The Functional Distance Between Mismatched HLA-DPB1 Increases Risks of Relapse and Mortality after Unrelated Donor Hematopoietic Cell Transplantation for AML, ALL and MDS: A Refinement of the T Cell Epitope Group Algorithm for Permissive Mismatches

Background: We and others have previously shown that non-permissive T cell epitope (TCE) group mismatches at HLA-DPB1 are associated with the risks of mortality after hematopoietic cell transplantation (HCT) from 10/10 HLA-matched unrelated donors (Fleischhauer et al, Lancet Oncol 2012; Pidala et al, Blood 2014). Moreover, we recently reported that TCE groups are reflected by a numerical score assignable to each HLA-DPB1 allele based on the combined median impact of 12 naturally occurring amino acid substitutions (AAS) on allorecognition of HLA-DPB1*09:01 as reference, termed functional distance (FD) (Crivello et al, Biol Blood Marrow Transplant 2015). Here we studied the association between the Delta in FD scores of HLA-DPB1 alleles present in the patient and in the donor (Delta-FD), and the clinical outcome of unrelated HCT.

Methods: 417 consecutive adult patients transplanted from a 10/10 HLA-matched unrelated donor AML (n=302 [72%]), ALL (n=58 [8%]), or MDS (n=57 [14%]) at the University Hospital Essen between the years 2005 and 2014 were included in the analysis. 37 pairs were matched for both HLA-DPB1 alleles (12/12 HLA matches) while the remaining 380 pairs were HLA-DPB1 mismatched. Among the latter, Delta-FD scores were calculated as the absolute number of [FD_patient-FD_donor] on the basis of previously described FD scores for each HLA-DPB1 allele (Crivello et al, Biol Blood Marrow Transplant 2015).

Results: The median Delta-FD score of HLA-DPB1 mismatched pairs was 1.64 (0.01-7.46). Receiver Operator Curves indicated stratification into 2 subgroups with Delta-FD scores <=2.665 (n=253 [66%]) and >2.665 (n=127 [34%]) as the best predictor of overall survival (OS). The 2 subgroups showed no significant differences for the distribution of major variables including diagnosis, disease status at transplant, immune prophylaxis and conditioning regimen, except for the percentage of permissive HLA-DPB1 TCE mismatches which was significantly higher in the subgroup with Delta-FD scores <=2.665 (p<0.0001).

With a median follow-up of 4 yrs for surviving pts, the 5-yrs OS in the entire HLA-DPB1 mismatched cohort was 48%. In the 2 Delta-FD subgroups, the Kaplan-Meier (KM) probabilities of OS were 52% for Delta-FD <=2.665 and 38% for Delta-FD >2.665 (p<0.008), compared to 50% and 44% (p=0.31) for permissive and non-permissive TCE mismatches, respectively. In multivariate analysis, independent predictors of OS were time-dependent acute GvHD (HR 3.41, p<0.0001) and chronic GvHD (HR 0.41, p<0.0001), the use of anti-thymocyte globulin (HR 0.58, p=0.0006), disease status at transplant (HR 1.27, p<0.007), patient age (HR 1.63 p<0.007), and the stratified Delta-FD score (HR 1.51, p<0.007). Moreover, Delta-FD scores >2.665 were associated with lower probability of event-free survival (HR 1.48, p<0.007), due to significantly higher risks of disease relapse (HR 1.52, p<0.03) and NRM (HR 1.50, p<0.045), but not of acute or chronic GvHD. No significant differences were observed for any of the endpoints between 12/12 HLA-matched and 10/10 HLA-matched pairs with Delta-FD <=2.665.

Conclusion: Stratification of HLA-DPB1 mismatches according to Delta-FD scores between donor and recipient represents a refinement of our previously published TCE algorithm of non-permissive mismatches with significant overlaps. In comparison to the latter, Delta-FD scores showed improved associations with the risks of mortality and relapse after 10/10 HLA-matched unrelated HCT in the patient cohort analyzed. If confirmed, these findings could provide a refined tool for donor-recipient matching for HLA-DPB1, and suggest that the combined impact of key AAS on T-cell alloreactivity, rather than AAS at individual positions, are relevant parameters for risk prediction in HLA-DPB1 mismatched HCT.