Detection and validation of hypoxia-related lncRNAs with associated ceRNA network involved in HCC prognosis, treatment responsiveness

Abstract Long non-coding RNAs can regulate hypoxia-induced tumor immune microenvironment remodeling and tumor progression. The present study aimed to build a risk model for predicting the prognosis of hepatocellular carcinoma (HCC) patients based on hypoxia-related lncRNAs and to explore its possible mechanisms. Based on the RNA-Seq and follow-up data downloaded from TCGA and GEO, we performed correlation analysis, differential analysis, and survival analysis to derive candidate hypoxia-related lncRNAs. In total, the 365 patients were used to randomly divide into the training and testing dataset at a ratio of 7:3. The Least Absolute Shrinkage and Selection Operator (Lasso)-Cox regression was applied to select and develop hypoxia-related lncRNAs signature (HRLS) in training datasets. We obtained a risk model consisting of 8 hypoxia-related lncRNAs which was systematically validated in the testing and GSE76427 dataset. HRLS was developed based on SNHG3, NRAV, AC073611.1, AL031985.3, AL049840.6, ZFPM2-AS1, AC074117.1, and MAFG-DT. Patients with low risk displayed a good prognosis, whereas those with high risk had a poor prognosis. Multivariate COX regression analysis also confirmed that the HRLS group was statistically significant after adjusting for clinical factors. Nomogram, time-dependent ROC curve, and decision curve analyses were performed to confirm the predictive ability. Furthermore, the high-risk patients had high-level infiltration of Macrophages and regulatory T cells. Next, AC073611.1 and AL031985.3 were found to be significantly decreased in Hep3B cells after hypoxia exposure (48 hours) in the GSE155505 dataset. Also, the expression of MAFG-DT was significantly upregulated in the sorafenib treatment responders of the GSE109211 dataset. The abnormal expression may be associated with hypoxia states. Finally, we constructed a novel ceRNA network and predicted their binding sites. We developed and validated a novel HRLS to accurately predict patient survival, assess immune infiltration, infer therapeutic benefits, and provide a new perspective for designing personalized therapies.