Correlation and agreement between the number of Class I and II eplet mismatches calculated using serological, low-intermediate and high resolution molecular human leukocyte antigen typing methods

Aim: To evaluate the agreement between eplet mismatches calculated by serological and two-digit typing methods compared to high-resolution four-digit typing method. Background: Structural human leukocyte antigen (HLA) matching at the eplet level can be identified by HLAMatchmaker, but requires the entry of four-digit allele typing, which are often not practicable in donor kidney allocation because of time constraint. Methods: In a cohort of 264 donor/recipient pairs, the evaluation of measurement error was assessed using intra-class correlation to confirm the absolute agreement between the number of eplet mismatches at class I (HLA-A, -B, C) and II loci (HLA-DQ and -DR) calculated using serological or two-digit molecular typing compared to four-digit molecular typing methods. The proportion of donor/recipient pairs with a difference of >5 eplet mismatches between the HLA typing methods was also determined. Results: Intra-class correlation coefficients between serological and four-digit molecular typing methods were 0.969 (95% confidence intervals [95%CI] 0.960-0.975) and 0.926 (95%CI 0.899-0.944), respectively; and 0.995 (95%CI 0.994-0.996) and 0.993 (95%CI 0.991-0.995), respectively between two-digit and four-digit molecular typing methods. The proportion of donor/recipient pairs with a difference of >5 eplet mismatches at class I and II loci was 4% and 16% for serological versus four-digit molecular typing methods, and <1% and 2% for two-digit versus four-digit molecular typing methods, respectively. Conclusions: Compared with serology, there is a high level of agreement in the number of eplet mismatches calculated using two- compared to four-digit molecular HLA-typing methods, suggesting that two-digit typing may be sufficient in determining eplet mismatch load in kidney transplantation.