Constructing a Panel of Common Predisposition Variants to Predict Women That Are at High-Risk of Developing Endometrial Cancer.

Introduction: Endometrial cancer (EC) is the most common gynaecological cancer in the UK and its incidence is rising. Major risk factors for EC include age, obesity, reproductive factors, and family history. Risk prediction models (RPMs) may predict future EC risk, however, genetic predisposition is currently overlooked in published models. Common susceptibility variants have been extremely successful, SNP18 for instance, at improving RPMs for breast cancer. We hypothesized that a risk prediction model comprised of multiple SNPs could identify women at increased risk of EC. Methods: To study the added benefits of utilising a SNP array to identify women at high-risk of EC, we systematically reviewed the literature to construct a 20- SNP guide panel and evaluated the distribution of polygenic risk score (PRS) in 1,526 control women, who were free of EC at point of entry. These were nested within PROCAS1 study and genotyped using OncoArray 500K. The study is in progress to analyse the genotype of a unique unselected cohort of 581 EC cases, obtained from studies across the faculty. Results: Our systematic review suggested many variants in a variety of biological pathways and processes including tumourigenesis and sex steroid pathway. The top 20 most robust variants were selected for our guide panel. Initial analyses are in line with expected values and show ideal distribution in controls when 16 of the directly genotyped SNPs (including 11 from the guide panel and 6 substitutes) were used to derive a PRS. The direction, significance power and risk allele frequency of each SNP were in agreement with previous studies. Conclusion: There is strong evidence for polygenic input from common variants in EC susceptibility. Our initial analysis reveals a PRS using 16 SNPs which mimics the distribution of the well-validated and adjusted SNP18 for breast cancer. In addition to the guide SNP panel, we are looking for any directly or imputed SNP above the significance threshold of 10-7. The resultant PRS will be adjusted by taking into account clinical factors. Formulating a predictive and multiplicative prediction tool that is stratified for each individual for targeted early detection and prevention strategies is of great importance as the direction of standard care moves towards personalised medicine.