Construction of an eight-gene signature for survival evaluation of papillary thyroid cancer patients

Background: Papillary thyroid carcinoma (PTC), accounting for 80% of all thyroid cancer cases, is the most frequent type of thyroid cancer. It is critical in identifying novel disease biomarkers, thus improving PTC patient management and making tailored therapeutic decisions. Methods: RNA-sequencing and survival data of PTC patients and normal thyroid tissues were retrieved from the Cancer Genome Atlas (TCGA). Univariate Cox analysis was utilized to investigate the relationship between expression levels of each differentially-expressed gene and survival. Multivariate Cox analysis was utilized for genes significantly associated with overall survival. With the co-efficient values of multivariate analysis, a gene prognostic signature was constructed by calculating risk scores for each subject. Results: A total of 1,757 genes were selected as significant differentially-expressed genes in PTC, compared to normal thyroid tissues. Fifteen genes were shown to be significantly associated with overall survival of PTC patients, according to univariate analysis. Multivariate Cox proportional hazards regression analysis recognized 8 differentially-expressed genes remarkably correlated with overall survival. Weighted by the corresponding co-efficient, an eight-gene prognostic signature was constructed: risk score = (0.3930)*Exp METTL7B + (-0.5943)*Exp ADRA1B + (-0.6605)*Exp RIPPLY3 + (-0.5314)*Exp FAM111B + (0.5030)*Exp PCOLCE2 + (-0.7236)*Exp LINC01208 + (-0.5016)*Exp ZSCAN4 + (0.4953)*Exp SALL3. The eight gene signature exhibited strong predictive power for 5-year survival of PTC patients. The area under the time-dependent ROC curve was 0.969. Conclusion: An eight-gene signature was identified. It can be used an independent prognostic marker, robustly predicting survival of PTC patients.