History of infertility and risk of breast cancer: a prospective cohort study.

We utilized data from the Nurses’ Health Study II (NHSII), a prospective cohort study of 116,429 female registered nurses that began in 1989 when participants were 25–42 years old. Participants were enrolled if they returned a mailed health questionnaire. Follow-up questionnaires have been sent every two years to participants and collect data on health and lifestyle factors. The study protocol was approved by the institutional review boards of the Brigham and Women’s Hospital and Harvard T.H. Chan School of Public Health, and those of participating registries as required. Infertility history was defined as whether participants reported having “tried to become pregnant for more than one year without success” on any questionnaire. For this analysis, participants who ever reported having experienced infertility before the age of 40 were considered to have a history of infertility for the entirety of the analysis [ 15 ]. Women who reported having experienced infertility were asked their age at first reported infertility, which was categorized as ≤25, 26–30, >30 years old. Participants were also asked about the cause of their infertility and were able to report multiple causes: tubal blockage, ovulatory disorder, endometriosis, cervical mucus factors, male factor infertility, not investigated, not found, and/or other. In our cohort, women have been able to validly recall their infertility diagnoses with high accuracy compared to medical records (ovulatory infertility, 95% concordance with medical records [ 16 ], endometriosis, 97% concordance with medical records [ 17 ]). The main analyses compared women with a history of infertility to gravid women with no history of infertility. On every questionnaire, participants reported whether they had become pregnant and the outcome of that pregnancy. Time-varying information on parity was incorporated into our definition of primary infertility (infertility occurring prior to first live birth) and secondary infertility (infertility occurring after first reported live birth). Participants self-reported information on cancer diagnoses on every questionnaire. Participants who reported an incident breast cancer diagnosis were asked to report their date of diagnosis. We then acquired permission to obtain their relevant medical record. If the participant was deceased, next of kin were contacted for permission to review medical records or data were linked to state cancer registries. Based on medical records, breast cancer diagnosis was confirmed and information on histopathology, size, invasiveness, grade, node status, and hormone receptor status was obtained. Our primary analysis was restricted to invasive breast cancer diagnoses. For our categorization of breast cancer hormone receptors status, given previous research indicating that ER-/PR+ is not a reproducible subtype [ 18 ], we utilized information on estrogen and progesterone status of the tumor from medical records and tissue microarrays (primary categorization) to create three categories: estrogen /progesterone receptor positive (ER+/PR+) breast cancer, ER+/PR- breast cancer, and ER-/PR- breast cancer. Breast cancer was defined as premenopausal vs. postmenopausal based on self-reported menopausal status at time of cancer diagnosis. Participants who reported having experienced breast cancer or other cancer (other than non-melanoma skin cancer) prior to baseline in 1989 were excluded from the analysis. Person months at risk were calculated from participants’ first report of either infertility or gravidity until confirmed i) death ii) breast cancer diagnosis, iii) other cancer (other than non-melanoma skin cancer), or iv) return of the 2015 questionnaire. We used Cox proportional hazard models jointly stratified by calendar time with age (months) as the time metameter to calculate the hazard ratios (HR) and 95% confidence intervals (CI) of invasive breast cancer (Model 1). The proportional hazard assumption was tested and met using a likelihood ratio test for the interaction between exposure and time. Model 1 a priori adjusted for known confounders between infertility and breast cancer: breast cancer family history (yes/no), race (non-white/white), BMI at 18 (continuous), age at menarche (≤10, 11, 12, 13, or ≥14 years), height (continuous), oral contraceptive use before age 18 (never, <2 months, 10+ months). Model 2 additionally adjusted for time-varying benign breast disease history (yes/no), physical activity (0, 0.1–1.0, 1.1–2.4, 2.5–5.9, or ≥6 h/week), current BMI (continuous), alcohol intake (0, 0–15, >15 g/d), recent mammogram or breast exam (yes/no), hormone therapy (never users, duration of use: <5y or 5y+, and type of hormone therapy: estrogen only, combined estrogen progesterone, mixed used, combined), and age at menopause (54, unknown; among the postmenopausal). We also investigated age at first report of infertility (≤ 25, 26–30, >30), primary vs. secondary infertility, and self-reported infertility diagnoses as potential exposures. We recognized that some covariates may be on the causal pathway between infertility and risk of breast cancer, such as breastfeeding history, parity, age at first birth, body mass index, and type 2 diabetes history. For these covariates, we used mediation analyses to quantify the proportion of the association between infertility and invasive breast cancer that was mediated by specific covariates [ 13 ]. Specifically, the proportion mediated was calculated as the indirect effect divided by the total effect and tests were conducted for evidence of mediation [ 19 , 20 ]. To test the robustness of our findings we conducted several sensitivity analyses. We compared women who first experienced infertility at varying ages to age-matched gravid women, we excluded women who self-reported endometriosis who did not report a history of infertility, we re-categorized individuals who reported only spousal factor infertility as unexposed, we stopped updating infertility history at age 35 years, we excluded women who reported fertility treatment (ovulation induction) but did not report infertility and thus had been in our unexposed group, and lastly, we expanded our unexposed group to include all women regardless of gravidity. All tests of statistical significance were 2-sided.

Over 26 years of follow-up, 26,208 women reported ever having experienced infertility. At the study baseline, women with infertility were more likely to have a BMI ≥30 kg/m 2 (12.0% vs. 10%), they also were less likely to be parous (66% vs. 83%) and among the parous, more likely to report a later age at first birth (27 vs. 25 years) and less likely to report cumulative breastfeeding for >12 months (20.9% vs. 28.7%) ( Table 1 ). Among women with a history of infertility, the most commonly reported causes of infertility were ovulatory disorder (29%), causes not found (21%), spouse/partner factor (19%), and endometriosis (16%). During 2,152,762 person-years of follow-up, we observed 3,201 newly diagnosed cases of invasive breast cancer, 1,173 of which were post-menopausal. Overall, we observed no statistically significant association between history of infertility and risk of invasive breast cancer (HR: 1.05, 95% CI:0.97–1.14) ( Table 2 ). This pattern was consistent among premenopausal breast cancer (HR:0.93, 95% CI:0.83–1.03). However, we did observe an association between history of infertility and risk of postmenopausal breast cancer (HR:1.13, 95% CI:1.00–1.28). There was no difference in breast cancer risk observed by tumor hormone receptor status (P-heterogeneity=0.23). In mediation analyses ( Table 3 ), 50.3% (CI: 8.2–92.0) of the relationship between infertility and risk of postmenopausal breast cancer was mediated through parity and age at first birth. The association with age at first birth was stronger and responsible for 36.9% of the association after adjusting for total parity. Among postmenopausal breast cancer cases, the association with infertility was strongest among women who first reported their infertility at younger ages (P-value, test for linear trend: 0.03; infertility at ages ≤ 25 HR: 1.15 95% CI:0.97–1.37; infertility at ages 26–30 HR: 1.20 95% CI:1.00–1.43; infertility at ages >30 HR: 0.98 95% CI:0.78–1.23) ( Table 4 ) and among women who experienced primary infertility (HR: 1.25, 95% CI:1.08–1.43) ( Supplemental Table 1 ). We observed no meaningful differences by self-reported underlying causes of infertility ( Supplemental Table 2 ). We observed that women with a history of infertility had an older age at first birth ( Supplemental Figure 1 ). In sensitivity analyses comparing age-matched gravid women to women experiencing infertility at varying ages ( Supplemental Table 3 ), we observed an overall similar pattern of association when compared to all gravid women; women who first experienced infertility at younger ages (≤ 25 years old) were at greater risk of postmenopausal breast cancer than women who first experienced infertility at older ages (> 30). The overall association between infertility and postmenopausal breast cancer remained consistent in sensitivity analyses testing the robustness of our comparison population ( Supplemental Table 4 ).

We observed that women with a history of infertility had a modestly increased risk of invasive postmenopausal breast cancer. However, we observed no association with invasive premenopausal breast cancer or overall invasive breast cancer. The relation between infertility and postmenopausal breast cancer risk was strongest among women who experienced infertility at younger ages and women with primary infertility. Consistent with this, a large portion of the association between infertility and risk of post-menopausal breast cancer was mediated by age at first birth and parity. Existing research on the association between infertility and risk of breast cancer has been inconclusive [ 21 , 22 ]. The greatest challenge to this area of research has been disentangling the association between infertility, infertility treatment, and other reproductive factors such as parity on risk of breast cancer. It is well established in the breast cancer literature that higher parity, younger age at first birth and longer duration of breastfeeding are associated with a lower risk of breast cancer [ 23 , 24 , 2 ]. In studying the association between infertility and breast cancer, these relations are complex to disentangle with very few studies being able to take this information into account. In our analyses, we performed mediation analyses [ 13 ] to statistically separate the influence of infertility from the influence of mediation variables. We observed that the association between infertility and risk of postmenopausal breast cancer was partially mediated through age at first birth and parity, with age at first birth being the stronger of the two mediators. We observed no meaningful mediation by breastfeeding history. Age at first birth has been hypothesized to contribute to breast cancer incidence through a variety of mechanisms. Pregnancy is associated with cell proliferation and therefore, it is hypothesized that women who have an older age at first pregnancy may have more initiated cells in their breast tissue than women who experience an earlier age at first pregnancy. A recent meta-analysis of breast cancer risk after childbirth among 15 prospective cohort studies estimated statistically significant heterogeneity in breast cancer risk by age at first pregnancy (p-value:0.01) with women whose age at first pregnancy was >35 having the highest breast cancer risk [ 25 ]. Indeed, we observed that women with infertility had lower parity and greater age at first birth ( Supplemental Figure 1 ). We observed that the association with post-menopausal breast cancer was strongest among women who reported a younger age at first experiencing infertility, as well as among women who experienced primary infertility. These women with infertility earlier in life may have more severe infertility compared to women who experienced infertility at older ages due to naturally diminishing ovarian reserve. Earlier age at first reported infertility is also associated with later age at first birth and lower total parity, which we observed to be substantial mediators of our association with postmenopausal breast cancer. Prior research has hypothesized an association between BRCA-1 mutations, infertility, and breast cancer as BRCA-1 has been associated with lower anti-Mullerian hormone levels, a marker of ovarian response [ 3 ]. However, most breast cancers are not of genetic origin and thus this mechanism could not explain our findings. There has also been conflicting prior research on specific infertility diagnoses, such as polycystic ovary syndrome (PCOS) or endometriosis [ 26 ] and risk of breast cancer. However, recent meta-analyses have suggested that neither PCOS nor endometriosis are associated with increased risk of breast cancer, although there was a suggestion that this varied by tumor characteristics [ 27 ], study population, and type [ 26 ]. In our analysis, we did not observe meaningful differences in the risk of breast cancer by infertility diagnoses. However, this may be influenced by women with PCOS being categorized in the heterogeneous group of ovulatory infertility. This study has important strengths including its longitudinal follow-up, and detailed information on both infertility as well as breast cancer. However, our study must also be interpreted within the context of its limitations. There are inherent complexities in defining infertility that impact all studies in this field. To be defined as infertile [ 28 ], a woman must have been trying to become pregnant and have been unsuccessful for 12 months. There will be women in our population who chose not to pursue pregnancy and thus were not at risk to be identified as either having experienced infertility or a pregnancy. Thus, our main analyses were restricted to women who experienced pregnancy and never reported infertility as the comparison group. Participants in NHSII are not a random sample of US women, so some findings may not be directly generalizable to the entire population. However, despite the known racial/ethnic differences in some breast cancer endpoints, it is unlikely that the biological relations in this cohort will differ from women in general [ 29 , 30 ]. The high level of education and interest in health in the NHSII cohort are distinct advantages that aid our ability to collect high quality, valid information on self-administered forms. Since all the participants are nurses, possible confounding by socioeconomic status is also reduced. In summary, we observed that women with a history of infertility were more likely to experience postmenopausal breast cancer. This association was primarily driven by later age at first pregnancy and was stronger among women who experienced primary infertility and who experienced infertility and earlier ages, which may be markers of infertility severity.

Previous research has demonstrated that endogenous hormone levels, use of exogenous hormones, and reproductive history (e.g. parity, age at first birth, breastfeeding) are associated with the risk of breast cancer [ 1 , 2 ]. However, research on the association between infertility and risk of breast cancer has been inconsistent [ 3 , 4 ]; some studies have observed an increased risk of breast cancer following infertility (trying to conceive for 12 months without success) [ 5 – 9 ], while other studies have suggested there is no association between infertility and risk of breast cancer [ 10 – 12 ]. Recent research from the Swedish Multi-Generation Register [ 12 ] observed a modest association between infertility and risk of overall breast cancer (HR:1.05) in age-adjusted models, but not in models adjusted for reproductive characteristics related to infertility such as parity, age at first birth, salpingectomy, hysterectomy, and bilateral oophorectomy, which may act as potential mediators [ 13 ]. Most prior research on infertility and risk of breast cancer has not adequately taken information on reproductive characteristics into account which may have contributed to conflicting findings. Valid analysis of the association between infertility and risk of breast cancer is complex. Many well-established reproductive risk factors for breast cancer, such as nulliparity, number of pregnancies, and age at first pregnancy [ 2 ], are also strongly correlated with infertility. Additionally, breast cancer is a heterogeneous disease, with varying risk factors by tumor hormone receptor status and age at cancer onset (premenopausal vs. postmenopausal) that has not been adequately accounted for in previous analyses [ 14 ]. Lastly, a large body of research has focused on infertility treatment and breast cancer risk [ 4 ] with limited research on whether the underlying indication for fertility treatment influences risk. The majority of studies on infertility and risk of breast cancer had a short duration of follow-up (<10 years), lacked information on specific infertility experience (e.g. diagnosis type, age on onset, primary vs. secondary infertility), and/or had small numbers of cancer cases, which may also have contributed to prior conflicting results. To overcome these outlined limitations existing in the current literature, we utilized data from the Nurses’ Health Study II, a prospective cohort study of over 116,000 women. We incorporated information on overall infertility, infertility diagnoses, age at onset of infertility, and whether infertility occurred in nulliparous (primary infertility) or parous (secondary infertility) women. We also incorporated detailed information on breast cancer, investigating differences according to menopausal status at breast cancer onset and tumor hormone receptor status.