• Occupational & Environmental Epidemiology Branch (OEEB), Division of Cancer Epidemiology and Genetics (DCEG), National Cancer Institute (NCI), Rockville, Maryland, USA.
• Organic Analytical Toxicology Branch, Division of Laboratory Sciences, National Center for Environmental Health, Centers for Disease Control and Prevention, Atlanta, Georgia, USA.
• Biostatistics Branch, DCEG, NCI.
• Huntsman Cancer Institute, University of Utah School of Medicine, Salt Lake City, Utah, USA.

BACKGROUND: Per- and polyfluoroalkyl substances (PFAS) are persistent, widespread environmental contaminants and some are endocrine-disrupting. Studies of gynecologic cancers are limited; we evaluated ovarian cancer, a rare, often fatal malignancy.

METHODS: This nested case-control study included 318 ovarian cancer cases and 472 individually matched female controls in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial, which recruited participants aged 55-74 years from 10 U.S. study centers (1993-2001). We ascertained cases through 2016 and quantitated eight PFAS in prediagnostic serum samples. We estimated ORs and 95% CIs for continuous (log2-transformed) and categorized PFAS concentrations via conditional logistic regression models implicitly adjusting for matching factors (age, center, randomization year, year of blood draw, race and ethnicity) and adjusted for smoking, body mass index, family history of cancer, menopausal hormone therapy and oral contraceptive use, parity, and number of freeze-thaws.

RESULTS: We found a positive association with ovarian cancer for a doubling in 2-(N-methyl-perfluorooctane sulfonamido) acetic acid (MeFOSAA) concentrations (ORperlog2=1.24, CI = 1.03-1.49) and 62% greater risk among those in the highest quartile (ORQ4vsQ1=1.62, CI = 1.03-2.54; p-trend = 0.02). Perfluorooctane sulfonic acid (PFOS) was associated with increased risk (ORperlog2=1.47, CI = 1.05-2.06) with no quartile trend (p-trend = 0.79). Associations with perfluorononanoic (ORperlog2=1.36, CI = 0.95-1.95) and perfluorodecanoic acid (ORperlog2=1.35, CI = 0.94-1.95) were suggested, with non-monotonic quartile trends (p-trend = 0.12-0.21). MeFOSAA associations were strongest in women aged 55-59 (ORperlog2=1.60, CI = 1.13-2.27), more moderate in those 60-64 (ORperlog2=1.31, CI = 0.90-1.90) and null among women 65 + (ORperlog2=1.02, CI = 0.73-1.43; p-heterogeneity = 0.22). Associations persisted in cases diagnosed ≥8 years after blood collection.

CONCLUSIONS: These findings offer novel evidence for PFAS as ovarian cancer risk factors, particularly PFOS and MeFOSAA, a PFOS precursor.