IMMU-06. Landscape of adaptive immunity of childhood brain cancers

T lymphocytes have a unique ability to recognize a vast array of antigens prompted by an enormous T cell receptor (TCR) repertoire. Characterization of tumor-infiltrating T cells (TILs) is key to understand MHC-restricted anti-tumor immunity and for developing T cell-centered immunotherapies such as adoptive cell therapy and tumor vaccines. In the current work, we investigated RNA-Seq data from 997 pediatric brain tumor patients and performed a large-scale comprehensive examination of the immunogenomic and TCR landscape of TILs across the entire spectrum of pediatric brain tumors. We show that the relative ratio between T cell diversity (clonality) and T cell abundance within each sample, represented by the clonal expansion index (CEI), is a strong predictor of prognosis both within and between tumor types. Interestingly, we show that CEI was strongly associated with molecular subgroups of medulloblastoma but not with known tumor-genomic features of these subgroups. Investigation of TCR clones recognizing a common recurrent tumor-antigen across patients based on CDR3 homology and characteristics, reveals 9 TCR clusters which are tumor type restricted with defined prognoses and HLA dominance. Using computational immunogenomics and machine learning-based investigations of these clusters yielded novel putative HLA-restricted tumor antigens which could bind and activate the clusters’ specific TCRs. Importantly, our framework grounded the foundations for developing a precision medicine approach of T cell-centered immunotherapies. These findings have major implications for understanding the interplay between T cell and tumor genomic, and for developing new immunotherapies for children with brain tumors.