The role of matching human leukocyte antigen (HLA) haplotypes in unrelated allogeneic bone marrow transplantation (allo-BMT) remains unclear. Here, we imputed the HLA haplotypes of 3657 patients who received unrelated single HLA allele-mismatched allo-BMT, included from the Transplant Registry Unified Management Program (TRUMP) database, the Japanese registry program for hematopoietic transplantation, using mathematical methods. We successfully imputed the HLA haplotypes of both patients and donors in 1365 cases (37.3%) with ≥90% probability. Of the patients, 1326 (97.1%) and 39 (2.9%) were categorized into one-haplotype-matched and no-haplotype-matched groups, respectively. Disease-free survival was significantly worse in the no-haplotype-matched group. Multivariate analyses revealed that no-haplotype-match was an independent risk factor for reducing disease-free survival (hazard ratio, 1.54 [95% confidence interval: 1.01-2.36]; p = 0.047). However, the overall survival did not significantly differ between the groups. The incidence of grade III-IV acute and chronic graft-versus-host disease did not significantly differ between the groups. Furthermore, there were no significant differences in the cumulative incidences of relapse and non-relapse mortality between the groups. Our findings suggest that imputing haplotypes using a mathematical approach can help to avoid transplanting patients with donors who do not share matching haplotypes, thereby improving the outcome of allo-BMT.
© 2021. The Author(s), under exclusive licence to Springer Nature Limited.