Cancer
Predicting risk of fatal ovarian cancer using a novel two-stage model Mary Townsend* Mary Townsend Shelley Tworoger Bernard Rosner
Background: Identifying risk factors for the most aggressive forms of ovarian cancer is important to guide prevention. We used a multi-state model to capture the additive effects of ovarian cancer risk factors on disease risk and death among those with ovarian cancer.
Methods: We used prospectively collected data from the Nurses’ Health Studies (three 12-year risk periods, 1980-2017). Putative risk factors assessed at baseline of each risk period included age at menarche, pre- and post-menopause duration, reproductive/hormonal factors, smoking, early life body mass index (BMI), family history of breast and ovarian cancer, and race. We used Fine-Gray models to estimate HRs and 95% CIs for risk of incident ovarian cancer among at-risk individuals and risk of mortality among those with ovarian cancer adjusted for competing risks. Estimates were combined using weights for incidence and mortality based on baseline hazard (incidence) and survival (mortality) functions over follow-up to estimate the log relative risk of fatal ovarian cancer over 12 years in cancer free individuals.
Results: We had 197,690 eligible participants, with 1678 confirmed cases of ovarian cancer and 693 deaths due to ovarian cancer. A SD increment in age at menarche (1.4 years), premenopausal (5.9 years) and postmenopausal (7.0 years) duration, duration of estrogen only hormone therapy (4.1 years), and BMI at age 10 (3.0kg/m2) as well as nulliparous vs. parous and 20+ pack-years vs. never smoking were associated with increased risk of fatal ovarian cancer (HRs 1.11-1.92, p<0.05). Duration of oral contraceptive use was associated with lower risk of fatal ovarian cancer (10+yr vs. never use HR=0.64, 95%CI=0.43-0.96).
Conclusion: Our results support that pre-diagnostic factors can predict risk of developing and dying of ovarian cancer. Model refinement and validation could identify high-risk populations who would benefit from intensive screening and possibly prophylactic bilateral salpingo-oophorectomy.