Computational discovery of non-mutational tumor-restricted antigens reveals evidence of immunoediting in head and neck squamous cell carcinoma.

Abstract

6564 Background: We previously identified 107 expression-based tumor antigens (EbTAgs) defined as genes with negligible expression in healthy tissue and overexpression in cancer. EbTAgs present novel targets for the adaptive anti-tumor immune response and exhibit evidence of immunoediting in highly immune infiltrated oral cavity tumors. To detail the landscape of EbTAgs in head and neck squamous cell carcinoma (HNSC) and further elucidate EbTAg immunoediting, we compared the expression EbTAgs in the context of tumor immune infiltration among four HNSCC subtypes: oral cavity (OC), HPV+ oropharyngeal (HPV+OP), HPV- oropharyngeal (HPV-OP), and laryngeal/hypopharyngeal (LH). Methods: Upper quartile FPKM gene expression values of all protein coding genes were calculated for all HNSC samples using RNAseq data from The Cancer Genome Atlas (TCGA). TCGA HNSC tumors were divided into subtypes and analyzed for EbTAg expression. Individual tumor sample immune infiltrate was determined using unsupervised clustering of 14 immune cell signature ssGSEA scores for the HNSC dataset as a whole and for each subtype. Results: LH tumors expressed significantly more EbTAgs than other subtypes (p=0.0014), specifically HPV+OP (p=0.0008, Tukey’s test). Immune clustering analysis showed that LH tumors were significantly more likely to be in the low than the high immune cluster whereas the reverse was true for HPV+OP tumors (p<0.0001). Hypothesizing that EbTAg expression was a function of tumor immune infiltration rather than HNSC subtype, we compared EbTAg expression between tumors in low and high immune clusters of the entire HNSC dataset as well as of each HNSC subtype. Significantly more EbTAgs were expressed in low immune tumors compared to high immune tumors of the HNSC dataset (p<0.0001). Similarly, significantly more EbTAgs were expressed in low immune tumors compared to high immune tumors of the OC, OP-all tumors, and HPV+OP datasets (p=0.0003, p<0.0001, p=0.0006) with a trend of more EbTAgs in the immune low versus high tumors in the HPV-OP and LH datasets (p=0.12, p=0.095). Conclusions: EbTAg expression in TCGA HNSC samples correlates with tumor immune infiltration resulting in lower expression under greater immunological pressure. These results reinforce the hypothesis that EbTAgs undergo immunoediting and are immunologically relevant. Exploration of EbTAgs as antigenic targets of modular vaccines or adoptive T-cell therapy as well as biomarkers of immune checkpoint inhibition therapy response is warranted.