Thyroid cancer (TC) ranks as the most common endocrine malignancy, and its incidence and mortality rates continue to rise annually. Increasing evidence have shown that DNA methylation, a kind of stable epigenetic modification, is associated with carcinogenesis, suggesting its potential as biomarkers for the early detection of tumors. With the aim of exploring likely DNA methylation biomarkers for TC diagnosis, we conducted a synthetic analysis of DNA methylation profiles based on 789 samples from the Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA) databases. In the discovery phase, we identified five CpG probes (cg11228682, cg01291854, cg06778183, cg01668008, and cg01702055) on the condition of DNA methylation data from GSE86961 (n = 82) and constructed a five-CpG signature-based diagnostic model for TC. In addition, we validated the diagnostic score formula in two independent training cohorts, GSE97466 (n = 141) and TCGA (n = 566), as well as the previous developing cohort GSE86961. Receiver operating characteristic analysis revealed that the five-CpG signature had a good diagnostic performance to distinguish TC samples from benign samples. In conclusion, our findings suggest that the five-CpG signature could provide a novel biomarker with useful applications in TC diagnosis.
Keywords: CpG signature; DNA methylation profiling; diagnosis; thyroid cancer.