Abstract A107: Ovarian cancer risk factors by tumor dominance, a surrogate for cell of origin

Abstract

Abstract Although some epidemiologic associations for ovarian cancer are established, such as the risk reduction due to parity or oral contraceptive use, the influence of other characteristics like endometriosis and non-steroidal anti-inflammatory use is less clear. Distinguishing categories of ovarian cancer based on shared pathways of development may clarify these associations and further our understanding of the disease. Traditionally, ovarian cancers have been thought to arise from the ovarian surface epithelium. Recent evidence in BRCA1 or 2 mutation carriers showing synchronous tumors in fallopian tubes and potential precursor lesions with p53 signatures that match tumors suggest that some “ovarian” cancers may originate in the fallopian tube. Tumor origin (e.g., ovarian vs. fallopian) may be classified by rigorous pathologic review protocol, including the identification of tubal lesions and p53 signatures, but this is not practical in population-based studies where the needed tissue is not always available. Alternatively, presence or absence of a two-fold difference in tumor size between ovaries (dominant ovarian mass) can be used to assign a greater or lesser probability of ovarian (versus tubal) origin. We applied this classification algorithm to ovarian cancer cases in a population based case-control study (NEC) and two prospective cohort studies (NHS/NHSII). Associations between known and suspected ovarian cancer risk factors were evaluated in relation to cases with dominant and non-dominant tumors using polytomous logistic regression (for NEC) and competing risks models (for NHS) adjusted for matching factors, study, oral contraceptive use (never, <5 years, > 5 years), parity (0, 1, 2, 3, 4+), tubal ligation, and family history of breast or ovarian cancer. Among the 1801 invasive epithelial ovarian cancer cases, we observed 1127 tumors (778 NEC, 246 NHS) with a dominant mass, indicating a greater likelihood of ovarian origin, and 674 (534 NEC, 140 NHS) with no dominant mass, indicating a greater likelihood of fallopian tube origin. The dominant cases were more likely to be mucinous, endometrioid, clear cell, or undifferentiated while the non-dominant cases were more likely to be serous ovarian cancers, except for serous borderline tumors which were more likely to be dominant. We observed stronger associations with dominant tumors than non-dominant tumors for tubal ligation, more than two births, and endometriosis (pheterogeneity=0.01, 0.01, and 0.0002, respectively). Interestingly, differences in non-steroidal anti-inflammatory use (pheterogeneity=0.08) suggested inverse associations for non-dominant tumors. These results suggest that risk factors for tumors putatively arising from ovarian versus fallopian tube sites may have different risk factors for disease. Acknowledgments: Supported by the DOD Ovarian Cancer Academy and National Cancer Institute grants R01CA54419, P50CA105009, P01CA87969, R01CA50385. Citation Information: Cancer Prev Res 2011;4(10 Suppl):A107.