Alternations of gene expression in PI3K and AR pathways and DNA methylation features contribute to metastasis of prostate cancer.

OBJECTIVE

The molecular heterogeneity of prostate cancer (PCa) gives rise to distinct tumor subclasses based on epigenetic modification and gene expression signatures. Identification of clinically actionable molecular subtypes of PCa is key to improving patient outcome, and the balance between specific pathways may influence PCa outcome. It is also urgent to identify progression-related markers through cytosine-guanine (CpG) methylation in predicting metastasis for patients with PCa.

METHODS

We performed bioinformatics analysis of transcriptomic, and clinical data in an integrated cohort of 551 prostate samples. The datasets included retrospective The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) cohorts. Two algorithms, Least Absolute Shrinkage and Selector Operation and Support Vector Machine-Recursive Feature Elimination, were used to select significant CpGs.

RESULTS

We found that PCa progression is more likely to occur after the third year through conditional survival (CS) analysis, and prostate-specific antigen (PSA) was a better predictor of Progression-free survival (PFS) than Gleason score (GS). Our study first demonstrated that PCa tumors have distinct expression profiles based on the expression of genes involved in androgen receptor (AR) and PI3K-AKT, which influence disease outcome. Our results also indicated that there are multiple phenotypes relevant to the AR-PI3K axis in PCa, where tumors with mixed phenotype may be more aggressive or have worse outcome than quiescent phenotype. In terms of epigenetics, we obtained CpG sites and their corresponding genes which have a good predictive value of PFS. However, various evidences showed that the predictive value of CpGs corresponding genes was much lower than GpG sites in Overall survival (OS) and PFS.

CONCLUSIONS

PCa classification specific to AR and PI3K pathways provides novel biological insight into previously established PCa subtypes and may help develop personalized therapies. Our results support the potential clinical utility of DNA methylation signatures to distinguish tumor metastasis and to predict prognosis and outcomes.