Qianwei Xing1, Shouyong Liu2, Jiaochen Luan2, Yi Wang3, Limin Ma4. 1. Department of Urology, Affiliated Hospital of Nantong University, Nantong 226001, Jiangsu Province, China. 2. Department of Urology, The First Affiliated Hospital of Nanjing Medical University, Nanjing 210029, Jiangsu Province, China. 3. Department of Urology, Affiliated Hospital of Nantong University, Nantong 226001, Jiangsu Province, China. Electronic address: wangyi_urology@163.com. 4. Department of Urology, Affiliated Hospital of Nantong University, Nantong 226001, Jiangsu Province, China. Electronic address: ntmalimin@163.com.
Abstract
BACKGROUND: Cancer precision medicine requires biomarkers or signatures to predict prognosis and therapeutic benefits. Driven by this, we established a biochemical recurrence (BCR) predictive model for prostate cancer (PCA) patients based on RNA-binding proteins (RBPs). METHODS: RNA-sequencing and corresponding clinicopathological data were downloaded from the Cancer Genome Atlas (TCGA) database and the Gene Expression Omnibus (GEO) database. Univariate COX, LASSO and multivariate COX regression analyses were carried out to develop the BCR predictive riskScore model. Survival analysis, ROC curve, independent prognostic analysis, nomogram were also performed to evaluate this signature internally and externally. RESULTS: A total of 13 RBPs including TRMT1L, WBP4, MBNL3, SMAD9, NSUN7, ENG9, PIWIL4, PEG10, CSDC2, HELZ2, CELF2, YBX2 and ESRP2 were eventually identified as BCR-related hub biomarkers and utilized to establish a riskScore. Further analysis including external and internal verification indicated that the patients with high riskScores had shorter time to BCR compared to those with low riskScores in both TCGA and GSE116918. The area under the curve (AUC) of the time-dependent receiver operator characteristic curve (ROC) of the predictive model exhibited a good predictive performance. The signature was also proven to be a valuable independent prognostic factor (all P < 0.05). We also established a nomogram based on the 13 RBPs to visualize the relationships between individual predictors and 1-, 3- and 5-year BCR for PCA. CONCLUSIONS: Our results successfully screened out 13 RBPs as a robust BCR-predictive signature in PCA by external and internal verification, helping clinician predict patients' cancer progression status and promoting the specific individualized treatment than original clinical parameters.
BACKGROUND: Cancer precision medicine requires biomarkers or signatures to predict prognosis and therapeutic benefits. Driven by this, we established a biochemical recurrence (BCR) predictive model for prostate cancer (PCA) patients based on RNA-binding proteins (RBPs). METHODS: RNA-sequencing and corresponding clinicopathological data were downloaded from the Cancer Genome Atlas (TCGA) database and the Gene Expression Omnibus (GEO) database. Univariate COX, LASSO and multivariate COX regression analyses were carried out to develop the BCR predictive riskScore model. Survival analysis, ROC curve, independent prognostic analysis, nomogram were also performed to evaluate this signature internally and externally. RESULTS: A total of 13 RBPs including TRMT1L, WBP4, MBNL3, SMAD9, NSUN7, ENG9, PIWIL4, PEG10, CSDC2, HELZ2, CELF2, YBX2 and ESRP2 were eventually identified as BCR-related hub biomarkers and utilized to establish a riskScore. Further analysis including external and internal verification indicated that the patients with high riskScores had shorter time to BCR compared to those with low riskScores in both TCGA and GSE116918. The area under the curve (AUC) of the time-dependent receiver operator characteristic curve (ROC) of the predictive model exhibited a good predictive performance. The signature was also proven to be a valuable independent prognostic factor (all P < 0.05). We also established a nomogram based on the 13 RBPs to visualize the relationships between individual predictors and 1-, 3- and 5-year BCR for PCA. CONCLUSIONS: Our results successfully screened out 13 RBPs as a robust BCR-predictive signature in PCA by external and internal verification, helping clinician predict patients' cancer progression status and promoting the specific individualized treatment than original clinical parameters.