An Integrative Analysis of DNA Methylation and Gene Expression to Predict Lung Adenocarcinoma Prognosis

Abnormal DNA methylation of gene promoters is an important feature in lung adenocarcinoma (LUAD). However, the prognostic value of DNA methylation remains to be further explored. We sought to explore DNA methylation characteristics and develop a quantifiable criterion related to DNA methylation to improve survival prediction for LUAD patients. Illumina Human Methylation450K array data, level 3 RNA-seq data and corresponding clinical information were obtained from TCGA. Cox regression analysis and the Akaike information criterion were used to construct the best-prognosis methylation signature. Receiver operating characteristic curve analysis was used to validate the prognostic ability of the DNA methylation-related feature score. qPCR was used to measure the transcription levels of the identified genes upon methylation. We identified a set of DNA methylation features composed of 11 genes (, , , , , , , , , and ). The feature score, calculated based on DNA methylation features, was independent of tumor recurrence and TNM stage in predicting overall survival. Of note, the combination of this feature score and TNM stage provided a better overall survival prediction than either of them individually. The transcription levels of all the hypermethylated genes were significantly increased after demethylation, and the expression levels of 3 hypomethylated proteins were significantly higher in tumor tissues than in normal tissues, as indicated by immunohistochemistry data from the Human Protein Atlas. Our results suggested that these identified genes with prognostic features were regulated by DNA methylation of their promoters. Our studies demonstrated the potential application of DNA methylation markers in the prognosis of LUAD.