Using families to understand the impact of genetic variation on prostate cancer

Prostate cancer (PCa) is the most common, non-cutaneous malignancy in men in the developed world. It is highly heritable, with twin studies suggesting that as much as 58% of disease risk can be explained by genetics. While more than 170 common genetic risk variants have been identified, these variants still only explain a minor portion of heritability, are largely of low to moderate effect size, and for many their function remains unclear. There has recently been significant success in the discovery of rare genetic variants contributing to complex disease through next-generation sequencing studies of large families. Mancuso and colleagues (2016) have estimated that as much as 42% of PCa risk is due to rare variants, but to date only 6% of this risk has been elucidated. With two-thirds of PCa heritability still unexplained, including the contribution of rare variants, we hypothesise that the utilisation of PCa families will aid in the identification of these rare variants. Germline risk variants and somatic tumour alterations have traditionally been regarded as unrelated events in cancer. However, there is now increasing evidence to suggest that specific germline variants may predispose some somatic tumour events, including copy number changes and gene fusions. Of particular interest in PCa, is the fact that germline variants have been reported to be significantly associated with the TMPRSS2:ERG fusion. Given the high frequency of these fusion events and accumulating evidence from previous studies, we also hypothesise that there are inherited determinants of somatic tumour variation, and this will be the second focus of this thesis. Family studies are proving highly valuable in the study of complex disease and here I will explore these hypotheses using the Tasmanian Familial Prostate Cancer Study cohorts, comprising genetic material from large families with multiple PCa cases and their relatives (Tasmanian Familial Prostate Cancer Cohort), as well as the Tasmanian Prostate Cancer Case-Control Study. To address the first hypothesis, whole-genome sequencing (WGS) was undertaken in five large Tasmanian PCa pedigrees to identify rare genetic variants contributing to disease risk. Variants were prioritised on a per-family basis by minor allele frequency, segregation with disease, mutation type and predicted functional consequence. Of the 20 prioritised rare variants, four were determined to be significantly associated with PCa risk in the Tasmanian population. This included rare variants in the genes RND1, WNT1, EZH2 and the known G84E HOXB13 variant. Both RND1 and WNT1 have been found to promote the growth and migration of cancer cells and, notably, in our study the variants appeared to be co-inherited. The EZH2 variant is a rare, intronic variant (rs78589034) present within a 3' splice consensus sequence. EZH2 encodes the histone methyltransferase enzyme and is constitutively overexpressed in a range of cancers, including PCa. EZH2 is a highly variable gene and multiple transcripts have been identified. In fact, Chen et al (2017) observed that alternative splicing involving the inclusion of exon 14 plays a major role in the tumourigenesis of renal cancer. While this variant was significantly associated with PCa risk in the Tasmanian population (OR=3.27, p=0.001), functional assays were unable to determine the potential impact of this variant on the splicing mechanisms of EZH2. The G84E HOXB13 variant (rs138213197) was initially observed in the WGS data and followup genotyping found a significant association with PCa risk in the larger Tasmanian Familial Prostate Cancer Study cohorts (OR=6.59, p=4.22x10\\(^{-5}\\)). Although multiple studies have demonstrated an association of the G84E variant with PCa risk, no study has assessed the functional impact of the variant on HOXB13 gene and protein expression. Here, no difference in HOXB13 gene or protein expression was observed between prostate tumours from G84E carriers and non-carriers, but interestingly, the variant allele was rarely transcribed in carriers. The unbalanced allele transcription did not appear to be caused by methylation differences and, thus, other mechanisms, such as DNA copy number variation at the HOXB13 site or rapid targeted degradation of the variant mRNA transcript, may underpin the observed allelic imbalance. Hence, questions remain regarding how this variant influences tumour development. Given the rarity of the G84E variant, achieving a sufficient sample size for analyses is challenging, therefore, through collaboration with members of the Prostate Cancer Association Group to Investigate Cancer Associated Alterations in the Genome (PRACTICAL) consortium, we aim to further explore the function of this variant. To address the second hypothesis, germline and tumour samples from PCa cases were utilised to explore inherited determinants of somatic tumour variation. Tumours from 14 PcTas9 cases were analysed using the TruSight RNA Fusion Panel (Illumina), identifying seven tumours as TMPRSS2:ERG fusion positive. Subsequently, analysis of the entire Tasmanian Prostate Tissue Pathology Resource showed that 31.5% of tumours were fusion positive. This event was more frequent in tumours from two families, PcTas2 and PcTas9 and, interestingly, was not identified in any of the eight sporadic tumours examined. These results suggest that there may be an underlying inherited genetic variant(s) predisposing to this fusion event. Subsequent work is focusing on screening for germline risk variants previously found to be associated with fusion positive tumours, including rare variants in POLI and ESCO1. Somatic copy number changes, including amplifications and deletions, are also common events in tumours, leading to the suggestion that they may also arise due to germline genetic variation. To explore this hypothesis, array comparative genomic hybridisation was applied to 12 PcTas9 prostate tumours to determine shared altered chromosomal regions. The most consistent alteration involved amplification of the EEF2 gene, which is a novel finding. EEF2 is highly expressed in human carcinoma tissue and has been suggested as a potential PCa biomarker. Immunohistochemistry of the Tasmanian Prostate Tissue Pathology Resource found that the EEF2 protein was overexpressed in 49% of malignant compared to matched benign tissue, but no difference was observed between tumours from PcTas9 cases and non-PcTas9 cases. However, gene expression assays found malignant cells from PcTas9 tumours had significantly higher EEF2 5'UTR/exon 2 expression compared to malignant cells isolated from non-PcTas9 tumours. Thus, these results suggest that the EEF2 amplification may be specific to PcTas9 and due to an inherited predisposition variant(s). To test this hypothesis, recent WGS data generated for this family will be utilised in linkage analysis based on EEF2 amplification status. Establishing rare variants as disease-causing requires analysis of large cohorts and secondly, comprehensive functional analyses. This study has identified four rare germline variants significantly associated with PCa risk in the Tasmanian population. Variant screening in larger cohorts of PCa cases and controls is required to determine their contribution to other populations. Moreover, the functional impact of the EZH2 and HOXB13 variants on gene and protein expression remains unclear and requires more comprehensive functional analyses. This study also identified recurrent somatic variations in the tumour genomes of Tasmanian PCa cases. The TMPRSS2:ERG fusion and amplification of the EEF2 gene is more apparent in tumours from the PcTas9 family, suggesting that these somatic tumour events could be underpinned by inherited predisposition. There is currently a strong push to implement polygenic risk scores based on common variants in the clinical setting, yet with only one-third of genetic predisposition explained, clinical implementation may be premature. Studies such as the one described here, aim to directly explore genetic contribution to PCa. Rare germline variants and somatic tumour variation are of great interest as potential screening biomarkers and therapeutic targets, and if we are to understand the genetic determinants of PCa development, a strong focus on fully characterising these factors is essential.