ω-3 and ω-6 Polyunsaturated Fatty Acid Intakes and the Risk of Breast Cancer in Mexican Women: Impact of Obesity Status

Breast cancer is the most frequent cancer among women with an estimated 1.38 million new cancer cases diagnosed in 2008 (23% of all cancers) and ranks second overall (10.9% of all cancers). It is now the most common cancer both in developed and developing regions with around 6,90,000 new cases estimated in each region (1). In Mexico, the estimated age-standardized incidence of breast cancer is 38.4 per 1,00,000 women (1). The increased incidence observed in Mexico during the last 20 years is linked in part to changes in the lifestyle of women, such as later age at first pregnancy, decreasing duration of lactation, more sedentary lifestyle, and diet (2).

Risk factors related to diet, obesity, and physical activity are often blamed for increasing breast cancer rates. High fat intake, high carbohydrate intake, low vegetable intake, and low soy intake have all been implicated, but the data are inconclusive (3). The role of fat intake in breast cancer etiology has been investigated for long but still remains one of the most controversial hypothesis in nutritional epidemiology (4). Experimental studies suggested strong tumor-enhancing effects of ω-6 polyunsaturated fatty acids (PUFA) whereas protective effects of ω-3 PUFA, present at high levels in fish oils, on mammary carcinogenesis, underlying the need to distinguish between the effects of ω-6 and ω-3 PUFA (5). Additional experimental studies suggest that high intakes of ω-3 PUFA could exert inhibitory effects on mammary tumorigenesis through competition with ω-6 PUFA (6). Meta-analysis of epidemiologic studies reported a significant increase in breast cancer risk with high saturated fat intake but failed to observe significant association with total PUFA (7) or ω-3 PUFA intakes (8). The hypothesis of a protective effect of ω-3 PUFA on breast cancer risk deserves further consideration in epidemiologic studies.

We analyzed the relationship of breast cancer risk to PUFA intake in a case–control study conducted in Mexico City. The analysis focused on ω-3 PUFA intakes which have been hypothesized to encompass a potential for preventive strategies. In addition, we investigated how the associations between PUFA intakes and breast cancer risk are influenced by obesity and menopausal status.

A Multicenter study, population-based case–control study (CAMA) was conducted by the National Institute of Public Health in Cuernavaca, Mexico. Women were recruited between 2004 and 2007 from 3 regions in Mexico and their surrounding metropolitan areas: Mexico City, Monterrey, and Veracruz. Breast cancer cases (n = 1,000) were women with newly diagnosed, histologically confirmed in situ (n = 20) or invasive breast cancer (n = 980), as previously described (9).

Cases received care from one of 12 participating hospitals from the 3 major health care systems in Mexico. The sample was, therefore, representative of the socioeconomic diversity of the general population of women living in these regions.

Cases were excluded in the following situations: if they had received breast cancer treatment (radiotherapy, chemotherapy, or hormone therapy) in the past 6 months; if they currently used aromatase inhibitors (exemestane, letrozole, or anastrozole) or megestrol, a progesterone derivative; if they were pregnant; or if they were HIV positive. The study protocol and data collection were approved by the Institutional Review Board at the National Institute of Public Health and by equivalent committees at the collaborating hospitals.

Controls (n = 1,074) were frequency matched to the cases according to age, health care system, and region. They were selected on the basis of a probabilistic multistage design, with the aim of sampling specific numbers of women in each 5-year category (range: 35–69 years) based on the age distribution of cases reported by the Mexican Tumor Registry in 2002. Within the 3 study regions, one or more geographic regions (from Spanish, Área Geoestadítica Básica) were selected for sampling.

Cases and controls provided written informed consent to participate in the study.

Project nurses conducted in-person interviews among the cases, obtained anthropometric measures (height, weight, and waist and hip circumference), and collected blood samples. Among controls, interviewers administered an in-person household survey and scheduled an appointment for a hospital visit during which anthropometric measurements were obtained, mammographic screening was carried out, and a blood sample was taken. Body mass index (BMI) was calculated as weight (kg) divided by height (m) squared. Women were classified into different BMI categories according to the World Health Organization guidelines as follows: women with a BMI between 18.5 and 24.9 kg/m² had normal weight, women with a BMI between 25.0 and 29.9 kg/m² were considered overweight, and women with a BMI of 30.0 kg/m² or higher were classified as obese (10). Waist to hip ratio (WHR), as indicator of central obesity, was calculated as waist circumference (cm) divided by hip circumference (cm). The median value (0.91) was used as cutoff point.

General health and lifestyle factors were addressed using a 243-item questionnaire. The questionnaire collected information on lifetime alcohol consumption, sociodemographic characteristics, reproductive/hormonal factors (e.g., age at menarche and menopause, pregnancies, pregnancy outcomes, lactation history, use of oral contraceptives, and hormone therapy), family history of breast cancer, smoking history, and physical activity. To measure physical activity, participants were asked about the time spent sleeping and engaging in physical activity (light, moderate, and vigorous) over a usual week prior to the onset of symptoms.

Cases were interviewed soon after diagnosis (median 3 days). Dietary information was obtained by asking cases about their food consumption the year prior to the onset of the symptoms and to the controls the year before the study started, using a separate 104-item semiquantitative food frequency questionnaire (FFQ) developed on the basis of consumption data from women living in Mexico City using methods described and already used (11). The relative validity compared with sixteen 24-hour recalls and reproducibility of the FFQ was assessed in 134 women in Mexico City (12). The procedures for secondary analyses of study data were approved by the Institutional Review Office at the Fred Hutchinson Cancer Research Center, Seattle, WA.

For this specific study, ω-6 and ω-3 PUFA, and energy intakes were computed from FFQ by multiplying the average daily frequency consumption by the nutrient content of commonly used portion sizes. The nutrient database developed by the National Institute of Nutrition in Mexico (13) and the U.S. Department of Agriculture food composition tables (14) were used to calculate intakes.

Subjects with unrealistic total caloric intake (<500 Kcal/d and >5,000 Kcal/d) were excluded from the analysis (n = 161). Twenty-three subjects were also excluded because of missing information on anthropometric values. The final number of cases and controls involved in the statistical analyses are 914 (91.4%) cases and 976 (90.9%) controls.

Baseline characteristics of the study population are compared by tertiles of ω-3 PUFA intakes. For continuous variables, F tests were used to test for significance of linear trend by assigning ordinal scores to each successive category and treating variables as continuous in the regression model. The Cochran–Armitage test for trend was used for categorical variables. ORs and 95% confidence intervals (CI) for breast cancer risk in relation to ω-6 and ω-3 PUFA and to ω-3 to ω-6 PUFA ratio were calculated by conditional logistic regression (SAS statistical software, version 9, SAS Institute), stratified by the case–control set. PUFA were divided into tertiles on the basis of the distribution among controls. Multivariate analyses were run controlling for potential confounders including BMI (continuous), height, family history of breast cancer, age at first menses, age at first full-term pregnancy, number of full-term pregnancies, breast feeding, age at menopause, ever use of hormone for menopause, ever use of oral contraceptive, physical activity (expressed as METS units), socioeconomic status, energy intake (continuous), alcohol consumption (yes/no), and menopausal status. Linear trend tests were determined on the score variables (tertile categories).

Subgroup analyses on the association between PUFA intakes and breast cancer risk were conducted by unconditional logistic regression, including matching variables in the model, stratified by BMI (normal weight, 18.5 < BMI < 25; overweight, 25 ≤ BMI < 30; and obese, BMI ≥ 30), WHR (median value as cutoff point), and menopausal status (pre- and postmenopause). Tests for heterogeneity in the associations among PUFA levels and breast cancer risk were carried out using χ² tests. Statistical tests were 2 sided and P values <0.05 were considered statistically significant.

Baseline characteristics of the study population by intakes of ω-3 PUFA are presented in Table 1. Women in the highest tertile of ω-3 PUFA had a higher energy intake, a higher ω-6 PUFA intake, a higher alcohol consumption, a higher folate intake, and a higher vitamin E intake than women in the lowest tertile of ω-3 PUFA intake (reference). In the highest tertile of ω-3 PUFA, the percentage of women in postmenopause was higher than those in premenopause. Women in the highest tertile of ω-3 PUFA were more highly educated and of higher socioeconomic level than those in the lowest ω-3 PUFA group.

The associations between PUFA intake and breast cancer risk in the overall population and stratified by menopausal status are presented in Table 2. Overall, there was no significant association between ω-3 PUFA intake and breast cancer risk (P = 0.31), whereas an increased risk of breast cancer was associated with increasing ω-6 PUFA intake (P = 0.04). Menopausal status did not change the risk estimate associated with ω-3 PUFA. The increased risk associated with increasing ω-6 PUFA intake appeared in premenopausal women (P = 0.02) but not in postmenopausal women (P = 0.91). Finally, there was a trend for a decreased risk of breast cancer associated with a high ω-3/ω-6 PUFA ratio, particularly in premenopausal women (P = 0.06).

The associations between PUFA intake and breast cancer risk stratified by BMI are presented in Fig. 1. A decreased risk of breast cancer was significantly associated with increasing ω-3 PUFA intake in obese women (P = 0.008; Fig. 1A) but not in overweight women (P = 0.23; Fig. 1B) nor in normal weight (P = 0.54, P_{heterogeneity} = 0.017; Fig. 1C).

Similarly, a decreased risk of breast cancer was associated with increasing ratio of ω-3 to ω-6 PUFA in obese women (P = 0.01), whereas no significant associations were observed in normal weight (P = 0.26) and in overweight women (P = 0.40, P_{heterogeneity} = 0.05). Obesity status did not significantly affect the positive association between breast cancer risk and ω-6 PUFA intake (P_{heterogeneity} = 0.46) even though the positive association reached statistical significance only in overweight women (P = 0.03).

The same trend was observed between ω-3 PUFA and breast cancer risk when stratifying by WHR but did not reach statistical significance (P_{heterogeneity} = 0.23).

This population-based case–control study conducted in Mexico reported an increased risk of breast cancer associated with increasing ω-6 PUFA, particularly among premenopausal women. This study showed no clear evidence of an inverse association between estimated ω-3 PUFA intake and breast cancer risk, in agreement with former meta-analyses of prospective studies (8). However, our study provided some indication that obesity status, defined by BMI measures, had an impact on risk estimates for ω-3 PUFA intake. A decreased risk of breast cancer associated with increasing ω-3 PUFA intake was found in obese women, whereas no significant inverse association was detected in normal weight and overweight women.

Excessive amounts of ω-6 PUFA, and a low ratio of ω-3 to ω-6 PUFA, as is found in today's Western diets, promote the pathogenesis of many diseases, including breast cancer (15, 16). Because of the increased amounts of ω-6 PUFA in the Western diet, the eicosanoid products from ω-6 PUFA, specifically prostaglandins, thromboxanes, leukotrienes, hydroxy fatty acids, and lipoxins, are formed in larger quantities than those formed from ω-3 PUFA. The eicosanoids from ω-6 PUFA are biologically active in very small amounts, and, if they are formed in large amounts, they contribute to the formation of inflammatory disorders and to proliferation of cells (16). Thus, the positive association between ω-6 PUFA and breast cancer risk may be related to increased production of proinflammatory products from ω-6 PUFA.

Overall, we found no inverse association between ω-3 PUFA intake and breast cancer risk, in agreement with most epidemiologic studies based on estimated intakes (17–19) or biomarkers (20, 21). As exceptions, inverse associations have been reported in Asian women having intakes up to 40 times greater than Western ones (22–24). In the present study, ω-3 PUFA intake was about 10 times lower than those reported in Western populations. In this context, clear inverse associations may not have been observed in our population study because ω-3 PUFA intake might have been below the threshold for a protective effect against breast cancer.

A population-based prospective cohort study conducted among postmenopausal breast cancer women revealed that obesity status may influence the association of breast cancer risk to dietary factors (25). Our population study presented a wide range in BMI measures, allowing us to stratify on BMI, in contrast to other studies with a small range in BMI measures. We found that overall obesity status, as estimated by BMI measures, had an impact on risk estimates for ω-3 PUFA intake. A decreased risk of breast cancer was associated to increasing ω-3 PUFA intake in obese women, whereas no significant inverse association was detected in normal weight and overweight women. The same inverse trend between ω-3 PUFA and breast cancer risk appeared in women according to WHR, as a measure of central adiposity, but did not reach statistical significance.

Differences in ω-3-breast cancer risk association between obese and nonobese women might be related to the anti-inflammatory effects of ω-3 PUFA. Indeed, increased adiposity leads to a chronic inflammation in adipose tissue, resulting in increased production of proinflammatory cytokines (i.e., monocyte chemotactic protein-1, interleukin-6, TNF-α, plasminogen activator inhibitor-1; ref. 26). Obesity is also associated with high levels of insulin, a known mitogen. Experimental studies showed that dietary supplementation with ω-3 PUFA was associated with reduced adipose tissue inflammation and increased insulin sensitivity in obese mice (27). In addition, preincubation of mammary tumor cells with proinflammatory TNF-α stimulated uptake of ω-3 PUFA compared with other fatty acids (28). Thus, ω-3 PUFA may have a protective effect on breast cancer risk in obese women which might be related to increased uptake of these fatty acids in cells and subsequent decreased inflammation and enhanced insulin sensitivity in adipose tissue.

Dysregulated adipokine secretion in obese subjects, particularly leptin and adiponectin (29), has been suspected to mediate the association of obesity with breast cancer (30). Growth of breast cancer cells could be regulated by various leptin-induced secondary messengers like STAT3, activator protein (AP-1), mitogen-activated protein kinase (MAPK), and extracellular signal-regulated kinases (ERK), involved in aromatase expression, generation of estrogens, and activation of estrogen receptor-α in malignant breast epithelium (31). Higher circulating levels of leptin found in obese subjects could be a growth-enhancing factor (as supported by in vitro studies), whereas lower levels of adiponectin found in obese women may allow growth-promoting effects of leptin (31). Fish oil rich in ω-3 PUFA has been shown to increase plasma levels of adiponectin in rodents and in human subjects and to decrease plasma leptin concentrations (26). The effect of ω-3 PUFA on plasma levels of adipokins may be in part a result of activation of peroxisome proliferator-activated receptor γ or inhibition of Toll-like receptor 4 (26). In this context, the possibility that ω-3 PUFA led to decreased breast cancer risk in obese women as a result, at least in part, of improved adiponectin and leptin levels altered in obesity, should be considered.

The discovery of the obesity-inflammation-aromatase axis in the mammary gland and visceral fat may provide insight into mechanisms underlying the inverse association between ω-3 PUFA and breast cancer risk in obese women. Elevated estrogen synthesis, as a consequence of increased aromatase expression in adipose tissue, is thought to be a growth factor associated with the obesity–breast cancer risk association. Analysis of the stromal vascular and adipocyte fractions of the mammary gland suggested that macrophage-derived proinflammatory mediators induced aromatase gene expression in obese mice (32). Aromatase expression in the breast has been shown to be upregulated by AMP-activated protein kinase and cyclic AMP responsive element binding protein–regulated transcription coactivator 2 in response to the altered adipokine milieu associated with obesity and may provide an important link between obesity and breast cancer risk (33). It is suggested that a high intake of ω-3 PUFA relative to that of ω-6 PUFA may decrease endogenous estrogen production via inhibition of aromatase activity/expression (34). However, no studies have yet directly addressed this issue in humans, and the potential of ω-3 PUFA to inhibit aromatase activity/expression altered in obesity needs to be investigated in the future.

Stratifying on menopausal status did not modify the risk estimates for ω-3 PUFA intake, but the positive association between ω-6 PUFA intake and breast cancer risk, or the negative association between the ratio of ω-3 to ω-6 PUFA and breast cancer risk, observed overall was restricted to the subgroup of premenopausal women. Few studies on the association between dietary fatty acids and breast cancer risk presented data stratified by menopausal status. In agreement with our findings, one study reported that the association between the ratio of ω-3 to ω-6 PUFA and breast cancer risk differed regarding to menopausal status, with a stronger association observed in the subgroup of premenopausal women compared with postmenopausal women (35). ω-6 PUFA may have effects opposite to those of the ω-3 series, and differences in breast cancer risk associated to ω-6 PUFA between pre- and postmenopausal women may be related to plasma estrogen levels, although the relationship between dietary PUFA and endogenous estrogen synthesis levels remains to be investigated in humans.

There are some limitations inherent to the case–control design. Case–control studies of diet and cancer are subject to recall bias when ascertaining past dietary information. Recall bias can produce differential measurement error, which can unpredictably bias OR. These results need to be confirmed by biomarkers of PUFA intakes, as it is planned in this population study in the future.

The underlying mechanisms of the inverse association between ω-3 PUFA intake and breast cancer risk among obese women is of particular interest for prevention strategies and warrant further investigation. In this context, experimental studies using models of obese rodents designed at investigating the potential of an enrichment of diet with ω-3 PUFA to prevent or delay the appearance of chemically induced mammary tumors would give more support to our original observation. Future studies of the relationship between ω-3 PUFA intake and breast cancer risk should consider stratification on obesity status.

No potential conflicts of interest were disclosed.

The authors thank all physicians responsible for the project in the different participating hospitals: Dr. Germán Castelazo (IMSS, Hospital de la Raza, Ciudad de México, DF), Dr. Sinhué Barroso Bravo (IMSS, Hospital siglo XXI, Ciudad de México, DF), Dr. Fernando Mainero Ratchelous (IMSS, Hospital de Gineco-Obstetricia N0 4. “Luis Castelaco Ayala”, Ciudad de México, DF), Dr. Hernando Miranda Hernández (SS, Hospital General de México, Ciudad de México, DF), Dr. Joaquín Zarco Méndez (ISSSTE, Hospital 20 de Noviembre, Ciudad de México, DF), Dr. Edelmiro Pérez Rodríguez (Hospital Universitario, Monterrey, Nuevo León), Dr. Jesús Pablo Esparza Cano (IMSS, Hospital N0. 23 de Ginecología, Monterrey, Nuevo León), Dr. Heriberto Fabela (IMSS, Hospital N0. 23 de Ginecología, Monterrey, Nuevo León), Dr. José Pulido Rodríguez (SS, Hospital Metropolitano Dr “Bernardo Sepulveda”, Monterrey, Nuevo León), Dr. Manuel de Jesús García Solis (SS, Hospital Metropolitano Dr “Bernardo Sepulveda”, Monterrey, Nuevo León), Dr. Fausto Hernández Morales (ISSSTE, Hospital General, Veracruz, Veracruz), Dr. Pedro Coronel Brizio (SS, Centro Estatal de Cancerología “dr. Miguel Dorantes Mesa”, Xalapa, Veracruz), Dr. Vicente A. Saldaña Quiroz (IMSS, Hospital Gineco-Pediatría N0. 71, Veracruz, Veracruz), and M.C. Teresa Shamah Levy, INSP, Cuernavaca Mor.

This work was supported by Consejo Nacional de Ciencia y Tecnologia CONACYT and the NIH.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.