Number of HLA-Mismatched Eplets Is Not Associated with Major Outcomes in Haploidentical Transplantation with Post-Transplantation Cyclophosphamide: A Center for International Blood and Marrow Transplant Research Study

Abstract

The number of haploidentical hematopoietic stem cell transplantations (haplo-HSCT) performed has increased substantially in recent years. Previous single-center studies using in silico algorithms to quantitively measure HLA disparity have shown an association of the number of HLA molecular mismatches with relapse protection and/or increased risk of acute graft-versus-host disease (GVHD) in haplo-HSCT. However, inconsistent results from small studies have made it difficult to understand the full clinical impact of molecular mismatch in haplo-HSCT. In this study, we investigated the potential of the HLA class I and II mismatched eplet (ME) score measured by HLAMatchmaker, as well as ME load at a specific locus to predict outcomes in a registry-based cohort of haplo-HSCT recipients. We analyzed data from 1287 patients who underwent their first haplo-HSCT for acute lymphoblastic leukemia, acute myelogenous leukemia, or myelodysplastic syndrome between 2013 and 2017, as entered in the Center for International Blood and Marrow Transplant Research database. ME load at each HLA locus and total class I and II were scored using the HLAMatchmaker module incorporated in HLA Fusion software v4.3, which identifies predicted eplets based on the crystalized HLA molecule models and identifies ME by comparing donor and recipient eplets. In the study cohort, ME scores derived from total HLA class I or class II loci or individual HLA loci were not associated with overall survival, disease-free survival, nonrelapse mortality, relapse, acute GVHD, or chronic GVHD (P < .01). An unexpected strong association was identified between total class II ME load in the GVH direction and slower neutrophil engraftment (hazard ratio [HR], 0.82; 95% confidence interval [CI], 0.75 to 0.91; P < .0001) and platelet engraftment (HR, 0.80; 95% CI, 0.72 to 0.88; P < .0001). This was likely attributable to ME load at the HLA-DRB1 locus, which was similarly associated with slower neutrophil engraftment (HR, 0.82; 95% CI, 0.73 to 0.92; P = .001) and slower platelet engraftment (HR, 0.76; 95% CI, 0.70 to 0.84; P < .0001). Additional analyses suggested that this effect is attributable to a match versus a mismatch in the graft-versus-host direction and not to ME load, as a dose effect was not identified. These findings contradict those of previous relatively small studies reporting an association between ME load, as quantified by HLAMatchmaker, and haplo-HSCT outcomes. This study failed to demonstrate the predictive value of ME from HLA molecules for major clinical outcomes, and other molecular mismatch algorithms in haplo-HSCT settings should be tested.