Translating transcriptome to immunophenotype in head and neck squamous cell carcinoma (HNSCC) to identify pathways promoting T-cell infiltration.

Abstract

e17542 Background: We sought to analyze the transcriptional landscape of HNSCC in an attempt to identify tumor-intrinsic oncogenic pathways that appear to mediate T-cell infiltration of tumor tissue. In this direction, we employ a methodology that integrates histopathology data of the tumor microenvironment with its corresponding transcriptome. Methods: 32 frozen HNSCCs were subjected to RNA-seq and corresponding FFPE were scored for plasma cells, tertiary lymphoid structures and CD8a+ TILs (center, invasive margin). RNA-seq data were analyzed to identify differentially expressed genes (DEGs) between tumors scored by immunohistochemistry (IHC) as CD8a high and CD8a low. Gene ontology analysis (GO) was performed based on DEGs > 1.5 fold expression change between CD8a high and CD8a low groups. Candidate genes were investigated by hierarchical clustering in TCGA RNA-seq data and further validated by IHC and quantitative RT-PCR in our cohort. Results: 32 HNSCCs were either scored as CD8a high or CD8a low based on IHC detection of CD8a+ cells in invasive margin of tumors. Comparative analysis of mRNA expression data between CD8a high and CD8a low groups in our cohort revealed that Muc1/16 overexpression and glycosylation was highly enriched in T cell infiltrated group of tumors. This finding was further validated using antibodies that detect glycosylated epitopes for both mucins. Analysis of TCGA RNA-seq data indicated that Muc1/16 overexpressing tumors share signatures of early keratinocyte differentiation and stem cell identity and co-express high levels of enzymes that promote Muc1/16 glycosylation. Interestingly, loss of CDH1 and acquisition of epithelial mesenchymal transition (EMT) markers in the cluster of Muc1/16 overexpressing tumors is strongly correlated with elevated CD8a, IDO1, CD274 and CXCL10 mRNA levels (P < 0.0001). Conclusions: Muc1/16 overexpressing tumors represent a very immunogenic HNSCC cluster. Previous studies have shown that mucins 1 and 16 in cancer cells expose glycosylated-specific epitopes that are recognized by T cells as cancer antigens. To this end, MUC1/16 expression may serve as predictive biomarkers for response to immunotherapy and MUC-targeted immunotherapy may function as an attractive partner to checkpoint inhibitors in HNSCC.