*Purpose: De novo donor-specific antibodies (dnDSA), especially to HLA-DQ, are one of the leading causes of late kidney graft failure. Many investigators have attempted to use molecular mismatch (MM) load quantification tools, such as HLA Matchmaker, to predict dnDSA formation. We assessed HLA-DQ MM load in a single center, well-characterized cohort of kidney transplant recipients to determine if risk for DSA formation could have been predicted with high accuracy at the time of transplant.

*Methods: Recipients (n=174) were confirmed to have no pre-transplant Class II HLA antibodies, and all subjects were high-resolution typed for DQA1/DQB1. Minimum follow up time was 3 years (median 4.5, maximum 10). We calculated the HLA-DQ eplet mismatch score (EpMM) in HLA Matchmaker, as well as the electrostatic mismatch score (EMS-2D) using the Cambridge Immunogenicity Algorithm. Analysis was performed by individually comparing each donor-mismatched DQA1/DQB1 allele (n=265) to the recipient typing, in order to account for immune presentation. These scores were then analyzed in relation to dnDSA formation against the corresponding allele.

*Results: Comparing DSA+ alleles (n=112) to DSA- alleles (n=153), there was a significant difference in mean EpMM score (16 vs 12, p<0.001) and mean EMS-2D score (40 vs 27, p<0.001). Importantly, though, these parameters were not prognostic for DSA formation, as 29% of DSA+ alleles had an EpMM score <12. Furthermore, 25% of the DSA- alleles (median follow up time of 4 years), had an EpMM score >16. ROC analysis, performed to quantify the ability of these scores to predict dnDSA formation, resulted in a relatively weak AUC of 0.67. We further evaluated a previously published EpMM cutoff of 8 with our cohort, resulting in good sensitivity (95%) but poor specificity (31%). Finally, to determine if high MM load correlated with either the strength or length of time to immune response, we compared both the EpMM and EMS-2D scores to the time to dnDSA formation, as well as the peak antibody titer, for the DSA+ alleles. No correlation was observed with either outcome, even when excluding patients with known subtherapeutic immunosuppression levels prior to dnDSA formation.

*Conclusions: In conclusion, while MM load assessment is strongly associated with dnDSA formation and rejection in many retrospective studies, these tools still lack the ability to determine prospectively which mismatched donor alleles will elicit an immune response in a given recipient. In addition, center-specific populations and standards of care likely influence rates of dnDSA formation, questioning current applicability of a set MM load threshold as a tool for purposes other than population enrichment approaches for clinical trial design.

« Back to 2020 American Transplant Congress