Human leukocyte antigen (HLA) DR15 is associated with reduced incidence of acute GVHD in HLA-matched allogeneic transplantation but does not impact chronic GVHD incidence

Antigen-specific T-cell responses play an important role in the pathophysiology of autoimmune cytopenias.^1,2  Oligoclonal T-cell populations correlating with clinical course have been identified in patients with aplastic anemia and myelodysplastic syndromes (MDSs).^3-5  Monoclonal T-cell populations are responsible for the cytopenias seen in proliferations of T-large granular lymphocytes (T-LGLs), a related autoimmune cytopenic state. The DR15 allele (serologic DR2) at the HLA DRB1 locus has been found to be a marker of disease susceptibility and clinical response to immunosuppressive therapy (IST) in autoimmune cytopenias.^6-9  In fact, DR15 is the strongest predictor of response to IST in MDS.¹⁰  Similarly, HLA DR4 is a marker for response to IST in T-LGLs.¹¹  Given the specificity and the HLA-restricted nature of T-cell responses, these results have led to the suggestion that there may be an immunodominant, DR15-restricted myeloid epitope driving T-cell proliferation in bone marrow failure syndromes. Extending this hypothesis, immune responses to DR15-binding epitopes from DRS-1, a candidate autoantigen in aplastic anemia, have been identified in patients with aplastic anemia.¹²  The alternative hypothesis is that DR15 could be a marker linked to generalized susceptibility to immunosuppression either through linkage disequilibrium to an immunogenetic profile or by differential binding affinity to the CD4 coreceptor.

Analogous to bone marrow failure syndromes, graft-versus-marrow and graft-versus-leukemia (GVL) effects in allogeneic blood or marrow transplantation (BMT) are mediated by donor T lymphocytes. Furthermore, GVL effects, as evidenced by progression-free survival and by likelihood of response to donor lymphocyte infusions (DLIs), are more pronounced in myeloid disorders such as chronic myeloid leukemia (CML), acute myeloid leukemia (AML), or MDSs than in the lymphoid malignancies, such as acute lymphoblastic leukemia (ALL). However, in contrast to bone marrow failure states, there is no known association between allogeneic stem cell transplantation outcomes and specific HLA antigens. If DR15 is associated with myeloid antigen presentation, then this would be expected to increase GVL; on the other hand, if DR15 represents an immune profile associated with susceptibility to IST, then the presence of DR15 would be expected to reduce the severity of graft-versus-host disease (GVHD). Thus, we analyzed myeloid malignancies (CML, AML, MDS) and excluded lymphoid malignancies to maximize the likelihood of detecting an immunodominant myeloid antigen effect. We present the relationship between BMT outcomes and the presence or absence of the HLA DR15 antigen.

This is a retrospective cohort study of 167 consecutive patients undergoing first allogeneic BMT for myeloid malignancies and disorders (AML, CML, and MDS) between January 1, 1991, and April 30, 2005, at Roswell Park Cancer Institute, Buffalo, NY. Patient characteristics included the following: AML (n = 84), CML (n = 63), and MDS (n = 20); median age, 42 years (range, 11-66 years); 104 male, 63 female; total body irradiation (TBI)–based conditioning regimens (n = 131); VPCyTBI (etoposide, cytoxan, TBI; n = 84), BuTBI (busulfan, TBI; n = 27), BuCy (busulfan, cytoxan; n = 22), CyTBI (cytoxan, TBI; n = 13), or other (n = 21). All patients received unmanipulated T-cell–replete HLA-matched (at A, B, and DR) bone marrow (n = 147) or peripheral blood (n = 20) grafts and standard immunosuppression prophylaxis. We excluded HLA-mismatched transplantations to avoid obscuring a relationship between HLA DR15 and transplantation outcomes by the powerful alloimmune response generated and the use of higher doses of IST in HLA-mismatched transplantations. Second or greater transplantations were excluded, as were patients with ALL and other diagnoses. Institutional review board approval was obtained at Roswell Park Cancer Institute to conduct this review of existing data and all individual patient data have been deidentified. Patients provided informed consent, in accordance with the Declaration of Helsinki.

HLA typing was performed using standard clinical laboratory techniques with serologic or molecular testing. Patients with serologic DR2 were included in the DR15-negative group. This was purposeful to reduce the risk of a chance finding.

The diagnosis and grading of acute and chronic GVHD were by the Seattle criteria.¹³ 

Statistical analysis was designed a priori and based on the null hypothesis that the presence or absence of HLA DR15 was not associated with transplantation outcomes. The frequency of HLA DR15 in the patient population was tested for Hardy-Weinberg equilibrium. Univariate analysis used the Pearson chi-square, Fisher exact test, or analysis of variance (ANOVA) where appropriate. Survival curves were generated according to Kaplan and Meier and were compared using the log-rank method.¹⁴  Survival analyses were stratified by patient and transplantation characteristics. Overall survival (OS) was calculated from the date of stem cell infusion (day 0) until death due to any cause. Progression-free survival (PFS) included all patients and was calculated from day 0 until the first sign of disease progression (if disease was present at the time of transplantation) or relapse after complete remission (CR). Patients were censored at the time of last follow-up or death if in continuous CR after transplantation. All P values are 2-sided, with P below .05 considered significant. Multivariate logistic regression analysis for the factors associated with acute GVHD was performed with forward step-wise selection conditional on the Wald score, with P below. 10 to enter and P above .10 to remove. Factors considered in the model were as follows: donor relation (related vs unrelated), DR15 status (positive vs negative), age, sex match (matched vs mismatched), diagnosis (AML vs CML vs MDS), GVHD prophylaxis (CsA or FK + Mtx vs CsA or FK + Mtx + Mp vs other), Karnofsky performance status (≤ 80 vs > 80), disease status (CR vs other), and conditioning regimen (TBI vs non-TBI). All statistical analyses used SPSS, v 11.0 (Chicago, IL).

Patient characteristics are summarized in Table 1. Ninety-one percent of the patient population was white. HLA DR15–positive and –negative groups were well matched for age, sex, performance status, diagnosis, disease status, conditioning regimen, the use of TBI, and relation to the donor. DR15-positive subjects were more likely to have received only double-agent GVHD prophylaxis (calcineurin inhibitor plus methotrexate), whereas DR15-negative subjects were more likely to have received triple-agent prophylaxis (calcineurin inhibitor plus methotrexate plus steroids; P = .03). Of the patients who had HLA DR15 at the DRB1 locus, a significantly greater proportion was designated by molecular typing than by serologic typing.

One-hundred and fifty-two subjects (91%) were non-Hispanic whites. Thirty-five patients carried one or both alleles of DR15 at the HLA DRB1 locus, an incidence of 21%, which is not significantly different from the DR15 incidence of the North-American white population (17%-25%).¹⁵  Allelic frequencies demonstrated Hardy-Weinberg equilibrium.

Acute GVHD grades II to IV was seen in 23% of patients with HLA DR15 and in 42% of patients with other HLA DRB1 alleles (P value .041 by Pearson chi-square) (Table 2).

There was no significant difference in the 2 groups in chronic GVHD incidence or severity, disease status, mortality rate, primary cause of death, or duration of median follow-up. Given that most patients who had HLA DR15 were identified by molecular rather than serologic typing, we performed a sensitivity analysis, by excluding those patients that were typed by serologic methods, which resulted in similar findings (data not shown). Multivariate analysis identified unrelated donor grafts (relative risk, 2.6; 95% CI, 1.3-5.3) and the absence of DR15 (relative risk, 2.4; 95% CI, 1.0-5.8) as the only 2 factors significantly associated with acute GVHD (Table 3).

There was no significant difference in OS or PFS between the DR15-positive and -negative groups. At a median follow-up of 58 months (4.8 years), the 5-year OSs for subjects with and without DR15 were 40% and 40%, respectively (P = .6); the 5-year PFSs for subjects with and without DR15 were 59% and 77%, respectively (P = .4). A trend toward late relapses after 3 years was observed in patients with HLA DR15. Sensitivity analysis to determine the impact of serologic versus molecular typing on OS and PFS yielded no difference (data not shown).

We found that acute GVHD grades II to IV was significantly reduced in subjects with HLA DR15. In addition, we found no difference in the incidence of chronic GVHD according to DR15 allele, consistent with different mechanisms for the development of acute and chronic GVHD. The decreased acute GVHD incidence in subjects with DR15 did translate into a lower, but not significantly different, PFS at 5 years in myeloid malignancies without an apparent impact on OS.

Several factors are known to affect the incidence of severe acute GVHD: donor-recipient antigenic disparity (HLA and non-HLA), the quantity and quality of donor T cells, the GVHD prophylaxis regimen, and conditioning regimen intensity. In this study, we controlled for HLA disparity by confining the analysis to HLA-matched pairs. All patients received unmanipulated T-cell–replete grafts with standard immunosuppressive prophylaxis. There were no significant differences in baseline characteristics between HLA DR15–positive or –negative subjects. In fact, DR15-positive subjects were more likely to have received only double-agent prophylaxis (calcineurin inhibitor plus methotrexate), whereas subjects without DR15 were more likely to have received triple-agent prophylaxis (calcineurin inhibitor plus methotrexate plus steroids). We show that the incidence of severe acute GVHD (grades II-IV) is nearly halved in patients with HLA DR15. This is the first reported association between a specific HLA antigen and acute GVHD. Despite the difference in acute GVHD, this study did not find statistically significant differences in the incidence or severity of chronic GVHD, either because chronic GVHD has a different pathogenesis or due to a lack of statistical power. Of interest, the reduction in acute GVHD did not result in a proportionate reduction in PFS. This may be due to the similar incidence of chronic GVHD in both DR15-positive and -negative populations. We found that the presence of HLA DR15 confers no significant benefit on overall or progression-free survival in myeloid malignancies. Moreover, there was a tendency toward late relapses in patients with HLA DR15, but this was not statistically significant. This finding disputes the hypothesis that immunodominant myeloid antigens are preferentially presented by HLA DR15. However, the study was insufficiently powered to look at survival end points and effectively test this particular hypothesis. GVHD (as opposed to GVL) is presumably not related to presentation of a myeloid-related epitope, but epitopes from other tissues. The decrease in acute GVHD could conceivably be related to a differential ability of host DR15-positive antigen-presenting cells (APCs) to effectively present nonmyeloid antigens as opposed to myeloid antigens.

The exact antigen-independent mechanism by which the presence of HLA DR15 protects against acute GVHD is unknown. DR15 could mediate immune dysregulation by linkage disequilibrium to other relevant genes. Several immune system genes, such as TNF-alpha, lie in close proximity to the HLA complex and are coinherited. This is supported by several studies showing that TNF-alpha promoter polymorphisms associated with up-regulation in TNF-alpha secretion are associated with organ failure and GVHD.^16,17  Reduced binding affinity of HLA DR15 to the CD4 coreceptor may be yet another possible explanation for the inherent susceptibility to IST. Peptide-MHC binding to the CD4 coreceptor may be responsible for amplification of T-cell signaling through the T-cell receptor.^18,19  However, this remains to be shown for HLA DR15. Further supporting immune dysregulation in HLA DR15 subjects is the finding that T cells derived from psoriatic subjects who have HLA DR15 have a reduced ability to produce IFN-gamma on stimulation with group A streptococcus cell wall proteins.²⁰  Like other HLA class II alleles, the presence or absence of HLA DR15 has been implicated in several autoimmune states.^21-33  The precise relevance of these associations is unknown in the context of BMT.

Generalization of the study is limited by the exclusion of HLA-mismatched pairs, absence of molecular typing for all patients, poor representation of peripheral blood stem cell transplantations, and a single-center study. The significance of HLA DR15 in HLA-mismatched transplantations is currently unknown. A large number of these transplantations were done at a time when molecular HLA typing was not available. Most transplantations in this analysis used bone marrow as the source of stem cells and the impact of HLA DR15 on peripheral blood stem cell transplantations, which have a greater likelihood of GVHD due to a higher T-cell dose, remains to be determined. Most subjects underwent standard myeloablative conditioning, and the impact of HLA DR15, if any, in the nonmyeloablative setting is also unknown. The smaller number of patients and the shorter median follow-up in the DR15-positive group reduces statistical power, thereby potentially obscuring the finding of a statistically significant difference in OS or PFS. Validation in a larger cohort of patients and in nonmyeloid malignancies is warranted and is being pursued. It would be premature to apply these results to predicting GVHD severity until further validation is complete.

In conclusion, we found that the presence of HLA DR15 in HLA-matched BMT for myeloid malignancies is associated with a decreased incidence of acute GVHD without appreciably altering survival. This supports a role for HLA DR15 in immune dysregulation in the BMT setting. If these results are confirmed, this would be the first association between a specific HLA antigen and GVHD development, thus suggesting the need for studying GVHD prophylaxis modification based on specific HLA antigens.

We gratefully acknowledge Dr Yasmin Thanavala and Dr Victor H. Englehard for reviewing the paper and for helpful discussion.