Meat, fish and egg intake and risk of breast cancer

Abstract

Intakes of animal protein, meat, and eggs have been associated with breast cancer incidence and mortality in ecological studies, but data from long-term prospective studies are limited. We therefore examined these relationships in the Nurses' Health Study. We followed 88,647 women for 18 years, with 5 assessments of diet by food frequency questionnaire, cumulatively averaged and updated over time. We calculated the relative risks (RR) and 95% confidence intervals (95% CI) for risk of developing invasive breast cancer, over categories of nutrient and food intake. During follow-up, 4,107 women developed invasive breast cancer. Compared to the lowest quintile of intake, the RR and 95% CI for the highest quintile of intake were 1.02 (0.92–1.14) for animal protein, 0.93 (0.83–1.05) for red meat and 0.89 (0.79–1.00) for all meat. Results did not differ by menopausal status or family history of breast cancer. We found no evidence that intake of meat or fish during mid-life and later was associated with risk of breast cancer. © 2003 Wiley-Liss, Inc.

In the search to explain international differences in breast cancer rates, differences in diet have been a major focus. In 1979, Gray et al.1 reported that among countries, per capita consumption of animal protein, meat and egg intakes were highly correlated with breast cancer incidence rates (r=0.80, 0.73 and 0.76, respectively), even after controlling for height, weight and age at menarche. Later ecological analyses have supported the association between total meat consumption and breast cancer incidence in Norway,2 and breast cancer mortality in Spain3 and several other countries.4 In contrast, several ecological studies have found an inverse relationship between fish or marine fat consumption and breast cancer mortality.5, 6, 7, 8, 9, 10, 11, 12 One ecological study in the U.S., which assessed diet in 1965 and 1966, found an inverse association between egg consumption and breast cancer mortality.13

Observational studies of meat intake and breast cancer have had mixed results. In a retrospective study based in the prospective Iowa Women's Health Study, although there was no association of breast cancer risk with total meat intake, there was a strong association with the method of cooking meat. Charring meat forms heterocyclic amines, substances known to cause mammary tumors in animals.14 Women consuming all meat cooked very well done had a more than 4-fold risk of breast cancer compared to women consuming all meat cooked rare or medium.15 This raises the issue that cooking methods may modify any association between meat intake and breast cancer.

We prospectively examined the association of meat, fish and egg intake and risk of developing breast cancer in the Nurses' Health Study. We also performed a sub-analysis examining the relationship of meat cooking methods with breast cancer risk. We followed women for 18 years, with 5 assessments of diet over time, during which 4,107 incident cases developed.

MATERIAL AND METHODS

The Nurses' Health Study Cohort

In 1976, the Nurses' Health Study (NHS) cohort was established when 121,700 female registered nurses from across the United States, aged 30 to 55 years, answered a mailed questionnaire on risk factors for cancer and cardiovascular disease. Every 2 years since, we have sent follow-up questionnaires to NHS participants. In 1980, a 61-item food-frequency questionnaire designed to assess dietary intake was added. In 1984, 1986, 1990 and 1994, an expanded food-frequency questionnaire was used. This analysis is based on 88,647 women who answered the 1980 diet questionnaire, who did not have implausible scores for total energy intake (<500 kcal or >3500 kcal/day, approximately 2% of returned diet questionnaires) and who did not have diagnosed cancer other than nonmelanoma skin cancer (3,093 cases excluded) prior to 1980.

The semiquantitative food-frequency questionnaires

The food-frequency questionnaires have been described in detail.16 A commonly used portion size was specified for each food (for example, one slice of bread or one egg). Participants were asked to average how frequently over the past year they had consumed that portion of food. The 9 pre-specified responses ranged from “never” to “six or more times per day.” The validity and reproducibility of the food frequency questionnaires for nutrients17 and foods18 have been documented elsewhere. In this population, the corrected correlation between the food frequency questionnaire and dietary records was 0.43–0.46 for meat, 0.52–0.66 for fish and 0.72–0.77 for eggs.

In this analysis, we examined the following types of animal source foods: fish, poultry (with and without skin), hamburger; beef, pork or lamb as a main dish or sandwich; and hot dogs, bacon, other processed meat and eggs (consumed alone, not in cooking). White meat was considered to be the sum of fish and poultry intake. On the questionnaires, the question for “other processed meat” came after the questions for bacon and hot dogs; examples given were sausage, salami and bologna. Total processed meat intake was considered to be the sum of hot dogs, bacon and other processed meat. Red meat intake was considered to be the sum of hamburger; beef, pork or lamb as a main dish; and beef, pork or lamb as a sandwich, and all processed meat. The total meat was the sum of white and red meat intakes. We also examined total protein, animal protein and vegetable protein.

In 1980, we asked participants, “What percentage of the time do you eat your meat cooked well-done?” In 1986, we asked, “How often do you eat meat that was charred during cooking? (e.g., during barbecuing or broiling).” The choices were the following: never, less than once per month, once per month, 2–3 times per month, once per week, and 2 or more times per week. Also, in 1986 we asked about diet during high school. The high school diet items that we examined were fish, eggs, hot dogs and other beef, pork or lamb dishes.

Identification of breast cancer cases

In each biennial questionnaire, participants were asked whether they had been diagnosed with breast cancer in the previous 2 years. Deaths were identified by a report from a family member, the postal service or the National Death Index; the ascertainment is estimated to be 98% complete. Follow-up of the diet cohort through 1998 is 98% complete. Medical records were obtained for breast cancer cases identified by either self-report or vital records, and over 99% of these records confirmed the self-report. We included 4,107 cases of invasive breast cancer in this analysis. Cases of carcinoma-in-situ were censored.

Statistical analysis

Each participant accumulated person-time beginning with the return of the 1980 questionnaire and ending with her cancer diagnosis, death or June 1, 1998, whichever came first. For analyses that involved cooking methods or high school diet, we began follow-up at the time of return of the 1986 questionnaire.

Cox proportional hazards models with age in months as the underlying time variable were used to calculate relative risks of breast cancer adjusted for other breast cancer risk factors. To take into account dietary changes over time, we calculated the cumulative average intake of foods and nutrients from all available dietary questionnaires up to the start of each 2-year interval. In this calculation, the 1980 diet was related to 1980–1984 breast cancer incidence; the average of the 1980 and 1984 diets was related to 1984–1986 breast cancer incidence; the average of the 1980, 1984 and 1986 diets was related to 1986–1990 breast cancer incidence; the average of the 1980, 1984, 1986 and 1990 diets was related to the 1990–1994 breast cancer incidence and the average of all 5 diets was related to 1994–1998 breast cancer incidence.

Protein intake was expressed as the percent of total energy intake and categorized into quintiles. Food intakes were energy-adjusted using the residual method17 and categorized into quintiles. For both protein and foods, the lowest quintile of intake was the reference category. Total energy intake was included in each regression model. Because exposures may affect risk of pre- and post-menopausal breast cancer differently, results were stratified by menopausal status at time of diagnosis. In alternative analyses, we used only the 1980 or the 1986 baseline diet.

The following nondietary covariates were updated every 2 years: age, history of benign breast disease, menopausal status, age at menopause, use and duration of use of postmenopausal hormones, parity, age at first birth and weight change since age 18. Age at menarche and height were determined at baseline, and information on family history of breast cancer was sought in 1976, 1982, 1988, 1992 and 1996. Body mass index (weight in kilograms divided by the square of height in meters) at age 18 was determined in 1980. Women of uncertain ovulatory status (mainly those who had undergone hysterectomy but had intact ovaries) were excluded from analyses that included stratification by menopausal status, until they reached the age at which 90% of the cohort had become post-menopausal; at which point they were considered post-menopausal. In tests for linear trend across quintiles of food and nutrient intake, ordinal rank was assigned to each category.

RESULTS

Mean age at the beginning of follow-up (1980) was 46.7 years (standard deviation = 7.2). We identified 4,107 incident cases of invasive breast cancer among 88,647 women between 1980 and 1998. Between 1980 and 1994, intake of fish and poultry increased substantially, intake of red meat and eggs decreased substantially, while total meat consumption decreased by only a small amount (Fig. 1). To control for these and other secular trends, all analyses were controlled for time period.

Figure 1. Mean change in servings over time.

Meat was the largest source of protein in this population. There was no evidence that total protein, animal protein or vegetable protein intake is associated with breast cancer in this population (Table I).

Table I. Multivariate Relative Risk of Breast Cancer (1980–1998), 95% Confidence Intervals and P-Value for the Linear Trend, Across Quintiles of Cumulatively Averaged and Updated Intake of Protein1

Nutrient	Percent of total energy/day 1994	All women (N = 88,647, cases = 4,107) RR (95% CI)	Pre-menopausal (N = 53,952, cases = 854) RR (95% CI)	Postmenopausal (N = 76,152, cases = 2,936) RR (95% CI)
1 Adjusted for age (continuous); 2 year time period; total energy intake (quintiles); alcohol intake (0, < 5, 5–14.9, ≥15g/day); parity and age at first birth categories (nulliparious, 1–2 births and age at first birth < 25 years, 1–2 births and 25≤ age at first birth < 30, 1–2 births and age at first birth ≥30, 3–4 births and age at first birth < 25, 3–4 births and 25≤ age at first birth < 30, 3–4 births and age at first birth 30, ≥5 births and age at first birth < 25, 5 births and 25≤ age at first birth < 30, missing); Body mass index at age 18 in kg/m^2 (< 19, 19–20.0, 21–22.9, 23, missing); weight change since age 18 in kg (< −2, −2 to 2, > 2 to 5, > 5 to 10, > 10 to 20, > 20 to 25, > 25); height in inches (< 63, 63–63.9, 64–65.9, ≥66); family history of breast cancer (yes/no); history of benign breast disease (yes/no); age at menarche in years (≤12, 13, ≥14); menopausal status, age at menopause and hormone replacement therapy (HRT) use categories (premenopausal, uncertain menopausal status, postmenopausal with age at menopause < 45 years and no HRT use, postmenopausal with age at menopause < 45 years and past HRT use, postmenopausal with age at menopause < 45 years and current HRT use < 5 years, postmenopausal with age at menopause < 45 years and current HRT use ≥5 years, postmenopausal with age at menopause 45–52 and no HRT use, postmenopausal with age at menopause 45–52 and past HRT use, postmenopausal with age at menopause 45–52 and current HRT use > 5 years, postmenopausal with age at menopause 45–52 and current HRT use ≥5 years, postmenopausal with age at menopause ≥53 and no HRT use, postmenopausal with age at menopause ≥53 and past HRT use, postmenopausal with age at menopause ≥53 and current HRT use < 5 years, postmenopausal with age at menopause ≥53 and current HRT use ≥5 years), duration of menopause (continuous).
Total protein	< 16.5	1.00 (—)	1.00 (—)	1.00 (—)
	16.5–17.8	0.97 (0.88–1.07)	1.06 (0.85–1.31)	0.92 (0.82–1.04)
	17.9–19.0	1.08 (0.98–1.20)	1.21 (0.98–1.49)	1.04 (0.93–1.17)
	19.1–20.6	1.03 (0.94–1.14)	1.10 (0.88–1.37)	1.01 (0.90–1.13)
	≥ 20.7	1.05 (0.95–1.16)	1.09 (0.86–1.37)	1.01 (0.90–1.14)
		P = 0.37	P = 0.60	P = 0.66
Animal protein	< 11.7	1.00 (—)	1.00 (—)	1.00 (—)
	11.7–13.1	0.99 (0.90–1.09)	1.07 (0.86–1.33)	0.97 (0.87–1.09)
	13.2–14.4	1.04 (0.95–1.15)	1.18 (0.95–1.46)	1.00 (0.89–1.13)
	14.5–16.1	1.01 (0.92–1.12)	1.20 (0.97–1.50)	0.97 (0.86–1.09)
	≥ 16.2	1.02 (0.92–1.13)	1.05 (0.83–1.33)	1.00 (0.88–1.12)
		P = 0.43	P = 0.59	P = 0.69
Vegetable protein	< 4.0	1.00 (—)	1.00 (—)	1.00 (—)
	4.0–4.3	1.06 (0.96–1.18)	1.20 (0.97–1.49)	1.01 (0.90–1.14)
	4.4–4.7	1.01 (0.91–1.11)	1.09 (0.88–1.36)	0.99 (0.88–1.12)
	4.8–5.2	1.02 (0.92–1.13)	1.09 (0.87–1.36)	0.99 (0.88–1.12)
	≥ 5.3	1.02 (0.92–1.13)	1.04 (0.82–1.31)	0.99 (0.87–1.12)
		P = 0.82	P = 0.87	P = 0.97

In the whole cohort and in both the pre- and post-menopausal groups, there was no association of breast cancer incidence with intakes of fish, poultry or red meat (Table II). A slightly lower risk of breast cancer was seen for higher intake of processed meat other than bacon or hot dogs, primarily in pre-menopausal women. Also, a borderline linear trend toward higher risk of breast cancer with higher egg intake was seen in pre-menopausal women.

Table II. Multivariate1 Relative Risk of Breast Cancer (1980–1998), 95% Confidence Intervals, and P-Value for the Linear Trend, Across Quintiles of Cumulatively Averaged and Updated Intake of Animal Source Foods

Food	Servings/day	All women (N = 88,647, cases = 4,107) RR (95% CI)	Pre-menopausal (N = 53,952, cases = 854) RR (95% CI)	Postmenopausal (N = 76,152, cases = 2,936) RR (95% CI)
1 Adjusted for same factors as Table I.
Fish	≤ 0.13	1.00 (—)	1.00 (—)	1.00 (—)
	0.14–0.20	0.98 (0.89–1.08)	1.20 (0.96–1.49)	0.95 (0.84–1.07)
	0.21–0.27	0.97 (0.87–1.08)	1.03 (0.79–1.34)	0.97 (0.85–1.09)
	0.28–0.39	0.99 (0.90–1.09)	1.06 (0.84–1.33)	0.98 (0.87–1.11)
	≥ 0.40	1.04 (0.93–1.14)	1.17 (0.92–1.50)	1.00 (0.89–1.12)
		p = 0.55	p = 0.71	p = 0.79
Poultry without skin	≤ 0.08	1.00 (—)	1.00 (—)	1.00 (—)
	0.09–0.14	0.97 (0.87–1.08)	0.99 (0.77–1.27)	0.99 (0.87–1.12)
	0.15–0.24	1.04 (0.94–1.14)	1.08 (0.88–1.31)	1.06 (0.94–1.18)
	0.25–0.36	0.95 (0.87–1.05)	0.92 (0.75–1.12)	0.98 (0.87–1.10)
	≥ 0.37	1.01 (0.92–1.11)	0.97 (0.79–1.20)	1.01 (0.90–1.13)
		p = 0.99	p = 0.56	p = 1.00
Poultry with skin	0	1.00 (—)	1.00 (—)	1.00 (—)
	0.01–0.05	1.04 (0.94–1.15)	0.93 (0.72–1.21)	1.10 (0.98–1.24)
	0.06–0.10	1.04 (0.95–1.15)	1.04 (0.84–1.36)	1.05 (0.94–1.18)
	0.11–0.17	1.06 (0.96–1.17)	1.03 (0.81–1.31)	1.11 (0.99–1.24)
	≥ 0.18	1.00 (0.90–1.10)	1.00 (0.78–1.28)	1.00 (0.89–1.12)
		p = 0.78	p = 0.67	p = 0.79
All poultry	≤ 0.17	1.00 (—)	1.00 (—)	1.00 (—)
	0.18–0.26	0.99 (0.89–1.09)	1.15 (0.92–1.44)	0.95 (0.84–1.07)
	0.27–0.35	1.03 (0.93–1.14)	1.13 (0.88–1.45)	1.02 (0.91–1.15)
	0.36–0.45	1.02 (0.92–1.12)	1.18 (0.93–1.51)	0.96 (0.85–1.07)
	≥ 0.46	1.01 (0.91–1.11)	1.08 (0.85–1.37)	1.00 (0.89–1.12)
		p = 0.69	p = 0.65	p = 0.97
White meat (fish and poultry)	< 0.36	1.00 (—)	1.00 (—)	1.00 (—)
	0.36–0.48	0.98 (0.89–1.08)	1.01 (0.81–1.27)	0.97 (0.86–1.09)
	0.49–0.62	0.96 (0.87–1.06)	1.02 (0.82–1.28)	0.95 (0.84–1.07)
	0.63–0.83	1.05 (0.95–1.16)	1.14 (0.91–1.54)	1.03 (0.91–1.15)
	≥ 0.84	1.00 (0.90–1.10)	1.01 (0.79–1.28)	0.97 (0.86–1.09)
		p = 0.56	p = 0.58	p = 0.99
Hamburger	≤ 0.08	1.00 (—)	1.00 (—)	1.00 (—)
	0.09–0.12	1.02 (0.92–1.13)	1.04 (0.79–1.35)	1.01 (0.90–1.14)
	0.13–0.16	0.98 (0.90–1.08)	1.01 (0.82–1.24)	0.99 (0.89–1.10)
	0.17–0.24	1.00 (0.90–1.11)	1.00 (0.76–1.31)	0.98 (0.87–1.11)
	≥ 0.25	0.95 (0.86–1.06)	1.02 (0.81–1.29)	0.91 (0.80–1.03)
		p = 0.36	p = 0.96	p = 0.15
Beef, pork, or lamb sandwich	≤ 0.07	1.00 (—)	1.00 (—)	1.00 (—)
	0.08–0.11	0.92 (0.83–1.02)	0.78 (0.62–0.99)	0.96 (0.86–1.08)
	0.12–0.17	1.02 (0.92–1.12)	0.88 (0.71–1.10)	1.05 (0.94–1.17)
	0.18–0.26	0.95 (0.85–1.05)	0.79 (0.61–1.02)	0.99 (0.87–1.11)
	≥ 0.27	1.00 (0.90–1.11)	0.95 (0.76–1.20)	1.00 (0.89–1.13)
		p = 0.85	p = 0.93	p = 0.84
Beef, pork, or lamb main dish	≤ 0.14	1.00 (—)	1.00 (—)	1.00 (—)
	0.15–0.21	1.04 (0.94–1.15)	1.20 (0.94–1.52)	1.00 (0.89–1.12)
	0.22–0.30	1.05 (0.94–1.16)	1.12 (0.86–1.46)	1.03 (0.92–1.16)
	0.31–0.44	1.07 (0.97–1.19)	1.29 (1.02–1.64)	1.04 (0.92–1.17)
	≥ 0.45	0.98 (0.87–1.09)	1.08 (0.82–1.42)	0.97 (0.86–1.11)
		p = 0.99	p = 0.40	p = 0.98
Hot dog	≤ 0.01	1.00 (—)	1.00 (—)	1.00 (—)
	0.02–0.05	0.97 (0.87–1.07)	0.95 (0.73–1.26)	0.97 (0.87–1.09)
	0.06–0.07	1.05 (0.96–1.15)	1.11 (0.90–1.36)	1.04 (0.94–1.16)
	0.08–0.11	0.99 (0.89–1.10)	1.08 (0.83–1.41)	0.98 (0.86–1.11)
	≥ 0.12	1.04 (0.95–1.15)	1.16 (0.94–1.44)	1.01 (0.90–1.14)
		p = 0.35	p = 0.11	p = 0.81
Bacon	0	1.00 (—)	1.00 (—)	1.00 (—)
	0.01–0.04	0.97 (0.88–1.08)	0.94 (0.73–1.22)	0.98 (0.87–1.11)
	0.05–0.07	1.01 (0.92–1.11)	1.08 (0.88–1.32)	1.03 (0.92–1.16)
	0.08–0.13	1.02 (0.93–1.12)	1.09 (0.89–1.34)	1.02 (0.91–1.14)
	≥ 0.14	0.96 (0.87–1.07)	0.93 (0.73–1.19)	1.01 (0.89–1.14)
		p = 0.92	p = 0.79	p = 0.68
Other processed meat	≤ 0.02	1.00 (—)	1.00 (—)	1.00 (—)
	0.03–0.06	0.98 (0.89–1.09)	0.69 (0.52–0.91)	1.06 (0.95–1.19)
	0.07–0.10	0.93 (0.84–1.01)	0.81 (0.66–0.99)	0.96 (0.86–1.07)
	0.11–0.20	0.93 (0.84–1.03)	0.92 (0.75–1.13)	0.91 (0.81–1.03)
	≥ 0.21	0.89 (0.80–0.98)	0.71 (0.57–0.88)	0.95 (0.85–1.08)
		p = 0.02	p = 0.05	p = 0.10
All processed meat (hot dog, bacon, and other processed meat)	≤ 0.10	1.00 (—)	1.00 (—)	1.00 (—)
	0.11–0.18	1.03 (0.94–1.13)	0.99 (0.78–1.25)	1.05 (0.94–1.17)
	0.19–0.28	1.01 (0.92–1.11)	1.08 (0.87–1.35)	1.02 (0.91–1.14)
	0.29–0.45	0.96 (0.87–1.06)	0.99 (0.79–1.25)	0.95 (0.84–1.06)
	≥ 0.46	0.94 (0.85–1.05)	0.86 (0.67–1.09)	1.00 (0.88–1.13)
		p = 0.12	p = 0.25	p = 0.45
Red meat	≤ 0.55	1.00 (—)	1.00 (—)	1.00 (—)
	0.56–0.76	0.99 (0.90–1.09)	0.95 (0.75–1.20)	1.03 (0.92–1.15)
	0.77–0.99	0.99 (0.89–1.09)	1.03 (0.82–1.30)	1.00 (0.89–1.13)
	1.00–1.31	1.00 (0.91–1.11)	1.11 (0.89–1.41)	0.99 (0.88–1.12)
	≥ 1.32	0.94 (0.84–1.05)	0.94 (0.72–1.22)	0.99 (0.86–1.13)
		p = 0.45	p = 0.90	p = 0.66
All meat (red and white)	≤ 1.11	1.00 (—)	1.00 (—)	1.00 (—)
	1.12–1.37	0.88 (0.80–0.97)	0.90 (0.71–1.13)	0.91 (0.81–1.02)
	1.38–1.63	0.98 (0.88–1.08)	1.09 (0.87–1.37)	0.95 (0.84–1.07)
	1.64–1.99	0.96 (0.87–1.07)	1.07 (0.84–1.37)	0.94 (0.83–1.07)
	≥ 2.00	0.89 (0.79–1.00)	0.90 (0.69–1.18)	0.88 (0.77–1.02)
		p = 0.29	p = 0.98	p = 0.21
Eggs	≤ 0.13	1.00 (—)	1.00 (—)	1.00 (—)
	0.14–0.22	0.95 (0.86–1.05)	1.00 (0.79–1.26)	0.97 (0.86–1.08)
	0.23–0.31	1.03 (0.93–1.14)	1.23 (0.97–1.55)	0.97 (0.86–1.09)
	0.32–0.43	1.04 (0.94–1.16)	1.22 (0.95–1.57)	1.02 (0.91–1.15)
	≥ 0.44	1.03 (0.93–1.15)	1.15 (0.90–1.47)	1.01 (0.89–1.14)
		p = 0.16	p = 0.08	p = 0.57

Quintiles of intake may not reflect categories extreme enough to detect an association. We therefore used women with moderate intakes of a particular food as the comparison group to examine breast cancer risk among women with zero intake and women with one or more servings per day of that food (representing the extreme 0.2–3.6% of the cumulatively averaged intake distribution in 1994) (Table III). There was no decrease in breast cancer risk for women consuming no fish, eggs, red meat or all meat, compared to women consuming moderate amounts of those foods. In fact vegetarians, (those consuming no meat) had a slightly increased (although not statistically significant) risk of breast cancer: relative risk (RR) [95% confidence interval (CI)] = 1.44 (0.84–2.44). Likewise, women consuming the highest amounts of fish (≥ 1 serving per day), eggs (≥ 1 serving per day), red meat (≥ 2 servings per day) and all meat (≥ 3 servings per day) had no increased risk of breast cancer compared to women consuming moderate amounts of these foods.

Table III. Multivariate1 Relative Risk (and 95% Confidence Intervals) of Developing Breast Cancer Associated with the Extremes of Cumulatively Averaged and Updated Intakes of Animal Source Foods, Compared with Moderate Intake, and P-Value for the Linear Trend Across Continuous Intakes

Food	Number of cases	Percent of women at this intake level in 1994	Intake	Relative risk	(95% Confidence interval)	p-value linear trend
1 Adjusted for same factors as Table I.
Fish	144	1.7	No intake	1.05	(0.89–1.25)
	3,894	96.4	0 < servings/day < 1	1.00	(—)
	69	1.8	≥ 1 serving/day	0.86	(0.67–1.10)	0.55
Eggs	119	1.6	No intake	1.10	(0.91–1.32)
	3,818	96.6	0 < servings/day < 1	1.00	(—)
	170	1.8	≥ 1 serving/day	1.12	(0.96–1.31)	0.15
Red meat	25	0.5	No intake	1.01	(0.68–1.51)
	3,879	95.9	0 < servings/day < 2	1.00	(—)
	203	3.6	≥ 2 servings/day	0.95	(0.82–1.10)	0.22
All meat	14	0.2	No intake	1.43	(0.84–2.43)
	3,993	97.5	0 < servings/day < 3	1.00	(—)
	100	2.3	≥ 3 servings/day	0.92	(0.75–1.13)	0.56

The 8-cohort international Pooling Project (of which the NHS is a part) recently reported a slightly increased risk of breast cancer for an increase of 2 eggs per day at baseline: RR=1.22, 95% CI = 1.03–1.45).19 We found little difference in risk between women with a cumulatively averaged intake of ≥ 2 eggs per day (RR=1.13, 95% CI = 0.75–1.69) and women consuming ≥ 1 egg per day (RR=1.12, 95% CI = 0.96–1.31) if each high intake category was compared to women consuming more than zero but less than 1 egg per day as a reference category. Because of the limited number of women with egg intake ≥ 2 per day over time (0.2% of the population), we were unable to examine this association stratified by menopausal status.

We performed a number of additional secondary analyses. The associations between the foods listed in Table II and breast cancer risk did not differ by family history of breast cancer in a mother or sister.

A latency period could exist between food intake and its affect on disease. The results were unchanged when we examined the associations of diet with breast cancer when diet was assessed only in 1980, and also when diet was simply updated over time, without cumulatively averaging in prior intake.

There may be a lag between the time a diet is consumed and when it may affect breast cancer risk. We found no association between cumulatively averaged and updated pre-menopausal meat intake and postmenopausal breast cancer risk 8 or more years later. It is also possible that the absolute amount of meat rather than the amount relative to total energy intake is important for breast cancer risk. We found no association between meat intake and breast cancer risk when we adjusted for all the covariates listed in Table Iexcept total energy intake.

Percent of time eating well-done meat in 1980 was not associated with breast cancer risk. Compared to those meat-eaters who consumed well-done meat ≤25% of the time, the relative risk (95% confidence interval) for those consuming well-done meat 26–50% of the time was 1.00(0.91–1.10), for consuming well-done meat 51–75% of the time was 1.02(0.92–1.13) and for consuming well-done meat 76–100% of the time was 0.98 (0.90–1.06) (p-value for linear trend=0.65). We also found no association between the amount of meat intake in 1980 and risk of breast cancer among the 14% of participants who said they always ate their meat cooked well-done. Similarly, frequency of eating charred meat in 1986 was not associated with breast cancer risk; RR (95% CI) for 2 or more times per week vs. never was 0.98 (0.78–1.22).

Intake of antioxidant vitamins may counteract the pro-apoptotic effects of omega-3 fatty acids.20 Therefore, we repeated all analyses of fish intake, additionally controlling for intake of multivitamins. Results did not change.

The decrease in meat consumption over time may have been accompanied by an increase in consumption of sugar, sweets and other high glycemic and insulinemic index foods, which might affect the risk of breast cancer. Therefore, we did additional analyses controlling for glycemic load.21, 22 Results did not change.

Health conscious women may have been more likely to eat less meat and also more likely to have screening mammograms. Therefore we did additional analyses restricted to only those women who reported having a screening mammogram prior to 1994. There were 75,166 such women (85% of the cohort) and 3,399 incident cases of breast cancer. Results in general did not change. However, there was a borderline association of increased breast cancer risk with increased egg intake: RR (and 95% CI) across increasing quintiles of egg intake were 1.00, 0.93, 1.05, 1.07 and 1.05 (0.94–1.18), p-value for linear trend = 0.05.

In 1986, we asked participants about their intake of eggs, hot dogs, beef and fish during high school. Intake of these foods was not associated with total risk of breast cancer nor with risk of pre-menopausal nor postmenopausal breast cancer.

DISCUSSION

Many case-control studies in multiple populations around the world have found a positive association between intake of meat,23, 24, 25 red meat,25, 26 high fat meat,27 beef,28, 29 pork,28, 30 sausage24 and animal protein26 and risk of breast cancer. However, no association between total meat consumption and breast cancer risk was seen in one case-control study based in Poland with 328 cases31 and another in Iowa, U.S. with 273 cases.15 A meta-analysis of 12 case-control studies published in 1993 by Boyd et al.32 found a summary RR (95% CI) of 1.14 (1.02–1.29) for the highest compared to the lowest category of meat intake.

The evidence from prospective studies is less consistent. The same meta-analysis mentioned above found a summary RR (95% CI) of 1.37 (1.07–1.76) for breast cancer comparing highest to lowest meat intake among 5 cohort studies published before 1993.32 Among prospective studies published after 1993, one based in New York state with 180 cases33 and one based in Norway with 248 cases34 showed positive associations with meat intake, with relative risks for the highest category of intake approximately double that of the lowest category of intake. However, the 8-cohort international Pooling Project (of which the NHS is a part) with more than 350,000 women and almost 7,400 breast cancer cases followed for up to 15 years recently reported no association of breast cancer risk with total, red or white meat intake; they report a summary RR (95%CI) for a 100 g increase in meat intake of 1.02 (0.97–1.08).20 Contrary to the predominant hypothesis, we also found no evidence of increased breast cancer risk with meat intake assessed several different ways. We found a borderline association of decreased breast cancer risk with increased intake of processed meat other than bacon or hot dogs (such as sausage, salami and bologna) but not bacon or hot dogs primarily in pre-menopausal women. As we have no mechanism or prior hypothesis for this association it may be a chance finding.

The current study extended the analysis of the Pooling Project in several ways. We had longer follow-up (18 years vs. 5–10 years), repeated measures of dietary intake over time (5 separate measures of food frequency vs. one) and we were able to examine cooking methods and high school diet as well.

The 8 cohort international Pooling Project reports no association of breast cancer risk with fish intake, RR (95% CI) for 100 g/day increment = 1.01 (0.87–1.16) (20). One case-control study based in Italy with 2,569 breast cancer cases and 2,588 controls found a modest inverse association with fish intake, particularly among post-menopausal women.35 With our extended follow-up we found a similar result to the Pooling Project.

The Pooling Project reported an unexpected weak positive association of breast cancer risk with egg intake, a RR (95% CI) of 1.22 (1.03–1.45) for an increment of approximately 2 eggs per day; this association apparently did not differ by menopausal status.20 We found a borderline linear trend toward higher risk of breast cancer with higher egg intake in pre-menopausal women only.

We did not find an association between meat cooked well-done and breast cancer risk in contrast to the positive relationship reported in a retrospective analysis in the Iowa Women Health Study.15 However, our assessment of cooking methods was much more limited.

A potential limitation of any study is that dietary intakes are measured with error. However the same dietary assessment used in our study has been able to detect a doubling of risk of colon cancer with high meat intake compared to low in this same population with only 150 cases and 6 years of follow-up.36 Also, our most complete dietary assessments occurred during mid-life. Although in our study our limited assessment of high school diet was not associated with breast cancer, more thorough evaluations of early life dietary exposures should be undertaken.

Our study had many strengths: it was prospective, included a large number of cases, had little loss to follow-up and was not prone to the biases of case-control studies. We were able to assess dietary intake multiple different ways. We were able to cumulatively average and update dietary intake assessed 5 different times over the 18 year follow-up period, which should reduce random error in long term dietary measurement. We were also able to assess whether baseline diet, simply updated diet and follow-up lagged 8 years was associated with breast cancer. None of these analyses changed the results.

We found no evidence that intake of meat or fish by mid-life and older women over 18 years of follow-up was associated with risk of breast cancer. A modest positive association of egg intake and breast cancer risk in pre-menopausal women should be assessed further.

Acknowledgements

Dr. Holmes was supported by a Career Development award by the National Cancer Institute Specialized Program of Research Excellence (SPORE) in breast cancer at Brigham and Women's Hospital.