Abstract 636P
Background
Despite early clinical stage prostate cancer (PCa) postoperative disease progresses slowly with a favorable survival profile, approximately 20% of patients occur biochemical recurrences even with negative surgical margins. Current clinical predictors of diagnosis have limitations to accurately establish recurrence and progression risk.
Methods
The expression data of 422 tumor and normal samples of PCa from TCGA was used to identify differentially expressed mRNAs of genes by Wilcoxon test. The mRNA expression data and clinical data in four public datasets (DFKZ2018, GSE70796, GSE107299, GSE54460) were used for external validation. The time-dependent receiver operating characteristic (ROC) curves for 1-, 2-, and 3-year overall survival(OS) predictions of eight gene signature was compared with three commercial gene expression panels (OncotypeDx, Decipher and Prolaris). Gene Set Enrichment Analysis (GSEA) was used to further study the underlying molecular mechanisms.
Results
An 8-gene prognosis signature model was established for predicting PCa prognosis. Patients was characterized with high-risk and low-risk sets using the model, where prognosis was significantly worse in the high-risk group across cohorts (P < 0.05). Multivariate Cox regression analysis indicated that this signature was an independent prognostic factor of PCa survival. Nomogram including the prognostic signature was constructed and showed better predictive performance in short year survival. In TCGA whole cohort, the eight-gene signature had a comparable or superior performance to that of the three commercial panels, whose AUCs for 1-, 2-, and 3-year OS predictions for the risk scores were 0.737, 0.760, and 0.737, respectively. In the subgroup analysis of clinical low-risk group (PSA < 10 ng/ml), the relapse-free survival (RFS) of low-risk group was still longer than high-risk group (P = 0.00024). GSEA revealed enrichment of immunoglobulin complex related terms and B cell mediated immunity in the low-risk group.
Conclusions
The 8-gene signature was not inferior to other commercial panels for predicting short-term recurrence risk in PCa, which might provide meaningful information to therapeutic customization.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
T. Chen, J. Miao, H. Chen, M. Huang: Financial Interests, Personal and Institutional, Member: 3D Medicines Inc. All other authors have declared no conflicts of interest.