Addition of HLA-DQA1/DQB1 trans heterodimers to single-molecule eplet mismatch analysis does not improve risk stratification for DSA, AMR, or Graft Loss

Aim: HLA-DQ antigens are usually considered as paired protein products from DQA1 and DQB1 alleles at the same chromosome (cis); however, trans (different chromosome) heterodimerization is possible within two evolutionarily distinct G1 & G2 groups [Petersdorf et al, Blood (2022)]. It has been hypothesized that trans heterodimers contribute to dnDSA formation and rejection, but most studies only consider cis heterodimers of HLA-DQ. We chose to assess DQ heterodimers using single-molecule eplet mismatch (SM epMM) scores from the HLAMatchmaker algorithm. Here we analyze the effects of additional DQA1/DQB1 trans heterodimers when predicting DSA, AMR, and Graft Loss with SM epMM algorithm scores.

Methods: We analyzed 4210 donor-recipient pairs from the Paris Institute for Transplantation and Organ Regeneration. High-resolution 9-gene HLA typing (HLA-A*, B*, C*, DRB1/3/4/5*, DQA1, DQB1) was imputed using an in-house imputation algorithm utilizing hlaR and custom R functions. Cis and trans HLA-DQA1/DQB1 heterodimers were determined based on haplotype frequencies. HLAMatchmaker was used to determine SM epMM using previously published methodology [Wiebe et al, AJT (2018)]. Statistical analysis and data visualization was conducted using RStudio.

Results: SM epMM scores increased in 134 cases (3.17%) when trans heterodimers were considered (Figure 1A). When previously published risk cutoffs [Wiebe et al, AJT (2018)] were applied, only 15 cases increased from intermediate to high risk (Figure 1B). Kaplan Meier results were similar when comparing cis and cis+trans for DQ DSA (cis: p<0.0001; cis+trans: p<0.0001; Figure 1C), AMR (p=0.0026; p=0.0024, Figure 1D) and Graft Loss (p=0.045; p=0.046, Figure 1E). In addition, ROC analysis demonstrated that the ability to predict DQ DSA was comparable with cis (AUC=0.5425) and cis+trans (AUC=0.5423) algorithms. Youden’s index (YI) derived cutoffs from ROC analysis resulted in a cis cutoff of 8 (YI=0.344) and a cis+trans cutoff of 7 (0.349) when assessing the full cohort.

Conclusion: Our results suggest that including trans heterodimers has minimal impact on risk stratification when utilizing SM epMM scores. Despite providing a more complete assessment on DQ mismatch, only a small proportion of cases result in changes of epMM scores and risk categories transitioning from intermediate risk to high risk.