TCGA data reveal a transcriptomic signature of Sox2 and Nanog targets associated with survival in head and neck cancer.

Abstract

e17546 Background: Head and neck cancers represent a diverse group of tumors linked with HPV infection, alcohol and smoking that contribute to cancer morbidity and mortality globally. Sox2 and Nanog are transcription factors that maintain pluripotency in embryonic stem cells, and their individual expression has been associated with prognosis in head and neck tumors previously. Methods: TCGA dataset containing 522 tumor samples with RNAseq data, clinical information (anatomically classified as 133 oral tongue, 117 laryngeal, 73 oral cavity, 63 floor of mouth, 45 tonsillar, 27 base of tongue and 70 other samples) and 44 normal samples with RNAseq data available was used to detect differentially expressed genes in head and neck tumors. The levels of each gene were then transformed to z-score per patient and classified as over- or under-expressed if those were 2 SD away from the mean. The expression levels were then correlated with survival in individual patients for all tumor combined. Subgroup analysis of oral tongue, laryngeal, oral cavity, floor of mouth, tonsillar cancers was carried out separately. Results: A total of 233 genes were significantly correlated with survival in head and neck tumors ( < 0.01). This geneset was subsequently analyzed for enrichment of common pathway dysregulation or for enrichment of shared regulatory elements. The expression level of a set of a total of 15 genes that represent targets of Nanog and Sox2 was shown to either increase or decrease survival directly dependent on either up- or down-regulated expression levels. Of note, neither Nanog nor Sox2 expression levels individually impacted survival. Subgroup analysis revealed individualized prognostic signatures in each anatomic location, with an overlap of individual genes with Nanog and Sox2 targets signature ranging from 6.25-20%. Conclusions: We describe a prognostic signature in head and neck tumors derived from a TCGA data consisting of genes that represent targets of critical stem cell regulators Sox2 and Nanog. On a subgroup analysis, we also derived prognostic signatures for five anatomic locations and head&neck tumors with various ranges of overlap with Sox2 and Nanog targets signature. Pending validation in additional datasets, those represent an attractive target for further mechanistic evaluation in tumorigenesis in this group of tumors.