Abstract
Introduction: A recently proposed model of ovarian carcinogenesis categorizes ovarian tumors into three types. Type I tumors include low-grade serous and mucinous carcinomas, are slower growing, and are characterized by activating mutations in KRAS or BRAF; Type II tumors are characterized by a hormonally-responsive precursor (e.g., endometriosis) and often have PTEN mutations. Type III tumors include high-grade serous, endometrioid and undifferentiated carcinomas, which present at an advanced stage, and are characterized by P53 mutations. Tumor type is an important factor in determining whether the tumor will ultimately be fatal to the patient, with increasing aggressiveness across the three tumor pathway types. In the Nurses' Health Study (NHS), we used time to death as a surrogate measure of type I/II vs. type III tumors and investigated differences in risk factor associations between the two groups.
Methods: If a woman died due to ovarian cancer within 3 years of diagnosis, her tumor was considered rapidly fatal. If a woman died due to ovarian cancer more than 3 years post-diagnosis, died due to any other cause, or did not die during follow-up, her tumor was considered less aggressive. We calculated hazards ratios (HR) and 95% confidence intervals (CI) for associations with rapidly fatal and less aggressive tumors using competing risks analysis in Cox proportional hazards models. We considered known or suspected risk factors for ovarian cancer, including age, parity, breastfeeding, oral contraceptive (OC) use, postmenopausal hormone (PMH) use, age at menopause, age at menarche, tubal ligation, height, and body mass index (BMI). All models were adjusted for these factors as well as age at first birth, menopausal status, smoking, and family history of breast and ovarian cancer.
Results: Over 30 years of follow-up, we observed 349 rapidly fatal cases and 404 less aggressive cases. Increased age was associated with greater risk of rapidly fatal cancer compared to less aggressive disease (HR: 1.08; 95% CI: 1.06–1.10 vs. HR: 1.02; 95% CI: 1.01–1.04, respectively; p-heterogeneity<0.001), whereas use of combined estrogen and progestin therapy was associated with increased risk of less aggressive cancers (HR: 0.91; 95% CI: 0.66–1.26), but not rapidly fatal disease (HR: 1.52; 95% CI:1.15–2.02; p-heterogeneity=0.02). OC use was associated with decreased risk of rapidly fatal cancers (HR for 1 year increase in OC use: 0.92; 95% CI: 0.88–0.96), but not less aggressive cancers (HR: 1.00; 95% CI: 0.97–1.03; p-heterogeneity=0.003). Ever having been parous was associated with decreased risk of rapidly fatal cancers (HR: 0.58; 95% CI: 0.39–0.88), but there was no increased protection for additional children among parous women (HR: 1.01; 95% CI: 0.94–1.09). Conversely, ever having children was not associated with risk of less aggressive tumors (HR: 0.92; 95% CI: 0.62–1.36), but there was decreased risk with each additional child among parous women (HR: 0.90; 95% CI: 0.83–0.97; p-heterogeneity=0.02). Furthermore, duration of breastfeeding was only associated with decreased risk of rapidly fatal disease (p-heterogeneity=0.03). There were no differences in association for height, tubal ligation, age at menarche, age at menopause, or BMI.
Conclusion: In the NHS, there are clear differences in risk factors for rapidly fatal vs. less aggressive ovarian tumors, surrogates for type I/II vs. type III tumors. This has important implications for targeting preventive measures. Validation of the observed associations is underway in two case-control studies.
Citation Information: Cancer Prev Res 2011;4(10 Suppl):A92.
RSS Feeds