Abstract 1220: Accurately genotyping HLA and KIR alleles using cfDNA assay and k-mer based algorithm for immunotherapy

Abstract

Background: HLA and KIR genotypes show great promise as emerging biomarkers for immune checkpoint inhibitors (ICIs) and understanding patient prognosis. Multiple studies have shown that HLA-I heterozygosity and high sequence divergence across alleles positively correlates with response to ICIs. However the high degree of polymorphism and allele sequence similarities in HLA and KIR present a challenge to accurate allele calling. To address these difficulties we developed kmerizer, a novel allele caller optimized for short fragments, such as reads from a cfDNA assay. Methods: We tested kmerizer on MHC class1 and class2 genes, and on KIR genes, on both simulated datasets and real samples (cell lines and plasma samples). To assess the capability of the algorithm to distinguish highly homologous allele pairs, we simulated cfDNA-like fragments with errors on randomly selected allele pairs, and on allele pairs with a high level of homology. Twelve plasma samples and 19 reference cell lines fragmented to cfDNA size were analyzed using a NGS cfDNA assay. For plasma samples, paired buffy coats were sent to an external vendor for HLA typing using multiplex PCR-based amplicon and sequenced by 300bp paired-end reads. Results: Of the 19 cell lines and 12 plasma samples for HLA typing (A, B, C, and DQB1 loci), kmerizer delivered 100% sensitivity with 98% specificity. On the simulated dataset, kmerizer achieved 99% sensitivity and specificity on all the MHC class1 and class 2 loci, and 90% sensitivity and specificity on all KIR loci, for both homozygous and heterozygous pairs. The novel allele caller kmerizer also demonstrated a lighter footprint on computational resource need: one deep-sequencing plasma sample on average can be processed in less than 2 minutes which is about 15 times faster than the most commonly used HLA typing tool HISAT21, which does not support KIR typing. Conclusions: As utilization of ICIs increases, the use of genetic and genomic information to accurately identify patients more likely to respond to ICIs will be critical. kmerizer is a fast and highly sensitive and specific allele caller, and it can effectively call alleles on both HLA and KIR. References: [1] Kim, D., Paggi, J. M., Park, C., Bennett, C., & Salzberg, S. L. (2019). Graph-based genome alignment and genotyping with HISAT2 and HISAT-genotype. Nature biotechnology, 37(8), 907-915. Citation Format: Sante Gnerre, Brian Yik Tak Tsui, Tingting Jiang, Yvonne Kim, Dustin Ma, Indira Wu, Rebecca Nagy, Han-Yu Chuang. Accurately genotyping HLA and KIR alleles using cfDNA assay and k-mer based algorithm for immunotherapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1220.