Abstract
BACKGROUND: Esophageal squamous cell carcinoma (ESCC) is an aggressive cancer with poor prognosis. We aimed to identify a panel of CpG methylation biomarkers for prognosis prediction of ESCC patients. METHODS: Illumina's GoldenGate methylation array, supervised principal components, Kaplan-Meier survival analyses and Cox regression model were conducted on dissected tumor tissues from a training cohort of 40 ESCC patients to identify potential CpG methylation biomarkers. Pyrosequencing quantitative methylation assay were performed to validate prognostic CpG methylation biomarkers in 61 ESCC patients. The correlation between DNA methylation and RNA expression of a validated marker, SOX17, was examined in a validation cohort of 61 ESCC patients. RESULTS: We identified a panel of nine CpG methylation probes located at promoter or exon1 region of eight genes including DDIT3, FES, FLT3, NTRK3, SEPT5, SEPT9, SOX1, and SOX17, for prognosis prediction in ESCC patients. Risk score calculated using the eight-gene panel statistically predicted poor outcome for patients with high risk score. These eight-gene also showed a significantly higher methylation level in tumor tissues than their corresponding normal samples in all patients analyzed. In addition, we also detected an inverse correlation between CpG hypermethylation and the mRNA expression level of SOX17 gene in ESCC patients, indicating that DNA hypermethylation was responsible for decreased expression of SOX17. CONCLUSIONS: This study established a proof-of-concept CpG methylation biomarker panel for ESCC prognosis that can be further validated by multiple cohort studies. Functional characterization of the eight prognostic methylation genes in our biomarker panel could help to dissect the mechanism of ESCC tumorigenesis.