Abstract 4511: Characterization of tumor neoantigen repertoire in patients with germline BRCA1/2 mutations

Abstract

Abstract Tumor neoantigens are peptides unique to the cancer genome that can induce endogenous T-cell activity and contribute to the efficacy of immunotherapy. We sought to characterize the neoantigen repertoire in 20 breast and 19 ovarian tumors from patients with germline BRCA1/2 mutations. Loss of BRCA1/2 function leads to genomic instability and increased mutational burden, a key correlate with clinical benefit from immunotherapy. To determine patient class I Human Leukocyte Antigen (HLA) haplotypes, whole exome sequencing data were analyzed using OptiType. Class I HLA alleles were typed to four-digit accuracy and were used in the prediction of tumor neoantigens using NetCTLpan 1.1. Briefly, non-synonymous variants from tumor exome data were translated to peptides of length 15, 17, and 19 amino acids centered on the variant of interest, where NetCTLpan analyzed all 8-, 9-, and 10-mer amino acid substrings, respectively, within the peptide sequences to predict HLA allele-specific binding affinities. IC50 values, predicted probability of binding to the TAP transporter, and probability of C-terminal cleavage were used to prioritize candidate neoepitopes of interest. When considering peptides with a predicted affinity less than 500nM, the minimum criterion for a likely binding occurrence in vivo, neoantigen loads varied extensively across both breast and ovarian patients, ranging from a minimum of 10 predicted binders in patients with the lowest neo-antigen load to greater than 300 predicted binders in patients with the highest load (mean 87, std. dev. 71). 10 of 20 breast cancer patients and 13 of 19 ovarian cancer patients harbored a moderate neoantigen load of 50-150 predicted neoantigens. In comparison, tumor types with low neoantigen load (e.g. acute myeloid leukemia) and high neoantigen load (e.g. melanoma) exhibit 8 and 480 predicted neoantigens on average, respectively, consistent with their missense mutational burden. Positive correlates of neoantigen load in our studies included mutational burden (p<0.01), genome gain/loss, tumor ploidy, and BRCA1/2 loss of heterozygosity (p<0.05), all related markers of genomic instability. These relationships will be further validated in an independent sample set of germline BRCA1/2 mutation carriers from the TCGA, and expression levels of these predicted neoantigens will be investigated by RNASeq. Further, additional neoepitope prediction tools will be used to generate a consensus list of top-ranking candidate neoepitopes with high predicted immunogenic potential, which will be functionally evaluated for T-cell reactivity. Our work demonstrates that breast and ovarian cancer patients with BRCA1/2 mutations vary substantially in terms of predicted immunogenicity and suggests that a subset of such patients may be responsive to immunotherapy. Citation Format: Adam A. Kraya, Kara N. Maxwell, Brandon M. Wenz, Bradley Wubbenhorst, Daniel DeSloover, Nicole Lunceford, Jennifer D. Morrissette, Michael Feldman, Robert H. Vonderheide, Susan M. Domchek, Katherine L. Nathanson. Characterization of tumor neoantigen repertoire in patients with germline BRCA1/2 mutations. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 4511.