1. Introduction
Experimental evidence suggests that phytoestrogen intake may modulate the risk of cancer and cardiovascular disease [
1,
2,
3,
4]. Phytoestrogens are plant-derived, naturally occurring non-steroidal polycyclic phenols that may have weak estrogenic effects when they are ingested and metabolized [
5]. Major classes of phytoestrogens are isoflavones, lignans, and coumestans. Diet is the main source of phytoestrogens in humans. Isoflavones are present in berries, soybeans, and other legumes [
6]. Lignans, primarily matairesinol and secoisolariciresinol, are found widely in many fiber rich foods such as fruits, vegetables, cereals, and flaxseeds [
7]. The main dietary sources of coumestans are alfalfa sprouts, followed by pinto and pea beans [
8]. Resveratrol, from the stilbene group, is found in wine, grape skins, and peanuts [
9].
The estimation of dietary consumption of phytoestrogens is limited by the scarcity of data on the content of food; therefore, food composition databases are complex to establish. Phytoestrogen analysis in food is complicated due to the variety of matrices and the different concentrations in foods. Thus, there is a need for a validated biomarker of phytoestrogen intake for epidemiological studies [
10]. An alternative approach for determining the intake of phytoestrogens, is measuring them in biological samples, such as urine.
Phytoestrogens in food are mainly in the form of several types of conjugates (β-glycosidic) and in smaller amounts as aglycones. Conjugate forms (inactive) are ingested and hydrolyzed to their aglycone forms (bioactive) by bacterial β-glucosidases in the intestine wall. Then, only the bioactive forms are absorbed by the intestinal tract, and can be further glucuronidated in the intestinal wall and liver [
11]. Overall, the main circulating and excreting forms of phytoestrogens are the glucuronidated metabolites. Urinary excretion of phytoestrogens varies with the type of diet, which may be related to differences in both human pharmacokinetics and metabolism by the intestinal bacteria [
12].
At population level, little information is available on the excretion of urinary phytoestrogens and their associations with dietary intakes in Western countries. Dietary intake of phytoestrogens has been associated with urinary excretion of related metabolites in observational studies [
1,
13,
14,
15] and controlled trials [
16,
17]. In Mexico, a study evaluated the phytoestrogens consumption in typical diets [
18]; however, a main limitation was the scarcity of foods with phytoestrogen concentrations in food composition databases. We have previously evaluated phytoestrogens in serum as biomarkers of intake in Mexican women, although limited correlations for individual compounds were found [
19]. Furthermore, there are no studies on the concentration of urinary phytoestrogens of the general population. Therefore, the ability to monitor objectively exposure to these compounds is important in order to understand the health impact of dietary intake of phytoestrogens in Mexican population. A better understanding of the relationship of phytoestrogen consumption and urine concentration in the population of Mexico is essential for carrying out epidemiological research regarding the association between phytoestrogens and health.
Therefore, the aim of this study was to determine phytoestrogens in urine as biomarkers of phytoestrogen intake, by measuring the association of urinary excretion of these compounds with recent and habitual dietary intakes in Mexican women. Estimates of dietary intakes of isoflavones (genistein, daidzein, equol, glycitein, formononetin, biochanin A), lignans (enterodiol, enterolactone, secoisolariciresinol, matairesinol), flavonols (quercetin, kaempferol), flavanones (naringenin), flavones (luteolin), stilbenes (resveratrol), and coumestans (coumestrol) were compared with their urinary compounds.
4. Discussion
In this study, we estimated habitual and recent dietary intakes of phytoestrogens, and measured the urinary excretion of these compounds and related metabolites in a group of Mexican adult women. We also evaluated the relationship between dietary and urinary phytoestrogens and observed that total dietary phytoestrogens, estimated by the FFQ, correlated to their urinary excretion. When we estimated dietary intake of phytoestrogens using the 24 h dietary recall, more correlations between dietary and urinary phytoestrogens were found, including genistein, naringenin, isoflavones, lignans, flavonoids, and total phytoestrogens.
Total phytoestrogens in our study included 16 individual compounds. No observational studies have examined simultaneously a wide range of phytoestrogens. The degree of correlation observed between total urinary phytoestrogens and our estimates of dietary intake were in the magnitude of 0.73 (corrected) for the FFQ, and 0.41 (corrected) for the 24 h recall. Correlations for the FFQ are higher than those obtained in previous study where the correlation between total dietary intake and urinary phytoestrogens was 0.54 [
29]. As observed, we obtained a higher correlation of total phytoestrogens using the FFQ, compared to the 24 h recall. This may be because the composition of phytoestrogens in certain food items that were consumed in the previous 24 h were not included in the food database; some phytoestrogens have not been analyzed for particular food components. Since total phytoestrogens consist of the sum of all the individual compounds, the lack of one or more of them contributes to a lower concentration of total phytoestrogens. On the other hand, the FFQ contains a larger list of foods, and many of them may have a complete composition of phytoestrogens in the database.
Some authors attribute the weak or null association of individual compounds to the extensive response categories used in the FFQ, and imprecise interpretation of the interviewer to describe “a few times a week” or “daily or almost daily” [
30]. Additionally, composition tables used are not representative of the foods that were consumed. Thus, dietary intake data in our study were based on estimated values and were therefore less accurate than intakes measured by urinary concentrations.
The information available on the dietary intake of phytoestrogens in the general population in Western countries is limited. The estimated habitual (3.2 mg/day) and recent (1.4 mg/day) intakes of isoflavones in women of our study were higher than those reported in previous studies. In postmenopausal White women who participated in the Framingham study, with less than 1 mg/d of isoflavone intakes [
31]. Women who participated as controls in a study of ovarian cancer, consumed in average 1.8 mg/day of isoflavones [
32], similar to our participants.
Compared to Asian women [
33], mean consumption of isoflavones in our study was approximately 8 times smaller. In terms of total lignans and coumestrol, our results from the FFQ were similar to the 1 mg/day and 1.4 mg/day, found in the Bandera et al. study [
32]. Naringenin and quercetin were the most consumed flavonoids in our study and their main dietary sources were citrus fruits and onion, respectively.
Overall, our results indicate that the intake of dietary isoflavones exceeded consumption of lignans, which is contrary to the results reported by previous studies in Western diets [
34,
35]. This discrepancy could be due to the use of more individual phytoestrogens that were added for the assessment by group. Another reason could be the continuous use of soy protein, soy isolated, and soy flour as food additives in the manufacture of soy-based cereal, frozen desserts, energy bars, and particularly meat substitutes [
36]. The FFQ included foods such as “Maizoro” cereal, commercial bread, breading, instant soups, and soy beverages, among others, that contributed to the estimates of isoflavone intakes.
Some studies have evaluated the relationship between estimates of dietary intake and urinary excretion of lignans. As in our study, in Australian women, no association between habitual lignan excretion and dietary intake was found for the FFQ [
37]. This is likely due to the wide variety of foods containing lignans, making estimations of lignan intake a challenge. The lack of analysis of the different types of lignans in the Western diet contributed to the underestimation of dietary lignans intake in our study, using the FFQ. The lignan or isoflavone concentrations can vary in the same food according to location, variety, crop season, and processing methods [
38,
39]. Therefore, measuring dietary intakes of phytoestrogens using a food-intake instrument and food composition databases is complicated and may not fully capture the intake of these compounds. Establishing a biomarker of phytoestrogen intake through the present study means that dietary intake can be estimated reasonably accurately from analysis of a 12 h urine collection.
This is the first study on the daily intake of phytoestrogens and their correlation with urinary excretion of these compounds in Mexican women. In a recent study from Northwest Mexico, the authors evaluated the use of serum phytoestrogens as a biomarker of phytoestrogen intake, and found correlations for some individual compounds, such as naringenin, luteolin, genistein, enterolactone, coumestrol, and resveratrol [
19]. However, no correlations were found for groups or for total phytoestrogens. We found correlations between recent intakes and urinary levels for total phytoestrogens, and for groups of isoflavones, lignans, and flavonoids, as well as for genistein and naringenin as individual compounds.
In a study that evaluated serum and urine as biomarkers of intake of phytoestrogens in the general US population, the authors discussed that the concentrations of phytoestrogens were lower in serum than in urine, which points to the fast clearance of these compounds from the body [
40]. The time of the collection of the blood sample is an important variable for the determination of phytoestrogens because of their short half-lives. Isoflavones, in the form of glucuronic, predominate in urine, and half-life is approximately 7 to 10 h [
41]. Enterolignans appear in circulation approximately 8 to 10 h after the ingestion of lignans derived from plants. In contrast, lignans derived from plants are in blood circulation after 2 h of consumption but their concentrations are lower than those of enterolignans [
42]. Thus, urine collections may be more useful for the complete estimation of lignan and isoflavone levels than blood samples. The authors from the study previously mentioned mentioned that the high correlation observed for urinary and serum levels of phytoestrogens validates that noninvasive collection techniques, such as those used for urine, can be used to assess phytoestrogen exposure [
40].
Another study in México estimated the dietary intake of phytoestrogens [
18]. The authors applied an adapted FFQ to measure the consumption of flavonoids (flavonols, flavones, and flavanols), lignans (secoisolariciresinol, matairesinol, lariciresinol, and pinoresinol), and coumestrol. Only the intake of coumestrol was similar to that obtained in our study. The same estimations were made in a study of Torres-Sánchez et al. [
43]. The differences in the intake of phytoestrogens between these two studies and our research could be due to the fact that they used different food composition databases.
In addition to dietary intake, metabolism by intestinal bacteria can also influence an individual’s urinary levels of phytoestrogens, especially equol, which is transformed from daidzein by gut microflora [
44]. The concentrations of equol that we found in the urine samples from the participants in our study (median, 23.1 ng/mL) were higher than those found in women from Hanoi, Vietnam (19 ng/mL), and exceeded the values of women in Japan (1.4 ng/mL) and USA (2.5 ng/mL) [
15]. However, daidzein concentrations were lower in our sample, although we found a significant correlation coefficient between urinary daidzein concentration and its metabolite equol. Diet may contribute to the ability to harbor equol-producing bacteria [
45,
46]; therefore, not all humans (30–60%) possess the gut flora that produces equol [
47].
When we estimated concentration ratios of equol to daidzein in urine samples, as an indicator of conversion efficiency, significantly higher ratios were found for our samples (ratio equol/daidzein = 0.61), compared to samples from women in Hanoi, Vietnam (ratio = 0.53), and Japan (ratio = 0.002) [
15]. These results indicate a more efficient biotransformation of daidzein into its metabolite equol by the Mexican women in our study. Equol has a higher estrogenic activity when compared to daidzein, and has been proposed as an important component of isoflavones for disease prevention [
48]. Biotransformation of daidzein to equol has been proposed as a key factor in the protective effects of phytoestrogens against breast cancer [
49,
50]. Thus, our results indicate that the population in Mexico comprised “good” equol producers, and this phenotype could have epidemiological implications for the reduction of some chronic diseases, such as breast cancer risk.
We also observed that the ratio of enterodiol to enterolactone, which can interconvert, was 0.25 (less than 1), indicating that enterolactone was excreted in higher amounts than enterodiol. This finding is in agreement with other human studies [
40,
51]. The bacterial synthesis of enterolactone occurs via dehydroxylation and demethylation of matairesinol. Enterolactone is also produced by oxidation of enterodiol, which is a product of secoisolariciresinol metabolism [
52]. The biological activity of enterolactone and enterodiol is different; enterolactone is a more potent aromatase inhibitor and has approximately 10 times the estrogenic activity than enterodiol [
53]. According to Liu et al. [
54], enterodiol and enterolactone both have potent inhibitory effects on ovarian cancer, but enterolactone possesses a more effective anti-cancer capability and fewer side effects than enterodiol. Enterolactone also has inhibitory effects on growth and metastasis in human breast cancer [
55]. Thus, differences in metabolism and exposure of these lignans may be of physiological importance in cancer prevention.
On the other side, variations between the estimated intake and the urinary levels of individual or group phytoestrogens, especially when we used the FFQ, could be due to the difficulty in estimating phytoestrogen intakes. The FFQ, by definition, is a semiquantitative questionnaire with a trend toward the overestimation of intake. Methods of dietary assessment are estimations of intake and rely on the veracity or reported intakes. Therefore, the accuracy of dietary data is influenced by memory recall, body size, sex, age, and ethnicity, as well as psychosocial and behavioral factors [
56,
57,
58,
59].
Studies in developing countries have shown that energy intake is underreported [
60,
61]. In fact, underreporting intake of some foods is apparent in our study, since energy intake from the 24 h recall was low, 1473 kcal/day. According to Scagliusi et al. [
62], individuals with a lower income might have greater difficulties in the reporting tasks. Body mass index is also an important variable. As obesity increases, especially in the female population, underreporting energy intake is more common. Irregular meal habits and low education have been also associated with the under-report of energy intake [
63]. In a study in New Zealand, the authors detected 265 unreported foods (often snacks) as revealed by the use of wearable cameras [
64].
According to the above, we might think that the under-report of energy intake may be due to the frequency of overweight and obese persons in our study population (37.3% and 38.8% respectively, according to body mass index). Low income and education are factors that could contribute to the underreporting of energy intakes, since 64% of women in our study were from a lower-income level and 58.6% had less than 9 years of schooling. Considering the possibility of the underestimation of intake in our study population, reported consumption of unhealthy foods that are rich in fat, high in sugar, or highly processed may be lower than habitual intake; thus, fruits and vegetables, which represents a source of phytoestrogens, may be overestimated.
A limitation of our study is the use of an FFQ validated for women in the region and modified for the study, although not validated specifically for phytoestrogens. The food list increased in the questionnaire, which could lead to an overestimation of phytoestrogen intake. The collection of 12 h instead of 24 h urine may be another limitation of our study. We used 12 h urine since it is a less burdensome method, and the 24 h urine collection was not well accepted by potential participants. Therefore, women collected their urine samples during the night, and we picked them up in the early morning. As an argument, we may say that, in the study of Grace et al. [
65], a spot urine was sufficient for obtaining good results. Because of unpredictable changes in urine flow during the day and since the synthesis and total elimination of creatinine is constant [
66], we adjusted the amount of phytoestrogens in urine to the concentration of creatinine. It is desirable that we could have collected urine samples for one or more days in each season of the year to consider seasonal variations in intakes. This could have allowed us to obtain better correlations between phytoestrogens from urine and dietary phytoestrogens estimated from the FFQ, because they represent the usual diet.
An advantage of using urine analysis could mean that we can avoid cases of omission or the overestimation of intake among participants, as well as the problem of not finding a food in the food database. Our quantification method is sensitive and reflects a change due to exposure, since the technique has high specificity and sensitivity. Therefore, the determination of urine levels provides a more accurate and objective measure of intake. The evaluation of 16 individual phytoestrogens, as well as the fact that there is no other study on this subject in Mexican women, is another advantage of our study. The results of our study could motivate other researchers to validate intakes of specific phytoestrogens.