Abstract
Background: Prospective epidemiologic studies in Asian populations consistently show that soy is protective against breast cancer.
Objective: The objective was to prospectively evaluate whether the protective effect of soy is due to soy isoflavones alone or to their combination with other beneficial dietary factors in an Asian population.
Design: Using principal components analysis, we previously identified a “meat–dim sum” pattern characterized by meat, starch, and dim sum items and a “vegetable-fruit-soy” pattern characterized by cruciferous vegetables, fruit, and tofu items in a population-based cohort of Singapore Chinese initiated between 1993 and 1998. Component scores representing intakes of each pattern were used in multivariable Cox regression models to analyze the relation between diet at baseline and breast cancer incidence.
Results: As of 31 December 2005, 629 incident breast cancer cases had been diagnosed among the 34,028 women. With greater intake of the vegetable-fruit-soy dietary pattern, we observed a dose-dependent trend (P < 0.01) for decreasing breast cancer risk among postmenopausal women [hazard ratio (HR): 0.70; 95% CI: 0.51, 0.95 for the fourth compared with first quartile]. A stronger association for the vegetable-fruit-soy pattern was observed among postmenopausal women with ≥5 y of follow-up (HR: 0.57; 95% CI: 0.36, 0.88; P for trend <0.01). No trend was observed for a greater intake of the meat–dim sum dietary pattern and increased breast cancer risk.
Conclusion: Our findings support the hypothesis that a diet characterized by vegetables, fruit, and soy has an early-acting protective effect on breast carcinogenesis.
INTRODUCTION
The use of a traditional nutritional epidemiology approach to identify foods and nutrients that are associated with breast cancer risk has yielded inconsistent findings. A comprehensive review of the literature in 1997 concluded that a diet high in vegetables and fruit was protective against breast cancer risk, with suggestive evidence that a diet high in total fat increased risk (1). However, an updated review of prospective studies, conducted largely in Western populations, concluded that there was no evidence for beneficial or adverse effects of dietary factors on breast cancer risk, with the exception of adverse effects with regular alcohol consumption (2). A shift in focus from evaluating individual foods and nutrients, to evaluating dietary patterns, may be more useful in determining whether there is a role for diet in the prevention or development of breast cancer.
As summarized in a recent review of 19 studies that evaluated dietary patterns and breast cancer risk, a range of risk from 0.66 to 0.84 and from 1.27 to 1.42 was reported for protective and detrimental dietary patterns, respectively (3). Overall, protective dietary patterns were characterized by vegetables, fruit, fish, and white meat, whereas detrimental patterns were characterized by high-fat– and high-sugar–containing foods (3). Of particular interest were the inverse associations reported for dietary patterns characterized by traditional diets among populations in Italy (4), Uruguay (5), and particular regions in the United States (eg, southeast and southwest) (6, 7). These findings highlight the importance of evaluating dietary patterns among populations that differ in their risk of breast cancer to identify unique dietary combinations that may be important for breast cancer prevention.
Some of the lowest rates of breast cancer worldwide have been reported in Asian countries (8). Soy foods and their isoflavones have received much attention as potential beneficial factors in the diet in Asian countries, where typical consumption patterns characterized by high lifelong intakes appear to be protective against breast cancer (9–14). In a recent population-based case-control study among Asian Americans in Los Angeles County, a dietary pattern characterized by soy and vegetables was inversely associated with breast cancer (odds ratio: 0.72; 95% CI: 0.54, 0.96) (15). To date, however, no prospective data have evaluated dietary patterns and breast cancer risk in Asian populations (3).
We previously identified 2 dietary patterns in a prospective cohort of Singapore Chinese that we labeled “vegetable-fruit-soy” and “meat–dim sum” (16). On the basis of our earlier finding of an inverse association with soy isoflavones (≥10 mg/d compared with <10 mg/d) and breast cancer risk among Singapore Chinese (17), we hypothesized that the dietary pattern characterized in part by soy foods would be inversely associated with breast cancer. By evaluating the role of these dietary patterns in a population with a historically low risk of breast cancer, such as Singapore Chinese, we will be able to gain further insight into dietary components that may protect against breast cancer development.
SUBJECTS AND METHODS
Study population
The design of the Singapore Chinese Health Study was previously described in detail (18). Briefly, the cohort consisted of 63,257 men and women recruited between April 1993 and December 1998, from permanent residents or citizens of Singapore aged 45–74 y and who resided in government-built housing estates (86% of the Singapore population reside in such facilities). We restricted the study to individuals belonging to the 2 major dialect groups of Chinese in Singapore: the Hokkiens and the Cantonese. Enrollment in the cohort entailed the completion of a baseline in-person interview in the participants’ homes. The questionnaire elicited information on diet, demographics, current physical activity, reproductive history (women only), occupational exposure, and medical history. For these analyses, we used data from the 34,028 women who did not have a history of cancer diagnosis at baseline, based on self-report and linkage with the Singapore Cancer Registry. The Institutional Review Boards at the University of Minnesota and the National University of Singapore approved this study.
Identification of incident breast cancer cases among cohort members were accomplished by record linkage of the cohort database with the population-based Singapore Cancer Registry. The nationwide cancer registry has been in place since 1968 and has been shown to be comprehensive in its recording of cancer cases (19). As of April 2008, only 27 cases were known to be lost to follow-up due to migration from Singapore. As of 31 December 2005, with an average of 10.7 y of follow-up, 629 female cohort participants had developed breast cancer. Histologic and staging information on all breast cancer diagnoses were confirmed by manual review of the pathology reports and clinical charts. On the basis of the same manual review, we were able to obtain estrogen receptor (ER), and/or progesterone receptor (PR) status for 63% (n = 394) of the breast tumors.
Identification of dietary patterns
At baseline, a 165-item quantitative food-frequency questionnaire (FFQ) developed for and validated in this population was administered to assess usual diet over the past year (20). Principal components analysis among women in the cohort was used to identify dietary patterns from the food frequency responses. The number of components retained for orthogonal rotation was based primarily on examination of scree plots and factor interpretability, but the eigenvalues (>1.0) and percentage variance explained were also considered (21). For each component, a score was computed as a linear composite of the foods with meaningful loading scores (eg, ≥0.30). Scores were calculated by taking the unweighted sum of standardized frequencies of intake for each food associated with the pattern. We labeled the 2 distinct dietary patterns: vegetable-fruit-soy and “meat–dim sum (see Supplemental Table 1 under “Supplemental data” in the online issue). Briefly, the vegetable-fruit-soy pattern was characterized by 33 foods including 23 vegetables, 5 soy foods, and 5 fruits. The meat–dim sum pattern contained 27 food items, including 7 meat items, 12 dim sum items (5 meat and 7 “other”), 4 starch items, 3 combined meat-starch items, and 1 egg item. No food items overlapped between the 2 patterns. When we previously conducted principal components analysis among the entire cohort (n = 63,257), we identified essentially the same 2 dietary patterns (16).
To directly compare our findings with those from a previous study among Asian Americans that reported an inverse association for a Mediterranean dietary pattern and breast cancer risk (15), we performed a secondary analysis using the same methods. Briefly, a value of 0 or 1 was assigned to each of 6 presumed beneficial dietary components (ie, vegetables, legumes, fruit/nuts, cereals, fish and seafood, and a high ratio of monounsaturated to saturated dietary fat) based on whether the subject's intake level was below or above the median value (g/1000 kcal) for all subjects. Conversely, a value of 1 or 0 was assigned to each of 4 presumed detrimental dietary components (ie, meat, dairy products, carbohydrates, and alcohol) based on whether the subject's intake level was below or above the median value for all subjects. The individual foods within each component are provided elsewhere (see Supplemental Table 2 under “Supplemental data” in the online issue). The scores across the 10 dietary components were summed to form the Mediterranean Diet Score (range: 0–10), with a high score indicating greater adherence to a typical Mediterranean diet.
Statistical methods
Person-years of follow-up were counted from the date of recruitment to the date of diagnosis of breast cancer, death, migration, or 31 December 2005, whichever occurred first. Proportional hazards regression methods were used to examine the associations between dietary patterns and breast cancer risk, measured by hazard ratios (HRs) and their corresponding 95% CIs (22). We also considered breast cancer subgroups defined by ER status (ER+ or ER−), PR status (PR+ or PR−), and the combination of ER and PR (ER+/PR+, ER−/PR−, ER+/PR−, or ER/PR unknown). There were too few cases (n = 9) to evaluate the ER−/PR+ combination. Dietary patterns were evaluated by using quartile variables with cutoffs based on the distribution among women in the baseline cohort.
The linear trend tests for dietary patterns–breast cancer associations were based on median values of the ordinal variable categories (0, 1, 2, and 3). In all analyses, we adjusted for the following potential confounders: age at baseline interview (y), year of interview (1993–1995 or 1996–1998), dialect group (Cantonese or Hokkien), level of education (no formal education, primary school, or secondary school or higher), one or more first-degree relatives with breast cancer (no or yes), number of full-term births (0, 1–2, 3–4, or ≥5), body mass index [BMI; in kg/m2: < or ≥ 23.2 (median)], and total caloric intake (quartiles). Additional inclusion of the following variables did not materially change any of the study results: menopausal status at baseline, ever use of menopausal hormones, age when period became regular, weekly physical activity, alcohol consumption, smoking history, and dietary n−3 polyunsaturated fatty acid (PUFA), n−6 PUFA, folate, soy isoflavone, and green tea intakes.
On the basis of previous analyses of our data and reports from the literature, we examined whether the association between dietary patterns and breast cancer varied by the following factors: menopausal status at baseline (pre- or postmenopausal), median BMI (< or ≥23.2), education (no formal education/primary school or secondary school or higher), and smoking status (ever/never) by stratified analyses and assessing the fitness of interaction terms in adjusted models. Statistical computing was conducted by using SAS version 9.2 (SAS Institute Inc, Cary, NC). All P values were 2-sided and were considered statistically significant if <0.05.
RESULTS
At baseline, 72% of our cohort was postmenopausal and had a median age of 55 y [interquartile range (IQR) = 13 y]. Younger median age, higher education, and being premenopausal were correlated with greater intakes of both the meat–dim sum and vegetable-fruit-soy dietary patterns (Table 1). The population was thin, with a median BMI of 23.2 (IQR = 3.5) and with no BMI variation by dietary pattern intake. Overall, few women ever smoked (8.7%) and most did not drink alcohol (90.9%), with fewer ever smokers and fewer nondrinkers among the fourth quartiles of the vegetable-fruit-soy and meat–dim sum dietary patterns, respectively (Table 1). Prevalence of moderate physical activity of ≥0.5 h/wk increased by increasing quartiles of vegetable-fruit-soy pattern intake, but did not vary much by quartiles of meat–dim sum pattern intake (Table 1). The prevalence of regular (ie, at least weekly) green and black tea drinkers was similar among the highest quartiles of both dietary patterns (Table 1).
TABLE 1.
Baseline characteristics by dietary pattern quartile (Q)1
| Meat–dim sum |
Vegetable-fruit-soy |
|||||||
| Q1 | Q2 | Q3 | Q4 | Q1 | Q2 | Q3 | Q4 | |
| No. of subjects | 8467 | 8491 | 8531 | 8539 | 8506 | 8527 | 8513 | 8482 |
| Person-years | 84,145 | 84,684 | 84,832 | 84,581 | 84,287 | 85,297 | 84,885 | 83,772 |
| Age (y) | 59 (13)2 | 56 (13) | 54 (12) | 52 (11) | 57 (14) | 55 (12) | 54 (12) | 54 (12) |
| Hokkien dialect group (%) | 54.6 | 52.8 | 51.1 | 50.1 | 57.0 | 53.1 | 49.6 | 48.9 |
| Highest level of education: secondary education or higher (%) | 14.3 | 18.8 | 21.9 | 27.7 | 13.8 | 18.2 | 23.2 | 27.0 |
| BMI (kg/m2) | 23.3 (3.0) | 23.2 (3.5) | 23.2 (3.5) | 23.2 (3.7) | 23.3 (3.1) | 23.3 (3.4) | 23.2 (3.6) | 23.1 (3.8) |
| Menopausal status (% premenopausal) | 19.2 | 25.2 | 31.0 | 37.3 | 22.7 | 27.3 | 30.2 | 32.6 |
| Ever use of menopausal hormones (%) | 4.9 | 5.4 | 6.0 | 6.2 | 3.9 | 5.4 | 6.3 | 6.9 |
| Number of full-term births (%) | ||||||||
| 0 | 7.7 | 6.5 | 6.9 | 7.1 | 7.8 | 7.2 | 6.3 | 6.7 |
| 1–2 | 24.0 | 27.6 | 28.7 | 32.4 | 26.4 | 26.8 | 29.0 | 30.5 |
| 3–4 | 34.1 | 34.1 | 36.9 | 38.7 | 32.8 | 37.2 | 39.2 | 39.2 |
| ≥5 | 34.2 | 29.0 | 25.8 | 21.8 | 33.0 | 28.7 | 25.6 | 23.6 |
| Family history: first-degree relative with diagnosis of breast cancer (%) | 1.1 | 1.3 | 1.3 | 1.5 | 1.2 | 1.2 | 1.3 | 1.5 |
| Ever smoker (%) | 9.3 | 8.5 | 8.4 | 8.7 | 12.5 | 9.0 | 6.9 | 6.5 |
| Alcohol intake (% nondrinkers) | 95.1 | 92.6 | 90.2 | 85.8 | 92.2 | 91.1 | 91.0 | 89.4 |
| Physical activity ≥0.5 h/wk (%) | 21.0 | 20.6 | 19.9 | 18.7 | 14.6 | 18.8 | 21.6 | 25.2 |
| Tea intake (%) | ||||||||
| Nondrinkers | 41.5 | 49.6 | 55.3 | 63.0 | 41.2 | 49.6 | 56.0 | 62.6 |
| Green tea, at least weekly | 19.3 | 23.9 | 26.8 | 32.2 | 15.6 | 22.6 | 28.5 | 35.6 |
| Black tea, at least weekly | 14.4 | 19.5 | 22.5 | 31.3 | 17.0 | 19.4 | 22.8 | 28.7 |
| Daily energy intake (kcal) | 1105 (422) | 1230 (454) | 1375 (489) | 1683 (638) | 1085 (446) | 1248 (451) | 1384 (493) | 1645 (625) |
| Daily intake | ||||||||
| Red meat (g/1000 kcal) | 11.0 (12.7) | 14.5 (12.6) | 17.5 (12.2) | 22.0 (13.4) | 16.1 (14.1) | 16.8 (13.6) | 16.8 (14.1) | 16.0 (13.9) |
| Preserved red meat (g/1000 kcal) | 0.06 (0.8) | 0.5 (1.4) | 1.0 (1.9) | 1.7 (2.7) | 0.5 (1.5) | 0.7 (1.8) | 0.8 (2.0) | 1.0 (2.3) |
| Poultry (g/1000 kcal) | 5.3 (8.7) | 9.3 (9.6) | 11.8 (10.6) | 14.7 (11.7) | 9.3 (11.0) | 10.4 (11.0) | 10.9 (11.3) | 10.9 (11.9) |
| Fish (g/1000 kcal) | 35.3 (26.7) | 36.2 (23.2) | 35.9 (21.7) | 34.9 (19.3) | 33.7 (24.5) | 35.7 (21.6) | 36.5 (21.4) | 36.4 (22.5) |
| Total vegetables (g/1000 kcal) | 74.7 (47.4) | 74.0 (44.1) | 73.6 (42.0) | 70.2 (38.2) | 51.1 (27.8) | 66.9 (30.7) | 80.0 (36.0) | 100.4 (48.2) |
| Cruciferous vegetables (g/1000 kcal) | 29.0 (21.7) | 28.6 (21.1) | 28.2 (20.0) | 25.5 (17.9) | 20.2 (14.9) | 25.7 (16.3) | 30.0 (19.1) | 37.1 (24.2) |
| Total fruit (g/1000 kcal) | 116.2 (133.8) | 116.7 (123.0) | 118.0 (116.5) | 118.5 (110.1) | 68.3 (93.7) | 106.1 (101.9) | 132.4 (110.5) | 164.5 (130.1) |
| Total soy isoflavones (mg/1000 kcal) | 8.8 (10.0) | 10.2 (10.1) | 11.0 (10.0) | 12.1 (9.7) | 6.9 (7.5) | 9.6 (8.4) | 11.8 (9.5) | 14.9 (11.7) |
| Dietary fiber (g/1000 kcal) | 8.8 (4.1) | 8.4 (3.5) | 8.4 (3.3) | 8.1 (2.9) | 6.9 (2.9) | 8.0 (3.0) | 8.7 (3.0) | 9.9 (3.3) |
| Folate (μg /1000 kcal) | 97.9 (43.1) | 98.3 (40.1) | 98.7 (38.6) | 99.3 (35.8) | 80.7 (31.3) | 92.9 (32.0) | 103.2 (34.7) | 117.5 (39.7) |
| Saturated fat (g/1000 kcal) | 7.5 (3.3) | 8.4 (3.2) | 9.1 (3.1) | 10.5 (3.1) | 8.0 (3.3) | 8.7 (3.3) | 9.1 (3.4) | 9.8 (3.7) |
| Marine n−3 PUFAs, (g/1000 kcal) | 0.18 (0.14) | 0.19 (0.12) | 0.19 (0.11) | 0.18 (0.10) | 0.17 (0.13) | 0.18 (0.11) | 0.19 (0.11) | 0.19 (0.12) |
| n−6 PUFAs (g/1000 kcal) | 4.0 (2.4) | 4.2 (2.3) | 4.4 (2.1) | 4.7 (2.0) | 3.7 (1.9) | 4.1 (1.9) | 4.5 (2.3) | 5.0 (2.6) |
PUFAs, polyunsaturated fatty acids.
Median; interquartile range in parentheses (all such values).
Differences between the dietary patterns were more evident when evaluating distribution of intake by foods and nutrients (Table 1). For example, median intake of red meat, preserved red meat, and poultry were higher among those in the fourth quartile of meat–dim sum pattern than in those in the fourth quartile of the vegetable-fruit-soy pattern. In contrast, the median intake of vegetables, fruit, soy isoflavones, dietary fiber, and folate were higher among those in the fourth quartile of the vegetable-fruit-soy pattern than among those in the fourth quartile of the meat–dim sum pattern. There were nominal differences in median total energy, fatty acid subgroups, or fish in comparisons of levels in the fourth quartiles of each dietary pattern.
With greater intake of the vegetable-fruit-soy dietary pattern, we observed a dose-dependent trend for decreasing breast cancer risk (Table 2). The inverse association was confined to postmenopausal women at baseline, where we observed a statistically significant, 30% decrease in risk for the fourth compared with first quartile of vegetable-fruit-soy pattern intake. There was no trend observed between greater intake of the meat–dim sum dietary pattern and breast cancer risk, regardless of menopausal status at baseline (Table 2). Stratified analyses by duration of follow-up showed a statistically significant trend with the vegetable-fruit-soy pattern and lower postmenopausal breast cancer risk among those with ≥5 y of follow-up (Table 3).
TABLE 2.
Hazard ratios (HRs) by dietary pattern quartile (Q) in relation to breast cancer by menopausal status at baseline
| Overall |
Premenopausal |
Postmenopausal |
||||
| No. of cases | HR (95% CI)1 | No. of cases | HR (95% CI)1 | No. of cases | HR (95% CI)1 | |
| Vegetable-fruit-soy | ||||||
| Q1 | 153 | 1.0 (reference) | 35 | 1.0 (reference) | 118 | 1.0 (reference) |
| Q2 | 182 | 1.12 (0.90, 1.40) | 52 | 1.22 (0.79, 1.89) | 130 | 1.08 (0.84, 1.39) |
| Q3 | 150 | 0.89 (0.70, 1.12) | 44 | 0.92 (0.58, 1.46) | 106 | 0.86 (0.65, 1.14) |
| Q4 | 144 | 0.82 (0.63, 1.05) | 59 | 1.09 (0.68, 1.73) | 85 | 0.70 (0.51, 0.95) |
| P for trend2 | 0.03 | 0.91 | 0.01 | |||
| Meat–dim sum | ||||||
| Q1 | 144 | 1.0 (reference) | 36 | 1.0 (reference) | 108 | 1.0 (reference) |
| Q2 | 166 | 1.12 (0.89, 1.40) | 45 | 0.98 (0.63, 1.52) | 121 | 1.17 (0.90, 1.51) |
| Q3 | 182 | 1.18 (0.94, 1.49) | 51 | 0.90 (0.58, 1.40) | 131 | 1.32 (1.01, 1.72) |
| Q4 | 137 | 0.84 (0.65, 1.10) | 58 | 0.79 (0.50, 1.26) | 79 | 0.85 (0.62, 1.17) |
| P for trend2 | 0.35 | 0.28 | 0.73 | |||
All HRs were adjusted for age at interview (y), dialect group (Cantonese or Hokkien), interview year (1993–1995 or 1996–1998), education (no formal education, primary school, or secondary school or higher), parity (0, 1–2, 3–4, or ≥5 births), BMI [in kg/m2; < or ≥23.2 (median)], first-degree relative with diagnosis of breast cancer (no or yes), and total daily energy intake (quartiles).
The linear trend tests for dietary patterns–breast cancer associations were based on median values of the ordinal variable categories (0, 1, 2, and 3).
TABLE 3.
Hazard ratios (HRs) by vegetable-fruit-soy dietary pattern quartile (Q) and duration of follow-up among postmenopausal women at baseline
| 0 to <5 y |
≥5 y |
|||||
| No. of cases | HR (95% CI)1 | P for trend2 | No. of cases | HR (95% CI)1 | P for trend2 | |
| Q1 | 58 | 1.0 (reference ) | — | 60 | 1.0 (reference ) | — |
| Q2 | 54 | 0.95 (0.65, 1.38) | — | 79 | 1.20 (0.85, 1.70) | — |
| Q3 | 47 | 0.83 (0.55, 1.24) | — | 59 | 0.90 (0.61, 1.31) | — |
| Q4 | 46 | 0.86 (0.56, 1.32) | 0.39 | 39 | 0.57 (0.36, 0.88) | <0.01 |
All HRs were adjusted for age at interview (y), dialect group (Cantonese or Hokkien), interview year (1993–1995 or 1996–1998), education (no formal education, primary school, or secondary school or higher), parity (0, 1–2, 3–4, or ≥5 births), BMI [in kg/m2; < or ≥23.2 (median)], first-degree relative with diagnosis of breast cancer (no or yes), and total daily energy intake (quartiles).
The linear trend tests for dietary patterns–breast cancer associations were based on median values of the ordinal variable categories (0, 1, 2, and 3).
Associations between the vegetable-fruit-soy pattern and breast cancer risk did not differ by ER/PR status. Results for ER+/PR− tumors are shown elsewhere (see Supplemental Table 3 under “Supplemental data” in the online issue). There was no evidence of effect modification by BMI, smoking status, or education level for the vegetable-fruit-soy dietary pattern–breast cancer association, overall or for postmenopausal breast cancer (data not shown). Additionally, we evaluated whether the most prominent food groups in the vegetable-fruit-soy pattern were associated with postmenopausal breast cancer and found no association with total vegetable, total fruit, soy food, or soy isoflavone intake (Table 4). We also found no association with greater adherence to a Mediterranean diet and breast cancer (HR: 0.96, 95% CI: 0.76, 1.21, for a dietary score ≥7 compared with 0–4).
TABLE 4.
Hazard ratios (HRs) for major food group contributors to the vegetable-fruit-soy dietary pattern and breast cancer among postmenopausal women by pattern quartile (Q)
| Q1 | Q2 | Q3 | Q4 | P for trend1 | |
| Total vegetables | |||||
| Median value (g/d) | 51.0 | 81.9 | 114.5 | 173.7 | — |
| HR (95% CI)2 | 1.0 (ref) | 0.91 (0.70, 1.18) | 0.81 (0.61, 1.07) | 0.86 (0.63, 1.16) | 0.22 |
| Total fruit | |||||
| Median value (g/d) | 39.0 | 120.2 | 201.0 | 357.0 | — |
| HR (95% CI)2 | 1.0 (ref) | 0.89 (0.68, 1.17) | 1.01 (0.76, 1.33) | 1.03 (0.77, 1.38) | 0.64 |
| Soy foods | |||||
| Median value (g/d) | 30.4 | 65.9 | 108.6 | 196.4 | — |
| HR (95% CI)2 | 1.0 (ref) | 0.94 (0.72, 1.21) | 0.74 (0.56, 0.98) | 0.83 (0.62, 1.11) | 0.09 |
| Soy isoflavones | |||||
| Median value (mg/d) | 4.6 | 10.6 | 18.1 | 33.9 | — |
| HR (95% CI)2 | 1.0 (ref) | 1.07 (0.83, 1.39) | 0.84 (0.64, 1.12) | 0.86 (0.64, 1.16) | 0.15 |
The linear trend tests for dietary patterns–breast cancer associations were based on median values of the ordinal variable categories (0, 1, 2, and 3).
All HRs were adjusted for age at interview (y), dialect group (Cantonese or Hokkien), interview year (1993–1995 or 1996–1998), education (no formal education, primary school, or secondary school or higher), parity (0, 1–2, 3–4, or ≥5 births), BMI [in kg/m2; < or ≥23.2 (median)], first-degree relative with diagnosis of breast cancer (no or yes), and total daily energy intake (quartiles).
DISCUSSION
From prospective analyses using data from a Singapore Chinese cohort, we observed a trend of decreasing breast cancer risk with increasing intake of a vegetable-fruit-soy dietary pattern among postmenopausal women. We also report stronger effects for the vegetable-fruit-soy dietary pattern among postmenopausal women with longer follow-up. Our findings lend further support for the hypothesis that certain dietary factors protect against breast cancer and that early-acting mechanisms on breast carcinogenesis may be involved (23–25).
All of the previously reported results from prospective analyses of dietary patterns and breast cancer were conducted among Western populations and most reported no overall association for a prudent-type dietary pattern characterized by vegetable, fruit, legumes, and sometimes fish intake (3, 26). However, inverse associations for prudent dietary patterns and breast cancer were reported for some subgroups, such as ever smokers (27), women with lower body size (4), and for ER− (28, 29) and ER+/PR− tumors (30). There was some evidence from a prospective cohort study by Velie et al (6) for an inverse association between postmenopausal breast cancer and a regional US southern traditional diet characterized in part by cooked greens, legumes, sweet potatoes, and cabbage. Their finding for the southern traditional diet supports other epidemiologic studies that found protective effects against breast cancer with a diet high in cruciferous vegetables (31–33).
Cruciferous vegetables and soy items are characteristic of the prudent-type dietary patterns identified in Asian populations (15, 34–37), including our Singapore Chinese cohort (16), unlike the prudent dietary patterns typically identified in Western populations (38). Cruciferous vegetables, as the main dietary sources of the glucosinolate precursors of isothiocyanates, may be beneficial for preventing breast cancer, in part because of their capacity as substrates for and inducers of the glutathione S-transferases M1 and T1 detoxification enzymes (39, 40). Soy isoflavones may reduce breast cancer risk by estrogen-dependent mechanisms related to preferential binding to ER-β relative to ER-α (41, 42). Soy isoflavones may also act via estrogen-independent mechanisms such as inhibition of nuclear transcription factor κB DNA-binding activity and the Akt signaling pathway (43), both of which are important for maintaining a homeostatic balance between cell survival and apoptosis (44, 45).
Although no prospective data from Asian populations have been previously reported for dietary patterns and breast cancer, conflicting results have been reported from case-control studies conducted in China (34), in Japan (35), and among Asian Americans in Los Angeles, CA (15). A population-based case-control study conducted among women in Shanghai did not report an overall association between a vegetable-soy pattern and breast cancer (34), whereas the hospital-based case-control study in Japan, as well as the population-based case-control study among Asian Americans, both reported modest, statistically significant inverse associations between a vegetable-soy dietary pattern and breast cancer (15, 35). Our inverse association for the vegetable-fruit-soy pattern and postmenopausal breast cancer supports the findings in earlier studies among Asian American and Japanese women.
We did not observe a trend of increasing breast cancer risk with our meat–dim sum pattern, despite the evidence of an adverse effect with a Western pattern, characterized by red and processed meats, sweets and desserts, and refined grains (5, 7, 28, 46). In addition, modest positive associations were reported among Asian Americans for a greater intake of an ethnic-meat-starch dietary pattern (15) and among Chinese in Shanghai for a meat-sweets dietary pattern (34), but not among Japanese for a “fatty” dietary pattern that was characterized primarily by red meat intake (35). Similar to the Japanese study population (35), the mean intake of red meat among Singapore Chinese women was much lower (16.4 g/1000 kcal/d) than the mean amounts consumed among the Asian American population (34.5 g/1000 kcal/d) (15). It also remains a possibility that we did not detect an association with the meat–dim sum pattern, because the degree of measurement error inherent in dietary variables was too great for us to observe a small positive association, if one truly existed.
The strongest inverse associations by ER/PR status for the vegetable-fruit-soy dietary pattern and postmenopausal breast cancer were for ER+/PR− tumors. Our findings support an inverse association reported from a prospective cohort of postmenopausal French women for a prudent pattern and ER+/PR− tumors (30), but do not support results from 2 prospective US studies that found inverse associations for ER− tumors (28, 29). Our findings, however, should be interpreted cautiously because we only had ER or PR status for 63% of the cases. Differences in lifetime patterns of exposure may contribute to the differences observed in the above studies between the prudent dietary pattern and ER+ compared with ER− tumors, assuming the associations were real.
The strengths of our study included the use of an FFQ that was created for and validated in our study population (20). A comparison of mean daily intakes of energy and major nutrients between the FFQ and multiple 24-h dietary recalls showed that the intakes were within 10% of each other (20). We also reported strong correlations between urinary soy isoflavones and self-reported soy intake (47) and between urinary and dietary isothiocyanates (48). Another strength included the prospective design of our study, which allowed the inference of a temporal relation between our vegetable-fruit-soy dietary pattern and a decreased risk of postmenopausal breast cancer.
Although the use of a principal components analysis to identify dietary patterns is a valuable tool in nutritional epidemiology as an alternative to evaluating individual foods and nutrients, it has limitations (3, 49). For example, the lack of dietary pattern reproducibility and the subjective nature of determining the number of patterns have been noted as limitations, but neither were sources of bias in our study, as we showed previously (16). Another notable limitation with the use of principal components analysis is that it is an a posteri method. We evaluated whether an a priori–determined Mediterranean dietary pattern was related to breast cancer in our data, based on recent findings that the Mediterranean diet may be protective against breast cancer among Asian Americans (15). We found no association with breast cancer for women with higher intakes of a Mediterranean diet.
It is possible that our observed inverse association between the vegetable-fruit-soy dietary pattern and breast cancer was due to correlations with nondietary aspects of a healthy lifestyle that are beneficial for breast cancer, such as being physically active, not smoking, and maintaining a healthy weight. Alternatively, our finding could represent a true combination of potentially beneficial dietary components for breast cancer, such as less red meat and alcohol and more cruciferous vegetables, soy isoflavones, folate, and dietary fiber. Finally, given the dichotomy between our 2 dietary patterns, that is, women who consume high amounts of vegetables, fruit, and soy items are inherently not likely to consume high amounts of meat and dim sum items, there remains a possibility that a diet high in meat and dim sum (ie, high starch, saturated fat, and sodium-containing snacks) has a small yet adverse effect on breast cancer risk that was not observed in our data.
In summary, we report a trend of decreasing postmenopausal breast cancer risk with greater intakes of a dietary pattern characterized by cruciferous vegetables, fruit, and soy in a prospective cohort of Singapore Chinese. We also observed a stronger inverse association among postmenopausal women with ≥5 y of follow-up. This protective effect was consistent with our previous finding that moderate soy isoflavone intake was inversely associated with breast cancer, especially among women with longer follow-up (17). Together, these findings support the beneficial effect of a diet characterized, in part, by soy and in cruciferous vegetables and fruit on early stages of breast carcinogenesis.
Supplementary Material
Acknowledgments
We thank Siew-Hong Low of the National University of Singapore for supervising the fieldwork of the Singapore Chinese Health Study and Kazuko Arakawa for the development of the cohort study database. We also thank the Singapore Cancer Registry for assistance with the identification of cancer outcomes.
The authors’ responsibilities were as follows—LMB and RW: conducted the statistical analyses; LMB and AWH: wrote the manuscript; and LMB, AHW, W-PK, J-MY, and MCY: interpreted the data and worked on subsequent drafts of the manuscript. J-MY and MCY were Principal Investigators of the Singapore Chinese Health Study. None of the authors had a conflict of interest.
REFERENCES
- 1.World Cancer Research Fund/American Institute for Cancer Research Food, nutrition and the prevention of cancer: a global perspective. Washington, DC: American Institute for Cancer Research, 1997 [DOI] [PubMed] [Google Scholar]
- 2.Michels KB, Mohllajee AP, Roset-Bahmanyar E, Beehler GP, Moysich KB. Diet and breast cancer: a review of the prospective observational studies. Cancer 2007;109:2712–49 [DOI] [PubMed] [Google Scholar]
- 3.Edefonti V, Randi G, La Vecchia C, Ferraroni M, Decarli A. Dietary patterns and breast cancer: a review with focus on methodological issues. Nutr Rev 2009;67:297–314 [DOI] [PubMed] [Google Scholar]
- 4.Sieri S, Krogh V, Pala V, et al. Dietary patterns and risk of breast cancer in the ORDET cohort. Cancer Epidemiol Biomarkers Prev 2004;13:567–72 [PubMed] [Google Scholar]
- 5.Ronco AL, De Stefani E, Boffetta P, Deneo-Pellegrini H, Acosta G, Mendilaharsu M. Food patterns and risk of breast cancer: a factor analysis study in Uruguay. Int J Cancer 2006;119:1672–8 [DOI] [PubMed] [Google Scholar]
- 6.Velie EM, Schairer C, Flood A, He JP, Khattree R, Schatzkin A. Empirically derived dietary patterns and risk of postmenopausal breast cancer in a large prospective cohort study. Am J Clin Nutr 2005;82:1308–19 [DOI] [PubMed] [Google Scholar]
- 7.Murtaugh MA, Sweeney C, Giuliano AR, et al. Diet patterns and breast cancer risk in Hispanic and non-Hispanic white women: the Four-Corners Breast Cancer Study. Am J Clin Nutr 2008;87:978–84 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Althuis MD, Dozier JM, Anderson WF, Devesa SS, Brinton LA. Global trends in breast cancer incidence and mortality 1973-1997. Int J Epidemiol 2005;34:405–12 [DOI] [PubMed] [Google Scholar]
- 9.Wu AH, Yu MC, Tseng CC, Pike MC. Epidemiology of soy exposures and breast cancer risk. Br J Cancer 2008;98:9–14 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Wu AH, Wan P, Hankin J, Tseng CC, Yu MC, Pike MC. Adolescent and adult soy intake and risk of breast cancer in Asian-Americans. Carcinogenesis 2002;23:1491–6 [DOI] [PubMed] [Google Scholar]
- 11.Shu XO, Jin F, Dai Q, et al. Soyfood intake during adolescence and subsequent risk of breast cancer among Chinese women. Cancer Epidemiol Biomarkers Prev 2001;10:483–8 [PubMed] [Google Scholar]
- 12.Thanos J, Cotterchio M, Boucher BA, Kreiger N, Thompson LU. Adolescent dietary phytoestrogen intake and breast cancer risk (Canada). Cancer Causes Control 2006;17:1253–61 [DOI] [PubMed] [Google Scholar]
- 13.Lee SA, Shu XO, Li H, et al. Adolescent and adult soy food intake and breast cancer risk: results from the Shanghai Women's Health Study. Am J Clin Nutr 2009;89:1920–6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Korde LA, Wu AH, Fears T, et al. Childhood soy intake and breast cancer risk in Asian American women. Cancer Epidemiol Biomarkers Prev 2009;18:1050–9 [DOI] [PubMed] [Google Scholar]
- 15.Wu AH, Yu MC, Tseng CC, Stanczyk FZ, Pike MC. Dietary patterns and breast cancer risk in Asian American women. Am J Clin Nutr 2009;89:1145–54 [DOI] [PubMed] [Google Scholar]
- 16.Butler LM, Koh WP, Lee HP, Tseng M, Yu MC, London SJ. Prospective study of dietary patterns and persistent cough with phlegm among Chinese Singaporeans. Am J Respir Crit Care Med 2006;173:264–70 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Wu AH, Koh WP, Wang R, Lee HP, Yu MC. Soy intake and breast cancer risk in Singapore Chinese Health Study. Br J Cancer 2008;99:196–200 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Yuan JM, Stram DO, Arakawa K, Lee HP, Yu MC. Dietary cryptoxanthin and reduced risk of lung cancer: the Singapore Chinese Health Study. Cancer Epidemiol Biomarkers Prev 2003;12:890–8 [PubMed] [Google Scholar]
- 19.Parkin DM, Whelan SL, Ferlay J, Teppo L, Thomas D. Cancer incidence in five continents. Lyon, France:International Agency for Research on Cancer, 2003 [Google Scholar]
- 20.Hankin JH, Stram DO, Arakawa K, et al. Singapore Chinese Health Study: development, validation, and calibration of the quantitative food frequency questionnaire. Nutr Cancer 2001;39:187–95 [DOI] [PubMed] [Google Scholar]
- 21.Hatcher L. A step-by-step approach to using SAS for factor analysis and structural equation modeling. Cary, NC: SAS Institute Inc, 1994 [Google Scholar]
- 22.Cox D. Regression models and life tables. J R Stat Soc 1972;B:187–220 [Google Scholar]
- 23.Jin Z, MacDonald RS. Soy isoflavones increase latency of spontaneous mammary tumors in mice. J Nutr 2002;132:3186–90 [DOI] [PubMed] [Google Scholar]
- 24.Constantinou AI, Krygier AE, Mehta RR. Genistein induces maturation of cultured human breast cancer cells and prevents tumor growth in nude mice. Am J Clin Nutr 1998;68(suppl):1426S–30S [DOI] [PubMed] [Google Scholar]
- 25.Giri AK, Lu LJ. Genetic damage and the inhibition of 7,12-dimethylbenz[a]anthracene-induced genetic damage by the phytoestrogens, genistein and daidzein, in female ICR mice. Cancer Lett 1995;95:125–33 [DOI] [PubMed] [Google Scholar]
- 26.Engeset D, Dyachenko A, Ciampi A, Lund E. Dietary patterns and risk of cancer of various sites in the Norwegian European Prospective Investigation into Cancer and Nutrition cohort: the Norwegian Women and Cancer study. Eur J Cancer Prev 2009;18:69–75 [DOI] [PubMed] [Google Scholar]
- 27.Adebamowo CA, Hu FB, Cho E, Spiegelman D, Holmes MD, Willett WC. Dietary patterns and the risk of breast cancer. Ann Epidemiol 2005;15:789–95 [DOI] [PubMed] [Google Scholar]
- 28.Fung TT, Hu FB, Holmes MD, et al. Dietary patterns and the risk of postmenopausal breast cancer. Int J Cancer 2005;116:116–21 [DOI] [PubMed] [Google Scholar]
- 29.Agurs-Collins T, Rosenberg L, Makambi K, Palmer JR, Adams-Campbell L. Dietary patterns and breast cancer risk in women participating in the Black Women's Health Study. Am J Clin Nutr 2009;90:621–8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Cottet V, Touvier M, Fournier A, et al. Postmenopausal breast cancer risk and dietary patterns in the E3N-EPIC prospective cohort study. Am J Epidemiol 2009;170:1257–67 [DOI] [PubMed] [Google Scholar]
- 31.Terry P, Wolk A, Persson I, Magnusson C. Brassica vegetables and breast cancer risk. JAMA 2001;285:2975–7 [DOI] [PubMed] [Google Scholar]
- 32.Lee SA, Fowke JH, Lu W, et al. Cruciferous vegetables, the GSTP1 Ile105Val genetic polymorphism, and breast cancer risk. Am J Clin Nutr 2008;87:753–60 [DOI] [PubMed] [Google Scholar]
- 33.Zhang CX, Ho SC, Chen YM, Fu JH, Cheng SZ, Lin FY. Greater vegetable and fruit intake is associated with a lower risk of breast cancer among Chinese women. Int J Cancer 2009;125:181–8 [DOI] [PubMed] [Google Scholar]
- 34.Cui X, Dai Q, Tseng M, Shu XO, Gao YT, Zheng W. Dietary patterns and breast cancer risk in the shanghai breast cancer study. Cancer Epidemiol Biomarkers Prev 2007;16:1443–8 [DOI] [PubMed] [Google Scholar]
- 35.Hirose K, Matsuo K, Iwata H, Tajima K. Dietary patterns and the risk of breast cancer in Japanese women. Cancer Sci 2007;98:1431–8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Cai H, Zheng W, Xiang YB, et al. Dietary patterns and their correlates among middle-aged and elderly Chinese men: a report from the Shanghai Men's Health Study. Br J Nutr 2007;98:1006–13 [DOI] [PubMed] [Google Scholar]
- 37.Cai H, Shu XO, Gao YT, Li H, Yang G, Zheng W. A prospective study of dietary patterns and mortality in Chinese women. Epidemiology 2007;18:393–401 [DOI] [PubMed] [Google Scholar]
- 38.Tseng M, DeVellis RF. Fundamental dietary patterns and their correlates among US whites. J Am Diet Assoc 2001;101:929–32 [DOI] [PubMed] [Google Scholar]
- 39.Hecht SS. Chemoprevention of cancer by isothiocyanates, modifiers of carcinogen metabolism. J Nutr 1999;129:768S–74S [DOI] [PubMed] [Google Scholar]
- 40.Grubbs CJ, Steele VE, Casebolt T, et al. Chemoprevention of chemically-induced mammary carcinogenesis by indole-3-carbinol. Anticancer Res 1995;15:709–16 [PubMed] [Google Scholar]
- 41.Kuiper GG, Carlsson B, Grandien K, et al. Comparison of the ligand binding specificity and transcript tissue distribution of estrogen receptors alpha and beta. Endocrinology 1997;138:863–70 [DOI] [PubMed] [Google Scholar]
- 42.Strom A, Hartman J, Foster JS, Kietz S, Wimalasena J, Gustafsson JA. Estrogen receptor beta inhibits 17beta-estradiol-stimulated proliferation of the breast cancer cell line T47D. Proc Natl Acad Sci USA 2004;101:1566–71 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Gong L, Li Y, Nedeljkovic-Kurepa A, Sarkar FH. Inactivation of NF-kappaB by genistein is mediated via Akt signaling pathway in breast cancer cells. Oncogene 2003;22:4702–9 [DOI] [PubMed] [Google Scholar]
- 44.Van Antwerp DJ, Martin SJ, Kafri T, Green DR, Verma IM. Suppression of TNF-alpha-induced apoptosis by NF-kappaB. Science 1996;274:787–9 [DOI] [PubMed] [Google Scholar]
- 45.Brunet A, Bonni A, Zigmond MJ, et al. Akt promotes cell survival by phosphorylating and inhibiting a Forkhead transcription factor. Cell 1999;96:857–68 [DOI] [PubMed] [Google Scholar]
- 46.Schulz M, Hoffmann K, Weikert C, Nothlings U, Schulze MB, Boeing H. Identification of a dietary pattern characterized by high-fat food choices associated with increased risk of breast cancer: the European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam Study. Br J Nutr 2008;100:942–6 [DOI] [PubMed] [Google Scholar]
- 47.Seow A, Shi CY, Franke AA, Hankin JH, Lee HP, Yu MC. Isoflavonoid levels in spot urine are associated with frequency of dietary soy intake in a population-based sample of middle-aged and older Chinese in Singapore. Cancer Epidemiol Biomarkers Prev 1998;7:135–40 [PubMed] [Google Scholar]
- 48.Seow A, Shi CY, Chung FL, et al. Urinary total isothiocyanate (ITC) in a population-based sample of middle-aged and older Chinese in Singapore: relationship with dietary total ITC and glutathione S-transferase M1/T1/P1 genotypes. Cancer Epidemiol Biomarkers Prev 1998;7:775–81 [PubMed] [Google Scholar]
- 49.Martinez ME, Marshall JR, Sechrest L. Invited commentary: factor analysis and the search for objectivity. Am J Epidemiol 1998;148:17–9 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.