Construction And Evaluation Of A Prognosis Lncrna Model For Hepatocellular Carcinoma

Current studies indicate that long non-coding RNA (lncRNA) is often abnormally expressed in hepatocellular carcinoma (HCC). We intend to generate a multi-lncRNA signal to improve the prognosis of HCC. By analyzing 12 pairs of HCC and adjacent normal mucosal tissues, 3900 differentially expressed lncrnas were identified as candidate biomarkers for the prognosis of HCC. Then, the 12-lncrna signature was constructed using the LASSO Cox regression method and verified in the TCGA training dataset. Finally, we established a novel 12-lncrna signature that was significantly associated with overall survival (OS) in the training data set. With the use of 12-lncrna markers, patients in the training cohort were divided into high-risk and low-risk groups with significant OV differences (P < .0001). Similar results were consistent in the TCGA verification dataset (P = .046). Multivariate Cox model was used to analyze and construct the risk scores of selected key lncRNA and AJCC stages. The results showed that, compared with AJCC stages, lncRNA-based risk scores were another important factor affecting the OS of patients. We found that risk scores based on lncRNA have a stronger prediction ability than the AJCC stage alone on 4-year OS. For 4-year survival rates, prediction combined with the lncRNA risk score and AJCC stage, model effectiveness (sensitivity and specificity) has reached to 0.750. To further explore the biological processes involved in prognostic lncRNA, all HCC samples in TCGA are divided into two groups according to the median lncRNA risk score, and analyzed the gene enrichment of high expression genes and low expression genes in KEGG data using goana in limma. The results suggest that the genes associated with tumor pathways, such as PI3K-Akt and ECM-receptor interaction, are highly expressed in the high risk group.