Identification of clinical trait-related lncRNA and mRNA biomarkers with weighted gene co-expression network analysis as useful tool for personalized medicine in ovarian cancer.

RELEVANCE

The pathogenesis and biomarkers of ovarian cancer (OC) remain not well-known in diagnosis, effective therapy, and prognostic assessment in OC personalized medicine. The novel identified lncRNA and mRNA biomarkers from gene co-expression modules associated with clinical traits provide new insight for effective treatment of ovarian cancer.

PURPOSE

Long non-coding RNAs (lncRNAs) are relevant to tumorigenesis via multiple mechanisms. This study aimed to investigate cancer-specific lncRNAs and mRNAs, and their related networks in OCs.

METHODS

This study comprehensively analyzed lncRNAs and mRNAs with associated competing endogenous RNA (ceRNA) network and lncRNA-RNA binding protein-mRNA network in the OC tissues in the Cancer Genome Atlas, including 2562 cancer-specific lncRNAs ( = 352 OC tissues) and 5000 mRNAs ( = 359 OC tissues). The weighted gene co-expression network analysis (WGCNA) was used to construct the co-expression gene modules and their relationship with clinical traits. The statistically significant difference of identified lncRNAs and mRNAs was confirmed with qRT-PCR in OC cells.

RESULTS

An lncRNA-based co-expression module was significantly correlated with patient age at initial pathologic diagnosis, lymphatic invasion, tissues source site, and vascular invasion, and identified 16 lncRNAs (ACTA2-AS1, CARD8-AS1, HCP5, HHIP-AS1, HOTAIRM1, ITGB2-AS1, LINC00324, LINC00605, LINC01503, LINC01547, MIR31HG, MIR155HG, OTUD6B-AS1, PSMG3-AS1, SH3PXD2A-AS1, and ZBED5-AS1) that were significantly related to overall survival in OC patients. An mRNA-based co-expression module was significantly correlated with patient age at initial pathologic diagnosis, lymphatic invasion, tumor residual disease, and vascular invasion; and identified 21 hub-mRNA molecules and 11 mRNAs (FBN3, TCF7L1, SBK1, TRO, TUBB2B, PLCG1, KIAA1549, PHC1, DNMT3A, LAMA1, and C10orf82) that were closely linked with OC patients' overall survival. Moreover, the prognostic model of five-gene signature (OTUD6B-AS1, PSMG3-AS1, ZBED5-AS1, SBK1, and PLCG1) was constructed to predict risk score in OC patients. Furthermore, starBase bioinformatics constructed the lncRNA-miRNA-mRNA and lncRNA-RNA binding protein-mRNA networks in OCs.

CONCLUSION

These new findings showed that lncRNA-related networks in OCs are a useful resource for identification of biomarkers in OCs.