Aim: HLA molecular mismatch (HLAMM) methods provide a detailed and accurate assessment of recipient/donor compatibility. Large cohorts with high-resolution HLA genotyping are necessary to further evaluate these methods, but historic transplants are often typed at the serological equivalent (SE) level. Here, we present a novel pipeline for imputing high-resolution HLA genotypes from SE data.
Methods: We used 1074 subjects (537 recipient/donor pairs) who were initially typed at low/intermediate resolution and reported as SE for HLA-A, B, C, DR, and DQ. They were retyped retrospectively with next-generation sequencing (NGS) for high-resolution at 9 HLA loci (HLA-A*, B*, C*, DRB1/3/4/5*, DQA1, DQB1). Imputation was conducted to obtain genotypes for 9 loci from donor and recipient serological typings. Our imputation methodology included 4 major steps: 1) convert SE to one-field molecular typing examining each ambiguity, 2) impute preliminary genotypes via hlaR, 3) resolve multiple genotypes (from serological to one-field conversion) by selecting the most frequent/common diplotype, and 4) re-impute Class II genotypes based on common haplotypes of DRB1/3/4/5* and DQA1/DQB1*. HLAMM was analyzed using HLAMatchmaker v4.0.
Results: After removal of donors/recipients (n=20/18) with discordant one-field typing the remaining 517 donors (96.3%) and 519 recipients (96.6%) were successfully imputed resulting in 500 complete pairs (93.1%). Donor allele level concordance ranged from 76.2% to 86.9% and recipients ranged from 71.7% to 85.9%. HLAMM results for eplet mismatch (epMM) were comparable whether the NGS typed or imputed dataset was used. The differences in epMM score were less than five for 91.6% of Class I, 94.8% DR, and 96.8% DQ. When risk categories were determined based on epMM score (Wiebe and Nickerson, AJT 2018), 95.1% of patients that were categorized as high risk by NGS data were also high risk with the imputed data, 75.8% remained intermediate (17.4% from intermediate to low), and 66.7% remained low risk (33.3% from low to intermediate).
Conclusion: Imputation of high-resolution genotypes from SE can be accomplished with a high degree of accuracy. Utilizing imputed data results in few changes in risk stratification for high-risk or low/intermediate-risk patients. Our imputation method allows for the application of HLAMM in advanced risk assessment on transplant recipients typed at the serological level.
