Infertility and Risk of Ovarian Cancer in the Women's Health Initiative.

The WHI ( n = 161,808) is a longitudinal cohort study comprised of postmenopausal participants (ages 50–79 years old) who were originally enrolled between 1993 and 1998 in one of three randomized clinical trials (CT) (hormone therapy, dietary modification, and calcium/vitamin D supplementation; n = 68,132) or an observational study (OS) ( n = 93,676) [ 21 – 23 ]. The WHI did not include women in either the CT or OS if they had medical conditions predictive of survival time of less than 3 years, or the CT if they had invasive cancer in the last 10 years or had breast cancer at baseline. The WHI was planned to be completed in March 2005; however, participants have been invited to complete extension studies with follow-up planned through 2027 ( n = 93,567 enrolled) using annual questionnaires using procedures similar to those used during the initial WHI observational study. For this analytic sample, we excluded participants if they had missing data on history of infertility ( n = 1,644) or missing data on prevalent ovarian cancer at baseline ( n = 1,376). We also excluded participants if they had prevalent ovarian cancer at enrollment ( n = 875), had missing outcome data over follow-up ( n = 660), had a bilateral oophorectomy ( n = 33,515), or were nulligravid with no history of infertility (18,869) (rationale described below), which gave us an analytic sample of 112,925 participants. This study was reviewed and approved by the IRB of the Fred Hutchinson Cancer Center in accordance with U.S. Department of Health and Human Services regulations at 45 CFR 46 (approval number: 3467-EXT), and this analysis was approved by the University of Arizona IRB (Protocol number: 2011237760). Participants provided written informed consent to participate. Additional consent to review medical records was obtained through signed written consent. At time of enrollment, participants in the WHI completed a reproductive history questionnaire. They were asked whether they had ever tried to become pregnant for more than one year without becoming pregnant as is a standard definition of infertility for epidemiologic research [ 24 , 25 ]. For this analysis, participants who responded “yes” were considered to have a history of infertility, as is consistent with other analyses in WHI [ 26 , 27 ]. In order to be eligible to receive a diagnosis of infertility, participants had to have been trying to get pregnant; thus, our primary comparison group was to “fertile” participants defined as gravid participants without a history of infertility. In sensitivity analyses, we expanded this comparison group to include all participants regardless of parity history who did not report a history of infertility. Participants were also asked whether they visited a doctor or clinic because they could not get pregnant and whether a reason was identified. Participants could self-report multiple reasons for infertility, including ovulatory, tubal or uterine, endometriosis, male, other, and not known. For our analysis, we analyzed self-reported infertility types (ovulatory, tubal or uterine, endometriosis, male, other, and not known) separately compared to fertile participants. All participants in the WHI completed self-reported questionnaires annually (OS and extension) or semi-annually (CT) for information on incident disease outcomes including ovarian cancer. Any participant who self-reported an ovarian cancer diagnosis was contacted by WHI staff to obtain additional details on the date of the health event, providers, and a signed release to obtain medical records. All invasive cancers were documented, coded according to primary site and adjudicated by trained physicians. Cases were coded according to the National Cancer Institute Surveillance, Epidemiology, and End Results (SEER) guidelines [ 28 ]. A blinded quality assurance sample was re-coded by a different coder to determine inter-coder variability. Unusual or difficult-to-code cases were reviewed with a reference cancer pathologist who performs a similar function for the Seattle-Puget Sound SEER registry. Vital status was collected periodically through follow-up of participants and surrogates by each clinical center. We were also able to separately investigate high grade serous, endometrioid, clear cell, and mucinous histotypes. We grouped endometroid and clear cell histotypes together given their well-established association with endometriosis [ 12 , 16 , 29 ]. We used Cox proportional hazards models stratified by age (months) to calculate the hazard ratios (HR) and 95% confidence intervals (CI) of postmenopausal ovarian cancer overall and by histotype (Model 1). The proportional hazard assumption was tested and met using a likelihood ratio test for the interaction between the covariates used and time. The WHI collected detailed self-reported information at study baseline on age, highest level of formal education, oral contraceptive use history, age at menarche, body mass index (BMI; kg/m 2 ), smoking status, marital status, physical activity, alcohol use, pregnancy history, and menstrual cycle regularity. Model 2 was additionally adjusted for variables a priori considered confounding variables [ 30 ] or variables associated with WHI study selection, including: education (less than high school, high school or GED completed, vocational training, technical school or some college, and college degree or more), oral contraceptive use (never, ever), age at menarche (10 or less, 11–12, 13–14, 15 +), parity (never pregnant, no term pregnancies, 1 term pregnancy, 2 term pregnancies, 3 or more term pregnancies), BMI (18.5–24.9, 25–29.9, ≥ 30 kg/m 2 ) at baseline, smoking status at baseline (never, former, current), marital status (married/partnered ever, never), physical activity (MET/hr/week), alcohol use at baseline (never, former, minimal, moderate to heavy), WHI trial arm (OS, CT), and menstrual cycle regularity (yes, sometimes, no). In secondary analyses, we investigated separately the association between specific infertility causes (ovulatory disorders, tubal or uterine factors, endometriosis, male factor, other, not known) and risk of ovarian cancer. Missing covariates were addressed with multiple imputation using chained equations (MICE) [ 31 ]. Competing risk analysis was used to examine whether the associations differed by histotype. Using a likelihood ratio test, a model with separate HR for each histotype was compared with a model with a common HR. In sensitivity analyses, we: (1) expanded our comparison group to include nulligravid participants without a history of infertility and (2) excluded nulliparous women from the participants with a history of infertility.

Among 112,925 participants at baseline, 19,350 reported having experienced infertility (17%), whereas 93,575 participants were fertile without a history of infertility ( Table 1 ). Participants with a history of infertility did not differ by age at baseline, or marital status. Participants with a history of infertility were more likely to be White (88.0 vs. 85.7%), be a college graduate (43.0 vs. 38.1%), to report irregular periods (12.0 vs. 6.4%), to be a current or former smoker (51.4 vs. 48.9%), and to report first birth after age 30 (17.3 vs. 7.9%) compared to fertile participants. Seventy five percent ( n = 14,481) of participants with infertility reported that they had visited a doctor or clinic because they could not get pregnant and of these, approximately 57% ( n = 8,192) reported that a reason was identified. The reasons reported included hormonal/ovulatory (17%), tubal/uterine factor (32%), endometriosis (12%), male/partner factor (34%), and other cause (18%) with participants able to report multiple reasons. Infertility was not statistically significantly associated with greater risk of ovarian cancer in crude and multivariable adjusted models (HR: 1.09, 95% CI 0.92–1.29) ( Table 2 ). Results were similar when the comparison group was expanded to include nulligravid women without a history of infertility (HR: 1.05, 95% CI 0.90–1.22) and when women who were nulliparous were excluded from the group with a history of infertility (HR: 1.09, 95% CI 0.92–1.29). When we examined ovarian cancer histotypes, participants with a history of infertility had nearly a twofold increased risk of endometrioid and clear cell ovarian cancer (HR: 1.90 95% CI 1.09–3.34) compared to fertile participants. We observed no association between history of infertility and risk of high grade serous or mucinous ovarian cancer ( Table 2 ), however the p for hetereogeneity between histotypes was not statistically significant ( p = 0.44). In analyses looking separately at specific infertility diagnoses, we did not observe statistically significant associations between infertility diagnoses and risk of ovarian cancer ( Supplementary Table 1 ); however, we were not able to examine infertility diagnoses by ovarian cancer histotypes due to small numbers.

Overall, we observed that postmenopausal participants who had a history of infertility were not at increased risk for ovarian cancer when all histotypes were grouped together. We did observe a statistically significant, nearly twofold greater risk of endometrioid/clear cell ovarian cancer histotypes for infertile participants compared to fertile participants. Specific infertility diagnoses were not found to be associated with ovarian cancer risk; however, our analyses had limited statistical power to detect these associations, as not all participants reported a reason for their infertility. Many prior studies of infertility and ovarian cancer risk have focused on populations receiving fertility treatments [ 3 , 5 – 9 , 32 – 34 ], making it difficult to separate the impacts of infertility from its treatments (i.e., IVF), and often examining a population with a lower age at ovarian cancer diagnosis than the average. For example, data linking ART patients in New York, Texas, and Illinois ( n = 167,085) with state cancer records observed a modest, but not statistically significant association with ovarian cancer (SIR: 1.18, 95% CI 0.87, 1.56); however, the mean follow-up was only 4.9 years from time of IVF [ 6 ]. Birth registry data from Norway ( n = 806,248) compared women who utilized ART to non-ART users and observed a modest risk of ovarian cancer among ART users (HR: 1.67 95% CI 1.02–2.74), which attenuated and was no longer statistically significant after covariate adjustment (HR: 1.56, 95% CI 0.94–2.60) and when follow-up was restricted to great than 10 years (HR: 0.96, 95% CI 0.39–2.33) [ 7 ]. Our population was ART naïve, allowing for the ability to separate infertility history from ART treatments and access to more specialized health services [ 35 ]. Similar to our results, the New England Case–Control (NECC) study of ovarian cancer, showed no statistically significant association between infertility and overall ovarian cancer risk for female or male factor infertility [ 1 ]. Ness et al. utilized data from eight population-based case–control studies seeking to separate the association between infertility and infertility treatments (Clomiphene and Gonadotropin utilizing treatments) with regard to ovarian cancer risk. Notably, they observed no association between fertility drug use and invasive ovarian cancer, but did observe increasing risk of ovarian cancer with years attempting to achieve pregnancy, potentially a marker of infertility severity [ 13 ]. Ovarian cancer is a heterogeneous disease that shows consistent differences in associations across risk factors by ovarian cancer histotypes [ 29 ]. A limitation of most prior studies of the infertility-ovarian cancer association was lack of examination of associations by histotype. In this study, we observed that participants with a history of infertility had a higher risk of endometrioid/clear cell (HR: 1.90 95% CI 1.09–3.34), while no association was observed with either high grade serous or mucinous. Consistent with our results, the NECC study observed an association between female factor infertility and risk of endometrioid/clear cell (odds ratio [OR]): 1.54; 95% CI = 1.09–2.16), but no associations with other histotypes [ 1 ]. These results may reflect the established association between endometriosis and risk of endometrioid and clear cell ovarian cancers [ 12 , 16 , 29 ]. Unfortunately we do not have the statistical power in our analysis to further disentangle whether endometriosis related infertility is driving the association with endometrioid and clear cell ovarian cancers. Endometriosis is a heterogeneous gynecologic condition and not all women with endometriosis will experience infertility [ 36 ]. While we did not observe an association with endometriosis as an infertility diagnosis and ovarian cancer risk overall, over 56% of participants reporting infertility did not report an infertility diagnosis, and this may reflect a time period in which a surgical diagnosis of endometriosis, which is considered the gold-standard for diagnosis, was less common. This analysis had numerous strengths including its incorporation of in-depth questionnaire data on infertility and relevant covariates, adjudicated ovarian cancer data, information on ovarian cancer histotype, and longitudinal follow-up among a large cohort of postmenopausal women. However, this analysis also has important limitations to acknowledge. Our exposure, infertility, was defined as having tried to become pregnant for one year without achieving pregnancy. To reduce the potential for misclassification of infertility among the comparison group, we compared participants who reported a history of having experienced infertility to participants who never reported a history of infertility and who were also gravid as the nulligravid group may include both those who had infertility (both diagnosed and undiagnosed) and those who were nulligravid by choice. Some gynecologic conditions which may influence infertility, such as endometriosis, may be undiagnosed in the comparison population. We would hypothesize that any undiagnosed endometriosis may attenuate observed associations. Another limitation was that due to the limited sample size we could not conduct analyses stratified by both specific infertility diagnosis and histotype. Our findings could have limited generalizability as the WHI study is not a random sample of the U.S. population; moreover, enrollment excluded participants with predicted survival of less than three years. Our questionnaire did not collect information on age at infertility or fertility treatment utilization including Clomiphene; however, given the age of our population, ART, including IVF, would not have been available for most participants. Prior research has not shown a strong association between Clomiphene exposure and risk of ovarian cancer [ 13 , 34 ]. While WHI did not collect information on Clomiphene, future research should investigate this further especially among women who remain nulligravid. Overall, in this population of postmenopausal participants, we did not observe a statistically significant association between infertility history and incident ovarian cancer, nor by type of infertility diagnosis, with the exception of clear cell and endometroid subtypes.

There has been a consistent body of research supporting reproductive factors, including lifetime ovulatory cycles, parity, and age at first birth, as being associated with risk of ovarian cancer [ 1 , 2 ]. However, infertility, which may reduce total lifetime parity, delay age at first birth, and be associated with both a higher and lower number of lifetime ovulatory cycles, has demonstrated conflicting results with ovarian cancer risk [ 1 , 3 – 9 ], necessitating additional research on overall infertility and risk of ovarian cancer. Moreover, prior research has suggested differences in ovarian cancer risk by gynecologic conditions that can impact fertility [ 7 , 8 , 10 – 15 ]. For example, it has been consistently shown that individuals with endometriosis, for whom endometriosis may impact fertility, have a two to threefold increased risk of clear cell, endometrioid, and low-grade serous ovarian cancer [ 14 , 16 ]. In contrast, polycystic ovary syndrome (PCOS), which can cause ovulatory infertility, is associated with lower risk of ovarian cancer [ 10 , 11 ], likely due to a reduced number of lifetime ovulatory cycles. Limitations in the prior literature may have contributed to the conflicting results for overall infertility and risk of ovarian cancer. Much of the literature on overall infertility and ovarian cancer has focused on the contribution of Assisted Reproductive Technologies (ART)—specifically In Vitro Fertilization (IVF)—on risk [ 3 , 5 – 9 ]. This emphasis may combine the contribution of the biologic experience of infertility with the ability to access ART and the exogenous hormones utilized in ART treatment. Additionally, few studies of overall infertility and risk of ovarian cancer have examined associations by ovarian cancer histotypes, an essential consideration given that this heterogeneity may dilute associations when exposure effects are not similar across histotypes [ 17 , 18 ]. Differences across studies could have also been influenced by different study populations (e.g., population-based vs. clinical population/registry of IVF) and short duration of follow-up, which has resulted in an emphasis on ovarian cancer diagnosed before menopause [ 6 , 8 ]. The median age of ovarian cancer diagnosis is 63 years; [ 19 ] therefore, prior research with limited follow-up may not be generalizable to the majority of ovarian cancer diagnoses, which occur among postmenopausal women. Information on parity, a known risk factor for ovarian cancer, has also not been adequately adjusted for in many registry studies, which may contribute to residual confounding of studies investigating overall infertility and ovarian cancer risk. To fill gaps in the prior literature, our objective was to first investigate history of overall infertility and risk of postmenopausal ovarian cancer among participants in the Women’s Health Initiative (WHI), in secondary analyses we will investigate contributions of infertility diagnoses. WHI a prospective cohort of over 160,000 postmenopausal women who have been followed for over 25 years. The rich data collection and disease adjudication, longitudinal follow-up among postmenopausal women, and population-based cohort design mitigate some of the limitations in prior literature and the sample size supports evaluation of ovarian cancer associations overall and by histotype (high grade serous, endometroid, clear cell, mucinous). Additionally, the age of participants at WHI enrollment in 1993–1998 (50–79 years) allows for investigation of this question among an ART-naïve population [ 20 ].

Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s10552-025-01962-z .