Impact of HLA evolutionary divergence and donor-recipient molecular mismatches on antibody-mediated rejection of kidney allografts

Abstract

Several in-silico methods emerge to assess HLA immunogenicity and stratify immunological risk, including HLA molecular mismatches and HLA evolutionary divergence (HED). However, their added value in risk-stratifying antibody-mediated rejection (AMR) remains uncertain. We include 5159 kidney transplant recipients from four centers. Thirty-three clinical and immunological parameters are assessed, including HLA eplet mismatches, PIRCHE-II scores, and HED. Their associations with AMR are evaluated using Cox models. AMR occurrs in 1024 patients (19.9%). Immunological determinants of AMR include anti-HLA DSA (MFI:500-1400, HR:1.87; MFI > 1400, HR:3.84, p < 0.001) and HLA Class II eplet mismatches (HR:1.02, p < 0.001). HLA-DQB1-derived (HR:1.01, p = 0.005) and HLA-DRB1-derived PIRCHE-II scores (HR:1.01, p = 0.001) are also associated with AMR, while HED is not. These findings remain consistent across centers and subpopulations, including non-sensitized patients (n = 4137). Our findings show that Class II molecular mismatches are independently associated with AMR. HED shows no association, suggesting limited utility for immune-risk stratification at a population level. Clinicaltrials.gov: NCT06436586.

Fig. 1. Correlations between baseline immunological parameters.

This plot illustrates the correlation between various HLA immunogenicity indicator variables, including HLA antigenic mismatches, HLA eplet mismatches, PIRCHE-II scores, recipient and donor HED for both HLA class I (A, B, C) and II (DQB1, DRB1), and anti-HLA DSA at the time of transplantation in kidney transplant cohort. Positive correlations are shown in shades of blue, and negative correlations are shown in shades of red. Correlation between variables was assessed using Pearson’s correlation coefficient. Abbreviations: DSA donor-specific antibodies, eplet MM eplet mismatches, HED HLA evolutionary divergence, HLA human leukocyte antigen, MFI mean fluorescence intensity, PIRCHE-II Predicted Indirectly Recognizable Human Leukocyte Antigen Epitopes-II.

Fig. 2. Parameters associated with AMR: multivariable Cox analysis.

Multivariable Cox models for AMR (n = 5,159 patients). Two separate multivariable cox models were performed, both including all clinical and immunological baseline parameters significantly (p < 0.1) associated in the univariate analyses. The only distinction between the models was the incorporation of PIRCHE-II scores: one model used total PIRCHE-II scores (a), while the other used locus-specific PIRCHE-II scores (b). a, b Presents the final results of these analyses. a Multivariable Cox result, including PIRCHE-II scores. b Multivariable Cox result, including locus-specific PIRCHE-II scores. Parameters in the final multivariate model were selected using stepwise backward elimination until each had a P < 0.05, assessed using the Wald test. Two-sided statistical tests were used. The forest plot shows hazard ratios (squares) and 95% CI (horizontal lines). Exact p values can be found in Source data. Abbreviations: DSA donor-specific antibodies, HLA human leukocyte antigen, MFI mean fluorescence intensity, PIRCHE-II Predicted Indirectly Recognizable Human Leukocyte Antigen Epitopes-II.