| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
ORIGINAL RESEARCH COMMUNICATION |
1 From TNO Quality of Life, Business Unit BioSciences, Zeist, Netherlands (EB); the Institute National de la Recherche Agronomique, GROUPE Ostéoporose, Saint-Genès Champanelle, France (BC); the Rowett Research Institute, Bucksburn, Aberdeen, United Kingdom (SR); the Laboratory of Organic Chemistry, Department of Chemistry, University of Helsinki, Helsinki, Finland (KW); the School of Medicine, University of East Anglia, Norwich, United Kingdom (AC); and the National Institute for Research on Food and Nutrition, Human Nutrition Unit, Rome, Italy (FB)
2 Supported by the European Commission (Phytos QLRT-2000-00431). 3 Reprints not available. Address correspondence to F Branca, National Institute for Research on Food and Nutrition, Human Nutrition Unit, Via Ardeatina 546, 00178 Rome, Italy. E-mail: fbr{at}euro.who.int.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONConclusionREFERENCES |
|---|
Objective: The objective was to investigate whether the consumption of isoflavone-enriched foods for 1 y affects bone mineral density, bone metabolism, and hormonal status in early postmenopausal women.
Design: This was a randomized, double-blind, placebo-controlled, parallel, multicenter trial. Two hundred thirty-seven healthy early postmenopausal women [mean (±SD) age of 53 ± 3 y and time since last menses of 33 ± 15 mo] consumed isoflavone-enriched foods providing a mean daily intake of 110 mg isoflavone aglycones or control products for 1 y while continuing their habitual diet and lifestyle. Outcome measures included bone mineral density of the lumbar spine and total body, markers of bone formation and bone resorption, hormones, isoflavones in plasma and urine, safety variables, and adverse events.
Results: Consumption of isoflavone-enriched products did not alter bone mineral density of the lumbar spine and total body or markers of bone formation and bone resorption. Hormone concentrations did not differ between the isoflavone and control groups. Consumption of isoflavone-enriched products resulted in increased isoflavone concentrations in plasma and urine, whereas control products did not. This finding indicated good compliance with treatment. Subgroup analysis did not support an effect of equol phenotype on bone density. The intervention had no effect on a range of safety variables and reported adverse events.
Conclusion: Consumption of foods containing 110 mg/d of soy isoflavone aglycone equivalents for 1 y did not prevent postmenopausal bone loss and did not affect bone turnover in apparently healthy early postmenopausal white women. This trial was registered at clinicaltrials.gov as NCT00301353.
Key Words: Isoflavones • bone mineral density • menopause • hormones
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONConclusionREFERENCES |
|---|
Furthermore, limited attention has been paid to the precise amounts and bioavailability of the isoflavones present in the intervention products, and compliance with the test food has often been overlooked. Finally, studies have been carried out in homogeneous populations, so that phenotypical differences in isoflavone metabolism could not be considered.
The present study was therefore designed to investigate the effects of isoflavone-enriched food consumption for 1 y on BMD and biomarkers of bone turnover in early postmenopausal women from 3 European countries. In addition, the effect of the intervention on hormonal status and safety variables was evaluated.
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONConclusionREFERENCES |
|---|
Ethics
Protocols were approved separately for the 3 clinical sites. In the Netherlands, the protocol was approved by the independent Medical Ethics Committee of TNO. The procedures followed were in accordance with the International Conference on Harmonisation of Technical Requirements for registration of Pharmaceuticals for Human Use, guidelines for Good Clinical Practice, with the Helsinki Declaration of 1975, as revised in 2000 and with the Dutch Regulations on Medical Research involving Human Subjects (WMO, 1999). In Italy, the protocol was approved by the Medical Ethics Committee of the University of Rome "La Sapienza." In France, biomedical research is in pursuance of the Huriet state law (88-1138 from 20 December 1988). The protocol was approved by the review board committee from Auvergne (Comité Consultatif de Protection des Personnes dans la Recherche Biomédicale).
Participants
Subjects were healthy white females recruited via advertisements, direct mail, and gynecologists. All subjects gave written informed consent. Subjects had been menopausal for 12–60 mo, as established by interview and confirmed by a follicle-stimulating hormone (FSH) concentration >20 IU/L; were nonosteoporotic, as determined by questionnaire and dual-energy X-ray (DXA) scan of the lumbar spine (exclusion: < –2 z scores for BMD); and had a BMI of 22 to 29. Subjects with severe scoliosis, with a history of medical or surgical events that might significantly affect the study outcome, or who were using concomitant medication known to affect bone metabolism were excluded. Additional exclusion criteria were reported food allergy, allergy to sunscreen products, or unexplained weight loss or gain in the month before screening. Subjects were excluded if they had changed smoking habits in the past 2 mo, consumed >21 units of alcohol/wk, followed a weight-reducing diet, regularly consumed soy-based foods, or were taking supplements containing isoflavones in the 3 mo before enrollment. Subjects were asked to not start taking calcium or vitamin D supplements during the study, but subjects who were already taking such supplements were asked not to stop. Professional sportswomen were not allowed to participate.
Study products
The study products included isoflavone-enriched biscuits and bars and control biscuits and bars which were identical in composition, taste and appearance. A soy isoflavone concentrate (Cerestar, Vilvoorde, Belgium) containing 40–50% of isoflavones by weight (60–75% genistein, 25–35% daidzein, and 1–5% glycitein) was incorporated into biscuits (Danone, Palaiseau Cedex, France) and cereal bars (Novartis, Revel, France). Throughout the study, 3 batches of biscuits and 2 batches of bars were used. Between-batch reproducibility of the IF content was <10% for biscuits and <5% for cereal bars (F Brouns, S Doat, N Joqueviel, K Wahala, F Branca, personal communication, 2002). The isoflavone-enriched biscuits contained 55 mg isoflavone aglycones per portion (12 g, 222 kJ) and the isoflavone-enriched cereal bars contained 76 mg isoflavone aglycones (30 g, 480 kJ). The composition is given in Table 1
.
|
Subjects were requested to consume their habitual diet, which was monitored 3 times during the study: at baseline and after 27 and 53 wk by 3-d dietary records. Lifestyle was evaluated with a short food questionnaire and physical activity questionnaire. Assessment of climacteric symptoms was performed using the Greene climacteric scale (5).
BMD measurements and standardization
DXA measurements were performed with the use of Hologic (Waltham. MA) instruments in France (QDR 4500A) and Italy (Delphi W, S/N 70556) and Lunar equipment in the Netherlands (Prodigy, DF + 13364; Madison, WI). Bone mineral content (in g) and BMD (in g/cm2) of both the whole body and the lumbar spine were assessed. DXA technicians from each center were trained together to perform data acquisition following a standardized protocol. Each instrument was subjected to daily checks of performance, including measurement of the manufacturer's phantom; no significant drift was observed. A cross calibration was performed with an anthropomorphic spine phantom (DPA/QDR-1, S/N 1479; Hologic) consisting of 4 vertebrae, with 10 scans of this control being performed before each period of measurement (weeks 1, 27, and 53). As a result, the data from Italy and France could be pooled, whereas data from the Netherlands had to be corrected by using a multiplier (0.864). The corrected values are presented. All data were collated at the end of the study within a single database (EXCEL, Microsoft, Redmond, WA).
Blood and urine measurements
Fasting venous blood samples were collected at baseline and after 14, 27, and 53 wk. Serum and plasma were stored at –80 °C until analyzed, except those for isoflavone analyses, which were stored at –20 °C. Whole-blood samples for genomic DNA isolation were collected at baseline and stored at 4 °C until analyzed. Additionally, 24-h urine samples were collected at baseline and at 14, 27, and 53 wk. Urine was stored at –20 °C until analyzed. Samples for a specific measurement from all centers were analyzed in the same laboratory, except for the safety variables and blood lipids.
DNA extraction for genotyping was performed on lysed white blood cells after red blood cell lysis via the addition of protein precipitation solution and separation with isopropanol. The extracts were absorbed on clean absorbent paper. The following genotypes were investigated: VDR (Fok 1), VDR (Taq 1), ERa (Xba 1), and ERa (Pvu II).
Parathyroid hormone, N-propeptide of procollagen type I (PINP), 25-hydroxyvitamin D, and 1,25-dihydroxyvitamin D concentrations in plasma were measured by using commercial radioimmunoassay kits. Bone ALP in plasma was measured by using an antibody capture activity assay (Quidel, San Diego, CA). The urinary pyridinium cross-links pyridinoline (PYD) and deoxypyridinoline were measured after full acid hydrolysis as previously described (6). Urinary calcium and creatinine were assayed with an autoanalyzer.
Genistein and daidzein were analyzed in plasma, in duplicate, by radioimmunoassay (TR-FIA) with the use of an internal standard tritiated estradiol β-glucuronide (7). The sensitivity limit of the method is 1.8 pg/µL (0.35 nmol/L), being recovery calculated after hydrolysis and extraction.
Genistein, daidzein, equol, and O-desmethylangolensin (O-DMA) were analyzed in duplicate in urine samples by gas chromatography–selective ion monitoring mass spectrometry with the use of deuterium-labeled internal standards (8).
Total and HDL cholesterol and triacylglycerols were measured with enzymatic techniques (Boehringer-Mannheim, Mannheim Germany; CV < 3%), and LDL cholesterol was calculated according to Friedewald et al (9). Serum lipoprotein(a) was analyzed by rate nephelometry on a Hitachi 911 analyzer (Roche Diagnostis Nederland BV, Almere, Netherlands). Hormones (estradiol, FSH, luteinizing hormone, and sex hormone–binding globulin), clinical chemical variables [aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase (ALP), 
-glutamyltransferase, total protein, albumin, bilirubin, creatinine, urea, glucose, iron, sodium, potassium, and chloride], and hematology variables were measured by using routine clinical techniques (CVs < 4%).
Vaginal cytology
Smears were obtained at baseline and week 53 from the proximal portion of the left lateral fornix of the vagina. The Maturation Index (MI) was calculated, and maturation of the squamous epithelium was expressed as the percentile relation of parabasal cells to intermediate cells to superficial cells.
Statistical analysis
Statistical analysis and results are reported based on the data of all completers. Non-normally distributed residuals were naturally log transformed before statistical analysis. Statistical outliers, defined as outside the 95% confidence limits of the normal probability plots, were removed before the analysis. In all statistical tests performed, the null hypothesis (no effect) was rejected at the 0.05 level of probability.
Statistical analysis of treatment effects was based on repeated measurements analysis. Differences in mean values of the BMD variables, the bone markers in urine and plasma, body weight, serum hormones, and isoflavone concentrations in plasma and urine were evaluated by using a split-split-plot analysis of variance (ANOVA), with country as a whole plot, treatment as a subplot, and time as a sub-subplot. The time component was subdivided into an orthogonal linear component, a quadratic component, and, in the case of >3 measurement occasions, a cubic component. The linear component describes the increase or decrease over time, the quadratic component may reveal a curve pattern over time, and the cubic component is used to detect fluctuations in the mean concentrations over time. The analysis is used not only to investigate the main effects of treatment, country, and time, but also to evaluate the interaction between these factors.
Besides using the split-split-plot ANOVA, the BMD variables and the bone markers in urine and plasma were also analyzed by using a split-plot ANOVA, with treatment as a whole-plot and time as a subplot. This analysis was performed both averaged over countries as well as per country.
Differences in mean concentrations of BMD variables and isoflavone concentrations in plasma and urine for equol responders and equol nonresponders were evaluated by using a split-split-plot ANOVA, with country as a whole plot, the response category as a subplot, and time as a sub-subplot. Equol producer phenotype was defined as a serum equol concentration >1000 nmol/L on 1 of the 4 measurement occasions (0, 14, 27, or 53 wk). All analyses were performed by using SAS version 8.02 (SAS/STAT version 8.0; SAS Institute Inc, Cary, NC).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONConclusionREFERENCES |
|---|
). The main reason for dropout was failure to comply with the study procedures. The subjects were all between 1 and 5 y postmenopausal and were neither obese nor osteoporotic (Table 2). There were no differences in BMI between countries. The French subjects had a lower t score for lumbar spine BMD than did the other subjects, and the Dutch had higher baseline concentrations of FSH (Table 2). No marked differences with respect to inclusion or exclusion, population characteristics, or diet and lifestyle habits were observed between subjects allocated to the IF+ or IF– group. Similarly, no differences were observed in the energy and nutrient intakes of either group, although intakes of some nutrients differed between countries (data not shown). The frequency of consumption of coffee, tea, and carbonated soft drinks was also not significantly different. The French women tended to be less active than were the Dutch and the Italians, and the Dutch spent more time in strenuous activities (data not shown). Finally, no significant differences in VDR (Fok 1), VDR (Taq 1), ERa (Xba 1), and ERa (Pvu II) genotypes between the IF+ and IF– groups were observed (data not shown).
|
|
). Urinary equol excretion was higher in the Netherlands and France than in Italy. In total, 64 subjects were defined as equol producers: 23 (51%) in the Netherlands, 16 (30%) in Italy, and 25 (74%) in France. Fifty-four subjects were defined as nonproducers: 22 (49%) in the Netherlands, 23 (70%) in Italy, and 9 (26%) in France. Plasma concentrations of genistein and daidzein were not significantly different between equol producers and nonproducers (data not shown).
|
Effects on BMD
Over the year of the study, there was a significant decrease in total-body BMD (Figure 2, Plinear < 0.0001) and total lumbar spine (Figure 3 A, Plinear < 0.0001) both in the IF+ and the IF– groups, with no significant group differences and no significant interaction between time and treatment. However, there were significant country differences in BMD change (for total lumbar spine: Plinear x country = 0.0011, Pquadratic x country = 0.0087; for total body: Plinear x country < 0.0001). When cohorts were analyzed separately, the decrease was not observed in France (Figure 3B). However, no interaction between country and treatment effects was observed, which indicated that the differences between countries were similar for both treatment groups. Also, there was no treatment response in BMD associated with time since menopause (data not shown).
|
|
).
Bone markers
Neither markers of bone formation nor of bone resorption were affected by the IF treatment (Table 4). Time trends and differences between countries were observed (time, country, or time x country effects), but only a few interactions with treatment. For bone ALP, PINP, and deoxypyridinoline/creatinine, a marginally significant interaction (Plinear x treatment < 0.10) between the linear time component and treatment was found. This means that the slope for the IF– group was different from the slope of the IF+ group. For PTH, the interaction between the quadratic component and the treatment was marginally significant (Pquadratic x treatment < 0.10). The difference between the linear time trend for the control group and that for the isoflavones group differed between countries for PTH (P = 0.0003), but these changes were not associated with plasma concentrations of 1,25-dihydroxyvitamin D or 25-hydroxyvitamin D.
|
Effects on safety variables
Total cholesterol, LDL cholesterol, HDL cholesterol, and triacylglycerols were unaffected by IF treatment, and similarly negative results were obtained for total leukocyte count, liver enzymes (alanine aminotransferase, ALP, -glutamyl transferase), and vaginal cytology (data not shown).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONConclusionREFERENCES |
|---|
Long-term trials performed so far with soy isoflavone extracts or pure genistein show conflicting results on bone health. An improvement in BMD after daily consumption of 35 to 54 mg aglycone equivalents during 6–12 mo was reported by Morabito et al (10) and Clifton-Bligh et al (11). Chen et al (12) showed that a daily intake of 50 mg, but not of 25 mg, aglycone equivalents for 12 mo increased bone mineral content in Chinese postmenopausal women without affecting BMD. Potter et al (13) showed an increase in BMD of the lumbar spine in postmenopausal women receiving 56 mg aglycone equivalents/d contained in soy protein isolate for 6 mo; a lower dose, however, was shown to have no effect. On the other hand, Gallagher et al (14) and Kreijkamp-Kaspers et al (15) could not demonstrate an effect of soy protein isolate on bone density using daily doses of 4–103 mg aglycone equivalents for 9 or 12 mo.
Factors possibly influencing the effects of isoflavones on bone health are characteristics of the study population, study products, study design, and primary endpoint measurements. Our study included a homogeneous study population; time since last menses was restricted to the previous 12–60 mo (mean: 33 mo) and, because this period is characterized by rapid bone loss, we hypothesized that such a population would be most suitable to detect the bone-sparing effect of the dietary intervention. Population characteristics, physical activity, and dietary habits differed only marginally between countries and were comparable for the IF treatment and control groups. Furthermore, no differences in genetic variation in the ER gene, which may influence response to IF treatment (16), were present between the IF+ and IF– groups.
The bioavailability of isoflavones in the study products was confirmed in a pilot study (B Chanteranne et al, personal communication, 2002), in which 100 mg isoflavones was administered to 14 volunteers in each of the cohorts, which resulted in mean plasma genistein concentrations ranging from 1440 to 1490 nmol/L for the tested products and mean plasma daidzein concentrations ranging from 522 to 536 nmol/L. The increased plasma concentrations of isoflavones in the present study confirm the good bioavailability of the IF-enriched foods over a prolonged period of use.
The effects of phytoestrogens on bone have been proposed to be dose dependent, possibly related to a balance between their interactions with estrogen and peroxisome proliferator–activated receptors (17). We administered a daily dose of 110 mg isoflavones as aglycones from soy concentrate, which consisted predominantly of genistein (60–75%). Morabito et al (10) reported that 54 mg pure genistein/d was as effective as hormone replacement therapy in protecting against bone loss, reducing bone resorption, and increasing bone formation. Weaver and Cheong (18) suggested that the presence of different isoflavones may act antagonistically, whereas Clifton-Bligh et al (11) observed an increase in BMD after daily consumption of a red clover extract (35–54 mg aglycone equivalents) containing genistein, daidzein, formonometin, and biochanin. Also, Potter et al (13) used a daily dose of 90 mg isoflavones (56 mg aglycone equivalents) as soy protein isolate for 6 mo and showed an increase in BMD. These data support the notion that the food source or dose used in our study was not responsible for the observed lack of effect on bone.
The 1-y duration of our study should be sufficiently long to detect effects on both BMD and bone markers. Repeated measurements of BMD by DXA have been the primary means of assessing fracture risk and are accepted as an intermediate endpoint in the evaluation of nutritional and pharmacologic regimens (19-22). Changes in BMD can often be measured meaningfully by DXA after 6–12 mo from baseline (23). Although DXA measurements were performed in 3 different centers, standardized procedures and quality control ensured optimal data.
The study was adequately powered because the pooled SD for the change over time for the Netherlands and Italy was 1.78 (control group), which is lower than the SD used for the initial power calculations. Power calculation based on an SD of 1.78 and an n value of 45 (Dutch cohort) resulted in a power of 88% to detect a change of 1.2% and for Italy this power was 84% (n = 39). Thus, numbers in the current study population were sufficient to detect the minimal assumed difference of 1.2% in BMD. The reported overall compliance was good, as confirmed by the plasma IF concentrations achieved in all cohorts.
Of the lifestyle factors possibly interfering with the results, both diet and physical activity were monitored throughout the study. Although there were some differences between countries, diet and physical activity pattern were not different between the IF+ and IF– groups during the study. This was confirmed by mean body weights remaining stable throughout the study in both the treatment (65.6–66.1 kg) and control (66.1–66.8 kg) groups.
Another reason for the wide variability in response in the studies conducted to date may relate to the metabolism of isoflavones (24). There is increasing evidence that the clinical efficacy of isoflavones in humans depends on the ability to produce a microbial metabolite of daidzein, equol. It is well established that only 30–50% of any given population group can produce equol after ingestion of soyfoods (25-28) and the capacity to produce equol has been associated with circulating reproductive hormones (25), inversely associated with mammographic density (29, 30) and positively associated with BMD in response to isoflavone intervention (24, 31, 32). The hypothesis that the biological effects of IF is dependent on their further metabolism through the production of equol is not, however, supported by post hoc analysis of the present study population. Although the soy concentrate used in this study predominantly contained genistein, the daily daidzein intake (50 mg) is a dose that was sufficient for equol production in previous studies (24, 33). The wide variability in equol production observed in this study supports the observed higher production of equol among older women (34) and fits with recent data suggesting that the frequency of equol producers may be higher than previously observed in some subgroups of the population; for example, the equol producer phenotype is higher in vegetarians (59%) and in adults living in Asian countries, where soy products are consumed with greater regularity (35). Thirty-six to seventy-one percent of the treated subjects had a measurable production of equol, but their bone density and bone turnover were not affected differently from the nonresponders or the control group. However, to correctly test this hypothesis we would need to prospectively recruit equol producers.
Robust criteria were followed to standardize the study population, and quality-control criteria were drafted and externally monitored to ensure optimal data quality. The IF-enriched food intervention did not reduce postmenopausal bone loss, nor was there any influence on bone turnover or hormonal status. These results, therefore, do not support the proposed use of isoflavone-enriched foods or supplements in early menopause to prevent bone loss. There were, however, no concerns over safety in sustaining the relatively high dosage of 110 mg/d for 1 y, as determined by the measured variables, ie, reported adverse events, blood chemistry, hematology, blood lipids, and vaginal maturation.
|
Conclusion |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONConclusionREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
The authors' responsibilities were as follows—EB, VC, and FB: responsible for the execution of the study and local (laboratory) analyses in the 3 separate centers; EB: responsible for the overall data management and statistical analysis and wrote the manuscript; VC: responsible for standardization and processing of the BMD measurements; FB: responsible for coordination of the complete project; SR: responsible for the centralized analysis of the bone markers; KW: responsible for the analysis of isoflavones; and AC: responsible for the laboratory analyses of the hormones. All authors approved the manuscript, were involved in the design of the study and took part in the discussion of the results, the writing process, and the final compilation of the manuscript. None of the authors reported a conflict of interest.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONConclusionREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  W. W Wong, R. D Lewis, F. M Steinberg, M. J Murray, M. A Cramer, P. Amato, R. L Young, S. Barnes, K. J Ellis, R. J Shypailo, et al. Soy isoflavone supplementation and bone mineral density in menopausal women: a 2-y multicenter clinical trial Am. J. Clinical Nutrition, November 1, 2009; 90(5): 1433 - 1439. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. M Kenny, K. M Mangano, R. H Abourizk, R. S Bruno, D. E Anamani, A. Kleppinger, S. J Walsh, K. M Prestwood, and J. E Kerstetter Soy proteins and isoflavones affect bone mineral density in older women: a randomized controlled trial Am. J. Clinical Nutrition, July 1, 2009; 90(1): 234 - 242. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Hooper, J.J. Ryder, M.S. Kurzer, J.W. Lampe, M.J. Messina, W.R. Phipps, and A. Cassidy Effects of soy protein and isoflavones on circulating hormone concentrations in pre- and post-menopausal women: a systematic review and meta-analysis Hum. Reprod. Update, July 1, 2009; 15(4): 423 - 440. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. W Lampe Is equol the key to the efficacy of soy foods? Am. J. Clinical Nutrition, May 1, 2009; 89(5): 1664S - 1667S. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
, n = 118) and control (IF–; 
, n = 119) groups. There was a significant decrease over time, Plinear < 0.0001 (split-split-plot ANOVA).
); IF–, n = 39 (•)], and France [IF+, n = 34 (
); IF–, n = 34 (
)]. C: Data for 118 IF+ subjects divided into 64 equol producers (
) and for 119 IF– subjects (•). There was a significant decrease over time, Plinear < 0.0001 (split-split-plot ANOVA).