Identification of a DNA methylome profile of esophageal squamous cell carcinoma and potential plasma epigenetic biomarkers for early diagnosis

Abstract

DNA methylation is a critical epigenetic mechanism involved in key cellular processes. Its deregulation has been linked to many human cancers including esophageal squamous cell carcinoma (ESCC). This study was designed to explore the whole methylation status of ESCC and to identify potential plasma biomarkers for early diagnosis. We used Infinium Methylation 450k array to analyze ESCC tissues (n = 4), paired normal surrounding tissues (n = 4) and normal mucosa from healthy individuals (n = 4), and combined these with gene expression data from the GEO database. One hundred and sixty eight genes had differentially methylated CpG sites in their promoter region and a gene expression pattern inverse to the direction of change in DNA methylation. These genes were involved in several cancer-related pathways. Three genes were validated in additional 42 ESCC tissues and paired normal surrounding tissues. The methylation frequency of EPB41L3, GPX3, and COL14A1 were higher in tumor tissues than in normal surrounding tissues (P < 0.017). The higher methylation frequency of EPB41l3 was correlated with large tumor size (P = 0.044) and advanced pT tumor stage (P = 0.001). The higher methylation frequency of GPX3 and COL14A1 were correlated with advanced pN tumor stage (P = 0.001 and P < 0.001). The methylation of EPB41L3, GPX3, and COL14A1 genes were only found in ESCC patients' plasma, but not in normal individuals upon testing 42 ESCC patients and 50 healthy individuals. Diagnostic sensitivity was increased when methylation of any of the 3 genes were counted (64.3% sensitivity and 100% specificity). These differentially methylated genes in plasma may be used as biomarkers for early diagnosis of ESCC.

Figure 1. Boxplots of log ratio of 12 samples' raw microarray data.

After the log₂ transfer and normalization, all samples showed very similar signal distributions.

Figure 2. Unsupervised clustering of samples.

Unsupervised hierarchical clustering on the entire DNA methylation data set showed the 2 clusters which were mainly composed by two different phenotypes: tumor and normal (including normal and adjcent normal tissues). This indicated that ESCC and normal samples might have different DNA methylation characters and patterns. In addition, there were some differences in DNA methylation profiles between adjacent normal and normal tissues.

Figure 3. Combination analysis between DNA methylation data and gene expression data identified 168 differentially methylated genes in ESCC.

Methylation data analysis showed 66,857 DMCs, which related to 6,200 genes (gene set 1). Expression data analysis showed 1,601 probe differences between ESCC and control, which related to 1.256 genes (gene set 2). There were 294 common genes between gene set 1 and gene set 2. 168 genes were selected due to inverse correlation between DNA methylation and expression change.

Figure 4. Unsupervised gene cluster of 168 genes.

The heat map of these genes shows different methylation between tumor tissues and normal tissues. Red indicates hypermethylation, while green indicates hypomethylation.

Figure 5. Representative results of MSP analysis of the candidate genes.

M, markers; N, negative control (water blank); P, positive control (universal methylated DNA); T, tumor tissue; S, paired surrounding normal tissue; TBP, blood plasma of tumor patients. Samples were scored as methylated when there was a clearly visible band with the methylated primers.