A large panel of molecular biomarkers have been identified to predict the prognosis of hepatocellular carcinoma (HCC), yet with limited clinical application due to difficult extrapolation. We here generated a genetic risk score system comprised of 12 HCC-specific genes to better predict the prognosis of HCC patients. Four genomics profiling datasets (GSE5851, GSE28691, GSE15765 and GSE14323) were searched to seek HCC-specific genes by comparisons between cancer samples and normal liver tissues and between different subtypes of hepatic neoplasms. Univariate survival analysis screened HCC-specific genes associated with overall survival (OS) in the training dataset for next-step risk model construction. The prognostic value of the constructed HCC risk score system was then validated in the TCGA dataset. Stratified analysis indicated this scoring system showed better performance in elderly male patients with HBV infection and preoperative lower levels of creatinine, alpha-fetoprotein and platelet and higher level of albumin. Functional annotation of this risk model in high-risk patients revealed that pathways associated with cell cycle, cell migration and inflammation were significantly enriched. In summary, our constructed HCC-specific gene risk model demonstrated robustness and potentiality in predicting the prognosis of HCC patients, especially among elderly male patients with HBV infection and relatively better general conditions.

Growing evidence indicates that long non-coding RNAs (lncRNAs) may be potential biomarkers and therapeutic targets for many disease conditions, including cancer. In this study, we constructed a risk score system of three lncRNAs (LOC101927051, LINC00667 and NSUN5P2) for predicting the prognosis of small hepatocellular carcinoma (sHCC) (maximum tumor diameter ≤5 cm). The prognostic value of this sHCC risk model was confirmed in TCGA HCC samples (TNM stage I and II). Stratified survival analysis revealed that the suitable patient groups of the sHCC lncRNA-signature included HBV-infected and cirrhotic patients with better physical conditions yet lower levels of albumin and higher levels of alpha-fetoprotein preoperatively. Besides, Asian patients with no family history of HCC or history of alcohol consumption can be predicted more precisely. Molecular functional analysis indicated that PYK2 pathway was significantly enriched in the high-risk patients. Pathway enrichment analysis indicated that the two lncRNAs (LINC00667 and NSUN5P2) associated with poor prognosis were closely related to cell cycle. The nomogram based on the lncRNA-signature for RFS prediction in sHCC patients exhibited good performance in recurrence risk stratification. In conclusion, we identified a novel three-lncRNA-expression-based risk model for predicting the prognosis of sHCC.

Pancreatic cancer (PC) is a highly malignant cancer with poor prognosis and high mortality. Aberrant DNA methylation plays a critical role in the occurrence, progression and prognosis of malignant tumors. In this study, we employed multiple datasets from APGI, TCGA and GEO to perform Multi-Omics analysis, including DNA methylation and expression profiling analysis. Three differentially expressed genes (SULT1E1, IGF2BP3, MAP4K4) with altered status of DNA methylation were identified and then enrolled into prognostic risk score model using LASSO regression. Univariate cox regression analysis indicated that high risk score was significantly associated with poor prognosis. Multivariate cox regression analysis proved the risk score was an independent prognostic factor for PC. In addition, time-dependent ROC curves indicated good performance of our model in predicting the 1-, 3- and 5-year survival of PC patients. Besides, stratified survival analysis revealed that the risk score model had greater prognostic value for patients of late stage with T3/T4 and N+. Pathway enrichment analysis suggested that these three genes might promote tumor progression by affecting signaling by Rho GTPases and chromosome segregation. In summary, three hypomethylated gene signature were significantly associated with patients' overall survival, which might serve as potential prognostic biomarkers for PC patients.

Non-small cell lung cancer (NSCLC) is the most common cancer and cause of cancer-related mortality globally. Increasing evidence suggested that the long non-coding RNAs (lncRNAs) were involved in cancer-related death. To explore the possible prognostic lncRNA biomarkers for NSCLC patients, in the present study, we conducted a comprehensive lncRNA profiling analysis based on 1902 patients from Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA) datasets. In the discovery phase, we employed 682 patients from the combination of four GEO datasets (GSE30219, GSE31546, GSE33745 and GSE50081) and conducted a seven-lncRNA formula to predict overall survival (OS). Next, we validated our risk-score formula in two independent datasets, TCGA (n=994) and GSE31210 (n=226). Stratified analysis revealed that the seven-lncRNA signature was significantly associated with OS in stage I patients from both discovery and validation groups (all P<0.001). Additionally, the prognostic value of the seven-lncRNA signature was also found to be favorable in patients carrying wild-type KRAS or EGFR. Bioinformatical analysis suggested that the seven-lncRNA signature affected patients' prognosis by influencing cell cycle-related pathways. In summary, our findings revealed a seven-lncRNA signature that predicted OS of NSCLC patients, especially in those with early tumor stage and carrying wild-type KRAS or EGFR.