney Int
25. Tambur AR, Leventhal JR, Friedewald JJ, et al. The complexity of human leukocyte antigen (HLA)-DQ antibodies and its effect on virtual crossmatching. Transplantation