The influence of the degree of compatibility of the patient and donor HLA genotype on the outcome of allogeneic hematopoietic stem cell transplantation

Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is one of the effective methods for treating hematological diseases. High accuracy of donor-recipient pair selection based on HLA-system reduces the risk of complications after allo-HSCT. Modern typing methods allow us to study HLA loci not only in high, but also in allelic resolution. The aim of the study was to evaluate the effect of compatibility at five HLA loci in allelic resolution and mismatches at the HLADPB1 locus of donor-recipient pairs on the outcome of allogeneic unrelated HSCT. Material and methods . The work included 38 donor-recipient pairs; before HSCT, HLA-typing at the HLA-A, -B, -C, -DQB1, -DRB1 loci was performed in high resolution. Retrospectively, all studied donor–recipient pairs were typed by NGS technology at allelic resolution using the AllType NGS II Loci Amplification Kit (One Lambda, USA) and NGSgo-MX11-3 (GenDX, Netherlands). Statistical data processing was performed using StatTech 4.8.3 software (developer – StatTech LLC, Russia). The DPB1 T-cell Epitope (TCE) Algorithm v2.0 calculator was used to classify the type of HLA-DPB1 mismatch. Results and discussion . When assessing the overall three-year survival for recipients compatible with donors 10/10 at allelic resolution and less than 10/10, no statistically significant differences were found (p = 0.912). Mismatch in HLA-DPB1 according to the classification did not have a statistically significant effect on the overall three-year survival (p = 0.589). The analysis revealed that the odds of developing chronic graft-versus-host disease in recipients with TCE-incompatible mismatch were higher (p = 0.045) compared with recipients whose donor was incompatible with TCE-compatible mismatch or fully compatible. Conclusion . Non-permissive HLA-DPB1 mismatch in donor-recipient pairs is one of the predictors of the development of chronic graft-versus-host disease in recipients. The model obtained by binary logistic regression had good discriminatory ability (area under the ROC curve 0.713; 95 % confidence interval 0.513–0.912).