Validation of a Clinical and Polygenic Risk Prediction Model for Ovarian Cancer in the Nurses' Health Study

Ovarian cancer is the leading cause of death among gynecologic cancers because of its late diagnosis, but risk stratification for targeted prevention can improve outcomes. To identify women at risk of ovarian cancer, pathogenic variant testing is used, especially for those with a family history of the disease. However, high-risk genetic variants are rare, accounting for <20% of ovarian cancer cases overall. More recently, clinical models that help identify ovarian cancer cases incorporate polygenic risk for improved risk stratification. Herein, we cross-validate a polygenic risk-integrated clinical model called the refined risk model in a nested case-control population from the Nurses' Health Study. The refined risk model is compared with a clinical risk score and polygenic risk score model, respectively. The ORs per SD were 1.26 (1.01, 1.48), 1.13 (0.96, 1.33), and 1.22 (1.04, 1.43) with corresponding AUCs of 0.56 (0.51, 0.61), 0.54 (0.49, 0.59), and 0.55 (0.50, 0.60) for the refined risk, clinical risk, and polygenic risk models, respectively. Population screening has been assessed over the years, and although stage-shifting in the screened population has occurred, the mortality benefit was not evident. However, as treatments for ovarian cancer improve, the benefits of a risk-stratified approach to screening are expected to grow.

In this study, we have validated the predictive performance of a combined clinical and polygenic model for ovarian cancer. Accurate risk assessment enables more efficient allocation of preventive interventions, including intensified surveillance and consideration of risk-reducing salpingo-oophorectomy among women at elevated risk.