Reproductive Factors, Age at Maximum Height, and Risk of Three Histologic Types of Breast Cancer

It is well established that some reproductive factors, including age at menarche, age at first full-term pregnancy, number of live births, and breast-feeding, are related to a woman's risk of breast cancer (1-4). There is also evidence that age at maximum height, as an indicator of the timing of the pubertal growth peak, is associated with a woman's risk of breast cancer (5-9). One mechanism through which all of these maturation and reproductive exposures influence breast cancer risk is their effect on lifetime number of ovulatory cycles a woman experiences. The number of cycles a woman experiences influences her lifetime exposure to endogenous ovarian hormones, which is strongly related to breast cancer risk (2, 10). Additionally, both pregnancy and breast-feeding induce the differentiation of breast epithelial cells, making them less susceptible to carcinogenic insults (11-13).

Although numerous studies have evaluated the relationship between maturation and reproductive factors and breast cancer risk, few have assessed how they are related to different histologic types of breast cancer among postmenopausal women, the age group that comprises the majority of new breast cancer diagnoses. Although there is evidence that ductal and lobular tumors have molecular and pathologic differences as well as differences in clinical characteristics (14-21), few studies have examined how factors relating to maturation and reproduction may affect risk of these histologic types differently (6, 22-28). Comparing results across these studies is challenging because different studies focused on different reproductive factors and many were heterogeneous regarding menopausal status of the study groups and classification of histologic case groups. Consequently, in an effort to further elucidate the differences in risk factors for ductal, lobular, and ductal-lobular breast carcinomas, we evaluated the associations of several maturation and reproductive factors with different histologic types of breast cancer using data from a recently completed population-based case-control study of postmenopausal breast cancer that involved a centralized pathology review. We included a histopathologic review of all available tissue samples in this study to accurately and consistently categorize cases by histologic type of invasive breast carcinoma.

We used data from the Seattle Area Hormone and Reproductive Epidemiology Breast Cancer Study, which is a population-based case-control study of invasive ductal, invasive lobular, and invasive ductal-lobular breast carcinoma among women ages 55 to 74 years living in the Seattle-Puget Sound region (King, Pierce, and Snohomish counties). This study was approved by the Fred Hutchinson Cancer Research Center Institutional Review Board and its methods have been described in detail elsewhere (29).

Cases were identified through the Cancer Surveillance System, a population-based cancer registry that is part of the Surveillance, Epidemiology, and End Results program of the National Cancer Institute. All women ages 55 to 74 years who were diagnosed with a primary invasive breast cancer from January 1, 2000 to April 31, 2004, who did not have a previous history of in situ or invasive breast cancer, and who had a landline home telephone were eligible for consideration as cases for this study. All lobular and ductal-lobular cases that met the eligibility criteria were approached to be part of the study. Because ductal cancer is the most common type of breast cancer (accounting for 70-75% of all cases), only a random 25% of ductal cases who met the eligibility criteria were contacted to be part of the study. Ductal cases were frequency matched to the combined lobular and ductal-lobular group 1:1 by 5-year age groups. A total of 1,251 cases were identified, and 83% of them were interviewed (524 ductal, 324 lobular, and 196 ductal-lobular cases). The reasons for eligible cases not being interviewed included refusal by the case or unknown location of the case (14%), death before interview (2%), and the case's physician not allowing his/her patient to be contacted for study participation (1%).

This study involved a centralized review of pathology reports for all cases and a centralized review of tumor tissue blocks for those cases with available tissue. A two-tiered process for assigning histologic type was used. Tumor tissue specimens from 869 (83%) cases were ascertained from local hospitals and centrally reviewed by pathologists in the Porter Laboratory at the Fred Hutchinson Cancer Research Center. Ductal and lobular tumors were classified based on the recommendations of the WHO (30). Ductal-lobular cases included cases with two separate areas of invasive ductal and invasive lobular tumors in the same breast and cases with one tumor with both ductal and lobular features. One of two pathologists assessed the histology of a case based on the review of all slides and the diagnostic criteria. All slides then underwent a second histology review by another study pathologist and any discrepancies were decided by consensus review or sent for review by a consultant pathologist. The study had 61 cases with discrepancies between study pathologists or between the review of pathology reports and the categorization based on tumor tissue review. These cases were reviewed by a consultant pathologist to determine a final histology category. For the 869 cases with available tumor tissue, the final histology classifications were made based on the tumor tissue review. When comparing the histology classifications based on tumor tissue review with those based on the review of pathology reports, there was 94% agreement (κ statistic = 0.91).

For the 175 cases for whom we could not obtain tissue, the final histologic classification was based on a review of pathology reports conducted by trained abstractors. If multiple pathology reports for a case were present (e.g., a biopsy report and a lumpectomy report), then the histology from the tumor specimen with the largest invasive component was given priority for the final histology classification. Again, cases were classified as ductal-lobular only if there was documentation that the tumor contained both invasive ductal and invasive lobular components. Cases with only invasive ductal tumors or only invasive lobular tumors were classified as ductal and lobular, respectively, regardless of the presence of any histologic type of in situ carcinoma. Overall, 61 cases with both invasive ductal carcinoma and lobular carcinoma in situ were classified as ductal cases and 63 cases with both invasive lobular carcinoma and ductal carcinoma in situ were classified as lobular cases.

Population-based controls who were female residents of King, Pierce, and Snohomish counties were identified through random-digit dialing using the Mitosky-Waksberg method with a clustering factor of 5 (31). Using one-step recruitment, controls were frequency matched 1:1 to the combined lobular and ductal-lobular case group by age (5-year age groups) and reference year (31, 32). Among the total of 29,735 random telephone numbers that were called, 9,876 were confirmed residential numbers or assumed to be residential numbers, and 87% of these numbers completed the screening questions for study eligibility. A total of 660 women ages 55 to 74 years who had never been diagnosed with invasive or in situ breast cancer were identified as eligible controls and 469 (71%) were interviewed.

All cases and controls provided written informed consent and completed an ∼90 min in-person interview with the assistance of a trained interviewer. The interview collected demographic, anthropometric, reproductive, contraceptive, medical history, and lifestyle information for the period before the reference date. For cases, the reference date was the date of breast cancer diagnosis, and for controls, the reference date was assigned to reflect the distribution of reference dates among the cases.

With respect to our primary exposures of interest, age at menarche was grouped as ages 8 to 11, 12 to 13, and ≥14 years. Age at maximum height was grouped as ages ≤14, 15 to 16, and ≥17 years. The interval between age at menarche and first live birth was grouped as 1 to 8, 9 to 12, and ≥13 years based on the tertile distribution of this variable. Number of live births was classified as 1, 2, 3, and ≥4. Age first breast-fed was categorized as never, ≤19, 20 to 24, 25 to 29, and ≥30 years. Age last breast-fed was categorized as never, ≤24, 25 to 29, 30 to 34, and ≥35 years. Age at first live birth and age at last live birth used the same age groupings as age first breast-fed and age last breast-fed, respectively. Ever breast-feeding was defined as breast-feeding for at least 1 month. Total breast-feeding was defined as total lifetime duration of breast-feeding measured in months and was grouped into the following categories: never, <1, 1.0 to 5.9, 6.0 to 11.9, 12.0 to 23.9, and ≥24.0 months.

Polytomous logistic regression was used to compare ductal, lobular, and ductal-lobular cases with controls. Odds ratios (OR) and 95% confidence intervals (95% CI) were computed as estimates of the relative risk, and all variables were modeled categorically. Tests for nonlinearity were conducted for variables with greater than two categories using a grouped linear term. For those variables that did not significantly depart from linearity, a test for linear trend was conducted. The never or zero variable category was excluded from the trend tests. All tests were two-sided and P values < 0.05 were considered to be statistically significant. All analyses were adjusted for age (5-year age groups) and reference year. In addition, analyses of the relationships between ever breast-feeding and duration of breast-feeding and breast cancer risk were adjusted for number of live births because it was a priori considered to be a confounder of these relationships.

The variables listed in Table 1, with the exception of estrogen receptor (ER) and progesterone receptor (PR) status, were systematically assessed as potential confounders of the relationships between age at maximum height and reproductive characteristics and risk of ductal, lobular, and ductal-lobular breast carcinoma. In addition, the following variables were assessed as potential confounders: age at menopause (<47, 47-50, 51-53, ≥54 years), income (<$20,000, $20,000-34,000, $35,000-69,000, ≥$70,000), marital status (married/living with male or female partner, divorced/separated, widowed, single never married), average grams of alcohol consumed per week (0, <24.75, ≥24.75), and smoking (ever, never). The maturation and reproductive factors of interest were also assessed as potential confounders of each other. A covariate was considered to be a confounder if it consistently altered the ORs across the covariate's categories for each histologic type by at least 10%, if the direction of the effect of the covariate on the exposure of interest was plausible, and if the covariate was not another characteristic of the exposure of interest (e.g., age at first breast-fed with a category for those who never breast-fed and duration of breast-feeding with a never category). None of the variables listed above or the reproductive factors of interest were found to be confounders of the relationship between age at menarche, age at maximum height, parity, number of live births, ever breast-fed, or duration of breast-feeding and development of ductal, lobular, or ductal-lobular breast carcinoma. Age at first live birth and age first breast-fed were not adjusted for each other because the two variables are highly correlated. Similarly, age at last live birth and age last breast-fed were not adjusted for each other. Thus, all of our final analyses were adjusted for age and reference date, and analyses of ever breast-fed and duration of breast-feeding were additionally adjusted for number of live births.

For covariates hypothesized a priori as possible modifiers of effects of interest, an interaction term was formed and tested. If the interaction term was significant (P < 0.05) for all histologic types when added to a model with the variable of interest, the covariate, and the matching variables, then the covariate was considered an effect modifier. For age at menarche, the following variables were considered as potential effect modifiers: number of live births, age at first live birth, ever breast-fed, duration of breast-feeding, and age first breast-fed. For number of live births, the variables considered were age at first live birth, ever breast-fed, duration of breast-feeding, and age first breast-fed. The potential effect modifiers considered for age at first live birth, age at last live birth, age first breast-fed, and age last breast-fed were number of live births and each other. For ever breast-fed and duration of breast-feeding, we considered number of live births, age at first and last live birth, and age first and last breast-fed. We did not observe statistically significant effect modification based on likelihood ratio testing. All analyses were done using Stata 9.2 (StataCorp).

Analyses of age at menarche, age at maximum height, and parity were conducted among all study women. Analyses of the interval between age at menarche and age at first live birth, number of live births, age at first live birth, age at last live birth, ever breast-fed, duration of breast-feeding, age first breast-fed, and age last breast-fed were conducted among women who reported having one or more live births (n = 456 ductal, 277 lobular, and 165 ductal-lobular cases and 431 controls). Differences in risk estimates across histologic type were assessed using P values from unconditional logistic regression models that treated either ductal cases or lobular cases as the reference category and excluded controls. Specifically, a variable was created with ductal cases as the reference group and lobular and ductal-lobular cases as separate comparison groups as well as a variable with lobular cases as the reference group and ductal and ductal-lobular cases as comparison groups. These variables were then used in the unconditional logistic regression models from the main analyses to obtain P values for the difference in risk estimates between the case groups.

Ductal-lobular cases were somewhat younger and more likely to have a body mass index < 25.0 kg/m² compared with ductal cases, lobular cases, or controls (Table 1). Lobular and ductal-lobular cases were somewhat more likely to have graduated from college than ductal cases or controls. Regardless of histologic type, cases were more likely than controls to have a first-degree family history of breast cancer, to have experienced natural menopause, and to be current users of combined estrogen and progestin postmenopausal hormones. The majority of ductal, lobular, and ductal-lobular cases had tumors that were ER positive and PR positive. The ductal-lobular case group contained the greatest percentage of ER-positive and PR-positive tumors (82.6%), whereas the ductal case group contained a greater percentage of ER-negative and PR-negative tumors (14.0%) than either the lobular (3.2%) or the ductal-lobular (4.1%) case groups (Table 1). Because of small numbers, the results stratified by ER and PR status are not presented here.

Age at menarche was not related to risk of ductal or lobular carcinoma but was inversely associated with risk of ductal-lobular carcinoma (P_trend = 0.04; Table 2). We compared the risk estimates for each histologic group to determine if they were statistically different than the risk estimates for other histologic groups. We found that the risk estimate for menarche at ages 8 to 11 years compared with ages ≥14 years for ductal-lobular carcinoma was statistically different than the risk estimate for lobular carcinoma (P = 0.02) and that the difference between the risk estimates for ductal-lobular and ductal carcinoma approached a statistically significant difference (P = 0.05). Women who reached their maximum height at an earlier age had an increased risk of ductal-lobular carcinoma (P_trend = 0.04) compared with controls; however, the individual risk estimates for ductal-lobular carcinoma were not statistically different than the risk estimates for ductal or lobular carcinoma. We also observed a suggestion of an association between age when maximum height was attained and risk of ductal carcinoma compared with controls, but this association was within the limits of chance (P_trend = 0.10). The interval between age at menarche and first live birth was not related to risk of ductal, lobular, or ductal-lobular carcinoma (Table 3).

Compared with nulliparous women, parous women had a 50% reduction in risk of ductal, lobular, and ductal-lobular carcinoma (Table 2). Twenty percent to 40% reductions in risk of ductal, lobular, and ductal-lobular carcinoma were observed among women who had 2, 3, or ≥4 live births compared with women who had 1 live birth; however, most of these risk estimates were within the limits of chance (Table 3). Women who had their first live birth when they were ages ≥30 years had increased risks of ductal, lobular, and ductal-lobular carcinoma compared with women who had their first live birth at ages ≤19 years; however, this association was within the limits of chance for ductal and lobular carcinoma. An older age at last live birth was associated with an increased risk of ductal-lobular (P_trend = 0.01) but not with ductal or lobular carcinoma. When comparing the risk estimates for the histologic case groups, we found no statistically significant differences between the risk estimates for each histologic type for parity, number of live births, age at first live birth, and age at last live birth.

Compared with parous women who never breast-fed, those who ever breast-fed had a reduced risk of ductal carcinoma (OR, 0.7; 95% CI, 0.5-0.9; Table 4), but not of lobular or ductal-lobular carcinoma, after adjusting for number of live births. However, no statistically significant differences between the risk estimates for each histologic type were observed. The decrease in risk of ductal carcinoma associated with breast-feeding was limited to women who breast-fed for ≥1 month, although no trend was observed (P_trend = 0.43). The risk estimate for ductal carcinoma associated with breast-feeding for at least 24 months (OR, 0.6; 95% CI, 0.3-1.0) was significantly different (P < 0.001) than the risk estimate for ductal-lobular carcinoma (OR, 1.9; 95% CI, 1.0-3.6). Both the ages when women first breast-fed and last breast-fed were unrelated to risk of ductal carcinoma and lobular carcinoma. However, risk of ductal-lobular carcinoma increased with age last breast-fed (P_trend = 0.004; Table 4). Specifically, women who were ages ≥35 years when they last breast-fed had a 2.3-fold (95% CI, 1.2-4.4) increased risk of ductal-lobular carcinoma compared with those who never breast-fed and this risk estimate was significantly different (P = 0.01) than the risk estimate for ductal carcinoma (OR, 1.1; 95% CI, 0.6-1.9).

The results from this analysis should be considered within the context of the limitations of this study. Among the eligible cases, 83% were interviewed; among eligible controls, 71% were interviewed. If women who declined to be interviewed were different than those who participated with respect to the maturation and reproductive exposures of interest, then the results could be biased. However, an analysis of potential nonresponse bias from a previous case-control study of breast cancer found that, with the exception of height, there was little effect on the risk estimates when adding available data from nonrespondents (33). Exposure information was gathered via self-report and required the participants to remember some maturation and reproductive events that occurred many years earlier. In an effort to improve the accuracy of responses, a life calendar was used to chart important events in a woman's life. If there was inaccurate recall, it is unlikely that it would differ by histologic case group and we expect that the resultant bias would be nondifferential; thus, bias our risk estimates toward the null. Additionally, tumor tissue was not obtained for 17% of cases, so histologic classification of these cases was based only on a centralized review of pathology reports. However, this difference is unlikely to bias our results appreciably because there was 94% agreement between the histology classification made from the tissue review and from the pathology report review for the 869 cases for whom both sources of data were available.

This is the first study to include a histopathologic review of tissue specimens and a centralized review of pathology reports, which allowed us to classify ductal-lobular tumors as those containing both an invasive ductal component and an invasive lobular component. In contrast, when classification is based on International Classification of Diseases for Oncology (ICD-O) codes, a ductal-lobular tumor (ICD-O histologic code 8522) is only required to have one component, either ductal or lobular, which is invasive (34). For instance, in studies using ICD-O codes to classify tumors, a tumor with an invasive ductal and a lobular in situ component as well as a tumor with an invasive lobular and a ductal in situ component are both classified as ductal-lobular. Additionally, some prior studies that examined histologic type of breast cancer by ICD-O codes included ductal-lobular tumors in the same group as invasive lobular tumors (ICD-O histologic code 8520; refs. 6, 23, 24). Therefore, for both of the above reasons, our approach for evaluating ductal-lobular tumors was unique and provided an opportunity to assess associations that may have been obscured in other studies due to the heterogeneity of the tumors included in the ductal-lobular category.

Both age at menarche and age at maximum height are markers of puberty, a time during which the breast is undergoing rapid development, and our study suggests that these ages are more strongly associated with risk of ductal-lobular carcinoma than with risk of ductal or lobular carcinoma. Although few of the case-case comparisons yielded statistically significant differences, this could be due to the relatively small sample size in the ductal-lobular case group, which limited our power to detect a difference in ductal-lobular compared with ductal or lobular risk estimates. Although we found that age at menarche was inversely associated with risk of ductal-lobular carcinoma but not ductal or lobular carcinoma, other studies of primarily or only postmenopausal women have had conflicting results. Two studies found a decreased risk of lobular carcinoma and not ductal carcinoma associated with a later age at menarche (6, 23), another found a slightly stronger decreased risk associated with ductal rather than lobular carcinoma (24), and a fourth study found a similar decreased risk for both ductal carcinoma and lobular carcinoma (25). One of these studies excluded ductal-lobular tumors (25), whereas the others classified histology based on ICD-O codes but combined the ICD-O categories for invasive lobular and ductal-lobular tumors because, when assessed separately, there were no significant differences between the two categories (6, 23, 24). In the only other study to our knowledge that assessed age at maximum height as a risk factor for breast cancer by histologic type among older women, Li et al. found an inverse association with risk of ductal but not lobular carcinoma, whereas we found that age at maximum height was inversely associated with risk of ductal-lobular carcinoma but not associated with risk of ductal or lobular carcinoma (6). It is difficult to compare our ductal-lobular results relating to reproductive factors and age at maximum height to prior studies because previous studies combined the ICD-O categories for invasive lobular and ductal-lobular tumors (6, 23, 24) rather than separately categorizing ductal, lobular, and ductal-lobular tumors via a centralized tumor tissue review.

There is also conflicting evidence regarding differences in risk of breast cancer by histologic type according to other reproductive factors including parity, number of live births, and age at first live birth. Our results suggest that parous women have a similar reduction in breast cancer risk across ductal, lobular, and ductal-lobular groups compared with nulliparous women, which is consistent with one previous report (25) but conflicts with another study that did not find an association between parity and risk of ductal or lobular carcinoma (24). We observed a suggestion of a similar risk reduction across histologic groups among women having two or more live births compared with those having one live birth. Li et al. found similar decreased risks of ductal and lobular carcinoma for women having four or more full-term births compared with those having one full-term birth (24). However, another study of older women found that having two or more births compared with one birth was associated with a decreased risk of ductal carcinoma but not lobular carcinoma (25). The majority of studies examining age at first live birth among primarily postmenopausal women have found a slightly stronger positive association with lobular carcinoma than ductal carcinoma for increasing age at first live birth; however, the differences by histology were not statistically significant (6, 23, 25). Our study, along with one other among postmenopausal women (24), found similar increases in risk across histologic types associated with having a first live birth at ages ≥30 years compared with ages ≤19 years.

To our knowledge, only one other published study has examined the effect of breast-feeding on risk of different histologic types of breast cancer among postmenopausal women and it found a reduced risk of ductal carcinoma but not lobular carcinoma among women who breast-fed for at least 24 months compared with those who never breast-fed, after accounting for number of live births (24). We found that ever breast-feeding was associated with a decreased risk of ductal but not lobular or ductal-lobular carcinoma compared with never breast-feeding and that this reduction in risk was limited to women who breast-fed for at least 1 month. Although it is already established that breast-feeding reduces the risk of invasive breast cancer in the aggregate, our observation suggests that the cell differentiation induced by breast-feeding may primarily exert its protective effect on the development of invasive ductal tumors rather than tumors of other histologic types. When examining other aspects of breast-feeding, we found that an increasing age when last breast-fed is associated with an increased risk of ductal-lobular carcinoma. No other studies have assessed this association among postmenopausal women thus far, and this finding requires confirmation.

Our findings suggest that early hormonal exposures associated with breast development and the beginning of ovulation during puberty may be associated more strongly with ductal-lobular rather than with ductal or lobular carcinoma, whereas the cell differentiation induced by breast-feeding may be more strongly associated with protection against ductal rather than lobular or ductal-lobular carcinoma. Our results also suggest that ductal-lobular carcinoma has distinct risk factors differing from ductal and lobular carcinoma. However, these results require replication and should be interpreted with caution considering the relatively small number of ductal-lobular cases in some categories and that the mechanisms underlying these differences are unclear. To further explore the risk factors associated with ductal-lobular carcinoma and the potential differences in etiology compared with ductal and lobular carcinoma, future studies are needed that classify cases via a centralized histopathologic review of tissue specimens and can thereby identify risk factors for tumors with both an invasive ductal and an invasive lobular component.

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.