Screening mitophagy-related lncRNAs based on WGCNA to identify and validate a prognostic signature for patients with ovarian cancer.

Abstract

e17554 Background: Mitophagy refers to the process by which cells selectively remove incomplete or damaged mitochondria through the autophagic mechanism. However, the value of mitophagy-related genes (MRGs) and mitophagy-related lncRNAs (MRLs) in ovarian cancer (OC) is still not clear. Methods: We explored the alterations of MRGs in four gynecological tumors using R packages. Based on weighted coexpression network analysis (WGCNA), MRLs were identified to carry out LASSO Cox regression analysis and then we established the prognostic MRL-signature. The prognosis and independent prognostic value of the MRL-signature was also validated by four independent datasets from GEO. The expression level of MRLs involved in the prognostic signature we established were detected by qPCR for the samples we collected. Analysis for characteristics of functional pathways, immunity status, and anti-tumor therapy were based on GSEA, ssGSEA, CIBERSORT, xCELL, MCPcounter, ESTIMATE, and TIDE. The ceRNA network was constructed applying miranda, miRWalk3.0, and Cytoscape. Results: Most MRGs were found at lower expression levels in four gynecological tumors. Among the four gynecological tumors, the levels of CNV and MRGs were positively correlated. Most MRGs were negatively correlated with methylation level. A total of 52 MRGs were differentially expressed in OC tissues. Nine MRLs were significantly associated with survival prognosis. Eight optimal lncRNA combinations were screened by LASSO Cox Regression algorithm. On the basis of Univariate Cox regression and Multivariate Cox regression analyses, the risk model based on MRLs was proved to be the independent prognostic factors. The qPCR assay showed that the differential expression trends of these MRLs in our collected OC samples were consistent with those in TCGA database. We found that TIDE score of OC patients with lower risk score was lower, suggesting that OC patients with low RSs are more sensitive to immune checkpoint blockade therapy. OC patients with higher risk score had higher immune cell scores and higher risk score correlated strongly with higher stromal and estimated scores, while lower risk score correlated with tumor purity. We found that the IC50 levels of paclitaxel and ABT.888 (Veliparib) in the high-risk group were observably higher than those in the low-risk group, suggesting a negative correlation between risk score and the drug susceptibility of OC. There were 539 miRNAs, 73 mRNAs and 8 lncRNAs in the ceRNA network. Conclusions: The integrative analysis of MRGs and MRLs revealed their roles in prognosis and therapeutic responsibility in chemotherapy and immunotherapy. Our findings demonstrate the important clinical significance of the signature based on MRLs, and will help improve clinical outcomes for patients with OC.