Construction and Validation of a Robust Cancer Stem Cell-Associated Gene Set-Based Signature to Predict Early Biochemical Recurrence in Prostate Cancer

BACKGROUND: Postoperative early biochemical recurrence (BCR) was an essential indicator for recurrence and distant metastasis of prostate cancer (PCa). The aim of this study was to construct a cancer stem cell- (CSC-) associated gene set-based signature to identify a subgroup of PCa patients who are at high risk of early BCR. METHODS: The PCa dataset from The Cancer Genome Atlas (TCGA) was randomly separated into discovery and validation set. Patients in discovery set were divided into early BCR group and long-term survival group. Propensity score matching analysis and differentially expressed gene selection were used to identify candidate CSC-associated genes. The LASSO Cox regression model was finally performed to filter the most useful prognostic CSC-associated genes for predicting early BCR. RESULTS: By applying the LASSO Cox regression model, we built a thirteen-CSC-associated gene-based early BCR-predicting signature. In the discovery set, patients in high-risk group showed significantly poorer BCR free survival than that patients in low-risk group (HR: 4.91, 95% CI: 2.75–8.76, P < 0.001). The results were further validated in the internal validation set (HR: 2.99, 95% CI: 1.34–6.70, P = 0.005). Time-dependent ROC at 1 year suggested that the CSC gene signature (AUC = 0.800) possessed better predictive value than any other clinicopathological features in the entire TCGA cohort. Additionally, survival decision curve analysis revealed a considerable clinical usefulness of the CSC gene signature. CONCLUSIONS: We successfully developed a CSC-associated gene set-based signature that can accurately predict early BCR in PCa cancer.