Infertility and risk of postmenopausal breast cancer in the women's health initiative.

The WHI is a longitudinal cohort study comprised of 161,808 women [ 29 ]. Participants 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)[ 30 ]. At baseline, participants were postmenopausal and between the ages of 50-79 years old [ 31 ]. Women were excluded at baseline from the CT and OS WHI studies if they had medical conditions predictive of survival time of less than 3 years. Women were excluded from the CT if they had invasive cancer in the last 10 years or had breast cancer at baseline. Participants were scheduled to complete the initial WHI study activities in March 2005. At that time, participants were invited to join the first WHI Extension Study for an additional 5 years of follow-up (n=115,400 enrolled). In 2010 and 2020, current participants were invited to continue for additional follow-up as part of the second and third Extension Studies, respectively. In the fourth Extension Study, follow-up will continue through 2027 (n=93,567 enrolled). Annual questionnaires were collected by mail using procedures similar to those used during the initial WHI observational study. For the present analysis, participants were excluded if they had missing data on infertility (n=1,644), missing data on prevalent cancer at baseline (n=1,153), had prevalent cancer at enrollment, other than non-melanoma skin cancer at baseline (n=14,151), had missing outcome data over follow-up (n=616), or were nulliparous with no history of infertility (12,460), leaving 131,784 women in the analytic cohort ( Supplemental Figure 1 ). 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 study baseline, participants completed a reproductive history questionnaire. Participants were asked whether they had ever tried to become pregnant for more than one year without becoming pregnant. Participants who responded “yes” were considered to have a history of infertility [ 32 , 33 ]. Participants were then asked whether they visited a doctor or clinic because they could not get pregnant and whether a reason was identified. Participants could identify multiple reasons for infertility (ovulatory, tubal or uterine, endometriosis, male, other, not known). We investigated heterogeneity in infertility diagnoses by looking separately at self-reported infertility types: ovulatory, tubal or uterine, endometriosis, male, other, not known. Infertility, by definition, can only be diagnosed among people trying to conceive; therefore, our primary comparison group was parous women without infertility. In sensitivity analyses, we expanded our comparison group to include all women who did not report a history of infertility. All participants in the CT underwent regular screening mammograms. Participants completed self-reported health assessment questionnaires annually (OS and Extension) or semi-annually (CT), which collected information on health outcomes, including breast cancer. Women who reported breast cancer were contacted by WHI field center staff to obtain additional details on health event dates, providers, and a signed release to obtain medical records, if needed. All invasive cancers were documented, coded according to primary site and adjudicated by trained physicians breast cancers were coded for histology and estrogen and progesterone receptors. Cases were coded according to the National Cancer Institute Surveillance, Epidemiology, and End Results (SEER) guidelines [ 34 ]. 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. Information on age, highest level of formal education, oral contraceptive use history, age at menarche, BMI at baseline, smoking status, marital status, physical activity at baseline, alcohol use, pregnancy history, and menstrual cycle regularity were collected at cohort baseline. Missingness is presented in Supplemental Table 1 . Missing covariates were addressed with multiple imputation using chained equations (MICE) [ 35 ]. We used Cox proportional hazards models stratified by age (months) to calculate the hazard ratios (HR) and 95% confidence intervals (CI) of postmenopausal breast cancer (Model 1). The proportional hazard assumption was tested and met using a likelihood ratio test for the interaction between the covariates used and time. Model 2 was additionally adjusted for variables a priori considered confounding variables [ 36 ] or variables associated with WHI study selection including, education (1: less than high school (reference group), 2: High school or GED completed, 3: vocational training, technical school or some college, 4: college degree or more), oral contraceptive use (never (reference group) or ever), age at menarche (10 or less, 11-12, 13-14 (reference group), 15+), BMI category (25-30kg/m 2 is reference group) at baseline, smoking status at baseline (never (reference group), former, current), marital status (married/partnered ever or never (reference group)), physical activity (MET/hr/week), alcohol use at baseline (never (reference group), former, minimal or moderate to heavy), and WHI trial arm (OS is reference group). Given that menstrual cycle irregularity is a consequence of some infertility diagnoses and is associated with risk of breast cancer [ 37 ] model 3 was further adjusted for menstrual cycle regularity (yes, sometimes, no (reference group)). In additional analyses, we investigated separately the association with invasive breast cancer and with specific infertility causes (ovulatory disorders, tubal or uterine factors, endometriosis, male factor, other, not known). Some reproductive health variables associated with breast cancer may be on the causal pathway between infertility and risk of postmenopausal breast cancer [ 38 ]. Therefore, for incident breast cancer we calculated the proportion mediated as the indirect effect divided by the total effect [ 38 - 40 ] for parity, postmenopausal hormone therapy use at baseline, age at first term pregnancy, age at menopause, breastfeeding history, and oophorectomy history. We also looked separately at breast cancer tumors by hormone receptor status (estrogen/progesterone receptor) and molecular subtypes using separate models. In sensitivity analyses, we expanded our comparison group to include nulliparous women without a history of infertility. We also performed secondary analyses using logistic regression models in which our outcome of interest was prevalent breast cancer at WHI baseline. For these analyses we no longer excluded prevalent cancers at baseline. These models adjusted for the same variables as the primary analyses.

Approximately 18% of participants reported having experienced infertility. We observed no meaningful differences in age, BMI at baseline, BMI at age 18, age at menarche, and history of a female relative with breast cancer between women with or without a history of infertility ( Table 1 ). Participants with a history of infertility were more likely to be White (88.4 vs. 87.0), to have a college degree or higher (41.5 vs. 36.8), to report irregular periods (12.2 vs. 6.6), to partake in moderate to heavy alcohol intake (38.9 vs. 36.7), to have an older age at first birth (30+ at first birth: 20.5 vs. 7.3), and to have had a bilateral oophorectomy (21.7 vs. 17.7) compared to parous women without a history of infertility. Women with a history of infertility were less likely to have been pregnant (84.6 vs. 100), to have had multiple live births (2 or more: 78.2 vs. 91.7), to have breastfed for 7 or more months (21.4 vs. 29.1), and to have used oral contraceptives (40.2 vs. 43.5) than parous participants without a history of infertility. We observed that women with a history of infertility had a modestly higher risk for postmenopausal breast cancer as compared to women without infertility (HR: 1.07, 95% CI: 1.02-1.13; HR Invasive breast cancer: 1.06, 95% CI: 1.02-1.12) ( Table 2 ). We observed no meaningful differences in risk by tumor hormone receptor types and molecular subtypes ( Supplemental Table 2 ). However, the effect estimates were slightly larger among ER+ tumors (HR: 1.06, 95% CI: 0.99-1.12) and Luminal A tumors (HR:1.08, 95% CI: 1.00-1.17). When investigating the association with specific infertility diagnoses, we did not observe associations with hormonal/ovulatory infertility, tubal factor infertility, endometriosis, or other causes of infertility ( Table 3 ). We did observe an association between male factor infertility and risk of postmenopausal breast cancer (HR: 1.21, 95% CI: 1.07-1.38). This association with male factor infertility was primarily observed among participants with an older age at first birth (>30 years HR: 1.50, 95% CI: 1.14-1.97; ≤ 30 years HR: 1.15, 95% CI: 0.95-1.40), however we did not observe statistically significant effect modification by age at first term birth (p-value, test for interaction: 0.12). When potential mediating factors were investigated, the association between infertility and risk of breast cancer was most strongly mediated by age at first term pregnancy (% mediated: 46.4%, 95% CI: 12.2-84.3%) ( Table 4 ). In sensitivity analyses where nulliparous women with no history of infertility were included in the comparison group (HR: 1.06, 95% CI: 1.01-1.12) ( Supplemental Table 3 ) the overall observed associations did not meaningfully change. In analyses investigating associations with prevalent breast cancers at baseline we observed a modest statistically significant association between the history of infertility and risk of breast cancer (OR: 1.12, 95% CI: 1.04-1.20) ( Supplemental Table 4 ). We also observed an association between infertility attributed to hormonal/ovulatory infertility (OR:1.27, 95% CI: 1.01-1.61), infertility attributed to tubal factor (OR:1.21, 95% CI: 1.02-1.43), and infertility attributed to male factor (OR:1.22, 95% CI: 1.03-1.44) and risk of breast cancer.

Overall, we observed that women with a history of infertility, as compared to those without a history of infertility, had higher risk of postmenopausal breast cancer. This risk was small and was predominately driven by later age at first birth. Future breast cancer research should continue to incorporate information on a spectrum of reproductive health risk factors, including consistently defined measures for infertility, to inform on clinical care among women experiencing infertility.

Overall, we observed that history of infertility was associated with a small risk of postmenopausal breast cancer. This association was primarily driven by the indirect association between infertility and age at first term pregnancy, where women with infertility were more likely to report an age at first term pregnancy over 30 years which mediated the association with breast cancer. The prior literature on infertility history and risk of breast cancer has been inconsistent, with many studies unable to separate the influence of IVF and infertility history overall [ 20 , 41 ] and very few studies able to investigate postmenopausal breast cancer. Much of the prior research has compared the expected breast cancer risk in the population to the breast cancer risk among individuals with a history of infertility or who have used fertility treatments, and most of these studies have observed a modest increased risk of breast cancer [ 2 , 4 , 18 , 42 , 43 ]. Recent research from the Swedish multi-generational registrar, which was able to incorporate postmenopausal breast cancers, observed an association between infertility and overall breast cancer (HR: 1.05, 95% CI:1.00-1.10) as well as with postmenopausal breast cancer (HR: 1.11, 95% CI:1.04-1.18) in age-adjusted models [ 11 ]. However, these associations attenuated completely in multivariable models that adjusted for age at first birth and parity (HR breast cancer overall: 0.96; HR postmenopausal breast cancer: 1.03). Recent research in the Nurses’ Health Study II, observed no association between history of infertility and overall invasive breast cancer (HR: 1.05, 95% CI: 0.97-1.14), but a modest association between infertility history and postmenopausal invasive breast cancer (HR:1.13, 95% CI: 1.00-1.28) [ 44 ], similar to the current findings. Consistent with our findings observed in the WHI, the findings in the NHSII were meaningfully mediated by total parity and age at first birth (% mediated: 50%, 95% CI: 8.2-91.0). The association between reproductive history, such as parity and age at first birth and risk of breast cancer is well established. Indeed, the National Cancer Institute’s Breast Cancer Risk Assessment Tool (BCRAT) already includes information on age at first live birth of child [ 45 ]. These tools fall short in the exclusion of fertility history. Prior research on infertility and risk of breast cancer rarely accounted for information on parity history due to the nature of registry data. Our results suggest that the observed association between infertility and risk of postmenopausal breast cancer in our population was primarily mediated by age at first term pregnancy (% mediated: 46.4%), where women with a history of infertility were more likely to be over age 30 y at first term pregnancy (age at first term pregnancy ≥ 30: 20.6%) compared to parous women without a history of infertility (7.3%). Prior research on infertility diagnoses has shown mixed associations, with some studies suggesting associations between endometriosis [ 46 , 47 ] and ovulatory infertility (with the most common diagnosis of polycystic ovary syndrome) [ 48 , 49 ] with risk of breast cancer. However, in our main analyses, we did not observe any statistically significant association between specific infertility diagnoses and risk of postmenopausal breast cancer except among individuals with a history of infertility attributed to their male partner. Women with male partner factor infertility may face additional obstacles to conceive including longer time to conceive and older age at first birth. We observed that the association between male factor infertility and risk of breast cancer was stronger among those who experienced their first birth after age 30. While low sample size precluded us from rigorously exploring in the present analysis, future research should investigate heterogeneity by race and/or ethnicity or other factors that could contribute to this finding with male factor infertility. In secondary analyses, where our primary outcome was prevalent (pre or post-menopausal) breast cancer at baseline, we observed that infertility attributed to ovulatory/hormonal factors and to tubal factors were associated with a small risk of breast cancer. This finding is consistent with some prior literature that has observed an association between polycystic ovary syndrome (PCOS) and premenopausal breast cancer [ 49 ] but not postmenopausal breast cancer. Our study has many strengths including its detailed information on infertility, large sample size, ability to consider potential mediating variables, and sufficient follow-up duration to investigate postmenopausal breast cancer. However, it also has important limitations. It is important to recognize the inherent complexities in defining infertility and related conditions that impact all studies in this field. To be defined as infertile [ 50 ], a woman must have been trying to become pregnant and have been unsuccessful for 12 months. There has been limited research on the validity of self-reported infertility after extended follow-up; however, research among women who underwent fertility treatment observed that most women were able to accurately recall their infertility diagnoses after 20 years [ 51 ], reducing concerns about the potential for misclassification. Moreover, to further minimize misclassification, our primary analysis defined our unexposed group as parous women who have never reported infertility. Unfortunately, we did not have information on age at infertility or primary/secondary infertility, which precludes us from studying the contribution of age at infertility on postmenopausal breast cancer risk, which has proved informative in other studies [ 44 ]. WHI is not a random sample of US women. However, participants were chosen to be representative of the racial and ethnic distribution of US women aged 50 to 79 when the cohort began. At the end of the recruitment period, approximately 17% of participants were from racial or ethnic minority groups [ 30 , 52 ]. Prior research has suggested that while the WHI population is representative of the US population of similar ages and cohort enrollment for BMI, WHI participants tended to be healthier (lower baseline prevalence of hypertension, diabetes, cardiovascular disease and medicated high cholesterol) and of higher socio-economic status (higher household income and level of education completed) when compared to the general US population [ 53 ]. Despite these differences, it is unlikely that the biological associations between infertility and risk of breast cancer will differ from US women in general [ 30 ]. Given that the diagnosis for some precursor lesions and early cancer may be contingent on whether cancer screening has occurred, there may have been over-reporting of cancer diagnoses. For this reason, all CT participants underwent regular screening mammograms at each time of contact during follow-up as part of the WHI study protocol. Initial WHI study exclusion criteria excluded women who had medical conditions predictive of survival time of less than 3 years. Women were excluded from the CT if they had invasive cancer in the last 10 years or breast cancer at baseline. In our primary analysis, we excluded women with history of cancer at study baseline. The median age of breast cancer diagnosis among US women is 62 years old [ 54 ]; therefore, our primary analysis is representative of a large proportion of breast cancer diagnoses. Moreover, in sensitivity analyses, we separately analyzed prevalent breast cancer diagnoses at baseline and observed similar patterns of associations. IVF was not widely available to the vast majority of WHI participants given their age. Therefore, the findings of our study may not be generalizable to infertile women of today’s age due to differences in fertility awareness, healthcare accessibility, and clinical practice between these two generations. Additionally, there is the possibility that some of the participants utilized other early fertility treatments available prior to IVF, including clomiphene and gonadotropins. The most comprehensive and largest studies on breast cancer [ 42 , 55 , 56 ], have found no clear association between these early fertility treatment types and cancer risk. Therefore, we expect any potential effect of not including this information to be minimal.

Reproductive factors such as low parity, older age at first birth, and never breastfeeding have been associated with an increased risk of breast cancer [ 1 ]. However, research on infertility (i.e., 12 months of trying to conceive without success), which is strongly associated with these reproductive breast cancer risk factors, and its relation to breast cancer risk has produced mixed findings [ 2 , 3 ]. A number of studies have reported an increased risk of breast cancer [ 4 - 8 ], while other studies have observed no clinically or statistically significant association [ 9 - 11 ]. Infertility is common, affecting 1.5 million women each year in the United States [ 12 , 13 ]. Aside from skin cancer, breast cancer is the most common cancer diagnosed in women in the United States [ 14 ], and greater than 85% of all breast cancer diagnoses are among postmenopausal women [ 15 ]. Prior studies on infertility and risk of breast cancer are limited by their short follow-up; as a result, those studies have focused on premenopausal breast cancer rather than postmenopausal breast cancer [ 5 , 16 - 22 ]. Additionally, pre-menopausal breast cancer has a different risk factor profile than postmenopausal breast cancer [ 23 ], leaving an important, unfilled gap in the existing literature on the association between infertility and postmenopausal breast cancer. Thus, advancing our understanding of the relationship between infertility and postmenopausal breast cancer is essential to establishing recommendations for care for all women experiencing infertility. Another limitation of available studies on infertility and breast cancer is the conflation of the diagnosis of infertility and treatments for infertility, namely in-vitro fertilization (IVF). Prior studies have shown a greater risk of breast cancer among women treated with IVF compared to the general population [ 24 - 26 ] with little emphasis on risk among women experiencing infertility independent of IVF. Therefore, separating the influence of infertility history and fertility treatment on cancer risk is an important contribution to the literature in this area. The Women’s Health Initiative (WHI) is a large, longitudinal study of postmenopausal women with over 25 years of follow-up. Through utilizing this data source, we can address many of the limitations in the prior literature. Specifically, given the timing of WHI enrollment (1993-1998) and sampling structure (age range 50-79, 30% of participants 50-59 years old), we would expect that none of our participants were eligible for IVF. Indeed, the first successful birth from IVF took place in 1981 in the US, and access to IVF was not geographically available to many until the 1990s [ 27 ], when our youngest participants were past reproduction and approaching menopause. Given that even today the majority of women who experience infertility in the US do not access IVF [ 28 ], our approach negates the influence of IVF and focuses on the influence of infertility history accounting for reproductive history. An additional strength is that our analysis focused on incident postmenopausal breast cancer, the most common timing for diagnosis, and adjusted for reproductive breast cancer risk-factors that may be associated with infertility.