ESPN 54th Annual Meeting

ESPN 2022

Assessment of HLA incompatibility at the molecular compared to antigenic HLA level enables better prediction of graft function deterioration in paediatric kidney transplantation.
JON JIN KIM 1 Alexander Fichtner 2 Hannah Copley 1 Caner Susal 2 Lars Pape 3 Jun Oh 4 Kai Krupka 2 Burkhard Toenshoff 2 Vasilis Kosmoliaptsis 1

2- University of Heidelberg
3- University of Hannover
4- University of Hamburg

Introduction: Immune recognition of donor HLA mismatches and subsequent graft rejection remain a major cause of graft deterioration in paediatric kidney transplantation. We aimed to evaluate the potential of computational methods of assessing HLA immunogenicity (molecular mismatching, molMM) and of classical antigen mismatching (antMM) to predict primary alloimmunity risk.

Material and methods:

Methods: We performed a retrospective analysis of 177 paediatric patients (median age 10.8 [IQR] 5-15 years) from the antibody-mediated rejection (ABMR) study from the CERTAIN registry, all of whom had prospective monitoring of donor-specific HLA antibody (DSA) posttransplant.  We compared four molMM methods: amino acid mismatch scores (AAMS, assessing T- and B-cell alloimmunity), electrostatic mismatch score-3-dimensional (EMS3D, assessing B-cell alloimmunity), and NetMHCIIpan (T-cell alloimmunity) at HLA-peptide affinity binding thresholds of 500 nM (netMHC) and 1000 nM (netMHC1k as implemented in the PIRCHE algorithm).


Results: De novo DSA incidence was highest against HLA-DQ (30/177, 17%, Table 1). Higher antMM and molMM were associated with DSA formation. DSA preferentially targeted the highest scoring molMM allele in each individual patient. MolMM methods were more predictive of DSA compared to antMM (Table 1, AUROC results), and EMS3D was the most consistent predictor across all HLA loci (AUROC 0.72-0.75). Biopsy-proven ABMR (11/177, 6%) was associated with increasing recipient age and class II molMM (defined by AAMS or EMS3D or netMHC). Late TCMR (>6 months postrtansplant) was diagnosed on for-cause biopsies in 23 (13%) patients. Accounting for the total HLA burden, neither molMM nor antMM scores were predictive of TCMR outcome. Stratifying HLA mismatches into low or high risk according to molMM score enabled patient classification into strata. TCMR was significantly associated with HLA-DQ molMM using T-cell algorithms - AAMS (HR 1.8, 95%CI 1.2-3.0 per risk stratum) and netMHC (HR 1.5, 95%CI 1.0-2.2 per risk statum).


Summary: Molecular HLA mismatching enabled better prediction of primary alloimmunity risk than conventional antigen mismatching. Incompatibility at the HLA-DQAB loci was the main driver for TCMR and ABMR.