| Background and Objective:Esophageal squamous-cell carcinoma(ESCC)accounts for 80%of esophageal cancer cases worldwide and has a 5-year survival rate of less than 30%.Treating this disease at early stages generally results in better prognosis than at late stages,but effective biomarkers that aid early detection and/or accurate prognosis prediction are currently lacking.Aberrant DNA methylation plays an important role in cancer initiation and progression and have been investigated to derive diagnostic/prognostic biomarkers for several types of human cancer.For cancer detection,DNA methylation is also considered better than other genetic features.Early works usually focused on a small number of aberrantly methylated genes instead of performing genome-wide search.More recent studies either ignored the genomic and transcriptomic contexts or investigated them in a different set of patients,likely due to a lack of matched multi-omics data.Therefore,we explore genome-wide DNA methylation anomalies using a multi-omic approach in 91 Chinese ESCC patients for systematic screening of diagnostic and prognostic DMC markers.We also aim to identify the important roles of aberrant DNA methylations in ESCC development and progression.Methods:In the present study,we continued to profile genome-wide DNA methylation of the same samples with Illumina 450K methylation chip and correlate DMCs with a variety of gene expression alterations as well as somatic and germline variants specific to ESCC.Applying random-forest and LASSO to these DMCs generated a diagnostic model of 12 markers.We also constructed a prognostic model of 4 markers based on how strong they were associated with the overall survival(OS)time of ESCC patients.These two models were verified with independent external datasets.To further validate the function implications of marker host genes,we knocked down the expression of MMP13,YEATS2,HOXC10 and NECAB2 in vitro,one at a time.Results:We identified 35,577 differentially methylated CpG sites(DMCs)and characterized their distribution patterns.Integrating whole-genome DNA and RNAsequencing data of the same samples,we found that the methylation status of 3,241(9.11%)promoter or gene-body DMCs were significantly correlated with the expression levels of their host genes in ESCC(|Spearman’s correlation coefficient r|>0.30,FDR q<0.05).Among protein coding genes differentially expressed between tumor and adjacent normal samples,90 downregulated and 44 upregulated genes were associated with 224 hyper-methylated and 70 hypo-methylated CpG sites at their promoter regions,respectively;274 downregulated and 70 upregulated genes were associated with 764 hypo-methylated and 221 hyper-methylated CpG sites in their gene-bodies,respectively.Using featured DMCs,we developed a 12-marker diagnostic panel capable of distinguishing ESCC from the normal esophagus with AUC>0.96.We also established a 4-marker prognostic panel capable of classifying our patients as having high or low prognostic risk where the high-risk patients had significantly shorter median OS than others(12 versus 33 months,Plog-rank=1.74e-4).Applying this model to the TCGA ESCC sample yielded a similar result:the predicted high-risk patients had significantly longer median OS than low-risk patients(23 versus 42 months,Plog-rank=0.032).In-vitro experiments validated the functions of 4 marker host genes.Conclusions:Our characterization of genome-wide DNA methylation anomalies using a multi-omic approach in 91 Chinese ESCC patients has supported that aberrant DNA methylation is an important part of ESCC development and progression and has targeted a small number of potentially functional methylation CpG sites able to distinguish tumors from normal tissues or classify patients into high or low-risk groups.Overall,this study demonstrates that the ESCC genome abounds with specific DNA methylation patterns that could be effective diagnostic or prognostic biomarkers and potential mediators of tumor development and/or progression. |
PDF Full Text Request