Unrelated donor transplants DPB1 disparities contribute to severe GVHD and reduced patient survival after unrelated donor bone marrow transplantation

Transcription

1 (2002) 30, Nature Publishing Group All rights reserved /02 $ Unrelated donor transplants DPB1 disparities contribute to severe GVHD and reduced patient survival after unrelated donor bone marrow transplantation P Loiseau 1,3, H Espérou 2, M Busson 3, R Sghiri 1, R Tamouza 1,3, M Hilarius 1, C Raffoux 1, A Devergie 2, P Ribaud 2, G Socié 2, E Gluckman 2 and D Charron 1,3 1 Service d Immunologie et d Histocompatibilité, Hôpital Saint Louis-AP, Paris, France; 2 Service d Hématologie et Greffe de Moelle, Hôpital Saint Louis-AP, Paris, France; and 3 INSERM U396, Hôpital Saint Louis-AP, Paris, France Summary: In order to evaluate the impact of HLA-DBP1 incompatibilities on the occurrence of acute graft-versus-host disease (GVHD) in unrelated hematopoietic cell transplantation, we studied 57 donor/recipient pairs characterized by their allelic identity for HLA-A, B, C, DRB1 and DQB1 and also for DRB3, 4, 5 loci and aimed to correlate DPB1 mismatches to already described risk factors for GVHD using multivariate Cox regression analysis. DPB1 identity between donor and recipient was observed in 24% and DPB1 compatibility (GVHD vector) in 42%. Two factors were independently associated with severe acute GVHD: two DP incompatibilities (RR = 8.25, 95% confidence interval (CI): , P = 0.010) and disease risk (RR = 10.23, 95% CI: , P = 0.012). Two DPB1 incompatibilities appeared also to be a factor in poorer survival independent of its effect on acute GVHD (RR = 4.97, 95% Cl: , P = 0.002). A correlation between acute GVHD and matching for each individual DPB1 polymorphic region and for residue 69 of the DP molecule, which seems to be a key residue in the alloimmune response, was not observed. Our data indicate that the outcome of unrelated hematopoietic cell transplantation in terms of GVHD but also survival, could be improved through HLA-DPB1 matching or at least by avoiding two DPB1 mismatches. (2002) 30, doi: /sj.bmt Keywords: HLA-DPB1; unrelated donor; acute graft-versus-host-disease; survival The outcome of hematopoietic cell transplantation from unrelated donors remains poorer than sibling bone marrow transplantation (BMT). This difference is related to an Correspondence: Dr P Loiseau, Service d Immunologie et d Histocompatibilité, Hôpital Saint Louis, 1, Avenue Claude Vellefaux, Paris Cedex 10, France Received 14 January 2002; accepted 20 June 2002 increased risk of graft-versus-host disease (GVHD) and delayed immune reconstitution. 1 HLA matching between donors and recipients is the major factor influencing the outcome of allogeneic BMT. Unrelated donor selection generally relies on matching for HLA-A, -B, -C, -DRB1 and -DQB1, but does not consider HLA-DP. With a complete HLA identity for these five loci between donors and recipients, HLA DPB1 disparity is observed in 60 70% of these cases. This is accounted for by the extreme polymorphism of DP genes (85 DPB1 alleles coding for 78 different DP 1 chains) and for the weak or absent linkage disequilibrium between DP and DR- DQ loci. HLA-DP molecules are MHC class II molecules from a structural point of view and can present peptides to T cells, 2 thereby provoking T cell responses. Nonetheless, the role of the HLA-DP in unrelated transplantation has been controversial and remains ill-defined. Several studies 3 8 aimed to correlate DPB1 mismatches with acute GVHD and survival with conflicting results. However, a recent publication using full molecular HLA typing 9 showed an important role of HLA-DP in the alloimmune response. In this study, we aimed to correlate DPB1 mismatches to already described risk factors of acute GVHD using a multivariate Cox regression analysis. For this purpose, we studied 57 donor/recipient transplant pairs fully matched for the alleles HLA-A, B, C, DRB1 and DQB1 and also for the alleles DRB3, 4 and 5 in order to eliminate the alloimmune impact of these HLA molecules. Materials and methods Patients Fifty-seven patients who received hematopoietic stem cell transplantation from an unrelated donor between 1994 and 1999 were included in the analysis. The donor/recipient pairs were selected to be identical for HLA-A, -B, -C, - DRB1, -DRB3, B4 or B5, -DQB1 alleles. Main patient-, donor-, disease-, and transplant characteristics are described in Table 1. The 57 patients were classified into two risk groups according to the disease. The 42 patients suffering from

2 498 Table 1 Patient-, donor-, disease- and transplant characteristics of the 57 pairs included in the study Variable Categories DPB1 mismatches Total P Gender recipient Male Female Age recipient Mean Median Min/Max Diagnosis AL CML Aplastic anemia Non-Hodgkin lymphoma MDS 2 2 Disease risk Low High GVHD Prophylaxis T depleted Non-T depleted CsA + MTX CsA + MTX + ALG CsA Age donor Mean Median Min/Max Gender donor/recipient Male Male Male Female Female Male Female Female CMV donor/recipient Pos Pos Pos Neg Neg Pos Neg Neg MtX = methotrexate, CsA = cyclosporin A; ALG = antilymphocyte globulin. malignant diseases were classified as low risk if grafted at an early stage of the disease (chronic phase for CML, first or second remission for acute leukaemia and NHL) (28 patients) and as high risk if not grafted at a early stage (14 patients). The remaining 15 patients with aplastic anemia are characterized by a similar survival and severe acute GVHD rates as the group of high risk malignant diseases (21% and 27% of survival at 1 year and 20% and 21.4% of severe acute GVHD, respectively). On this basis, aplastic anemias were pooled with high risk malignant diseases for statistical analysis. Conditioning regimens varied with disease and disease stage. For GVHD prophylaxis, cyclosporin A (CsA) was administered to the 16 patients who received a CD34 + selected bone marrow and antilymphocyte globulin (ALG) was added to cyclosporine A + methotrexate in five of the 41 patients who received unmanipulated grafts. All patients received bone marrow, except one who received mobilized peripheral blood hematopoietic stem cells. Methods Histocompatibility typing of HLA class I: Serological typing for HLA-A and -B was performed using the standard microlymphocytotoxicity method 10 with monoclonal antibodies (One Lambda, Inc., Canoga Park, CA, USA) which defined 24 HLA-A and 48 HLA-B antigens. High resolution typing of HLA-A, -B and -C: HLA-A, - B and -C high resolution typings were performed using specific allele PCR-SSP kits (Dynal, Compiègne, France). Histocompatibility typing of HLA class II: HLA-DRB1- DQB1 and HLA-DPB1 medium resolution typings were performed using the PCR-SSO reverse dot blot kits from Innogenetics (Innolipa DRB key and Innolipa DQB kits) (Zwijndrecht, Belgium) and HLA-DRB1, -B3, -B4 and - B5 high resolution typings using a local PCR-SSP/RFLP technique 11 or using allele specific PCR-SSP kits from One Lambda. Persistent HLA-DPB1 ambiguities of typing were resolved by a local PCR-RFLP typing. 3,12 Because of the strong linkage disequilibria between DPA1/DPB1 8,12 and DQA1/DQB1 alleles, we did not perform the DPA1 and DQA1 typings for the evaluation of the donor/recipient matching. Statistical methods Grading the severity of acute GVHD was performed according to Glucksberg s criteria, 13 with modifications introduced in Survival rates and complication cumulative incidence rates were estimated using the Kaplan Meier method 15 and compared curve by curve by the logrank test. The influence of factors on disease free survival or on GVHD cumulative incidence was examined using the proportional hazard model. 16 If the relative death, or GVHD rates in two adjacent categories were not substantially different, these categories were grouped together.

3 Variables changing with time, such as acute GVHD, were used in the proportional hazard model as time-dependent covariates. In the multivariate procedure, a P value less than 0.05 was considered as significant. Relative risks of dying or of GVHD occurrence were presented as mean estimates with a 95% confidence interval (95% CI). The Statistical Package for Social Scientists (SPSS) was used for data management and for survival analyses. The variables retained for the multivariate analysis of GVHD were: recipient and donor age, disease risk, donor and recipient CMV status, donor and recipient gender (sex match or mismatch) and DPB1 incompatibility; for survival, the variables retained were: recipient and donor age, disease risk, donor and recipient CMV status, donor and recipient gender, DPB1 incompatibility, grade of acute GVHD and T cell depletion. Results Cumulative risk of acute GVHD P = inc. (n = 11) 0 or 1 inc. (n = 46) Time (days) Figure 1 Effect of DPB1 incompatibilities on severe GVHD (grades III IV) in 57 unrelated bone marrow transplants identical for HLA-A, B, C, DRB and DQB1 alleles. Inc = incompatibility. 499 Demographic characteristics Patient and donor characteristics are displayed in Table 1. Recipients having 0, 1 or 2 DPB1 incompatibilities with their donors constituted three groups who were balanced for the different variables included in Table 1 (age, disease risk, CMV status, sex matching, GVHD prophylaxis and conditioning regimen) Grade 0 I Grade II Grade III IV Donor/recipient matching for HLA-DPB1 Among the 57 transplant pairs selected for their HLA-A, - B, -C, DRB, and -DQB1 allelic identity, DPB1 identity was observed in 24% (14/57) and compatibility in 42% (24/57). DPB1 compatibility with respect to GVHD was considered to be those transplants where no DPB1 difference exists between recipient and donor (DP identical) and those where the recipient is homozygous for one allele that is shared by the donor. One DPB1 incompatibility was observed in 39% (22/57) and two in 19% (11/57). Analysis of factors influencing acute GVHD Of the 57 patients, 35 developed a grade 0 I GVHD, 15 developed a grade II GVHD and seven developed a grade III IV GVHD. The distribution of severe acute GVHD (grade III IV) was different between the group of patients with 0 1 and the group with two DPB1 disparities (3/46 or 7% vs 4/11 or 36%, respectively, P = 0.02) (Figures 1 and 2). In a multivariate Cox regression analysis studying the following factors: recipient and donor age, disease risk, donor and recipient CMV status, donor and recipient gender (sex match or mismatch) and DPB1 incompatibility, two factors were significantly associated with severe GVHD (grade III IV): two DPB1 incompatibilities (0, 1 vs 2) (RR = 8.25, 95% CI: , P = 0.010) and the disease risk (low vs high) (RR = 10.23, 95% CI: , P = 0.012) (Table 2). The absence of GVHD in patients receiving T cell-depleted grafts meant that Cox regression analysis of the impact of T cell depletion on GVHD was not possible. These results clearly showed that two DPB inc. 1 inc. 2 inc. Number of DPB1 incompatibilities Figure 2 Number of transplants developing grade 0 I, II, III IV acute GVHD according to 0, 1 or 2 HLA-DPB1 disparities. Inc = incompatibility. Table 2 Factors influencing severe GVHD (grade III IV) (multivariate Cox analysis) Variable Relative risk (95% CI) P Disease risk (low vs high ( ) risk) DPB1 incompatibility 8.25 ( ) (0 1 vs 2) disparities between donor and recipient is a risk factor of severe acute GVHD. Analysis of factors influencing the survival Survival at 1 year post transplant was 40%, 49% and 9% in the groups of patients with 0, 1 and 2 DPB1 disparities, respectively, and the survival rates were significantly different between the groups with 0 and 2 DPB1 incompati

4 500 Cumulative survival vs 2: P = vs 2: P = inc (n = 22) 0 inc (n = 24) aa 55 57, aa 65 69, aa 76 and aa 84 87). The biological significance of each HVR matching on GVHD and survival was tested and no correlation was observed. Whilst it appears that glutamic acid at position 69 (E69) in the DP 1 molecule is a key residue in the alloimmune response and antigen presentation, 20 we did not detect any effect of E69 incompatibility on either survival or acute GVHD. Discussion Inc. (n = 11) Time (months) Figure 3 Effect of DPB1 incompatibilities on survival in 57 unrelated bone marrow transplants identical for HLA-A, B, C, DRB and DQB1 alleles. bilities (P = 0.04) and the group with 1 and 2 incompatibilities (P = 0.01) (Figure 3). In a multivariate Cox analysis studying the following factors: recipient and donor age, disease risk, donor and recipient CMV status, donor and recipient gender, DPB1 incompatibility, grade of acute GVHD and T-depletion, four factors were independently associated with poorer survival: two DPB1 incompatibilities (RR = 4.97, 95% CI: , P = 0.002), high risk (RR = 4.16, 95% CI: , P 10 3 ), recipient s CMV serostatus (RR = 3.51, 95% CI: , P = 0.001) and acute GVHD (grade II IV) (RR = 1.71, 95% CI: , P = 0.044) (Table 3). These results showed that two DPB1 incompatibilities is predictive of poor survival and that the effect on survival is not only through the impact on GVHD. DPB1 mismatches and relapse Eleven patients out of the 42 patients eligible for this complication underwent a relapse. Among these 11 patients, six presented no DPB1 mismatches and five presented one or two DPB1 mismatches. Statistical analysis did not show any significant impact of DPB1 mismatches on relapse. DPB1 polymorphism The DPB1 polymorphism is restricted to six hypervariable regions (HVR) of exon 2 (amino-acids (aa) 8 11, aa 33 36, Table 3 Factors influencing patient survival (multivariate Cox analysis) Variable Relative risk (95% CI) P Acute GVHD (0 I vs II IV) 1.71 ( ) Disease risk (low risk vs 4.16 ( ) 10 3 high risk) DPB1 incompatibility ( ( ) vs 2 incompatibilities) Recipient s CMV serostatus 3.51 ( ) (negative vs positive) HLA-DP molecules are not only class II molecules from a structural point of view, but they are fully functional: they can present peptides to T cells, 20 can induce an in vitro 3,19,21 23 and an in vivo 24,25 T cell response and DP polymorphisms are susceptibility factors in certain diseases These observations suggest that the DP molecule could act as a transplantation antigen. Several studies have shown that mismatching at the HLA-DPB1 gene is very frequent in unrelated hematopoietic stem cell transplantations (70 to 80%). 6,8 This is due to the extreme DPB1 polymorphism and to the low or absent linkage disequilibrium between DPB1 and DR/DQ loci. Among the 57 transplant pairs identical for HLA-A, - B, -C, -DRB1, B3, B4 or B5 and -DQB1 alleles, 24% of the transplant pairs were DPB1 identical and 42% were compatible. The increased compatibility is due mainly to DPB1*0401 homozygosity which is the most common DPB1 allele with an allelic frequency of 50%. The percentages of DBP1 identity and compatability observed in this work are twice as high as those observed in Varney s study 8 who reported 11% and 27% of transplant pairs to be DPB1 identical and compatible. As the transplant pairs of the present study were chosen to be identical for the alleles HLA-A, -B, -C, -DRB and -DQB1, a higher rate of transplants with conversed haplotypes was probably included in this study explaining the higher proportion of DBP1 identity or compatability. This hypothesis is confirmed by Petersdorf s study, 9 in which the percentage of DPB1 identical pairs was similar to our cohort. Using an univariate and multivariate Cox analysis, a significantly increased frequency of severe GVHD was observed in the two DPB1 mismatched compared with the matched and single allele-mismatched transplants which were otherwise fully allelic HLA matched in order to eliminate any contribution from these genes to the clinical endpoints. However, as the hematopoietic stem cell transplantation with T cell-depleted bone marrow are characterized by a very low rate of acute GVHD, two DPB1 disparities are risk factor for severe GVHD in non-t cell-depleted BMT. The isolation of DP restricted T cells invading a skin lesion at the onset of GVHD in an otherwise HLA matched unrelated bone marrow transplant is further evidence for DP incompatibility as risk factor of GVHD. 24 Two DPB incompatibilities also appeared to be a factor in poorer survival in our study independent of GVHD, and to have no impact on relapse. The way in which DPB1 mismatches impair survival independently is unclear. Two possible explanations are: (1) an ineffective presentation of peptides by the DP molecule to host cells leading to a less effective anti-infectious response; (2) delayed immune

5 reconstitution through an attack of the DP mismatched cells by the host. The impact of HLA-DP mismatches on the outcome of unrelated hematopoietic stem cell transplantation has been controversial for a long time. 3,4,6 8 However, Petersdorf et al have shown recently an important role for the DP molecule in the induction of severe GVHD in patients suffering from CML, who received hematopoietic stem cells from an unrelated donor molecularly matched for HLA-A, -B, -C, - DRB1 and -DQB1. 9 Our study confirms the impact of DPB1 mismatches on severe GVHD in patients fully HLAmatched including HLA-DRB3, -B4 and -B5 loci. Indeed, mismatches for the second DR gene which occur in 5% of fully HLA matched individuals (except for DPB1) (data not shown) could contribute to the clinical end-points and give confusing conclusions. Moreover, our work which studied patients suffering from diverse diseases (malignant and non-malignant), confirms that DPB1 impact on GVHD is independent of the diagnosis. In contrast to our findings, Petersdorf et al did not observe a significant difference in survival between matched and mismatched DPB1 grafts. However, as GVHD is the main risk factor for poor survival and as DPB1 mismatches are a risk factor for GVHD, an impact of DPB1 mismatches on survival is expected. In conclusion, the results of this study provide evidence of a deleterious effect of two DPB1 mismatches on the occurrence of severe acute GVHD and on patient survival in unrelated hematopoietic stem cell transplantations that are matched for HLA-A, -B, -C, -DRB and -DQB1. alleles. As an effect of multilocus mismatching for class I and class II (DR and DQ) on GVHD and survival has been shown, 31 it is tempting to speculate that in partially matched grafts for the other HLA loci, DPB1 mismatches could participate in this multilocus mismatching effect. Our data indicate that the results of unrelated hematopoietic stem cell transplantations could be improved through matching of donors and recipients for HLA-DPB1 or at least by avoiding two DPB1 mismatches. Acknowledgements We thank V Bajik, N David, M Dourche, G Henry, F Fecamp, N Probel, F Marzais for technical work and N Decolasse for secretarial assistance. We also thank Nuala Mooney for reading the paper. The work was supported by the ARS 2000 FRM grant. References 1 Szydlo R, Goldman J, Klein J et al. Results for allogeneic bone marrow transplants for leukaemia using donors other than HLA-identical siblings. J Clin Oncol 1997; 15: Crumpton MJ, Bodmer JG, Bodmer WF et al. Biochemistry of class II antigens: Workshop report. In: Albert ED, Baur MP, Mayr WR (eds). Histocompatibility Compatibility Testing Springer-Verlag: New York, 1984, pp Al Daccak R, Loiseau P, Soulie A et al. HLA-DP genotyping in HLA-A, -B and -DR identical intrafamilial bone marrow transplantations. Leukemia 1990; 4: Nomura N, Ota M, Kato S et al. Severe acute graft-versus- host disease by HLA-DPB1 disparity in recombinant family of bone marrow transplantation between serologically HLAidentical siblings: an application of the polymerase chain reaction-restriction fragment length polymorphism method. Hum Immunol 1991; 32: Eiermann TH, Fakler J, Goldmann SF. The incidence of DPB1 differences between serological and mixed lymphocyte culture matched unrelated individuals: implications for selection of bone marrow donors. Bone Marrow Transplant 1992; 9: Petersdorf EW, Smith AG, Mickelson EM et al. The role of HLA-DPB1 disparity in the development of acute graft-versus-host disease following unrelated donor marrow transplantation. Blood 1993; 87: Howell WM, Sage DA, Kohler CA et al. Incidence and clinical significance of HLA-DP patient donor mismatches in allogenic bone marrow transplantation. Bone Marrow Transplant 1995; 15: Varney MD, Lester S, McCluskey J et al. Matching for HLA- DPA1 and DPB1 alleles in unrelated bone marrow transplantation. Hum Immunol 1999; 60: Petersdorf E, Gooley T, Malkki M et al. The biological significance of HLA-DP gene variation in haematopoietic cell transplantation. Br J Haematol 2001; 112: Terasaki PI, Verbici D, Park LS et al. Microdroplet testing for HLA-A, -B, -C and -D antigens. Am J Clin Pathol 1978; 69: Mercier B, Al Daccak R, Samaan A et al. HLA-DRB1 generic and HLA-DQB1 typing by PCR amplification using sequence specific primers (PCR-SSP). Assessment after one year of routine use by three laboratories. Eur J Immunogenet 1994; 21: Al Daccak R, Wang FQ, Theophile D et al. Gene polymorphism of HLA-DPB1 and DPA1 loci in Caucasoid population: frequencies and DPB1-DPA1 associations. Hum Immunol 1991; 31: Glucksberg H, Storb R, Fefer A et al. Clinical manifestation of graft-vs-host disease in human recipients of marrow from HLA-matched sibling donors. Transplantation 1974; 18: Przepiorka D, Weisford D, Martin P et al Consensus conference on acute GVHD grading. Bone Marrow Transplant 1995; 15: Kaplan EL, Meier P. Non parametric estimation from incomplete observations. J Am Stat Assoc 1958; 53: Cox DR. Regression models and life-tables (with discussions), series B. J Roy Stat Soc 1972; 34: Portolicchio I, Brookes PA, Madrigal A et al. HLA-DPB1 mismatch at position 69 is associated with high helper T lymphocyte precursor frequencies in unrelated bone marrow transplant pairs. Transplantation 1996; 62: Diaz G, Catalfamo M, Coiras MT et al. HLA-Dpbeta residue 69 plays a crucial role in allorecognition. Tiss Antigen 1998; 52: Nicholson I, Varney M, Kanaan C et al. Alloresponses to HLA-DP detected in the primary MLR: correlation with a single amino acid difference. Hum Immunol 1997; 55: Fontenot AP, Torres M, Marshall WH et al. Beryllium presentation to CD4+ T cells underlies disease-susceptibility HLA- DP allele chronic beryllium disease. Proc Natl Acad Sci USA 2000; 97: Penzes M, Rajczy K, Gyodi E, Petranyi GG. Influence of HLA-DPB1 mismatches on MLR responses: the role of high resolution HLA Class II typing and MLC in unrelated donor selection for BMT. Bone Marrow Transplant 1998; 22: S34 S Jeras M, Delbosc A, Bohinjec M, Tongio MM. Limiting 501

6 502 dilution analysis of human IL-2 producing T-cell precursors. Determination of hierarchical alloantigenic potential of human leukocyte antigens and evaluation of alloresponses to single HLA-DP molecules. Pflugers Arch 2000; 439: R104 R Ibisch C, Gallot G, Vivien R et al. Recognition of leukemic blasts by HLA-DPB1-specific cytotoxic T cell clones: a perspective for adjuvant immunotherapy post-bone marrow transplantation. Bone Marrow Transplant 1999; 23: Gaschet J, Lim A, Vivien R et al. Acute graft versus host disease due to T lymphocytes recognizing a single HLA- DPB1*0501 mismatch. J Clin Invest 1996; 98: Gaschet J, Gallot G, Ibisch C et al. Acute graft-versus-host disease after bone marrow transplantation with a single HLA- DPB1*1001 mismatch: involvement of different TCRBV subsets. Bone Marrow Transplant 1998; 22: Lympany PA, Petrek M, Southcott AM et al. HLA-DPB polymorphisms: Glu 69 association with sarcoidosis. Eur J Immunogenet 1996; 23: Yamasaki K, Horiuchi I, Minohara M et al. HLA-DPB1* 0501-associated opticospinal multiple sclerosis: clinical, neuroimaging and immunogenetic studies. Brain 1999; 122: Fukazawa T, Yamasaki K, Ito H et al. Both the HLA-CPB1 and -DRB1 alleles correlate with risk for multiple sclerosis in Japanese: clinical phenotypes and gender as important factors. Tiss Antigen 2000; 55: Nishimaki K, Kawamura T, Inada H et al. HLA-DPB1*0201 gene confers disease susceptibility in Japanese with childhood onset type I diabetes, independent of HLA-DR and DQ genotypes. Diabet Res Clin Pract 2000; 47: Noble JA, Valdes AM, Thomson G, Erlich HA. The HLA class II locus DPB1 can influence susceptibility to type 1 diabetes. Diabetes 2000; 49: Petersdorf E, Gooley T, Anasetti C et al. Optimizing outcome after unrelated marrow transplantation by comprehensive matching of HLA Class I and II alleles in the donor and recipient. Blood 1998; 92: