Unrelated Stem Cell Donor HLA Match Likelihoods in the US Registry Incorporating HLA-DPB1 Permissive Mismatching

Abstract

Donor-recipient HLA matching at the DPB1 locus improves the outcomes of hematopoietic cell transplantation (HCT). Retrospective outcomes studies found that in transplants matched for all 8 alleles of the A, B, C, and DRB1 loci at high resolution (8-of-8 match), few transplants were also allele-matched at the DPB1 locus. DPB1 allele matching was thought to be logistically impractical, however a DPB1-permissive mismatch model based on T-cell epitope (TCE) reactivity expands the proportion of suitable donors. To understand the likelihood of success for finding a DPB1-permissive donor, we sought to expand population genetic match likelihood models for the US unrelated donor registry, National Marrow Donor Program (NMDP). After extending HLA haplotype frequency estimates to include the DPB1 locus, our models found that the likelihood of having a DPB1-permissive donor was not much lower than the 8-of-8 match likelihoods. A maximum of 5 additional donors would need to be typed to find a more optimal DPB1-permissive donor at least 90% of the time. Linkage disequilibrium patterns between the DPB1 locus and other classical HLA loci varied markedly by haplotype and population, indicating that the known recombination hotspot between DQ and DP gene complexes has not had uniform impact, thus DPB1-permissive donors are easier to identify within minority populations. DPB1 TCE categories were highly predictable from HLA typing at other loci when imputed with extended haplotype frequency data. Our overall results indicate that registry search strategies that seek a more optimally matched HCT donor encompassing HLA-DPB1 permissibility are likely to be highly productive.

Efficient test for deviation from Hardy Weinberg Equilibrium with known or ambiguous typing in highly polymorphic loci

“…We studied 8,078,224 donors from the National Marrow Donor Program (NMDP) registry [10]. The dataset provided for each donor an HLA typing and the 21 detailed populations that combine into 5 broad populations (S1 Table ) if known, any other typed race was converted to unknown.…”