Phytoestrogens: End of a tale?

Abstract

Phytoestrogens are plant‐derived hormone‐like diphenolic compounds of dietary origin that are present at high levels in plasma of subjects living in areas with low atherosclerosis and cancer incidence. The term phytoestrogen is commonly applied to the soy isoflavones genistein, daidzein and glycitein. As outlined in a previous review article in this journal by Adlercreutz and Mazur , these compounds are weakly estrogenic and appear to influence the cardiovascular system, the production, metabolism and biological activity of sex‐hormones, as well as malignant cell proliferation, differentiation and angiogenesis. Recently skepticism has developed concerning the true potential of phytoestrogens to beneficially modify these processes. A critical analysis of the early findings from supplementing the diet with soy protein has failed to confirm phytoestrogens as the responsible agent for beneficial cardiovascular effects, be it by way of lipid reduction, vasodilation or lipoprotein oxidation. Furthermore, contrasting data have been reported on the potential of phytoestrogens to prevent hormone‐dependent cancers (e.g. breast and prostate) and to successfully treat post‐menopausal complaints, an indication for which they are widely used. These potentially negative findings have led health authorities in several countries to suggest maximum daily intake levels for phytoestrogens. There is now growing interest in the use of soy products containing low levels of phytoestrogens and in research on other phytoestrogen free legumes such as lupin.

• Atherosclerosis
• isoflavones
• menopause
• osteoporosis
• phytoestrogens
• soy
• regulatory issues
• toxicity

Introduction

Phytoestrogen is a general term given to a large number of plant‐derived estrogen‐like compounds found in legumes, seeds, fruits and vegetables. The attitude of the scientific community towards these compounds has generally been positive: for example a 1997 review article in Annals of Medicine 1 discussed a large number of findings suggesting potentially beneficial effects of phytoestrogen consumption, in particular against arterial disease and cancer. The first ever literature report on the effects of phytoestrogen intake, however, described a potentially toxic activity, related to multiple fertility problems in sheep grazing on pastures based on red clover 2. This plant contains high amounts of two phytoestrogens, formononetin and biochanin A. Subsequent studies showed that these and other phytoestrogens interact with the estrogen receptors ERα or ERβ, due to their structural similarity to 17β‐estradiol 3.

Currently, four groups of phenolic compounds are classified as phytoestrogens 3: the isoflavones, stilbenes, coumestans, and lignans (Figure 1). The main stilbene is resveratrol, found primarily in grapes and peanuts 3. Although there are two isomers, cis and trans, only the latter shows estrogenic activity. Resveratrol is biosynthesized only in grape skin; therefore white wines contain only traces, whereas higher levels of resveratrol are found in red wines, fermented with skins 4, 5. Only a few coumestans are characterized by estrogenic activity, the most important being coumestrol, whose main dietary source is legumes, although it has also been reported in other vegetables, such as Brussels sprouts and spinach 6. ‘Lignans’ is a general term for a large family of compounds. One important example, matairesinol, is a non‐estrogenic dimer, converted by gut microflora to enterolactone, which is estrogenic and easily absorbed 7. The main dietary source of lignans is flax‐seed, but they are present also in whole wheat flour, fruits, and tea 3. Isoflavones are a subclass of the flavonoids, which are found almost exclusively in soybeans and in a few other legumes including red clover. The main isoflavones are genistein, daidzein and glycitein, which in soybean and soy‐based foods exist as 7O‐glucosides 3. As isoflavones are the most widely investigated phytoestrogens, unless specifically indicated, throughout this review the term ‘phytoestrogens’ will refer specifically to this class of compounds.

Figure 1 Chemical structures of major phytoestrogens compared to 17β‐estradiol. Genistein and the chemically related isoflavones daidzein and glycitein are the most widely used compounds. Genistein, daidzein and glycitein are generally labelled as “phytoestrogens” and they are thus the major topic of the present review article.

Key messages

• Critical analysis of clinical findings from studies on soy‐treated subjects has failed to identify phytoestrogens as responsible agents for cardiovascular benefit, anti‐cancer properties, and beneficial effects on post‐menopausal complaints.

• Proteins from soy seem to provide essentially all of the clinical benefit, at least for cardiovascular indications.

• Regulations in some nations restrict use of soy phytoestrogens to relatively low daily intakes.

Phytoestrogens: Metabolism and intake

Phytoestrogens are primarily ingested in the form of their glycosides, genistin, daidzin and glycitin. De‐conjugation occurs in the gastrointestinal tract through the activity of microflora to produce the free aglycones genistein, daidzein and glycitein, which are readily absorbed, re‐conjugated in the liver and then undergo enterohepatic circulation 8. It is however the minor fraction of free and sulphated forms that are considered biologically active 9. Excretion occurs primarily through the urine (conjugated forms) and feces (unconjugated forms).

Differences in individual gastrointestinal metabolism may be an important factor determining the efficacy of these compounds for disease risk reduction. Apart from deconjugation, they can also be further metabolized within the intestine by resident microflora. For example, in around a third of the population (‘good equol producers’) daidzein is readily converted via dehydroequol to equol, leading to high levels of equol detectable in the urine 10–12. This conversion appears to depend on an individual's endogenous gastrointestinal microflora, which may in turn be modulated by habitual diet 11. All laboratory rats can however convert daidzein to equol 13 and this may help explain some of the discrepancies in the effects of phytoestrogens in rats compared to humans. It has been proposed that the ability to produce equol may be associated with specific health benefits 13 (see the next two sections).

The phytoestrogen intake in the average East and Southeast Asian diet is estimated to be around 20–50 mg/day 14, 15, whereas in Western countries due to limited soy product consumption, it is much lower. Intakes ranging from 0.15 mg/d to >3 mg/d have been reported in the United States 16. Data regarding European countries are very scarce: typical values range from 0.63 to 1.00 mg/d in men and from 0.49 to 0.66 mg/d in women 17, 18.

Growing knowledge of the physiological effects of phytoestrogens has gone hand in hand with a rapid growth in the use of soy in Western diets. The growth of soy consumption was further catalyzed by the United States Food and Drug Administration (US FDA) decision to allow labeling of soy protein rich foods with the indication that 25 g/day can help reduce cardiovascular disease risk 19.

As a result of the wide popularity of soy in general, phytoestrogens, as one of the key soy components, have also gained a reputation for being beneficial to health. This hypothesis arose from a number of observations, ranging from the recognition that phytoestrogens have estrogen‐like activity in some animal models, to the low prevalence of cardiovascular disease and hormone‐dependent cancers of the breast, endometrium and prostate in Asia, where phytoestrogen intake is relatively high. In addition, a large number of studies have demonstrated the hypocholesterolemic properties of soy proteins 20, frequently containing isoflavones. These findings have wrongfully led to the conclusion that phytoestrogens are the active beneficial agents within soy protein.

Potential cardiovascular benefits

A number of epidemiological observations have supported a protective role of phytoestrogens in modulating cardiovascular disease (CVD) risk markers. Consumption of soy products has been associated with reduced serum cholesterol21 and lignans have also shown a moderately protective effect on triglyceridemia 22. The US FDA decision to allow promotion of soy protein for heart health was based on a meta‐analysis indicating a significant total and low density lipoprotein (LDL) cholesterol reduction following an average soy protein intake of 47 g/day 23. However, studies reporting significant cholesterol reduction within the meta‐analysis (that were essentially the basis for the approval of the health claim by the US FDA) were on relatively severe cases of hypercholesterolemia and were mainly carried out in Italy and Switzerland using soy proteins with very low isoflavone levels 24, 25. A possible role of phytoestrogens in cholesterol reduction was however indicated by clinical studies in which they were removed from the soy protein preparations hot ethanol extraction 26. This rather extreme chemical treatment resulted in a dramatic reduction of the hypocholesterolemic properties of the phytoestrogen‐free soy protein in animal models of hypercholesterolemia 27. However, in a study on ovariectomized adult female cynomolgous monkeys, a phytoestrogen rich ethanol extract of soybean did not exert any lipid lowering effect 28. This finding and the fact that ethanol treatment removes bioactive compounds other than phytoestrogens and may denature the protein, tends to invalidate any conclusion on the hypocholesterolemic activity of phytoestrogen based on the ethanol extraction studies.

Little support for the role of phytoestrogens in modulating cholesterolemia has been gained from human randomized placebo‐controlled studies using isolated phytoestrogens in tablet/capsule form or soy protein with varying levels of phytoestrogens 29–31. These studies have generally failed to show an effect of phytoestrogens on cholesterolemia independent from that of soy protein.

Recently, a comparison of different commercial soy protein isolates performed by some of the authors of this review using proteomic techniques indicated that soy protein isolates, used in clinical studies in the US, showed extensive proteolysis 32: this was even more significant in samples submitted to ethanol extraction. The damage of the protein structure may possibly be related to the loss of the hypocholesterolemic activity, a fact that has certainly been underestimated in available literature. In fact, a soy protein product, in which the isoflavones had been removed by mild column chromatography, proved to maintain its hypocholesterolemic properties 33, thus providing evidence that specific components of the proteins are most likely responsible for the cholesterol lowering properties of soybeans 34.

At a mechanistic level, low density lipoprotein receptor (LDL‐R) stimulation is believed to be the route by which soy protein preparations elicit their hypocholesterolemic activity 35–38. There have been a number of in vivo investigations showing this same effect suggesting that peptides resulting from soy protein digestion may enter the circulation and exert beneficial effects on cholesterol metabolism in the liver. An early study at the University of Milan 39 showed an 8‐fold increase in LDL receptor activity in isolated lymphomonocytes from severely hypercholesterolemic patients treated with soy proteins. A similar finding, but with lesser increase of LDL receptors was reported by Baum et al. 40 in post‐menopausal women with less dramatic elevations of LDL cholesterolemia. In rats the purified 7S α' component of soy protein has been found to elicit a 10‐fold higher hypocholesterolemic activity than that exerted by a synthetic hypocholesterolemic drug 41. The major soy phytoestrogen genistein, however, at concentrations of up 1 mg/ml, failed to demonstrate any LDL‐R stimulatory activity in studies on HepG2 cells 38.

A recent consensus paper 42 indicates that both soy protein and isoflavones may be needed for the maximal cholesterol lowering effect of soy, also recommending a diet low in saturated fat and cholesterol to promote heart health. It is however difficult at present to identify possible mechanisms whereby phytoestrogens may exert any additional effects to those exerted by soy protein. It has been considered that by binding to estrogen receptors, phytoestrogens may stimulate LDL receptors. By using phytoestrogen concentrations elevated far beyond physiological levels, Borradaile et al. 43 show reduced secretion of apolipoprotein B and increased LDL receptor activity in liver cells: the clinical relevance of these findings is however questionable.

Soy phytoestrogens do exert some vasodilatory activity in special conditions. Acute intravenous administration of genistein or daidzein was evaluated in healthy humans of both sexes 44. Genistein was infused for 6 minutes at concentrations of between 10 and 300 nmol/min. At the two highest doses a significantly increased forearm arterial flow was observed in both sexes. Similar effects were exerted by equimolar amounts of 17β‐estradiol. Both genistein and 17β‐estradiol effects were antagonized by a nitric oxide (NO) synthase antagonist. The plasma genistein concentrations after i.v. administration were, however, 8–10‐fold higher than those observed after oral intake of high phytoestrogen soy proteins. In a randomized double blind trial, genistein intake (54 mg/day) significantly increased flow‐mediated endothelium dependent vasodilation in post‐menopausal women 45 with a significant increase in serum nitrates after 6 months of treatment along with a 50% reduction of plasma endothelin‐1. The discovery of this NO dependent mechanism was supported by a one‐year investigation on post‐menopausal women where genistein (54 mg/day) was compared to standard hormone replacement therapy 46. Flow‐mediated dilatation significantly increased from 3.9% to 7% supporting a potential protective role of genistein in conditions of altered vasomotility. It should, however, be pointed out that another study in a similar series of women given purified phytoestrogens failed to confirm these findings and again did not report any plasma lipid changes 47. Similarly, a recent randomized controlled trial of phytoestrogen‐containing soy protein versus casein in 202 post‐menopausal women found no beneficial effect of the soy diet on vascular function 48.

There is no clinical evidence currently available on the effects of isoflavones on atherosclerosis, as assessed by the standard clinical procedures of carotid intima‐media thickness, or coronary angiography. A report in diabetic and non‐diabetic monkeys has however reported a significantly reduced delivery of lipoproteins into monkey arteries after a phytoestrogen rich soy diet 49: no effort was however made to assess the separate contributions of proteins and phytoestrogens.

Another beneficial mechanism attributed to phytoestrogens is their potential antioxidant activity. The consumption of soy proteins compared to casein does lead to a significant decrease in arterial lipid peroxidation, a putative mechanism of atherosclerosis development/progression. In vitro and in vivo studies investigating the antioxidant effects of phytoestrogens 50 concluded that they act as antioxidants directly or indirectly by enhancing the anti‐oxidant enzyme activities of catalase, superoxide dismutase, glutathione peroxidase and glutathione reductase. Arterial lipid peroxidation levels were ∼17% lower in post‐menopausal non‐human primates fed a soy protein isolate containing phytoestrogens, compared to animals fed casein and lactalbumin as the protein sources 51. Tikkanen et al. 52 examined the effects of feeding soy protein containing 60 mg phytoestrogens/day on copper‐induced LDL oxidation in six healthy volunteers. Two weeks of soy consumption significantly prolonged the LDL oxidation lag time by ∼ 20 min.

More recently the effects on in vivo lipid peroxidation and resistance of LDL to in vitro oxidation in men and post‐menopausal women fed a textured soy protein diet with either high or low phytoestrogen content were reported. The phytoestrogen‐rich diet induced significantly lower levels of the F2 isoprostane biomarker of lipid peroxidation, 8‐epi‐prostaglandin F_2α (−19.5%) and, in addition the lag time for copper‐mediated LDL oxidation was significantly prolonged 53. However, in a study using phytoestrogens in pill form (86 mg/day) 54, there was no evidence of reduced LDL oxidation. This latter finding suggests a direct antioxidant activity of soy proteins per se, an hypothesis also supported by the findings of Castiglioni et al. 55 that showed a powerful hypolipidemic and anti‐atheromatous activity of an essentially phytoestrogen free soy protein diet in rabbits. Moreover, a very recent study in mildly hypercholesterolemic individuals found very little effect of soy protein or phytoestrogens on plasma antioxidant capacity or biomarkers of oxidative stress 56. Similar inconclusive findings were also observed in a large Dutch study (Dutch Prospect‐EPIC cohort) on 16,165 women from 49 to 70 years of age that carefully evaluated phytoestrogen intake. In this study phytoestrogens were not associated with decreased CVD risk. However when stratifying for ‘ever’ versus ‘never’ smokers, CVD risk did decrease with increasing intake of lignans in ‘ever’ smokers 57.

Recently, some evidence has arisen that the metabolism of genistein to equol that occurs in some humans may be related to a reduced cardiovascular disease risk 13. In a crossover dietary intervention study of 26 mildly hypercholesterolemic and/or hypertensive volunteers, Meyer et al. 10 found no difference in the effect on serum lipids of a three‐week consumption of a phytoestrogen‐containing soy protein compared to dairy protein. However when the data of the 8 good equol producers were analysed separately, significantly lower levels of total cholesterol, LDL‐cholesterol, triglycerides and lipoprotein(a) were found after the soy protein diet. In addition, equol reportedly is a better antioxidant than daidzein ex vivo58 and unlike daidzein, dehydroequol has demonstrated vasodilatory properties in humans 59. In the study by Kreijkamp‐Kaspers et al. 48, beneficial changes (though not statistically significant) in blood pressure and endothelial function after a phytoestrogen‐containing soy‐supplemented diet were observed only in the sub‐set of equol producing subjects.

A summary of the randomized clinical studies investigating the effect of isolated phytoestrogens or low versus high phytoestrogen‐containing soy on risk factors for cardiovascular disease is given in Table I. At present, there appears little evidence for a major contribution to cardiovascular health of soy components other than the protein itself. Recent experimental findings indicate that some peptide fractions from soy protein given orally may in fact reduce cholesterolemia in animals to a greater extent even than lipid lowering drugs 41 and appropriately genetically modified soy proteins can reduce blood pressure in animals at doses achievable in man 60. In addition, naturally phytoestrogen‐free proteins from other legumes, e.g. lupin, lower cholesterolemia to a similar extent as soy proteins 61.

Table I. Summary of randomized clinical studies investigating the effect of isolated phytoestrogens or low versus high phytoestrogen‐containing soy on risk factors for cardiovascular disease.

Source	Design	Subjects	Intervention	Results	Conclusion
Nestel et al. 1997 29	Placebo‐controlled cross‐over, 5–10 week treatments	21 women	Phytoestrogens (80 mg/day) and placebo	No effect on plasma lipids but significant improvement in systemic arterial compliance in phytoestrogen group compared with placebo	Phytoestrogens did not improve plasma lipids but improved measures of arterial health
Hodgson et al. 1998 30	Parallel, randomized, double‐blind, placebo‐controlled, 8 weeks	46 men and 13 postmenopausal women not receiving hormone replacement therapy	Phytoestrogen tablets (55 mg/day) or placebo tablets	Serum lipids did not differ significantly between groups	In subjects with average serum cholesterol levels phytoestrogens do not beneficially modify serum lipids
Samman et al. 1999 54	Randomized cross‐over, four menstrual cycles (∼ 4 months)	14 healthy pre‐menopausal women	Phytoestrogens (86 mg/day) and placebo	Phytoestrogen intake did not lead to changes in LDL oxidizability or plasma levels of total cholesterol or triacylglycerols	Did not support hypothesis that phytoestrogens are a major active hypocholesterolemic agent
Simons et al. 2000 47	Double‐blind, placebo‐controlled, randomized crossover, 8 weeks each treatment	20 healthy, post‐menopausal women, with evidence of endothelial dysfunction	Soy phytoestrogen tablets (80 mg/day) or placebo	Blood pressure, plasma lipids and flow‐mediated endothelium‐dependent dilatation were not affected by phytoestrogens	No significant effect of phytoestrogens on plasma lipids or endothelial function
Wiseman et al. 2000 53	Randomized, crossover, 17 days each treatment	24 subjects (men and women aged 19–40 years)	Soy diets low (1.9 mg/day) or high (56 mg/day) in phytoestrogens	Biomarkers of in vivo lipid peroxidation lower and lag time for copper‐ion‐induced LDL oxidation longer after high phytoestrogen soy	High phytoestrogen soy demonstrated increased antioxidant action
Dewell et al. 2002 31	Parallel, randomized, 8 weeks	36 moderately hypercholesterolemic, elderly, postmenopausal women	Phytoestrogen (150 mg/d) or placebo	No effects on serum lipids	Phytoestrogens may not be effective at reducing cardiovascular disease risk in the population tested
Squadrito et al. 2002 45	Parallel, double‐blind, placebo controlled, randomized, 6 months	60 healthy postmenopausal women	Genistein (54 mg/day) or placebo	Genistein led to increased level of nitrites/nitrates, decreased level of plasma endothelin‐1 and increased flow‐mediated, endothelium‐dependent vasodilatation	Flow‐mediated endothelium dependent vasodilatation was improved by phytoestrogen therapy
Squadrito et al. 2003 46	Parallel, double‐blind, controlled, randomized, 1 year	79 healthy postmenopausal women	Genistein (54 mg/d), estrogen/progestin therapy or placebo	Genistein increased levels of nitrites/nitrates, decreased plasma endothelin‐1 levels and improved brachial artery flow‐mediated dilation similarly to estrogen/progestin therapy	Genistein and estrogen/progestin improve endothelium function to a similar extent
Vega‐Lopez et al. (2005) 56	Randomized crossover design	42 hypercholesterolemic men and postmenopausal women aged >50 years	Soy protein (trace or 50 mg phytoestrogens/ 1000 kcal) or animal protein (trace or 50 mg phytoestrogens/1000 kcal)	No significant effect of phytoestrogens on LDL oxidation, urinary F(2)‐isoprostanes, or protein carbonyl groups	Plasma antioxidant capacity and biomarkers of oxidative stress are not modified by soy‐derived phytoestrogens

Hormone dependent tumors

The endocrine effect of phytoestrogens was suggested as a potential mechanism to explain the epidemiological observation that Asian women have a significantly lower rate of breast cancer (BC) than Western women. The soy phytoestrogen genistein is known to bind rather weakly to the classical estrogen receptor (ERα), but binds with much higher affinity to ERβ 62 and may, therefore, display more potent effects in tissues with a higher ERβ expression such as the arteries. After endothelial denudation 63, expression of both the ERα and ERβ are both increased, but ERβ is over‐expressed to a much greater extent (∼30‐times more), versus the case of the uterus. Treatment of rats with various doses (0–2.5 mg/kg) of either 17β‐estradiol or genistein resulted in protection against cell proliferation, confirming a selective inhibition by genistein of the migration and proliferation of vascular smooth muscle cells 64, 65.

A number of data have indicated an inverse association between BC risk and phytoestrogen intake 66, both in relation to consumption and urinary excretion. Further, Asian women who move to the US and adopt a Western diet lose the lower risk of BC 67. Lignans have also received attention for their contribution to reduced BC risk 68. However, epidemiological studies have not provided consistent results. There was no relationship between soy food consumption and BC risk in a case‐control study in Chinese women 69 or in a more recent prospective study, in Japanese women 70. In this last study, however, consumption of miso soup several times per week was associated with a reduced BC risk. BC risk was not affected by phytoestrogen intake in a case‐control study involving multi‐ethnic American women 71 and similar inconclusive findings were reported in post‐menopausal Dutch women 72.

Lowered BC risk may be related to the ability of some humans to metabolize daidzein to equol, independent of level of isoflavone intake. Urinary excretion of equol (used as a surrogate of intake) was found to be inversely proportional to breast cancer risk in a case controlled study 73. Equol excretors also exhibited hormonal profiles more indicative of lower breast cancer risk than non‐excretors 12. This possible cancer protective effect of equol could be related to differences in the estrogenic properties of equol compared to its parent molecule 74, 75.

In the light of these contradictory findings, it has been hypothesized that phytoestrogens may protect against BC by a mechanism independent of their activity on estrogen receptors. They can, in fact, inhibit enzymes involved in the synthesis of steroid hormones, including aromatase and 17 β‐hydroxysteroid dehydrogenase 76. In addition, genistein has the capacity to inhibit tyrosine kinases, DNA topoisomerase, and angiogenesis 77. One last confusing issue is that of age of exposure. Lamartiniere et al. 78 demonstrated that the maximal protection against chemically induced BC in adult rats is obtained from exposure to genistein pre‐pubertally and again in adulthood. Therefore, possible use of phytoestrogens for BC protection could be complicated by the need to expose very young individuals where other safety issues (described below) may be present. Further, caution has been recently been suggested when using phytoestrogens in women treated with tamoxifen, since low intakes of phytoestrogen may antagonize the therapeutic effect of this estrogen antagonist 79.

Contradictory findings have also been reported for the potential role of phytoestrogens in protection against prostate cancer (PC). There appears to be a lower rate of PC in Asian men relative to Western men 80 and this, like BC risk in women, is associated with a higher intake of phytoestrogens. Elevated phytoestrogen levels can be detected in blood, urine and prostatic fluids of Asian compared to Western men 8, and an increased risk of developing PC occurs when Asian men are exposed to a Western diet 81. However, epidemiological studies relating intake of phytoestrogen‐rich soy and PC risk have not been consistent. Prospective studies in Japanese‐Hawaiian men and California's Seventh Day Adventists indicated a reduced PC risk following consumption of tofu or soy milk 82 and two US case‐control studies found a significant inverse relationship between soy food consumption and PC risk 83. However, two case‐control and one prospective study in Asian countries could not find an association between soy consumption and PC risk 84–86. Further, there has been no evidence that soy consumption may reduce prostate specific antigen (PSA) levels 87, 88, although an international study focusing specifically on lignan‐rich flaxseed in patients awaiting prostate surgery did find some reduction in testosterone, free androgen index and tumor proliferation index, but again with no change in PSA 89. There remains the possibility that PC protection may be related to mechanisms such as inhibition of 5α‐reductase, responsible for converting testosterone to dihydrotestosterone, or other enzymes regulating steroid hormone biosynthesis. Nevertheless the evidence of a protective effect of phytoestrogens on PC is far from proven.

Results of studies investigating the activity of phytoestrogens on other types of cancer have been even less conclusive. There is an apparent relationship between consumption of soy foods and gastric cancer; this appears to depend on whether or not the soy food is fermented 90. Fermented soy foods may possibly increase the risk of stomach cancer because of their high content of N‐nitroso compounds or other unidentified pro‐carcinogenic components 90. Soy intake and colon cancer have been found to be inversely related, but some data on miso also indicated an increased risk of rectal cancer 91, 92. Finally consumption of soy products has been linked to a decreased risk of endometrial cancer in a case‐control study performed in Hawaii 93.

Osteoporosis and bone health

Reducing the risk of osteoporosis is among the most relevant health topics in post‐menopausal women, with estrogen production being a major contributing factor. Hormone replacement therapy (HRT) can ameliorate the loss of estrogen in the menopause and provide clear benefit against osteoporosis 94, and a synthetic phytoestrogen, ipriflavone, was found to reduce bone loss in post‐menopausal women 95. There is substantial interest in alternative therapies for osteoporosis risk reduction, particularly those of dietary origin. A bone‐preserving effect of phytoestrogens has been supported by a number of epidemiological studies, based on observations of a significantly lower risk of fractures among Asian women compared to Caucasian women 96. Observational studies have also indicated that soy intake in pre‐ and post‐menopausal women is significantly associated with improved bone mineral density (BMD) 97, 98. In a Japanese study, in 478 post‐menopausal women, there was evidence of significantly different BMD adjusted to post‐menopausal years in the group with highest versus lowest daily intake of phytoestrogens (P<0.01) 99. These findings were not however supported by a 10‐year follow‐up study of post‐menopausal women in the Netherlands where no association was found between bone changes and the excretion of phytoestrogens 100.

While observational studies have provided some evidence of the osteoporosis‐protective role of phytoestrogens, controlled investigations have provided inconsistent results. Potter et al. 101 investigated supplementation with two different doses of phytoestrogens for 6 months on the effect on BMD. This double‐blind study noted significant increases (P<0.005) in both bone mineral content and density in lumbar spine (but not elsewhere), in the high dose phytoestrogens group (90 mg/day) versus controls. In a controlled study, also on pre‐ and post‐menopausal women, soy protein isolates containing different amounts of phytoestrogens (from 8 to 130 mg/day) failed to induce significant changes in the markers of bone turnover, though a possible reduction of osteocalcin, insulin‐like growth factor (IGF)‐1 and IGFB‐3, following high‐dose phytoestrogens was seen in the post‐menopausal group 102. Therefore this study did not support any useful effect of phytoestrogens on bone turnover. The conclusion of these earlier studies was that post‐menopausal women should not be advised to replace HRT with phytoestrogens in order to improve bone health 103.

More recently some beneficial effects of phytoestrogens have been noted when evaluating lumbar spine BMD and bone mineral content (BMC) by dual‐energy X‐ray absorptiometry. In the study by Alekel et al. 104, women were treated with a phytoestrogen‐rich soy (SOY+, 80.4 mg aglycones/day) a low phytoestrogen soy (SOY−, 4.4 mg/day) or a whey protein control. Both SOY+ and SOY− groups experienced no loss of BMD and BMC in the lumbar spine, whereas a loss in BMD and BMC occurred in the whey protein control group (−1.28% and −1.73% respectively, both P<0.005). Regression analysis identified significant positive effects of SPI+ on BMD (+5.6%) and BMC (+10.1%). A 2‐year study on the effects of a soy milk analogue (SOY+, 76 mg/day phytoestrogens) compared to transdermal progesterone (TPD+), the combination of the two or a placebo control was carried out on post‐menopausal Danish women with established osteoporosis or at least 3 risk factors 105. BMD and BMC were measured in the lumbar spine and hip using X‐ray absorptiometry. Confirming the study by Alekel et al. 104, increases in the percentage changes in lumbar spine BMD and BMC after 2 years were observed in the SOY+ group (+1.1% and +2.0% respectively) and the TDP+ group (+1.1% and +0.4% respectively). A significant bone loss occurred in the group on no treatment (−4.2% and −4.3% respectively) though also in the combined treatment group (SOY+/TDP+) (−2.8% and −2.4% respectively). This study therefore provides evidence that two glasses of soy milk a day, containing 76 mg/phytoestrogens, may prevent lumbar spine bone loss with similar efficacy to transdermal progesterone. However, the negative effect of the combination of the two treatments is difficult to explain. Further, the authors did not evaluate a classical HRT but just a progesterone addition. Somewhat similar findings were also reported by Chen et al. 106 and in abstract form by Vitolins et al. again in a 2 year study 107. In contrast, two recent, large randomized trials on post‐menopausal women showed no clear activity of phytoestrogens on bone changes 108, 109.

A very recent study on calcium metabolism carried out in 15 post‐menopausal women using metabolic balance and kinetic modeling 110 did not confirm earlier conclusions that phytoestrogens have at best a calcium sparing effect 8 and, as reported in one study 111, an apparent stimulatory effect on osteoblastic activity. This randomized crossover investigation of three, one‐month controlled dietary interventions was carried out 110: soy protein with phytoestrogens (SOY+, 82 mg/day), soy protein without phytoestrogens, (SOY−) and a casein/whey protein enriched diet. There was evidence of a lower urinary calcium excretion (P<0.01) with consumption of either soy based diet (SOY+ and SOY−) i.e. 85 and 80 mg/day, compared to the control diet (121 mg/day). However, fractional calcium absorption was unaffected by the treatments and overall no effects on bone deposition, resorption and calcium retention were seen, indicating no overall effect of phytoestrogens on calcium metabolism.

A summary of randomized clinical studies investigating the effect of phytoestrogens on osteoporosis and bone metabolism markers is given in Table II. The reported effects of phytoestrogens on osteoporosis appear mixed, some studies showing mild benefit others no effect. The majority of trials have not, however, strongly supported clear beneficial effects 112, 113.

Table II. Summary of randomized clinical studies investigating the effect of phytoestrogens on osteoporosis and bone metabolism markers.

Source	Design	Subjects	Intervention	Results	Conclusion
Potter et al. 1998 101	Parallel double blind, 6 months	66 free‐living postmenopausal women	40 g protein/day ‐ either casein or isolated soy protein (1.39 or 2.25 mg phytoestrogens/g protein)	Significant increases in bone mineral content and density in the lumbar spine but not elsewhere for higher phytoestrogen group only	Only the higher phytoestrogen level protected against spinal bone loss.
Wangen et al. 2000 102	Randomized, cross‐over, 3 month each treatment	30 pre‐ and post‐menopausal women	3 soy protein isolates containing 8, 65, and 130 mg/day phytoestrogens	Only small, not clinically relevant changes observed in markers of bone turnover	Did not provide evidence that phytoestrogens beneficially affect bone turnover
Alekel et al. 2000 104	Parallel, 24 weeks	69 perimenopausal women	Phytoestrogen‐rich (80.4 mg/d) soy, phytoestrogen‐poor soy or whey control	High phytoestrogen treatment had a positive effect on lumbar spine bone mineral density and content	Lumbar spine bone loss slower due to phytoestrogens
Lydeking‐Olsen et al. 2004 105	Double‐blind parallel, 2 years	89 postmenopausal women with established osteoporosis or at least 3 risk‐factors	Soy containing phytoestrogens (76 mg/day), transdermal progesterone, the combination of the above two or placebo	Lumbar spine bone mineral density and content lost in placebo and combination group but not in soy containing phytoestrogen and progesterone group	Phytoestrogens prevented lumbar spine bone loss alone but not in combination with progesterone
Chen et al. 2004 106	Parallel, randomized, 1 year	203 postmenopausal women	Mid‐dose phytoestrogens (40 mg/day), high‐dose (80 mg/day) and placebo	Significant beneficial effect of phytoestrogens on bone mineral content (total hip and trochanter) in late postmenopausal women with lower body weight or lower calcium intake	Beneficial effects of phytoestrogens increased in low body weight and low calcium intake women and those in later menopause
Spence et al. 2005 110	Randomized, crossover design, 1 month each treatment	15 postmenopausal women	Soy protein isolate with phytoestrogens, soy protein isolate devoid of phytoestrogens, and dairy protein diets	Urinary calcium excretion was significantly less on either of the soy diets. Fractional calcium absorption, total fecal calcium excretion, bone deposition and resorption, and calcium retention were unaffected by treatment	Soy phytoestrogens did not significantly affect calcium metabolism.
Arjmandi et al. 2005 109	Parallel, randomized, 1 year	87 postmenopausal women	Soy‐containing foods (60 mg phytoestrogens/day) or control foods with no soy	Both soy and control foods reduced whole body and lumbar bone mineral density and content but positively affected markers of bone formation	Results do not support a role of soy or its phytoestrogens in bone protection

Post‐menopausal complaints

Observational studies have indicated that post‐menopausal vasomotor symptoms (e.g. hot flushes and/or night sweats) in Japanese women are nearly 10‐fold lower than in US or other Western women. This reduction of symptoms has been related to the 100‐fold higher excretion of phytoestrogens in the urine of Japanese women compared to their Western counterparts 114, 115.

HRT in post‐menopausal women is generally associated with a reduction in the number and severity of vasomotor symptoms and a potentially beneficial effect against cognitive decline was described by some authors 116 but not recently confirmed 117. Only a relatively small percentage of post‐menopausal women take HRT (12% to 21% of US women in 2000) and this percentage has recently declined after the negative results of randomized studies evaluating cardiovascular effects of HRT 118.

Clinical studies on the effects of phytoestrogens on vasomotor symptoms of menopause have provided mixed results. Modest reductions in the frequency and severity of hot flushes was reported in a number of studies. In an open study of soy flour versus wheat flour, both diets reduced hot flushes (by 40% and 25% respectively) and menopausal symptom scores 119. A few studies using higher doses of phytoestrogens (50–80 mg/day) in women with a high incidence of vasomotor symptoms at baseline (4–7 symptoms/day) have shown mildly beneficial effects on self‐reported frequency and severity of vasomotor symptoms 120, 121. In the largest of these studies, a double blind, parallel, multicenter, placebo controlled trial, 51 post‐menopausal women took 60 g of soy protein isolate (with 76 mg/d of phytoestrogens) and 53 patients took 60 g of placebo (casein) for 12 weeks. Patients taking soy had a 45% reduction in daily hot flushes versus 30% obtained with the placebo (P<0.01) 122. However, in a cross‐over double blind study involving 177 pre‐ and post‐menopausal women with hot flushes and a history of mammary carcinoma, soy‐derived capsules rich in phytoestrogens at a dose of 150 mg phytoestrogens/day for 4 weeks did not result in any significant symptom change compared to the placebo 123. Similar inconclusive findings were reported by other groups on women with breast cancer 124, 125. Decreases in the number and severity of hot flushes in smaller studies on post‐menopausal women treated with soy containing phytoestrogens have been described but, overall, results have been inconsistent. Some studies have reported a significant reduction of hot flushes 126–130, whereas others reported no significant effects, secondary to an apparent placebo activity 131, 132.

A summary of randomized clinical studies investigating the effect of phytoestrogens on menopausal vasomotor symptoms is given in Table III. There appears some evidence of a possible role of phytoestrogens in the management of menopausal hot flushes, but this is clouded by a large placebo effect and specific effects due to soy phytoestrogens appear modest in degree 133. A consensus opinion of the Northern American Menopause Society only recommended that menopausal women consume whole foods that contain phytoestrogens (especially for the cardiovascular benefits). It also suggested a level of caution to be observed in making these recommendations 134.

Table III. Summary of randomized clinical studies investigating the effect of phytoestrogens on menopausal vasomotor symptoms.

Source	Design	Subjects	Intervention	Results	Conclusion
Murkies et al. 1995 119	Parallel, randomized, double blind, 12 weeks	58 postmenopausal women with at least 14 hot flushes per week	Soy flour (phytoestrogen level not reported) versus wheat flour	Hot flushes reduced by 45% in soy and 25% in wheat flour groups. Increased urinary excretion of phytoestrogens in soy group	Both soy (presumably containing phytoestrogens) and wheat flour significantly reduced hot flushes
Albertazzi et al. 1998 122	Parallel double‐blind, multicenter, randomized placebo‐controlled, 12 weeks	104 postmenopausal women	40 g of isolated soy protein (76 mg phytoestrogens) or 60 g of casein placebo daily	Soy significantly reduced hot flushes by 45% versus 30% with the placebo	Soy protein isolate added daily to the diet substantially reduced the frequency of hot flushes
Washburn et al. 1999 126	Randomized, double‐blind crossover, 6 weeks each treatment	51 perimenopausal women	20 g/day of soy protein (34 mg/day of phytoestrogens) given in a single dose or split into two doses versus 20 g/day carbohydrate placebo	Significant improvement in severity of vasomotor symptoms in the twice‐daily soy group compared with carbohydrate placebo group.	Soy with phytoestrogens gave significant improvements in perceived severity of vasomotor symptoms.
Baber et al. 1999 131	Randomized cross‐over, 12 week and 14 week treatments	51 postmenopausal women	Phytoestrogen tablets (40mg/day) and placebo	No significant difference between phytoestrogen and placebo groups in the reduction in hot flushes	Phytoestrogens did not provide a therapeutic benefit to hot flushes
Knight et al. 1999 132	Parallel, randomized, double‐blind, placebo‐controlled, 12 weeks	37 postmenopausal women with symptoms of estrogen deficiency	Phytoestrogen tablets (40 or 160 mg/day) or placebo	No significant difference in hot flushes between groups	Large placebo response may have obscured effect of phytoestrogens
Quella et al. 2000 123	Randomized, double‐blind crossover, 4 weeks each treatment	177 breast cancer survivors with substantial hot flushes	Soy‐derived phytoestrogen tablets (150 mg/day) or placebo	No difference between soy and placebo periods for hot flushes	Phytoestrogens did not reduce hot flushes in the breast cancer survivors
Van Patten et al. 2001 124	Parallel, randomised, placebo‐controlled, double‐blind, 12 weeks	123 postmenopausal women with moderate hot flushes previously treated for early‐stage breast cancer	Soy beverage (90 mg phytoestrogens/day) or placebo rice beverage	No significant differences in the number or severity of hot flushes. Strong placebo effect	In women with breast cancer, soy beverage with phytoestrogens did not reduce hot flushes any more than the placebo
Faure et al. 2002 127	Parallel, multicenter, randomized, double‐blind, placebo‐controlled, 16 weeks	75 women in natural or surgical menopause with at least seven hot flushes per day	Soy phytoestrogen extract (70 mg/day) or placebo	Patients taking soy extract 61% reduction in hot flushes versus 21% reduction with placebo	Phytoestrogens may help to reduce the frequency of hot flushes
Han et al. 2002 128	Parallel, double‐blind, placebo‐controlled, 16 weeks	80 postmenopausal women	Soy phytoestrogens (100 mg/day) and placebo	Significantly lower menopausal symptoms in phytoestrogen than placebo group	Phytoestrogens may be an effective treatment for menopausal symptoms
Colacurci et al. 2004 129	Parallel, randomised, 12 weeks	75 postmenopausal women	Phytoestrogens (50 or 75 mg/day orally or 6 or 12 mg/day transdermally), and a control group of no treatment.	Phytoestrogens gave significant reduction in hot flushes (except 6 mg/day transdermally)	Phytoestrogens given orally and transdermally reduce menopausal neurovegetative symptoms slightly or moderately
Petri Nahas et al. 2004 130	Parallel, double‐blind placebo‐controlled, 6 months	50 postmenopausal women	Soy germ phytoestrogen capsules (60 mg/day) and placebo	The reduction in hot flushes was significantly greater for the phytoestrogen than the placebo groups	Beneficial effects on vasomotor symptoms are provided by soy germ phytoestrogens
McGregor et al. 2005 125	Parallel, randomized, 12 weeks	72 postmenopausal women diagnosed with breast cancer having menopausal symptoms	Soy phytoestrogen capsules (70 mg/day) or placebo	No statistical difference in menopausal symptom scores between two treatments	No benefit of phytoestrogen supplementation compared to placebo

Potential toxic effects of phytoestrogens

Any consideration of the potential protective effects of phytoestrogens needs to be weighed against their potential toxicity 113. It has been reported that the major soy phytoestrogen genistein, at concentrations about 10‐fold higher than that found in humans after an average daily intake of soy products 135, leads to a clear mutagenic effect in human cells 136. The relevance of this finding becomes clear when it is considered that regulations on drug safety state that a drug is rated as a ‘mutagen’ when mutagenic effects occur at levels below 10‐fold of those found in plasma.

In addition to the potential of genistein to exert a clastogenic activity in human chromosomes, this compound has also been associated with an inhibition of purified topoisomerase II activity, which may result in DNA strand breakage and arrested cell growth in human leukemia and gastric cancer cell lines 137. This kind of activity is potentially also exerted at the thymic level thus reducing cell‐dependent immunity 138. There have been somewhat controversial data reported on the possibility that consumption of soy in infancy may lead to an increased risk of allergy development, possibly because of a prevalence of the Th2 activity in the periphery, resulting in immunological alterations in adult life, requiring a more frequent intake of antihistamine medications 139.

Two recent publications have suggested some other potentially important side effects of soy phytoestrogens on infant development. In one case, a meta‐analysis of soy intake suggested a thyrotoxic activity, possibly resulting in thyroid alterations in young individuals 140. While these findings were not clearly supported by evidence of an increased intake of thyromimetic drugs, they still called for caution. More recently, a study in piglets, comparing soy infant formulas with those from cow's milk, showed a 50% lower number of proliferating cells in the intestine after soy intake, possibly resulting in intestinal immaturity 141. Newborn piglets appear to be a good model for human infants and concentrations of genistein in the piglets' blood were similar to those of babies fed soy formulas that contain relatively high concentrations of genistein (32 to 46 mg/L of reconstituted formula) 142. In fact, when fed these soy formulas, babies are exposed to 6–11‐fold higher levels of phytoestrogens per kg body weight 135 than women receiving phytoestrogens for menopausal complaints.

The possibility of high levels of dietary phytoestrogen intake in infants has led health authorities in several countries to recommend maximum levels of daily intake 143. Countries such as UK, Australia, Canada, Ireland, New Zealand, and Switzerland, have recommended the preferential use of infant formulas based on cow's milk, when breast feeding is not possible, and that only women advised by their doctor should continue to use soy infant formulas. A very short time ago, French authorities also issued a detailed public statement cautioning against the indiscriminate use of phytoestrogens for any major therapeutic indication: they indicated a maximal daily intake of 1 mg/kg of body weight and suggested the use of infant formulas based on soybean with less than 1 mg/L of phytoestrogens 144.

Phytoestrogens are also available as dietary supplements that may contain more than 100 mg per tablet. Recent data, indicative of a small but significant endometrial hyperplasia in women receiving 150 mg daily of phytoestrogens for 5 years, questioned the safety of these treatments 145. Due to the potential toxicity of phytoestrogens at very high intakes and their increasing use by women for menopausal complaints, the Italian Health Authorities have advised the public to maintain daily intake of phytoestrogens as dietary supplements below 80 mg 146, which again represents a maximal daily intake of about 1 mg/kg of body weight.

In light of the Italian recommendation, a recent study has considered the phytoestrogen content of Italian soy food products and the daily intakes for some specific classes of consumers. Total phytoestrogen contents of soy foods were in the range of 21–803 µg/g dry weight (dw) and were particularly high in soy‐based imitation dairy and meat products 147. The phytoestrogen content of gluten‐free products was surprisingly high, being in the range of 77–220 µg/g dw, due to the inclusion of soy protein for technological purposes. Infant formulas contained 121–427 µg/g dw: these levels may possibly lead to daily intakes of about 2–3 mg/kg body weight. These intakes are far higher than the maximal daily intakes of 1 mg/kg established by the Italian and French Health Authorities 144, 146.

Conclusions

Based on available data, the major therapeutic activity of soy appears to reside exclusively in the proteins, whereas the role of phytoestrogens appears to be minimal. Yet even now it is somewhat perplexing to note the frequent confusion between phytoestrogens and proteins as the major hypocholesterolemic components of soy based diets, even in highly reputed journals 148.

Phytoestrogens have yet to be proven as beneficial to health and their potential negative effects may not be acceptable in the absence of any clear benefit. Substantial effort and financial support has been given to studies aiming at elucidating the beneficial activities and mechanisms of action of phytoestrogens. For example the European Commission has financed several projects that have cost around 12 million euros to European citizens (for information see the CORDIS database at http://ica.cordis.lu/search/). Unfortunately these studies have produced very little published material in support of the health benefits of phytoestrogens.

At present soy still appears to have great potential for treatment in a variety of conditions and the US FDA claim for CVD protection remains well supported. However, it is clear that the benefit of soy intake resides essentially in the protein itself, whereas phytoestrogens provide little if any additional benefit. From a practical point of view, promotion of the cultivation of soybean varieties with low phytoestrogen contents for manufacturing human food products is warranted. In addition the efficacy of using other legumes 149 characterized by much lower phytoestrogen contents, such as peas, beans, chickpeas or lupin 61, 150, 151, as sources of food ingredients and nutraceuticals for the prevention of cardiovascular disease is worthy of further investigation. The phytoestrogen tale may be essentially over.