Immunologic risk stratification of pediatric heart transplant patients by combining HLA-EMMA and PIRCHE-II.

Human leukocyte antigen (HLA) molecular mismatch is a powerful biomarker of rejection. Few studies have explored its use in assessing rejection risk in heart transplant recipients. We tested the hypothesis that a combination of HLA Epitope Mismatch Algorithm (HLA-EMMA) and Predicted Indirectly Recognizable HLA Epitopes (PIRCHE-II) algorithms can improve risk stratification of pediatric heart transplant recipients. Class I and II HLA genotyping were performed by next-generation sequencing on 274 recipient/donor pairs enrolled in the Clinical Trials in Organ Transplantation in Children (CTOTC). Using high-resolution genotypes, we performed HLA molecular mismatch analysis with HLA-EMMA and PIRCHE-II, and correlated these findings with clinical outcomes. Patients without pre-formed donor specific antibody (DSA) (n=100) were used for correlations with post-transplant DSA and antibody mediated rejection (ABMR). Risk cut-offs were determined for DSA and ABMR using both algorithms. HLA-EMMA cut-offs alone predict the risk of DSA and ABMR; however, if used in combination with PIRCHE-II, the population could be further stratified into low-, intermediate-, and high-risk groups. The combination of HLA-EMMA and PIRCHE-II enables more granular immunological risk stratification. Intermediate-risk cases, like low-risk cases, are at a lower risk of DSA and ABMR. This new way of risk evaluation may facilitate individualized immunosuppression and surveillance.

Second update of the International Registry of HLA Epitopes. I. The HLA-ABC Epitope Database.

The International Registry of HLA Epitopes (http://www.epregistry.com.br) is a website-based resource for HLA epitopes important in transplant rejection and platelet transfusion refractoriness. Its primary goal is to document epitopes that are verified experimentally with specific antibodies. Such epitopes can be defined by single eplets and by eplets paired with certain polymorphic residues within a 15-Å radius, the dimension of the corresponding structural epitope. This report is an update of the HLA-ABC repertoire including descriptions of 72 antibody-verifications of epitopes defined by eplets and/or eplet pairs. The newly updated version 2.0 EpRegistry shows also the polymorphic residue compositions of structural epitopes corresponding to eplets shared between groups of alleles. At present, 151 eplets have not been antibody-verified, and we ranked them with a so-called ElliPro score as a potential predictor of immunogenicity. Sixty eplets with low ElliPro scores might be considered non-epitopes incapable of inducing specific antibodies.

Proteasome Inhibitor Carfilzomib-Based Therapy for Antibody-Mediated Rejection of the Pulmonary Allograft: Use and Short-Term Findings.

We present this observational study of lung transplant recipients (LTR) treated with carfilzomib (CFZ)-based therapy for antibody-mediated rejection (AMR) of the lung. Patients were considered responders to CFZ if complement-1q (C1q)-fixing ability of their immunodominant (ID) donor-specific anti-human leukocyte antibody (DSA) was suppressed after treatment. Treatment consisted of CFZ plus plasma exchange and immunoglobulins. Fourteen LTRs underwent CFZ for 20 ID DSA AMR. Ten (71.4%) of LTRs responded to CFZ. DSA IgG mean fluorescence intensity (MFI) fell from 7664 (IQR 3230-11 874) to 1878 (653-7791) after therapy (p = 0.001) and to 1400 (850-8287) 2 weeks later (p = 0.001). DSA C1q MFI fell from 3596 (IQR 714-14 405) to <30 after therapy (p = 0.01) and <30 2 weeks later (p = 0.02). Forced expiratory volume in 1s ( FEV1 ) fell from mean 2.11 L pre-AMR to 1.92 L at AMR (p = 0.04). FEV1 was unchanged after CFZ (1.91 L) and subsequently rose to a maximum of 2.13 L (p = 0.01). Mean forced expiratory flow during mid forced vital capacity (25-75) (FEF25-75 ) fell from mean 2.5 L pre-AMR to 1.95 L at AMR (p = 0.01). FEF25-75 rose after CFZ to 2.54 L and reached a maximum of 2.91 L (p = 0.01). Responders had less chronic lung allograft dysfunction or progression versus nonresponders (25% vs. 83%, p = 0.04). No deaths occurred within 120 days and 7 patients died post CFZ therapy of allograft failure. Larger prospective interventional studies are needed to further describe the benefit of CFZ-based therapy for pulmonary AMR.

Detection of newly antibody-defined epitopes on HLA class I alleles reacting with antibodies induced during pregnancy.

The determination of HLA mismatch acceptability at the epitope level can be best performed with epitopes that have been verified experimentally with informative antibodies. The website-based International Registry of HLA Epitopes (http://www.epregistry.com.br) has a list of 81 antibody-verified HLA-ABC epitopes but more epitopes need to be added. Pregnancy offers an attractive model to study antibody responses to mismatched HLA epitopes which can be readily determined from the HLA types of child and mother. This report describes a HLAMatchmaker-based analysis of 16 postpregnancy sera tested in single HLA-ABC allele binding assays. Most sera reacted with alleles carrying epitopes that have been antibody-verified, and this study focused on the reactivity of additional alleles that share other epitopes corresponding to eplets and other amino acid residue configurations. This analysis led in the identification of 16 newly antibody-defined epitopes, seven are equivalent to eplets and nine correspond to combinations of eplets in combination with other nearby residue configurations. These epitopes will be added to the repertoire of antibody-verified epitopes in the HLA Epitope Registry.

First report on the antibody verification of MICA epitopes recorded in the HLA epitope registry.

The International Registry of HLA Epitopes (http://epregistry.com.br) has been recently established as a tool to understand antibody responses to HLA mismatches. These epitopes are defined structurally by three-dimensional molecular modelling and amino acid sequence differences between HLA antigens. A major goal was to identify HLA epitopes that have been verified experimentally with informative antibodies. This report addresses the identification of MICA epitopes. Our analysis included published information about MICA antibody reactivity in sera from sensitized patients as well as data from our own laboratories. This report describes twenty-one MICA epitopes verified with antibodies which have primarily been tested in Luminex assays with single alleles. The epitopes correspond to distinct eplets that are often defined by single residues. The Registry is still a work-in-progress and will become a useful resource for HLA professionals interested in histocompatibility testing at the epitope level and investigating antibody responses to HLA mismatches in transplant patients.

First report on the antibody verification of HLA-ABC epitopes recorded in the website-based HLA Epitope Registry.

The International Registry of Antibody-Defined HLA Epitopes ( http://www.epregistry.com.br) has been recently established as a tool to understand humoral responses to human leukocyte antigen (HLA) mismatches. These epitopes are defined structurally by three-dimensional molecular modeling and amino acid sequence differences between HLA antigens. So-called eplets represent essential components of HLA epitopes and they are defined by polymorphic residues. A major goal is to identify HLA epitopes that have been verified experimentally with informative antibodies. Our analysis has also included data in many publications. As of 1 November 2013, 95 HLA-ABC antibody-verified epitopes have been recorded, 62 correspond to eplets and 33 are defined by eplets paired with other residue configurations. The Registry is still a work-in-progress and will become a useful resource for HLA professionals interested in histocompatibility testing at the epitope level and investigating antibody responses to HLA mismatches in transplant patients.

16th IHIW: a website for antibody-defined HLA epitope Registry.

The concept that HLA antibodies are specific for epitopes rather than HLA antigens is important not only for the determination of mismatch acceptability for sensitized patients but also for a better understanding of the antibody response to an HLA mismatch. Numerous publications describe epitope-specific antibodies, but there is no standardized information about the repertoire of clinically relevant HLA epitopes. Under auspices of the 16th IHIW, we have developed a website-based registry of antibody-verified HLA epitopes. Epitope notations are based on HLA molecular modelling of amino acid residues in polymorphic sequence positions. Informative epitope-specific antibodies had been induced by a transplant, transfusion or pregnancy and were monoclonal antibodies or eluates of sera absorbed with single HLA alleles. Antibody reactivity was determined in binding assays with single-allele panels. Antibody producer/immunizer HLA types enhanced the characterization of specific epitopes. The Registry also includes epitopes described in original research publications. Based on the extent of antibody reactivity information, we assigned epitope status as confirmed (well documented) or provisional (more data are needed). At present, the Registry has 69 HLA-ABC, 53 DRB1/3/4/5, 17 DQ, 8 DP and 22 MICA antibody-verified epitopes and will be updated on a quarterly basis. Laboratories worldwide continue to submit data about previously unreported antibody-specific epitopes. For each epitope, the website shows its amino acid composition and HLA alleles that share the epitope. Links show antibody reactivity patterns, sensitization information and references. Other links show molecular modelling of corresponding structural epitopes and polymorphic residue information for epitope-carrying alleles. The website will also have a link to epitope frequency information in different populations. Search functions will list mismatched epitopes on mismatched alleles for selected HLA types. The HLA Epitope Registry will become a valuable resource for researchers interested in HLA compatibility at the epitope level and investigating antibody responses to HLA mismatches.

Correlations between Terasaki's HLA class II epitopes and HLAMatchmaker-defined eplets on HLA-DR and -DQ antigens.

Human leukocyte antigen (HLA) class II-specific antibodies increase the risk of transplant failure, and their characterization must consider epitopes rather than antigens. There are two strategies to determine HLA epitope structure. Terasaki's group has analyzed antibody reactivity patterns with single antigen panels with a computer program based on shared amino acid residues of reactive alleles. HLAMatchmaker is a theoretical algorithm that predicts HLA epitopes on the HLA molecular surface from stereochemical modeling of epitope-paratope interfaces of antigen-antibody complexes. Our epitope repertoire is based on so-called 'eplets' representing 3-A patches of at least one polymorphic residue on the molecular surface. This report describes how 49 of 53 Terasaki's HLA-DR epitopes correspond to HLAMatchmaker-defined eplets. Most of them are equivalent to single eplets (n = 33) or two or more possible eplets (n = 10), but six had corresponding eplet pairs. There were 10 cases whereby eplets have permissible residue combinations, and in 5 cases, we found that eplet specificity might be influenced by nearby hidden residues. We could assign corresponding eplets to 17 of 18 Terasaki's HLA-DQ epitopes. This study demonstrates how the HLAMatchmaker interpretation of amino acid residues shared between antibody-reactive antigens can increase our understanding of the structural basis of HLA epitopes.

Correlations between Terasaki's HLA class I epitopes and HLAMatchmaker-defined eplets on HLA-A, -B and -C antigens.

Although the determination of human leukocyte antigen (HLA) antibody specificity has traditionally been directed toward HLA antigens, there is now increasing attention to structurally defined HLA epitopes. An understanding of the HLA epitope repertoire is important to acceptable mismatching for sensitized patients and to a new epitope-based matching algorithm aimed to reduce antibody-mediated rejection. There are two strategies to determine the HLA epitope repertoire. Terasaki's group has used an empirical method to analyze the reactivity of single allele Luminex panels with mouse monoclonal antibodies (mAbs) and absorbed/eluted alloantibodies with a computer program based on shared residues in the amino acid sequences of reactive alleles. HLAMatchmaker is a theoretical algorithm that predicts HLA epitopes on the HLA molecular surface from stereochemical modeling of epitope-paratope interfaces of antigen-antibody complexes. Our epitope repertoire is based on so-called 'eplets' representing 3-A patches of at least one polymorphic residue on the molecular surface. A comparative analysis has shown that 81/103 Terasaki's HLA class I epitopes are equivalent to individual eplets (n = 50) or pairs of eplets (n = 31) separated far enough to serve as potential contact sites for two complementarity-determining regions of antibody. An additional 12 Terasaki's epitopes (TerEps) correspond to eplets with permissible residue combinations that do not seem to affect epitope specificity. We could not identify corresponding eplets for the remaining 10 TerEps, including 8 that might be considered xeno-epitopes defined by mouse mAbs. Conversely, HLAMatchmaker has 38 additional eplets in well-exposed surface positions that do not have equivalent TerEps, and for many of them, we have found specific antibodies. These findings strengthen the concept that eplets are essential basic units of HLA epitopes and that they provide a better understanding of HLA immunogenicity (i.e. ability to induce an antibody response) and antigenicity (i.e. reactivity with specific antibody).

Molecular determinants of the physicochemical properties of a critical prion protein region comprising residues 106-126.

Prion diseases are marked by the cerebral accumulation of conformationally modified forms of the cellular prion protein (PrP(C)), known as PrP(res). The region comprising the residues 106-126 of human PrP seems to have a key role in this conformational conversion, because a synthetic peptide homologous with this sequence (PrP106-126) adopts different secondary structures in different environments. To investigate the molecular determinants of the physicochemical characteristics of PrP106-126, we synthesized a series of analogues including PrP106-126 H(D), PrP106-126 A and PrP106-126 K, with l-His-->d-His, His-->Ala and His-->Lys substitutions respectively at position 111, PrP106-126 NH(2) with amidation of the C-terminus, PrP106-126 V with an Ala-->Val substition at position 117, and PrP106-126 VNH(2) with an Ala-->Val substitution at position 117 and amidation of the C-terminus. The analysis of the secondary structure and aggregation properties of PrP106-126 and its analogues showed the following. (1) His(111) is central to the conformational changes of PrP peptides. (2) Amidation of the C-terminal Gly(126) yields a predominantly random coil structure, abolishes the molecular polymorphism and decreases the propensity of PrP106-126 to generate amyloid fibrils. (3) PrP106-126 V, carrying an Ala-->Val substitution at position 117, does not demonstrate a fibrillogenic ability superior to that of PrP106-126. However, the presence of Val at position 117 increases the aggregation properties of the amidated peptide. (4) Amyloid fibrils are not required for neurotoxicity because the effects of PrP106-126 NH(2) on primary neuronal cultures were similar to those of the wild-type sequence. Conversely, astroglial proliferation is related to the presence of amyloid fibrils, suggesting that astrogliosis in prion encephalopathies without amyloid deposits is a mediated effect rather than a direct effect of disease-specific PrP isoforms.

Determination of HLA-A,B residue mismatch acceptability for kidneys transplanted into highly sensitized patients: a report of a collaborative study conducted during the 12th International Histocompatibility Workshop.

BACKGROUND: During the 12th International Histocompatibility Workshop, a collaborative study between 35 laboratories was conducted on a group of highly allosensitized patients who had received a kidney transplant from 1981 to 1995. The major goal of the study was to assess how serum screening against a large cell panel could determine donor HLA mismatch acceptability in relation to graft outcome. METHODS: Twenty laboratories participated in an extensive screening of 92 high panel-reactive antibody (PRA) sera from patients at 29 transplant centers worldwide; each patient had received a kidney allograft from an HLA-A,B mismatched unrelated donor. Screening was done by complement-dependent lymphocytotoxicity and antihuman globulin augmentation techniques using a common protocol and shared standardized reagents. After an extensive quality-control assessment, we selected data from 14 participants who had screened the sera against a combined panel of 535 HLA-typed cells. RESULTS: With the 2x2 table-based Multiscreen computer program, we could readily determine for virtually every patient the significant correlations between serum reactivity and the presence of panel cell markers, including private and public HLA-A,B epitopes and amino acid residues assigned from published sequencing data. Donor mismatch acceptability was assessed at the amino acid residue level. In the complement-dependent lymphocytotoxicity (n=49; PRA=84.1+/-12.1%) and antihuman globulin (n=60; PRA=92.5+/-5.8%) groups, the 3-month graft survivals were 28% and 30% lower for unacceptable residue mismatches. CONCLUSIONS: These studies underscore the importance of a comprehensive serum screening analysis in the selection of appropriately mismatched donors for highly sensitized transplant patients.

Multicenter evaluation of the flow cytometry T-cell crossmatch: results from the American Society of Histocompatibility and Immunogenetics-College of American Pathologists proficiency testing program.

BACKGROUND: The performance characteristics and interlaboratory comparisons of the T-cell flow cytometry crossmatch remain largely unknown. METHODS: This study was performed using data from the ASHI-CAP proficiency testing program. Four unknown sera and two unknown cells were sent to participating laboratories twice a year for 4 years. RESULTS: In one survey in which different crossmatch techniques were compared, flow cytometry was slightly more sensitive than the antiglobulin method and considerably more sensitive than direct cytotoxicity. However, the proportion of participants in any given survey detecting antibodies in all sera expected to be positive was 50-60% and has not changed over the years. Failure to detect antibodies correlated with low antibody concentration, diluting the unknown serum by the testing laboratory, and with the instrument used. False positive results with normal sera were infrequent. Fluorescence intensity values were not standardized and were highly variable, but when fluorescence units reported by individual laboratories were divided by their own positive-negative cutoff values, results from different centers were more comparable. In general, fluorescence-to-cutoff ratios >5 correlated with complement binding activity, whereas values <5 denoted concentrations below those required to fix complement. CONCLUSIONS: Flow cytometry, as used by most centers, is highly sensitive and allows relative antibody quantitation. Furthermore, the data define objective parameters that may help to standardize the test and improve its predictive value in clinical transplantation.

Progress report on the ASHI/CAP Proficiency Survey Program in Histocompatibility Testing. I. HLA-A,B,C typing, antibody screening, and lymphocytotoxicity crossmatching. American Society for Histocompatibility and Immunogenetics. College of American Pathologists.

The Histocompatibility Survey Program was organized in 1982 as a joint project by the ASHI and CAP to evaluate laboratory performance in HLA typing, lymphocytotoxicity crossmatching, and antibody analysis. This report summarizes the experience with the HS surveys on HLA class I serology. During a 12-year period, the number of participating laboratories increased from 150 to 285 and HLA typing was done with 90 survey specimens representing 20 HLA-A and 35 HLA-B antigens. Most unsplit antigens were correctly identified in more than 90% of the laboratories. For many antigens, a high percentage of participants reported a split and there was generally a high consensus of a correct assignment. Nevertheless, several antigens were difficult to define, as shown by low consensus rates. During recent years, the assignments of Bw4/6 and HLA-C antigens have significantly improved. Lymphocytotoxicity crossmatching was analyzed for 138 cell-serum combinations tested by an average of 143 laboratories. Comparisons between four techniques (basic NIH, Amos modified, LI, and AHG) showed consistent results (greater than 90% crossmatch compatibility or incompatibility) for 71% of the cell-serum combinations. The crossmatch results with the remaining combinations were more variable for one or more of the crossmatch techniques. Serum antibody identification showed a continued improvement during recent years, and the average consensus for assigning acceptable antibody specificity reached 88%. A performance grading system based on a 90% consensus rate among participants is used to satisfy requirements for laboratory accreditation.

Progress report on the ASHI/CAP Proficiency Survey Program in Histocompatibility Testing. II. HLA-DR, DQ serologic typing, antibody identification, and B-cell crossmatching. American Society for Histocompatibility of Immunogenetics. College of American Pathologists.

This report summarizes the 8-year experience of the DR survey program designed to evaluate the performance of histocompatibility laboratories in the serologic typing of cell specimens for HLA-DR and HLA-DQ polymorphisms and the HLA class II antibody identification and B-cell crossmatching of serum specimens. The number of participants increased from 45 in 1985 to 214 in 1992. Although the performance criteria are based on laboratory consensus, the availability of DNA typing since 1990 has enabled a critical assessment of the reliability of serologic HLA-DR, DQ typing. The survey results shows that unsplit HLA class II antigens DR1-DR8, DR52/53, and DQ1-3 are generally correctly identified in over 90% of the participating laboratories. DR9 and DR10 have not yet been tested and testing for DQ4 has not yet achieved this level on consensus. In contrast, the assignments of serologic subtypes of HLA-DR and HLA-DQ are less consistent and frequently unreliable. Although the B-cell crossmatches show generally high laboratory consensus rates, the serum screening results show frequently inconsistent results regarding HLA class-II-specific antibody identification.

Intravenous gamma globulin decreases platelet-associated IgG and improves transfusion responses in platelet refractory states.

In an attempt to improve platelet transfusion responses, intravenous immunoglobulin (IV-IgG) was administered to 19 patients who were refractory to random and best available HLA-matched platelets. A response to IV-IgG was defined as two or more successive transfusions of HLA-matched products that provided recoveries greater than 30%. Thirteen of 19 (68%) patients responded to therapy at a median time of 7 days after initiation of IV-IgG (range = 2-17). Baseline platelet associated IgG levels (PalgG) were elevated in both the responders (61.6 +/- 76.2) (mean +/- SD) and the non-responders (47.0 +/- 46.3 fg/plt). Post-therapy, PalgG levels remained unchanged in the nonresponders but were decreased significantly (p = 0.05) to 11.1 +/- 6.2 fg/plt in the responders. The latter levels were similar to those (11.6 +/- 8.2 fg/plt) measured in a series of 36 transfusion responsive patients. This apparent decline in PalgG was not explained by differences in lymphocytotoxic antibodies (LCT-Ab) after therapy. Moreover, a high degree of alloimmunization was associated with a poorer response to IgG. Only two of eight patients with LCT panel-reactive antibody (PRA) of greater than 85% were responders. By contrast, improved transfusion outcomes were seen uniformly in patients with PRA greater than or equal to 85%. Improved recoveries were obtained using LCT-Ab compatible but not incompatible platelets. The median increment (% predicted) with compatible platelets before therapy was 6.0 +/- 9.9 (SD). Post-IgG, median recoveries were 37.0 +/- 31.2 percent, P less than 0.001. These findings suggest that IV-IgG may alter destructive mechanisms that affect the survival of compatible platelets in refractory patients.

A1, Cw7, B8, DR3 HLA antigen combination associated with rapid decline of T-helper lymphocytes in HIV-1 infection. A report from the Multicenter AIDS Cohort Study.

108 seropositive homosexual men were examined for associations between HLA phenotype and progression of human immunodeficiency virus type 1 (HIV-1) infection. Among men of predominantly European ethnic origin, 49 with very rapid 2-year declines in CD4+ lymphocyte counts showed significant differences in antigen frequencies from 59 men matched for ethnic background, study centre, and initial CD4+ cell count but with little or no decline in CD4+ cells. Relations of varying strength (odds ratios 6.1-10.3) were seen with several HLA antigens often linked in the A1-Cw7-B8-DR3 haplotype. The strongest relation was with the A1, Cw7, B8 combination (odds ratio 10.3). Associations between these antigen combinations and development of AIDS were weaker. The frequency of HLA A24 was also significantly higher in rapid than in slow decliners (odds ratio 4.3). These findings strengthen the suggested link between the product of a gene in the A1-Cw7-B8-DR3 haplotype and HIV-1-related disease.

Progress report on the ASHI/CAP Histocompatibility Survey Program, 1981-1986.

Two histocompatibility testing surveys were conducted over a five-year period by the American Society for Histocompatibility and Immunogenetics and the College of American Pathologists. More than 200 laboratories participated in the HS survey, which consisted of four shipments of cells and serum samples to be tested for ABO type, HLA-A and -B types, antibody identification, and lymphocytotoxicity crossmatching. About 100 laboratories participated in the DR survey, which consisted of four annual shipments to be tested for HLA-DR and -DQ types, antibody identification, and crossmatching. The results of the analysis of the HLA-typing results showed high consistency (generally greater than 90%) among laboratories in the definition of most recognized HLA antigens. In contrast, the HLA antibody screening was generally less consistent between laboratories. On the basis of 90% or greater consensus among participants, it was possible to develop a performance grading system.

Alloreactive T-cell recognition of two DQ beta allelic specificities associated with DQw1.

Two sets of alloreactive T-cell clones from different original mixed lymphocyte reactions (MLR) discriminate between two DQ beta allelic forms when these are expressed in association with the same DQ alpha form. From these results we concluded that it is possible to guide a DQ specific response of alloreactive T-cell clones by setting up an appropriate responder-stimulator combination in the original MLR; DQw1 is associated with two allelic forms of the DQ molecule recognized as lymphocyte activating determinants by alloreactive T-cell clones; alloreactive T cells recognize specific alpha-beta chain combination of the DQ molecule; the allelic forms that differentiate the alpha or the beta chains of the DQ molecule from each other can also be recognized at the molecular level by DQ alpha and DQ beta RFLP analysis.

Relationship between DQ alpha and DQ beta RFLP and cell surface polymorphisms of class II HLA antigens.

DQ alpha and beta DNA probes of the human major histocompatibility complex (MHC) were hybridized to restriction enzyme-digested genomic DNA with the aim of establishing a correspondence between the polymorphisms recognized by classical serology and DNA restriction fragment length polymorphisms (RFLP). In DR homozygous human cell lines, three distinct PstI fragments were recognized by the DQ alpha probe and four PstI fragments were recognized by the DQ beta probe. Each fragment was associated with a different group of DR antigens. Three allelic forms of either DX alpha or beta genes were identified, but none showed any strong association with DR or DQ. Family segregation analysis at the DNA level further confirmed the DR linkage of the DQ alleles in estimations of gene frequencies of different alleles of DQ alpha, DQ beta, DX alpha and DX beta. Evidence was presented that the DQ alpha and DQ beta allelic forms described at the DNA level correspond to polymorphic determinants at the cell surface which can be defined serologically or in cellular assays. Our data suggest that the HLA-DQ subregion-encoded alloantigens should be defined at the individual alpha and beta chain levels.