Gene Expression Analysis Of Human Papillomavirus Positive Head And Neck Cancer Primary Tumor Samples Reveals Two Distinct Subgroups

Introduction: Human Papillomavirus (HPV) infection is recognized as a major cause for an increasing percent of oropharyngeal squamous cell carcinomas (OPSCC). Patients with HPV-induced OPSCC have a better prognosis than those who are HPV-negative, which implies distinct molecular mechanisms. However, predicting prognosis with the HPV-positive population remains elusive. In this study, we aimed to characterize molecular aspects of HPV expression, HPV integration, and host gene expression to later correlate with survival and recurrence as those data become available for our study population. Materials and methods: Thirty-six primary head and neck SCC (HNSCC) tumor samples were collected between 2011-2013 at University of Michigan hospital, of which 18 were determined to be HPV+ (type16, 18 or 35). Paired-end RNA-Seq libraries were generated using Illumina HiSeq for all samples. The sequences were aligned to both the human and HPV genome. Gene expression levels were quantified for both human and HPV genes using TopHat2 and HTSeq, and HPV integration events were determined using Virus-seq. Results: Hierarchical clustering of the gene expression levels showed three distinct clusters among the samples. The HPV- samples were first separated from the HPV+ samples, and then there were two robust subgroups within the HPV+ samples. Subsequent analysis revealed one subgroup had significantly higher HPV expression levels than the other (p-value: 0.03; Wilcoxon test). The subgroups were also highly correlated with HPV-integration status (8 of the 16 tumors had at least one identified HPV integration event). Differentially expressed genes between HPV-integration positive and negative samples were found to be enriched for interesting pathways, including “keratinocyte differentiation”, “cell adhesion” and “growth factor binding”, suggesting possible differences in carcinogenesis We also examined the HPV integration sites and found the majority of them were transcription factors. These also include several cancer-related genes, such as “BIRC3”, “CD274” and “PBX1”. One gene, KLF12, contained an integration event in one of our tumors and in one of the 36 HPV+ tumors from The Cancer Genome Atlas (TCGA). Network analysis of the genes harboring an integration event in our population or the TCGA HNSCC population (38 total genes) revealed interactions around p63 and ETS. Conclusion: Our data is an important addition to TCGA HNSCC cohort data, because as opposed to TCGA, we did not enforce selection bias towards large tumor size. With our data, we identified two distinct HPV+ subgroups characterized by RNA-Seq expression levels and strongly correlated with HPV viral expression levels and integration status. A significant portion of the HPV+ tumors (50%) had integrated HPV genome, implying important oncogenic potentials for integration events. Citation Format: Yanxiao Zhang, Lada A. Koneva, Shama Virani, Alisha Virani, Katie M. Rentschler, Thomas E. Carey, Laura S. Rozek, Maureen A. Sartor, the University of Michigan Head and Neck SPORE. Gene expression analysis of human papillomavirus positive head and neck cancer primary tumor samples reveals two distinct subgroups. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 2184. doi:10.1158/1538-7445.AM2015-2184