Dairy Consumption in Adolescence and Early Adulthood and Risk of Breast Cancer

Breast cancer remains the most common type of cancer in women. During 2016, 246,660 new cases of invasive breast cancer were estimated, with 40,450 breast cancer deaths in the U.S. women (1). Among numerous dietary components that have been related to breast cancer, dairy products have been hypothesized to play a role by increasing hormone levels such as estrogen and insulin-like growth factors (2–5). However, epidemiologic studies assessing dairy intake and breast cancer risk have been inconsistent (6–23), reporting no (6–15), higher (16, 17), or even lower (18–23) breast cancer risk. Moreover, the findings vary across dairy products (6–10, 12–16, 18, 20–23). A meta-analysis of prospective cohort studies found an inverse association between total dairy intake and breast cancer risk (24). Notably, almost all of these studies used dietary assessments during mid-life and later, with little focus on diet earlier in life (9, 12). In early life, carcinogenic exposure may be critical for subsequent breast cancer risk. For example, the atomic bombings of Hiroshima and Nagasaki and radiotherapy for Hodgkin lymphoma demonstrated evidence for early life exposure and subsequent risk of breast cancer. In both cases, exposure to radiation in childhood and early adult life was associated with risk of breast cancer, whereas exposure after age 30 had little effect when observed among women exposed to ionizing radiation (25–27). Moreover, our studies using data from the Nurses' Health Study II (NHSII) indicated that a high red meat intake during adolescence and early adulthood was more strongly associated with breast cancer risk than in later life (28, 29). We also found a stronger inverse association with adolescent fruit intake as well as adolescent and early adulthood fiber intake with breast cancer risk, compared with consumption later in life (30, 31). In a previous analysis of the NHSII (12, 17), whereas dairy intake during adolescence was not significantly associated with subsequent breast cancer risk (12), premenopausal high-fat dairy intake was associated with higher risk of breast cancer (17). It is, however, not clear whether this increased risk was due to dietary assessment at early age or the relatively young age of women at diagnosis of breast cancer.

Dairy products are a diverse food group that consists of several nutrients potentially influencing breast cancer risk. Calcium and vitamin D components in dairy products may have anticarcinogenic activities (32). However, in most prospective studies, the association between intake of calcium or vitamin D and risk of breast cancer is inconsistent (19, 22, 33–38), lacking data from early adult life. Other components of dairy products such as lactose might also be responsible for the associations between dairy and breast cancer risk. Although a few studies have suggested that lactose intake is associated with an increased risk of ovarian cancer (39, 40), little is known about the relationship with breast cancer (22, 41, 42). To date, the relationship between early life lactose intake and breast cancer has yet to be investigated.

Furthermore, breast cancer is a heterogeneous disease and breast tumors vary by estrogen and progesterone receptor status. Therefore, the relation between dairy foods and breast cancer may differ by hormone receptor status (14, 19, 23).

Therefore, based on our previous analyses (12, 17), but with longer follow-up and a larger number of cases, the associations between consumption of adolescent and early adulthood dairy products and breast cancer were examined. We also evaluated the association between vitamin D, calcium, and lactose intake with breast cancer incidence. The associations between dairy consumption and breast cancer were also examined by menopausal and hormone receptor status.

The NHSII was established in 1989 with a total enrollment of 116,429 female registered nurses ages 25 to 42 years. From baseline enrollment, the participants have been followed up every two years to update lifestyle and medical information. In 1991, women reported dietary intake through a semi-quantitative food-frequency questionnaire (FFQ). Among the 97,813 women who returned the FFQ, we included the participants who were premenopausal in 1991 and had total energy intake between 600 and 3500 kcal/day. We excluded the participants who had missing information on age, or had left more than 70 food items blank, or had left all items on dairy foods blank, or had prevalent cancer (except nonmelanoma skin cancer) in 1991 or before, leaving a total population of 90,503 women for early adulthood diet analysis. The follow-up rate was 96% of total potential person-years from 1991 through 2013.

In 1997, women were asked if they would complete a supplemental FFQ about diet during high school (HS-FFQ; 13–18 years). Among the 47,355 women who returned the HS-FFQ in 1998, women were excluded if they were diagnosed with cancer (except nonmelanoma skin cancer) before 1998, or had total energy intake under 600 or over 5,000 kcal/day, or left more than 70 food items blank, or left all items on dairy foods blank, leaving a total population of 44,264 women for adolescent diet analysis. Among women who provided data on adult diet, there were minimal differences in baseline demographic characteristics and breast cancer rate between those who completed the HS-FFQ compared with those who did not (12). The follow-up rate was 98% of total potential person-years from 1998 through 2013.

This analysis was approved by the Human Subjects Committee at Brigham and Women's Hospital and the Harvard T.H. Chan School of Public Health (Boston, MA). All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study.

Usual dietary habits from the past year were evaluated in 1991 and every four years thereafter by validated semi-quantitative FFQs with approximately 130 food items (available at http://www.nurseshealthstudy.org/participants/questionnaires). Women reported the frequency of dairy consumption in nine possible responses ranging from “never or less than once per month” to “6 or more times per day” (Supplementary Table S1). The validity of the adult dietary questionnaire used in the NHSII has been extensively documented (43).

Administered in 1998, the adolescent diet was ascertained from the HS-FFQ with 124 food items. Women reported the foods typically consumed between 1960 and 1980 when participants attended high school (Supplementary Table S2). For validity, of the 47,355 women who completed the HS-FFQ, the mothers of 272 of these women also completed the HS-FFQ for diet comparison. The mean Pearson correlation for nutrients was 0.40 (range, 0.13–0.59; ref. 44). The validity was also evaluated by comparing the responses of 80 women in high school from another study, using three, 24-hour recalls and two HS-FFQs with the same HS-FFQ administered 10 years later. The mean of corrected correlation coefficients for energy-adjusted nutrient intake was 0.58 (range, 0.40–0.88) calculated using the average of two HS-FFQs and the HS-FFQ from 10 years later, and 0.45 (range, 0.16–0.68) calculated using the average of three, 24-hour recalls and the HS-FFQ (10 years later; ref. 45).

Nutrient values in foods were obtained from the United States Department of Agriculture, food manufacturers, and independent academic sources (46–48). The food composition database was updated every 4 years to account for changes in the food supply. The intake of calcium, vitamin D, and lactose was energy-adjusted using the residuals from the regression of nutrient intake on total energy intake (49).

In biennial questionnaires, participants reported the diagnosis of breast cancer and the date of diagnosis. When a case of breast cancer was identified, we obtained the medical records and pathology reports to confirm the breast cancer diagnosis. Because of the high accuracy of self-reporting (99%), diagnoses with unavailable medical records (n = 384) were included in the analysis. Estrogen and progesterone receptor status of tumors was extracted from medical records. Deaths were identified by family members, the postal service, or the National Death Index.

At baseline enrollment and every 2 years thereafter, we inquired about potential risk factors for breast cancer, including age, weight, smoking, history of benign breast disease, family history of breast cancer, menopausal status, menopausal hormone use, and oral contraceptive use. Information on race, age at menarche, weight at age 18, adult height, and adolescent alcohol consumption was obtained from the 1989 questionnaire. Women were considered postmenopausal if they reported natural menopause or had undergone bilateral oophorectomy. We defined women of unknown menopausal status or who had hysterectomy without bilateral oophorectomy as premenopausal if they were under 46 years and smokers, or under 48 years and nonsmokers; and as postmenopausal if they were 54 years or older and smokers, or 56 years or older and nonsmokers (50).

Dairy intake reported in the baseline FFQ (1991) was considered as early adulthood dietary intake (27–44 years). To evaluate early adulthood diet and breast cancer, we computed person-years of follow-up for each participant from the date of return of the 1991 questionnaire until the date of any cancer diagnosis except nonmelanoma skin cancer, death from any cause, or the end of follow-up (June 1, 2013), whichever came first. For adolescent dairy consumption, similarly, person-years were calculated except that follow-up began with return of the adolescent diet questionnaire in 1998. Cox proportional hazards regression was used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for breast cancer associated with dairy product, calcium, vitamin D, and lactose consumption. Participants were divided into quintiles according to food groups. Adolescent and adult intake of ice cream and adult intake of yogurt and cream were divided into tertiles. As teenagers, 74% of the women did not consume yogurt, and less than 2% consumed five or more servings/week. The adolescent yogurt intake was divided into two categories: women who reported “never or <1 serving/month” or “1 serving/month or more.” We examined quintiles of dairy consumption to allow for nonlinearity and modeled the median value in each quintile as a continuous term to test for linear trend. For yogurt, ice cream, and cream, we used the median value in each tertile as a continuous term to test for linear trend. All regression models were stratified by age in months and calendar year of the current questionnaire cycle. Multivariable models were also simultaneously adjusted for history of breast cancer in mother or sisters, history of benign breast disease, smoking, height, body mass index (BMI) at age 18, weight change since age 18, age at menarche, parity and age at first birth, oral contraceptive use, menopausal status, menopausal hormone use, age at menopause, physical activity, alcohol consumption, and energy intake. For adolescent dairy consumption and breast cancer risk, we further included adolescent intake of alcohol and energy, rather than early adulthood energy intake in the multivariable models. To better represent long-term diet and minimize the effect of random measurement errors by using repeated dietary assessments, we calculated the premenopausal cumulative average of dietary intake from our repeated FFQs in 1991, 1995, 1999, 2003, 2007, and 2011, terminating the update of dietary information when a woman reached menopause. In addition, the average of adolescent and early adulthood (1991) dairy consumption was calculated among participants with information from these two periods.

To address whether the observed associations were independent of other dietary factors, we controlled for the Alternate Healthy Eating Index (AHEI; ref. 51), total red meat, fruit and vegetable, or animal fat intake. To determine whether associations with adolescent intake were independent of adulthood dietary intake, we adjusted for dietary intake during adult life (cumulative average of dietary intake). To address whether the associations between dairy consumption during adolescence or early adulthood and breast cancer risk were modified by age and BMI at age 18, a cross-product term of the ordinal score was included in the multivariable model. The tests for interactions were obtained from a likelihood ratio test. We used the Cox proportional cause-specific hazards regression model with a duplication method for competing risk data to calculate the HR of tumor subtypes, including both estrogen and progesterone receptors positive (ER⁺/PR⁻), both estrogen and progesterone receptors negative (ER⁻/PR⁻), and estrogen receptor positive and progesterone receptor negative (ER⁺/PR⁻; ref. 52). Because estrogen receptor negative and progesterone receptor positive (ER⁻/PR⁺) is not a reproducible breast cancer subtype (53), we did not evaluate the association between dairy food intake and this subtype of breast cancer. The association between adolescent dairy intake and adult height was evaluated using regression models adjusted for age, race, and adolescent consumption of total red meat, and total fruits and vegetables. All P values were two-sided. SAS version 9.3 (SAS Institute, Inc.) was used for all analyses.

Among women with early adulthood dietary intake, 3,191 cases of invasive breast cancer were identified from 1991 through 2013. Among women with adolescent dietary intake, we documented 1,318 cases of invasive breast cancer from 1998 through 2013. In early adulthood, higher total dairy consumption was associated with a lower prevalence of smoking, using oral contraceptives, and nulliparity, higher consumption of animal fat, red meat, fruits, and vegetables, and lower consumption of fiber (Table 1). Women with higher adolescent total dairy consumption were less likely to smoke, to be nulliparous, and to consume fiber, and more likely to consume animal fat, red meat, fruits, and vegetables (Supplementary Table S3). Also, women who consumed an average of 5.4 servings/day of total dairy products during adolescence were 1.7 cm taller than those who consumed an average of 1.0 serving/day (Supplementary Table S3). Moreover, in regression analysis, milk consumption during teenage years was significantly associated with attained height after adjusting for age, race, adolescent consumption of total red meat, and total fruits and vegetables; women were 0.41 cm taller on average for each additional serving/day of milk (P < 0.0001).

Adolescent total dairy consumption was not associated with overall breast cancer risk [HR (highest vs. lowest) = 1.08, 95% CI, 0.88–1.33; P_trend = 0.46] nor with premenopausal or postmenopausal breast cancer (Table 2). No significant associations were noticed between breast cancer risk and adolescent intake of low-fat/high-fat dairy products, milk, cheese, ice cream, or yogurt (Table 2; Supplementary Table S4). Results were similar after additional adjustment for adolescent red meat, fruit and vegetable, fiber, or animal fat intake or for adult total dairy intake.

Early adulthood total dairy intake was not associated with overall breast cancer risk [HR (highest vs. lowest) = 1.04, 95% CI, 0.92–1.18; P_trend = 0.73] nor with premenopausal or postmenopausal breast cancer (Table 3). No association was found between risk of breast cancer and consumption of low-fat or high-fat dairy, milk, cheese, yogurt, ice cream, or cream (Table 3; Supplementary Table S5). Results were similar after additionally adjusting for intake of fiber, total fruits and vegetables, total red meat, animal fat, or the AHEI score. The cumulative average of premenopausal total dairy intake was also not associated with risk of overall breast cancer [HR (highest vs. lowest) = 1.07, 95% CI, 0.94–1.21; P_trend = 0.74], premenopausal breast cancer [HR (highest vs. lowest) = 1.06, 95% CI, 0.89–1.26; P_trend = 0.66], or postmenopausal breast cancer [HR (highest vs. lowest) = 0.94, 95% CI, 0.76–1.16; P_trend = 0.26]. No significant associations were found for the cumulative average of premenopausal low-fat dairy, high-fat dairy, milk, cheese, or yogurt intake and breast cancer risk.

Adolescent and early adult (1991) total dairy intake was modestly correlated (r = 0.35). Among women with both early adulthood and adolescent dietary data (n = 41,092), adolescent and early adulthood dairy intake was averaged. We observed no significant association with risk of premenopausal or postmenopausal breast cancer. However, an inverse association was noted between average adolescent and early adulthood yogurt intake and overall (for ≥1 serving/month vs. never or <1 serving/month: HR = 0.88, 95% CI, 0.79–1.00) and premenopausal breast cancer risk (for ≥1 serving/month vs. never or <1 serving/month: HR = 0.82, 95% CI, 0.68–0.99). This inverse association did not reach a significant level for postmenopausal breast cancer (for ≥1 serving/month vs. never or <1 serving/month: HR = 0.91, 95% CI, 0.77–1.08).

We did not observe any significant association between breast cancer risk and calcium, vitamin D, or lactose intake during either adolescence or early adulthood (Supplementary Tables S6 and S7).

We had information on ER status for 82% (n = 2,617) of breast cancers and PR status for 82% (n = 2,604) of breast cancers. Associations between adolescent consumption of total dairy and high-fat dairy products and risk of breast cancer differed by ER/PR status (Table 4). High adolescent total dairy intake was associated with higher risk of ER⁻/PR⁻ cancer (each serving/day for total dairy product HR = 1.11, 95% CI, 1.00–1.24; P_{for difference by receptor status} = 0.04). During adolescence, high intake of high-fat dairy products was associated with increased risk of ER⁻/PR⁻ cancer (each serving/day HR = 1.17, 95% CI, 1.04–1.31; P_{for difference by receptor status} = 0.0004) and decreased risk of ER⁺/PR⁺ tumors (each serving/day HR = 0.91, 95% CI, 0.86–0.97). These associations were minimally changed after additional adjustment for animal or total fat intake. We did not observe significant heterogeneity between early adulthood dairy intake and tumor receptor status in either premenopausal or postmenopausal breast cancer (Table 4). To evaluate whether missing data on ER/PR status may affect the results, we examined the associations between dairy food intake and overall breast cancer incidence when only cases with ER/PR data were included. Compared with analyses including all cases of breast cancer, we observed similar results for both adolescent and early adulthood dairy food intake when the analyses included only cases with ER/PR data.

In addition, we examined whether the association between total dairy and high-fat dairy consumption during adolescence and early adulthood, and risk of breast cancer differed by BMI at age 18 (<21 or ≥ 21 kg/m²). No significant interaction was observed. Because we found a significant positive association between high-fat dairy and breast cancer in a previous analysis using NHSII data and we did not find this result in the current analysis, we examined whether the positive association was due to the younger age of participants at diagnosis, we looked at the association in two groups of women: younger than 45 years and 45 years or more (age updated during follow-up). We observed a significant positive association between early adulthood high-fat dairy consumption and breast cancer risk only among women younger than 45 years (each serving/day HR = 1.15 95% CI, 1.01–1.30).

Adolescent or early adulthood dairy intake was not associated with overall breast cancer risk. Although analysis of two time periods (adolescence and early adulthood) separately did not support the beneficial effects of yogurt for breast cancer prevention, the average adolescent and early adulthood yogurt consumption was associated with lower risk of breast cancer, specially before menopause. Moreover, higher dairy consumption, especially high-fat dairy during adolescence, was associated with a higher risk of ER⁻/PR⁻ and lower risk of ER⁺/PR⁺ breast cancer. Adolescent or early adulthood calcium, vitamin D, and lactose intake was not associated with incident breast cancer.

Prospective studies have been inconsistent concerning a role of dairy consumption on breast cancer incidence. Although a few studies reported higher (16, 17) or lower (18–23) breast cancer risk, most studies have not reported significant relationships between dairy consumption and breast cancer risk (6–15). Similar to the current study, in an earlier analysis of the NHSII, adolescent dairy intake was not associated with breast cancer risk (12). Although we did not observe a significant association between early adulthood high-fat dairy foods and risk of breast cancer overall, a significant positive association was noted between high-fat dairy and breast cancer among women younger than 45 years. Therefore, it is possible that the positive association observed in the Cho and colleagues (17) study was primarily due to the young age of participants (mean age at diagnosis = 43 years) rather than the early-life dietary evaluation.

Fewer studies evaluated whether the association between dairy products and breast cancer varied by hormone receptor status (14, 19, 23). While lower risk of ER⁺ breast cancer was observed with higher dairy intake among postmenopausal women in the Cancer Prevention Study II Nutrition Cohort (23), Lin and colleagues (19) and Genkinger and colleagues (14) did not observe any difference by ER/PR status. In our study, high adolescent intake of total and high-fat dairy products was associated with higher risk of ER⁻/PR⁻, but not ER⁺/PR⁺ breast cancer. The biological mechanisms are not clear for dairy products by hormone receptor status. Dairy products naturally contain endogenous estrogens and estrogen metabolites (54), which account for approximately 50% of the estrogens consumed (55). In one study, higher dairy consumption was associated with higher total and free estradiol concentrations among postmenopausal women (2). Because factors related to estrogen metabolism are more strongly related to ER⁺ tumors, our a priori hypothesis was that development of hormone receptor positive tumors might be more sensitive to dairy consumption. Further, insulin-like growth factors likely play an important role in the etiology of breast cancer (56–58), and dairy intake may promote neoplasms by raising insulin-like growth factor I concentrations (3–5, 59). Plasma levels of insulin-like growth factor I were mainly associated with higher ER⁺ tumor risk, but not ER⁻ tumors (60, 61). In addition, N-glycolylneuraminic acid found in dairy foods may play a role in breast tumors (62, 63); however, its role in relation to tumor subtypes is not clear. Therefore, the variation in associations by ER/PR status might have occurred by chance and needs confirmation.

Although prospective cohort studies have not shown any significant association between either yogurt or fermented milk intake and incident breast cancer (8, 14, 20, 22), a recently published meta-analysis with case–control and cohort studies (64) reported a 9% lower breast cancer risk with high yogurt intake. Our study suggested that the average of adolescent and early adulthood yogurt consumption might reduce risk of breast cancer, but this finding should be interpreted cautiously as an association with yogurt was not seen when adolescent and adult intakes were examined separately. Lactic acid bacteria that are found in yogurt may play a role in decreasing breast cancer risk, at least partially through immunomodulatory mechanisms (65).

Our study has several strengths. We were able to evaluate the role of dairy intake in specific life periods, including adolescence, early adulthood, and the cumulative average of the premenopausal time span. The large number of cases showed modest differences in risk and examined breast cancer by menopausal and hormone receptor status.

There are several limitations to our study. The participants were predominantly white nurses who do not represent a random sample of U.S. women; however, biological mechanisms underlying these associations might not differ by race or education. In previous studies, we observed that the associations for breast cancer and other diseases were very similar to those found in the other broadly based U.S. populations. Because women ages 33 to 52 years were asked to recall high school diet, the assessment of adolescent dietary intake might be imprecise. Nevertheless, the HS-FFQ showed reasonable validity (45). In addition, adolescent milk intake was positively associated with increased adult height in the current study, as observed prospectively in the Growing Up Today Study, supporting the validity of our measurement of adolescent milk intake (66). Finally, residual confounding is always of concern in observational studies. With the detailed NHSII questionnaires, we adjusted for all known breast cancer risk factors.

In summary, we found no association between dairy consumption during adolescence or early adulthood with overall breast cancer risk. However, our findings suggested that adolescent high dairy intake might be associated with higher ER⁻/PR⁻ and lower ER⁺/PR⁺ cancer risk. Intake of calcium, vitamin D, and lactose during adolescence or early adulthood was not associated with breast cancer incidence. Further studies are warranted to investigate these relationships, especially the associations between dairy intake and risk of breast cancers classified by hormone receptor status.

No potential conflicts of interest were disclosed.

Conception and design: M.S. Farvid, E. Cho, W.Y. Chen, W.C. Willett

Development of methodology: M.S. Farvid

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.H. Eliassen, E. Cho, W.C. Willett

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.S. Farvid, E. Cho, W.Y. Chen

Writing, review, and/or revision of the manuscript: M.S. Farvid, A.H. Eliassen, E. Cho, W.Y. Chen, W.C. Willett

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.S. Farvid, A.H. Eliassen

Study supervision: M.S. Farvid, A.H. Eliassen, W.C. Willett

The Nurses' Health Study II was supported by the NIH grants (R01 CA050385 and UM1 CA176726; W.C. Willett) and a grant from The Breast Cancer Research Foundation (W.C. Willett).

We would like to thank the participants and staff of the NHSII for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY.

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.