Identification of potential CpG sites for oral squamous cell carcinoma diagnosis via integrated analysis of DNA methylation and gene expression.

Background Oral squamous cell carcinoma (OSCC) accounts for more than 90% of the oral carcinomas and has a high fatality rate. This study aimed to identify potentially diagnostic biomarkers of OSCC through integrated analysis of DNA methylation and gene expression profiles. Methods The DNA methylation profiles of OSCC patients from The Cancer Genome Atlas (TCGA) were analyzed to screen patients with CpG island methylator phenotype (CIMP) and investigate the relationship between CIMP and survival probability of OSCC patients. Differential methylation and expression analyses of the paired OSCC tumor and paracancerous samples from TCGA were performed. Logistic regression model was established, and the accuracy of this diagnostic model for OSCC was evaluated in validation sets from Gene Expression Omnibus (GEO). Results OSCC patients with CIMP had lower survival probability than those without CIMP. The cg02860732 and cg04342955 were determined as candidate diagnostic methylation sites for OSCC. Logistic regression model was established based on cg02860732 and cg04342955 showed relatively high diagnostic accuracy in OSCC. Conclusions A diagnostic model for OSCC was identified based on the methylation sites cg02860732 and cg04342955, which might be favorable for the diagnosis of OSCC.