Obesity is highly prevalent in the USA. Its incidence in adults was 13 % in 1962 and climbed to 34·9 % in 2010, as reported by the Centers for Disease Control and Prevention( Reference Ogden, Carroll and Kit 1 ). Moreover, in 2011–2, 16·9 % of American children and adolescents were obese, which was a serious public health problem( Reference Ogden, Carroll and Kit 1 ). Obesity can cause several co-morbidities, such as CVD and respiratory disease, type 2 diabetes, certain cancers, and the metabolic syndrome( Reference Mokdad, Ford and Bowman 2 , Reference Collins 3 ). Approximately 50 % of overweight adults have already been reported to be overweight in childhood, and most obese children remain obese as adults( Reference Deshmukh-Taskar, Nicklas and Morales 4 ). Obesity is most commonly caused by a combination of genetic susceptibility, unhealthy behaviours and harmful environmental factors( Reference Ogden, Yanovski and Carroll 5 ). Previous studies have linked environmental endocrine-disrupting chemicals such as diethylstilboestrol, bisphenol A, phyto-oestrogens, phthalates and organotins with adverse health consequences, including obesity and diabetes( Reference Newbold 6 ).
Phyto-oestrogens, a family of plant-derived xeno-oestrogens, include two main forms, isoflavones and lignans, both of which are common in the human diet. Due to their structural similarity with oestradiol, they can cause oestrogenic and/or anti-oestrogenic effects in organisms( Reference Yildiz 7 ). Phyto-oestrogens have been used to prevent postmenopausal bone loss( Reference Lagari and Levis 8 ). However, the biological effect of phyto-oestrogens remains unclear. Several studies have reported that phyto-oestrogen administration is protective against breast and prostate cancers and CVD( Reference Adlercreutz, Heinonen and Penalvo-Garcia 9 ). Phyto-oestrogens cannot be considered nutrients because no characteristic deficiency syndrome is caused by their absence from the diet, and they do not cause any essential physiological malfunction.
To our knowledge, few data on the relationship between phyto-oestrogens and obesity in human subjects have been reported. Several studies that used National Health and Nutrition Examination Survey (NHANES) data from 1999 to 2004 have suggested that urinary enterolignan concentrations were significantly associated with serum TAG levels, HDL-cholesterol levels( Reference Penalvo and Lopez-Romero 10 ) and the presence of the metabolic syndrome( Reference Struja, Richard and Linseisen 11 ). However, the data were not stratified by age and sex, which is necessary because phyto-oestrogens have shown activities similar to oestradiol. In addition, the sample size was not large. Therefore, in the present study, we investigated the association of urinary phyto-oestrogens with obesity and metabolism-related indicators such as TAG levels, insulin levels and the metabolic syndrome in adults aged 20–60 years. The results were obtained from the NHANES 2001–10 and stratified by sex.
Materials and methods
Study population
The NHANES studies were conducted by the US National Center for Health Statistics (Centers for Disease Control and Prevention, Atlanta, GA, USA). The NHANES is a cross-sectional study that used a stratified, multi-stage sample, representative of non-institutionalised civilians in the USA, which was approved by the National Center for Health Statistics Research Ethics Review Board. The subjects enrolled in the present study were obtained from five cycles of the NHANES (2001–2, 2003–4, 2005–6, 2007–8 and 2009–10). Interviews and substantial physical examinations, which included blood and urine collection, were performed in the Mobile Examination Centers. We examined the association of urinary phyto-oestrogens with obesity and metabolism-related indicators in children and adults who participated in the NHANES.
Participants in the NHANES over a range of 10 years were selected to form a random subgroup for the measurement of urinary phyto-oestrogen and serum metabolite levels. Pregnant women (n 383) were excluded from the study because they are considered to have an abnormal physiological status that prevents the accuracy of the original BMI. Women who had undergone an ovariectomy (n 74) were also excluded. The final study sample consisted of 1294 children (6–19 years of age, 694 males and 600 females) and 3661 adults ( ≥ 20 years of age, 1851 males and 1810 females) from this subgroup.
Evaluation of fatness: BMI and waist circumference
BMI was computed as weight divided by height squared (kg/m2). Waist circumference was measured by professionals using a standardised protocol. Among children, individuals were classified as being overweight or obese according to cut-off points based on sex and age( Reference Cole, Bellizzi and Flegal 12 ). Among adults, subjects were classified as overweight if they had a BMI value ranging between 25 and 30 kg/m2, and obese if they had a BMI ≥ 30 kg/m2. For waist circumference, the American Heart Association/National Heart, Lung, and Blood Institute (AHA/NHLBI) recommends a single set of sex-specific cut-offs: >102 cm for men and >88 cm for women( 13 ).
The metabolic syndrome
The metabolic syndrome was diagnosed by incorporating the International Diabetes Federation and AHA/NHLBI definitions from 2009( Reference Alberti, Eckel and Grundy 14 ). In brief, the diagnosis of the metabolic syndrome includes any three of the following five criteria: elevated waist circumference (according to population- and country-specific cut-offs: ≥ 102 cm (men) or ≥ 88 cm (women) in the USA); serum TAG level >1·70 mmol/l ( ≥ 150 mg/dl); serum HDL-cholesterol level < 1·03 mmol/l ( < 40 mg/dl) (men) or < 1·30 mmol/l ( < 50 mg/dl) (women); systolic blood pressure ≥ 130 mmHg or diastolic blood pressure ≥ 85 mmHg; fasting plasma glucose level ≥ 5·55 mmol/l ( ≥ 100 mg/dl).
Laboratory analysis
Phyto-oestrogen metabolites were measured in spot urine samples from participants aged ≥ 6 years. Spot urine samples were obtained in collection cups at the Mobile Examination Centers and were quickly frozen at −20°C. Phyto-oestrogen concentration was measured by two different methods. In the survey years 2001–2, phyto-oestrogen samples were analysed by a Sciex API III heated nebulizer–atmospheric pressure chemical ionisation interface coupled with MS/MS( Reference Barnes, Coward and Kirk 15 ). In the following years, the samples were analysed by HPLC–atmospheric pressure photoionisation–tandem MS( 16 ) (see Laboratory procedure manual available at http://www.cdc.gov/nchs/data/nhanes/nhanes_09_10/Phyto_F_met_phytoestrogens.pdf). The two methods are comparable, and both incorporated enzymatic deglucuronidation before a solid-phase extraction was conducted. Subsequently, the components were resolved by reverse-phase HPLC. MS with internal isotope-labelled standards was used to ensure proper accuracy and detection limit. No differences between the two HPLC methods for urinary phyto-oestrogens were reported in the NHANES protocols, and there were no biologically significant differences (NHANES 1999–2002 v. 2003–4)( Reference Penalvo and Lopez-Romero 10 ). Information on other covariates, such as age, sex, race/ethnicity, education level, and poverty:income ratio (PIR), was obtained from a standardised questionnaire.
Covariates
The PIR is a measure of socio-economic status and defined as the calculated ratio of family income to the poverty threshold after adjusting for inflation and family size. Mean dietary fibre intake and energy intake values were calculated from 2 d dietary interviews. Energy intake was classified as ‘normal’ or ‘excessive’ based on the US Department of Agriculture( 17 ) guidelines, stratified by age and sex. Urinary creatinine, a measure of urinary dilution, was included in the analyses as an independent variable. Serum cotinine, a marker of exposure to environmental cigarette smoke, was measured and divided into quartiles. Serum total cholesterol, serum LDL-cholesterol and fasting insulin were divided into higher weighted ( ≥ 90 %) and lower weighted ( < 90 %) values. TAG level >1·70 mmol/l ( ≥ 150 mg/dl) and the use of a cholesterol-lowering medication, serum HDL-cholesterol level < 1·03 mmol/l ( < 40 mg/dl) (men) or < 1·30 mmol/l ( < 50 mg/dl) (women), and fasting plasma glucose level ≥ 5·55 mmol/l ( ≥ 100 mg/dl) were classified according to the International Diabetes Federation and AHA/NHLBI guidelines.
Statistical analyses
We used multinomial logistic regression models to estimate adjusted OR with 95 % CI for the association of obesity and overweight status and waist circumference as distinct outcomes (v. normal/underweight) with categorical phyto-oestrogen exposure. The lowest quartile (quartile 1) was used as the reference value. As age and sex are extremely important variables, we conducted separate analyses stratified by age (6–19, 20–60 and >60 years) and sex. We controlled for the following a priori confounding factors for the associations of phyto-oestrogens with BMI and waist circumference: age; race/ethnicity; energy intake; dietary fibre intake; PIR; serum cotinine level; urinary creatinine level; education level (in adults). We also adjusted for age, dietary fibre intake, urinary creatinine level and serum cotinine level to explore the relationship between phyto-oestrogens and serum metabolites. All statistical analyses were performed using the Statistical Analysis Systems software package version 9.2 (SAS Institute, Inc.). A P value < 0·05 was designated as the cut-off for statistical significance.
Results
General characteristics
Table 1 presents the baseline characteristics and mean concentrations of urinary phyto-oestrogen metabolite levels among the participants (12–19, 20–60 and >60 years) included in the present study from the NHANES 2001–10 database. The response rates for daily hours of television, video games and computer use were approximately 50 %, so they were not included in the present analysis. Table 2 presents the serum metabolite levels of the participants.
Table 1 Weighted characteristics of the study participants in the NHANES (National Health and Nutrition Examination Survey) 2001–10* (Mean values with their standard errors)
PIR, poverty:income ratio; NA, not applicable.
* Participants were excluded if they had a positive laboratory pregnancy test or self-reported to be pregnant at exam (n 383), and had undergone an ovariectomy (n 74).
Table 2 Serum metabolites of the study participants aged 20–60 years (Mean values with their standard errors)
NA, not applicable.
Overweight, obesity and high-risk waist circumference
As a whole, the adjusted OR from the multinomial logistic regression models adjusted for factors such as age, race/ethnicity, urinary creatinine, dietary fibre intake, PIR, serum cotinine level, energy intake and education level (in adults) for enterolactone are presented in Table 3. Data for daidzein, O-desmethylangolensin, equol, enterodiol genistein are summarised in online supplementary Table S1. Furthermore, the association between BMI and quartiles of urinary enterolactone in men is shown in Fig. 1. The present results showed strong inverse associations with obesity and waist circumference for enterolactone in men aged 20–60 years. Notably, the highest concentration of enterolactone was associated with obesity status (OR 0·30, 95 % CI 0·17, 0·54; P< 0·001) and high-risk waist circumference (OR 0·32, 95 % CI 0·21, 0·51; P< 0·001) in 20- to 60-year-olds. However, in women, the highest concentration of enterolactone was only inversely associated with overweight status (OR 0·43, 95 % CI 0·24, 0·77; P= 0·005).
Table 3 OR for the association between the quartiles of urinary enterolactone and overweight, obesity and high-risk waist circumference v. normal/underweight and low risk* (Odds ratios and 95 % confidence intervals)
* Multinomial logistic regression models were used, adjusted for age, race/ethnicity, urinary creatinine, dietary fibre intake, poverty:income ratio, serum cotinine, energy intake and education level (in adults).
† BMI cut-offs based on age and sex.
‡ Quartiles (Q) of enterolactone (ng/ml): Q1: ≤ 127; Q2: 127–420; Q3: 420–979; Q4: >979.
Fig. 1 BMI trend for the quartiles of urinary enterolactone levels in men aged 20–60 years in the NHANES (National Health and Nutrition Examination Survey) 2001–10. Values are means, with their standard errors represented by vertical bars.
Serum metabolites and the metabolic syndrome
Because obesity status increases the risk of several physical conditions, we further explored the association between some serum profiles and the metabolic syndrome for enterolactone that exerts potential actions in 20- to 60-year-olds. The results are shown in Table 4 and positive results are shown in Fig. 2. Higher enterolactone levels had a stronger relationship with HDL-cholesterol levels, TAG levels, fasting glucose levels and fasting insulin levels, particularly in men. In addition, the highest concentration of enterolactone was linked to a lower risk of the metabolic syndrome in men (OR 0·46, 95 % CI 0·30, 0·70; P< 0·001).
Table 4 OR for the association between the quartiles of urinary enterolactone and serum metabolites of adults aged 20–60 years* (Odds ratios and 95 % confidence intervals)
* Multinomial logistic regression models were used, adjusted for age, urinary creatinine, dietary fibre intake, serum cotinine. Serum total cholesterol, serum LDL-cholesterol and fasting insulin were divided into higher weighted ( ≥ 90 %) and lower weighted ( < 90 %) values. TAG level >1·70 mmol/l ( ≥ 150 mg/dl) and the use of a cholesterol-lowering medication, serum HDL-cholesterol level < 1·03 mmol/l ( < 40 mg/dl) (men) or < 1·30 mmol/l ( < 50 mg/dl) (women), fasting plasma glucose level ≥ 5·55 mmol/l ( ≥ 100 mg/dl) were classified according to the International Diabetes Federation and the American Heart Association/National Heart, Lung, and Blood Institute.
† Quartiles (Q) of enterolactone (ng/ml): Q1: ≤ 127; Q2:127–420; Q3: 420–979; Q4: >979.
Fig. 2 Adjusted OR for (a) serum TAG levels (P for trend = 0·001), (b) HDL-cholesterol levels (P for trend < 0·001), (c) fasting insulin levels (P for trend = 0·004), (d) fasting glucose levels (P for trend = 0·001) and (e) the metabolic syndrome (P for trend < 0·001) by increasing quartiles of urinary enterolactone levels in men aged 20–60 years. Values are OR, with 95 % CI represented by vertical bars, adjusted for age, urinary creatinine, dietary fibre intake and serum cotinine.
Discussion
In a large-scale, nationally representative sample of US children and adults, we found independent associations between various levels of urinary enterolactone and obesity, and high-risk waist circumference. These associations were significant in men aged 20–60 years. Further analysis of serum biochemical parameters showed negative associations between urinary enterolactone levels, serum TAG levels, fasting glucose levels and fasting insulin levels, and a positive association with HDL-cholesterol. Interestingly, higher urinary enterolactone levels correlated with a lower risk of the metabolic syndrome in men aged 20–60 years.
Previously, only a few studies have investigated the association between phyto-oestrogen and metabolic profiles. For example, one small population study conducted in 155 Canadian women showed that women with high enterolactone levels have greater insulin sensitivity( Reference Morisset, Lemieux and Veilleux 18 ), which is inconsistent with our findings. In another study conducted in 299 pregnant American women, circulating fasting glucose concentrations have been shown to be inversely associated with urinary total isoflavone level( Reference Shi, Ryan and Jones 19 ). Furthermore, a randomised controlled trial has found that daidzein administration may aid in the reduction of LDL-cholesterol levels( Reference Liu, Ho and Chen 20 ). A larger population study that included 1748 participants from the NHANES showed that plasma TAG levels were lower in participants in the highest enterolactone quartile( Reference Struja, Richard and Linseisen 11 ). A cross-sectional analysis of 1492 adults has found that enterolignan levels were associated with higher HDL-cholesterol concentrations and lower serum TAG concentrations in American adults( Reference Penalvo and Lopez-Romero 10 ). In the present study, we analysed the whole urinary phyto-oestrogen levels measured by the NHANES. Using these data, we were the first to find an association between urinary enterolactone levels and obesity in males in a cross-sectional study of 1273 adult males aged 20–60 years. Moreover, we showed that an elevation of enterolactone levels was inversely associated with TAG levels, fasting insulin levels and the metabolic syndrome, but positively associated with HDL-cholesterol levels in adult males.
Several studies on the association between phyto-oestrogens and obesity have been conducted. In a recent epidemiological study, O-desmethylangolensin, the metabolite of daidzein, has been shown to be negatively associated with obesity( Reference Frankenfeld, Atkinson and Wahala 21 ). Daidzein, one of the most abundant phyto-oestrogens in the human diet, can regulate lipid and carbohydrate homeostasis and could have health benefits related to human obesity( Reference Orgaard and Jensen 22 , Reference Newbold, Padilla-Banks and Jefferson 23 ). Bhathena & Velasquez( Reference Bhathena and Velasquez 24 ) proposed that dietary phyto-oestrogens have a beneficial role in obesity. On the contrary, we did not find that daidzein was significantly associated with obesity. As expected, we showed a negative association of another phyto-oestrogen (enterolactone) with obesity in men and of overweight in women. Other phyto-oestrogens such as isoflavone have been reported to affect central appetite mechanisms( Reference Yamori 25 ). A related study has also reported that lower doses of genistein are adipogenic and that pharmacological doses of genistein inhibit adipose deposition( Reference Penza, Montani and Romani 26 ).
The potential mechanism for the association between phyto-oestrogens and obesity has not been fully elucidated. Endocrine disrupters interact with other factors that affect fetal and postnatal growth and can cause obesity( Reference Keith, Redden and Katzmarzyk 27 , Reference Heindel and vom Saal 28 ). These disrupters might play roles in multiple pathways( Reference Grun and Blumberg 29 ): (1) metabolic sensors, such as the PPAR and the 9-cis retinoic acid receptor; (2) sex-steroid receptors; (3) the hypothalamic–pituitary adrenal axis; (4) epigenetic process. Some phyto-oestrogens can act as endocrine disrupters by influencing oestrogen biosynthesis( Reference Mense, Hei and Ganju 30 ). These include isoflavones, lignans, stilbenes and coumestans. Among the six phyto-oestrogens included in the present study, four were isoflavones (daidzein, O-desmethylangolensin, equol, and genistein) and two were lignans (enterodiol and enterolactone)( Reference Lampe 31 ). An in vivo study by Cederroth et al. ( Reference Cederroth, Vinciguerra and Kuhne 32 ) discovered that soya-derived phyto-oestrogens have promoting effects on obesity in mice. Furthermore, Taxvig et al. ( Reference Taxvig, Specht and Boberg 33 ) found that phyto-oestrogens have an inhibitory effect on adipose cell differentiation in vitro, and this phenomenon did not seem to be entirely the result of PPARγ activation/inhibition. Phyto-oestrogens may activate the AMPK pathway, subsequently mediating the advantageous effects in peripheral tissues( Reference Cederroth, Vinciguerra and Gjinovci 34 ). This may explain the structural similarities of phyto-oestrogens with endogenous oestrogens. Phyto-oestrogens are able to compete with 17β-oestradiol by binding to the intranuclear oestrogen receptors in different tissues, subsequently causing weak oestrogenic effects( Reference Orgaard and Jensen 22 ).
Phyto-oestrogens are ingested as precursors and then metabolised in the gut by the intestinal flora( Reference Moutsatsou 35 ). Thus, different dietary structures and the diverse composition and abundance of the gut microbiota can result in a variety of circulating phyto-oestrogens in the human body, which have a variety of biological functions. A large number of studies have reported that obesity is associated with the composition of the gut microbiota( Reference Everard and Cani 36 ). This could be another explanation contributing to the underlying mechanisms of obesity. However, the present study had some limitations. First, the different ethnicities of the study subjects prevented us from comparing the results of the present study with previous ones. Second, some confounders, such as physical activity level and genetic susceptibility, may influence the associations in the study. In addition to the common covariants, we analysed other covariants that may affect the development of obesity, such as PIR, education level (in adults), and blood cotinine level; performing this covariant analysis ensured that the present results reflected reality. Because of the cross-sectional nature of the present study, it is not possible to estimate whether phyto-oestrogens affected obesity or vice versa. However, some evidence from in vivo and in vitro studies has suggested that phyto-oestrogen administration may prevent obesity by inhibiting adipocyte differentiation and promoting the apoptosis of mature adipocytes( Reference Cederroth, Vinciguerra and Gjinovci 34 , Reference Hwang, Park and Shin 37 ).
Conclusion
We found that urinary enterolactone was strongly associated with obesity, waist circumference, serum TAG level, fasting glucose level and fasting insulin level, and positively associated with HDL-cholesterol in 20–60-year-old American males. Most of these factors are key components of the metabolic syndrome. Interestingly, we also found a negative association between enterolactone with the metabolic syndrome in men only. Although the exact mechanism of how phyto-oestrogen contributes to obesity remains unclear, the present results may contribute to a better understanding of obesity from an epidemiological standpoint.
Supplementary material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S0007114514004115
Acknowledgements
The authors thank Yuqiu Ge (Department of Environmental Genomics, Jiangsu Key Laboratory of Cancer Biomarkers, Prevention and Treatment, Cancer Center, Nanjing Medical University, China) for her help with statistics.
The present study was supported by the National Natural Science Foundation of China (to A. G., grant no. 81172694; to Z.-Y. J., grant no. 81270537); the Outstanding Youth Fund of Jiangsu Province (to A. G., grant no. SBK2014010296), the Research Project of the Chinese Ministry of Education (to A. G., grant no. 213015A), the Practice Innovation Training Program Projects for the Jiangsu College Students (2012JSSPTTP1018); the Jiangsu Province's Qinglan project (to A. G., grant no. JX2161015124); and the Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions.
The authors' contributions are as follows: Q. L. and C. X. conducted the data analyses; Z.-Y. J. and A. G. were involved in the design of the study and interpretation of the results, and critically reviewed the manuscript. All authors approved the final manuscript.
The authors declare that they have no competing interests.
