| dc.description.abstract | Personalized cancer vaccine strategies directed at tumor neoantigens derived from somatic mutations in the DNA are currently under prospective evaluation (Ott et al. 2017; Sahin et al. 2017). Alterations in tumor RNA, rather than DNA, may also represent a previously-unexplored source of neoantigens. Here, we show that intron retention, a widespread feature of cancer transcriptomes (Dvinge and Bradley 2015; Jung et al. 2015), represents a novel source of tumor neoantigens. We developed an in silico approach to identify retained intron neoantigens from RNA sequencing data and applied this methodology to tumor samples from patients with melanoma treated with immune checkpoint blockade (Snyder et al. 2014; Hugo et al. 2016), discovering that the retained intron neoantigen burden in these samples augments the DNA-derived, somatic neoantigen burden. We validated the existence of retained intron derived neoantigens by implementing this technique on cancer cell lines with mass spectrometry-derived immunopeptidome data (Barretina et al. 2012; Ritz et al. 2016), revealing that retained intron neoantigens were complexed with MHC I experimentally. Unexpectedly, we observed a trend toward lack of clinical benefit from immune checkpoint blockade in high retained intron load-tumors, which harbored transcriptional signatures consistent with cell cycle dysregulation and DNA damage repair. Our results demonstrate the contribution of transcriptional dysregulation to the overall burden of tumor neoantigens, provide a foundation for augmenting personalized cancer vaccine development with a new class of tumor neoantigens, and demonstrate how global transcriptional dysregulation may impact selective response to immune checkpoint blockade. | |
