Folate Intake and Ovarian Cancer Risk among Women with Endometriosis: A Case–Control Study from the Ovarian Cancer Association Consortium

This article is featured in Highlights of This Issue, p. 995

Endometriosis is an estrogen-dependent, chronic inflammatory gynecologic condition characterized by endometrial-like tissue that grows outside the uterus and often presents as pelvic pain or infertility. It affects an estimated 10% of reproductive-age women (1); however, this is likely to be an underestimate of the true prevalence due to the difficulty in diagnosing the condition (2). Women with endometriosis have a 2- to 3-fold (3, 4) higher risk of developing endometrioid and clear cell ovarian cancers, which account for approximately 20% of all epithelial ovarian cancers (5).

Folate, a water-soluble B vitamin, plays an important role in DNA, RNA, and protein synthesis and is necessary for cell division (6). Folic acid is the synthetic form of folate used in supplements and food fortification and is better absorbed than folate from food sources due to differing bioavailability (6). The effectiveness of folate supplementation in the prevention of neural tube defects in early pregnancy led to the introduction of mandatory folate fortification of specified foods, typically flour and bread products, beginning in the United States in 1998, followed by other countries including Canada and Australia (7).

Previous epidemiologic and laboratory studies have suggested a possible dual role of folate in carcinogenesis: higher intakes may be protective for healthy epithelial cells, but may promote progression of precursor neoplastic lesions such as preneoplastic colorectal epithelial cells (8, 9). High folate intake has not been associated with an increased risk of ovarian cancer overall (10–13), although there is some evidence of effect modification by other factors including alcohol intake (14, 15). It is, however, possible that high folate intake may promote ovarian cancer in the presence of endometriosis—a known precursor lesion (16).

The aim of this study was to determine whether folate intake is associated with an increased risk of ovarian cancer, particularly the endometrioid and clear cell types, among women with and without endometriosis.

We pooled primary data from six case–control studies participating in the Ovarian Cancer Association Consortium (OCAC) that could provide data on folate intake and endometriosis status. This included five studies from the United States [Diseases of the Ovary and their Evaluation Study (DOV), ref. 17; Hawaii Ovarian Cancer Study (HAW), ref. 18; New England Case–Control Study of Ovarian Cancer (NEC), ref. 19; New Jersey Ovarian Cancer Study (NJO), ref. 20; Los Angeles County Case–Control Studies of Ovarian Cancer (LAC), ref. 21] and one study from Australia [Australian Ovarian Cancer Study (AUS), ref. 22].

Eligible cases included women ages 18 years or older who were diagnosed with invasive epithelial ovarian cancer (including fallopian tube and primary peritoneal cancers). Women with no prior personal history of ovarian cancer and who had at least one ovary at recruitment were included as controls. All studies obtained institutional ethics committee approval and followed recognized ethnical guidelines, including the Declaration of Helsinki, the Belmont Report, and/or the US Common Rule, and all study participants provided written informed consent.

For this analysis, a total of 16,755 women (7,144 cases and 9,611 controls) from the six studies were eligible for inclusion. Women were excluded if they were missing nutrient data (n = 2,713) or had implausible energy intakes (n = 131), defined as more than three standard deviations from the mean natural logarithm of total energy among the control group for that study (23). An additional 53 women were excluded as they were missing data on endometriosis status, leaving 13,858 women (5,741 cases and 8,117 controls) included in the study population. For analyses assessing folate supplementation, women missing information (or from studies that did not collect information) on supplement use were excluded, leaving a total of 9,072 women (3,759 cases and 5,313 controls). Supplementary Fig. S1 details the exclusions applied to obtain the final study population.

Dietary information was acquired through the Multidisciplinary Ovarian Cancer Outcomes Group—a group created out of OCAC to explore, among other aims, dietary associations with survival after ovarian cancer diagnosis. Folate intake was estimated using validated food frequency questionnaires (FFQs) for AUS (24), DOV (25), HAW (26), LAC (26), NEC (27), and NJO (20). Participants were asked to report their usual frequency of consumption of a range of food items (range, 120–200 on the various FFQs) in the year or two prior to diagnosis for cases or prior to interview for controls. This information was used to estimate nutrient intakes using Australian (AUS) or US food tables. Measurement of folate intake using FFQs has been shown to be reliable with a correlation of 0.63 compared with plasma folate levels reported for the Willett FFQ (variations of which were used by AUS and NEC; ref. 28). Four studies (AUS, HAW, NEC, and NJO) and one phase (of three) of the LAC study additionally collected information on supplement use.

Folate intake was defined using three main measures: (i) dietary folate intake including folate that occurs naturally in foods as well as folic acid from fortified foods, (ii) folate intake from supplements, and (iii) total folate intake (from both diet and supplements). For dietary folate intake, we additionally differentiated between naturally occurring folate from foods such as fruits and vegetables, and synthetic folate from foods fortified with folate, including flour. As the bioavailability of natural folate is lower than that of folic acid, we calculated dietary folate equivalents (DFE) for measures including a component of folic acid intake, whereby 1 μg of folic acid was assumed to contribute 1.7 DFE (29).

All measures of dietary folate data were energy-adjusted using the residual method (30). Dietary and total folate intake were categorized using study-specific tertile cutoff points while folic acid intake from supplements was categorized based on the recommended daily intake (RDI) of 400 μg DFE for the general population as 0, <400 and 400+ μg (Supplementary Table S1). Alcohol intake (none, <10, 10+ grams/day) was also assessed using the FFQs.

Endometriosis status was self-reported via questionnaire. Four studies (DOV, HAW, NJO, and LAC) asked if a woman was ever told by a doctor/health professional that they had endometriosis and two studies (AUS and NEC) asked if a woman had ever had endometriosis prior to the reference/diagnosis date.

Dietary data were merged with information potentially relevant to ovarian cancer risk or folate intake from the OCAC core database. These variables, which had been harmonized centrally, included case–control status, age at diagnosis (or comparable reference date for controls), education (high school or less, some college, college graduate, graduate or professional degree), smoking status prediagnosis (never, former current), body mass index (BMI) (<25, 25–29, and ≥30 kg/m²) measured one year (AUS, NEC, NJO, LAC) or five years prior to diagnosis or interview date (DOV and HAW), first-degree family history of breast or ovarian cancer, oral contraceptive pill (OCP) use, parity, breastfeeding history, tubal ligation, endometriosis status, aspirin and nonsteroidal anti-inflammatory drug (NSAID) use. Clinical information included histotype (high-grade serous, low-grade serous, mucinous, endometrioid, clear cell, and other).

Logistic regression models were used to estimate odds ratios (OR) and 95% confidence intervals (CI) for the association between the three main measures of folate intake and ovarian cancer risk, separately for women with and without endometriosis. We conducted analyses for all invasive cancers combined and then separately for endometrioid and clear cell (END/CCC) cancers, as they are most strongly linked to endometriosis, and high-grade serous cancers (HGSC). We also compared results of logistic regression models to equivalent generalized linear mixed models to allow random effects between OCAC sites.

Directed acyclic graphs (DAG) were generated a priori to identify potential confounders of the relationship between folate intake and ovarian cancer; these were retained in models if they altered the beta coefficients for folate intake by >10%. Based on the DAGs, all models were adjusted for age and total energy intake (log) and stratified by study site. Parity was included in the following models for women with endometriosis as its inclusion altered the folate estimates by >10%: dietary folate intake, total folate intake, and END/CCC subtype supplement analyses. Other potential confounders including education, BMI, smoking, OCP use, alcohol consumption, breastfeeding, race, family history, and fortification exposure (whether women completed the FFQ before or after the introduction of mandatory folate fortification) were not included in the final models as they did not alter the folate estimates appreciably.

To assess heterogeneity between studies, study-specific ORs comparing medium/high intake to low dietary folate intake were combined using random effects meta-analysis, and I² and P values for heterogeneity (from chi-square tests) were calculated. To assess whether the folate–cancer association differed between women with and without endometriosis, we reran models including an interaction term between the folate variable and endometriosis. A P < 0.05 for the interaction term was considered statistically significant.

To assess whether any association between folate intake (medium/high vs. low) and ovarian cancer risk was modified by other factors, we stratified by potential modifiers. These included alcohol use (none, <10 g/day, 10+ g/day), BMI (<25, 25–29, ≥30 kg/m²), folate fortification status, NSAID and aspirin use [regular use (at least once per week vs. less often)]. These variables were chosen because they can interfere with the bioavailability of folate (as reported for alcohol; ref. 31) or affect inflammation (aspirin and NSAIDs, ref. 32, and as suggested for BMI, ref. 33). Factors related to inflammation were investigated as folate may play a role in inflammatory processes (34), and endometriosis is an inflammatory condition, so it is possible any associations may be modified by pro- or anti-inflammatory factors.

In post hoc analyses, we also examined the association between glycemic index (GI), glycemic load (GL), and intake of grains (total, whole and refined)—which are likely to include a high proportion of folate-fortified foods—and ovarian cancer risk. This was to assess whether associations seen for dietary folate, particularly for the synthetic component, were potentially due to the types of foods that are fortified rather than folate itself. Models were run for women with and without endometriosis and GI, GL, and grain intake were categorized using study-specific tertile cutoff points. We used chi-squared tests to assess associations between dietary folate intake and grain intake, GL and GI.

Analyses were performed using SAS version 9.4 (SAS Institute) and Stata version 15 (StataCorp LP).

Given possible issues with recall bias and dietary assessment in case–control studies, we also used Mendelian randomization (MR) to evaluate this association. Although we knew this would be underpowered for women with endometriosis, our primary goal was to determine if results were consistent with those from the observational analyses.

We used two-sample MR to assess the associations between folate and ovarian cancer risk using genetic markers as a proxy for serum folate levels. We used publicly available summary data for 4 single-nucleotide polymorphisms (SNP) associated with serum folate levels (predicting 1.3% variance) in the largest published genome-wide association study (GWAS) to date (ref. 35; Supplementary Table S2). Summary estimates for the association between the SNPs and ovarian cancer were not available by endometriosis status, so we estimated these using individual level data from 1,740 women with endometriosis and 19,145 women without endometriosis from 18 OCAC studies including the six studies in the dietary analysis. DNA samples had been genotyped as previously described (36). We estimated the association between each SNP and ovarian cancer risk by fitting logistic regression models adjusted for the participants’ study country of origin and ancestral principal components (between 1 and 9 depending on the genotyping platform) to account for population structure (36). All women were of genetically determined European ancestry.

We used the beta coefficients and standard errors for the SNP–folate and SNP–ovarian cancer associations to estimate ORs and 95% CIs for the effect of folate on ovarian cancer. Estimates were obtained for each SNP by dividing the SNP–outcome association by its SNP–folate association (Wald ratio).The individual SNP estimates were then combined using an inverse-variance weighted MR model (37).

We undertook sensitivity analyses to assess potential violations of the MR assumptions, including MR-Egger (38), weighted-median MR (39), and MR-PRESSO (40). We calculated Cochran's Q-statistic for between-SNP heterogeneity of effects. We checked whether SNPs were associated with other relevant traits using the NHGRI-EBI GWAS Catalog (41) and PhenoScanner (42, 43). Analyses were performed using the MendelianRandomization package (44) and MR-PRESSO package (40) implemented in the R software (R Foundation for Statistical Computing).

Data described in the article cannot be made publicly available due to privacy and ethical limitations imposed by the original studies in which these data were collected, but can be shared upon approval of a data request form by the OCAC Data Access Coordinating Committee and with appropriate human subjects approval and data transfer agreements.

Table 1 shows the characteristics of cases and controls with and without endometriosis in the observational analysis. The majority of cases and controls were white and, as expected, compared with controls, cases were more likely have a shorter duration of OCP use, more likely to be nulliparous, less likely to have breastfed and less likely to have had tubal ligation. Women who reported a previous diagnosis of endometriosis were younger, more likely to be nulliparous and had a longer duration of OCP use than those without endometriosis. Cases with endometriosis were more likely to have endometrioid or clear cell cancers (38.1%) than those without endometriosis (20.6%).

Estimated dietary folate intake and supplement use varied across study sites (Table 2). Folate intake was higher after the introduction of fortification programs and supplement use was more common in the USA than in Australia.

Table 3 shows the associations between folate intake and risk of invasive epithelial ovarian cancer for women with and without endometriosis. Among women with endometriosis, there was a suggestion that higher dietary folate intake was associated with an increased risk of invasive ovarian cancer [tertile 2 (T2): OR 1.29 (95% CI, 0.95–1.75); T3: 1.37 (1.01–1.86) vs. T1; P_trend 0.045]. No increased risk was seen for women without endometriosis. Among women with endometriosis, the association was stronger for synthetic folate from dietary sources [OR for T2: 1.73 (1.17–2.56); T3: 1.36 (0.92–1.99)]; there was also a suggestion that synthetic folate was associated with an increased risk of ovarian cancer for women without endometriosis [OR for T2: 1.17 (1.04–1.30); T3: 1.10 (0.98–1.23)]. Conversely, an inverse association was seen for naturally occurring folate (P_trend < 0.001) for women without endometriosis. There were no significant associations for folic acid from supplements or total folate intake (diet and supplements) for women in either group. The patterns did not differ when we considered endometrioid/clear cell cancer and HGSC separately (Supplementary Table S3). There was little difference in results when we used generalized linear mixed models.

The associations between dietary folate intake and ovarian cancer risk were consistent across the study sites, with a 41% increased risk of ovarian cancer (95% CI, 1.07–1.85, I² = 11%, P = 0.3) associated with medium/high dietary folate intake among women with endometriosis, but no association (OR, 0.97; 0.90–1.05) among those without endometriosis (P_interaction = 0.0001; Fig. 1).

Figure 2 shows the associations between folate intake (dietary and supplemental) and risk of ovarian cancer stratified by alcohol intake, BMI, fortification status, NSAID, and aspirin use, for women with endometriosis. Among women who took NSAIDs at least once per week, supplemental folate intake was associated with an increased risk of ovarian cancer (OR 2.29; 95% CI, 1.08–4.84), whereas among nonusers there was a decreased risk (OR, 0.65; 95% CI, 0.42–0.98). This difference was not seen for dietary folate intake or total folate intake. A similar increased risk was seen for dietary folate among regular aspirin users (OR, 2.16; 95% CI, 0.98–4.77). There was no strong evidence of effect modification for the other variables. For women without endometriosis, the only evidence of effect modification was for folate intake from supplements and alcohol intake (Supplementary Fig. S2).

Post hoc analyses (Supplementary Table S4) to assess whether observed associations were potentially due to the types of foods fortified, rather than folate itself, showed no association between grain intake (total, refined, or whole) and ovarian cancer in women with or without endometriosis. Among those without endometriosis, higher GI and GL but not grain intake were associated with an increased risk of ovarian cancer (GI, high intake: OR 1.31 (1.13–1.53); GL, high intake: OR 1.38 (1.20–1.58)], which is consistent with published literature (45). This was not observed among women with endometriosis. Dietary folate intake was significantly associated with grain intake (total, refined, and whole), GL and GI (all P < 0.005).

Although the number of women with endometriosis was relatively small, so the SNP–cancer estimates for this group were imprecise, we saw the same pattern with a suggested increased risk of ovarian cancer for higher genetically predicted folate levels in women with endometriosis [OR, 2.22 (0.80–6.17) per 1 standard deviation higher folate] but no association among those without endometriosis [OR, 0.90 (0.60–1.34); Supplementary Fig. S3]. Results from the sensitivity analyses were broadly consistent (Supplementary Table S5) and no outlying SNPs were identified using MR-PRESSO.

Our results, from both self-reported intake and genetically predicted measures of serum folate, support our a priori hypothesis that higher folate intake increases the risk of ovarian cancer among women with endometriosis but not those without. There was also a suggestion that higher intake of synthetic folate, folic acid added to foods during fortification, but not naturally occurring folate in foods was associated with increased risk. However, this may have been a chance finding as we did not see any association with folic acid intake from supplements.

Previous reports (including one each from AUS, ref. 10 and NEC, ref. 11) and a meta-analysis with 12 studies (including AUS and NEC) have shown no association between high folate intake and risk of ovarian cancer (12, 13), but due to the relatively low prevalence of endometriosis in the population (approximately 10%; ref. 1), most women included in these studies would not have had endometriosis. Our results for women without endometriosis are consistent with this. Two previous studies (14, 15) have suggested possible effect modification with alcohol intake, with the association between folate and reduced ovarian cancer risk limited to those with higher alcohol intake. Our results suggested that there was variation by alcohol intake for both women with and without endometriosis, but this was limited to supplement use and suggested increased risk at higher alcohol intakes. However, it is important to note that alcohol intake was low across the study populations so we had limited power to assess effect modification with high intakes. Our observations that the associations might also vary by aspirin or NSAID use in women with endometriosis are interesting but, given the lack of consistency between associations for dietary and supplemental intake and for aspirin and other NSAIDs, it is hard to draw any definitive conclusions.

Folate may have dual effects on cancer development and progression. Epidemiologic and laboratory studies in colorectal cancer have shown that, in normal cells, folate deficiency may lead to cancer progression through impaired DNA repair and increased mutations, while higher levels may be protective (9). However, in vitro and in vivo studies have shown that in the presence of established premalignant colorectal lesions with rapidly replicating cells, folate supplementation may accelerate progression to cancer by promoting further proliferation and progression (9). Although observations found in colorectal cancer may not necessarily be relevant for ovarian cancer, our results support a similar hypothesis.

Strengths of this study include the large sample size and pooled design. The retrospective self-reporting of dietary intake is a limitation and may potentially introduce measurement error. However, it is unlikely this would differ by endometriosis status, so it is unlikely to explain the observed difference in the ORs between women with and without endometriosis. It remains possible that the observed associations are due to unknown or unmeasured confounding factors; however, our model accounts for known risk factors associated with folate intake and risk of ovarian cancer. Associations with dietary folate intake could also be due to the kinds of foods that are fortified (e.g., bread). However, we also saw no association between grain intake, GI and GL, and ovarian cancer risk for women with endometriosis, so it is unlikely that the observed associations are due to food type rather than folate intake. We saw a similar pattern, although the individual estimates were not statistically significant, using MR. This consistency adds confidence that overall findings are unlikely to be due to bias or confounding.

It is possible that the null associations for supplement use could be due to residual confounding by socioeconomic factors that might affect access to and use of supplements (46) or potential misclassification by participants in the type or brand of supplement used. Additionally, we were unable to directly assess the impact of duration of supplement use, and it is possible that any increased risk may be observed only in long-term users.

A further limitation is that there are likely to be women with undiagnosed endometriosis in the no endometriosis group. Additionally, endometriosis status was self-reported and laparoscopic confirmation was not required, so some women may have been misdiagnosed. However, a recent study showed good agreement (84%) between self-reported endometriosis and medical records in general (47). Both situations would tend to make the groups look more similar leading to underestimates of the difference between women with and without endometriosis.

Age at endometriosis diagnosis was not routinely collected across studies, so we were unable to assess the relevance of timing of endometriosis diagnosis in relation to folate intake. However, in studies that had information, most women were diagnosed with endometriosis prior to dietary assessment (median = 19 years). Using MR, although underpowered, helped to address this issue by providing an unbiased estimate of folate intake unrestricted to a particular time point.

It is important to note that ovarian cancer is more common in older women; thus, the majority of women in this analysis were postmenopausal. Given the established benefits of folate supplementation during pregnancy for preventing neural tube defects and the rarity of ovarian cancer in women of reproductive age, women with endometriosis who plan to conceive should follow established guidelines for folic acid supplementation.

In summary, our results suggest that higher folate intake, particularly from dietary sources, may be associated with increased risk of ovarian cancer among women with endometriosis. There is a need for additional research to better understand the role of dietary and supplementary folate sources in ovarian cancer risk, especially the potential cancer-promoting effect of high folate intake in women with endometriosis.

K.L. Terry reports grants from the NIC and grants from Congressionally Directed Medical Research Program during the conduct of the study. B. Qin reports grants from NIH during the conduct of the study. F. Modugno reports grants from National Science Foundation and from the Department of Defense during the conduct of the study. H.A. Risch reports grants from NCI, NIH during the conduct of the study. P.M. Webb reports grants from U.S. Army Medical Research and Materiel Command, National Health and Medical Research Council of Australia, and Cancer Councils of New South Wales, Victoria, Queensland, South Australia and Tasmania, Cancer Foundation of Western Australia during the conduct of the study; grants from AstraZeneca and personal fees from AstraZeneca outside the submitted work. No disclosures were reported by the other authors.

K. Gersekowski: Formal analysis, methodology, writing–original draft, writing–review and editing. T.I. Ibiebele: Data curation, writing–review and editing. J.A. Doherty: Resources, writing–review and editing. H.R. Harris: Resources, writing–review and editing. M.T. Goodman: Resources, writing–review and editing. K.L. Terry: Resources, writing–review and editing. A.H. Wu: Resources, writing–review and editing. E.V. Bandera: Resources, writing–review and editing. B. Qin: Resources, writing–review and editing. J.-S. Ong: Methodology, writing–review and editing. J.P. Tyrer: Resources, writing–review and editing. S.C. Dixon-Suen: Methodology, writing–review and editing. F. Modugno: Resources, writing–review and editing. H.A. Risch: Resources, writing–review and editing. P.M. Webb: Conceptualization, resources, supervision, writing–original draft, writing–review and editing.

OCAC was funded by a grant from the Ovarian Cancer Research Fund (OCRF) and we are also grateful to the family and friends of Kathryn Sladek Smith for their generous support of the Ovarian Cancer Association Consortium through their donations to the OCRF. The OCAC OncoArray genotyping project was funded through grants from the NIH (CA1 × 01HG007491-01, U19-CA148112, and R01-CA149429), NIH (R01-CA058598; to M.T. Goodman), Canadian Institutes of Health Research (MOP-86727), and the OCRF. The COGS project was funded through a European Commission's Seventh Framework Programme grant (agreement number 223175—HEALTH-F2-2009-223175) and in part by the U.S. National Cancer Institute GAME-ON Post-GWAS Initiative (U19-CA148112). This study made use of data generated by the Wellcome Trust Case Control consortium that was funded by the Wellcome Trust under award 076113. Individual studies were funded by AUS: U.S. Army Medical Research and Materiel Command (DAMD17-01-1-0729; to P.M. Webb) and National Health and Medical Research Council of Australia (199600, 400413, 400281, and 1173346; to P.M. Webb); DOV: NIH R01-CA112523 and R01-CA87538 (to J.A. Doherty); HAW: NIH (R01-CA58598, N01-CN-55424, and N01-PC-67001; to M.T. Goodman); HOP: University of Pittsburgh School of Medicine Dean's Faculty Advancement Award (F. Modugno), Department of Defense (DAMD17–02–1-0669; to F. Modugno) and NCI (K07-CA080668, R01-CA95023, P50-CA159981, MO1-RR000056, and R01-CA126841; to F. Modugno); LAC: NIH (P01CA17054, P30CA14089, N01PC67010, and N01CN025403; to A.H. Wu), NIH (R01CA61132, R03CA113148, and R03CA115195), and California Cancer Research Program (00-01389V-20170; 2II0200; to A.H. Wu); MAC: National Institutes of Health (R01-CA122443, P30-CA15083, P50-CA136393 to E.L. Goode); Mayo Foundation; Minnesota Ovarian Cancer Alliance; Fred C. and Katherine B. Andersen Foundation (to E.L. Goode); Fraternal Order of Eagles; MAY: National Institutes of Health (R01-CA122443, P30-CA15083, P50-CA136393 to E.L. Goode); Mayo Foundation; Minnesota Ovarian Cancer Alliance; Fred C. and Katherine B. Andersen Foundation (to E.L. Goode); NCO: National Institutes of Health (R01-CA76016) and the Department of Defense (DAMD17-02-1-0666); NEC: National Institutes of Health R01-CA54419 and P50-CA105009 and Department of Defense W81XWH-10-1-02802 (to K.L. Terry); NJO: National Cancer Institute (NIH-K07 CA095666, R01-CA83918, NIH-K22-CA138563, and P30-CA072720; to E.V. Bandera) and the Rutgers Cancer Institute of New Jersey (to E.V. Bandera); NTH: Radboud University Medical Centre; OPL: National Health and Medical Research Council (NHMRC) of Australia (APP1025142, APP1120431 to P.M. Webb) and Brisbane Women's Club (to P.M. Webb); SEA: Cancer Research UK (C490/A10119 C490/A10124); UK National Institute for Health Research Biomedical Research Centres at the University of Cambridge; TBO: National Institutes of Health (R01-CA106414-A2), American Cancer Society (CRTG-00-196-01-CCE), Department of Defense (DAMD17-98-1-8659), Celma Mastry Ovarian Cancer Foundation; TOR: NIH grants R01 CA063678 and R01 CA063682 (to H.A. Risch); UCI: NIH R01-CA058860 and the Lon V Smith Foundation grant LVS-39420; UKO: The UKOPS study was funded by The Eve Appeal (The Oak Foundation) with salary contribution to Usha Menon and Aleksandra Gentry-Maharaj through MRC core funding (MC_UU_00004/01). The pooling and harmonizing of dietary information was supported by the Department of Defense Congressionally Directed Medical Research Programs (grant number W81XSH-16-2-0010). Funding was also provided by the National Science Foundation (DGE-2217399-Modugno; to F. Modugno). The views expressed are those of the authors and do not necessarily reflect the views of the National Science Foundation.

We thank all the study participants who contributed to this study and all the researchers, clinicians, technical and administrative staff who have made this work possible. We also acknowledge Linda Kelemen who contributed to the early stages of this work.

Acknowledgements for individual studies: AUS: The Australian Ovarian Cancer Study gratefully acknowledges additional support from Ovarian Cancer Australia and the Peter MacCallum Foundation. We acknowledge the Australian Ovarian Cancer Study Group for their contribution, as well as the cooperation of the participating institutions in Australia, and the contribution of the study nurses, research assistants and all clinical and scientific collaborators. The complete AOCS Study Group can be found at www.aocstudy.org. We would like to thank all of the women who participated in this research program; GER: The German Ovarian Cancer Study (GER) thank Ursula Eilber for competent technical assistance; NJO: We thank Drs. Sara Olson, Lisa Paddock and Lorna Rodriguez, and all participants and research staff at the Rutgers Cancer Institute of New Jersey, the New Jersey State Cancer Registry, and Memorial Sloan-Kettering Cancer Center; OPL: Members of the OPAL Study Group (http://opalstudy.qimrberghofer.edu.au/); SEA: SEARCH team, Craig Luccarini, Caroline Baynes, Don Conroy; UKO: We particularly thank I. Jacobs, N. Singh, M.Widschwendter, E. Wozniak, A. Ryan, J. Ford and N. Balogun for their contribution to the study.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.