Exploring the genetic differences between Primary and Metastatic Prostate Cancer using Bioinformatic Approaches: A Preliminary Study

Abstract

Screening of prostate cancer (PCa) by measuring prostate cancer antigen has proven beneficial in reducing the mortality and progression of prostate cancer. However, its level can be affected if patients are taking certain drugs and/or suffering from certain medical conditions, causing a false negative. This can lead to PCa being undetected, where when untreated can lead to metastatic prostate cancer (MPC). Hence, in this study, genetic differences between PCa and MPC were explored using bioinformatics approaches to predict potential biomarkers for MPC. The study was divided into two parts, where the first involves feature selection and principal component analysis to differentiate PCa and MPC based on mRNA gene expression. Additionally, top 20 mutated genes for MPC were determined using odds ratio (OR). In the second phase, a predictive model was built using outcome of the mRNA gene expression analysis. The results showed that the mRNA expression of 26 identified genes could differentiate between PCa and MPC. This was further corroborated by the predictive model, where a sensitivity and specificity of 0.616 and 0.017 respectively was achieved. While importance is placed on sensitivity over specificity, further improvements involving more data need to be made to increase the specificity rate. Additionally, genes such as PAG24, BOP1 and GRWD1 should be investigated further as both potential biomarkers as well as potential pathways in MPC progression, based on further protein-protein interaction analysis. OR and protein-protein interaction suggests that androgen signalling pathway may crosstalk with NF-κB signalling and breast cancer pathway. This preliminary study shows that bioinformatics approaches could aid in understanding MPC, which could lead to the discovery of novel targeted therapy and potential biomarkers.