P27-14EXPRESSED HPV INTEGRATION EVENTS IN HEAD AND NECK CANCER: WHERE THEY OCCUR AND THEIR EFFECT ON SURVIVAL AND MOLECULAR SIGNATURES

27. HPV and oropharynx / Head and neck cancer
M. Sartor 1, L. Koneva 1, Y. Zhang 1, S. Virani 1, P. Hall 1, J. Mchugh 1, D. Chepeha 1, G. Wolf 1, T. Carey 1, L. Rozek 1.
1University of Michigan (United States)

Background / Objectives

The incidence of human papillomavirus (HPV)-related oropharyngeal cancer has risen to epidemic levels in the United States. One common molecular event during HPV-related oncogenesis is full or partial integration of the HPV genome into the host genome. Expression of the integrated HPV early genes has clinically-relevant effects. However, the specific nature of where these integration events occur, their effect on patient survival, and the extent to which the switch to integrated HPV integration affects differentiation, the host immune response, and DNA copy number changes is unknown. Previously, our group characterized two HPV-related tumor subtypes, finding that they differed by HPV integration status, immune response, keratinization, EMT signatures, type and number of DNA copy number alternations, and PIK3CA mutations1. Interestingly, many of the differences could be explained by HPV integration.


Methods

In this study, we identified and characterized expressed HPV integration events from 84 HPV-positive oral cavity and oropharyngeal tumors (18 from the University of Michigan Hospital (UMH) and 66 from The Cancer Genome Atlas (TCGA)). Flash frozen tumor tissue was collected at UMH for 36 oral cavity and oropharyngeal cases, and H&E slides were assessed for degree of cellularity (minimum 70%) and necrosis (less than 10%). Upon mRNA-seq using 100nt paired-end reads on an Illumina HiSeq, half (18/36) were identified as HPV-positive. HPV integration events were detected using RNA-seq data, captured by HPV-host fusions transcripts. Raw HNSC TCGA data was downloaded and re-processed the same as for the UMH samples. Virus-seq was used to detect HPV-host fusion events, the edgeR R package was used to determine differential expression, and a Cox proportional hazards model was used for survival analysis.


Results

We found 320 virus-host integration breakpoints in 51 (61%) of the 84 samples. Integration events were strongly overrepresented near known head and neck, lung, and urogenital cancer genes, with five recurrent genes (including PD-L1). They were enriched in certain classes of repetitive regions, and a significant number of genes harboring an integration were found to interact with Tp63, ETS, or FOX1A. Patients with no detected integration had significantly better survival than those with a detected integration and HPV-negative patients.


Conclusion

Our results suggest that in HPV-related tumors of the upper aerodigestive tract, there is strong natural selection for cells with expressed HPV integration events in/near key oncogenes and tumor suppressors. The survival benefit of those patients having no expressed integration event provides a candidate cohort for de-escalated therapy.


References

[1] Zhang Y, Koneva LA, Virani S, Arthur AE, Virani A, Hall PB, Warden CD, Carey TE, Chepeha DB, Prince M, McHugh JB, Wolf GT, Rozek LS, Sartor MA. Subtypes of HPV-positive head and neck cancers are associated with HPV characteristics, copy number alterations, PIK3CA mutation, and pathway signatures. Clinical Can Res. 2016; 22(18):4735-45.