Which is the best predictor of de novo donor specific antibodies in a cohort of non-sensitized first kidney transplantation: antigenic, allelic, epitope or physiochemical HLA mismatches?

Background Assessment of human leukocyte antigen (HLA) matching by using high-resolution allele typing and knowledge of HLA molecule structure may lead to better prediction of de novo donor-specific antibody (dnDSA) development. Methods We conducted a single-center cohort study among 150 non-sensitized first kidney transplant recipients to compare the association between antigenic (Ag), allelic (Al), eplet (Ep), amino acid (AAMS) HLA matching and electrostatic (EMS) and hydrophobic (HMS) mismatch scores, and the development of dnDSA. Results After a mean follow-up time of 49.3 +/- 17.7 months, 18 patients (12%) developed dnDSA. The number of HLA mismatches (MM) was significantly associated with the development of dnDSA. The optimal threshold, determined by Harrell's C-index, varied according to the method (5 MM for Ag, P = 0.006; 6 for Al, P = 0.009; 22 for Ep, P = 0.005; 42 for AAMS, P = 0.0007; 45 for EMS, P = 0.009 and 44 for HMS, P = 0.026). C-indices were similar for all matching approaches, suggesting a similar prediction of dnDSA development. Conclusion In this cohort of low immunological risk transplant patients, the use of Al or Ep matching did not improve the prediction of dnDSA development in comparison with the traditional approach.