Abstract
Background: Dairy foods are complex mixtures which include nutrients and non-nutrient substances that could potentially influence cancer etiology, including breast cancer.
Objective: The purpose of this study was to examine associations between the types and quantity of dairy foods consumed and the risk of breast cancer among women participating in the Roswell Park Cancer Institute Data Bank and BioRepository (DBBR) between 2003 and 2014.
Methods: Archived clinical and questionnaire data were obtained from the DBBR from 1941 women diagnosed with breast cancer between December 2003 and October 2014, and 1237 control participants. Intakes of dairy foods were queried with a self-administered food-frequency questionnaire and grouped into monthly intakes of total dairy, milk, yogurt, low-fat cheese, other cheese, and sweet dairy. ORs and 95% CIs were estimated with unconditional logistic regression adjusting for age, race, body mass index, menopausal status, energy intake, type of milk usually consumed, cigarette smoking status, and family history of breast cancer.
Results: Total dairy intakes were associated with a non-significant 15% reduction in breast cancer risk (P = 0.11). Higher intakes of yogurt were associated with reduced risk of breast cancer (OR: 0.61; 95% CI: 0.46, 0.82) and higher intakes of American, cheddar, and cream cheeses were associated with a marginally significant increased risk (OR: 1.53; 95% CI: 0.99, 2.34; P = 0.05). Associations with dairy foods were mixed when stratified by estrogen receptor (ER) status, and in general reflected those of overall breast cancer. However, we observed positive associations between milk intake and risk of ER− breast cancer (OR: 1.58; 95% CI: 1.05, 2.37) and inverse associations between sweet dairy and ER+ breast cancer (OR: 0.52; 95% CI: 0.29, 0.95).
Conclusions: Specific dairy foods may contribute to breast cancer risk in women, although the risk varies by source of dairy. Future studies are warranted to confirm the protective potential of yogurt in this type of cancer.
Keywords: cancer, cancer epidemiology, dietary intake, dietary patterns, breast cancer, dairy, estrogen receptor
Introduction
Several lines of evidence suggest that foods that contain calcium and vitamin D, particularly dairy, may be important in the etiology of several cancers (1). Vitamin D, as measured by circulating concentrations of 25-hydroxyvitamin D, has been inversely associated with the risk of colorectal cancer and, to a lesser degree, with the risk of breast cancer (2), yet the evidence for associations with calcium intake has been inconsistent (1). Despite some supporting evidence from observational studies, supplemental low-dose vitamin D and calcium was not shown to be preventive against breast cancer incidence in the Women's Health Initiative (2, 3). Aside from studies of vitamin D and calcium as single nutrients, evidence for dairy products in association with breast cancer risk is also variable and dependent upon dose, dairy food form, and period of consumption (youth or adult) (4). This complexity is probably not unexpected, because dairy foods are complex mixtures and include several nutrients and nonnutrient substances that could potentially influence cancer etiology through either increases or decreases in risk. Healthy lifestyle factors that may accompany dairy food consumption could further confound the associations, although to our knowledge this has not been directly documented in the literature. To better understand these complex relations and to facilitate the synthesis of evidence-based dietary recommendations, more epidemiologic studies are necessary. The purpose of this study was to examine the associations between the types and quantity of dairy foods consumed and breast cancer in women participating in the Roswell Park Cancer Institute Data Bank and BioRepository (DBBR) between 2003 and 2014.
Methods
Archived clinical and questionnaire data were obtained from the DBBR at Roswell Park Cancer Institute (RPCI) from 1941 women diagnosed with breast cancer between December 2003 and October 2014, and 1237 control participants. DBBR control participants had no reported history of cancer, and were recruited from those who were accompanying cases (friends and family members), were community volunteers, or were employees of RPCI. Controls were randomly selected from this pool of healthy participants and frequency-matched on 10-y age strata to cases. The DBBR is a shared resource at Roswell Park Cancer Institute that provides biospecimens and linked data for studies of cancer etiology and prognosis (5). The protocol for the DBBR was approved by the RPCI Institutional Review Board, and all participants provided signed informed consent. Demographics, anthropometric measurements, medical history, lifestyle variables, and food habits were ascertained with a self-administered extensive epidemiologic questionnaire, and clinical characteristics for women with breast cancer were obtained from the RPCI tumor registry through linkage with the DBBR. We excluded women with an energy intake <2092 kJ/d or >18,828 kJ/d (n = 119), leaving 1857 cases and 1202 controls for analyses.
As part of the self-administered questionnaire, participants completed a detailed FFQ that queried the usual frequency of consumption of 110 foods and beverages in the year before diagnosis. Nutrient intake was calculated from the FFQ with the use of USDA food composition data and standard nutrient calculation algorithms. Dairy foods queried on the FFQ included fluid milk, yogurt, cheese (American, cheddar, or cream cheese), low-fat cheese, ricotta or cottage cheese, ice cream, low-fat frozen desserts, and pudding. Dairy foods were classified according to nutrient content and culinary use, and were expressed as servings/mo calculated from the FFQ as the product of frequency of use and portion size summed across group members. Separate questions were not queried for milk by fat content; rather, a single qualitative question was included that asked what types of milk were usually consumed that allowed multiple choices to be recorded but did not capture the proportion of each type of milk consumed. Therefore, we were unable to examine associations with milk by fat content. Monthly dairy consumption was categorized to represent typical daily serving sizes, and ranges varied according to dairy group. Dairy intake was not normally distributed; therefore, dairy consumption was evaluated as a categorical variable.
Statistical analyses were conducted with SAS 9.3 for Windows. All tests were 2-sided and considered to be statistically significant at P < 0.05. Menopause was defined as self-reported cessation of menses either as natural menopause or hysterectomy with bilateral oophorectomy. Differences in characteristics between cases and controls were assessed with standard descriptive statistics: differences in continuous characteristics between cases and controls were assessed with Student's t tests and with Pearson's chi-square for categorical variables. ORs and 95% CIs for the associations of breast cancer with each dairy group and total dairy were estimated with unconditional logistic regression adjusting for age (continuous), race (Caucasian or other), BMI (continuous), menopausal status (pre- or postmenopausal), energy intake (continuous), type of milk usually consumed (nonfat, low-fat, whole, or nondairy milk), cigarette smoking status (never, former, or current), and family history of breast cancer (yes or no). Although unadjusted ORs were similar to the adjusted estimates, adjustment for the above variables slightly strengthened the observed associations. In the interest of clarity, only the adjusted estimates are presented. Additional covariates (education, food groups other than dairy, alcohol, physical activity, and other lifestyle variables) were assessed for inclusion, but did not substantially modify the risk estimates. Analyses were conducted for breast cancer overall and further assessed by estrogen receptor (ER) status (ER+ and ER−). If data for ER status were missing, those cases were excluded for that analysis. Finally, because stratification by menopausal status is conventional in breast cancer research, we had initially conducted stratified analyses. However, estimates were similar between pre- and postmenopausal women; therefore, overall breast cancer estimates are presented herein.
Results
The descriptive characteristics of the women with breast cancer and healthy controls selected from the RPCI DBBR and included in this analysis are shown in Table 1. Among pre- and postmenopausal women, those with breast cancer tended to be older than those without breast cancer (mean ± SD: 45.5 ± 5.9 y compared with 44.7 ± 5.9 y and 64.6 ± 8.9 y compared with 62.3 ± 8.6 y, pre-and postmenopausal cases and controls, respectively). Postmenopausal women with breast cancer had a higher mean BMI (in kg/m2) than did postmenopausal women without breast cancer (29.2 ± 6.4 compared with 28.4 ± 6.1). Age at menarche was comparable between cases and controls, but cases were less likely than controls never to have had children. Premenopausal cases were more likely to be never smokers than were premenopausal controls, whereas postmenopausal cases were more likely to be current smokers than were postmenopausal controls. As expected, family history of breast cancer was higher in women with breast cancer than in those without breast cancer.
TABLE 1.
Descriptive characteristics of breast cancer cases and controls, Roswell Park Cancer Institute Data Bank and BioRepository, 2003–20141
| Premenopausal | Postmenopausal | |||
|---|---|---|---|---|
| Cases (n = 601) | Controls (n = 470) | Cases (n = 1256) | Controls (n = 732) | |
| Age, y | 45.5 ± 5.9* | 44.7 ± 5.9 | 64.6 ± 8.9** | 62.3 ± 8.6 |
| BMI, kg/m2 | 27.0 ± 6.3 | 27.4 ± 6.0 | 29.2 ± 6.4 | 28.4 ± 6.1 |
| Age at menarche, y | ||||
| ≤11 | 96 (16.0) | 94 (20.0) | 259 (20.6) | 146 (20.0) |
| 12 | 202 (33.6) | 153 (32.6) | 390 (31.1) | 230 (31.4) |
| 13 | 177 (29.5) | 123 (26.2) | 356 (28.3) | 212 (29.0) |
| 14 | 60 (10.0) | 61 (13.0) | 140 (11.2) | 81 (11.1) |
| 15 | 36 (6.0) | 20 (4.3) | 57 (4.5) | 29 (4.0) |
| ≥16 or never had period | 30 (5.0) | 19 (4.0) | 54 (4.3) | 34 (4.6) |
| Age at first birth, y | ||||
| Nulliparous | 133 (22.1)* | 131 (27.9) | 203 (16.2)* | 134 (18.3) |
| ≤19 | 53 (8.8) | 57 (12.1) | 205 (16.3) | 80 (10.9) |
| 20–24 | 149 (24.8) | 100 (21.3) | 466 (37.1) | 258 (35.3) |
| 25–29 | 140 (23.3) | 102 (21.7) | 250 (19.9) | 175 (23.9) |
| 30–34 | 99 (16.5) | 58 (12.3) | 99 (7.9) | 64 (8.7) |
| ≥35 | 27 (4.5) | 22 (4.7) | 33 (2.6) | 21 (2.9) |
| Smoking status | ||||
| Never | 353 (58.7)** | 262 (55.7) | 613 (48.8)* | 387 (52.9) |
| Former | 166 (27.6) | 144 (30.6) | 531 (42.3) | 300 (41.0) |
| Current | 82 (13.6) | 64 (13.6) | 112 (8.9) | 45 (6.2) |
| Family history of breast cancer | ||||
| No | 471 (78.4)** | 403 (85.7) | 950 (75.6) | 603 (82.4) |
| Yes | 130 (21.6) | 67 (14.3) | 306 (24.4) | 129 (17.6) |
Values are means ± SDs or n (%). Excludes women with implausible dietary data. Differences between cases and controls assessed with Student's t test for continuous variables and Pearson's chi-square for categorical variables. *P < 0.05; **P < 0.01.
The clinical characteristics of women with breast cancer are detailed in Table 2. The majority of breast cancer was stage 2 or lower among both pre- and postmenopausal women. ER− cancers were present in 20.0% of premenopausal and 16.3% of postmenopausal women. Approximately 25% of tumors were progesterone receptor negative (PR−). Human epidermal growth factor receptor 2 (HER2) status was positive in 12.8% of premenopausal and 7.2% of postmenopausal women.
TABLE 2.
Clinical characteristics of women with breast cancer, Roswell Park Cancer Institute Data Bank and BioRepository, 2003–20141
| Premenopausal (n = 601) | Postmenopausal (n = 1256) | |
|---|---|---|
| Stage | ||
| 0 | 92 (15.3) | 156 (12.4) |
| 1 | 225 (37.4) | 591 (47.1) |
| 2 | 181 (30.1) | 318 (25.3) |
| 3 | 65 (10.8) | 80 (6.4) |
| 4 | 10 (1.7) | 25 (2.0) |
| Unknown/missing | 28 (4.7) | 86 (6.9) |
| Estrogen receptor status | ||
| Negative | 120 (20.0) | 205 (16.3) |
| Positive | 413 (68.7) | 880 (70.1) |
| Missing | 68 (11.3) | 171 (13.6) |
| Progesterone receptor status | ||
| Negative | 149 (24.8) | 341 (27.2) |
| Positive | 384 (63.9) | 745 (59.3) |
| Missing | 68 (11.3) | 169 (13.5) |
| HER2 | ||
| Borderline | 0 | 2 (0.2) |
| Negative | 374 (62.2) | 818 (65.1) |
| Positive | 77 (12.8) | 90 (7.2) |
| Missing | 150 (25.0) | 346 (27.6) |
Values are n (%). HER2, human epidermal growth factor receptor 2.
Although we were unable to examine the associations between breast cancer and milk by fat content, adjustment for types of milk usually consumed significantly affected the majority of the estimates; thus, it was included in all models. In our sample, preference was distributed as follows: 14% drank no milk, 24% usually drank nonfat milk, 20% usually drank 2% milk, 11% usually drank 1% milk, 4% usually drank whole milk, and 27% drank various combinations of milk types (data not shown).
ORs and 95% CIs for associations between dairy intake and breast cancer are shown in Table 3. For total dairy, women in the highest compared with lowest category of intake (>42 servings/mo compared with <14 servings/mo) had a marginally significant 15% lower risk of breast cancer (OR: 0.85; 95% CI: 0.68, 1.06; P = 0.11). The inverse association between total dairy intake and breast cancer appeared to be mainly attributable to higher yogurt intake (OR: 0.61; 95% CI: 0.46, 0.82). Contrary to our observations between yogurt and breast cancer, we observed a marginally significant 53% increased risk of breast cancer associated with higher “other cheese” (American, cheddar, and cream cheese) intake (OR: 1.53; 95% CI: 0.99, 2.34; P = 0.05). No associations were observed with the remaining dairy groups and breast cancer in these data.
TABLE 3.
ORs and 95% CIs for associations between total dairy and specific dairy foods and breast cancer, Roswell Park Cancer Institute Data Bank and BioRepository, 2003–20141
| Servings/mo | Cases, n | Controls, n | OR (95% CI) |
|---|---|---|---|
| Total dairy | |||
| 0–14 | 424 | 245 | 1.00 |
| >14 to 28 | 457 | 250 | 1.10 (0.88, 1.38) |
| >28 to 42 | 360 | 241 | 0.93 (0.73, 1.18) |
| >42 | 616 | 466 | 0.85 (0.68, 1.06) |
| Milk | |||
| 0 | 495 | 322 | 1.00 |
| >0 to 14 | 852 | 535 | 0.94 (0.78, 1.14) |
| >14 to 28 | 250 | 172 | 0.88 (0.69, 1.13) |
| >28 | 260 | 173 | 0.96 (0.75, 1.24) |
| Yogurt | |||
| 0 | 417 | 196 | 1.00 |
| ≤14 | 1268 | 843 | 0.78 (0.64, 0.96) |
| >14 | 172 | 163 | 0.61 (0.46, 0.82) |
| Low-fat cheese | |||
| 0 | 191 | 115 | 1.00 |
| ≤14 | 1513 | 958 | 0.99 (0.76, 1.27) |
| >14 | 153 | 129 | 0.84 (0.60, 1.19) |
| Other cheese | |||
| 0 | 82 | 63 | 1.00 |
| >0 to 14 | 1613 | 1034 | 1.28 (0.91, 1.82) |
| >14 | 162 | 105 | 1.53 (0.99, 2.34) |
| Sweet dairy | |||
| 0 | 74 | 41 | 1.00 |
| >0 to 14 | 1537 | 974 | 0.89 (0.60, 1.33) |
| >14 to 28 | 178 | 136 | 0.75 (0.47, 1.18) |
| >28 | 68 | 51 | 0.73 (0.42, 1.26) |
ORs and 95% CIs estimated with unconditional logistic regression while adjusting for age, race, BMI, menopausal status, energy intake, type of milk usually consumed (nonfat, low-fat, whole, or nondairy milk), cigarette smoking status, and family history of breast cancer.
Associations between the intake of total dairy and specific dairy foods and ER+ and ER− breast cancer are shown in Table 4. We observed a borderline significant 18% reduction in the risk of ER+ cancer in women with highest compared with lowest total dairy intake (OR: 0.82; 95% CI: 0.64, 1.04; P = 0.10). As with breast cancer risk overall, yogurt intake was statistically significantly negatively associated with the risk of both ER+ and ER− breast cancer (ER+ OR: 0.65; 95% CI: 0.48, 0.89; ER− OR: 0.61; 95% CI: 0.38, 0.99). However, the estimates were comparable, supporting no effect of ER status on the observed associations (P-heterogeneity = 0.73). The intake of low-fat cheese was inversely associated with the risk of ER− breast cancer (OR: 0.54; 95% CI: 0.29, 0.99), although the estimates were not statistically significantly different from those associated with ER+ breast cancer (P-heterogeneity = 0.23). Associations between milk intake and ER status were significantly different (P-heterogeneity = 0.04), with no associations observed for ER+ breast cancer, but an increased risk of ER− breast cancer (OR: 1.58; 95% CI: 1.05,2.37). Finally, we observed significant heterogeneity in associations between sweet dairy foods and ER status (P = 0.01) wherein a higher intake was inversely associated with ER+ breast cancer (OR: 0.52; 95% CI: 0.29, 0.95), but appeared to increase the risk of ER− breast cancer, although this association was not statistically significant (OR: 1.55; 95% CI: 0.89, 2.70).
TABLE 4.
ORs and 95% CIs for associations between total dairy and specific dairy foods and breast cancer by ER status, Roswell Park Cancer Institute Data Bank and BioRepository, 2003–20141
| ER+ (n = 1293) | ER− (n = 205) | |||||
|---|---|---|---|---|---|---|
| Servings/mo | Controls (n = 1202), n | n | OR (95% CI) | n | OR (95% CI) | P-heterogeneity |
| Total dairy | 0.24 | |||||
| 0–14 | 245 | 301 | 1.00 | 72 | 1.00 | |
| >14 to 28 | 250 | 326 | 1.11 (0.87, 1.42) | 70 | 0.97 (0.66, 1.43) | |
| >28 to 42 | 241 | 249 | 0.91 (0.70, 1.18) | 64 | 0.99 (0.66, 1.46) | |
| >42 | 466 | 417 | 0.82 (0.64, 1.04) | 119 | 1.00 (0.68, 1.46) | |
| Milk | 0.04 | |||||
| 0 | 322 | 355 | 1.00 | 76 | 1.00 | |
| >0 to 14 | 535 | 586 | 0.90 (0.74, 1.11) | 142 | 1.03 (0.74, 1.42) | |
| >14 to 28 | 172 | 178 | 0.87 (0.66, 1.14) | 46 | 1.12 (0.74, 1.71) | |
| >28 | 173 | 174 | 0.88 (0.67, 1.16) | 61 | 1.58 (1.05, 2.37) | |
| Yogurt | 0.73 | |||||
| 0 | 196 | 290 | 1.00 | 74 | 1.00 | |
| ≤14 | 843 | 883 | 0.81 (0.66, 1.00) | 218 | 0.72 (0.52, 0.98) | |
| >14 | 163 | 120 | 0.65 (0.48, 0.89) | 33 | 0.61 (0.38, 0.99) | |
| Low-fat cheese | 0.23 | |||||
| 0 | 115 | 125 | 1.00 | 40 | 1.00 | |
| ≤14 | 958 | 1062 | 1.06 (0.80, 1.40) | 264 | 0.82 (0.55, 1.23) | |
| >14 | 129 | 106 | 0.92 (0.63, 1.35) | 21 | 0.54 (0.29, 0.99) | |
| Other cheese | 0.92 | |||||
| 0 | 63 | 58 | 1.00 | 13 | 1.00 | |
| ≤14 | 1034 | 1125 | 1.32 (0.90, 1.93) | 283 | 1.31 (0.70, 2.42) | |
| >14 | 105 | 110 | 1.56 (0.97, 2.50) | 29 | 1.44 (0.68, 3.03) | |
| Sweet dairy | ||||||
| 0 | 41 | 56 | 1.00 | 0.01 | ||
| >0 to 14 | 974 | 1074 | 0.82 (0.54, 1.26) | 276 | 1.002 | |
| >14 to 28 | 136 | 124 | 0.68 (0.42, 1.11) | 29 | 0.82 (0.53, 1.26) | |
| >28 | 51 | 39 | 0.52 (0.29, 0.95) | 20 | 1.55 (0.89, 2.70) | |
ORs and 95% CIs estimated with unconditional logistic regression while adjusting for age, race, BMI, menopausal status, energy intake, type of milk usually consumed (nonfat, low-fat, whole, or nondairy milk), cigarette smoking status, and family history of breast cancer. ER, estrogen receptor.
Categories collapsed to 0–14, >14 to 28, and >28 for ER negative cancer.
Discussion
In a recent meta-analysis, total dairy and yogurt consumption were inversely associated with breast cancer risk, especially among premenopausal women (4). Similarly, in this hospital-based case-control study of usual adult dairy consumption and breast cancer, we observed total dairy consumption to be negatively associated with breast cancer, and especially ER+ cancer. Whereas milk consumption was weakly negatively associated with breast cancer overall, a higher intake was strongly positively associated with ER− postmenopausal breast cancer. Unexpectedly, the consumption of sweetened dairy foods (pudding, low-fat frozen yogurt, and ice cream) was inversely related to ER+ breast cancer.
Dairy foods are important sources of several nutrients that could favorably affect cancer risk, including vitamin D, calcium, conjugated linoleic acid, butyrate, and other nutrients and phytochemicals, but they also contain substances, such as insulin-like growth factor 1 (IGF-1) and other growth hormones, that may adversely affect risk (6, 7). In the Nurses' Health Study, calcium and vitamin D were inversely related to breast cancer incidence in premenopausal women (8). The multivariable RR for highest calcium intake compared with lowest calcium intake was 0.80 (95% CI: 0.58, 1.12). When further separated into dairy compared with nondairy calcium, dairy calcium was associated inversely with risk. The consumption of >800 mg Ca compared with <200 mg Ca had an RR of 0.69 (95% CI: 0.48, 0.98). Total vitamin D intake was also associated with a lower risk of breast cancer in premenopausal women (RR: 0.72; 95% CI: 0.55, 0.94) in the Nurses' Health Study.
However, the Women's Health Initiative, a large randomized trial of vitamin D and calcium in postmenopausal women, did not show a beneficial effect of supplementation with these nutrients on breast cancer risk in women already consuming supplements (3), a common practice among postmenopausal women. However, associations with dairy foods and breast cancer were not assessed in the Women's Health Initiative, and, to our knowledge, clinical trials testing the impact of dairy foods on cancer risk have not been conducted. Given that dairy foods are complex mixtures of nutrients and nonnutrient substances that could be negatively as well as positively associated with risk, an examination of whole foods rather than single nutrients is warranted.
Whereas dairy foods may be important contributors of nutrients with anticarcinogenic potential, inverse associations may be partly explained by the healthier lifestyle adopted by those who consume low-fat dairy products. These factors could include tobacco avoidance, being physically active, use of dietary supplements, and interest in health-promoting behaviors, although the literature is sparse concerning documentation of these associations. However, adjustment for lifestyle factors such as physical activity did not have a large impact on our observed estimates, and we were unable to examine dairy consumption according to fat content, because our FFQ was not designed to query this level of detail. However, adjustment for usual type of milk consumed was an important covariate and suggests that part of our associations could be due to reduced fat consumption from low-fat dairy products, because the majority of milk consumed was nonfat or reduced fat.
Despite the fact that fluid milk consumption provided the largest contribution to total dairy intake in our study (R2 = 0.72), we observed fairly substantial inverse associations with yogurt consumption. Yogurt provides nutrients and nonnutrient compounds beyond those found in fluid milk, including probiotics and prebiotics that could promote a healthy gut bacterial community structure (9). The gut bacterial community has been implicated in both innate and adaptive immunity, which suggests that dysbiosis may contribute to suboptimal immune function and subsequent disease development (10, 11). Therefore, it is possible that higher yogurt consumption may be favorably affecting immune function and subsequent cancer risk.
Contrary to our expectations, we observed inverse associations between the intake of sweet dairy foods and ER+ breast cancer, although there were no associations with breast cancer overall. The sweet dairy group included ice cream, frozen yogurt, low-fat frozen desserts, and pudding, and it therefore contributed primarily to added sugar intake. A higher sugar intake could increase insulin secretion, which has been associated with cancer etiology; therefore, one would expect a positive association between the intake of these foods and breast cancer. Furthermore, the intake of these foods was low and did not contribute a large proportion of the variation in total dairy intake (R2 ≤ 0.04). Although there may be unknown mechanisms responsible for this finding, it is also possible that it is due to chance.
Whereas the majority of associations between dairy intake and breast cancer in this study were inverse, a strong positive association was noted between high and low consumption of milk and ER− postmenopausal breast cancer. Fluid milk is relatively high in IGF-1 as a consequence of the growth hormones used to increase milk production (12). Because IGF-1 and the ER participate in substantial crosstalk, positive associations would be more likely between ER+ breast cancer and dairy, which is contrary to our current findings (13–15). Alternatively, milk protein consumption has been shown to increase postprandial hyperinsulinemia, which could potentially increase cell growth and proliferation, independently of the ER (16).
The current study is subject to limitations common in hospital-based case-control studies. Case-control studies may be susceptible to recall bias if the cases are more likely than controls to remember an exposure. Both cases and the majority of controls were recruited at RPCI or at cancer-focused community events, and participants completed the questionnaire at home. Whereas the controls did not have a cancer diagnosis, many were family members or friends of nonbreast-cancer patients seeking care at RPCI, and thus also may have been more aware of the role of diet in cancer etiology, thus reducing the differential recall between the groups. Another possible limitation is that all cases were patients of RPCI and all controls were friends, relatives, or community members from the surrounding area. Therefore, the results may not be generalizable to all women, but should be generalizable to the western New York catchment area.
Dietary intake was queried with a self-administered FFQ that included the majority of commonly consumed dairy foods. Accurate completion of an FFQ requires the averaging of estimated intake of a fairly large number of foods, and, therefore, intake could be under- or overestimated by this method. Measurement of absolute intake was not a goal of our study; rather, we ranked participants on reported intake, which is standard epidemiologic methodology. Dairy intake in our sample was comparable to that reported in the NHANES (17), and we are confident that intake estimates are adequate for US populations. Finally, although the time period of interest was specified to be in the few years before diagnosis, this may not be the relevant time period for breast cancer development, particularly if growth hormones are of interest. Timing of dairy consumption was examined in the meta-analysis by Zang et al. (4), and childhood consumption was not strongly associated with subsequent breast cancer risk.
In conclusion, we found inverse associations between total dairy intake and yogurt intake and breast cancer, and positive associations between other cheese and breast cancer, as well as between milk and ER− breast cancer. Our study suggests that specific dairy foods may influence breast cancer risk, although the direction of the associations varied by food source. Future studies are also warranted to explore the mechanisms by which yogurt could contribute to risk reduction for breast cancer.
Acknowledgments
The authors' responsibilities were as follows—SEM, JH, and CBA: designed and conducted the research; SEM and JH: analyzed the data, wrote the manuscript, and had primary responsibility for the final content; and all authors: contributed to the review and refinement of the manuscript after initial analyses, and read and approved the final manuscript.
Abbreviations
- DBBR
Data Bank and BioRepository
- ER
estrogen receptor
- HER2
human epidermal growth factor receptor 2
- IGF-1
insulin-like growth factor 1
- PR−
progesterone receptor negative
- RPCI
Roswell Park Cancer Institute
References
- 1. World Cancer Research Fund/American Institute for Cancer Research Food, nutrition, physical activity, and the prevention of cancer: a global perspective. Washington (DC): AICR; 2007. [Google Scholar]
- 2. Jacobs ET, Kohler LN, Kunihiro AG, Jurutka PW.. Vitamin D and colorectal, breast, and prostate cancers: a review of the epidemiological evidence. J Cancer 2016;7:232–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Bolland MJ, Grey A, Gamble GD, Reid IR.. Calcium and vitamin D supplements and health outcomes: a reanalysis of the Women's Health Initiative (WHI) limited-access data set. Am J Clin Nutr 2011;94:1144–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Zang J, Shen M, Du S, Chen T, Zou S.. The association between dairy intake and breast cancer in Western and Asian populations: a systematic review and meta-analysis. J Breast Cancer 2015;18:313–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Ambrosone CB, Nesline MK, Davis W.. Establishing a cancer center data bank and biorepository for multidisciplinary research. Cancer Epidemiol Biomarkers Prev 2006;15:1575–7. [DOI] [PubMed] [Google Scholar]
- 6. Chagas CE, Rogero MM, Martini LA.. Evaluating the links between intake of milk/dairy products and cancer. Nutr Rev 2012;70:294–300. [DOI] [PubMed] [Google Scholar]
- 7. Qin LQ, He K, Xu JY.. Milk consumption and circulating insulin-like growth factor-I level: a systematic literature review. Int J Food Sci Nutr 2009;60Suppl 7:330–40. [DOI] [PubMed] [Google Scholar]
- 8. Shin MH, Holmes MD, Hankinson SE, Wu K, Colditz GA, Willett WC.. Intake of dairy products, calcium, and vitamin d and risk of breast cancer. J Natl Cancer Inst 2002;94:1301–11. [DOI] [PubMed] [Google Scholar]
- 9. Merenstein DJ, Smith KH, Scriven M, Roberts RF, Sanders ME, Petterson S.. The study to investigate the potential benefits of probiotics in yogurt, a patient-oriented, double-blind, cluster-randomised, placebo-controlled, clinical trial. Eur J Clin Nutr 2010;64:685–91. [DOI] [PubMed] [Google Scholar]
- 10. Honda K, Littman DR.. The microbiota in adaptive immune homeostasis and disease. Nature 2016;535:75–84. [DOI] [PubMed] [Google Scholar]
- 11. Thaiss CA, Zmora N, Levy M, Elinav E.. The microbiome and innate immunity. Nature 2016;535:65–74. [DOI] [PubMed] [Google Scholar]
- 12. Collier DA, Arranz MJ, Li T, Mupita D, Brown N, Treasure J.. Association between 5–HT2A gene promoter polymorphism and anorexia nervosa. Lancet 1997;350:412. [DOI] [PubMed] [Google Scholar]
- 13. Lee AV, Jackson JG, Gooch JL, Hilsenbeck SG, Coronado-Heinsohn E, Osborne CK, Yee D.. Enhancement of insulin-like growth factor signaling in human breast cancer: estrogen regulation of insulin receptor substrate-1 expression in vitro and in vivo. Mol Endocrinol 1999;13:787–96. [DOI] [PubMed] [Google Scholar]
- 14. Yee D, Lee AV.. Crosstalk between the insulin-like growth factors and estrogens in breast cancer. J Mammary Gland Biol Neoplasia 2000;5:107–15. [DOI] [PubMed] [Google Scholar]
- 15. Key TJ, Appleby PN, Reeves GK, Roddam AW.. Insulin-like growth factor 1 (IGF1), IGF binding protein 3 (IGFBP3), and breast cancer risk: pooled individual data analysis of 17 prospective studies. Lancet Oncol 2010;11:530–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Ludwig DS.. The glycemic index: physiological mechanisms relating to obesity, diabetes, and cardiovascular disease. JAMA 2002;287:2414–23. [DOI] [PubMed] [Google Scholar]
- 17. Cifelli CJ, Houchins JA, Demmer E, Fulgoni VL.. Increasing plant based foods or dairy foods differentially affects nutrient intakes: dietary scenarios using NHANES 2007–2010. Nutrients 2016;8:e422. [DOI] [PMC free article] [PubMed] [Google Scholar]