Abstract 5444: Analytical validation of HLA class I allele typing in a liquid biopsy platform.

Abstract Background: Human Leukocyte Antigen (HLA) genotyping is an emerging biomarker for immunotherapy efficacy in several solid tumor types but can be particularly challenging to genotype with standard NGS due to the locus’ inherent variability. Accurate HLA typing from cell-free DNA (cfDNA) enables non-invasive biomarker assessment for therapy selection and clinical trial matching and can be provided as incidental to standard liquid biopsy without the additive cost of white blood cell sequencing. Using a novel algorithm to enable HLA genotyping as part of Guardant360 Liquid (Guardant Health, Palo Alto, CA), we present analytical validation of HLA class-I allele typing (HLA-A, -B, -C) and demonstrate concordance with paired tissue assay results. Methods: The cfDNA genotyping algorithm leverages perfect read-pair alignments against a database of more than 16000 known HLA alleles, identifying the most supported allele pairs (germline configuration) based on the read coverage from amongst reference alleles. Analytical validation assessed accuracy using 32 samples with orthogonal HLA data from a validated commercial assay. Limit of detection (LoD) was established through in silico downsampling at 5 coverage levels ranging from high to challenging inputs near the test minimum using 4 samples with 5 replicates per level. Precision was evaluated using 6 clinical samples at 5ng input tested across 5 operator, reagent lot, and instrument combinations. Concordance with HLA genotyping in tissue was assessed by comparing 80 paired patient samples run on both Guardant360 Liquid and Guardant360 Tissue. Results: Guardant360 Liquid demonstrated high accuracy with 97.3% positive percent agreement (PPA) 95% CI: 93.8-99.1% for HLA class-I alleles. Among evaluable results, 100% concordance was achieved for homozygous alleles (16/16) and 97.0% for heterozygous alleles (163/168). LoD was established at the lowest input level allowed for the assay with 100% detection rate across all tested levels. Precision at 1-5X LoD demonstrated 100% PPA across multiple testing conditions. Paired tissue-liquid concordance analysis showed 97.5% overall PPA (231/237 alleles) across HLA-A (80/80, 100%), HLA-B (77/80, 96.3%), and HLA-C (74/77, 96.1%), with discordances primarily attributed to tumor-specific loss of heterozygosity or minor SNP differences between closely related alleles. Conclusions: Guardant360 Liquid HLA typing demonstrates robust analytical performance with high accuracy, sensitivity, and precision in cfDNA with accelerated turnaround times compared to standard genotyping assays. Strong concordance between paired tissue and liquid samples validate the algorithm's performance in cfDNA. This non-invasive HLA typing capability enables comprehensive biomarker profiling for immunotherapy selection and clinical trial eligibility assessment in patients with advanced solid cancers. Citation Format: Adrian Bubie, Sante Gnerre, Marisa Juntilla, Reagan Barnett, Matthew Ellis, Tingting Jiang. Analytical validation of HLA class I allele typing in a liquid biopsy platform abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 5444.