Increasing awareness of the potential hormonal effects of endocrine disruptors motivated us to evaluate possible estrogenic effects in female infants consuming soy-based formulas, a rich source of phytoestrogens. The major phytoestrogens in soy formulas are the isoflavones genistein and daidzein, 2 heterocyclic phenols with structural homology to estradiol. These compounds bind with different affinities to estrogen receptors α and β (1,2), and have both estrogen agonist and antagonist activity (3,4). Several studies in animal models and human adults have demonstrated estrogenic effects of isoflavones on estrogen-responsive organs in both the reproductive system and mammary glands.
Pre- and postnatal exposure of mice and rats to isoflavones induced an increase in uterine weight (5,6), ovarian multioocyte follicles (7), and earlier vaginal opening (6). In a recent study in postmenopausal women, a small percentage of the participants developed endometrial hyperplasia in response to long-term exposure to dietary soy phytoestrogens (8). The biological effects of isoflavones on mammary glands also are limited mostly to animal studies. Exposure of pregnant mice to genistein increased the density of mammary glands' terminal buds in the female offspring (9). Similarly, intrauterine exposure of rats to genistein, which lasted through lactation, induced ductal/alveolar hyperplasia of the mammary glands (10). In addition, early mammary gland differentiation was observed in female rats in response to 5 days' administration of formononetin, an isoflavonoid phytoestrogen that is not present in soy (11), and in response to genistein administration (12). A single study in premenopausal women showed that short-term dietary soy has a weak estrogenic effect on the breast, as measured by the expression of nipple aspirate apolipoprotein D and pS2, 2 estrogen-regulated proteins (13). Similar studies in infants are lacking, despite long-term exposure to soy-based formulas during the early months of life. There are, however, a few other endocrine disruptors reported to be associated with breast development in infancy and childhood. Estrogen-containing hair and cosmetic products have been associated with breast development in infants (14,15), and premature breast development in young Puerto Rican girls was attributed to phthalates (16), plasticizers with known estrogenic and antiandrogenic activity that are used in numerous consumer products, commodities, and building materials.
The isoflavone content in different soy-based infant formulas ranges from 32 to 47 mg/L, compared with negligible levels in human breast milk and cow's milk–based formulas (17,18). Consequently, circulating plasma concentrations of isoflavones in 4-month-old infants fed exclusively soy-based formula were found to be 100 to 200 times higher than plasma levels in infants fed breast milk and dairy-based formulas, and 13,000 to 22,000 times higher than plasma estradiol concentration in early life (19). When adjusted for body weight, these infants are exposed to a daily intake of isoflavones that is 4- to 13-fold higher than that reported to exert significant physiological effects on the hormonal regulation of women's menstrual cycles (19). However, despite their abundance in soy-based formulas and high plasma concentrations in soy-fed infants, the estimated estrogenicity of isoflavones is 2 to 4 orders of magnitude lower than that of estradiol (20), making their role as endocrine disruptors questionable.
To assess possible endocrine effects of soy-based formulas during infancy, we conducted a cross-sectional study in a group of female infants and evaluated the association between soy-based diets and the presence of breast buds, the latter serving as a measure of estrogenic influence. The rationale for studying this particular outcome was derived from the above-mentioned animal studies that showed the effects of isoflavones on female breast development and maturation.
PATIENTS AND METHODS
Female infants (n = 694) ages 3 months to 24 months from 10 general pediatric clinics in central Israel were consecutively enrolled in the study between January 2003 and December 2004, none of them having been referred for breast development. Between 60 and 70 participants were recruited from each clinic. All of the participating pediatricians were instructed and trained by 1 of the authors (A.Z.) to create unified methods for the study. Breast status was initially evaluated as part of the physical examination, by laying the patients in a supine position and assessing the breast diameter by pinching the breast tissue between thumb and index finger. Breast bud was defined as a palpable tissue of at least 1.5 cm in diameter. The examiners were aided by the smallest Israeli coin, which is exactly 1.5 cm in diameter. Breast status was initially documented as part of the physical examination, followed by a parent's report on the type of formula or breast milk supplemented to the infant from birth, and the age (in months) of beginning and termination of each type of diet. The order of documentation was purposely selected to eliminate any potential bias in breast definition from a prior knowledge of the type of diet consumed by the infant. Infants that had continuously consumed soy-based formulas for more than 3 months were enrolled in the soy group, whereas infants consuming breast milk and cow's milk–based formula (both of which contain negligible amounts of phytoestrogens (17,18)) were enrolled together in the milk group. The study was approved by the ethics committee of our institute, and verbal informed consent was obtained from the parents.
Differences between prevalence data and proportions were analyzed by chi-square test (when sample size was sufficiently large) or by Fisher exact test. Age of soy initiation and length of exposure in different groups were compared by t test and Mann-Whitney rank sum test for nonparametric data. Differences between means of length of exposure within the soy group were tested by Kruskal-Wallis analysis of variance nonparametric test and pairwise multiple comparisons (Dunn method).
The soy and milk groups consisted of 92 (13.3%) and 602 (86.7%) infants, respectively. The mean age of the milk and soy groups was similar: 11.5 ± 5.6 months and 12.2 ± 5.3 months, respectively, with a range of 3.1 to 24 months in both groups. Breast bud prevalence was similar between groups (Table 1). However, when the data was analyzed separately for each year, breast buds were more prevalent in the soy versus milk group during the second year of life (P = 0.02, Table 1), with an odds ratio (OR) of 2.45 (95% confidence interval [CI] 1.11–5.39). A significant decline in the prevalence of breast buds was found between the first and second year in milk- but not soy-consuming infants (P < 0.001, Table 1). None of the participants had any other signs of sexual development, such as pubic hair.
Most soy formula–fed infants also had consumed cow's milk–based formula, breast milk, or both prior to their clinic attendance (Table 2). In 17 infants (18.5% of the soy group) that were exclusively soy-fed, the duration of soy formula consumption was longer than in the other subgroups with mixed feeding. Nevertheless, in all of the soy subgroups, the prevalence of breast buds was similar (Table 2). Similarly, more than half of the infants in the milk group consumed both breast milk and dairy-based formula either simultaneously or sequentially, whereas 97 infants consumed breast milk and 196 infants consumed milk-based formula almost exclusively (ie, were exposed to the other type of diet for ≤1 month). Most of those infants were observed during the first year of life. Breast bud prevalence in these 2 subgroups was similar: 13.4% in the breast-fed infants and 19.4% in those consuming milk-based formula (P = 0.267).
Infants in the soy group who were examined during their second year of life had longer exposure to soy-based formulas than infants examined during their first year of life (11.7 ± 5.3 vs 6.4 ± 2.5 months; P < 0.001). Nevertheless, there was no difference in the prevalence of breast buds between participants grouped by their length of exposure to soy-based infant formula (Table 3). Similarly, although infants examined in the second year started soy formula feeding later than infants in the first year (4.3 ± 4.0 vs 2.0 ± 1.8 months; P = 0.007), the starting age of soy formula consumption had no effect on the prevalence of breast buds (Table 4).
All but 1 infant in the soy group were started on soy formula in their first year of life, and most infants in this group still consumed soy-based formulas at the time of examination: 33 of 42 infants (78.6%) and 38 of 50 infants (76.0%) in the first and second year, respectively.
Soy-based infant formulas have been widely used since the turn of the 20th century (21). A substantial body of literature, which we have reviewed (22), demonstrates that infants fed soy formulas thrive and develop normally. In line with these observations, a reassessment of a cohort of young men and women who had been assigned as infants to either soy- or cow's milk–based formulas provides some reassurance that soy-related significant adverse effects are unlikely (23). In that study by Strom et al (23), those given soy formula as infants did not differ from those given cow's milk–based formula in self-reported parameters of growth, puberty, and fertility. However, in spite of the lack of evidence for endocrine effects in humans from soy-based formulas during infancy (20,24), some concern does exist, mainly with respect to the exposure of infants during a critical period of development to the potentially estrogen-like effects of isoflavones in soy-based formulas. This concern has been fueled by studies in animal models and women, but similar studies in neonates are lacking. In a single study in girls from Puerto Rico that was conducted following a 3-fold increase in the prevalence of premature thelarche in this population, Freni-Titulaer et al (25) found an association between premature breast development (before 2 years of age) and 3 environmental factors, 1 of which was soy-based infant formula. Although this association cannot be considered causal, because many of the girls were not exposed to any of the environmental factors, some role for soy-derived phytoestrogens in a multifactorial condition is possible.
Given the uncertainty and contradictory views regarding the safety of soy-based infant formulas, a number of governmental and professional bodies have adopted a precautionary approach to this issue (24,26–28). The main recommendation of all of these regulatory bodies is that soy-based infant formulas should be reserved for specific medical indications (hereditary lactose intolerance and galactosemia), whereas the use of these formulas for “cultural” reasons such as vegetarianism is not recommended.
The findings of our study support this approach because we show for the first time a possible endocrine effect of soy-based formulas in female infants, manifested in a higher prevalence of breast buds during the second year of life. We postulate that this effect is related to the estrogenic effect of the isoflavones in soy. The absence of differences in breast prevalence between soy and milk groups during the first year may derive from the relatively high plasma concentrations of endogenous estrogens, especially during the first several months of life (minipuberty) (29), which may blur the minor estrogenic effects of phytoestrogens. It is only when estradiol levels decline later in the first year of life that the estrogenic effects of soy-derived isoflavones become apparent. Alternatively, during the first year of life, isoflavones may actually function as partial estrogen antagonists, an effect reported when high concentrations of isoflavones were combined with physiological concentrations of estradiol (4,8,30). During the second year of life, however, isoflavones may express mainly their estrogen agonist effects, when endogenous estradiol levels are low.
It is conceivable that during the second year of life, infants fed soy-based formulas are exposed to smaller amounts and maintain lower plasma concentrations of isoflavones compared with first-year infants, given that their diet is more varied. Nevertheless, when adjusting the estimation of Setchell et al (19) in 4-month-olds to an 18-month-old infant (6 kg vs 11 kg body weight, and an estimated daily infant formula intake of 900–1000 mL vs 250 mL formula at least once per day), the calculated soy intake at the second year of life is comparable, on a body-weight basis, to the amount that induces hormonal effects in adults (19). Furthermore, it should be emphasized that even when serum isoflavone levels are low, target organs such as the ovary, uterus, and mammary gland can store these compounds at relatively higher concentrations (13,31). It is therefore an oversimplification to predict the estrogenic effects of isoflavones based only on serum levels.
Two observations in this study lend support to the conclusion that soy-derived isoflavones do not induce breast development de novo, but rather maintain its presence: Although breast bud prevalence was similar in both dietary groups during the first year of life, it declined only in the milk group during the second year of life, and remained stable in the soy group; and although the prevalence of breast buds in the second year in the soy group was higher than in the milk group, most soy-fed infants did not develop breast buds, in either the first (81%) or second (78%) year. We postulate that only those who developed breast buds in their first year due to endogenous estrogens (about 20%) retain it when continuously exposed to soy-derived isoflavones.
Several limitations of our study should be addressed. It is not a population-based study, there is no control of possible confounding factors such as exposure to other unknown endocrine disruptors, and there may be misclassification of the exposure to different diets because collected data was based only on verbal reports by the parents. A possible bias in our study is overestimation of breast buds prevalence due to misdiagnosis of infant fat as breast bud. This is particularly important because breast-fed girls have been reported to gain weight more slowly during the first 9 months compared with formula-fed girls (32). However, we did not find a significant difference in breast buds prevalence between predominantly breast-fed infants and those predominantly fed milk-based formula. Furthermore, differences in breast bud prevalence between milk and soy groups were observed only during the second year of life, when differences in body composition are no longer apparent (32). In addition, it is difficult to standardize the measurement of breast tissue, as has been demonstrated by a certain degree of interobserver variation (33). To minimize this variability and to reduce misinterpretation of fat tissue as breast tissue, we defined a diameter of at least 1.5 cm for the diagnosis of breast bud. This definition reflects the upper limit of breast size in female infants shortly after birth (34), at 3 months (33), and at 5 months to 8 months of age (34).
The prevalence of soy-based formula consumption in our study (13.3%) is similar in general terms to the reported prevalence in Britain (7%) (35), New Zealand (13%) (35), and the United States (10–20%) (36). The prevalence of soy-based consumption in Israel recently was evaluated in a telephone survey among mothers of 1803 infants attending well-baby clinics (37). The calculated prevalence was 10.4% at 2 months and 31.5% at 12 months, and about 70% of the soy-fed infants were given this formula for more than 6 months. By comparison, the prevalence of soy-based formula consumption around 12 months in our study was 21.9% (a similar comparison to the reported prevalence at 2 months of age was impossible because of age limits in our study). The difference between those figures may derive from different methods of data collection and from somewhat different cultural and socioeconomic characteristics of the populations in the 2 studies.
The average isoflavone concentration in a common brand of prepared Israeli soy-based infant formula is 31.7 mg/L (according to the manufacturer), which is comparable to the reported isoflavone concentrations in 5 commercially available soy-based formulas in the United States (range 32–47 mg/L) (17,19) and higher than levels found in soy formulas from Australia (17.2–21.9 mg/L) (38).
In conclusion, we suggest that in female infants, signs of endogenous estrogen effect that are initiated in the first few months of life wane more slowly in infants fed soy formula than in those fed breast milk or cow's-milk formula. This relatively subtle effect of isoflavones is easily ignored, which may explain the absence of documented estrogenic effects of soy over decades of soy-based formula consumption. A longitudinal study is needed to confirm our conclusions.
We wish to thank the pediatricians who took part in this study: Doron Dushinski, Sophia German, Irena Fishelev, Weida Agis, Neta Ben-Israel, Vered Lazaros, Ayala Yaron, Levana Sinai, Amit Nachum, and Joseph Aladjem.
1. Kuiper GG, Carlsson B, Grandien K, et al
. Comparison of the ligand binding specificity and transcript tissue distribution of estrogen receptors alpha and beta. Endocrinology 1997; 138:863–870.
2. Kuiper GG, Lemmen JG, Carlsson B, et al
. Interaction of estrogenic chemicals and phytoestrogens
with estrogen receptor β. Endocrinology 1998; 139:4252–4263.
3. Dixon RA, Ferreira D. Genistein. Phytochemistry 2002; 60:205–211.
4. Barnes S, Peterson G, Coward L. Rationale for the use of genistein-containing soy matrices in chemoprevention trials for breast and prostate cancer. J Cell Biochem 1995; 22:181–187.
5. Doerge DR, Twaddle NC, Banks EP, et al
. Pharmacokinetic analysis in serum of genistein administered subcutaneously to neonatal mice. Cancer Let 2002; 184:21–27.
6. Lewis RW, Brooks N, Milburn GM, et al
. The effects of the phytoestrogen genistein on the postnatal development of the rat. Toxicol Sci 2003; 71:74–83.
7. Jefferson WN, Couse JF, Padilla-Banks E, et al
. Neonatal exposure to genistein induces estrogen receptor (ER) alpha expression and multioocyte follicles in the maturing mouse ovary: evidence for ERbeta-mediated and nonestrogenic actions. Biol Reprod 2002; 67:1285–1296.
8. Unfer V, Casini ML, Costabile M, et al
. Endometrial effects of long-term treatment with phytoestrogens
: a randomized, double-blind, placebo-controlled study. Fertil Steril 2004; 82:145–148.
9. Hilakivi-Clarke L, Cho E, Clark R. Maternal genistein exposure mimics the effects of estrogen on mammary gland development in female mouse offspring. Oncol Rep 1998; 5:609–616.
10. Delclos KB, Bucci TJ, Lomax LG, et al
. Effects of dietary genistein exposure during development on male and female CD (Sprague-Dawley) rats. Reprod Toxicol 2001; 15:647–663.
11. Wang W, Tanaka Y, Han Z, et al
. Proliferative response of mammary glandular tissue to formononetin. Nutr Cancer 1995; 23:131–140.
12. Lamartiniere CA, Moore J, Holland M, et al
. Neonatal genistein chemoprevents mammary cancer. Soc Exp Biol Med 1995; 208:120–123.
13. Hargreaves DF, Potten CS, Harding C, et al
. Two-week dietary soy supplementation has an estrogenic effect on normal premenopausal breast. J Clin Endocrinol Metab 1999; 84:4017–4024.
14. Beas F, Vargas L, Spada RP, et al
. Pseudoprecocious puberty in infants caused by a dermal ointment containing estrogens. J Pediatr 1969; 75:127–130.
15. Tiwary CM. Premature sexual development in children following the use of estrogen- or placenta-containing hair products. Clin Pediatr (Phila) 1998; 37:733–739.
16. Colon I, Caro D, Bourdony CJ, et al
. Identification of phthalate esters in the serum of young Puerto Rican girls with premature breast development. Environ Health Perspect 2000; 108:895–900.
17. Setchell KDR, Zimmer-Nechemias L, Cai J, et al
. Isoflavone content of infant formulas and the metabolic fate of these phytoestrogens
in early life. Am J Clin Nutr 1998; 68:1453S–1461S.
18. Irvine CHG, Shand N, Fitzpatrick MG, et al
. Daily intake and urinary excretion of genistein and daidzein by infants fed soy- or dairy-based infant formulas. Am J Clin Nutr 1998; 68:1462S–1465S.
19. Setchell KDR, Zimmer-Nechemias L, Cai J, et al
. Exposure of infants to phyto-oestrogens from soy-based infant formula. Lancet 1997; 350:23–27.
20. Chen A, Rogan WJ. Isoflavones in soy infant formula: a review of evidence for endocrine and other activity in infants. Annu Rev Nutr 2004; 24:33–54.
21. Hill LW, Stuart HC. A soy bean food preparation for feeding infants with milk idiosyncrasy. JAMA 1929; 93:985–987.
22. Zung A, Reifen R, Kerem Z, et al
: the pediatric perspective. J Pediatr Gastroenterol Nutr 2001; 33:112–118.
23. Strom BL, Schinnar R, Ziegler EE, et al
. Exposure to soy-based formula
in infancy and endocrinological and reproductive outcomes in young adulthood. JAMA 2001; 286:807–814.
24. Tuohy PG. Soy infant formula and phytoestrogens
. J Paediatr Child Health 2003; 39:401–405.
25. Freni-Titulaer LW, Cordero JF, Haddock L, et al
. Premature thelarche in Puerto Rico. A search for environmental factors. Am J Dis Child 1986; 140:1263–1267.
26. Committee on Toxicity of Chemicals in Food, Consumer Products, and the Environment. Phytoestrogens and Health
. London: Food Standard Agency, 2003.
27. Agence Francaise de Securite des Aliments (French Food Safety Agency). Report of the working group on phytoestrogens
2005 [French]. http://www.afssa.fr
. Accessed December 2007.
28. Agostoni C, Axelsson I, Goulet O, et al
. Soy protein infant formulae and follow-up formulae: a commentary by the ESPGHAN Committee on Nutrition. J Pediatr Gastroenterol Nutr 2006; 42:352–361.
29. Rosenfield RL. Puberty in the female and its disorder. In: 2nd ed. Edited by Sperling MA. (ed). Pediatric Endocrinology
. Philadelphia: Saunders 2002; pp. 458–460.
30. Arici A, Bukulmez O. Phyto-oestrogens and the endometrium. Lancet 2004; 364:2081–2082.
31. Doerge DR. Metabolism and disposition of genistein, the principal soy isoflavone. In: Gilani GS, Anderson JJB, editors. Phytoestrogens
and Health. Champaign, IL: AOCS Press; 2002. pp. 196–208.
32. Butte NF, Wong WW, Hopkinson JM, et al
. Infant feeding mode affects early growth and body composition. Pediatrics 2000; 106:1355–1366.
33. Schmidt IM, Chellakooty M, Haavisto AM, et al
. Gender difference in breast tissue size in infancy: correlation with serum estradiol. Pediatr Res 2002; 52:682–686.
34. McKiernan JF, Hull D. Breast development in the newborn. Arch Dis Child 1981; 56:525–529.
35. Essex C. Phytoestrogens
and soy based infant formulas. BMJ 1996; 31:507–508.
36. Gillooly M, Torrance JD, Bothwell TH, et al
. The relative effects of ascorbic acid on iron absorption from soy-based and milk-based infant formulas. Am J Clin Nutr 1984; 40:522–527.
37. Berger-Achituv S, Shohat T, Romano-Zelekha O, et al
. Widespread use of soy-based formula
without clinical indications. J Pediatr Gastroenterol Nutr 2005; 41:660–666.
38. Knight DC, Eden JA, Huang JL, et al
. Isoflavone content of infant food and formulas. J Paediatr Child Health 1998; 34:135–138.