Published online Nov 7, 2020. doi: 10.3748/wjg.v26.i41.6414
Peer-review started: July 20, 2020
First decision: August 8, 2020
Revised: August 17, 2020
Accepted: September 10, 2020
Article in press: September 10, 2020
Published online: November 7, 2020
Gastric cancer (GC) is a heterogeneous disease with genetic and epigenetic alterations. Robust biomarkers for management and survival prognosis of GC patients are lacking. Deoxyribonucleic acid (DNA) methylation is a major epigenetic event that participates in early stage of GC and is suggested to be associated with survival in many cancers including GC.
Exploring molecular subtypes of GC can improve understanding of this heterogeneous cancer and contribute to better management and prognosis prediction. Studies on DNA methylation subtypes of GC are lacking.
To identify the specific DNA methylation sites that influence the prognosis of GC patients by integrating epigenetic and clinical information. We also aimed to establish a prognostic model based on subtypes of DNA methylation.
Data of GC patients were obtained from The Cancer Genome Atlas and the University of California Santa Cruz cancer browser. Prognostic DNA methylation sites were identified by integrating DNA methylation profiles and clinical data. We used unsupervised clustering to identify distinct subgroups based on methylation status. A risk score model was built and further validated in a test set.
In this study, we identified three subtypes based on DNA methylation profiles using methylation sites that were significantly associated with survival. These methylation subtypes were associated with clinical features, patient outcomes, and potential responses to therapy. Enrichment analysis of specific hyper- or hypomethylation sites revealed that they were mainly involved in pathways related to carcinogenesis and tumor growth and progression. A prognostic model consisting two methylation sites was subsequently generated. The high-risk group showed a significantly poorer prognosis compared to the low-risk group in both the training (hazard ratio = 2.24, 95% confidence interval: 1.28-3.92, P < 0.001) and test (hazard ratio = 2.12, 95% confidence interval: 1.19-3.78, P = 0.002) sets. More samples are needed to optimize the model performance.
This study indicates that DNA methylation-based classification reflects the epigenetic heterogeneity of GC. A prediction model based on methylation subtypes can predict the OS of GC patients.
Our study can help predict prognosis and increase our understanding of the heterogeneity of GC patients. This is a retrospective analysis of GC patients from public database, so prospective studies are needed to validated the findings.