Phytochemicals are compounds present in plant-derived foods, including soya, grain cereals, seeds and berries, that can be grouped into families based upon similarities in chemical structure: flavonoids (isoflavones, flavones, flavonols, flavan-3-ols, flavanones and anthocyanins), dietary lignan precursors (secoisolariciresinol, matairesinol, lariciresinol and pinoresinol), cinnamic acid (quinnic acid and caffeine acid), and coumestans (coumestrol)(Reference Manach, Scalbert, Morand, Remesey and Jiménez1). All of the aforementioned compounds are polyphenols, which have antioxidant, antiproliferative and antiangiogenic biological activity. Some of these compounds have weak oestrogenic activity, which may confer a protective effect against breast cancer (BC)(Reference Manach, Scalbert, Morand, Remesey and Jiménez1, Reference Webb and McCullough2).
The incidence of BC worldwide is characterized by important geographical differences, which suggests that environmental factors, specifically diet, may play a role in its aetiology. This is consistent with the increased risk of BC reported for women emigrating from areas of low to high BC incidence(3).
Both the consumption of soya(Reference Yamamoto, Soube, Kobayashi, Sasaki and Tsugane4, Reference Shu, Jin, Dai, Wen, Potter, Kushi, Ruan, Gao and Zheng5), a rich source of isoflavones, and the subsequent excretion of isoflavones have been associated with a decreased risk of BC among Asian women(Reference Zheng, Dai, Custer, Shu, Wen, Jin and Franke6, Reference Dai, Franke, Jin, Shu, Hebert, Custer, Cheng, Gao and Zheng7). Information regarding other phytochemicals consumed in Western countries is limited and inconsistent. In prospective studies conducted in Western populations, a beneficial effect has not been found for the consumption(Reference Keinan-Boker, van Der Schouw, Grobbee and Peeters8–Reference Horn-Ross, Hoggatt and West11) or high urinary excretion(Reference den Tonkelaar, Keinan-Boker, Veer, Arts, Adlercreutz, Thijssen and Peeters12) of specific isoflavones and flavonols. In contrast, beneficial effects have been observed in retrospective studies measuring the consumption of flavones(Reference Peterson, Lagiou, Samoli, Lagiou, Katsouyanni, La Vecchia, Dwyer and Trichopoulos13, Reference Bosetti, Spertini and Parpinel14), flavonols, flavones and flavan-3-ols(Reference Fink, Steck and Wolff15), as well as the intake(Reference Linseisen, Piller, Hermann and Chang-Claude16) and urinary excretion of daidzein and genistein(Reference Murkies, Dalais, Briganti, Burger, Healy, Wahlqvist and Davis17). Studies determining the protective effect of lignans (the phyto-oestrogens present in highest amounts in Western diets) have measured dietary consumption(Reference Keinan-Boker, van Der Schouw, Grobbee and Peeters8, Reference Fink, Steck and Wolff15, Reference Linseisen, Piller, Hermann and Chang-Claude16, Reference McCann, Muti, Vito, Edge, Trevisan and Freudenheim18), serum concentration(Reference Piller, Chang-Claude and Linseisen19, Reference Hulten, Winkvist, Lenner, Johansson, Adlercreutz and Hallmans20) and urinary metabolites (enterolactone and enterodiol)(Reference Ingram, Sanders, Kolybaba and Lopez21).
Dietary patterns in Mexico play a role in BC risk. Particularly the intake of lettuce, spinach, onion and citrus fruits(Reference Torres Sánchez, López Carrillo, López Cervantes, Rueda and Wolff22), as well as vitamin E, polyunsaturated fat(Reference Bonilla-Fernandez, Lopez-Cervantes, Torres-Sanchez, Tortolero-Luna and Lopez-Carrillo23) and vitamin B(Reference Lajous, Lazcano-Ponce, Hernandez-Avila, Willett and Romieu24), are protective for BC; in contrast, carbohydrate consumption is related to increased BC risk(Reference Romieu, Lazcano-Ponce, Sanchez-Zamorano, Willett and Hernandez-Avila25). Some of these foods (lettuce, onion, citrus fruits) contain phyto-oestrogens(Reference Galvan-Portillo, Wolff, Torres-Sanchez, Lopez-Cervantes and Lopez-Carrillo26). We extended these efforts by establishing a data set of phytochemical contents in Mexican foods, to perform an evaluation of the intake of selected phytochemicals and BC risk in Mexican women.
Materials and methods
The present study is a secondary analysis of a hospital-based, case–control study evaluating the effect of exposure to DDT (dichlorodiphenyltrichloroethane) and its main metabolite, p,p′-DDE (dichlorodiphenyldichloroethylene), on BC risk. It was conducted in three hospitals in Mexico City, Central Region of Mexico (National Institute of Cancer Research, General Hospital of Mexico and Hospital Manuel Gea González), between March 1994 and April 1996. A total of 141 BC cases were identified and age-matched (±3 years) with 141 hospital controls. The inclusion criterion for all participants was a minimum of 20 years’ residence in Mexico City(Reference Lopez-Carrillo, Blair, Lopez-Cervantes, Cebrian, Rueda, Reyes, Mohar and Bravo27).
Participants
Cases were all women between 21 and 79 years of age, with a first-time diagnosis of histologically confirmed BC, who had received no previous cancer treatment. Tumour stage was classified using the tumour–nodes–metastasis system.
For each case, an age-matched (±3 years) female control with no mammary pathology was identified. Subject selection was carried out in all clinical services (except oncology) of participating hospitals. Principal diagnoses of control subjects were as follows: injuries (17·7 %), genitourinary diseases (14·9 %), gastrointestinal disorders (11·3 %), ophthalmic diseases (5·6 %), non-specific pathologies (4·3 %), blood disorders (4·2 %), respiratory diseases (3·6 %), musculoskeletal and connective tissue disorders (3·5 %), disorders of the central nervous system (2·8 %), dermatological diseases (2·8 %), infectious diseases (2·7 %), endocrine disorders (2·6 %), CVD (2·6 %) and auditory diseases (2·4 %), with other diagnoses in smaller proportions.
Information was obtained from each participant regarding diet, sociodemographic characteristics, reproductive history and family antecedents of BC via a direct interview conducted in the hospital by trained interviewers, who knew the case–control status but not the study’s hypothesis. Anthropometric measurements (weight and height) were collected at the time of the interview to calculate BMI. Participation rates among cases and controls were 81 % and 72 %, respectively. All participants signed an informed consent letter.
Dietary phytochemical consumption
A semi-quantitative FFQ, previously updated and validated(Reference Galvan-Portillo, Wolff, Torres-Sanchez, Lopez-Cervantes and Lopez-Carrillo26), was used to estimate the daily dietary consumption of phytochemicals. Cases were asked about dietary habits during the 12-month period prior to the onset of symptoms, and controls were asked the same questions in reference to the 12 months before the interview. The intake of nutrients derived from the consumption of food and beverages was estimated by means of the Food Intake Analysis System 3·0 software (Texas University School of Public Health, Houston, TX, USA)(Reference Bonilla-Fernandez, Lopez-Cervantes, Torres-Sanchez, Tortolero-Luna and Lopez-Carrillo23, Reference Galvan-Portillo, Wolff, Torres-Sanchez, Lopez-Cervantes and Lopez-Carrillo26).
To estimate the dietary consumption of flavonols, flavan-3-ols, flavones, cinnamic acid, lignans and coumestrol, we created a database using published nutritional data for the phytochemical content in ninety-five different foods(Reference Galvan-Portillo, Wolff, Torres-Sanchez, Lopez-Cervantes and Lopez-Carrillo26). The values of flavonols, flavan-3-ols, flavones and cinnamic acid were expressed in milligrams, while lignans (lariciresinol, pinoresinol, secoisolariciresinol and mataresinol) were in micrograms per 100 g of edible portion per day.
Statistical analysis
Reproductive and sociodemographic differences between cases and controls were evaluated using the t test and χ 2 statistic, while differences in the medians of macronutrient and phytochemical consumption were evaluated using a Mann–Whitney test. To determine the association between the dietary consumption of each phytochemical and the risk of BC, we used tertiles based on the distribution in the control group. Odds ratios for the effect of phytochemicals on BC risk were calculated in the total population and stratified by menopausal status using unconditional logistic regression models.
The following known reproductive characteristics associated with BC were considered as potential confounders: age at menarche (years), four categories of parity and age at birth of first child (≥3 children and <20 years for first child, <3 children and <20 years for first child, ≥3 children and ≥20 years for first child, <3 children and ≥20 years for first child, no children), lifetime lactation (months), family history of BC, alcohol consumption (yes/no) and BMI. Each one of them was added to the corresponding crude model, and those that changed the odds ratio by more than 10 % were kept in the final multivariate model(Reference Greenland28).
The dose–response relationship was evaluated by including the tertiles of each phytochemical in the model as an ordinal variable; additionally, interaction terms between menopausal status and phytochemicals were checked. We considered as statistically significant a P value of <0·05. The analysis was performed with the STATA 9·0 statistical software package (Stata Corporation, College Station, TX, USA).
Results
Compared with BC cases, controls were significantly younger at the birth of their first child, had more pregnancies and greater lifetime lactation. The proportion of women with a family history of BC and the average BMI were marginally smaller among controls than cases. No differences were found regarding the remainder of the variables included in Table 1. Most of the breast tumours were stage II.
Table 1 Selected characteristics of the study population: hospital-based, case–control study on intake of selected phytochemicals and breast cancer risk, Mexico City, 1994–1996
†For postmenopausal women only.
‡P from t test.
§P from χ 2 test.
The median dietary consumption of macronutrients and selected phytochemicals, as well as the 10th and 90th percentiles, are presented in Table 2. Dietary consumption of flavonols (31·0 v. 27·8 mg/d) and flavones (2·6 v. 2·5 mg/d) was significantly greater in controls than cases.
Table 2 Distribution of the daily consumption of nutrients and phytochemicals in the study population: hospital-based, case–control study on intake of selected phytochemicals and breast cancer risk, Mexico City, 1994–1996
Median values were significantly different from those of the control group (Mann–Whitney test): *P < 0·05, **P < 0·01.
†Quercetin, quercetin-3-rutinoside, free quercetin, quercetin-3-glucoside, quercetin-4-glucoside, quercetin-3,4-diglucoside, myricetin, kaempferol, conjugated isorhamnetin and isorhamnetin-4-glucoside.
‡Conjugated luteolin and apigenin.
Postmenopausal women in the highest tertiles of flavonols and flavones consumption showed a statistically significant decrease in BC risk (OR = 0·21, 95 % CI 0·07, 0·60 and OR = 0·29, 95 % CI 0·10, 0·82) compared with women in the lowest tertiles. In both cases the tests for trend were significant (P for trend = 0·004 and 0·025, respectively). Among premenopausal women, high consumption of the lignans lariciresinol and pinoresinol (high v. low tertile: OR = 0·32, 95 % CI 0·10, 0·99, P for trend = 0·051 and OR = 0·19, 95 % CI 0·06, 0·62, P for trend = 0·006, respectively) was associated with a significant reduction in BC risk (Table 3). There were no significant interactions between menopausal stage and phytochemical dietary consumption.
Table 3 Effect of dietary consumption of selected phytochemicals on breast cancer risk among premenopausal and postmenopausal women: hospital-based, case–control study, Mexico City, 1994–1996
†Adjusted for age, total energy consumption (kcal), lifetime lactation (months) and menopausal stage.
‡Adjusted for age, total energy consumption (kcal) and lifetime lactation (months).
§95 % CI not calculated (reference category).
All previous results remained when p,p′-DDE values were added in the models (data not shown).
Discussion
The present study is the first one conducted in a Mexican population about specific phytochemical intake and BC risk. Regarding the consumption of flavonoids, the inverse association between flavonols and flavones consumption and the risk of BC is consistent with previous studies conducted in Greece(Reference Peterson, Lagiou, Samoli, Lagiou, Katsouyanni, La Vecchia, Dwyer and Trichopoulos13), Italy(Reference Bosetti, Spertini and Parpinel14) and the USA(Reference Fink, Steck and Wolff15), which reduces the likelihood that our results may be due to chance. In the two first studies, a reduction in BC risk was associated with an increase in dietary flavones consumption, while in the last study risk reduction was associated with the consumption of flavonols, flavones, flavan-3-ols and lignans.
Additionally we showed a protective effect of lariciresinol and pinoresinol among premenopausal women. A recent study also found a similar effect among postmenopausal women with oestrogen receptor- and progesterone receptor-positive BC tumours(Reference Touillaud, Thiébaut, Fournier, Niravong, Boutron-Ruault and Clavel-Chapelon29). This suggests that these compounds operate through an oestrogen antagonist mechanism(Reference Mueller, Simon, Chae, Metzler and Korach30); however, further research is needed to reach a definite conclusion about the protective role of these phyto-oestrogens.
We also found in our study population that onion and hot sauce (made with tomatoes, onion and chilli pepper) are sources of flavonols and flavones, while banana, melon, lettuce and tomatoes contain lariciresinol and pinoresinol (data not shown). A previous study in Mexico showed a protective effect of BC due to onion consumption(Reference Torres Sánchez, López Carrillo, López Cervantes, Rueda and Wolff22). Besides flavonoids and some antioxidants (e.g. Se(Reference Arnault and Auger31) and β-sitosterol(Reference Mizushina, Nakanishi and Kuriyama32)), onion is a source of lariciresinol and secoisolariciresinol(Reference Milder, Arts, van de Putte, Venema and Hollman33). To assess whether flavonoids or other compounds may be responsible for the favourable effect of onion in BC, a model adjusting by flavonol, lariciresinol and secoisolariciresinol intake (not included in the tables) was run. Results showed a significant, protective effect of onion intake among postmenopausal women (OR = 0·17, P for trend = 0·02), suggesting that other compounds present in onion may also be exerting a protective role in BC risk. Thus, the phyto-oestrogen composition of Mexican foods deserves further attention to advance understanding of their role in breast carcinogenesis.
In addition to their antioxidant properties, antiangiogenic activity and apoptosis promotion, flavonoids exhibit an anticarcinogenic function by inhibiting enzymes that participate in cell differentiation and tumour growth (DNA topoisomerases and kinases)(Reference Knight and Eden34). They also inhibit aromatases, which are involved in the synthesis of endogenous oestrogens(Reference Adlercreutz35). Owing to either their aromatase inhibition activity or their competence as oestrogenic receptors, high circulating amounts of isoflavones have been negatively associated with serum oestrogen concentration(Reference Low, Taylor and Grace36).
Although lignans share the same hormonal and non-hormonal mechanisms as flavonoids, a modification of their protective effect has been observed depending on the oestrogen receptor status of mammary tissue. This suggests a hormone-independent mechanism(Reference Webb and McCullough2, Reference Olsen, Knudsen, Thomsen, Loft, Stripp, Overvad, Moller and Tjonneland37) such that mammary cells without oestrogen receptors are subject to apoptosis. This may explain the differential effect of pinoresinol and lariciresinol intake by menopausal status.
Some methodological considerations should be deliberated in the interpretation of our results. Our estimation of dietary flavonoids and lignans consumption using an FFQ should be taken into consideration with regard to its reproducibility/reliability. However, our instrument captures 18 % and 47 % of the actual daily dietary intake for flavan-3-ols and lariciresinol, respectively(Reference Galvan-Portillo, Wolff, Torres-Sanchez, Lopez-Cervantes and Lopez-Carrillo26). Thus, our results may be a conservative estimation of the actual effect of phytochemicals on BC risk. Regardless, there is little reason to suspect that such an underestimation could be differential in relation to cases and controls, because interviewers and participants were blinded to the study hypothesis.
On the other hand, we consider the possibility of a selection bias to be low, given that the consumption of macronutrients observed among our controls, compared with that reported in the National Survey of Nutrition of female residents of Mexico City(Reference Rivera Dommarco, Shamah Levy, Villalpando Hernández, González de Cossio, Hernández Prado and Sepúlveda38), was similar. Furthermore, the diagnoses of the clinical controls were not statistically related to consumption of the studied phytochemicals (data not shown), while other types of cancer were not included because of their potential relationship with phytochemical intake. Moreover, the rates of participation in our study were high (around 80 %), reducing the possibility that non-participating individuals differed substantially in dietary habits and other characteristics compared with participants.
No information was available for the present study regarding the use of oral contraceptives (OC) and hormone replacement therapy (HRT). Although national statistics show that around 20 % of women are OC users, there is no reason to consider an association between OC and/or HRT and phytochemical consumption that could bias our results(39).
In Mexico the age-adjusted incidence rate of BC is four times smaller than that reported in the USA (26·4 v. 101·1 per 100000)(Reference Ferlay, Bray, Pisani and Parkin40). In the USA the incidence rate of BC among states does not significantly vary(Reference Jemal, Siegel, Ward, Murray, Xu, Smigal and Thun41), whereas in Mexico large differences exist by region. The northern frontier states of Mexico have BC incidence rate similar to the USA and approximately three to four times greater than that observed in the southern zone of the country(42). While many factors could explain this difference, it is important to note that dietary patterns observed in both regions are quite different(Reference Barquera, Hotz, Rivera, Tolentino, Espinoza, Campos and Shamah43). The diet in the southern and central region of Mexico is characterized as being lower in fat and red meats, but rich in foods that are high in flavones and lignans (e.g. corn, beans and green vegetables). The diet in the northern region is more similar to the US diet, characterized by a high consumption of fats, sugars and red meats, and a low consumption of vegetables. Although the previous perspective is ecological, it is consistent with our results.
Dietary phytochemical consumption could play an important protective role in BC development, as well as in the differences in BC incidence observed among regions in Mexico. Several phytochemical food sources and types of phytochemicals across the world should be considered for evaluation in future studies.
Acknowledgements
Study design and collection of information were financed by Consejo Nacional de Ciencia y Tecnología de México (CONACyT). Funding from the Fogarty International Center of the National Institutes of Health (D43TW00640) to Mount Sinai School of Medicine/Queens College International Training and Research in Environmental and Occupational Health Program financed the collection, analysis and interpretation of data, as well as writing of the manuscript. The authors declare that they have no competing interests.
Authors’ contributions: L.T.-S. participated in statistical analysis planning and writing the manuscript. M.G.-P. performed the statistical analysis of data and collaborated in writing the manuscript. M.S.W. participated in study design, exposure estimations and critical revisions of the manuscript. L.L.-C. conceptualized the study, obtained the funds for the study and made critical revisions of the manuscript.
