SNP-SNP interactions as risk factors for aggressive prostate cancer

Prostate cancer (PCa) is one of the most significant male health concerns worldwide. Single nucleotide polymorphisms (SNPs) are becoming increasingly strong candidate biomarkers for identifying susceptibility to PCa. We identified a number of SNPs reported in genome-wide association analyses (GWAS) as risk factors for aggressive PCa in various European populations, and then defined SNP-SNP interactions, using PLINK software, with nucleic acid samples from a New Zealand cohort. We used this approach to find a gene x environment marker for aggressive PCa, as although statistically gene x environment interactions can be adjusted for, it is highly impossible in practicality, and thus must be incorporated in the search for a reliable biomarker for PCa. We found two intronic SNPs statistically significantly interacting with each other as a risk for aggressive prostate cancer on being compared to healthy controls in a New Zealand population.


Introduction
Prostate cancer (PCa) is highly prevalent, and around 1 in 6 patients are at risk of developing the aggressive form of the disease 1 . It has become one of the most significant male health concerns worldwide 2 . An individual is diagnosed as having high-risk or aggressive PCa based on the classification by the American Urological Association 3 , when the clinical T stage ≥cT2c, and/or the Gleason score ≥8, and/or the serum prostate serum antigen (PSA) level >20ng/ml 4 .
Although a hereditary aspect is well known for this disease 5 , various studies have also shown that genetic interactions with biological and behavioral factors play an important role in the overall risk and prognosis of PCa 6-8 . Variations in the genome are a major contributor to the differences in disease susceptibly amongst individuals 9 . Single nucleotide polymorphisms (SNPs) are the most commonly identified variations in a genome.
Analysing the role of SNP-SNP interactions and epistasis 10 is very appealing among researchers working on risk factors for various cancers 11-13 , including prostate cancer 14 . Here we have identified a SNP-SNP interaction as a risk factor for aggressive PCa, by comparing the data generated after carrying out SNP genotyping using the SEQUENOM MassARRAY iPLEX ® assay, and the TaqMan ® assay (depending on the gene of interest) from the DNA extracted from blood samples. These samples were taken from a New Zealand cohort of men with self-reported European ethnicity that have been clinically diagnosed with aggressive and non-aggressive PCa, and healthy controls with no reported symptoms of the disease. Symptoms include increased urination during night time along with a frequent urge to urinate problems maintaining a steady flow of urine, hematuria and dysuria 15 . Our results indicate a strong influence of gene x environment interaction in overall gene expression and epistasis.

Study population
Patients with a clinically established diagnosis of PCa (aggressive and non-aggressive) from the Auckland Regional Urology Registry (Auckland, Middlemore, and North Shore hospitals), and certain private practices in the Waikato region of New  Table 1 shows the statistically significant SNP-SNP interaction discovered in patients with aggressive PCa when compared to healthy controls. The results obtained for other categorical analyses are not discussed here, as they were not statistically significant in our study and have been mentioned in Supplementary Table 2. The SNP rs2121875, an intronic SNP present in chromosomal position 5p12 near the fibroblast growth factor 10 (FGF10) gene 24 , has been identified to be associated with the SNP rs4809960, an intronic SNP present in chromosomal position 20q13 near the gene cytochrome P450 family 24 subfamily A member 1 (CYP24A1) 25 , such that the latter SNP raises the odds of having the prior.

Discussion
Epistatic effects that are crucial to define various biologicallyintuitive models of interaction between two SNPs have already been observed in a variety of species 11 . We believe this is the first study on SNP-SNP interactions associated with aggressive PCa carried out with patients from a New Zealand population.
The SNP rs4809960 in the gene CYP24A1 has been reported by Holt et al., (2010) to be associated with prostate cancer-specific mortality, and was not evolutionarily conserved 25 . It was also found to have an effect on the body mass index (BMI), but due to a small sample size the hazard ratios for the BMI strata were not considered reliable enough to be reported 25 . The protein encoded by CYP24A1 initiates the degradation of the physiologically active form of Vitamin D3 (VD3) 26 . VD3 is an important hormone that is actively involved in regulating cell proliferation in the prostate, and has also been identified to have increased expression in PCa cell lines 27 . It is well established that, with ageing, the skin cannot synthesize VD3 as effectively as desirable and the kidney's ability to convert VD3 to its active form decreases 28 . This is of relevance because PCa has always been considered as a disease of elderly men 29 who have had less exposure to sunlight and thereby Vitamin D3 30 . It is even more intriguing for the other epistatic SNP to be identified in FGF10.
According to Paul et al. (2013), during mesenchymal development, eFGF10 protein can trigger PCa development through increased androgen receptor expression in the neoplastic epithelium 31 . It is also worthy to mention that FGF10 is closest to FGF7 based on its evolutionary history 32 , and according to Emoto et al. (1997), is suggested to have no activity for fibroblasts 32 . We do not agree with this, because fibroblasts in certain organs, senesce due to aging 33 , and can promote tumour invasion 34 . This logical progression of ageing-led senescence and promotion of tumour invasion holds true for ageing and risk of aggressive PCa 16 as well.
We suggest that the intronic SNP rs2121875 in the gene FGF10 may be causing alterations in gene expression, perhaps due to the prevalent external/environmental conditions in the elderly men with PCa. Our theory is based on the recent discovery in a study by Zhang et al. (2007) that even intronic SNPs (such as the ones identified in FGF10 and CYP24A1) can change the outcome and usage of exons 35,36 . This unique and novel epistatic finding emphasizes the fact that intronic SNPs (and SNP-SNP interactions) can also have a significant effect on the risk of diseases such as aggressive PCa, and need to be investigated further. Author contributions VV and VN planned and carried out the experiments. VV wrote the manuscript. VV and VN did the data cleaning and statistical analysis, respectively. VV interpreted the data. VV, NK, AJ, RP, GM and LRF conceived the idea of the discussion chapter and proofread the manuscript.

Competing interests
No competing interests were disclosed.

Grant information
The author(s) declared that no grants were involved in supporting this work. The authors have carried out SNP genotyping in from a New Zealand cohort of men with self-reported European ethnicity that have been clinically diagnosed with aggressive and non-aggressive PCa, and healthy controls. They have identified a number of SNPs from the GWAS from various European populations and described SNP-SNP interaction as a risk factor for aggressive PCa in a New Zealand cohort. The intronic SNP rs2121875, an on chromosomal position 5p12 near the fibroblast growth factor 10 ( ) gene24, has been identified to be associated with the intronic SNP rs4809960, an intronic FGF10 SNP present in chromosomal position 20q13 near the gene cytochrome P450 family 24 subfamily A member 1 ( ). The protein encoded by initiates the degradation of the physiologically CYP24A1 CYP24A1 active form of Vitamin D3 (VD3) 26 which is an important hormone that is actively involved in regulating cell proliferation in the prostate, and has also been identified to have increased expression in PCa cell lines. The epistatic effect of the SNP-SNP interactions suggested by the authors may be relevant in view of many recent studies showing intronic mutations which can exert their effect on protein coding exons. The recent observation of decreasing Vitamin D3 levels worldwide further support the role of environmental factors in these gene environment interactions. Future studies may help in understanding the role of SNPs and environmental interactions.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Partly

Are the conclusions drawn adequately supported by the results? Yes
No competing interests were disclosed.
I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.