Journal of Exposure Analysis and Environmental Epidemiology (2003) 13, 276–282 r 2003 Nature Publishing Group All rights reserved 1053-4245/03/$25.00 www.nature.com/jea Urinary and serum concentrations of seven phytoestrogens in a human reference population subset LIZA VALENTI´ N-BLASINI, BENJAMIN C. BLOUNT, SAMUEL P. CAUDILL, AND LARRY L. NEEDHAM National Center for Environmental Health, Centers for Disease Control and Prevention, Atlanta, GA 30341, USA Diets rich in naturally occurring plant estrogens (phytoestrogens) are strongly associated with a decreased risk for cancer and heart disease in humans. Phytoestrogens have estrogenic and, in some cases, antiestrogenic and antiandrogenic properties, and may contribute to the protective effect of some diets. However, little information is available about the levels of these phytoestrogens in the general US population. Therefore, levels of phytoestrogenswere determined in urine (N ¼ 199) and serum (N ¼ 208) samples taken from a nonrepresentative subset of adults who participated in NHANES III, 1988– 1994. The phytoestrogens quantified were the lignans (enterolactone, enterodiol, matairesinol); the isoflavones (genistein, daidzein, equol, O- desmethylangolensin); and coumestrol (urine only). Phytoestrogens with the highest mean urinary levels were enterolactone (512 ng/ml), daidzein(317 ng/ ml), and genistein (129 ng/ml). In serum, the concentrations were much less and the relative order was reversed, with genistein having the highest mean level (4.7 ng/ml), followed by daidzein (3.9 ng/ml) and enterolactone (3.6 ng/ml). Highly significant correlations of phytoestrogen levels in urineand serum samples from the same persons were observed for enterolactone, enterodiol, genistein, and daidzein. Determination of phytoestrogen concentrations in large study populations will give a better insight into the actual dietary exposure to these biologically active compounds in the US population. Journal of Exposure Analysis and Environmental Epidemiology (2003) 13, 276–282. doi:10.1038/sj.jea.7500278 Keywords: phytoestrogens, lignans, isoflavones, human, exposure, urine, serum. 1. Introduction oxidative balance (Phipps et al., 1993; Wei et al., 1995). Although the physiologic relevance of some of these proper- Decreased risks for heart disease and cancer are associated ties is questionable, consumptionof foods rich inphytoestro- with certaindiets. This protective effect may result from gens is associated with significantly reduced risk for cancer, dietary micronutrients, macronutrients, biologically active osteoporosis, cardiovascular disease, and severity of meno- non-nutrients, or a combination of these and other factors. pause (Adlercreutz and Mazur, 1997; Kurzer and Xu, 1997). The human diet contains a complex array of naturally Additionally, certain lignans (Saarinen et al., 2001) and occurring bioactive compounds called phytochemicals. isoflavones reduce carcinogen-induced cancer in laboratory Among these phytochemicals are phytoestrogens, a broad animals (Lamartiniere et al., 1995a, b). group of plant-derived compounds of nonsteroidal structure Although considerable health benefits are associated with that canmimic estrogen.The largest dietary dosage of diets rich inphytoestrogens, little informationis available phytoestrogenic activity results from two classes of com- about the actual phytoestrogen concentrations in the general pounds: lignans and isoflavones. These groups modulate US population.Phytoestrogenexcretioninurinevaries with many biologically relevant processes, including sex hormone the type of diet (Adlercreutz et al., 1986; Hutchins et al., productionandmetabolism (Adlercreutz et al., 1987, 1988), 1995a, b; Horn-Ross et al., 1997). Sources for lignans intracellular enzyme levels (Adlercreutz et al., 1993), protein includewholegrains,flax,andsomefruitsandvegetables. synthesis, malignant cell proliferation, cellular differentiation Isoflavone diets consist primarily of legumes, such as (Adlercreutz and Mazur, 1997; Kurzer and Xu, 1997), and soybeans and soy-based products (Adlercreutz et al., 1991; Kirkmanet al., 1995; Hutchinset al., 1995a, b). The phytoestrogens are ingested in their natural beta-glycosidic 1. Address all correspondence to: Dr Liza Valentı´ n-Blasini, National forms, which are hydrolyzed to their aglycones in the Center for Environmental Health, Centers for Disease Control and intestine, absorbed, and then glucuronidated in the intestinal Prevention, 4770 Buford Highway, MS F-17, Atlanta, GA 30341, USA. Tel.: +1-770-488-7902; fax: +1-770-488-4609. wall. The major circulating forms of the isoflavones are the E-mail: [email protected] glucuronidated metabolites, and the glucuronidated forms Received 5 November 2002; accepted 3 March 2003 also predominate in the urine (Setchell et al., 2001). The half- Urinary and serum phytoestrogens levels in reference populations Valentı´n-Blasini et al. lives of the phytoestrogens are reported to be in the range of deconjugation internal standards (4-methylumbelliferone 3–10 h (Setchell et al., 2001). Previous studies have reported glucuronide and 4-methylumbelliferone sulfate, Sigma Che- urinary phytoestrogen levels in various populations, but they mical, St. Louis, MO, USA). Serum was diluted and only examined a very few people in each population buffered with ammonium acetate (1 ml of 250 mM, pH 5). (Adlercreutz et al., 1995). The largest reported phytoestrogen Conjugated analytes were hydrolyzed by the addition of b– exposure assessment study measured urinary levels of glucuronidase/sulfatase (Helix pomatia, H-1, Sigma Chemi- phytoestrogens in an ethnically diverse population of 50 cal)andincubatedovernightat371C. Serum deconjugated youngwomeninonegeographic area, the SanFrancisco Bay samples were extracted with Oasis HLB SPE (60 mg HLB, Area (Horn-Ross et al., 1997). We measured the serum and 3 ml, Waters Scientific, Beverly, MA), and urine samples urinary levels of the lignans (matairesinol, enterolactone, and were extracted using C18 SPE column (Varian 500 mg, 3 ml, enterodiol); isoflavones (genistein, daidzein, O-desmethylan- 120 mm pore size); the concentrated extract was resuspended golensin (O-DMA), and equol); and coumestrol (in urine inmobile phase. The phytoestrogensinthe extracted samples only) in a larger and more geographically diverse US were separated by reversed-phase HPLC (50 mm Prism population of adults of different race/ethnicity and both column, Keystone Scientific, Bellafonte, PA, USA) and genders with different dietary habits. measured by tandem mass spectrometry (Sciex API III, Perkin-Elmer Sciex Instruments, Wellesley, MA, USA) using heated nebulizer atmospheric pressure chemical ionization in Materials and methods thenegativeionmode. Urinary concentrations of the phytoestrogens are reported Study subjects as nanograms of phytoestrogen per milliliter of urine (ng/ml) Urine and serum samples were obtained from a subset of and as micrograms of phytoestrogens per gram of urinary samples from adult participants of the Third National Health creatinine (mg/g); serum concentrations are reported as ng/ and Nutrition Examination Survey (NHANES III), 1988– ml. Quality control samples (spiked serum or urine) and 1994. This subset was chosenat random from individuals standards were analyzed along with unknown samples. To who volunteered to give an extra blood and urine sample. diminish the high degree of skewness of the data, the The samples were takenat various times of the day anddo concentrations (mg/g creatinine) were graphed using a not represent fasting samples. This population subset was not logarithmic scale. Geometric means were determined with designed to be representative of the US population but rather the values corrected by urinary creatinine concentrations. to serve as a reference range group of civilian, noninstitutio- For statistical purposes, nondetectable values were treated as nalized adults, age 20–58 years (mean of 36.7710.2 years). the limit of detectiondivided by two (LOD/2). Detection The gender distribution was 61% female. Race/ethnicity of limits were inthe low ng/ml (Valentı ´ n-Blasini et al., 2000). these samples was as follows: non-Hispanic white 29%; non- Statistical analysis of the correlation of each of the Hispanic black 43%; Mexican American 23%; and other phytoestrogen concentrations in urine and serum from the 5%. The paired urine and serum samples from 66 people same individuals was performed using a Spearman’s rank were simply based onthe availability of samples for which we correlation analysis. Multivariate statistical analysis (annual had adequate volume of both matrices. income, education, age, ethnicity, rural or urban residence, and gender) was accomplished using PROC GLM in SAS Method (SAS User’s Guide: Statistics, 1982 Edition. SAS Institute, Randomly selected urine (N ¼ 199) and serum (N ¼ 208) Inc., Cary, NC, USA). samples were analyzed for phytoestrogens using solid-phase extraction (SPE) followed by HPLC and tandem mass spectrometry, as previously described (Valentı´ n-Blasini et al., Results 2000). All of these samples had beenstored at À701Csince collection. Genistein, coumestrol, and daidzein were pur- Urinary Levels of Phytoestrogens chased from Indofine Chemical (Somerville, NJ, USA). The Urinary levels (in ng/ml) of phytoestrogens in our reference remaining phytoestrogens and the labeled phytoestrogens, group are summarized inTable 1. Sincecoumestrol was except for daidzein, were purchased from Professor Kristina detected in only nine urine samples, its