Abstract
Neoantigen-directed therapy lacks preclinical models recapitulating neoantigen characteristics of original tumors. It is urgent to develop a platform to assess T cell response for neoantigen screening. Here, immunogenic potential of neoantigen-peptides of tumor tissues and matched organoids (n = 27 pairs) are analyzed by Score tools with whole genome sequencing (WGS)-based human leukocyte antigen (HLA)-class-I algorithms. The comparisons between 9203 predicted neoantigen-peptides from 2449 mutations of tumor tissues and 9991 ones from 2637 mutations of matched organoids demonstrate that organoids preserved majority of genetic features, HLA alleles, and similar neoantigen landscape of original tumors. Higher neoantigen load is observed in tumors with early stage. Multiomics analysis combining WGS, RNA-seq, single-cell RNA-seq, mass spectrometry filters out 93 candidate neoantigen-peptides with strong immunogenic potential for functional validation in five organoids. Immunogenic peptides are defined by inducing increased CD107aCD137IFN-γ expressions and IFN-γ secretion of CD8 cells in flow cytometry and enzyme-linked immunosorbent assay assays. Nine immunogenic peptides shared by at least two individuals are validated, including peptide from TP53R90S . Organoid killing assay confirms the antitumor activity of validated immunogenic peptide-reactive CD8 cells, which is further enhanced in the presence of immune checkpoint inhibitors. The study characterizes HLA-class-I neoantigen landscape in hepatobiliary tumor, providing practical strategy with tumor organoid model for neoantigen-peptide identification in personalized immunotherapy.