Identification of DNA Methylation Signatures for Diagnosis of Lung Adenocarcinoma

Abstract

Lung adenocarcinoma is the leading cause of death among men and women with cancer worldwide. Here, we performed an analysis of Illumina HumanMethylation450K data from TCGA to identify DNA methylation markers for lung adenocarcinoma diagnosis. We examined the DNA methylation landscape of lung adenocarcinoma and investigated the relationship between DNA methylation and clinical features. We then extracted differentially methylated cytosines in CpG island promoter regions, and then adopted machine learning techniques to determine the final methylation markers. As a result, we identified three methylation subtypes of lung adenocarcinoma, and found that the methylation status was not significantly related with the prognosis of lung adenocarcinoma. We finally identified two novel lung adenocarcinoma methylation markers including cg14823851 (TBX4) and cg07792478 (MIR124-2) with the AUCs of 100%, 100%, 98.3%, and 100% on support vector machine, logistic regression, decision tree, and random forest, respectively. Overall, our study demonstrates the potential use of methylation markers in lung adenocarcinoma diagnosis and may boost the development of new epigenetic therapies.