Work

1088 Identification and validation of novel shared neoantigen candidates utilizing whole exome sequencing of >340,000 patient samples

Year: 2026

Type: conference-abstract

Abstract: Background CD8 + T cells identify neoplastic cells by detecting peptides presented on the cell surface by major histocompatibility complex class I (MHC-I). Among these peptides are neoantigens – peptides derived from mutated proteins – which are promising targets for immunotherapies. Personalized neoantigen vaccines have shown impressive results in recent clinical trials. However, personalized immunotherapy development is currently time-intensive and cost-prohibitive at scale, illustrating the growing need to identify shared neoantigens that can be targeted by off-the-shelf immunotherapies. Using large-scale mutation frequency screening from Caris Life Sciences’ clinical dataset, we have developed a pipeline to identify high-impact shared neoantigen candidate targets for use in cancer-specific immunotherapy development.Methods Using de-identified whole-exome data from Caris Life Sciences’ clinical dataset, unique somatic mutations from ~20,000 protein-coding genes across 52 tumor lineages were identified, and mutation frequency was calculated within both the sample lineage and the complete dataset. Reference transcripts were then mutated in silico for the top 10,000 mutations, followed by in silico translation to identify the resulting mutant amino acid sequences. An in-house tiering system utilizing NetMHCpan 4.1 EL and MHCflurry 2.01 2 was used to identify and prioritize 8- to 14-mer peptides that were most likely to be presented by HLA alleles common in the United States. The top 16 candidate peptides were synthesized and screened for cell-free MHC-I binding affinity using peptide-receptive recombinant MHC-I proteins.Results Comprehensive WES screening of >340,000 patient samples resulted in over 60 TB of germline-subtracted unique mutations that were then ranked by mutational frequency. Of the top 10,000 mutations screened, 490 mutations resulted in peptides that were predicted to be presented by a common HLA allele. Of these, 16 peptides were shortlisted based on the presence of a significant patient population harboring both the mutation and appropriate HLA allele. Of the 16 peptides that were synthesized and screened, 9 demonstrated significant (<μM) cell-free binding to the predicted recombinant HLA allele.Conclusions Analysis of Caris Life Sciences’ extensive clinical sequencing dataset resulted in the identification of thousands of somatic mutation neoantigen candidates that have not been previously investigated. MHC-I presentation prediction algorithms combined with patient cohort analysis enabled prioritization of peptides with high presentation potential and a targetable patient cohort size. Cell-free MHC-I binding validation was performed on 16 potential neoantigens, of which 9 displayed sub-micromolar binding to the predicted HLA allele. Further validation, including immunogenicity testing, TCR profiling, and detection in primary tumor samples, is ongoing.References Birkir Reynisson, Bruno Alvarez, Sinu Paul, Bjoern Peters, Morten Nielsen. NetMHCpan-4.1 and NetMHCIIpan-4.0: improved predictions of MHC antigen presentation by concurrent motif deconvolution and integration of MS MHC eluted ligand data. Nucleic Acids Research. 2020; Volume 48, Issue W1, Pages W449–W454.O’Donnell TJ, Rubinsteyn A, Laserson U. MHCflurry 2.0: Improved Pan-Allele Prediction of MHC Class I-Presented Peptides by Incorporating Antigen Processing. Cell Systems. 2020;11(1):42–48.e7.

Source: Journal for ImmunoTherapy of Cancer

DOI: 10.1136/jitc-2026-sitc2026.1088