The Alpha-beta Soup of HLA-DQ: A Recipe for Alloimmune Perspectives

A landmark work by Wiebe et al1 demonstrated that clinical and histopathological changes leading to antibody-mediated rejection (AMR) in kidney transplant recipients are strongly associated with the formation of de novo donor-specific HLA antibodies. This work has revolutionized posttransplant patient care, leading to the implementation of DSA monitoring as a routine procedure, with the hope to minimize the damage caused by these antibodies. The introduction of the single antigen bead assay and its improved characterization of antibody specificity allowed for meticulous and systematic antibody evaluation. Multiple studies implicated HLA class II antibodies, and specifically HLA-DQ de novo donor-specific HLA antibodies, as the leading cause of AMR and allograft loss. Yet, it is my perception that the precision of HLA-DQ antibody analysis was tempered by well-meaning but poorly informed information technology (IT) experts that designed the single antigen bead analysis software. At first glance, the IT approach seemed quite plausible. For HLA class I antigens (HLA-A, -B, -C), 1 genetic locus was given per each antigen, with antibody responses designated as 1 antibody per each locus. For HLA-DR, the vendor provided information only for the DRB1 locus, and none for DRA1 (each class II molecule is encoded by 2 different genes: DRA1 and DRB1; DQA1 and DQB1; DPA1 and DPB1). Immunologically speaking, that stands to reason, since the DRA1 gene product, or DRα chain, is not polymorphic. The IT experts therefore proceeded to designate antibody responses as 1 antibody per 1 locus. Unfortunately, they carried over this approach also for analysis of antibodies against HLA-DQ and HLA-DP. This is where immunology, rather than IT, knowledge should have come to play. In difference with the DRA1 gene, DQA1 and DPA1 genetic loci are polymorphic. Moreover, since they form half of the DQ (or DP) antigenic complex, they are critical for determining the specific identity of the DQ (or DP) antigens. Indeed, the vendor included the identity of the DQA1 and DPA1 genes, together with their DQB1 and DPB1 counterparts, encoding the molecules attached to the relevant beads. The IT experts, however, were oblivious to the significance of this pairing. Using the logic of 1 genetic locus associated with 1 target for 1 antibody, the IT experts proceeded to design the analysis software for HLA-DQ and HLA-DP designating 1 antibody to the DQA1 locus and 1 to the DQB1 locus (similarly for HLA-DP). Unfortunately, both Terasaki’s (Deng et al2) and Duquesnoy’s3 groups adopted the IT approach as they described their DQA1 and DQB1 “epitopes.” Indeed, antibodies may recognize a target that is marshaled by different regions of the HLA molecule; however, neither the product encoded by the DQA1 gene, nor the 1 encoded by the DQB1 gene, can function as an HLA antigen independently. They must pair together to form a heterodimer (termed DQαβ or DPαβ) to be viewed by the immune system as a functional molecule. HLA class II antigens are not the sole example where molecules encoded by several genes must form a final structure before obtaining a functional capacity. The most befitting example is an antibody. The heavy chain and the light chain of an antibody are each encoded by a different gene. In normal, physiologic, conditions, an antibody light chain dose not function on its own, unless it is paired with a heavy chain, and vice versa. Of note, HLA-DQ molecules have been a focus of interest in the study of several autoimmune diseases. Studies in this field clearly address HLA-DQ as a heterodimer composed of the relevant DQα and DQβ chains and have a naming convention to characterize this.4 The lack of appreciation of the physiology and functionality of HLA class II molecules, led to an arbitrary separation between DQA1 and DQB1 for conducting HLA-DQ (and HLA-DP) antibody analysis. Some investigators have highlighted this point earlier, cautioning about potential ramifications.5,6 Alas, our perception of the role of HLA-DQ in transplant immunology is still deficient. It is with this perspective that we should find the recent work from Ursule-Dufait et al,7 published in this issue of Transplantation, of great interest. The authors performed a retrospective analysis of a large consecutive cohort of kidney transplant recipient from 3 transplant centers in Paris. Their goal was to evaluate the role of preformed HLA-DQα DSA. Because of the retrospective nature of the study, patients with potential HLA-DQα DSA were presumed to be included within a subcohort that was prospectively identified, at the time of transplant evaluation, as having HLA-DQβ DSA. Of significance, 202 patients were originally categorized as having preformed DQ DSA based on DQβ-only analysis. However, when DQα information was added for the purpose of this study, it was demonstrated that in 80 patients (39.6%!) the antibodies recognized a different, non-DSA, HLA-DQ target. This means that in more than one third of the cases, the clinicians presumed they were transplanting across preformed DSA, treated the patients as such, but those antibodies were not relevant to the transplant. This highlights the knowledge gap in current practice for analyzing HLA-DQ antibodies in transplant centers that do not consider the complete DQαβ heterodimer in their analyses. While this may explain some of the negative complement-dependent cytotoxicity crossmatch results reported by the investigators (mean fluorescence intensity [interquartile range, 677–2896] of the presumed DSA), the authors clearly demonstrated a strong association between the preformed antibodies and higher risk for AMR. The negative impact of these antibodies was observed already at the 3-mo protocol biopsy, shown as high microvascular inflammation scores. A negative role for preformed HLA DSA was previously demonstrated by Senev et al.8 The focus of that study was to compare the impact of preformed antibodies that either resoled or persisted posttransplant. In their study, both forms of antibodies were associated with high incidence of histological picture of AMR (54%–59% of these patients). While not the focus of that study, the authors clearly demonstrated that most of the antibodies that persisted were directed at HLA-DQαβ DSA. Taken together, both studies send a strong message for the need to consider potential ramifications of memory responses when evaluating risk for transplant outcome. This convincingly reiterates the recommendation from the STAR (Sensitization in Transplantation: Assessment of Risk) working group.9 Ursule-Dufait et al7 should be congratulated for using specifically information from the bead representing the donor DQαβ heterodimer in their (retrospective) analysis. This practice is not followed by many laboratories and as the authors have clearly demonstrated here, may lead to misinterpretation of HLA-DQ antibody information. It may therefore be prudent to assess whether prior analysis schemes may have affected available literature and may have led to underappreciation of the role of HLA-DQ in solid organ transplantation. One example might be analyses derived from large historic datasets such as the Scientific Registry of Transplant Recipients that recorded DQ information only at the serological (DQβ equivalent) level. Overall, this work sends a clear message to the transplant community. There is a need to incorporate evaluation of HLA-DQα as part of the DQαβ heterodimer. It should be an integral part of any pretransplant as well as in posttransplant antibody evaluation. It may help in guiding patient management, and it may inform of the need to add HLA-DQ into organ allocation algorithms.10