Effect of Reproductive Factors and Oral Contraceptives on Breast Cancer Risk in BRCA1/2 Mutation Carriers and Noncarriers: Results from a Population-Based Study

Women with mutations in BRCA1/2 have a high risk of developing cancers of the breast and ovaries (1-3). The exact mechanisms that account for why BRCA1/2 mutation carriers predominantly develop cancers of these hormonally regulated organs are not clear, but it is possible that BRCA1/2 may interact with estrogen in breast carcinogenesis. Estrogen is involved in breast carcinogenesis by promoting cell proliferation and/or by acting as a genotoxic agent, through its metabolites, generating mutagenic DNA damage (4, 5). BRCA1 and BRCA2 are involved in several cellular functions important in carcinogenesis including DNA damage repair as well as cell cycle checkpoint (6). Therefore, it seems plausible that the cancer-promoting effects of estrogen would be even stronger in the absence of functioning BRCA1/2. Further, BRCA1 has been shown to repress estrogen-dependent and estrogen-independent transactivation activity of estrogen receptor-α (7-10).

Reproductive factors such as multiparity, early age at first full-term pregnancy, breast-feeding, and late age at menarche have been consistently found to protect against breast cancer (11, 12). Parity and early age at first birth predominantly protect against estrogen receptor-positive and progesterone receptor-positive tumors, whereas breast-feeding and possibly late age at menarche may protect against both receptor-positive and receptor-negative disease (13, 14). Thus, it is possible that these reproductive factors act through different hormonal mechanisms, some of which may involve estrogen, progesterone, or sex hormone binding globulin (15, 16). Oral contraceptive use has also been associated with a slightly increased risk of breast cancer (17), although the effect is modest, possibly restricted to current use, and not observed in all studies.

Although the overall effects of these reproductive and hormonal risk factors are well established, the extent to which these risk factors contribute to breast cancer risk in women with BRCA1 and BRCA2 mutations is less clear. Studies of the role of reproductive factors on breast cancer in BRCA1/2 mutation carriers have generated somewhat mixed results. Although the protective effect of multiparity seems quite consistent in studies of BRCA1 mutation carriers, the results from studies of BRCA2 mutation carriers are not clear (18-26). The results on breast-feeding and age at menarche have been mixed for BRCA1 mutation carriers, whereas there was no statistically significant association for BRCA2 in any of the previous studies (21, 23, 27-29). Studies of the role of oral contraceptives, particularly the low-dose oral contraceptive preparations, on breast cancer risk among BRCA1/2 mutation carriers have also generated mixed results (30-34).

Most previous studies of BRCA1/2 mutation carriers were based on studies of women from high-risk families, with multiple relatives with breast and/or ovarian cancer, many whom were recruited from genetic counseling centers. The extent to which such recruitment schemes influenced the results of these studies is not known. In this report, we describe results from a population-based study in Los Angeles County where we investigated the effect of hormonal and reproductive lifestyle factors and oral contraceptives on breast cancer risk in BRCA1/2 mutation carriers as well as in noncarriers.

Data collection methods used in conducting the Women's Learning the Influence of Family and Environment Study have been described previously (35). In brief, female patients diagnosed with histologically confirmed first primary invasive breast cancer were identified through the Los Angeles County Cancer Surveillance Program, a population-based registry sponsored by the National Cancer Institute Surveillance, Epidemiology and End Results program. Eligible cases were (a) U.S.-born and English-speaking, (b) White (including Hispanic) or African American, (c) ages 20 to 49 years at the time of diagnosis, (d) with no history of breast cancer before the index diagnosis, and (e) residents of Los Angeles County at diagnosis. A total of 2,882 eligible cases were identified (2,534 Whites and 348 African Americans) between February 1998 and May 2003. Because the recruitment of African Americans began after the initiation of the study, eligible African American cases were diagnosed from January 2000 to May 2003. Among the 2,882 potentially eligible cases, 1,794 (62%) were interviewed (1,585 White and 209 African American). Reasons for nonparticipation were patient refusal (n = 428), no longer living in Los Angeles County (n = 37), not located (n = 88), deceased (n = 38), serious illness or disability (n = 18), physician refusal (n = 50), or inability to schedule an interview within 18 months of diagnosis (n = 429).

Because this study was initially designed as a case-case study to investigate gene-environment interactions for breast cancer risk, control subjects were recruited for a subset of case patients diagnosed between July 2000 and March 2003. The control subjects were matched to the case patients on race and age (within 5 years and ages 20-49 years). The eligibility criteria for control subjects were the same as those of case patients, except that they had never been diagnosed with invasive or in situ breast cancer. We identified the control subjects using a neighborhood walk algorithm that has been reported previously (35). We identified 603 eligible control subjects for the 1,108 case patients (1,018 White and 90 African American) during the recruitment period. Among them, we interviewed 444 (74%) subjects (409 White and 35 African American). Reasons for nonparticipation were subject refusal (n = 77), no longer living in Los Angeles County (n = 18), deceased (n = 1), serious illness (n = 2), or inability to schedule the interview within 18 months of initial household contact (n = 61). On average, we canvassed 20 houses to find an eligible control subject who agreed to be interviewed.

The in-person interview in the Women's Learning the Influence of Family and Environment Study used a structured questionnaire, which was modified from that used in a previous case-control study, Women's Contraceptive and Reproductive Experiences Study (36), and asked for detailed information on demographic and environmental factors such as oral contraceptive use, reproductive history, and breast-feeding. We also obtained data on family history and ethnic origin. We obtained information up to the date of diagnosis of the case patient's invasive breast cancer or the date of initial household contact of the control subjects. Blood specimens were successfully collected from 1,519 (85%) cases around the time of the interview and brought to the Norris Cancer Center Genetics Core Laboratory in Styrofoam containers on frozen ice packs. For the first 50 samples, buffy coat was immediately extracted and stored, and for the remaining samples, we stored whole blood. The study was approved by the University of Southern California Institutional Review Board. All participants provided written informed consent. The consent for the cases explained that we would perform BRCA testing after anonymizing the samples.

We sequenced BRCA1/2 genes for 1,469 of 1,519 blood specimens. We were unable to sequence the remaining 50 specimens due to insufficient DNA. The detailed procedures of BRCA1/2 sequencing were reported previously (37). Briefly, we sequenced all BRCA1 and BRCA2 exons (except BRCA1 exons 1 and 4 and BRCA2 exon 1) as well as all exon-intron boundaries. Exon 1 was not sequenced for either gene because it is located upstream of the translation start site in both genes. BRCA1 exon 4 was not sequenced because it is not found in the normal BRCA1 mRNA transcript. DNA extraction, amplification, and sequencing were done in the University of Southern California Genomics Core Laboratory using a protocol similar to that described previously (38). Thirty-three randomly selected, blinded samples were resequenced for quality-control purposes. The discordance rate was 0.19%: 16 discordant sequencing results of the total 8,646 variant sites sequenced (262 variant sites for each of the 33 samples). We considered women with a truncating mutation in the BRCA1/2 genes as the mutation carriers. Women with a missense mutation that have been reported previously to affect the protein function (BRCA, M1775R and A1708E) were also considered as mutation carriers (39-41).

We classified women based on their family history of breast or ovarian cancer as follows: first-degree (mother or full sisters), second-degree (mother's or father's full sisters; grandmothers), none, or unknown. We considered women who had unknown second-degree family history but had no first-degree family history, as having no family history. Women who started their oral contraceptive use in or after 1975 were assumed to have used low-dose oral contraceptives only (31-33); therefore, we will refer to the use of oral contraceptives during or after 1975 as use of low-dose oral contraceptives in this report.

When there was indication that the results for the BRCA1 and BRCA2 were dissimilar, we describe the results from separate analyses. Otherwise, we present the results with BRCA1/2 combined. We compared demographic and hormone-related factors depending on the breast cancer status and BRCA1/2 mutation status of the breast cancer patients using ANOVA tests for continuous variables and Pearson's χ² tests for categorical variables. When the overall P value was statistically significant, we did pairwise comparisons with Bonferroni adjustment. We did case-control analyses comparing the BRCA1/2 mutation carrier and noncarrier cases with all control women to investigate the association between the breast cancer risk and the various hormonal factors. In our study, only a subset of cases had matched controls. We have shown previously in this study that the use of unconditional logistic regression with adjustment for the matching factors (race, age, and education) generated essentially identical results to that of unconditional logistic regression stratified by the matching factors and that of conditional logistic regression when limited to the matched pairs (35). Therefore, we broke the matched case-control pairs and analyzed all cases and controls using unconditional logistic regression adjusted for matching factors. Trend tests were based on likelihood ratio tests. We also did polytomous logistic regression to compare BRCA1/2 mutation carrier and noncarrier cases to the control women at the same time. Because the results were essentially the same, we present results from two separate unconditional logistic regression analyses. To examine whether mutation status modified the effect of each hormonal factor, we did case-case analyses comparing the BRCA1/2 mutation carrier cases with noncarrier cases on these hormonal factors.

All models were adjusted for age at reference date (<35, 35 to <40, 40 to <45, ≥45 years), education (high school or less, technical school or some college, college graduate), family history of breast or ovarian cancer (none, first-degree, second-degree, unknown), race (White, African American), self-identified Ashkenazi Jewish origin (yes, no), and number of full-term pregnancies (never pregnant, 1-2, ≥3, non-full-term pregnancy). When examining the effect of breast-feeding or age at last full-term pregnancy, we also adjusted for age at first full-term pregnancy. All P values are two-sided. All analyses were done using Stata 8.2 (Stata) and SAS 9.1 (SAS Institute).

The breast cancer case patients with BRCA1/2 mutations were slightly younger, more likely to have a first-degree family history of breast or ovarian cancer, and be of Ashkenazi Jewish origin compared with case patients without such mutations or to control subjects (Table 1). Control subjects had more education, had menarche at a later age, were more likely to have breastfed, and breastfed for a longer duration than the noncarrier cases, but there were no statistically significant differences in these factors between the two case groups or between the control subjects and the carrier cases. More control subjects reported having used oral contraceptives than mutation carrier cases.

An increasing number of full-term pregnancies was associated with a decreased risk of breast cancer regardless of BRCA1/2 mutation carrier status (Table 2), although the trends did not reach statistical significance. For both carriers and noncarriers, women who had four or more full-term pregnancies had about half the breast cancer risk of nulliparous women. The protective effect of full-term pregnancies was mainly limited to women who had their first full-term pregnancy before age 25 years, although the numbers became sparse, particularly among the mutation carriers (data not shown).

Later age at menarche was associated with a decreased risk of breast cancer in the BRCA1/2 mutation noncarriers, but no such protective association was observed in the carriers (Table 2). However, when the analyses were separated according to whether the mutation was in BRCA1 or BRCA2, later age at menarche among the BRCA1 mutation carriers was associated with lower risk (P_trend = 0.011), and the odds ratio (OR) for menarche at age ≥14 years (versus age ≤11 years) was 0.31 [95% confidence interval (95% CI), 0.11-0.87; data not shown]. No such protective effect of later age at menarche was seen for BRCA2 mutation carriers, but the sample size was small.

Among parous women, longer breast-feeding was associated with a decreased breast cancer risk in the BRCA1/2 mutation noncarriers (P_trend = 0.002; Table 2). In contrast, among the BRCA1/2 mutation carriers, risk did not decline with increasing duration of breast-feeding. This apparent effect modification by BRCA1/2 mutation status was not statistically significant (P value from case-case comparison = 0.23).

Any use of oral contraceptives, duration of oral contraceptive use, and the time since last oral contraceptive use were not associated with breast cancer risk in either BRCA1/2 mutation carriers or noncarriers (Table 3). Results for analyses of high-dose versus low-dose oral contraceptives suggest a protective association between the use of low-dose oral contraceptives and breast cancer risk; however, among the BRCA1/2 mutation carriers, this association did not reach statistical significance (Table 3). When examining the risk among BRCA1 mutation carriers and BRCA2 mutation carriers separately, the OR (95% CI) for low-dose oral contraceptive use was 0.55 (0.22-1.39) among the BRCA1 mutation carriers and that of the BRCA2 mutation carriers was 0.94 (0.28-3.14; data not shown). Use of high-dose oral contraceptives was statistically nonsignificantly associated with increased breast cancer risk in both carriers and noncarriers among younger women (age <45 years; Table 4). However, among BRCA1/2 mutation carriers ages <45 years, only 5 case patients had used high-dose oral contraceptives, and the test for interaction between the high-dose oral contraceptive and age was not statistically significant.

Because of the concern that oral contraceptive use at an early age might increase breast cancer risk, we assessed the effects of duration of oral contraceptive use before and after age 30 years. This was done by fitting a model with variables for duration of oral contraceptive use before and after age 30 years. We found no indication that oral contraceptive use before age 30 years increases breast cancer risk in the BRCA1/2 mutation carriers and noncarriers (data not shown). Compared with never users, the OR (95% CI) for using oral contraceptives for ≥5 years before age 30 years was 0.80 (0.57-1.14) among the BRCA1/2 mutation noncarriers and 0.84 (0.40-1.78) among the BRCA1/2 mutation carriers.

In this population-based study of breast cancer in young women, we found that an increasing number of full-term pregnancies was associated with a decreased breast cancer risk regardless of BRCA1/2 status and that breast-feeding was protective in the BRCA1/2 mutation noncarriers but not clearly so in the carriers. Oral contraceptive use overall, particularly the low-dose oral contraceptive use, was not associated with increased risk of breast cancer regardless of BRCA1/2 status.

Individuals with a BRCA1/2 mutation are more likely to develop cancers of the breast and ovary and possibly of the colon and prostate (1). Because hormones are considered to play a role in the etiology of these cancers, it seems likely that BRCA1/2 may be important regulators of growth and differentiation in hormonally responsive epithelial cells. Consistently, it has been shown that expression of BRCA1/2 is elevated in highly proliferating and differentiating cells at puberty and pregnancy (42, 43). Estrogen and progesterone stimulate breast cell proliferation (4). Considering that BRCA1/2 is involved in DNA repair and that unrepaired DNA damage in proliferating cells can be tumorigenic, the effect of estrogen and progesterone on breast cancer risk might be even stronger among BRCA1/2 mutation carriers than among the general population. Our results suggest, however, that this may not be the case and that at least parity and oral contraceptive use have similar effects on breast cancer in BRCA1/2 mutation carriers and noncarriers.

Our observation that an increasing number of full-term pregnancies was protective in the BRCA1/2 mutation carriers, although not statistically significant, is consistent with the results from most studies of BRCA1/2 mutation carriers (18, 19, 21, 25, 26), except a few studies where such protective effects were not found among BRCA2 mutation carriers (20, 23). Several mechanisms have been proposed to explain the protective effect of pregnancy in the general population: decreased levels of estrogen (15) and progesterone (16), increased levels of sex hormone-binding globulin (15), and pregnancy-induced differentiation of breast tissue (44). Our observation suggests that these protective mechanisms of pregnancy may also work in BRCA1/2 mutation carriers.

We observed that increased duration of breast-feeding is associated with a decreased breast cancer risk in the noncarriers but not among the BRCA1/2 mutation carriers. This observation is consistent with the results from a large collaborative study, showing that breast-feeding was protective against breast cancer in the general population but to a lesser degree among those with a first-degree family history (12). Previous studies of breast-feeding in BRCA2 mutation carriers are generally consistent with our findings (21, 27), although an Icelandic study found a suggestive but nonsignificant protective effect (23). In this latter study, breast-feeding was not protective in BRCA2 mutation noncarriers. The reasons for the difference between this Icelandic study and the other studies are not clear. The two studies of breast-feeding in BRCA1 mutation carriers yielded contradictory results, with one study observing a protective effect (27), whereas the other did not (21).

Several mechanisms underlying the role of breast-feeding in breast cancer risk have been proposed, including postponed resumption of ovulatory menstrual cycles, breast tissue differentiation (45), and decreased estrogen levels (46). Other proposed mechanisms include excretion of carcinogens from the breast ductal tissue (47). Whatever the true mechanisms are, our data support most studies indicating that breast-feeding is not protective against breast cancer in BRCA2 mutation carriers and possibly not in BRCA1 mutation carriers. It is possible that the protective effect of breast-feeding works through different hormonal (and/or nonhormonal) mechanisms than from that of pregnancy (13, 14), such that this protective mechanism is hampered in BRCA1/2 mutation carriers. It has been reported that BRCA1 mutation carriers are more likely to experience poor milk production and stop breast-feeding (48). It is not clear whether BRCA2 mutation carriers have similar problems or whether this modifies the protective effect of breast-feeding on the breast tissue.

Our observation that late age at menarche is protective in BRCA1 mutation carriers but not in BRCA2 mutation carriers is consistent with data from one familial study (28). In contrast, the International BRCA1/2 Carrier Cohort Study did not find a statistically protective effect in either BRCA1 or BRCA2 mutation carriers (29). However, case patients in the latter study were older than control women (8-year difference in mean age). Further, the duration between diagnosis and interview was long (average, ∼10 years) in the latter study (21, 29), raising concerns of survival bias.

Although a modest increase in breast cancer risk among current oral contraceptive users has been reported in a large pooled analysis and a meta-analysis (17, 49), not all studies have observed this increased risk. In the Women's Contraceptive and Reproductive Experiences Study and a previous Los Angeles case-control study of women ages ≤40 years, breast cancer risk was not associated with oral contraceptive use overall (36, 50). Consistent with this, among the noncarrier women in our current study, we observed that oral contraceptive use did not increase risk for breast cancer and even decreased risk of breast cancer among users of low-dose oral contraceptives.

In the earlier study of women ages ≤40 years, we identified 50 BRCA1/2 mutation carriers among Ashkenazi Jewish women with results suggesting that breast cancer risk was increased for mutation carriers who had used oral contraceptives (30). This was not observed in the current, much larger study. The results from the earlier, smaller study may have been due to chance, to the fact that the participants in that study were very young or to the fact that they were diagnosed in the 1980s, meaning that they may have had greater use of high-dose oral contraceptives and that they may have been relatively recent oral contraceptive users. Our current results among the young (ages <45 years) women suggesting that high-dose oral contraceptives may be associated with increased risk of breast cancer support this latter possibility, although it was not statistically significant (Table 4).

The results from our current study of no increased risk associated with overall oral contraceptive use in BRCA1/2 mutation carriers are consistent with a recent population-based study of BRCA1/2 mutation carriers (32). Further, our observations that low-dose oral contraceptive use was associated with a seemingly decreased risk of breast cancer in BRCA1/2 mutation noncarriers (statistically significant) and BRCA1 mutation carriers (statistically nonsignificant), but not in BRCA2 mutation carriers, are also consistent with the results from the same population-based study (32). In contrast, a study of BRCA1/2 mutation carrier cases and controls from breast cancer family registries has reported that the low-dose oral contraceptive use was associated with an increased risk of breast cancer in BRCA2 mutation carriers but not among BRCA1 mutation carriers (33). A third study of BRCA1/2 mutation carrier cases and controls who opted for genetic testing found a positive association between low-dose oral contraceptive use and breast cancer risk (34). However, in the latter studies based on family registries or high-risk women (33, 34), the mean interval between diagnosis and interview was rather long. Oral contraceptive use has been associated with localized breast cancer (17). Therefore, although the authors in the latter studies did restricted analyses of women diagnosed within 5 years observing the same association, the concerns for survival bias cannot be completely ruled out. Further, many of the BRCA1/2 mutation carrier controls in the Haile et al. study were recruited because their relatives (the probands) were cancer cases who had a mutation in BRCA1/2 (33). Therefore, the motivation to recall oral contraceptive use may have been higher for the probands, which could explain the positive association in the latter studies. Additional larger population-based studies would be useful to confirm whether use of low-dose oral contraceptives is truly not associated with increased breast cancer risk in BRCA1/2 mutation carriers.

Some differences between BRCA1- and BRCA2-related tumors exist; these include that BRCA1-related tumors are more likely to be estrogen receptor/progesterone receptor-negative than sporadic tumors, whereas BRCA2-related tumors are not (51). However, it is not clear how this could explain the different roles of low-dose oral contraceptive use between the BRCA1 mutation carriers and the BRCA2 mutation carriers. Due to the limited sample size of estrogen receptor/progesterone receptor-positive BRCA1 mutation carriers, it was not possible to isolate these aspects in this study.

Oral contraceptives have been found to lower ovarian cancer risk in BRCA1/2 mutation carriers as well as in noncarriers (52-54). However, a concern has been whether oral contraceptive use increases breast cancer risk. If low-dose oral contraceptives are safe for the breast in mutation carriers as suggested by our results, this finding would be clinically very important. Given the high risk of ovarian cancer in mutation carriers, oral contraceptive use could perhaps be recommended to BRCA1/2 mutation carriers as a preventive measure.

The strengths of our study include our sampling of a large number of population-based breast cancer patients, sequencing of whole BRCA1/2 genes in cases, and extensive collection of data on main exposure variables as well as important confounding factors during an in-person interview, which used the creation of life-event calendars during the interview to assist recall. Further, our study is less likely to be prone to survival bias because our case subjects were interviewed within 18 months of diagnosis. A weakness of our study is the relatively small number of BRCA1/2 mutation carriers, which may have been responsible for many of our statistically nonsignificant results. Further, given that our study participants were ages <50 years, our results may not be generalizable to postmenopausal women.

We compared BRCA1/2 mutation carrier cases and noncarrier cases with the same population-based controls. Given that the BRCA1 or BRCA2 mutation prevalence in the general population has been estimated as 0.45% (55), few control subjects are expected to be BRCA1/2 mutation carriers, suggesting that our OR estimates in noncarriers are reasonable. We used a case-case approach to examine the potential interaction between BRCA1/2 status and these reproductive factors. Case-case studies have been shown to be good alternatives to examine gene-environment interactions when the gene variants are rare but need an assumption of independence between genotype and environmental exposure (56). This means that, in our study, we assume mutation carriers were not more or less likely to have given birth or to have used oral contraceptives than noncarriers. We believe this assumption held in our study, especially because we adjusted for family history of breast or ovarian cancer.

The protective effect of parity was similar in BRCA1/2 mutation carriers as in noncarriers, whereas the effect of breast-feeding was not. Use of oral contraceptives that were presumed to be low dose was not associated with increased breast cancer risk regardless of BRCA1/2 mutation status. However, due to the relatively small number of participants, especially the BRCA1/2 mutation carriers, many of our results did not reach statistical significance. Further confirmation of the role of low-dose oral contraceptives is important from a public health perspective given the high prevalence of oral contraceptive use in the United States.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The authors are deeply grateful to the interviewers of this study and to Ms Juliana Bamrick for managing the data collection.