Endocrine-disrupting activities of phyto-oestrogens in vertebrates and human subjects

An EDC is defined by the Endocrine Society as a compound that interferes with any aspect of hormone action⁽ Reference Diamanti-Kandarakis, Bourguignon and Giudice ^{20 )}. The word ‘interferes’ is important because many things can have hormone action as part of maintaining homeostasis and interacting with the environment, such as the simple act of eating or standing in sunlight. An EDC is a compound that interferes with the way in which the pancreas responds to a meal, or disrupts the vitamin-D-producing capacity of sunbathed skin. In the case of isoflavones, the target of this ‘interference’ is primarily thyroid hormone and oestrogen. Although a formal definition has not yet been established, the term neuroendocrine disruption has been used to broadly describe chemical impacts on endocrine-related brain development and function⁽ Reference Waye and Trudeau ^{21 ,} Reference Gore and Patisaul ^{22 )}. In the case of phyto-oestrogens, the vast majority of research effort has concentrated on the reproductive neuroendocrine system, which includes the hypothalamic–pituitary–gonadal (HPG) axis. Importantly, neuroendocrine disruption is distinct from, and should not be conflated with, neurotoxicity, which characterises processes contributing to neuronal cell death and related downstream consequences (e.g. dopaminergic cell death and Parkinson's like symptoms) and peripheral neuropathies. Isoflavones are not neurotoxic.

That phyto-oestrogens are endocrine disrupting has been known since at least the 1940s when ewes grazing on clover rich pastures in Australia were observed to have abnormally high rates of infertility, abortion and reproductive abnormalities in their offspring⁽ Reference Bennetts, Underwood and Shier ^{23 –} Reference Adams ^{25 )}. Consequently, management of phyto-oestrogen levels has been the subject of grazing/feeding practices within the agricultural community for decades, including, most recently, in aquaculture⁽ Reference Green and Kelly ^{26 ,} Reference Gontier-Latonnelle, Cravedi and Laurentie ^{27 )}. Phyto-oestrogens have proven to be potently endocrine disrupting for a wide range of vertebrates, including rodents⁽ Reference Patisaul and Jefferson ^{1 ,} Reference Whitten, Naftolin and Hochberg ^{28 )}, birds⁽ Reference Leopold, Erwin and Oh ^{29 )}, cheetahs⁽ Reference Setchell, Gosselin and Welsh ^{30 )}, multiple species of fish⁽ Reference Sassi-Messai, Gibert and Bernard ^{31 ,} Reference Clotfelter and Rodriguez ^{32 )}, and grazing mammals such as cattle, sheep and even the southern white rhinoceros⁽ Reference Bennetts, Underwood and Shier ^{23 ,} Reference Adams ^{25 ,} Reference Tubbs, Hartig and Cardon ^{33 )}.

Evidence of endocrine disruption by soya in human subjects also dates back decades. Soya has been known to be goitrogenic for nearly a century⁽ Reference McCarrison ^{34 ,} Reference Divi, Chang and Doerge ^{35 )} necessitating the addition of iodine to soya infant formula and other soya-rich foods. Both genistein and daidzein potently block thyroxine synthesis by serving as alternate substrates and blocking thyroid peroxidase catalysed tyrosine iodination. Soya also decreases absorption of synthetic thyroid hormone⁽ Reference Messina and Redmond ^{36 )} potentially necessitating higher doses in hypothyroid patients. Thus, for these and other patients at risk for clinical or subclinical hypothyroid, compensatory iodine intake is advisable if soya is part of the regular diet. Additionally, although research regarding the relationship between soya intake and thyroid levels during pregnancy is extremely limited⁽ Reference Li, Teng and Wang ^{37 )}, because thyroid hormone is essential for normal brain development, pregnant women regularly consuming soya should be particularly mindful of this endocrine-disrupting property of soya.

Soya can also impact reproductive function in women. Suppression of circulating steroid hormone levels and attenuation of the preovulatory gonadotropin surge have been repeatedly observed and a 2009 meta-analysis concluded that isoflavone intake moderately increases cycle length and suppresses luteinising hormone and follicle-stimulating hormone levels⁽ Reference Hooper, Ryder and Kurzer ^{38 )}. A 2008 clinical case report described three women (aged 35–56 years) experiencing a suite of symptoms related to excessive soya intake (estimated to exceed 40 g/d), including abnormal uterine bleeding, endometrial pathology and dysmenorrhea, all of which resolved when soya intake was discontinued or reduced⁽ Reference Chandrareddy, Muneyyirci-Delale and McFarlane ^{39 )}. Importantly, as for all EDC, timing of exposure is important when considering the potential for long-term effects. The youngest of the three patients had been on a soya-rich diet since age 14 years and was experiencing secondary infertility, a condition that resolved and resulted in a pregnancy once she cut back on her soya consumption. Of even greater concern is what might happen in infant girls who consume high levels of soya, while their reproductive systems are still developing. Exposure earlier in life may have more lasting effects because disruption of the organisational actions of hormones may produce permanent structural and/or functional changes⁽ Reference Gore, Chappell and Fenton ^{40 )}.

The earliest evidence for developmental reproductive health effects came from two studies, conducted in the mid-1980s, which associated neonatal phyto-oestrogen exposure with thelarche before age 2 years in a population of Puerto Rican girls. A number of confounding factors, however, including the consumption of meat that had been fattened with potent oestrogens, including the notorious endocrine disruptor diethylstilbestrol, make the data problematic and difficult to interpret⁽ Reference Freni-Titulaer, Cordero and Haddock ^{41 ,} Reference Schoental ^{42 )}. A highly cited retrospective cohort study of 952 women found that young women reared on soya-based infant formula (248 women) as part of a controlled, University of Iowa feeding study, reported longer menstrual bleeding and menstrual discomfort than those who were fed a non-soya based formula (563 women)⁽ Reference Strom, Schinnar and Ziegler ^{43 )}. At the time the study was conducted, the women were too young to comprehensively examine pregnancy or fertility outcomes, but, now that nearly a decade has past, this area is ripe for reevaluation. Soya formula consumption has also been linked to a greater risk of developing uterine fibroids⁽ Reference D'Aloisio, Baird and DeRoo ^{44 )}. A prospective study reported oestrogenised vaginal epithelium in female soya formula-fed infants, an important finding confirming soya infant formula is oestrogenic in human subjects⁽ Reference Bernbaum, Umbach and Ragan ^{45 )}. Other studies, however, have found no link between soya infant formula and developmental reproductive parameters, including breast, ovarian or testes volume⁽ Reference Bernbaum, Umbach and Ragan ^{45 ,} Reference Andres, Moore and Linam ^{46 )}, and impacts of soya formula intake and on age at menarche are mixed⁽ Reference D'Aloisio, DeRoo and Baird ^{47 ,} Reference Adgent, Daniels and Rogan ^{48 )}. That soya is hormonally active is irrefutable. Whether or not soya intake, particularly during infancy, can have long-term health effects remains the subject of debate, but parents should be made aware of possible oestrogenic effects if they choose to feed their infants a soya-based formula.

Mechanisms of endocrine disruption by isoflavones

EDC can act via a myriad of mechanisms but the most fundamental include: (1) mimicking the effects of natural hormones by acting as a ligand at their binding sites; (2) antagonising the effect of these hormones by blocking their interaction with their physiological binding sites; (3) reacting directly and indirectly with the hormone in question; (4) altering the natural pattern of synthesis/degredation of hormones; or (5) altering cellular hormone receptor levels⁽ Reference Gore, Chappell and Fenton ^{40 ,} Reference Frye, Bo and Calamandrei ^{49 ,} Reference Patisaul and Adewale ^{50 )}. Isoflavones have been shown to interfere with oestrogen action via all of these. They also have other biological activities, which is not atypical as one of the hallmarks of EDC is that they simultaneously affect multiple hormonal systems, and act by multiple mechanisms. Genistein is thought to slow tumourigenesis, for example, via inhibition of protein tyrosine kinases and inhibition of DNA topoisomerases I and II, along with other chemoprotective mechanisms⁽ Reference Patisaul and Jefferson ^{1 ,} Reference Messina ^{6 ,} Reference Mahmoud, Yang and Bosland ^{51 )}. Phyto-oestrogens are also good antioxidants and anti-inflammatory agents.

The primary mode of isoflavone endocrine disruption is interference with oestrogen. At almost the same instant that a second subform of the nuclear oestrogen receptor (ER) was discovered (termed ERβ) it was recognised that isoflavones bind and activate transcription via both forms (ERα and ERβ), but generally have a higher relative binding affinity for ERβ⁽ Reference Kuiper, Lemmen and Carlsson ^{52 –} Reference Kuiper, Carlsson and Grandien ^{55 )}. Potency estimates vary by assay, but most isoflavones bind nuclear ER far more readily than their synthetic endocrine-disrupting counterparts including bisphenol a⁽ Reference Kuiper, Lemmen and Carlsson ^{52 )}. Exposure is also consistently higher, often orders of magnitude higher, making them one of the most significant EDC in the human landscape⁽ Reference Patisaul and Jefferson ^{1 ,} Reference Rappaport and Smith ^{56 )}. Once bound, isoflavones act as partial agonists, with activity varying across tissue types and local levels of endogenous oestrogen. ER subtype distribution varies across tissues and cell types, particularly in the brain, changes over the lifespan, and is sexually dimorphic⁽ Reference Cao and Patisaul ^{57 –} Reference Schwarz, Nugent and McCarthy ^{59 )}. Because ERα and ERβ are differentially distributed throughout the body and the brain, including neuroendocrine pathways, which coordinate reproductive function, that isoflavones are more bioactive via ERβ is functionally significant⁽ Reference Shughrue, Lane and Scrimo ^{60 –} Reference Handa, Ogawa and Wang ^{64 )}. ERα and ERβ regulate different aspects of reproduction, behaviour and neuroendocrine function across the lifespan, although their relative roles are more clearly elucidated in animal models than in human subjects, in some tissues than others and, in some cases, one sex than another⁽ Reference Rissman ^{65 –} Reference Jia, Dahlman-Wright and Gustafsson ^{68 )}. For example, ERβ in the paraventricular nucleus of the hypothalamus (PVN), a region important for the coordination of reproductive, social and stress-related behaviours, suppresses anxiety-related behaviours and enhances production of the neuropeptide oxytocin (OT)⁽ Reference Lund, Rovis and Chung ^{69 –} Reference Acevedo-Rodriguez, Mani and Handa ^{71 )}. ERβ is also expressed at higher levels than ERα in the basal forebrain, hippocampus, dorsal raphe and cerebral cortex in the adult⁽ Reference Shughrue, Lane and Scrimo ^{60 ,} Reference Zhang, Cai and de Zhou ^{72 ,} Reference Fan, Xu and Warner ^{73 )}, all brain regions critical to neuroendocrine function and mood-related behaviours. ERβ is particularly abundant in the prenatal brain and plays a key role in brain morphogenesis by affecting cortical layering and interneuron migration⁽ Reference Fan, Xu and Warner ^{73 )}.

Once bound to ER, phyto-oestrogens can initiate transcription classically through interactions with the oestrogen response element or by binding early immediate genes, such as Jun and Fos⁽ Reference Kushner, Agard and Greene ^{74 )}. Steroid hormones, particularly oestrogens, can also initiate rapid, non-genomic actions at the cell surface via a range of mechanisms, including the binding of specialised steroid membrane receptors or ion channel subunits⁽ Reference Levin ^{75 –} Reference Kow and Pfaff ^{78 )}. The vast majority of rapid actions are thought to originate at oestrogen-binding sites at the extracellular surface of the cell membrane, meaning that a potential EDC does not have to enter the cell to be active. Binding then activates second messenger pathways leading to cellular responses such as increased intracellular calcium or cAMP levels, or promoting nitric oxide release resulting in the stimulation of signal transduction pathways important for neuronal signalling, differentiation and other cellular processes⁽ Reference Belcher and Zsarnovszky ^{79 )}. The best-known transmembrane ER, the G-protein-coupled oestrogen receptor, was cloned as the orphan receptor GPR30 two decades ago and is now known to be capable of binding a wide range of EDC, including genistein⁽ Reference Thomas and Dong ^{80 )}. The functional significance of this pathway, or its disruption, has yet to be fully described but G-protein-coupled oestrogen receptor plays an important role in rapid vascular oestrogen signalling along with ERα and ERβ⁽ Reference Barton ^{81 )}. Emerging data reveals that phyto-oestrogens have epigenetic activity and can alter activities of DNA and histone methyltransferases, NAD-dependent histone deacetylases and other modifiers of chromatin structure⁽ Reference Shanle and Xu ^{82 –} Reference Labinskyy, Csiszar and Veress ^{84 )}.

Phyto-oestrogens have also been shown to interfere with the enzymes needed for steroid biosynthesis and/or degradation. Coumestrol, for example, attenuates the conversion of [3H]-estrone to [3H]-estradiol in vitro by inhibiting the enzyme 17β-hydroxysteroid oxidoreductase Type 1 in a dose-dependent fashion⁽ Reference Franke and Custer ^{85 )}. Genistein, though weaker, has a similar dose-dependent inhibitory effect. In rats, genistein can alter folliculogenesis, an outcome postulated to result, at least in part, from dysregulation of steroidogenic enzymes⁽ Reference Patel, Zhou and Rattan ^{86 )}. In porcine granulosa cells, genistein decreases the activity of cholesterol side-chain cleavage enzyme (P450scc) and 3β-hydroxysteroid dehydrogenase⁽ Reference Piasecka-Srader, Kolomycka and Nynca ^{87 )}. Genistein has also been characterised as a non-competitive inhibitor of 11β-hydroxysteroid dehydrogenase type 1, which produces bioactive glucocorticoids, such as cortisol, from inactive precursors⁽ Reference Tagawa, Kubota and Kobayashi ^{88 )}. Disruption of aromatase and 5α-reductase by a number of phyto-oestrogens has also been demonstrated in vitro but this potential activity in mammalian tissues remains controversial⁽ Reference Jefferson, Patisaul and Williams ^{2 )}. Disruption of biosynthetic/degradative enzymes could significantly alter local endogenous hormone levels but not manifest as a change in circulating hormone levels. This may be particularly important for brain and hormone-sensitive subregions such as the hypothalamus as growing evidence strongly suggests that neural cells have the capacity to synthesise steroid hormones de novo ⁽ Reference Taves, Plumb and Sandkam ^{89 –} Reference Fokidis, Adomat and Kharmate ^{91 )}.

Another mechanism by which phyto-oestrogens can perturb steroid bioavailability and transport is by altering sex hormone-binding globulin synthesis and availability. Isoflavones have long been known to appreciably stimulate sex hormone-binding globulin production, particularly in individuals who have levels on the low range of normal⁽ Reference Pino, Valladares and Palma ^{92 )}. Heightened sex hormone-binding globulin levels are thought to be one mechanism by which soya may lower breast cancer risk because bioavailable levels of circulating oestrogens are concomitantly reduced⁽ Reference Low, Dunning and Dowsett ^{93 )}. Similarly, suppression of circulating androgens, particularly dihydrotestosterone by equol, is hypothesised to be one way in which soya might protect against prostate cancer⁽ Reference Mahmoud, Yang and Bosland ^{51 ,} Reference Tanaka, Fujimoto and Chihara ^{94 )}. Notably, a subset of studies have found no impact of isoflavones on circulating sex hormone-binding globulin or steroid hormone levels in human subjects (e.g.⁽ Reference Maskarinec, Williams and Inouye ^{95 )}). One found suppressed luteal oestrogen levels following increased soya intake, but only in women of Asian descent⁽ Reference Wu, Stanczyk and Hendrich ^{96 )}, indicating ethnicity and/or the capacity to produce equol could be an underappreciated factor-mediating interindividual variability in responsiveness⁽ Reference Atkinson, Newton and Stanczyk ^{97 ,} Reference Jackson, Greiwe and Schwen ^{98 )}.

Endocrine-disrupting effects of soya isoflavones on the adult neuroendocrine system

Impacts on the mature reproductive axis in human subjects and other vertebrates have already been summarised and include altered serum hormone levels and suppression of ovulation. Elevated urine levels of genistein and daidzein have been associated with idopathic infertility and lower semen quality in Chinese men⁽ Reference Xia, Chen and Zhu ^{99 )}, and a slightly lower percentage of normal sperm in US men whose partners were attempting pregnancy⁽ Reference Mumford, Kim and Chen ^{100 )}, but supporting evidence in other populations or species for effects on spermatogenesis is limited. In contrast, the animal literature has explored a wider age range and a more diverse array of endocrine-disrupting effects.

Work in our laboratory focuses on neuroendocrine pathways underlying sexually dimorphic behaviours and, using a variety of animal models, we and others have shown that isoflavone intake interferes with oestrogen-mediated behaviours, including female sexual motivation. For example, consumption of a commercially prepared isoflavone supplement to adult female rats, at a dose that results in serum levels between those seen in Western and Asian (human) adults, attenuated lordosis (a reflexive posture indicating sexual receptivity) to the same degree as tamoxifen⁽ Reference Patisaul, Luskin and Wilson ^{101 ,} Reference Patisaul, Dindo and Whitten ^{102 )}. The supplement also suppressed proceptive behaviours even more profoundly than tamoxifen suggesting that soya isoflavones can suppress female sexual motivation and solicitation (Fig. 2). Administration of genistein alone did not recapitulate these effects⁽ Reference Patisaul, Melby and Whitten ^{103 )}. Whether or not libido is altered in human subjects appears to be completely unknown. Remarkably, a Pubmed search with the keywords ‘soya’ and ‘libido’ produced only nine published papers, not all of which were relevant. One was a case report describing a case of bilateral gynecomastia, erectile dysfunction and loss of libido in a 60-year-old man, which resolved when he discontinued drinking three quarts of soya milk daily⁽ Reference Martinez and Lewi ^{104 )}. Another reported a beneficial effect of soya protein dietary supplements on libido in post-menopausal women but there was an equally beneficial placebo effect suggesting that the soya effect was spurious⁽ Reference Kotsopoulos, Dalais and Liang ^{105 )}. Further inquiry revealed no studies, which have tackled this question in younger populations, or with a large-enough sample size to achieve reasonably robust statistical power. Given that soya appears to have a consistently suppressive effect on circulating steroid hormone levels it is not implausible that libido may also be suppressed but this appears to be, by and large, an unanswered question.

Fig. 2.

(Colour online) In ovariectomised, hormone replaced female rats, sexual behaviour is suppressed by a soya isoflavone supplement. (a) Lordosis is a hallmark receptive posture in the rat and the frequency of lordosis in response to male mounting, which can be induced in ovariectomised females with progesterone (P) and estradiol benzoate (E), but not P alone. (b) In the presence of E and P tamoxifen (E + T+) or a soya supplement (E + S+) significantly decrease lordosis in female rats. (c) Similarly, proceptive behaviour, including hopping and darting, is also suppressed in hormonally replaced female rats on tamoxifen (E + T+) and, to an even greater degree, the soya isoflavone supplement (E + S+). *P ≤ 0·05; **P ≤ 0·01; means ± sem. Figure adapted from Patisaul et al.⁽ Reference Patisaul, Luskin and Wilson ^{101 )}

Mood and anxiety-related behaviours may also be impacted by adult soya intake⁽ Reference Lephart, Setchell and Handa ^{106 )}. In human subjects, nearly all studies along these lines have focused on post-menopausal women and evidence for improvement of mood is minimal and sporadic⁽ Reference Thomas, Ismail and Taylor-Swanson ^{107 )}. Results across animal studies are mixed and sex dependent with females generally showing decreased anxiety and males showing heightened⁽ Reference Patisaul, Blum and Luskin ^{108 ,} Reference Patisaul ^{109 )}. This pattern tends to abrogate or reverse expected sex differences in assessments of anxiety-related behaviours⁽ Reference Patisaul, Sullivan and Radford ^{110 )} and appears to involve the neuropeptides OT and vasopressin (AVP). For example, male cynomolgus monkeys fed soya protein isolate containing 1·88 mg isoflavones/g protein over 18 months demonstrated higher frequencies of intense aggressive (67 % higher) and submissive (203 % higher) behaviours as well as a decreased proportion of time (68 % reduction) spent in physical contact with other monkeys⁽ Reference Simon, Kaplan and Hu ^{111 )}. Male rats maintained on a diet containing 150 µg/g genistein and daidzein displayed increased anxiety and elevated stress-induced plasma AVP and corticosterone levels⁽ Reference Hartley, Edwards and Spiller ^{112 )}. Increased hypothalamic AVP has also been reported in Sprague–Dawley rats fed a diet containing 1250 ppm genistein across the lifespan in a study run at a US Food and Drug Association research center⁽ Reference Scallet, Wofford and Meredith ^{113 )}. We found that the same isoflavone supplement found to suppress reproductive behavior and motivation in female rats (described earlier) abrogated the oestrogen-dependent up-regulation of OT receptors in the ventromedial nucleus of the hypothalamus and heightened ERβ expression levels in the PVN, an effect opposite to that of estradiol⁽ Reference Patisaul, Dindo and Whitten ^{102 )}.

Involvement of the PVN is consistent with both the oestrogen and thyroid-disrupting properties of soya, and the high potency of isoflavones at ERβ. The PVN, which contains little to no ERα but high levels of ERβ⁽ Reference Hrabovszky, Kallo and Hajszan ^{114 )}, is a primary site of OT and AVP production, peptide hormones important for social behaviour and the facilitation of sexual behaviour⁽ Reference Ross and Young ^{115 )}, as well as thyroid hormone releasing hormone and corticotropin-releasing hormone. Oestrogen-dependent stimulation of PVN OT and AVP production requires ERβ⁽ Reference Patisaul, Scordalakes and Young ^{116 ,} Reference Nomura, McKenna and Korach ^{117 )}. OT then binds to its receptor throughout the brain, including the ventromedial nucleus, a nucleus critical for mediating the lordosis response in females⁽ Reference Pfaff ^{118 )}. Isoflavone-related effects on these and other oestrogen dependent systems in the adult rodent brain have previously been reviewed⁽ Reference Patisaul ^{109 ,} Reference Lephart, West and Weber ^{119 –} Reference Patisaul and Polston ^{121 )} but a concerted focus on OT/AVP systems in human subjects remains lacking.

Evidence for developmental neuroendocrine disruption in animals and human subjects

Neuroendocrine disruption by soya isoflavones in mature neuroendocrine systems is by and large reversible with dietary modification and thus, with the exception of some hypersensitive groups such as hypothyroid and oncology patients, soya likely poses no long term health risk and may even confer modest benefits. Of greater concern is that phyto-oestrogens may interfere with the organisational role of oestrogen in the developing brain and reproductive system. Data from a diversity of animal models have repeatedly shown that manipulation of oestrogen during specific critical windows of development throughout gestation and early infancy leads to a myriad of adverse outcomes in the HPG axis including malformations in the ovary, uterus, mammary gland and prostate, early puberty, reduced fertility, disrupted brain organisation, and reproductive tract cancers⁽ Reference Simerly ^{66 ,} Reference Newbold ^{122 –} Reference Crain, Janssen and Edwards ^{126 )}. The disruptor diethylstilbestrol story also starkly illustrates the broad spectrum of sex-specific consequences on neuroendocrine systems following fetal oestrogen exposure⁽ Reference Reed and Fenton ^{127 ,} Reference Newbold ^{128 )}. Although isoflavones and other EDC are far less potent than disruptor diethylstilbestrol, human exposure is ubiquitous and there is growing acceptance that EDC are contributing to adverse reproductive health trends in Western nations including median age at menarche, first breast development, and sexual precocity⁽ Reference Gore, Chappell and Fenton ^{40 ,} Reference Aksglaede, Sorensen and Petersen ^{129 –} Reference Mouritsen, Aksglaede and Sorensen ^{131 )}. Advanced pubertal onset in girls adopted from developing countries by Western parents supports a role for environmental factors⁽ Reference Aksglaede, Sorensen and Petersen ^{129 )}. Emerging but controversial data suggest that EDC may also be shifting age at puberal onset in boys⁽ Reference Zawatski and Lee ^{132 )}. Among men, sperm counts in the USA and Europe appear to have declined by approximately half over the past 50 years⁽ Reference Swan, Elkin and Fenster ^{133 ,} Reference Hauser ^{134 )} with upwards of 30 % in the subfertile range in places like Denmark where exposure to persistent environmental pollutants is particularly high⁽ Reference Joensen, Jorgensen and Rajpert-De Meyts ^{135 )}. A provocative but limited study associated increased incidence of hypospadias (malformation of the male external genitalia) with maternal vegetarianism⁽ Reference North and Golding ^{136 )} but this effect has not been replicated. Synthetic EDC which interfere with androgen biosynthesis or activity are also associated with disorders of male genital development⁽ Reference Skakkebaek, Rajpert-De Meyts and Main ^{137 ,} Reference Fisher ^{138 )} thus it is not implausible that equol may be endocrine disrupting in this regard. Increased prevalence of reproductive health disorders is likely not attributable to a single factor, not even a single environmental factor, but EDC are causal to some degree and isoflavones are hypothesised to play a contributing role⁽ Reference Cederroth, Zimmermann and Nef ^{5 ,} Reference Gore, Chappell and Fenton ^{40 ,} Reference Massart, Parrino and Seppia ^{139 –} Reference West, Anderson and McClure ^{141 )}.

How much is too much: human isoflavone intake, metabolism and excretion

Ultimately risk of harm comes down to two primary factors: dose and timing of exposure. Undoubtedly, development is the most sensitive period for the endocrine-disrupting consequences of soya isoflavone exposure, thus it is not surprising that concerns have been expressed regarding the safety of soya-based infant formula. Initially developed as an alternative to bovine milk formulas for babies with milk allergy, use of soya infant formula in the USA is a popular choice and constitutes an estimated 25 % of the formula market⁽ Reference Badger, Ronis and Hakkak ^{185 –} Reference Barrett ^{187 )}. The safety of soya formula has been rigorously discussed from several perspectives, and a litany of review articles and position papers have been published on the subject⁽ Reference Bhatia and Greer ^{188 –} Reference Chen and Rogan ^{195 )}. Societies including the American Academy of Pediatrics and the European Society for Pediatric Gastroenterology Hepatology and Nutrition Committee on Nutrition have issued guidelines recommending against the exclusive use of soya formula except in the rare cases of true milk allergy or lactose intolerance. The US National Toxicology Programme completed its most recent safety assessment of soya infant formula in 2010 (monograph available at http://ntp.niehs.nih.gov/pubhealth/hat/noms/formula/index.html) and concluded there is ‘minimal concern for adverse developmental effects.’ For comparison, this is the same level of concern initially expressed for bisphenol a until the Food and Drug Association elevated that advisory to ‘some concern’ in January, 2010 based on new data (and then subsequently lowered it again). Notably, the National Toxicology Programme could not issue a conclusive recommendation regarding potential long-term reproductive effects of soya infant formula largely because of limited and poor-quality human data. An apparent lack of adverse effects is one reason why so many consumers, clinicians and public health agencies consider regular use of soya formula to be safe, even beneficial. However, the absence of evidence is not evidence of safety so this problematic data gap regarding the long-term impacts of soya formula use remains in serious need of attention.

When considering the potential safety of soya formula, it is frequently argued that Asian populations have been consuming soya for centuries, with no obvious consequences. This argument fails to recognise, however, that exposure patterns differ in key ways between Asians consuming a traditional soya-rich diet and Caucasians eating a typical Western diet⁽ Reference Badger, Ronis and Hakkak ^{185 )}. This timing of exposure is critical. In a traditional Asian diet, soya consumption is moderate across the entire lifespan, but because isoflavones do not effectively transfer via lactation, exposure in breastfeeding infants is extremely low. By contrast, Western babies on soya infant formula have their highest exposures in the first year of life then exposure rapidly plummets. In that regard the two populations are not comparable because their exposure patterns during a critical window of development are so dramatically different. Other diet and lifestyle differences may also be confounding when evaluating the potential health benefits and risks of soya. For example, Asian populations on traditional diets eat less processed foods, considerably higher levels of seafood and lower levels of animal fat than Western populations.

So how much is too much? There is no ‘typical’ level of isoflavone intake as consumption patterns vary widely across populations, and geographic regions. For Asians, vegetarians and other groups in which soya is foundational to the diet, isoflavone consumption can be as high as 100 mg/d (intake range is about 0·3–1·5 mg/kg body weight)⁽ Reference Messina ^{6 ,} Reference Setchell, Zimmer-Nechemias and Cai ^{192 ,} Reference Franke, Custer and Tanaka ^{196 –} Reference Mortensen, Kulling and Schwartz ^{198 )}. Western diet intake estimates range from 1 to 3 mg/d⁽ Reference Mortensen, Kulling and Schwartz ^{198 –} Reference Horn-Ross, Barnes and Lee ^{201 )}. For their weight, infants exclusively fed soya-based formula have the highest mean daily consumption of total isoflavones, ranging from 6 to 9 mg/kg body weight per d in 4-month-old infants, an amount that is up to seven times higher than Asians consuming a traditional soya-based diet.

The isoflavone content of an array of foods and food products is now available via online databases (reviewed in:⁽ Reference Schwarz, Sontag and Plumb ^{202 )}) including one maintained by the USDA^{( 203 )}. Food isoflavone content varies widely, even in the same foods, because of local and/or seasonal differences in growing conditions so its difficulty to accurately estimate intake⁽ Reference Mortensen, Kulling and Schwartz ^{198 )}. Additionally, soya is found in upwards of 60 % of processed foods and ground meats^{( 204 )}. Textured soya protein (50–70 % soya protein) is used as a meat substitute or filler for hotdogs, hamburgers, sausages and other meat products⁽ Reference Thomas and Lutz ^{205 ,} Reference Senti ^{206 )}, while soya protein isolate (90 % soya protein) is frequently used to enrich energy bars and sports drinks (particularly those advertising high protein levels), cereals, granola bars, infant formula, imitation dairy products, ice cream and cheese. Soya isoflavones and other phyto-oestrogens are also widely available as dietary supplements⁽ Reference Setchell, Brown and Desai ^{207 ,} Reference Piotrowska, Jakobkiewicz-Banecka and Wegrzyn ^{208 )}, typically containing concentrations far higher than those found in food⁽ Reference Cao, Calafat and Doerge ^{209 )}.

Not surprisingly, blood isoflavone levels also vary widely, and can be orders of magnitude different between individuals based on dietary preferences and individual differences in phyto-oestrogen absorption and metabolism⁽ Reference Hutchins, Slavin and Lampe ^{210 –} Reference Rolwand, Wiseman and Sanders ^{212 )}. Blood genistein levels are generally in the range of 25 ng/ml for Asian women, slightly less for vegetarian women, and under 2 µg/ml for US women⁽ Reference Verkasalo, Appleby and Allen ^{213 )}. Isoflavones can pass from mother to fetus through the placenta, and have been found in human umbilical cord blood and amniotic fluid at levels comparable with concentrations seen in maternal plasma, demonstrating that fetal exposure can be significant⁽ Reference Adlercreutz, Yamada and Wahala ^{214 )}. Infants on soya formula can have plasma levels exceeding 1000 ng/ml⁽ Reference Cao, Calafat and Doerge ^{209 )} which is 13 000–22 000 times higher than their own endogenous oestrogen levels, 50–100 times higher than oestradiol levels in pregnant women, and 3000 times higher than oestradiol levels at ovulation⁽ Reference Badger, Ronis and Hakkak ^{185 ,} Reference Winter, Hughes and Reyes ^{215 ,} Reference Setchell, Zimmer-Nechemias and Cai ^{216 )}. In contrast, infants fed cow's milk formula or human breast milk have plasma isoflavone levels of 9·4 and 4·7 ng/ml, respectively⁽ Reference Setchell, Zimmer-Nechemias and Cai ^{192 ,} Reference Franke, Custer and Tanaka ^{196 ,} Reference Setchell, Zimmer-Nechemias and Cai ^{216 )}. Notably, levels in infants and vegetarians easily far surpass, sometimes by several orders of magnitude, internal levels other endocrine disruptors of concern, including bisphenol a and phthalates⁽ Reference Crain, Janssen and Edwards ^{126 )}.

Conclusions and recommendations

Soya isoflavones are clearly endocrine disrupting, but although they are similar to their synthetic brethren in terms of their cellular and molecular mechanisms of action on neuroendocrine structure and function, and the scope of adverse outcomes they can inflict, society embraces these compounds at the same time it rejects, often with vigour, exposure to their synthetic brethren. Thus, phyto-oestrogens both challenge our attitudes regarding EDC and highlight how profoundly the direction and interpretation of research and available data can be influenced by source. While some beneficial effects might be conferred by including moderate levels of dietary soya, particularly in adults eating a diet high in saturated fat and animal protein, the potentially adverse effects of these compounds for some groups are likely underappreciated. An abundance of animal data unequivocally demonstrates that soya isoflavone exposure, at doses and plasma concentrations attainable in human subjects, including soya-reared infants, can permanently alter the structure and function of neuroendocrine pathways in both sexes. Infants fed soya formula have the highest exposure to any non-pharmacological source of oestrogen-like compounds, and yet greater anxiety surrounds compounds like bisphenol a and the phthalates which have far lower potency on neuroendocrine targets and to which exposure is far lower. Although relatively few adverse effects have been reported, that is somewhat a consequence of lack of data rather than lack of measurable effects. Although unsatisfying, a parsimonious approach to soya intake is to follow the classic adage and consume in moderation. Development of dietary guidelines should consider the endocrine-disrupting properties of soya and other hormonally active foods, particularly for vulnerable groups such as pregnant women and hypothyroid individuals.