Diet and breast cancer

A review of the prospective observational studies

Abstract

The role of diet for the risk of breast cancer is of great interest as a potentially modifiable risk factor. The evidence from prospective observational studies was reviewed and summarized on selected dietary factors, gene-diet interactions, and breast cancer incidence. Dietary factors were considered that, based on their nutritional constituents, are of particular interest in the context of breast cancer: fat intake, biomarkers of fat intake, fruit and vegetable consumption, antioxidant vitamins (vitamins A, C, E, and beta-carotene), serum antioxidants, carbohydrate intake, glycemic index and glycemic load, dairy consumption (including vitamin D), consumption of soy products and isoflavones, green tea, heterocyclic amines, and adolescent diet. The PubMed database was searched for all prospective studies that relate these dietary items to the incidence of breast cancer or consider gene-diet interactions. Among the prospective epidemiologic studies conducted on diet and breast cancer incidence and gene-diet interactions and breast cancer incidence, to date there is no association that is consistent, strong, and statistically significant, with the exception of alcohol intake, overweight, and weight gain. The apparent lack of association between diet and breast cancer may reflect a true absence of association between diet and breast cancer incidence or may be due to measurement error exceeding the variation in the diet studied, lack of sufficient follow-up, and focus on an age range of low susceptibility. The risk of breast cancer can be reduced by avoidance of weight gain in adulthood and limiting the consumption of alcohol. Cancer 2007. © 2007 American Cancer Society.

Rationale

A role for diet in cancer etiology has been suggested in part because of the large international variation in cancer rates and may be ascribed to the antioxidant properties of selected nutrients, influence on DNA repair, DNA mutations, DNA adducts, metabolic detoxification, stimulation of growth factors, and potential antiestrogenic influence of some nutrients.1 Conversely, some foods and nutrients have been suggested to increase the risk of breast cancer through an increase in circulating levels of endogenous estrogen, insulin-like growth factor 1, or other growth factors. Energy balance, the interplay of caloric intake, physical activity, and metabolic rate is another important factor impacting breast risk through mechanisms not entirely understood.

Type of Studies

The majority of epidemiologic studies of diet and breast cancer are case-control studies. Some ecologic studies have been conducted and in more recent years data from cohort studies have become increasingly available. Few randomized clinical trials on diet have been conducted. Whereas some important leads may arise from case-control studies, they have the potential for recall bias because individuals with breast cancer may well associate their malignancy with a previous “unhealthy diet” or “bad diet” and thus overreport foods considered less healthy, whereas healthy control subjects may not have selective recall. By using prospectively and retrospectively collected diet data, Giovannucci et al.2 found that fat intake was associated with the risk of breast cancer only if the dietary data were retrospectively assessed, but not if they were collected prospectively. A similar study nested within the Canadian National Breast Screening Study did not evidence recall bias,3 suggesting that the degree of bias is likely to differ between studies. Furthermore, selection bias is a problem in case-control studies of diet and cancer. Selecting a comparable control group for a cancer case series is difficult and participation rates among individuals approached to be controls are often lower than participation rates among cases, introducing substantial bias. Ecologic studies are subject to confounding because correlations are made on a population level and cannot provide estimates of causal associations. Migrant studies can provide important information on the role of environmental factors but are again not able to provide specific information on the role of diet and cancer.

Randomized control trials of diet are problematic (unless randomizing dietary supplements) because the randomized dietary scheme has to be adhered to for many years; it is difficult for healthy participants to change their diet substantially and maintain this altered diet over a long period of time. The recently released results from the Women's Health Initiative (WHI) Randomized Controlled Dietary Modification Trial support this concern: As evidenced by the lack of difference in biomarkers of dietary intake between the intervention and control groups, women in the intervention group were largely unable to follow the low-fat diet (20% of calories from fat) they were assigned to, having consumed a high-fat diet (≥38% of calories) most of their lives.4 This lack of adequate contrast in diet between the intervention and control groups leaves the WHI results difficult to interpret. The number of randomized trials on diet and breast cancer is limited. It is possible that randomizing women with breast cancer to different dietary regimens to study survival or recurrence may be more successful because individuals with a severe illness may be more compliant with an assigned food regimen than healthy individuals. The gold standard would be a trial providing participants with all meals; however, such an approach is not feasible for cancer outcomes due to the required duration of intervention.

Given the wealth of studies available on diet and breast cancer, we deemed it appropriate to restrict the present review to cohort studies and nested case-control studies in which diet intake is assessed before diagnosis of disease.

Diet Assessment Methods

The valid and reproducible assessment of dietary intake in free-living populations is difficult because of the variety of foods and complexity of dishes consumed. With more meals prepared away from home, reporting of ingredients becomes increasingly difficult for participants in an observational study. All dietary instruments are subject to misclassification of food consumption, nutrient intake, and total caloric intake.

The most commonly used dietary assessment instrument in observational studies is the food frequency questionnaire (FFQ).5 The FFQ requests information on the frequency of consumption of a prespecified list of between 50 and 200 food items. The semiquantitative FFQ provides simple measures of portion size such as a glass or a cup. The strengths of this dietary assessment instrument are that dietary preferences and average frequencies of consumption of individual food items are generally captured reasonably well.5 On the other hand, because of the restricted list of food items, other food items, which were consumed but not listed, may be missed and nutrient calculations may be misclassified, especially if the contribution to a particular nutrient is high.5 The structured format of the FFQ with a specified list of selected foods allows the FFQ to be easily scanned and transformed into a food data file. Nutrient intake can be calculated using a standard nutrient database.5

The 7-day diet diary (7DD) collects more detailed information on food consumption during a limited period of time.5 The 7DD requires participants to keep records of every food item and beverage they consume during 1 week. The free format of the 7DD makes it much more laborious to computerize as the list of recorded foods is indefinite and dieticians have to determine foods that are part of prepared dishes and ready-to-eat meals and calculate their nutrient contents. Hence, the 7DD is rarely used in large-scale epidemiologic studies, with the notable exception of Epic Norfolk.6

The 24-hour recall provides a 1-day snippet of a person's diet, which can be compromised by day-to-day and seasonal variation of diet.5 Whereas the 24-hour recall can provide adequate population means, it can be substantially misclassified on the individual level. For this reason, often several recalls are administered on different weekdays and during different seasons of the year.5

For the reasons delineated above, and because it is easiest and most cost-efficient to administer, the FFQ has emerged as the most popular dietary assessment instrument in large-scale observational studies.

Analysis of Dietary Data

With a large number of foods and nutrients ascertained with a diet questionnaire the optimal analytic model is not obvious. Statistical models including a large number of foods suffer from collinearity, whereas models including only 1 food or 1 nutrient at a time are subject to confounding by foods and nutrients not included in the model. Several analytic approaches have attempted to overcome these limitations including the use of dietary patterns (using factor or cluster analysis),7 dietary indices, food groups (eg, fruits and vegetables), or the use of more complex statistical models such as hierarchical models that account for correlation of foods and nutrients using a 2-stage model.8

FINDINGS FROM COHORT STUDIES ON DIET AND BREAST CANCER

Outline

A large number of studies have addressed the association between diet and breast cancer. The results from these studies have dampened previous optimistic expectations that adult life diet may play an important role in breast cancer etiology.

In this review we summarize the most important findings from prospective cohort studies on diet and breast cancer. We will restrict our review to the following foods and nutrients of particular interest in the context of breast cancer: fat, fruits and vegetables, antioxidants, carbohydrates, glycemic index and glycemic load, dairy and vitamin D, soy and isoflavones, green tea, and heterocyclic amines. Because the relations between weight, body mass, and the incidence of breast cancer and regular alcohol consumption and the incidence of breast cancer are fairly well established9 we will only provide brief summaries of the available evidence. On the basis of our review we will give recommendations for future directions of research on diet and breast cancer.

Methods

Search strategy

We searched PubMed database for all articles in English published in peer-reviewed journals from January 1950 through May 2005 for evidence relevant to diet and breast cancer. We only included studies that were cohorts, nested case-control studies, meta-analyses, or pooled analyses and that reported a point estimate with an appropriate confidence interval (CI). Among nested case-control studies we only included those that had collected information on diet prospectively before disease occurrence. We specifically examined 11 dietary exposures: fat intake, biomarkers of fat intake, fruit and vegetable consumption, antioxidant vitamins (vitamins A, C, E, and beta-carotene), serum antioxidants, carbohydrate intake, glycemic index and glycemic load, dairy consumption (including vitamin D), consumption of soy products and isoflavones, green tea, heterocyclic amines, and adolescent diet. For this review we were interested in intake of antioxidants from diet and not from supplements. Studies that included only information from supplements were excluded.

We were also interested in studies that examined gene-environment interactions and searched the Medline database from January 1966 to May 2005, using the same criteria stated above. These molecular epidemiologic investigations represent efforts to examine the associations between dietary exposures and breast cancer risk among subgroups of women that are considered to be susceptible or nonsusceptible to the potential protective or harmful effects of these exposures on risk. For instance, a potential protective effect of high antioxidant intake might be more pronounced among women with genetic traits associated with high DNA repair capacity. On the other hand, women with reduced capacity to detoxify and excrete dietary carcinogen may be at greater risk of breast cancer than women with genetic traits associated with enhanced detoxification capacity.

All diet and breast cancer search strategies were limited to humans and included a phrase for breast cancer, study design, and the specific dietary exposure (Table A). Gene-environment interaction strategies were limited to humans and included a phrase for breast cancer, gene-environment interaction, and the specific dietary exposure.

Study selection

The search strategy identified a total of 1477 articles. We reviewed the abstracts of these articles to determine whether they met our criteria for review. Once we included a specific article in our review, we examined reference lists for additional articles. We did not attempt to identify unpublished articles or abstracts from scientific conferences. If there were 2 or more reports from 1 cohort, we included the report with the most up-to-date analysis. For fat intake we included 19 cohort studies,10–28 1 meta-analysis,29 and 1 pooled analysis.30 The meta-analysis and the pooled analysis included several of the 19 cohorts identified in our review.

For fruits and vegetables we included 6 cohorts,31–36 1 meta-analysis37 and 1 pooled analysis38; 11 cohorts for carbohydrates11–13, 18, 21, 22, 39–43; 11 cohorts for antioxidants16, 18, 32, 33, 35, 36, 44–48; and 11 cohorts for milk10, 13, 14, 24, 25, 27, 31, 49–52; 2 cohorts for vitamin D51, 53; 5 cohorts for soy18, 31, 54–56; 5 cohorts31, 57–59 and 1 pooled analysis for green tea58; 1 cohort for heterocyclic amines60; 2 retrospective cohorts for adolescent diet49, 51; and 4 nested case-control studies for biomarkers of fat composition.61–64 We also included 8 studies analyzing biomarkers of antioxidants.65–72 We also included 1 meta-analysis for milk, biomarkers of fat composition, and green tea.73–75

We also identified 7 nested case-control studies that examined gene-environment interactions.59, 76–81

Data extraction and synthesis

For each of the dietary exposures we have summarized the evidence in tables and presented evidence separately for pre- and postmenopausal status, if available. Information for each table was taken directly from the published manuscript of each individual study. When determining the size of the population for each study, we extrapolated information from the abstract or the methods section of the article. If the numbers differ, the number was identified from the methods section that best described the diet cohort for the study that was used in the analysis. We provide covariate-adjusted measures of association. In all studies, age and total caloric intake was adjusted unless noted. Although we have listed all variables that were either considered confounders or adjusted for in the analysis, we systematically record adjustment for body mass index (BMI) and family predisposition to breast cancer, which are factors likely associated with dietary patterns. We have not provided information about the participation rates because most of the studies are part of larger cohorts, ie, the Nurses' Health Study (NHS) or the Iowa Women's Study, in which no information is given about the original participant rate. The average follow-up time in the cohort studies was 8 to 9 years. Table B contains all the abbreviations for the studies included in the review. In addition, a critical review for each study was entered in a database (accessible at www.silentspring.org/sciencereview and www.komen.org/environment).

Studies on Diet and Breast Cancer

Weight, body mass, and weight gain

Epidemiologic evidence consistently demonstrates that a high BMI increases the incidence of breast cancer after menopause.82 Pooling data from 8 prospective cohort studies including 337,819 women and 3208 incident cases of invasive breast cancer after menopause resulted in a relative risk (RR) = 1.26 (95% CI, 1.09–1.46) for postmenopausal breast cancer comparing women with a BMI above 28 kg/m2 to women with a BMI of less than 21.83 Adult weight gain has been associated with postmenopausal breast cancer incidence in several studies.84–87

Alcohol

Regular alcohol consumption has been consistently linked to a modest increase in the incidence of breast cancer. In a recent pooled analyses of 6 prospective cohort studies including 322,647 women and 4335 cases of incident invasive breast cancer, consumption of each additional 10 g of alcohol per day was associated with a 9% (95% CI, 4%–13%) increase in the risk of breast cancer.9

Fat intake

Fat intake may be related to the risk of breast cancer because it may raise endogenous estrogen levels. A total of 19 cohort or nested case-control studies have been conducted on the association between fat intake and breast cancer incidence (Table 1).10–28 Two of these studies considered premenopausal women separately17, 24 and 6 studies considered postmenopausal women separately.16, 17, 22, 23, 26, 27 Overall fat intake was not related to the incidence of breast cancer. The most notable exception is NHANES I, which found a marked decrease in the risk of breast cancer associated with total fat intake (RR = 0.34; 95% CI, 0.16–0.73), but diet was assessed using a 24-hour recall.20 An increase in postmenopausal breast cancer incidence associated with total fat intake was found in the Rancho Bernardo study (RR = 2.01 per 28 g increase in fat intake; 95% CI, 1.19–3.41),12 and for the highest category of total fat intake in the Italian ORDET study (RR = 3.47; 95% CI, 1.43–8.44).11 Associations of borderline statistical significance were reported from the Nurses' Health Study II (NHS II)13 (RR = 1.25; 95% CI, 0.98–1.59) and among postemenopausal women of the Malmo Diet Cancer Cohort (MDC)23 (RR = 1.36; 95% CI, 0.96–1.94), comparing the highest versus the lowest quintile of intake. NHS II participants in the highest quintile of animal fat intake had a relative risk of breast cancer of 1.33 (95% CI, 1.02–1.73) compared with women in the lowest quintile.13 Among other studies, Gaard et al.14 observed an increase in the association between monounsaturated fat intake and breast cancer (RR = 1.72; 95% CI, 1.19–2.49) in a Norwegian study.

Table 1. Studies on Fat intake and Breast Cancer Incidence

Author, year, country, parent cohort	Size of cohort/no. of cases; years followed	Type of fat intake	Comparison of highestvs lowest quantile (quantity of intake)	Relative risk (95% CI)	Adjustment for confounding variables
					BMI	Family predisposition	Other variables
BCCDP indicates Breast Cancer Detection Demonstration Project Follow Up Cohort Study; CTS, California Teacher's Study; FSIIMCHS, Finnish Social Institution's Mobile Clinic Health Survey; IWHS, Iowa Women's Health Study; MDC, Malmo Diet and Cancer; NHANES, National Health and Nutrition Examination Survey I; NHEFS, National Health Epidemiologic Follow-up Study; NCS, Netherlands Cohort Study; NHS, Nurses' Health Study; NHS II, Nurses' Health Study II; NYUWHS, New York University's Women's Health Study; ORDET, Hormone and Diet Etiology of Breast Cancer; SMSC, Swedish Mammography Screening Cohort. * The effect of the variable on the measure of association was tested but the variable was not included in the final model.
Premenopausal and postmenopausal women combined
Jones20 1987	5485/99	Total	H vs l quartile (≥74 vs <38 g)	0.34 (0.16–0.73)	Y	Y	Age at menarche, menopausal status-age at menopause, income, education, parity, age at first birth*
USANHANES 1	1971/1975 (10 y mean follow-up)	Saturated	H vs l quartile (≥27 vs <13 g)	0.29 (0.12–0.67)
		Monounsaturated	H vs l quartile (≥29 vs <14 g)	0.59 (0.30–1.13)
		Polyunsaturated	H vs l quartile (≥9 vs <3 g)	0.73 (0.39–1.36)
Mills24 1989USA	20,341/2151976–1982	Total	H vs 1 quartile (no information given)	1.21 (0.81–1.81)	Y	Y	Calories, history of benign breast disease, education
Seventh Day Adventist Study
Knekt21 1990	3988/54	Total	H vs 1 tertile (≥97.3 vs <71.2 g)	1.72 (0.61–4.82)	Y*	Y*	Energy, parity, menopausal status, stature, smoking, region type, vitamin A, vitamin C, vitamin D, vitamin E, carotene*
Finland	1967–1986	Saturated	H vs 1 tertile (≥55.4 vs <39.6 g)	1.36 (0.50–3.73)
FSIIMCHS		Monounsaturated	H vs 1 tertile (≥31.1 vs <22.7 g)	2.70 (0.99–7.37)
		Polyunsaturated	H vs 1 tertile (≥6.8 vs <4.6 g)	1.23 (0.55–2.75)
Howe19 1991Canada	56,837/519 (1182 controls)Nested case-control study	Total	H vs 1 quartile (no information given)	1.30 (0.90–1.88)	Y*	Y*	Other sources of calories
CNBSS	1982–1987	Saturated	H vs l quartile (no information given)	1.08 (0.73–1.59)			Education, age of menarche, age at first pregnancy, nulliparity, surgical menopause, history of benign breast disease, breast cancer in first-degree relatives*
		Monounsaturated	H vs 1 quartile (no information given)	1.23 (0.81–1.89)
		Polyunsaturated	H vs 1 quartile (no information given)	1.30 (0.93–1.82)
Toniolo25 1994USA	14,291/180 (829 controls)Nested case-control	Total	H vs 1 quintile (mean in controls 123 vs 28 g/d)	1.49 (0.89–2.48)	Y*	Y*	Energy adjustedHeight, menopausal status (matched), date of enrollment (matched), age at menarche, age at first full term pregnancy, number of full-term pregnancies, history of benign breast disease, race, religion*
NYUWHS	1985–1991	Saturated	H vs 1 quintile (mean in controls 50 vs 11 g/d)	1.47 (0.88–2.46)
		Monounsaturated	H vs 1 quintile (mean in controls 42 vs 10 g/d)	1.57 (0.90–2.71)
		Polyunsaturated	H vs 1 quintile (mean in controls 19 vs 4 g/d)	1.13 (0.65–1.98)
Gaard14 1995	31,209/248	Total	H vs 1 quartile (≥68 vs <41 g)	1.25 (0.86–1.81)	Y	N	Energy, smoking, height, menopausal status
Norway	1977–1983	Saturated	H vs1 quartile (≥30 vs <17 g)	1.01 (0.75–1.57)
Norwegian National Health Screening Services		Monounsaturated	H vs1 quartile (≥23 vs <14 g)	1.72 (1.19–2.49)
Byrne10 1996	6156/53	Total	H vs1 quartile (>59.41 vs <36.9 g)	0.68 (0.3–1.5)	Y*	Y*	Calories, number of breast biopsies, alcohol, physical activity, oral contraceptives, age at menarche, menopausal hormones, race, education, region of country, area of residence, family income*
USA	1982/1984–1986/1987
NHEFS/NHANES 1
Wolk28 1998	61,471/674	Total	H vs 1 quartile (>50.2 vs <40.3 g/d)	1.0 (0.76–1.32)	Y	Y		Parity, age at first child's birth, education, cholesterol, fiber, alcohol, total energy intake
Sweden	1987–1991 (mean 4.2 y)	Saturated	H vs 1 quartile (>21.7 vs <16.3 g/d)	1.20 (0.89–1.63)
SMSC		Monounsaturated	H vs 1 quartile (>18.4 vs <14.4 g/d)	0.80 (0.52–1.21)
		Polyunsaturated	H vs 1 quartile (>7.7 vs <5.3 g/d)	1.18 (0.85–1.64)
Holmes17 1999	88,795/2956	Total	Continuous variable; 5% increase	0.97 (0.94–1.00)	Y	Y	Energy, energy-adjusted vitamin A intake, protein intake, alcohol, time period, height, weight change since age 18 years, benign breast disease, parity, age at first birth, age at menopause, menopausal status, use of hormonal replacement therapy, age of menarche
USA	1980–1994	Animal	Continuous variable; 5% increase	0.98 (0.96–1.01)
NHS		Vegetable	Continuous variable; 5% increase	0.97 (0.93–1.02)
		Polyunsaturated	Continuous variable; 5% increase	0.94 (0.88–1.01)
		Saturated	Continuous variable; 5% increase	0.94 (0.88–1.00)
		Monounsaturated	Continuous variable; 5% increase	0.91 (0.79–1.04)
		Trans-Unsaturated fat	Continuous variable; 1% increase	0.92 (0.86–0.98)
Horn-Ross18 2002	111,526/711	Total	H vs 1 quartile (no information given)	0.8 (0.6–1.2)	Y	Y	Race, daily caloric intake, age at menarche, nulliparity/age at first full-term pregnancy, physical activity, interaction for BMI and menopausal status
USA	1995/1996–1998	Saturated	H vs 1 quartile (no information given)	0.8 (0.6–1.2)
CTS		Monounsaturated	H vs 1 quartile (no information given)	0.9 (0.6–1.2)
		Polyunsaturated	H vs 1 quartile (no information given)	0.9 (0.7–1.3)
Cho13 2003USA	90,655/7141991–1999	Total	H vs 1 quintile (median 38% vs 24% of energy)	1.25 (0.98–1.59)	Y	Y	Smoking, height, parity and age at first birth, age at menarche, history of benign breast disease, oral contraceptive use, menopausal status, alcohol intake, energy, protein
NHS II		Animal	H vs 1 quintile (median 23% vs 12% of energy)	1.33 (1.02–1.73)
		Vegetable	H vs 1 quintile (median 19% vs 9% of energy)	0.97 (0.76–1.24)
		Saturated	H vs 1 quintile (median 14% vs 8% of energy)	1.06 (0.74–1.53)
		Monounsaturated	H vs 1 quintile (median 15% vs 9 % of energy)	1.10 (0.75–1.62)
		Polyunsaturated	H vs 1 quintile (median 7% vs 4 % of energy)	0.96 (0.73–1.27)
		Trans-unsaturated	H vs 1 quintile (median 2.3% vs 0.9% of energy)	0.96 (0.70–1.31)
Gago-Dominguez15 2003China	34,734/3141993–1998	Total	H vs 1 quartile (≥29.44 vs ≤21.87 % kcal)	0.94 (0.68–1.31)	Y*	Y	Energy intake, year of interview, dialect group, education, daily alcohol intake, age of menarche, number of live births, age when periods became regular
Singapore Chinese Health Study		Saturated	H vs 1 quartile (≥10.73 vs ≤7.18 % kcal)	0.92 (0.67–1.26)
		Monounsaturated	H vs 1 quartile (≥10.00 vs ≤7.23 % kcal)	1.02 (0.73–1.43)
		Polyunsaturated	H vs 1 quartile (≥6.27 vs ≤3.95 % kcal)	1.27 (0.92–1.74)
Premenopausal women
Holmes17 1999	437,613 person y/784	Total	Continuous variable; 5% increase	0.99 (0.93–1.05)	Y	Y	Energy, energy-adjusted vitamin A intake, protein intake, alcohol, time period, height, weight change since age 18 years, benign breast disease, parity, age at first birth, age at menopause, menopausal status, use of hormonal replacement therapy, age of menarche
USA	1980–1994	Animal	Continuous variable; 5% increase	1.01 (0.96–1.06)
NHS		Vegetable	Continuous variable; 5% increase	0.99 (0.91–1.07)
		Polyunsaturated	Continuous variable; 5% increase	0.98 (0.87–1.11)
		Saturated	Continuous variable; 5% increase	1.02 (0.91–1.15)
		Monounsaturated	Continuous variable; 5% increase	0.99 (0.77–1.27)
		Trans-unsaturated fat	Continuous variable; 1% increase	1.00 (0.88–1.13)
Postmenopausal women
Graham16 1992	18,586/359	Total	H vs 1 quintile (≥2344 vs ≤1268 g)	0.99 (0.69–1.41)	N	N	Education
USA	1980–1987	Animal	H vs 1 quintile (≥1780 vs ≤893 g)	1.12 (0.78–1.61)
NY State Cohort		Vegetable	H vs 1 quintile (≥699 vs ≤3108 g)	1.07 (0.76–1.50)
Kushi22 1992	34,388/459	Total	H vs 1 quartile (median 80.7 vs 56.6 g)	1.16 (0.87–1.55)	Y	Y	Total energy intake, age at menarche, age at menopause, age at first birth, waist-to-hip ratio, history of benign breast disease, BMI at age 18, alcohol intake
USA	1986–1989	Saturated	H vs 1 quartile (median 29.3 vs 18.8 g)	1.10 (0.83–1.46)
IWHS		Monounsaturated	H vs 1 quartile (median 30.7 vs 20.3 g)	1.08 (0.80–1.46)
		Polyunsaturated	H vs 1 quartile (median 15.8 vs 9.2 g)	1.15 (0.87–1.52)
Barrett-Connor12 1993	575/15	Total	Continuous variable; 28 g increase	2.01 (1.19–3.41)	Y	N	Energy, age at menopause, parity, alcohol, total calories, total carbohydrates, total protein
USA	1972–1987
Holmes17 1999	620,329 person y/1913	Total	Continuous variable; 5% increase	0.96 (0.93–1.00)	Y	Y	Energy, energy-adjusted vitamin 1 intake, protein intake, alcohol, time period, height, weight change since age 18 years, benign breast disease, parity, age at first birth, age at menopause, menopausal status, use of hormonal replacement therapy, age of menarche
USA	1980–1994	Animal	Continuous variable; 5% increase	0.98 (0.94–1.02)
NHS		Vegetable	Continuous variable; 5% increase	0.96 (0.91–1.02)
		Polyunsaturated	Continuous variable; 5% increase	0.88 (0.74–1.04)
		Saturated	Continuous variable; 5% increase	0.93 (0.85–1.02)
		Monounsaturated	Continuous variable; 5% increase	0.91 (0.84–0.99)
		Trans-unsaturated fat	Continuous variable; 1% increase	0.91 (0.84–0.99)
Velie26 2000USA	40,222/9961987–1993 (mean 5.3 y)	Total	H vs 1 quartile (mean 45.4% vs 23.9% energy)	1.07 (0.86–1.32)	Y	Y	Total energy, height, education, alcohol use, parity, age at first birth, educational level, alcohol use, age at menarche, history of benign breast disease, other fat subtypes
BCCDP		Saturated	H vs 1 quintile (mean 15.7% vs 11.5% energy)	1.12 (0.87–1.45)
		Unsaturated	H vs 1 quintile (mean 30.6% vs 23.5% energy)	1.13 (0.88–1.45)
		Monounsaturated	H vs 1 quintile (mean 11.2% vs 7.3% energy)	0.88 (0.62–1.25)
		Polyunsaturated	H vs 1 quintile (mean 16.0% vs 12.0% energy)	1.05 (0.82–1.34)
Sieri11 2002	3367/56 (214 controls)	Total	H vs 1 tertile (≥62.8 vs <54.3 g)	3.47 (1.43–8.44)	Y	N		Energy intake, parity, place of birth, level of education, total fat (for all analyses except total fat)Matched on daylight-saving period, recruitment center, recruitment data
Italy	Nested case-control	Animal	H vs 1 tertile (≥36.3 vs <27.6 g)	1.84 (0.63–5.43)
ORDET	1987–1992	Vegetable	H vs 1 tertile (≥30.0 vs <22.3 g)	0.88 (0.29–2.66)
		Saturated	H vs 1 tertile (≥22.2 vs <18.3 g)	1.12 (0.31–4.04)
		Monounsaturated	H vs 1 tertile (≥30.0 vs <23.5 g)	2.96 (0.70–12.6)
		Polyunsaturated	H vs 1 tertile (≥7.7 vs <6.3 g)	2.03 (0.68–6.03)
		Linoleic	H vs 1 tertile (≥6.18 vs <5.07 g)	1.39 (0.51–3.80)
		Linolenic	H vs 1 tertile (≥0.99 vs <0.86 g)	0.71 (0.20–2.55)
Voorrips27 2002	62,573/941 (1598 controls)	Total	H vs 1 quintile (median 86 vs 61 g/d)	1.13 (0.84–1.52)	Y	Y	History of benign breast disease, age at menarche, age at menopause, oral contraceptive use, parity, age at first childbirth, education, alcohol use, current cigarette smoking, total energy intake, total adjusted fat intake
Netherlands	Nested case-control	Animal	H vs 1 quintile (median 71 vs 35 g/d)	1.05 (0.79–1.40)
NCS	1986–1992	Vegetable	H vs 1 quintile (median 38 vs 5 g/d)	1.02 (0.75–1.38)
		Saturated	H vs 1 quintile (median 38 vs 22 g/d)	1.40 (0.97–2.03)
		Monounsaturated	H vs 1 quintile (median 27 vs 18 g/d)	0.61 (0.38–0.96)
		Polyunsaturated	H vs 1 quintile (median 24 vs 8 g/d)	0.88 (0.65–1.21)
		Trans-unsaturated	H vs 1 quintile (median 3.6 vs 1.5 g/d)	1.30 (0.93–1.80)
Mattisson23 2004Sweden	11,726/3421991/1996–2001 (mean 7.6 y)	Total	H vs 1 quintile (mean 45.4% vs 29.6% of total energy)	1.36 (0.96–1.94)	N	N	Energy, interviewer, season of diet interview, change of dietary habits, height, weight, current hormone use, age at birth of first child, age at menarche, physical activity, smoking, drinking, education
MDC

Two studies pooled the prospective evidence (Table 2): Boyd et al.29 in 1993 conducted a meta-analysis including 7 prospective studies with a total of about 3000 cases of breast cancer and found an overall RR = 1.03 (95% CI, 0.92–1.96) for the highest category of intake compared to the lowest. Smith-Warner et al.30 in 2001 pooled 8 prospective studies on fat intake and breast cancer and derived an overall RR = 1.00 (95% CI, 0.98–1.03) per 5% increase in fat. In this pooled analysis the RR associated with intake of saturated fat was 1.09 (95% CI, 1.00–1.99) and of animal fat 1.01 (95% CI, 0.96–1.06) per 5% increase in fat.

Table 2. Meta-analysis/Pooled Analysis of Studies on Fat Intake And Breast Cancer Incidence

Author, year, parent cohorts	Size of cohort/no. of cases; typeof study	Type of fat intake	Comparison of highest vs lowest quantile (quantity of intake)	Relative risk(95% CI)	Adjustment for confounding variables
					Age	BMI	Family predisposition	Other variables
Boyd29 1993 (includes Graham16 1992; Howe19 1991; Jones20 1987; Knekt21 1990; Kushi22 1992; van den Brandt^109 1993; Willett^110 1992)	252,765/3007	Total	H vs l level (different for each study)	1.03 (0.92–1.16)	N	N	N	Energy intake, other breast cancer risk factors
	Meta-analysis
		Saturated		0.95 (0.84–1.08)
		Monounsaturated		0.95 (0.84–1.08)
		Polyunsaturated		1.00 (0.89–1.13)
Smith-Warner30 2001 (includes Toniolo25 1994; Graham16 1992; Holmes17 1999; Howe19 1991; Kushi22 1992; Mills24 1989; van den Brandt^109 1993; Wolk28 1998)	351,821/7321	Total	Continuous variable; 5% increase in energy	1.00 (0.98–1.03)	Y	Y	Y	Percent of energy from protein, percent of energy from alcohol, age at menarche, parity, age at birth of first child, menopausal status at diagnosis, postmenopausal hormone use, oral contraceptive use, history of benign breast disease, smoking status, education, body mass index and menopausal status interaction, height, fiber intake, energy intake
	Pooled analysis
		Animal	Continuous variable; 5% increase in energy	1.01 (0.96–1.06)
		Vegetable	Continuous variable; 5% increase in energy	1.01 (0.98–1.04)
		Saturated	Continuous variable; 5% increase in energy	1.09 (1.00–1.19)
		Monounsaturated	Continuous variable; 5% increase in energy	0.93 (0.84–1.03)
		Polyunsaturated	Continuous variable; 5% increase in energy	1.05 (0.96–1.16)

Biomarkers of fat intake

The use of biochemical markers that reflect dietary intake has potential advantages compared with the assessment of dietary intake through self-reports, as reporting errors and limitations of food composition tables are avoided. However, levels of biomarkers can be affected by several factors other than diet, such as smoking and metabolic factors. Furthermore, biomarkers that reflect dietary intake with sufficient accuracy are available for only a limited number of foods and nutrients. In particular, few recovery biomarkers have been identified. Recovery biomarkers are based on the balance between intake and output of a specific nutrient in a certain time period and can be translated into estimates of absolute intake over that period.88 Only urinary nitrogen as a biomarker for protein intake and potassium excretion have been identified as recovery biomarkers with adequate validity. Conversely, concentration biomarkers do not reflect absolute intake and correlations with dietary intake are lower, but in the absence of recovery biomarkers they provide approximations of intake. Examples of concentration biomarkers are fatty acid compositions and carotenoid concentrations.5

Dietary fatty acid intake is reflected by the concentration of particular fatty acids in adipose tissue, in fractions of triglycerides, phospholipids, and cholesteryl ester in serum, plasma or erythrocyte membranes, and by free fatty acids.89, 90 The fatty acid composition of adipose tissue reflects long-term intake, whereas the fatty acid profile of serum or plasma phospholipids reflects medium-term intake of dietary fat.91 These blood fractions can be used as biomarkers of habitual dietary intake of fatty acids in observational studies.

Four studies have considered biomarkers of fat intake and incidence of breast cancer using a nested case-control design (Table 3).61–64 Only the ORDET Study conducted in Italy found a statistically significant association between monounsaturated fatty acids measured in erythrocyte membranes and breast cancer (odds ratio [OR] = 5.21; 95% CI, 1.95–13.9) and polyunsaturated fat (OR = 0.34; 95% CI, 0.15–0.79).63

Table 3. Studies of Biomarkers of Fat Intake and Breast Cancer Incidence

Author, year, parent cohort	No. of controls/no. of cases; years followed	Type of fat	Comparison of highest vs lowest quantile (quantity of intake)	Odds ratio (95% CI)	Adjustment for confounding variables
					BMI	Family predisposition	Other variables
MDC indicates Malmo Diet Cancer Cohort; MONICA, Monitoring of Trends in Cardiovascular diseases; MSP, Mammary Screening Project; ORDET, Hormone and Diet Etiology of Breast Cancer; NYUWHS, New York University Women's Health Study; VIP, Vasterbotten Intervention Project. * The effect of the variable on the measure of association was tested but the variable was not included in the final model.
Premenopausal and postmenopausal women combined
Chajes61 1999	388 controls/196	Saturated	H vs 1 quartile (no information given)	1.15 (0.46–2.85)	N	N	Age at menarche, height, weight, age at first full-term pregnancy, number of children, use of hormone-replacement therapy
Sweden	Nested case-control	Monounsaturated		1.78 (0.81–3.92)
VIP/MSP/MONICA	1986–1997	Polyunsaturated (n-6)		0.91 (0.40–2.06)
		Polyunsaturated (n-3)		0.58 (0.27–1.28)
Saadatian-Elahi64 2002	197 controls/197	Saturated	H vs 1 quartile (no information given)	1.46 (0.74–2.83)	N	Y	Age at first full-term birth, history of benign breast disease, total cholesterol
USA	Nested case-control	Monounsaturated		1.15 (0.60–2.18)
NYUWHS	1985/1991–1998	Polyunsaturated		0.59 (0.31–1.09)
Premenopausal women
Saadatian-Elahi64 2002	91 controls/91	Saturated	H vs 1 quartile (no information given)	1.66 (0.56–4.89)	N	Y	Age at first full-term birth, history of benign breast disease, and total cholesterol
USA	Nested case-control	Monounsaturated		1.13 (0.42–3.04)
NYUWHS	1985/1991–1998	Polyunsaturated		0.60 (0.24–1.54)
Postmenopausal women
Pala63 2001ItalyORDET	141 controls/71 Nested case-control1987/1992–1995 (mean 5.5 y)	Saturated	H vs 1 tertile (≥17.42 vs <16.16% phospholipid fatty acid composition)	1.01 (0.45–2.29)	Y*	N	Age at menarche, age at first child birth, age at menopause, months of location, parity, educational level*
		Monounsaturated	H vs 1 tertile (≥43.61 vs <42.44% phospholipid fatty acid composition)	5.21 (1.95–13.91)
		Polyunsaturated	H vs 1 tertile (≥35.16 vs <33.69% phospholipid fatty acid composition)	0.34 (0.15–0.79)
Saadatian-Elahi64 2002	106 controls/106	Saturated	H vs 1 quartile (no information given)	1.96 (0.73–5.25)	Y*	Y	Age at first full-term birth, history of benign breast disease, total cholesterolAge at menarche*
USA	Nested case-control	Monounsaturated		1.38 (0.55–3.49)
NYUWHS	1985/1991–1998	Polyunsaturated		0.42 (0.17–1.08)
Wirfalt62 2004	673 controls/237	Stearic (saturated)	Continuous variable; 1 unit change	1.04 (0.84–1.27)	Y	N	Alcohol habits, current hormone therapy, age at birth of first child, waist circumference, height
Sweden	1991/1996–1999	Monounsaturated (Oleic)		0.99 (0.85–1.15)
MDC		Polyunsaturated (Linoleic)		0.99 (0.89–1.10)

In a pooled analysis of these studies including 268 postmenopausal cases, high levels of monounsaturated fatty acids were associated with an elevated RR, in particular among postmenopausal women (RR = 2.20; 95% CI, 1.93–2.52)75; this association was restricted to cohort studies and not seen in case-control studies. High levels of n3-polyunsaturated fatty acids (OR = 0.58; 95% CI, 0.52–0.64), n6-polyunsaturated fatty acid (RR = 0.67;95% CI, 0.59–0.75), and linoleic acid (RR = 0.88; 95% CI, 0.78–0.98) were associated with a decreased RR of breast cancer, again, particularly among postmenopausal women (Table 4).75

Table 4. Meta-analyses of Studies of Biomarkers of Fat Intake and Breast Cancer Incidence

Author, year, parent cohorts	No. of controls/no. of cases	Type of fat	Comparison of highest vs lowest quantile (quantity of intake)	Odds ratio(95% CI)	Adjustment for confounding variables
					Age	BMI	Family predisposition	Other variables
Premenopausal and postmenopausal women combined
Saadatian-Elahi75 2004 (includes Chajes61 1999; Pala63 2001; Saadatian-Elahi64 2002)	728/464	Saturated	H vs 1 quartile (no limits given)	1.36 (0.84–2.19)	N	N	N	N
	Monounsaturated		1.93 (1.03–3.61)
	Polyunsaturated (n-6)		0.67 (0.44–1.02)
		Polyunsaturated (n-3)		0.61 (0.40–0.93)
		Linoleic acid		0.91 (0.53–1.57)
Postmenopausal women
Saadatian-Elahi75 2004 (includes Pala63 2001; Saadatian-Elahi64 2002)	531/268	Saturated	H vs 1 quartile (no limits given)	1.26 (1.10–1.45)	N	N	N	N
		Monounsaturated		2.20 (1.93–2.52)
		Polyunsaturated (n-6)		0.67 (0.59–0.75)
		Polyunsaturated (n-3)		0.58 (0.52–0.64)
		Linoleic acid		0.88 (0.78–0.98)

Consumption of fruits and vegetables

Fruit and vegetable consumption may prevent breast cancer through their antioxidants, fiber, and other nutrients. Six prospective cohort studies have considered the relation between fruit and vegetable consumption and the incidence of breast cancer (Table 5)31–36; 3 studies were restricted to postmenopausal women.33, 35, 36 The only significant association was reported from the Nurses' Health Study (NHS), which reported an inverse association for regular consumption of 5 or more vegetables per day and premenopausal breast cancer incidence compared with consuming less than 2 vegetables per day (RR = 0.64; 95% CI, 0.43–0.95).36 In a meta-analysis including 5 cohorts and 2608 cases an RR of 0.73 (95% CI, 0.64–0.83) for breast cancer was associated with high vegetable consumption,37 but a pooled analysis of data from 8 prospective cohorts including 7377 cases resulted in an RR = 0.96 (95% CI, 0.89–1.04) for vegetable consumption and RR = 0.93 (95% CI, 0.86–1.00) for consumption of fruits and vegetables comparing the women in the highest versus the lowest category of intake (Table 6).38

Table 5. Studies on Fruit and Vegetable Consumption, and Breast Cancer Incidence

Author, year, country, parent cohort	Size of cohort/no. of cases; years followed	Type of fruit and vegetable consumption	Comparison of highest vs lowest quantile (quantity of intake)	Relative risk (95% CI)	Adjustment for confounding variables
					BMI	Family predisposition	Other variables
CNBSS indicates Canadian National Breast Screening Study; EPIC, European Prospective Investigation Into Cancer and Nutrition; NCS, Netherlands Cohort Study; NHS, Nurses' Health Study. * The effect of the variable on the measure of association was tested but the variable was not included in the final model.
Premenopausal and postmenopausal women combined
Rohan32 1993	56,837/519 (1182 controls)	Fruits	H vs 1 quintile (cut point: 491 vs 189 g/d)	0.81 (0.57–1.14)	N	Y	Age at menarche, surgical menopause, age at first live birth, years of education, history of benign breast disease, other contributors to total food intake
Canada
CNBSS	Nested case-control	Vegetables	H vs 1 quintile (cut point: 433 vs 203 g/d)	0.86 (0.61–1.23)
	1982–1987
Key31 1999	34,759/427	Fruits	H vs 1 tertile (≥5/wk vs ≤1 wk)	0.95 (0.71–1.27)	N	N	Calendar period, city, age at time of bombing, radiation dose
JapanRadiation Effects Research Foundation's Life Span Study	1969/1970–1993; 1979/1980–1993
van Gils34 2005	285,526/3659	Fruits	H vs 1 quintile (>367 vs ≤114 g/d)	1.09 (0.94–1.25)	Y*	N	Center, energy intake, alcohol intake, saturated fat intake, height, weight, age at menarche, parity, current oral contraceptive use, current use of hormone therapy, menopausal status, smoking status, physical activity, education
EuropeEPIC	1992/1998–2002 (mean 5.4 y)	Vegetables	H vs 1 quintile (>309 vs <109 g/d)	0.98 (0.84–1.14)
Premenopausal women
Zhang36 1999	23,808/784	Fruits and vegetables	H vs 1 quintile (≥5.0 vs <2 servings/d)	0.77 (0.58–1.02)	Y at age 18	Y	Length of follow up, total energy intake, parity, age at first birth, age at menarche, history of benign breast disease, alcohol, weight change from age 18, height
USA	1980–1994
NHS		Fruits	H vs 1 quintile (≥5.0 vs <2 servings/d)	0.74 (0.45–1.24)
		Vegetables	H vs 1 quintile (≥5.0 vs <2 servings/d)	0.64 (0.43–0.95)
Postmenopausal women
Shibata33 1992	45,941 person y/219	Fruits and Vegetables	H vs 1 tertile (≥8.3 vs <5.9 servings/d)	0.87 (0.63–1.21)	N	N	Smoking
USA	1981–1989
		Fruits	H vs 1 tertile (≥3.7 vs <2.4 servings/d)	0.82 (0.60–1.12)
		Vegetables	H vs 1 tertile (≥4.8 vs <3.2 servings/d)	0.96 (0.69–1.34)
Verhoeven35 1997	62,573/650 (1716 sub-cohort)	Fruits	H vs 1 quintile (median 343.1 vs 124.0 g/d)	0.76 (0.54–1.08)	N	Y	Energy intake, alcohol intake, history of benign breast disease, age at menarche, age at menopause, age at first birth, parity
Netherlands	1986–1990
NCS		Vegetables	H vs 1 quintile (median 303.0 vs 108.0 g/d)	0.94 (0.67–1.31)
Zhang36 1999USANHS	59,426/1913 1980–1994	Fruits and vegetablesFruitsVegetables	H vs 1 quintile (≥5.0 vs <2 servings/d)H vs 1 quintile (≥5.0 vs <2 servings/d)H vs 1 quintile (≥5.0 vs <2 servings/d)	1.03 (0.81–1.31)0.84 (0.64–1.09)1.02 (0.85–1.24)	Y at age 18	Y	Length of follow up, total energy intake, parity, age at first birth, age at menarche, history of benign breast disease, alcohol, weight change from age 18, height, age at menopause, HRT

Table 6. Meta-Analysis/Pooled Analysis of Studies of Fruit and Vegetable Consumption, and Breast Cancer Incidence

Author, year, parent cohorts	Size of cohort/no. of cases; type of study	Type of fruit and vegetable consumption	Comparison of highest vs lowest quantile (quantity of intake)	Relative risk (95% CI)	Adjustment for confounding variables
					Age	BMI	Family predisposition	Other variables
Premenopausal and postmenopausal women combined
Gandini37 2000 (includes Graham16 1992; Hunter^111 1993; Kushi22 1992; Rohan32 1993; Verhoeven35 1997)	208,904/2608Meta-analysis	Vegetables	High consumption vs low consumption (>1 portions/d vs <3 to 4 portions/wk)	0.73 (0.64–0.83)	N	N	N	N
Smith-Warner38 2001 (Toniolo25 1994; Graham16 1992; Holmes17 1999; Howe19 1991; Kushi22 1992; Mills24 1989; van den Brandt^109 1993; Wolk28 1998)	351,825/7377Pooled analysis	Fruits and vegetables	Continuous variable; increase in 100 g/d	1.00 (0.98–1.01)	Y	Y	Y	Age at menarche, interaction between parity and age at birth of first child, oral contraceptive use, history of benign breast disease, menopausal status at follow-up, postmenopausal hormone use, smoking status, education, BMI and menopausal status interaction, height, alcohol intake, energy intake
		Fruits	Continuous variable; increase in 100 g/d	1.00 (0.97–1.02)
		Vegetables	Continuous variable; increase in 100 g/d	1.00 (0.97–1.02)
		Fruits and vegetables	H vs 1 quartile (varied by study)	0.93 (0.86–1.00)
		Fruits	H vs 1 quartile (varied by study)	0.93 (0.84–1.02)
		Vegetables	H vs 1 quartile (varied by study)	0.96 (0.89–1.04)

Dietary antioxidants

Eleven cohort studies have been conducted on dietary vitamin A, vitamin C, vitamin E, and/or beta-carotene intake and the incidence of breast cancer (Table 7).16, 18, 32, 33, 35, 36, 44–48 There was no consistent association between any of these antioxidants and breast cancer incidence among these studies. In the Danish Diet Cancer and Health Cohort an inverse association was observed for vitamin E intake among postmenopausal women (RR = 0.59; 95% CI, 0.37–0.95).48

Table 7. Studies of Dietary Vitamin A, Vitamin C, Vitamin E, β Carotene and Breast Cancer Incidence

Author, year, country, parent cohort	Size of cohort/no. of cases; years followed	Type of antioxidant	Comparison of highest vs lowest quantile (quantity of intake)	Relative risk(95% CI)	Adjustment for confounding variables
					BMI	Family predisposition	Other variables
CNBSS indicates Canadian National Breast Screening Study; CTS, California Teacher's Study; IWHS, Iowa Women's Health Study; NCS, Netherlands Cohort Study; NHS, Nurses' Health Study; NHS II, Nurses' Health Study II; SMSC, Swedish Mammography Screening Cohort. * The effect of the variable on the measure of association was tested but the variable was not included in the final model.
Premenopausal and postmenopausal combined
Rohan32 1993	1182/519	Dietary vitamin A	H vs 1 quintile (14,136 vs 6133 IU/d)	0.80 (0.55–1.17)	N	Y	Energy intake, menarche, surgical menopause, age at first live birth, education, history of benign breast disease, smoking, alcoholMatched on screening center, date of enrollment
Canada	Nested case-control	Dietary β carotene	H vs 1 quintile (8441 vs 3446 IU/d)	0.77 (0.53–1.10)
CNBSS	1982–1987	Dietary vitamin C	H vs 1 quintile (220 vs 101 mg/d)	0.88 (0.62–1.26)
		Dietary vitamin E	H vs 1 quintile (25 vs 12 mg/d)	0.96 (0.63–1.45)
Michels47 2001Sweden	59,039/12711987/1990–1997	Dietary β carotene	H vs 1 quintile (median daily energy adjusted intake 5.10 vs 0.97 mg)	0.94 (0.78–1.14)	Y	Y	Height, education, parity, age a first birth, total caloric intake, alcohol, fiber, monounsaturated fatty acid
SMSC		Dietary vitamin C	H vs 1 quintile (median daily energy adjusted intake 109.7 vs 30.7 mg)	1.01 (0.84–1.22)
		Dietary vitamin E	H vs 1 quintile (median daily energy adjusted intake 9.3 vs 3.8 mg)	0.83 (0.60–1.14)
Terry45 2002CanadaCNBSS	56,837/1458 (5422 subcohort)1980/1985–1993 (average 9.5 y)	Dietary β carotene	H vs l quintile (energy adjusted median 9832± 3817 vs 2205 ± 610 μg/d)	1.01 (0.7–1.33)	Y	Y	Screening center, allocation, smoking, physical activity, education, history of benign breast disease, menarche, parity, menopausal status, OC, HRT, practice breast self examination, multivitamin use, total energy intake, alcohol, dietary fiber, folate, calcium
Horn-Ross18 2002	11,526/711	Dietary β carotene	H vs l quintile (80^th 4,652 vs 20^th 1465 μg/d)	1.1 (0.9–1.4)	Y	Y	Race, daily caloric intake, menarche, nulliparity, age at first pregnancy, physical activities, interaction term for menopausal status
USA	1995/1996–1998	Dietary vitamin C	H vs l quintile (80^th 653 vs 20^th 390 μg/d)	1.1 (0.8–1.3)
CTS		Dietary vitamin E	H vs l quintile (80^th 204 vs 20^th α-TE/d)	1.1 (0.9–1.4)
Cho44 2003	90,655/714	Dietary β carotene	H vs l quintile (median 7701 vs 1675) μg/d)	0.96 (0.75–1.22)	Y	Y	Smoking, height, parity and age at first birth, age at menarche, history of benign breast disease, oral contraceptive use, menopausal status, alcohol intake, energy, animal fat
USA	1991–1999	Dietary vitamin A	H vs l quintile (median 17801 vs 4895 IU/d)	0.92 (0.72–1.17)
NHSII		Dietary vitamin C	H vs l quintile (200 vs 69 mg/d)	1.30 (1.00–1.69)
		Dietary vitamin E	H vs l quintile (10 vs 6 IU/d)	1.17 (0.92–1.50)
Premenopausal women
Zhang36 1999	23,808/784	Dietary β carotene	H vs 1 quintile (median 7609 vs 1683 μg/d)	0.84 (0.67–1.05)	Y at age 18	Y	Length of follow up, total energy intake, parity, age at first birth, menarche, history of benign breast disease, alcohol, weight change from age 18, height
USANHS	1980–1994	Dietary vitamin A	H vs 1 quintile (median 17,073 vs 5,293 mg/d)	0.82 (0.65–1.04)
		Dietary vitamin C	H vs 1 quintile (median 202 vs 70 mg/d)	1.01 (0.81–1.26)
		Dietary vitamin E	H vs 1 quintile (median 10 vs 5 IU/d)	0.81 (0.4–1.02)
Postmenopausal women
Shibata33 1992	11,580/219	Dietary β carotene	H vs l tertile (≥9800 vs <4800 μg/d)	0.79 (0.57–1.10)	N	N	Smoking
USA	1981–1989	Dietary vitamin C	H vs l tertile (≥225 vs <155 mg/d)	0.86 (0.63–1.18)
Graham16 1992USA	18,586/3591980–1987	Vitamin A (not clear if only from diet)	H vs l quintile (limits: 513–3333 vs 0–174 1000 IU/mo)	0.96 (0.68–1.34)	Y*	N	EnergyParity, age at first pregnancy, total calorie intake*
NY State Cohort		β carotene (not clear if only from diet)	H vs l quintile (limits: 347-2030 vs 0–115 1000 IU/mo)	0.89 (0.63–1.26)
		Vitamin E (not clear if only from diet)	H vs l quintile (limits: 278–2036 vs 30–130 mg/mo)	0.86 (0.61–1.21)
		Vitamin C (not clear if only from diet)	H vs l quintile (limits: 79–498 vs 0–34 100 mg/mo)	0.81 (0.59–1.12)
Kushi103 1995USA	34,387/8791986–1992	Dietary vitamin A	H vs 1 quintile (limits: ≥20,343 vs <7254 IU/d)	1.15 (0.85–1.56)	Y	Y	Energy intake, menarche, age menopause, age first birth, parity, BMI at 18, history of benign breast disease, alcohol, education
IWHS		Dietary carotenoid	H vs 1 quintile (limits: ≥ 13,470 vs <4,426 IU/d)	1.17 (0.87–1.56)
		Dietary vitamin C	H vs 1 quintile (limits: ≥392 vs <112 IU/d)	1.06 (0.77–1.47)
		Dietary vitamin E	H vs 1 quintile (limits: ≥35.66 vs <5.66 IU/d)	1.08 (0.74–1.58)
Verhoeven35 1997Netherlands	62,573/650 (1716 subcohort)1986–1990	Dietary β carotene	H vs l quintile (median 0.719 vs 0.197 mg/d)	1.01 (0.72–1.42)	N	Y	Energy intake, history of benign disease, menarche, age at menopause, age at first birth, parity
NCS		Vitamin C (not clear if only from diet)	H vs l quintile (median 165.3 vs 58.6 mg/d)	0.77 (0.55–1.08)
		Dietary Vitamin E	H vs l quintile (median 19.82 vs 5.96 mg/d)	1.25 (0.85–1.85)
Zhang36 1999USA	59,426/19131980–1994	Dietary β carotene	H vs 1 quintile (median 7609 vs 1683 μg/d)	0.94 (0.81–1.09)	Y at age 18	Y	Length of follow up, total energy intake, parity, age at first birth, menarche, history of benign breast disease, alcohol, weight change from age 18, height, age at menopause, HRT
NHS		Dietary vitamin A	H vs 1 quintile (median 17,073 vs 5293 mg/d)	1.03 (0.89–1.19)
		Dietary vitamin C	H vs 1 quintile (median 202 vs 70 mg/d)	1.06 (0.91–1.22)
		Dietary vitamin E	H vs 1 quintile (median 10 vs 5 IU/d)	0.96 (0.83–1.11)
Nissen48 2003Denmark	29,875/418 (394 controls)Nested case-control	Dietary vitamin A	H vs second l (limits: >4000 vs 800–2000 mcg/d)	1.29 (0.81–2.05)	Y	N	Total intake of other two vitamins, parity, age at first birth, history of benign breast disease surgery, education, duration of HRT, alcohol intakeMatched on postmenopausal status, use of HRT at baseline, age at baseline
Danish Diet, Cancer and Health	1993–1997 (mean 4.7 y)	Dietary vitamin C	H vs second l (limits: >300 vs 60–150 mg/d)	1.69 (1.12–2.57)
		Dietary vitamin E	H vs second l (limits: >25 vs 10–15 mg/d)	0.59 (0.37–0.95)

Blood antioxidants

The association between serum antioxidants vitamins, which are biomarkers of intake, and breast cancer incidence was considered in 8 nested case-control studies (Table 8).65–72 For serum levels of beta-carotene, the Finnish Social Institution's Mobile Clinic Health Survey (FSIIMCHS) reported a decreased risk of breast cancer among women with low serum levels (OR = 0.3; 95% CI, 0.1–1.0),68 whereas the New York University Women's Health Study found an OR of 2.21 (95% CI, 1.29–3.79) among women in the lowest quintile71 and Sato et al.69 reported an OR of 0.41 (95% CI, 0.22–0.79) among women in the highest quintile of serum beta-carotene compared with those in the lowest, but in several other studies no association was found. High serum levels of alpha-tocopherol (vitamin E) were inversely associated with breast cancer risk in 2 studies.66, 72

Table 8. Studies of Biomarkers of Vitamin A, Vitamin C, Vitamin E, β carotene Intake and Breast Cancer Incidence

Author, year, country, parent cohort	Size of cohort/no. of cases; years followed	Type of biomarker	Comparison of highest vs lowest quantile (quantity of intake)	Odds ratio (95% CI)	Adjustment for confounding variables
					BMI	Family predisposition	Other variables
FSIIMCHS indicates Finnish Social Institution's Mobile Clinic Health Survey; MONICA, Monitoring of Trends in Cardiovascular diseases; MSP, Mammary Screening Project; NHS, Nurses' Health Study; NYUWHS, New York University Women's Health Study; VIP, Vasterbotten Intervention Project. * The effect of the variable on the measure of association was tested but the variable was not included in the final model.
Premenopausal and postmenopausal combined
Wald72 1984UK	5,004/39 (78 controls)Nested case-control	Serum β carotene	H vs 1 quintile (mean serum level in controls 50 μg/L)	0.54 (P not significant)	N	Y matched	Duration of storage of serum samples
	1968/1975–1982	Serum vitamin E	H vs 1 quintile (mean serum level in controls 6.0 mg/L)	0.5 (P < .01)			Matched on parity, previous history of benign breast disease, menopausal status menopausal status (if premenopausal: day of menstrual cycle, if post-menopausal: number of years of menopause)
Knekt67 1988Finland	15,093/67 (123 controls)Nested case-control	Serum vitamin E (α tocopherol)	4 H quintiles vs l (≥7.91 vs 7.9 mg/L)	1.03 (P < .05)	Y	N	Smoking, Cholesterol, hematocrit, parity, occupationMatched on municipality
FSIIMCHS	1968/1971–1977 (mean 8 y)
Knekt68 1990	36,265 men and women/67 (123 controls)	Serum β carotene	L vs h quintile (≤40 vs >180 μg/L)	0.3 (0.1–1.0)	Y*	N	Smoking
Finland	Nested case-control						Matched on sex and municipality
FSIIMCHS	1968/1971–1977 (mean 8 y)
Dorgan65 1998USA	6426/105 (203 controls)Nested case-control	Serum β carotene	Quartile H vs l (0.69–2.20 vs ≤0.29 μmol/L)	1.1 (0.5–2.4)	Y	Y*	Serum cholesterol, pack cigarettes smoked/d
Breast Cancer Serum Bank in Columbia, Missouri	1977/1987–1989 (median 2.7 y)	Serum α tocopherol	Quartile H vs l (31.33–107.69 vs ≤21.59 μmol/L)	1.2 (0.5–2.8)			Matched on date of blood donation, diagnosis of benign breast diseases during the 2 years prior to enrollment
							Height, parity, menarche, menopausal status, exogenous estrogen use, education*
Toniolo71 2001	14,275/270 (270 controls)	Serum β carotene	L vs h quintile (no information given)	2.21 (1.29–3.79)	Y*	Y	Age at first full term pregnancy, history of benign breast disease, total cholesterolMatched on menopausal status and date of last menstruation, date of blood sampling
USA	Nested case-control
NYUWHS	1985/1991–1995
	1985/1991–1995						Menarche and parity*
Hulten66 2001SwedenVIP/MSP/MONICA	47,194/201390 (controls)Nested case-control	Plasma β carotene	H vs l quartileVIP + MONICA (limits for controls: 0.51 vs 0.25 μ/L)	0.8 (0.5–1.4)	Y	N	Plasma total cholesterol, triglycerides
							Age at menarche, parity, age at full term pregnancy, HRT, cotinine, hours of fasting
			MSP (limits for controls: 0.55 vs 0.6 μg/L)
	VIP: 1985–1997 (median: 1.9 y)	Plasma α tocopherol	H vs l quartile	1.3 (0.6–2.7)
	MONICA: 1986/1994–1997 (median 2.4 y)MSP:1995–1997 (median <1 mo)		VIP + MONICA (limits for controls: 24.89 vs 18.93 μg/L)
			MSP (limits for controls: 30.03 vs 22.49 μ/L)
Sato69 2002	1974–1994:23,850/244 (244 controls)	Serum β carotene	H vs 1 quintile (≥22 vs <0.9 μg/L)	1974: 0.41 (0.22–0.79)	Y*	Y*	County, race, menopausal status (if premenopausal: last menstrual cycle), month and year of blood donationAge at first birth, menarche, alcohol, smoking, duration of lactation, education, time since last meal, cholesterol*
USA	1989 – 1994: 25,080/115 (115 controls)			1989: 0.62 (0.27–1.42)
Blood Collection Campaign	Nested case-control	Serum α tocopherol	H vs 1 quintile (≥0.32 vs <0.15 mg/dL)	1974: 0.94 (0.52–1.73)
				1989: 0.67 (0.28–1.62)
Tamimi70 2005USA	32,826/969969 (controls)	Plasma β carotene	H vs 1quintile (median 1.15 vs 0.18 μmol/L)	0.73 (0.53–1.02)	Y at age 18	Y	Age at menopause, weight gain since age 18 years, age at menarche, history of benign breast cancer, age at first birth, parity, HRT, alcohol consumptionMatched on menopausal status, postmenopausal hormone use, month, time of day and fasting status at blood collection
NHS	Nested case-control1989/1990–1998 (median 4 y)	Plasma α tocopherol	H vs 1quintile (median 46.94 vs 16.42 μmol/L)	0.79 (0.57–1.08)
Premenopausal women
Hulten66 2001Sweden	47,194 (all)/57 (93 controls)Nested-case control	Plasma β carotene	H vs l quartile VIP 1 MONICA (limits for controls: 0.51 vs 0.25 μg/L)	1.6 (0.5–5.4)	Y	N	Plasma total cholesterol and triglycerides, age at menarche, parity, age at full term pregnancy, HRT, cotinine *
VIP/MONICA	VIP: 1985–1997 (median: 1.9 years)MONICA: 1986/1994–1997 (median 2.4 y)	Plasma α tocopherol	H vs l quartileVIP 1 MONICA (limits for controls : 24.89 vs 18.93 μg/L)	0.5 (0.0–6.6)
Postmenopausal women
Hulten66 2001Sweden	47,194 (all)/67 (109 controls)Nested-case control	Plasma β carotene	H vs l quartile VIP 1 MONICA (limits for controls: 0.51 vs 0.25 μg/L)	0.7 (0.2–1.9)	Y	N	Plasma total cholesterol and triglycerides
							Age at menarche, parity, age at full term pregnancy, HRT, cotinine, hours of fasting*
VIP/MONICA/MSP	VIP: 1985–1997 (median: 1.9 y)	Plasma β carotene	H vs l quartile	0.4 (0.1–1.2)
	MONICA: 1986/1994–1997 (median 2.4 y)MSP: 1995–1997 (median: <1 mo)		MSP (limits for controls: 0.55 vs 0.26 μg/L)
		Plasma α tocopherol	H vs l quartile	2.9 (0.7–12.4)
			VIP+ MONICA (limits for controls: 24.89 vs 18.93 μg/L)
			H vs l quartile	2.3 (0.3–17.2)
			MSP (limits for controls: 30.03 vs 22.49 μg/L)

Carbohydrate intake, glycemic index, and glycemic load

Carbohydrate intake may influence breast cancer risk by affecting insulin resistance and plasma levels of insulin and glucose. Eleven cohort studies have been conducted on the association between carbohydrate intake, glycemic index, glycemic load, and the incidence of breast cancer (Table 9).11, 12, 18, 21, 22, 39–43, 92 Most studies did not reveal an association between these measures of carbohydrate intake and breast cancer incidence. The Finnish FSIIMCHS study reported an RR of 0.5 (95% CI, 0.25–1.00) among women in the highest tertile of carbohydrate intake compared with the lowest, but carbohydrates in Finland are qualitatively different from those regularly consumed in Asia or South America, as they are higher in fiber.21 Among postmenopausal women, the Italian ORDET study also reported an inverse association (highest vs lowest tertile of carbohydrate intake: RR = 0.42; 95% CI, 0.18–0.95),11 whereas results from 3 studies indicated an increased incidence among women with high carbohydrate intake: In the Rancho Bernardo study the RR per 66 g increase in carbohydrates was 1.93 (95% CI, 1.18–3.16),12 in the Nurses' Health Study, the RR associated with the highest quintile of glycemic index was 1.15 (95% CI, 1.02–1.30) compared with the lowest quintile of intake,40 and in the Canadian National Breast Screening Study the RR for breast cancer in the highest quintile of glycemic index was 1.87 (95% CI, 1.18–2.97).43

Table 9. Studies of Dietary Carbohydrates Glycemic Index, Glycemic Load, and Breast Cancer Incidence

Author, year, country, parent cohort	Size of cohort/no. of cases; years followed	Type of carbohydrate	Comparison of highest vs lowest quantile (quantity of intake)	Relative risk(95% CI)	Adjustment for confounding variables
					BMI	Family predisposition	Other variables
CNBSS indicates Canadian National Breast Screening Study; CPS II, Cancer Prevention Study II Nutrition Cohort; CTS, California Teacher's Study; FSIIMCHS, Finnish Social Institution's Mobile Clinic Health Survey; IWHS, Iowa Women's Health Study; NHS, Nurses' Health Study; NHS II, Nurses' Health Study II; ORDET, Hormone and Diet Etiology of Breast Cancer; WHS, Women's Health Study. * Includes both premenopausal and postmenopausal women, whereas cases are only either premenopausal or postmenopausal. † The variable was considered as a confounding variable but was found to be not significant.
Premenopausal and postmenopausal women combined
Knekt21 1990	3988/54	Carbohydrate	H vs 1 tertile (≥278 vs <208)	0.50 (0.25–1.00)	N	N	Fat, protein
Finland	1967–1986
FSIIMCHS
Horn-Ross18 2002USA	111,526/7111995/1996–1998	Carbohydrate	H vs l quintile (90^th percentile 240 vs 20^th percentile 128 g/d)	0.8 (0.5–1.2)	Y	Y	Race/ethnicity, daily caloric intake, menarche, parity, age at first pregnancy, menopausal status, physical activity
CTS
Cho92 2003USA	90,655/7141991–1999	Cumulative average dietary carbohydrate	H vs l quintile (median 59.4% vs 41.2% energy)	0.89 (0.63–1.26)	Y	Y†	Smoking, height, parity and age at first birth, menarche, history of benign breast disease, OC, menopause, alcohol, energy intake, animal fat intake
NHS II		Cumulative average glycemic index	H vs l quintile (median 82% vs 70% energy)	1.05 (0.83–1.33)
		Cumulative average glycemic load	H vs l quintile (median 211 % vs 138% energy)	1.06 (0.78–1.45)
Higginbotham39 2004	39,876/946	Glycemic load	H vs l quintile (median 143 vs 92 g/d)	1.01 (0.76–1.35)	Y	Y	Alcohol, smoking, menarche, age first pregnancy, number pregnancy, OC, HRT, physical activity, total energy, energy-adjusted total fat, energy adjusted total fiber, energy adjusted total folate
USA	1993/1995 (mean 6.8 y)
WHS		Glycemic index	H vs l quintile (median 55 vs 50 g/d)	1.03 (0.84–1.28)
Holmes40 2004USA	88,678/40921980–1998	Carbohydrate	H vs l quintile (median 240 vs 159 energy-adjusted g/d)	0.97 (0.87–1.08)	Y	Y	2-year time period, total energy intake, alcohol, parity, age at first birth, height, history of benign breast disease, menarche, HRT
NHS		Glycemic load	H vs l quintile (median 81 vs 69 energy-adjusted g/d)	1.08 (0.97–1.19)
		Glycemic index	H vs l quintile (median 186 vs 116 energy-adjusted g/d)	0.99 (0.89–1.10)
Silvera43 2005	49,111/1450	Carbohydrate	H vs l quintile (>249 vs <143 g/d)	0.93 (0.70–1.22)	Y	Y	Menopausal status, alcohol, use of HRT, use of oral contraceptive, parity, age at menarche, age at first birth, history of benign breast disease, energy intake, energy-adjusted total fiber, study center, randomization group
Canada	1980/1985–1998/2000 (mean 16.6 y)	Glycemic load	H vs l quintile (>175 vs <119 g/d)	0.95 (0.79–1.14)
CNBSS		Glycemic index	H vs l quintile (>96 vs <60 g/d)	0.88 (0.63–1.22)
Premenopausal women
Higginbotham39 2004	39,876* /354	Glycemic load	H vs l quintile (median 143 vs 92 g/d)	1.27 (0.79–2.03)	Y	Y	Alcohol, smoking, menarche, age first pregnancy, number pregnancy, OC, HRT, physical activity, total energy, energy-adjusted tot fat, energy adjusted total fiber, energy adjusted total folate
USA	1993/95 (mean 6.8 y)
WHS		Glycemic index	H vs l quintile (median 55 vs 50 g/d)	1.29 (0.92–1.81)
Holmes40 2004	53,891/852	Carbohydrate	H vs l quintile (median 240 vs 159 energy-adjusted g/d)	0.98 (0.78–1.23)	Y	Y	2-year time period, total energy intake, alcohol, parity ad age at first birth, height, history of benign breast disease, menarche, HRT
USA	1980–1998
NHS		Glycemic load	H vs l quintile (median 81 vs 69 energy-adjusted g/d)	1.02 (0.82–1.28)
		Glycemic index	H vs l quintile (median 186 vs 116 energy-adjusted g/d)	0.87 (0.70–1.12)
Silvera43 2005	400,673 person y/670	Glycemic load	H vs l quintile (>169 vs <125 g/d)	0.96 (0.76–1.22)	Y	Y	Alcohol, use of oral contraceptive, parity, age at menarche, age at first birth, history of benign breast disease, energy intake, energy-adjusted total fiber, study center, randomization group
Canada	1980/85–1998/2000 (mean 16.6 y)	Glycemic index	H vs l quintile (>92 vs <63 g/d)	0.78 (0.52–1.16)
CNBSS
Postmenopausal women
Kushi22 1992	34,388/459	Carbohydrate	H vs 1 quartile (median 252.7 vs 181.0 g)	0.90 (0.70–1.21)	Y	Y	Total energy intake, age at menarche, age at menopause, age at first birth, waist-to-hip ratio, history of benign breast disease, BMI at age 18, alcohol intake
USA	1986–1989
IWHS
Barrett-Connor12 1993	575/15	Carbohydrate	Continuous variable, 66 g increase	1.93 (1.18–3.16)	Y	N	Energy, age at menopause, parity, alcohol, total calories, total carbohydrates, total protein
USA	1972–1987
Sieri11 2002	3367/56 (214 controls)	Carbohydrate	H vs 1 tertile (≥217.6 vs 190.2 g)	0.42 (0.18–0.95)	Y	N	Energy intake, parity, place of birth, level of educationMatched on daylight-saving period, recruitment center, recruitment data
ItalyORDET	Nested case-control 1987–1992
Jonas41 2003USA	63,307/14421992–1997/1998	Glycemic load	H vs l quintile (mean score 147 vs 83)	0.90 (0.76–1.08)	Y	Y	Menarche, age menopause, numberof live births, HRT, OC, benignbreast cysts, education, adultweight gain, location of bodyweight gain, height, physicalactivity, energy, diethylstilbestroluse, alcohol, race, smoking, totalfat, protein, total fiber, fiber fromgrain or vegetable, vegetable, fruits†
CPS II Nutrition Cohort		Glycemic index	H vs l quintile (mean score 85 vs 65)	1.03 (0.87–1.22)
Higginbotham39 2004USA	39,876*/5891993/1995 (mean 6.8 y)	Glycemic load	H vs l quintile (median 143 vs 92 g/d)	0.90 (0.63–1.31)	Y	Y	Alcohol, smoking, menarche, age first pregnancy, number pregnancy, OC, HRT, physical activity, total energy, energy-adjusted total fat, energy adjusted total fiber, energy adjusted total folate
WHS		Glycemic index	H vs l quintile (median 55 vs 50 g/d)	0.89 (0.67–1.17)
Holmes40 2004USA	76,200/29241980–1998	Carbohydrate	H vs l quintile (median 240 vs 159 energy-adjusted g/d)	0.96 (0.84–1.09)	Y	Y	2-year time period, total energy intake, alcohol, parity ad age at first birth, height, history of benign breast disease, menarche, HRT
NHS		Glycemic load	H vs l quintile (median 81 vs 69 energy-adjusted g/d)	1.15 (1.02–1.30)
Nielsen42 2005Denmark	23,870/6341993/1997–2002 (mean 6.6 y)	Glycemic index	H vs l quintile (median 186 vs 116 energy-adjusted g/d)	1.03 (0.90–1.16)			Parity, age at first birth, education, use of HRT, duration of HRT, alcohol intake
Diet, Cancer and Health Study		Carbohydrate	50 g increment in daily intake	1.06 (0.97–1.16)	Y	N
		Glycemic load	10 units increment for daily intake	0.94 (0.80–1.10)
		Glycemic index	100 units increment for daily intake	1.04 (0.90–1.19)
Silvera43 2005Canada	300,048 person y/5751980/1985–1998/2000 (mean 16.6 y)	Glycemic loadGlycemic index	H vs l quintile (>169 vs <125 g/d)H vs l quintile (>92 vs <63 g/d)	1.08 (0.82–1.41)1.87 (1.18–2.97)	Y	Y	Alcohol, use of HRT, use of oral contraceptive, parity, age at menarche, age at first birth, history of benign breast disease, energy intake, energy-adjusted total fiber, study center, randomization group
CNBSS

Dairy and vitamin D consumption

Milk consumption induces a rise in endogenous insulin-like growth factor I levels, at least in the short term, which may affect breast cancer risk.93 Because dairy is fortified with vitamin D in some countries, like the US, but not in many others, like most European countries, the association between milk, vitamin D, and breast cancer may be more difficult to disentangle. Furthermore, bovine growth hormones are used in the US, but not in Europe.

Twelve studies have been conducted on the association between consumption of diary products, vitamin D, and breast cancer incidence (Table 10).10, 13, 14, 24, 25, 27, 31, 49–52 Among studies that considered both pre- and postmenopausal women, the results were inconsistent. The New York University Women's Health Study found an inverse association between breast cancer incidence and intake of diary products (highest vs the lowest quintile: RR = 0.59; 95% CI, 0.35–0.99).25 The NHSS reported an RR of 2.91 (95% CI, 1.38–6.14) for breast cancer among women in the highest quintile of whole milk consumption vs the lowest quintile.14 In the Finnish FSIIMCHS cohort, women in the highest tertile of intake of all dairy products had an RR of breast cancer of 0.42 (95% CI, 0.23–0.78) compared with those in the lowest tertile; the respective RR for milk consumption was 0.42 (95% CI, 0.24–0.74).50 Women in the highest quintile of high-fat dairy food consumption had a relative risk of breast cancer of 1.36 (95% CI, 1.07–1.75) in NHS II.13 No consistency in associations was found according to differences in milk fortification practices.

Table 10. Studies of Dairy Product Consumption, Vitamin D Intake, and Breast Cancer Incidence

Author, year, country, parent cohort	Size of cohort/no. of cases; years followed	Type of dairy	Comparison of highest vs lowest quantile (quantity of measure)	Relative risk (95% CI)	Adjustment for confounding variables
					BMI	Family predisposition	Other variables
FSIIMCHS indicates Finnish Social Institution's Mobile Clinic Health Survey; NHS, Nurses' Health Study; NHS II, Nurses' Health Study II; NHSS, Norwegian National Health Screening Services; NOWAC, Norwegian Women and Cancer Study; NYUWHS, New York University Women's Health Study. * Includes both premenopausal and postmenopausal women, while cases are only either premenopausal or postmenopausal. † The effect of the variable on the measure of association was tested but the variable was not included in the final model.
Premenopausal and postmenopausal women combined
Mills24 1989	20,341/215	Whole milk	H vs 1 tertile (≥daily vs none)	0.94 (0.66–1.33)	Y	Y	Age at first live birth, age at menarche, menopausal status, history of benign breast disease, education
USA	1976–1982
Seventh Day Adventist Study
Ursin52 1990	2679/29 1967 (11.5 y follow-up)	Milk consumption	H vs 1 tertile (≥2 vs 1 glass/d)	1.48 (Ptrend = .40)	N	N	Place of residence
Norway
Toniolo25 1994	14,291/180 (829 controls)	Dairy products	H vs l quintile (mean cases 285 vs 37 g/d)	0.59 (0.35–0.99)	Y†	Y†	Energy adjusted Height, menopausal status (matched), date of enrollment (matched), age at menarche, age at first full term pregnancy, number of full-term pregnancies, history of benign breast disease, race, religion†
USA	Nest case-control
NYUWHS	1985–1991
Gaard14 1995	24,897/248	Milk	H vs 1 quintile (≥5 vs 1 glass/d)	1.71 (0.86–3.38)	N	N
Norway		1977/1983–1990 (mean 10.4 y)	Whole Milk	H vs 1 quintile (≥5 vs 1 glass/d)	2.91 (1.38–6.14)
NHSS
Byrne10 1996	6156/53	Whole milk	H vs 1 (>7 vs ≤7 servings/wk)	0.5 (0.1–2.1)	Y†	Y†	No of breast biopsies, alcohol, physical activity, oral contraceptives, age at menarche, menopausal hormones, race, education, region of country, area of residence, family income†
USA NHANES 1	1982/1984–1986/1987
NHANES 1
Knekt50 1996	4697/88	All dairy products	H vs 1 tertile (no information given)	0.42 (0.23–0.78)	N	N	N
Finland	1966/1972 (25 y follow-up)
FSIIMCHS		Milk	H vs 1 tertile (≥620 vs <370 g/d)	0.42 (0.24–0.74)
Key31 1999	34,759/427	Milk	H vs 1 tertile (≥5 vs ≤1 servings/wk)	0.87 (0.66–1.16)	N	N	Calendar period, city, age at time of bombing, radiation dose
Japan	1969/1970–1993; 1979/1980–1993
Radiation Effects Research Foundation's Life Span Study
John53 1999	4747/179	Dietary vitamin D	H vs l tertile (≥200 vs <100 IU)	0.85 (0.59–1.24)	Y	Y†	Education, age at menarche, age at menopause, alcohol consumption, physical activity, calcium intake, income, nulliparity, sunlight exposure, age at birth†
USA	1971–1992 (mean 17.3 y)
NHANES 1
Cho13 2003	90,655/714 1991–1999	Total dairy foods	H vs 1 quintile (median 4.0 vs 0.7 servings/d)	1.03 (0.79–1.36)	Y	Y	Smoking, height, parity and age at first birth, age at menarche, history of benign breast disease, oral contraceptive use, menopausal status, alcohol intake, energy
USA
NHS II		High-fat dairy foods	H vs 1 quintile (median 2.2 vs 0.2 servings/d)	1.36 (1.07–1.75)
Premenopausal women
Hjartaker49 2001	48,844/317	Milk	H vs 1 quartile (≥3 glasses/d vs none)	0.56 (0.31–1.01)	Y	Y	Age at menarche, number of children, age at first birth, current use of oral contraceptives, years of education, level of physical activity, alcohol consumption
Norway	1991/1992 (mean 6.2 y)
NOWAC		Milk (adult and child)	H vs 1 tertile (no information given)	0.51 (0.27–0.96)
Shin51 2002	88,691*/827	Total dairy	H vs 1 tertile (≥3 vs ≤1 servings/d)	0.80 (0.63–1.03)	Y	Y	Time period, physical activity, history of benign breast disease, height, weight change since age 18, age at menarche, parity, age at first birth, alcohol intake, total energy intake, total fat intake, glycemic index, beta carotene intake, total active E intake, vitamin D intake
		USA	1980–1996	Total milk	H vs 1 quartile (>1 8-oz glass/d vs ≤3 8 oz glasses/mo)	0.78 (0.60–1.01)
		NHS
				Whole milk	H vs 1 quintile (>1 8 oz glass/d) vs never)	0.87 (0.59–1.28)
		Dietary vitamin D	H vs 1 quintile (>300 vs ≤75 IU/d)	0.66 (0.43–1.00)
Postmenopausal women
Mills24 1989	20,3411*/171	Whole milk	H vs 1 tertile (≥daily vs none)	0.98 (0.66–1.45)	Y	Y	Age at first live birth, age at menarche, menopausal status, history of benign breast disease, education
USA	1976–1982
Seventh Day Adventist Study
Voorrips27 2002	1598/941 (mean 6.3 y)	Milk and milk products	H vs 1 quintile (median 532 vs 72 g/d)	0.91 (0.67–1.24)	Y	Y	History of benign breast disease, age at menarche, age at menopause, oral contraceptive use, parity, age at first childbirth, education, alcohol use, current smoking, and energy intake
		Netherlands
Netherlands Cohort Study		Whole milk and products	H vs 1 quintile (median 232 vs 0 g/d)	0.90 (0.66–1.22)
Shin51 2002	88,691*/2345	Total dairy	H vs 1 tertile (≥3 servings/d vs ≤1 servings/d)	0.97 (0.85–1.12)	Y	Y	Time period, physical activity, history of benign breast disease, height, weight change since age 18, age at menarche, parity, age at first birth, alcohol intake, total energy intake, total fat intake, glycemic index, B-carotene intake, age at menopause, post menopausal hormonal use, total active E intake
USA	1980–1996
NHS		Total milk	H vs 1 quartile (>1 8-oz glass/d vs ≤3 8-oz glasses/mo)	1.01 (0.87–1.17)
		Whole milk	H vs 1 quintile (>1 8 oz glass/d vs never)	0.87 (0.69–1.10)
		Dietary vitamin D	H vs 1 quintile (>300 vs ≤75 IU/d)	1.06 (0.85–1.34)

Two studies focused on the association between dairy products and breast cancer incidence among premenopausal women.49, 51 In a study from Norway the risk of breast cancer associated with highest vs the lowest tertile was 0.56 (95% CI, 0.31–1.01) for milk consumption in early adult life and 0.51 (95% CI, 0.27–0.96) for milk consumption during childhood and early adulthood.49 In the Nurses' Health Study, the RR of breast cancer associated with the highest quartile of milk consumption compared with the lowest quartile was 0.78 (95% CI, 0.60–1.01).51 The RR of breast cancer associated with the highest quintile of dietary vitamin D compared with the lowest quintile was 0.66 (0.43–1.00). In the 3 cohort studies that were restricted to postmenopausal women no statistically significant association between milk consumption and breast cancer incidence emerged.24, 27, 51

In a pooled analysis of cohort studies on the consumption of dairy products and breast cancer incidence the overall RR of breast cancer associated with the highest quartile of total dairy fluid vs the lowest quartile was 0.93 (95% CI, 0.84–1.03) (Table 11).73

Table 11. Pooled Analysis of Studies on Dairy Consumption and Breast Cancer Incidence

Author, year, parent cohort	Size of cohort/no. of cases menopausal status	Type of dairy measure	Comparison of highest vs lowest quantile (quantity of measure)	Relative risk (95% CI)	Adjustment for confounding variables
					Age	BMI	Family predisposition	Other variables
Missmer73 2002 (includes Toniolo25 1994; Graham16 1992; Holmes17 1999; Howe19 1991; Kushi22 1992; Mills24 1989; van den Brandt^109 1993; Wolk28 1998)	351,041/7379	Total dairy fluids	H vs 1 quartile	0.93 (0.84–1.03)	Y	Y	Y	Age at menarche, interaction between parity and age at first birth, oral contraceptive use, history of benign breast disease, menopausal status, the interaction of body mass index and menopausal status, postmenopausal hormone use, smoking status, education, height, alcohol intake, total energy intake
		Total dairy solids	H vs 1 quartile	(0.93–1.09)
		Whole milk	100 g/d increment	0.99 (0.96–1.01)
	Premenopausal Women	Total dairy fluids	100 g/d increment	1.00 (0.98–1.01)
	Postmenopausal women	Total dairy solids	100 g/d increment	1.05 (0.94–1.16)
		Total dairy fluids	100 g/d increment	0.96 (0.90–1.02)
		Total dairy solids	100 g/d increment	0.87 (0.68–1.11)

Soy products and isoflavones

The chemical structure of isoflavones is similar to that of estrogen. Isoflavones bind to and activate the estrogen receptor, competing with estrogen. Both an increase and a decrease in risk of breast cancer associated with isoflavone intake are plausible. Five cohort studies considered the association between intake of soy products, isoflavones, and breast cancer incidence (Table 12).18, 31, 54–56 Three of these studies were conducted in Japan,31, 54, 56 where the mean intake of isoflavone is substantially higher than in North America or Europe. Mean intakes of the isoflavones genistein and daidzein in Japan has been reported to be 30–700 times the average intake in the US.94 The mean intake of genistein was 10 times higher among Japanese than Caucasian women living in the US in 1 study.95 The Japan Public Health Center-based Prospective Study on Cancer and Cardiovascular Disease found an RR of breast cancer of 0.46 (95% CI, 0.25–0.84) associated with the highest quartile of isoflavone intake compared with the lowest quartile.56 This RR was stronger among women with postmenopausal breast cancer (RR = 0.32; 95% CI, 0.14–0.71). In the California Teachers Study women in the highest quintile of the lignan secoisolariciresinol intake had an RR for breast cancer of 1.4 (95% CI, 1.0–1.8).18 Overall, evidence is inconclusive.

Table 12. Studies of Soy Consumption, Isoflavone Intake, and Breast Cancer Incidence

Author, year, country, parent cohort	Size of cohort/no. of cases; years followed	Type of phytoestrogen	Comparison of highest vs lowest quantile (quantity of intake)	Relative risk (95% CI)	Adjustment for confounding variables
					BMI	Family predisposition	Other variables
CTS, California Teacher's Study; EPIC, European Prospective Investigation into Cancer and Nutrition; JPHC, Japan Public Health Center-Based Prospective Study on Cancer and Cardiovascular Diseases. * Breast cancer mortality. † The effect of the variable on the measure of association was tested but the variable was not included in the final model.
Premenopausal and postmenopausal women combined
Hirayama54 1990	142,857/241*	Dietary soy bean paste soup	Daily vs nondaily	0.85 (0.68–1.06)	N	N
Japan	1965 (17 y follow-up)
Key31 1999	34,759/427	Dietary Tofu	H vs l tertile (≥5/wk vs <1/wk)	1.19 (0.62–2.29)	Y†	N	Radiation dose, city Hiroshima and Nagasaki, calendar period
Japan	1969/1970–1993; 1979/1980–1993
Radiation Effects Research Foundation's Life Span Study		Miso soup	H vs l tertile (≥5/wk vs <1/wk)	0.74 (0.43–1.28)			Age at bombing†
Horn-Ross18 2002	111,526/711	Genistein (isoflavone)	H vs l quintile (80^th percentile 1100 vs 20^th percentile 290 μg/d)	1.0 (0.7–1.3)	Y†	Y†	Race/ethnicity, daily caloric intake, menarche, parity, age at first pregnancy, menopause, physical activity
USA	1995/1996–1998
CTS		Daidzein (isoflavone)	H vs l quintile (80^th percentile 906 vs 20^th percentile 301 μg/d)	0.9 (0.7–1.2)
		Secoisolariciresinol (lignan)	H vs l quintile (80^th percentile 2343 vs 20^th percentile 821 μg/d)	1.4 (1.0–1.8)
Yamamoto56 2003	21,852/179	Miso soup	H vs l quartile (≥3 vs <1 cup/d)	0.60 (0.34–1.1)	Y†	Y†	Menarche, parity, menopause, age first pregnancy, smoking (active-passive) alcohol, physical activity, education, total energy, meat, fish, vegetable, fruit consumption
Japan	1990–1999	Soy food	H vs 1 tertile (Almost daily vs <2 times/wk)	0.81 (0.49–1.3)
JPHC
		Isoflavone	H vs l quartile (25.3 vs 6.9 mg/d)	0.46 (0.25–0.84)
							History of benign disease, exogenous hormone†
Keinan-Boker55 2004	15,555/280	Isoflavone	H vs l quartile (0.77 vs 0.19mg/d)	0.98 (0.65–1.48)	Y†	Y†	Age at first pregnancy, height, weight, parity, physical activity, oral contraceptive or HRT, marital status, education, daily energy intake, smoking, BMI, fat, fiber, vegetables consumption†
Netherlands	1993/1997–2001 (median 5.2 y)	Lignans	H vs l quartile (0.01vs 4.9 g/d)	0.70 (0.46–1.09)
EPIC
Premenopausal women
Yamamoto56 2003	93,628 person y/89	Isoflavone	H vs l quartile (25.3 vs 6.9 mg/d)	0.66 (0.25–1.7)	Y†	Y†	Menarche, parity, menopause, age first pregnancy, smoking (active-passive) alcohol, physical activity, education, total energy, meat, fish, vegetable, fruit consumption, history of benign disease, exogenous hormone†
Japan	1990–1999
JPHC
Postmenopausal women
Yamamoto56 2003	111,637 person y/87	Isoflavone	H vs l quartile (25.3 vs 6.9 mg/d)	0.32 (0.14–0.71)	Y†	Y†	Menarche, parity, menopause, age first pregnancy, smoking (active-passive) alcohol, physical activity, education, total energy, meat, fish, vegetable, fruit consumption†
Japan	1990–1999
JPHC
							History of benign disease, exogenous hormone

Green tea

Green tea components, such as epigallocatechin gallate, may have antiproliferative effects on tumor cells. Five cohort studies have considered the association between green tea and breast cancer incidence (Table 13),31, 57–59 as well as 1 pooled analysis58 and 1 meta-analysis74 (Table 14). None of these studies indicated an important association between green tea consumption and the incidence of breast cancer.

Table 13. Studies on Green Tea Consumption and Breast Cancer Incidence

Author, year, country, parent cohort	Size of cohort/no. of cases; years followed	Type of food	Comparison of highest vs lowest quantile (quantity of intake)	Relative risk (95% CI)	Adjustment for confounding variables
					BMI	Family predisposition	Other variables
* The effect of the variable on the measure of association was tested but the variable was not included in the final model.
Premenopausal and postmenopausal women combined
Key31 1999	34,759/427	Green tea	H vs l tertile (≥5 vs ≤1 times/d)	0.86 (0.62–1.21)	Y*	N	Radiation dose, city (Hiroshima and Nagasaki), calendar period, age at bombing
Japan	1969/1970–1993;
Radiation Effects Research	1979/1980–1993
Foundation's Life Span Study
Nagano57 2001	23,667/276	Green tea	H vs l tertile (≥5 vs ≤1 cup/d)	1.0 (0.67–1.6)	Y	N	City, radiation exposure, smoking, drinking, education level, calendar time
Japan	1981–1994
Life Span Study
Suzuki58 2004	14,409/103	Green tea	H vs l tertile (≥5 vs <1 cup/d)	1.17 (0.67–2.05)	Y	Y	Types of health insurance, age at menarche, menopausal status, age at first birth, parity, smoking, alcohol, frequencies of black tea and coffee
Japan	1984 – 1993
Suzuki58 2004	20,595/119	Green tea	H vs l tertile ≥5 vs <1 cup/d)	0.61 (0.26–1.06)	Y	Y	Types of health insurance, age at menarche, menopausal status, age at first birth, parity, smoking, alcohol, frequencies of black tea and coffee
Japan	1990–1999

Table 14. Meta-Analysis/Pooled Analysis of Studies of Green Tea Consumption and Breast Cancer Incidence

Author, year, country. parent cohorts	Size of cohort/no. of cases; type of study	Type of food	Comparison of highest vs lowest quantile (quantity of intake)	Relative risk(95% CI)	Adjustment for confounding variables
					BMI	Family predisposition	Other variables
* Includes both men and women.
Suzuki58 2004 Japan	35,004/222	Green tea	H vs l tertile (≥5 vs <1 cup/d)	0.84 (0.57–1.24)	Y	Y	Types of health insurance, age at menarche, menopausal status, age at first birth, parity, smoking, alcohol, frequencies of black tea and coffee
	Pooled analysis
Seely74 2005 (includes Key31 1999; Nagano57 2001; Suzuki58 2004)	115,601*/925	Green tea	(≥5 vs ≤1 cup or times/d)	0.89 (0.71–1.10)	Y	N	City, radiation exposure, smoking, drinking, education level, calendar time
	Meta-analysis

Heterocyclic amines

Heterocyclic aromatic amines are formed in meat during high-temperature cooking and are mutagenic and carcinogenic in animals. The association between fried meat and incidence of breast cancer was studied in the Finnish FSIIMCHS cohort (Table 15).60 The RR for breast cancer for the highest vs the lowest tertile of fried meat consumption was 1.80 (95% CI, 1.03–3.16).60 More evidence is needed in this area.

Table 15. Studies of Heterocyclic Amines and Breast Cancer Incidence (Premenopausal and Postmenopausal Women Combined)

Author, year, country, parent cohort	Size of cohort/no. of cases; years followed	Type of food Processing	Comparison of highest vs lowest quantile (quantity of intake)	Relative risk(95% CI)	Adjustment for confounding variables
					BMI	Family predisposition	Other variables
FSIIMCHS indicates Finnish Social Institution's Mobile Clinic Health Survey; MPCDRF, Monitoring Project on Cardiovascular Disease Risk Factors; NHS, Nurses' Health Study.
Knekt60 1994	95,611 person y/94	Fried meat	H vs 1 tertile (no information given)	1.80 (1.03–3.16)	Y	N	Smoking, geographical area, occupation, parity, intake of energy, other meats, cereals, potatoes, vegetables, fruits and berries, margarine, dairy products, fish and eggs
Finland	1966/1972 (25 y follow-up)
FSIIMCHS

Adolescent diet and breast cancer

The mammary gland is most susceptible to environmental influences in early life before accelerated cell differentiation during puberty and first pregnancy. The association between adolescent diet and breast cancer risk has been assessed in only a few cohorts (Table 16). Data are available from 3 retrospective cohort studies.49, 51, 96 Two studies considered the association between milk consumption during adolescence or high school age and breast cancer among premenopausal women, but no statistically significant association was observed.49, 51 We did not include data from NHS because breast cancer cases diagnosed before diet data collection were included in the analysis.96

Table 16. Studies of Adolescent Diet And Breast Cancer Incidence

Author, year, country, parent cohort	Size of cohort/no. of cases; years followed	Type of adolescent diet	Comparison of highest vs lowest quantile (quantity of intake)	Relative risk (95% CI)	Adjustment for confounding variables
					BMI at age 18 y	Family predisposition	Other variables
NHS indicates Nurses' Health Study; NOWAC, Norwegian Women and Cancer Study. * Includes both premenopausal and postmenopausal women, while cases are only either premenopausal or postmenopausal.
Premenopausal women
Hjartaker49 2001	48,844/317	Milk consumption during childhood	H vs 1 quartile (>7 glasses/d vs none)	0.64 (0.22–1.87)	Y (time not specified)	Y	Age at menarche, number of children, age at first birth, current use of oral contraceptives, years of education, level of physical activity, and alcohol consumption
Norway	1991/1992 (mean 6.2 y)
NOWAC
Shin51 2002	88,691*/827	Milk consumption during high school years	H vs 1 quartile (>3 vs ≤ 0.5 8-oz glasses/d)	0.81 (0.51–1.28)	Y	Y	Time period, physical activity, history of benign breast disease, height, weight change since age 18, age at menarche, parity, age at first birth, alcohol intake, total energy intake, total fat intake, glycemic index, B-carotene intake, total active E intake, and vitamin D intake
USA	1980–1996
NHS
Postmenopausal women
Shin51 2002	88,691*/2345	Milk consumption during high school years	H vs 1 quartile (>3 vs ≤ 0.5 8-oz glasses/d)	1.02 (0.82–1.26)	Y	Y	Time period, physical activity, history of benign breast disease, height, weight change since age 18, age at menarche, parity, age at first birth, alcohol intake, total energy intake, total fat intake, glycemic index, B-carotene intake, age at menopause, post menopausal hormonal use, and total active E intake
USA	1980–1996
NHS

Gene-diet interactions

A growing body of literature focuses on the effect of genetic polymorphisms on the associations between various dietary exposures and breast cancer risk. Such investigations hold the promise to detect associations between dietary exposures and breast cancer risk, as they are aimed at examining such links among women who are considered genetically susceptible or not susceptible to the beneficial or harmful effects of these exposures. To remain consistent with the body of this article, we will describe results from case-control studies nested within prospective cohorts (Table 17). These studies have largely concentrated on antioxidant-rich foods, but some have investigated the associations between genetic polymorphisms and meat consumption.

Table 17. Studies on Gene-Diet Interactions

Author, year, country, parent cohort	Size of cohort/no. of cases; years followed	Type of food/polymorphism	Comparison of highest vs lowest quantile (quantity of intake) gene-environment classification	Relative risk (95% CI)	Adjustment for confounding variables
					BMI	Family predisposition	Other variables
FSIIMCHS indicates Finnish Social Institution's Mobile Clinic Health Survey; MPCDRF, Monitoring Project on Cardiovascular Disease Risk Factors, NHS = Nurses' Health Study; d, day. * The effect of the variable on the measure of association was tested but the variable was not included in the final model.
Tamimi80 2004	32,826/968 (1205 controls)	Polymorphism	Gene-environment joint classification with reference category of Val/Val and lowest tertile of antioxidant	Ala/Ala genotype and highest tertile of α carotene: 0.93 (0.58–1.47) Ala/Ala genotype and lowest tertile of α carotene: 1.42 (0.90–2.25)	Y at age 18	Y	Weight gain since age 18, age at menarche, age at menopause, parity/age at first birth, history of benign breast disease, duration of HRT use
USA		MnSOD:Val/Val, Val/Ala, Ala/Ala
NHS	Nested case-control 1989/1990–1998
		Plasma α carotene	H vs 1 tertile (limits not described)				Matched on menopausal status, postmenopausal hormone use, month, time of day, fasting status at blood collection
				Associations/interactions between genotype and remaining antioxidants (β carotene, β cryptoxanthin, lycopene, lutein/zeaxanthin, α tocopherol, γ tocopherol, retinol, total carotenoids) not significant
Han78 2004	32,826/994 (995 controls)	Polymorphism: XRCC2	Gene-environment joint classification with reference category of 188H non-carrier and lowest quartile of antioxidant	Non-carriers of 188H in highest quartile of α carotene: 0.55 (0.40–0.75) Carriers of 188H in highest quartile of α carotene: 0.92 (0.51–1.66)	Y at age 18	Y	Weight gain since age 18, age at menarche, age at menopause, parity/age at first birth, history of benign breast disease, alcohol use, smoking, duration of HRT use
USA
NHS	Nested case-control 1989/1990–1998	R188H: 188H carriers or non-carriers					Matched on menopausal status, postmenopausal hormone use, month, time of day, fasting status at blood collection
		Plasma α carotene	H vs l quartiles (limits not described)
				No significant associations of risk for other haplotypes assessed on XRCC3 and Ligase IV
							Other measured antioxidants (β cryptoxanthin, lycopene, lutein/zeaxanthin, α tocopherol, γ tocopherol)*
Han77 2003	32,826/994 (995 controls)	Polymorphism: XRCC1	Gene-environment joint classification with reference category of ^194Trp non-carrier and lowest quartile of antioxidant	Carriers of ^194Trp and in highest quartile of α carotene: 0.40 (0.21–0.77)	Y at age 18	Y	Weight gain since age 18, age at menarche, age at menopause, parity/age at first birth, history of benign breast disease, duration of HRT use Matched on menopausal status, postmenopausal hormone use, month, time of day, fasting status at blood collection
USA				Carriers of ^194Trp and in highest quartiles of β carotene: 0.32 (0.16–0.61)
NHS	Nested case-control 1989/1990–1998	Arg^194Trp: ^194Trp carriers or noncarriers
		Plasma α carotene	H vs l quartile (control median 0.11 μmol/L)
		Plasma β carotene	H vs l quartile (control median 0.45 μmol/L)
				No significant associations of risk for other haplotypes assessed: C26602T, Arg^399Gln, Gln^634Gln
							Other measured antioxidants (β cryptoxanthin, lycopene, lutein/zeaxanthin, α tocopherol, γ tocopherol)*
Yuan59 2005	63,257 men and women/297 (665 controls)	Green tea	H vs 1 tertile (>weekly drinker vs nondrinker)	OR = 0.91 (0.66–1.26)	N	N	Age at recruitment, year of recruitment, dialect group, level of education, age when periods became regular, number of live births, black tea consumption Soy*
Singapore		Polymorphism ACE: low activity (TT and/or DD) or high activity (AT/AA and ID/II)		High activity genotype: OR = 0.29 (0.10–0.79)
Singapore Chinese Health Study			Gene-environment interaction assessed by stratification of high activity and low activity genotypes
	Nested case-control 1994–1999			Low activity genotype: OR = 1.11 (0.79–1.57)
				Significant interaction term (P = .01) between genotype and green tea consumption
Gertig76 1999	32,826/466 (466 controls)	Polymorphism: NAT2, slow acetylator or rapid acetylator	Gene-environment joint classification with reference category of lowest number of servings and slow acetylators	Charred meat consumption ≥1/wk and rapid acetylator: OR = 1.2 (0.6–2.3)	Y	Y	Age at menarche, parity, age at first birth, BMI, family history of breast cancer, and history of benign breast disease
USA
NHS	Nested-case control 1989/1990–1994
		Charred meat	H vs l tertile (≥1/wk vs <1/mo servings)				Matched on year of birth, time of day of blood draw, fasting status, month of blood sampling, HRT use, and menopausal status
van der Hel81 2004	36,000 men and women/251 (300 controls)	Polymorphisms NAT1 (slow or rapid acetylator), NAT2 (slow or rapid acetylator), GSTM1 (present or null), GSTT1 (present or null)	Gene-environment joint classification with reference category of GSTM1 present and lowest number of servings	≥100 g/d total meat consumption and GSTM1 null: OR = 1.66 (0.87–3.17)	Y*	N	Residence, menopausal status, and energy intake
Netherlands
MPCDRF							Matched on five-year age interval, menopausal status, and residence
	Nested case-control1987/1991–1997		H vs l tertile (≥100 vs <75 g/d)				Smoking, alcohol, age at menarche*
		Total meat consumption (type of preparation not noted)

Table A. Search Strategies for Systematic Review

Topic	Search strategy
Diet And Breast Cancer Search Strategy In PubMed
Overall search strategy	(“Breast Neoplasms”[MeSH] OR “breast cancer”[All Fields]) AND (“Cohort Studies”[MeSH] OR “meta analysis”[Publication Type] OR cohort* OR “Retrospective Studies”[MeSH])
Fat	(“Dietary Fats/adverse effects”[MeSH] OR “Fatty Acids/adverse effects”[MeSH] OR “fat”[All Fields])
Biomarkers of fat intake	(“Fatty Acids/blood”[MeSH]); fruits and vegetables (“Fruit”[MeSH] OR “Vegetables”[MeSH] OR “fruit and vegetable”[All Fields] OR fruit [text word] or fruit* OR vegetable [text word] OR vegetable*)
Antioxidants	(“Antioxidants”[MeSH] OR “Vitamin A”[MeSH] OR “Ascorbic Acid”[MeSH] OR “Vitamin E”[MeSH] OR “Carotenoids”[MeSH] OR antioxidants OR vitamin A OR ascorbic acid OR vitamin E or carotenoids)
Carbohydrates	(“Glycemic Index”[MeSH] OR “Carbohydrates”[MeSH] OR glycemic index or carbohydrates)
Dairy	(“Milk” [MeSH] OR “Dairy Products”[MeSH] OR (“milk”[MeSH Terms] OR milk[Text Word]) OR dairy[All Fields])
Vitamin D	(“Vitamin D”[MeSH] OR (“vitamin d”[MeSH Terms] OR vitamin D[Text Word]))
Soy	(“Soy Foods”[MeSH] OR “Isoflavones”[MeSH] OR “Phytoestrogens”[MeSH] OR soy* [All Fields] OR (“isoflavones”[MeSH Terms] OR isoflavones[Text Word]) OR (“phytoestrogens”[MeSH Terms] OR “phytoestrogens”[Pharmacological Action] OR phytoestrogens[Text Word]))
Heterocyclic amines	(“Heterocyclic Compounds”[MeSH] OR heterocyclic[All Fields] AND (“amines”[MeSH Terms] OR amines[Text Word])) OR fried[All Fields] AND (“meat”[MeSH Terms] OR meat[Text Word]) OR (well[All Fields] AND (“meat”[MeSH Terms] OR meat[Text Word]))
Adolescent diet	((“Adolescent”[MeSH] or adolescence or young adult or childhood AND (“diet”[MeSH Terms] OR diet[Text Word] or nutrition))
Gene-Environment Interaction Search Strategy in Medline
Breast cancer	Subject heading: breast neoplasms
	Keywords: breast cancer
Gene-Environment interaction	Subject heading: genotype; polymorphism (genetics); epidemiology, molecular
	Keywords: gene-environment interactions; polymorphism; molecular epidemiology; CYP; COMT; MnSOD; GST; SULT; EGFR; XRCC, MTHFR, MPO, PROGINS, MAT
Diet	Subject heading: diet, Mediterranean; diet, surveys; diet, cariogenic; diet, protein-restricted; diet, reducing; diet, therapy; diet, atherogenic; diet, macrobiotic; diet, records; diet, fads; diet, diabetic; diet, vegetarian; diet, sodium-restricted; diet, fat-restricted
	Keywords: Diet; adolescent diet
Diet specific nutrients, exposures, or measures	Subject heading: fats, unsaturated; fats, saturated; dietary fats; fats; fruit; vegetables; antioxidants; dietary carbohydrates; carbohydrates; glycemic index; dietary fiber; food handling; meat; meat products; heterocyclic compounds; milk; dairy products; soy; phytoestrogens; vitamin D Keywords: monounsaturated fat; saturated fat; polyunsaturated fat; fruit and vegetable; food processing; heterocyclic amines; dairy; soy milk; soy foods

Table B. Abbreviations of Studies in Tables

Abbreviation	Study
BCCDP	Breast Cancer Detection Demonstration Project Follow Up Cohort Study
CNBSS	Canadian National Breast Screening Study
CPS II	Cancer Prevention Study II Nutrition Cohort
CTS	California Teacher's Study
EPIC	European Prospective Investigation Into Cancer and Nutrition
FSIIMCHS	Finnish Social Institution's Mobile Clinic Health Survey
JPHC	Japan Public Health Center-based Prospective Study on Cancer and Cardiovascular Disease
IWHS	Iowa Women's Health Study
MDC	Malmo Diet Cancer Cohort
MONICA	Monitoring of Trends in Cardiovascular diseases
MPCDRF	Monitoring Project on Cardiovascular Disease Risk Factors
MSP	Mammary Screening Project
NCS	Netherlands Cohort Study
NHANES I	National Health and Nutrition Examination Survey I
NHEFS	National Health Epidemiologic Follow-up Study
NHS	Nurses' Health Study
NHS II	Nurses' Health Study II
NHSS	Norwegian National Health Screening Services
NOWAC	Norwegian Women and Cancer Study
NYUWHS	New York University Women's Health Study
ORDET	Hormone and Diet Etiology of Breast Cancer
SMSC	Swedish Mammography Screening Cohort
VIP	Vasterbotten Intervention Project
WHS	Women's Health Study

A recent case-control study nested within the NHS examined the effect of the MnSOD genetic polymorphism on the association between dietary fruit and vegetable intake and breast cancer risk. The MnSOD enzyme scavenges and neutralizes reactive oxidative species and thereby confers protection from oxidative DNA damage. Tamimi et al.80 observed a nonsignificant increase in risk for premenopausal women with the variant Ala/Ala genotype and the highest consumption of fruits and vegetables (OR = 1.87; 95% CI, 0.67–5.20), but no such effect was seen for postmenopausal women with these characteristics (OR = 0.88; 95% CI, 0.62–1.24). However, no modifying effect of the MnSOD polymorphisms on the association between dietary vitamin C intake and risk was found in this investigation. Furthermore, no significant effect of the MnSOD genotype on the relation between serum antioxidants was found. In contrast, Han et al.77 reported a protective effect of high plasma α-carotene levels and β-carotene levels most pronounced among women carrying the ¹⁹⁴Trp allele of the XRCC1 gene. In a subsequent report that focused on the influence of a panel of DNA repair genes on the association between plasma antioxidants and risk,78 this group demonstrated that a risk reduction associated with high α-carotene levels was apparent for women who carried the XRCC2 188H allele (OR = 0.55; 95% CI, 0.40–0.75), but not among noncarriers (OR = 0.92; 95% CI, 0.51–1.66). Interestingly, a report from the Singapore Chinese Health Study demonstrated a protective association of green tea consumption restricted to women with the ACE high-activity genotype (OR = 0.29; 95% CI, 0.10–0.79) and not apparent among women with the low-activity genotype (OR = 1.11; 95% CI, 0.79–1.57).59 The ACE gene codes for an enzyme that is involved in the conversion of angiotensin II, which is important in reduction of cell adhesion and invasion.97 Polymorphisms in the ACE gene have been linked to increased risk of breast cancer98, 99 and may modify the association between green tea consumption and breast cancer risk among Chinese women.59

To determine whether or not the potential adverse effect of heterocyclic aromatic amines might be restricted to women with genetically determined poor detoxification capacity, Gertig et al.76 conducted a case-control study nested in the NHS. However, they reported similar associations between red or charred meat consumption and risk among women with the slow or rapid acetylator genotypes of the NAT2 gene, which is involved in the detoxification of activated carcinogens.76 Similarly, results from a case-control study nested in a Dutch prospective cohort did not provide evidence for the notion that the NAT1 or NAT2 genotypes modify the association between meat consumption and breast cancer risk.81 Nevertheless, in the same report the authors note that women with 2 intact GSTM1 alleles and the highest total meat intake were not at elevated risk of breast cancer (OR = 1.02; 95% CI, 0.52–2.00), whereas women with the GSTM1 null genotype and highest intake had a nonsignificant increase in risk (OR = 1.66; 95% CI, 0.87–3.17).

Conclusions, Recommendations, and Future Directions

Among the prospective epidemiologic studies conducted on diet and breast cancer to date there is no association that is consistent, strong, and statistically significant, except for regular alcohol consumption, overweight, and weight gain. There are several possible explanations for this apparent lack of association between diet and breast cancer incidence:

• 1
  No causal association: Diet during adult life may not be an important predictor of breast cancer risk.
• 2
  Measurement error: Self-reported diet is generally assessed with considerable measurement error. Any true association between diet and breast cancer is likely to be small to moderate. It is possible that a modest or moderate association such as a true RR of 1.5 or 0.7 cannot be detected with our current diet assessment methods.
• 3
  Variation in diet: Most studies on diet and breast cancer have been conducted in industrialized countries, primarily North America, Europe, and Japan. Although there is substantial variation in diet across Northern and Southern European countries, the international variation in dietary patterns across industrialized and developing countries is even greater. The role of diet for breast cancer risk in developing countries has been less explored. Because the variation in diet studied needs to exceed the measurement error, diet studies may need to take advantage of the international variation in diet when exploring the association with cancer. Many studies restricted to 1 homogeneous population may not have sufficient variation in intake to overcome the random noise introduced by measurement error.
• 4
  Timing of diet: Almost all studies on diet and breast cancer have focused on diet during adult life. It is possible that the window of opportunity is a different one than we have focused on in the past. The human breast may be most susceptible to dietary influences during early life, in particular before puberty. Currently available data do not allow inferences about the role of diet before or during puberty and the risk of breast cancer.
• 5
  Follow-up: The average follow-up time in the cohort studies was 8 to 9 years. This time period may be too short to capture a role of diet as an initiator or initial preventive factor of cancer; however, because most people do not change their dietary preferences much over time, inferences may be possible to the more distant past. Diet may also promote or prevent further growth of a cancer that is already initiated.
• 6
  Subgroups of women characterized by estrogen- or progesterone-receptor status or genetic, epigenetic, or hormonal status may be specifically susceptible to the influence of diet. This has not been sufficiently explored.
• 7
  It is possible that a protective effect of some foods such as fruits and vegetables on breast cancer risk is countered by a harmful effect of food residues such as pesticides.100 Furthermore, the association of more restrictive dietary patterns such as organic foods, whole foods, raw foods, or a vegan diet with breast cancer incidence has not been sufficiently studied.

Future studies may address some of the concerns raised above by developing improved dietary assessment instruments, improving analytic methods to adequately correct for measurement error, and developing new biomarkers, in particular recovery biomarkers of dietary intake. The biology of the mammary gland and preliminary evidence calls for studies to evaluate the role of diet during early life, especially puberty.7 The reasons for the stark contrast in breast cancer rates in developed and developing countries is insufficiently explored. Furthermore, there is evidence that consumption of organic foods prevents the accumulation of persistent toxic chemicals from food in the body,101 but any impact of organic foods on cancer prevention remains unexplored.

Data on the role of diet in premenopausal breast cancer are still sparse. In addition, more diet studies are needed that consider breast cancer outcomes more specifically according to estrogen- and progesterone-receptor status. There is evidence that diet may play a more important role in ER− breast cancer than in ER+ breast cancer, but such associations may not be detected in analyses of overall breast cancer.102–104

Some leads from observational studies will deserve further attention in the next decade. Intake of animal and saturated fat have been associated with increased breast cancer incidence in some studies and more evidence is needed on the role of these subtypes of fat.13, 30 Similarly, consumption of high-fat dairy foods has been associated with pre- but not postmenopausal breast cancer.13, 51 Soy intake may play a role in breast cancer etiology if consumption starts in adolescence and continues in adulthood—an example of the importance of the critical time period and of long-term intake.105

Future studies of gene-diet interactions are also warranted and perhaps particularly important in the field of diet and breast cancer, because most of the existing evidence has not revealed strong associations with risk. It is possible that the beneficial or harmful effects of dietary exposures are restricted to subgroups of women defined by specific genetic characteristics with direct relevance to the biological pathways underlying the associations under investigation. Such studies should be guided by the recent recommendations for studies of genetic variations and cancer by Rebbeck et al.79 Specifically, future studies of gene-diet interactions should be solidly grounded in biological plausibility, which requires that 1) the gene under investigation is biologically related to the dietary exposure of interest, and 2) the gene under investigation has demonstrated functional significance. To be informative, future studies of gene-diet interactions must have sufficient statistical power to detect suspected modifying effects of candidate genetic polymorphism on associations between dietary exposures and breast cancer risk. Thus, investigators will have to consider the prevalence of the variant genotype, expected risk elevation, and available sample size. Finally, future investigations should use appropriate, yet creative statistical methods, as well as transparent, well-described, and reproducible laboratory methodologies.

Among dietary items, evidence is consistent for a modest but consistent effect of alcohol on increasing breast cancer risk9 and the observation that this effect may be neutralized by folate.36, 106 Finally, the impact of total caloric intake, energy balance, and weight gain on the risk of breast cancer107, 108 indicates a role for overall diet and dietary patterns that may not be captured in studies of individual foods and nutrients.

Acknowledgements

Supported by Susan G. Komen for the Cure as part of the Environmental Factors and Breast Cancer Science Review project led by Silent Spring Institute with collaborating investigators at Harvard Medical School, Roswell Park Cancer Institute, and the University of Southern California.