Pooled Analysis Of Methylome Data Identifies A DNA Methylation Signature Highly Predictive For HPV Status In Head And Neck Squamous Cell Carcinomas

Davide DEGLI ESPOSTI, International Agency for Research on Cancer, France
SKLIAS A. 1 , BEGHELLI-DE LA FOREST DIVONNE S. 2 , HERNANDEZ-VARGAS H. 1 , CROS M. 1 , CAHAIS V. 1 , CUENIN C. 1 , SUDAKA A. 2 , VAN OBBERGHEN-SCHILLING E. 2 , ZDENKO H. 1

1 Epigenetics Group, International Agency for Research on Cancer
2 University of Nice Sophia Antipolis, CNRSUMR7277-InsermU1091

Purpose. Oncogenic human papilloma viruses (HPV) have been found to be causally associated with a subset of head and neck squamous cell carcinomas (HNSCC). Several lines of evidence argue that some viruses including HPV may promote tumour development via an ‘‘epigenetic strategy’’, although the precise underlying mechanisms remains poorly understood. In this study we aimed to address whether HPV affects the DNA methylome in HNSCCs and if any specific methylation signature can be used for differentially diagnosis with HNSCCs associated with other aetiologies.
Methods. DNA methylation raw data from Illumina 450k platform of two different cohorts (TCGA and University College of London, UCL) were retrieved and analyzed together with a subset of cases from a French cohort for which we generated data from the same platform. A total of 326 cases were analyzed, of which 63 were HPV (mainly subtype 16) positive. Machine learning algorithms were used to identify a DNA methylation signature able to identify a HPV specific signature and the signature was tested on the entire cohort of French cases.
Results. After normalization of the data, global methylation profiles were able to distinguish HPV positive from HPV negative cases .We identified a signature of only 5 CpGs which was able to discriminate between the two groups of tumours with a sensitivity of 90-95% and a specificity of 98%. Conclusions. HPV infection is associated with a specific DNA methylation signature with higher predictive value of HPV status compared with current detection methods in HNSCCs. These results have an important relevance for the early diagnosis and treatment of these tumours. DNA methylation signatures may be used to identify and improve the understanding of the carcinogenic mechanisms of different environmental exposures.
Funding source. This research was funded by the INCA, PAIR VADS programme and by the Epigenetics Group at IARC.