Genetic and genomic technologies and diagnosis of aggressive prostate cancer

Abstract

Prostate cancer (PCa) is the second-most commonly diagnosed cancer among men worldwide. In the year 2012 itself, approximately 1.1 million men were diagnosed with PCa, accounting for 15% of all the new cancer cases registered in men worldwide. With the alarming estimate that at least 1 in 6 PCa patients is in the risk of developing aggressive form of the disease, the identification of a predictive biomarker for PCa is of much importance. More so for New Zealand’s perspective, because the highest recorded rate of men with PCa relative to the overall population of men is observed in this region in addition to higher mortality rates compared to records from United States of America (USA). Meanwhile, there are a number of unique factors about the environment and lifestyle of the New Zealand population that need to be considered when analyzing various diseases. There are a high number of tobacco smokers in New Zealand, deficiency of trace elements such as selenium in the New Zealand soil impacting selenium nutrition from locally produced food, and New Zealand recording an overweight population. This motivated us to look into various gene x environment interactions and the risk of aggressive PCa. For this SNP-based study, the PubMed database was screened for research articles based on Genome-Wide Association Studies (GWAS) and/or case-control studies published on or after the year 2000. Attention was given to relevant research papers describing SNP association with PCa among patients with European ethnicity only. We herewith present the data and analysis of gene x environment interaction and risk of aggressive PCa with 136 single nucleotide polymorphisms (SNPs) present in various genes and undefined regions collected from 197 men with clinically diagnosed aggressive PCa, 57 men clinically diagnosed with non-aggressive PCa and 369 healthy controls. We have identified several SNPs that have risk associations, both, with and without environmental interactions. These include certain SNPs present in or near genes associated with obesity and diabetes mellitus such as FADS2 (Fatty acid desaturase 2), LEP (Leptin), PPAR-γ (Peroxisome Proliferator-Activated Receptor gamma), as well as selenoproteins SEP15 (Selenoprotein 15KDa) and SEPS1 (Selenoprotein S) were a significant risk for PCa along with a Cytochrome P450 Family 24 Subfamily A Member 1 (CYP24A1) involved in the degradation of Vitamin D3. We have also observed that the SNPs that may be vulnerable to environmental conditions may be playing a role in the initiation of non-aggressive PCa, as all the SNPs that were identified as statistically significant lost their power when adjusted for the effect of the environmental factors. These analyses were carried out by using PLINK software version 1.07. As the progression of PCa was mapped, we found an increasing role of environmental factors interacting with a panel of SNPs in increasing the risk of aggressive PCa. Among these SNPs, several including those on or near the genes Myeloma Overexpressed (MYEOV), Microseminoprotein B (MSMB), Fatty Acid Desaturase 2 (FADS2), Peroxisome proliferator-activated receptor gamma (PPAR-γ), SEP15 and Kallikrein-3 (KLK3) showed increased risk of aggressive PCa with MYEOV and PPAR-γ identified to have the highest statistical significance. These results motivated us to consider a wider approach, and work further on the data to identify if there were any other possibilities of finding potential SNP biomarkers missed by us. We, therefore, considered the statistical adjustments for ageing and each lifestyle factor individually and SNP-SNP epistasis as risk for aggressive PCa. With only some of the prior identified statistically significant SNP biomarkers detected after a number of analyses, we proceeded with a machine-learning approach employing Artificial Neural Networks (ANN) produced by Waikato Environment for Knowledge Analysis (WEKA) version 3.8.1 to understand our data better. Comparing the statistically significant SNP genotyping risk for aggressive PCa in our cohort between the results obtained by using PLINK analysis and ANN, we successfully narrowed down our list of SNPs to only 3 (rs17793693 in the gene PPAR-γ, rs10896438 in the gene MYEOV and rs10244329 in the gene LEP) from 97 (mentioned in the Appendix). This was a major end-point of this exercise. The role of MYEOV- a putative oncogene, PPAR-γ (gene involved with fat metabolism) and Anaplastic Lymphoma Kinase (ALK) was verified by analysing the expression of these three proteins in three PCa cell lines (and additional non-prostate cancer cell lines). Of interest here was the involvement of the gene ALK, which had previously not been studied by us. The gene ALK is involved with non-small lung cancer, and since tobacco smoking was identified as an important risk factor, we decided to involve the gene ALK as well while plotting the phylogenetic tree and doing protein expression analysis. This work helped us identify a unique link between risk of PCa and the gene ALK. We performed a multi-dimensional analysis and these findings were not only interesting, but also very novel in its own way. Overall, we conclude that multidimensional/holistic approach can be designed to discover the hidden biomarkers in biomedical ailments which are still in want of one. Using this approach successfully, we herewith propose SNP biomarkers in three genes for aggressive PCa– MYEOV, LEP and PPARG for NZ cohort. By extending the findings of multidimensional approach, we were able to report the over-expression of an oncogene- MYEOV and a kinase- ALK that were identified to be in close proximity when analysed using phylogenetic analysis in aggressive PCa.