HLA-DPB1 Mismatch and Chronic Graft-Versus-Host Disease Are Favorable Independent Risk Factor for Graft-Versus-Leukemia Effect in Leukemia Patients Transplanted Marrow from HLA-A, -B, -C, -DRB1 and -DQB1 Matched Unrelated Donor

Although the impact of HLA-A, B, C and DRB1 matching in unrelated hematopoietic stem cell transplantation (HSCT) has been well reported, the role of HLA-DPB1 matching for transplant related immunological events, especially chronic GVHD (c-GVHD), graft-versus-leukemia (GVL) effect and their relations has not been well elucidated. Accumulation of large scale unrelated HSCT patient-donor pairs and retrospective 6 HLA locus allele data including HLA-DPB1 in Japan Marrow Donor Program enable us to analyze the effects of HLA-DPB1 matching. 2177 patients met the following criteria were included: HLA-A, B, C, DRB1 and DQB1 allele matched unrelated donor, diagnosis of ALL (N=738), AML (N=1040) or CML (N=399), non-T cell depleted bone marrow without use of ATG, and survived more than 100 days after transplantation. 632 patients were matched for HLA-DPB1 allele, 1181 one allele mismatched, and 364 two allele mismatched in GVH direction. Tacrolimus-based regimen was employed in 1158 patients and cyclosporine-based regimen in 1007. Myeloablative regimen was conditioned in 1956 patients and non-myeloablative regimen in 221. Multivariable competing risk regression analyses were conducted to evaluate the impact of acute GVHD, chronic GVHD and leukemia relapse after transplantation. Confounders considered were combinations of patient age (linear), donor age (linear), risk of leukemia relapse (low and high), GVHD prophylaxis, preconditioning, sex matching and transplanted year. C-GVHD was treated as time-dependent covariates. Hazard risk (HR) of one allele and two alleles mismatch at HLA-DPB1 locus was 0.70 (95% CI 0.58-0.84) and 0.54 (0.41-0.71) respectively compared with DPB1 match (p<0.001). HR of limited type c-GVHD (n=350) and extensive type c-GVHD (n=563) were 0.55 (0.42-0.74) and 0.45 (0.36-0.58) respectively (p<0.001) compared with no c-GVHD (n=1264). Interaction between these two factors was not significant (HR=1.25, p=0.278). HLA-DPB1 mismatch had no impact for c-GVHD (one allele mismatch HR of 0.98 (0.85-1.14) p=0.84 and two allele mismatch 1.03 (0.85-1.25) p=0.73, compared with HLA-DPB1 match. Therefore, HLA-DPB1 matching and c-GVHD were an independent factor to induce GVL effect. Although HLA-DPB1 mismatch showed significant risk factor of a-GVHD, a-GVHD showed no significant HR for leukemia relapse (HR of grade 1 a-GVHD (n=640) 1.11 p=0.32, grade 2 (n=596) 0.91 p=0.45, and grade 3-4 (n=266) 0.78 p=0.14 compared with grade 0 (n=762). Retrospective analysis of HLA-A, B, C, DRB1 and DQB1 complete match unrelated HSCT elucidated the biological effect of HLA-DPB1 matching without information biases at transplantation, showing that GVL effect was induced by HLA-DPB1 mismatch which was independent of c-GVHD.