An Integrated miRNA-lncRNA Signature Predicts the Survival of Stage II Colon Cancer.

Objective: By in silico analysis of colon cancer data from the Cancer Genome Atlas (TCGA) database, we develop a prognostic signature to improve the stratification of high-risk stage II colon cancer patients.
Method: RNA sequencing (RNA-Seq) data from 187 stage II colon cancer patients was obtained from the TCGA data portal. We excluded cases without a sufficient amount of survival data (n=21), leaving 166 stage II colon cancer patients to be selected for further survival analysis. Differentially expressed lncRNAs and miRNAs were unveiled by the edgeR package of R. A comprehensive ceRNAs regulatory network was constructed using the Cytoscape. Cox regression analysis was performed to screen prognostic RNAs and develop a prognostic signature. The Multi Experiment Matrix and Gene Ontology were undertaken to assess the prognostic lncRNA MIR31HG function.
Results: The multivariate analysis indicates that 2 lncRNAs (WASIR2 and MIR31HG) and 2 miRNAs (hsa-mir-200a and hsa-mir-155) exhibited an independently significant prognostic value for stage II colon cancer. The 4 lncRNA-miRNA signatures for predicting the overall survival (OS) was constructed with the formula: Risk score=exp WASIR2*(0.213)+exp MIR31HG*(0.152)+exp hsa-mir-200a*(-0.329)+exp hsa-mir-155*(0.300). The area under the curve in the receiver operating characteristic analysis was 0.810. Kaplan-Meier survival curves confirm that the low-risk group had a low death rate, with a 5-year OS rate at 87.7%. However, the high-risk group had a low 5-year OS of 23.1% ( P =0.000). The correlative genes of MIR31HG were found to be enriched in the epithelial-to-mesenchymal transition pathway, and the VEGFR3 signaling in lymphatic endothelium pathways.
Conclusions: These findings indicate that the 4 lncRNA-miRNA prognostic signature could be a marker for survival of stage II colon cancer patients.
(© 2019 by the Association of Clinical Scientists, Inc.)