Six-long non-coding RNA signature predicts recurrence-free survival in hepatocellular carcinoma.

BACKGROUND Recent evidence shows that long non-coding RNAs (lncRNAs) are closely related to hepatogenesis and a few aggressive features of hepatocellular carcinoma (HCC). Increasing studies demonstrate that lncRNAs are potential prognostic factors for HCC. Moreover, several studies reported the combination of lncRNAs for predicting the overall survival (OS) of HCC, but the results varied. Thus, more effort including more accurate statistical approaches is needed for exploring the prognostic value of lncRNAs in HCC. AIM To develop a robust lncRNA signature associated with HCC recurrence to improve prognosis prediction of HCC. METHODS Univariate COX regression analysis was performed to screen the lncRNAs significantly associated with recurrence-free survival (RFS) of HCC in GSE76427 for the least absolute shrinkage and selection operator (LASSO) modelling. The established lncRNA signature was validated and developed in The Cancer Genome Atlas (TCGA) series using Kaplan-Meier curves. The expression values of the identified lncRNAs were compared between the tumor and non-tumor tissues. Pathway enrichment of these lncRNAs was conducted based on the significantly co-expressed genes. A prognostic nomogram combining the lncRNA signature and clinical characteristics was constructed. RESULTS The lncRNA signature consisted of six lncRNAs: MSC-AS1, POLR2J4, EIF3J-AS1, SERHL, RMST, and PVT1. This risk model was significantly associated with the RFS of HCC in the TCGA cohort with a hazard ratio (HR) being 1.807 (95% CI [confidence interval]: 1.329-2.457) and log-rank P-value being less than 0.001. The best candidates of the six-lncRNA signature were younger male patients with HBV infection in relatively early tumor-stage and better physical condition but with higher preoperative alpha-fetoprotein. All the lncRNAs were significantly upregulated in tumor samples compared to non-tumor samples (P < 0.05). The most significantly enriched pathways of the lncRNAs were TGF-beta signaling pathway, cellular apoptosis-associated pathways, etc. The nomogram showed great utility of the lncRNA signature in HCC recurrence risk stratification. CONCLUSION We have constructed a six-lncRNA signature for prognosis prediction of HCC. This risk model provides new clinical evidence for the accurate diagnosis and targeted treatment of HCC.