IMMU-08. THE ANTIGENIC LANDSCAPE OF GLIOBLASTOMA - REFINING THE TARGETS FOR IMMUNOTHERAPY

Abstract

Listen

Translate article

Abstract We provide a comprehensive analysis of the antigenic landscape of glioblastoma using a multi-omics approach including ligandome mapping of the Human Leukocyte Antigen (HLA) ligandome, next generation sequencing (NGS) as well as an in-depth characterization of tumor-infiltrating lymphocytes (TIL) using mass cytometry and ultra-deep sequencing of the T-cell receptor (TCR). Tumor-exclusive HLA class I and class II ligands (immune precipitation and LC-MS/MS) of 24 isocitrate dehydrogenase 1 wild type glioblastoma samples and 10 autologous primary glioblastoma cell lines were defined in comparison to an HLA ligandome normal tissue reference database (n > 418). We found 11,496 glioblastoma exclusive HLA class I ligands (2,064 shared with cell lines; 3,754 on ≥ 2 glioblastoma samples). On the source protein level, 239 glioblastoma exclusive proteins were identified; among them 54 were also found in cell lines. For HLA class II ligands the analysis revealed 11,870 glioblastoma exclusive peptides (444 shared with cell lines; 3,420 on ≥ 2 glioblastoma samples) and 278 glioblastoma exclusive proteins; among which 18 were present also in cell lines. Moreover, whole-exome sequencing and whole RNA sequencing of 13 tumor samples was performed with the aim to predict neoantigens. On average 5,662 somatic missense effects were identified per patient (min: 4,258; max: 7,479). Candidate peptides are grouped into (i) in silico predicted neoepitopes, (ii) tumor-exclusivity on HLA, (iii) gene expression (e.g. cancer testis antigens). Top-ranking candidates from each group will be tested with regards to their immunogenicity in an autologous setting (TIL, peripheral blood mononuclear cells, patient derived tumor cells). Finally, the peptide and immunogenicity data is correlated with the immune phenotype of the TIL compartment as well as the TCR repertoire of the sample.