# Population-agnostic real-time molecular mismatch estimation using rSSO-defined HLA allele strings: minimal impact of allelic discordance with NGS

## TL;DR

This study shows that a specific HLA typing method can reliably estimate molecular mismatches in transplants, even when high-resolution data is not available.

## Contribution

The study introduces a population-agnostic method using rSSO MAC strings as a practical surrogate for high-resolution HLA typing in mismatch assessment.

## Key findings

- The first allele in rSSO MAC strings showed over 94% concordance with NGS alleles for most HLA loci.
- Most discordant alleles did not significantly increase eplet mismatches, with over 77% having two or fewer mismatches.
- The method works reliably across different racial and ethnic groups, with minimal clinically meaningful mismatch increases.

## Abstract

Molecular HLA mismatch at the epitope level is strongly associated with donor-specific antibody formation, antibody-mediated rejection, and graft failure after transplantation. Accurate molecular mismatch assessment requires high-resolution HLA typing; however, in clinical practice—particularly in deceased donor transplantation—typing is often performed at low or intermediate resolution. Haplotype-based imputation methods have been developed to infer high-resolution alleles but show variable accuracy, particularly in underrepresented populations. We evaluated whether the first allele listed in the reverse sequence-specific oligonucleotide (rSSO)–derived National Marrow Donor Program (NMDP) multiple allele code (MAC) string can serve as a practical surrogate for high-resolution typing in molecular mismatch assessment.

We analyzed 4,738 individuals who underwent dual HLA typing by rSSO and next-generation sequencing (NGS) across 11 classical HLA loci. Concordance between the first allele of the rSSO MAC string and the corresponding two-field NGS allele was assessed. For discordant allele pairs, molecular disparity was quantified by calculating eplet mismatches using the HLA Eplet Mismatch Calculator.

Allele-level concordance exceeded 94% for 9 of 11 loci, including HLA-A (97.9%), -B (98.5%), -C (95.2%), -DRB1 (96.8%), -DRB3 (94.5%), -DRB5 (99.6%), -DQB1 (98.3%), -DPB1 (96.8%), and -DPA1 (98.7%). Lower concordance was observed for HLA-DQA1 (86.7%) and -DRB4 (72.0%). Most discordant NGS alleles, except for the DPB1 locus, were present in the rSSO MAC string. Among discordant allele pairs, 30.4% had zero eplet mismatches and 77.6% had two or fewer mismatched eplets. Allele pairs with more than two mismatched eplets accounted for less than 0.3% of all comparisons across loci. No significant differences in concordance were observed across racial or ethnic groups.

The first allele listed in the rSSO-derived MAC string provides a reliable, population-agnostic surrogate for high-resolution HLA typing in molecular mismatch assessment. Although allele-level discordance occurs, it rarely results in clinically meaningful increases in eplet mismatch burden. This laboratory-based approach enables real-time molecular mismatch estimation in time-sensitive settings and facilitates retrospective analysis of cohorts lacking sequencing-based HLA data, supporting broader integration of molecular mismatch into precision transplantation practice.