Abstract PR13: A new high-performance HLA ligand identification strategy enables prediction of T-cell tolerance to neoepitopes

Abstract

T-cell responses against neoepitopes presented by human leukocyte antigen (HLA) complexes represent a critical effector of anticancer immunity. However, detection of neoepitopes by mass spectrometry is still challenging as is identification of the fraction of neoepitopes that elicit immune responses in vitro and in vivo. To address these problems, we developed a strategy to identify HLA ligands combining the peptide identification algorithm Byonic with the epitope binding predictor netMHCpan. We obtained up to 4-fold increases in unique HLA ligand identifications compared to standard approaches, false discovery rates below 0.6% and over 17,000 unique HLA ligand identifications in a single experiment. Even with only 10 million cancer cells we were able to detect over 5,000 unique HLA ligands, which allows our approach to be applied to small tumor samples in a clinical setting. Of note, phosphorylated as well as glycosylated HLA ligands were identified with the same accuracy and allowed a better characterization of the rules for presentation of post-translationally modified HLA ligands. Furthermore, by reanalyzing mass spectrometry samples with our novel approach, we confirmed high-confidence, rejected low-confidence and identified additional neoepitopes including the first-time detection of a phosphorylated neoepitope. Finally, we used our broadened knowledge of the immunopeptidome to create rules for prediction of non-immunogenicity of neoepitopes based on high biochemical similarity with unmutated HLA ligands, which showed high specificity and positive predictive value when validated with two datasets derived from neoepitope-based clinical studies. Altogether, our methods substantially improved sensitivity and specificity for detection of native, modified and mutated HLA ligands, propose an explanation for tolerance to neoepitopes, and should in combination facilitate the design of neoepitope-based therapies. Citation Format: Martin G. Klatt, Ron S. Gejman, Sung S. Moon, Tatyana S. Korontsvit, Tao Dao, David A. Scheinberg. A new high-performance HLA ligand identification strategy enables prediction of T-cell tolerance to neoepitopes [abstract]. In: Proceedings of the Fourth CRI-CIMT-EATI-AACR International Cancer Immunotherapy Conference: Translating Science into Survival; Sept 30-Oct 3, 2018; New York, NY. Philadelphia (PA): AACR; Cancer Immunol Res 2019;7(2 Suppl):Abstract nr PR13.