Infant feeding with soy formula milk: effects on the testis and on blood testosterone levels in marmoset monkeys during the period of neonatal testicular activity

Richard M. Sharpe,1, Bronwen Martin, Keith Morris, Irene Greig, Chris McKinnell, Alan S. McNeilly and Marion Walker

MRC Human Reproductive Sciences Unit, Centre for Reproductive Biology, 37 Chalmers Street, Edinburgh EH3 9ET, UK


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
BACKGROUND: This study has addressed concerns about possible effects of feeding human infants soy formula milk (SFM). METHODS: This is a feeding study in marmosets, using a mainly co-twin design. From 4–5 until 35–45 days of age, co-twin males were fed by hand with either standard (cow) formula milk (SMA = controls) or with SFM for ~8 h each day (2 h at weekends) and intake related to bodyweight. Blood samples were collected at 18–20 and at 35–45 days of age in 13 sets of co-twins plus two non-twin males per group and, at the later age, seven sets of co-twins were killed and the testes and pituitary gland fixed for cell counts. RESULTS: Weight gain and formula intake were similar in both feeding groups. SMA-fed males had mean testosterone levels of 2.8–3.1 ng/ml, typical of the ‘neonatal testosterone rise’, whereas SFM-fed males exhibited consistently lower mean levels (1.2–2.6 ng/ml); paired comparison in SMA-and SFM-fed co-twins at day 35–45 revealed 53–70% lower levels in 11 of 13 co-twins fed with SFM (P = 0.004). Further evidence for suppression of testosterone levels in SFM-fed males came from comparison of the frequency of low testosterone levels (<0.5 ng/ml). In historical controls aged 35–45 days, two out of 22 values were <0.5 ng/ml, a similar frequency as found in control SMA-fed males (one out of 15 values <0.5 ng/ml). In contrast, 12 out of 15 values for SFM-fed males were <0.5 ng/ml (P < 0.001). There was no consistent relationship between SFM intake/g and testosterone levels. Paradoxically, the mean number of Leydig cells per testis was increased by 74% (P < 0.001) in co-twins fed SFM, when compared with their SMA-fed brothers, whereas no significant changes were found in numbers of Sertoli and germ cells. Because of the lack of gonadotrophin assays, the number of immunopositive LHß and FSHß cells in the pituitary gland, and their ratio, were determined but no consistent difference was found between SMA- and SFM-fed twins. CONCLUSIONS: Based on the average isoflavone content of the SFM brand used, intake of isoflavones was estimated at 1.6–3.5 mg/kg/day in the SFM-fed marmosets which is 40–87% of that reported in 4 month human infants fed on a 100% SFM diet. It is therefore considered likely that similar, or larger, effects to those shown here in marmosets may occur in human male infants fed with SFM. Whether the changes described result in longer-term effects is under investigation.

Key words: co-twins/marmoset/soy formula milk/Leydig, Sertoli and germ cells/testosterone


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Within the last half-century, there have been major changes in Western countries in the type of feeding offered to human infants prior to weaning. This has involved predominantly a switch from breastfeeding to formula milk feeding based on cow's milk but, more recently, powdered milk formula derived from soybean extracts has been introduced and has been used for human infant feeding in various countries. The use of soy formula milk (SFM) was originally advocated or prescribed for infants exhibiting features of intolerance to cow's milk formula (Polack et al., 1999Go). However, based on current estimates of the prevalence of feeding SFM to infants (2–3% in the UK and Australia, 20–25% in the USA and Canada) (UK Department of Health, 1996Go; Australian College of Paediatrics, 1998Go; American Academy of Pediatrics Committee on Nutrition, 1998Go; Canadian Pediatric Society, Dieticians of Canada, Health Canada, 1998Go), it is apparent that this is no longer the case and that SFM is often the breastmilk substitute of choice for many mothers and is frequently used within the first months of life (Polack et al., 1999Go). Although careful studies have demonstrated comparable growth rates of infants fed on either SFM or a more conventional cow's milk formula (Jung and Carr, 1977Go; Kohler et al., 1984Go; Mimouni et al., 1993Go; Lasekan et al., 1999Go), concerns have been expressed as to whether or not the high isoflavenoid phytoestrogen content of SFM might exert adverse effects on the developing infant (Irvine et al., 1995Go, 1998aGo,Irvine et al., bGo; Setchell et al., 1997Go, 1998Go). These concerns have arisen because of a number of factors. First, evidence that isoflavenoid phytoestrogens, such as genistein and daidzein, or other phytoestrogens, can exert a range of effects in animals such as birds, rodents, cheetahs and sheep (Irvine et al., 1995Go; Whitten et al., 1995Go; Whitten and Naftolin, 1999Go). These effects can apply to both sexes and, in highly susceptible species such as the sheep, can include irreversible infertility (in the female) (Whitten and Naftolin, 1999Go); in rodents, neonatal exposure of female pups directly to phytoestrogens or indirectly via their mother's milk can cause ovulatory cycle problems later in life (Whitten et al., 1995Go; Whitten and Naftolin, 1999Go). Second, the growing evidence from laboratory animal studies that even brief neonatal exposure to (potent) estrogens can induce a spectrum of changes to the testis (Sertoli and Leydig cells) and reproductive tract and their function (Toppari et al., 1996Go; Atanassova et al., 1999Go; Williams et al., 2001Go). Third, the finding that ~4 month human infants fed on a 100% SFM diet have blood levels of isoflavenoids that are >=5-fold higher than those of adult humans eating a soy rich diet (Setchell et al., 1997Go, 1998Go; Irvine et al., 1998aGo,bGo). The latter studies concluded that such levels are almost certain to exert (unknown) biological effects, though opinion is divided as to whether these are likely to be adverse (Irvine et al., 1998aGo,bGo) or beneficial (Setchell et al., 1997Go).

In contrast to the above concerns, there is also interest in the potential beneficial effects of consuming phytoestrogens/soy. This is because Oriental countries in which the traditional diet is rich in soy products, such as Japan and China, have a low incidence of breast cancer, invasive prostate cancer and cardiovascular disease, and it has been argued, though not proven, that this stems from soy and/or phytoestrogen consumption (Adlercreutz, 1990Go; Adlercreutz and Mazur, 1997Go). If soy consumption does have beneficial health effects it is unclear whether these result from lifetime exposure or might, for example, result from exposure during a critical period during development. Based on this possibility, the feeding of SFM to infants could have potential lifelong health beneficial effects (Setchell et al., 1997Go), though it is emphasized that feeding of soy products, including SFM, to infants in the first 4–6 months has not been traditional in Oriental countries (Guy and Yeh, 1938Go).

There are no reports in the literature on human infants fed with SFM that help to distinguish between the two opposing possibilities given above, though a recent (retrospective) cohort study, based on telephone interviews, reported no overt reproductive problems in young adults fed as infants with SFM (Strom et al., 2001Go). As no major adverse effects of feeding SFM have come to light since its widespread introduction more than 30 years ago, it has been argued that adverse effects are unlikely (Klein, 1998Go). Though numerous studies in infant rodents have examined the effects of exposure to phytoestrogens during the neonatal period (Whitten and Naftolin, 1999Go), it is unknown whether these are relevant to human infants for two important reasons. First, the route and mode of administration of phytoestrogens are generally not comparable. In rodents, the compounds have usually been injected (once daily) rather than being consumed orally by the pups, and though some studies have administered soy, SFM or phytoestrogens to the lactating mother, this is not directly comparable with its consumption by the pups themselves (Setchell et al., 1997Go; Irvine et al., 1998aGo). Second, neonatal rodents are relatively undeveloped when compared with neonatal human infants and may therefore be more vulnerable to effects of hormonally active compounds. With regard to male human infants there is a third and potentially important difference from rodents, namely the occurrence of a prolonged ‘neonatal testosterone rise’ when testosterone levels increase to low adult levels during the first 4–6 months of life, associated with activation of gonadotrophin secretion (Winters et al., 1975Go; Forest, 1990Go; Mann and Fraser, 1996Go; Andersson et al., 1998Go), Sertoli cell proliferation (Cortes et al., 1987Go; Sharpe, 1999Go) and a steep rise in the secretion of inhibin B by the latter cells (Andersson et al., 1998Go). In contrast, if a neonatal testosterone rise occurs in rodents it is restricted to a matter of hours around the day of birth. The role of the neonatal testosterone rise in the human male remains unclear, though effects on the immune system (Mann and Fraser, 1996Go), on sexually dimorphic brain development (Mann and Fraser, 1996Go), the testis and reproductive tract (Mann and Fraser, 1996Go; Sharpe et al., 2000Go; McKinnell et al., 2001Go), and genitalia (Wallen et al., 1995Go; Mann and Fraser, 1996Go; Brown et al., 1999Go) are distinct possibilities, based on studies in non-human primates which also exhibit a neonatal testosterone rise (Wallen et al., 1995Go; Mann and Fraser, 1996Go; Brown et al., 1999Go; McKinnell et al., 2001Go).

Because of the widespread prevalence of feeding human infants with SFM, it is clearly important to establish whether this is likely to have any biological consequences, in particular to identify or rule out the possibility of adverse effects (Irvine et al., 1998aGo). To address this issue, we have undertaken a feeding study in a non-human primate, the marmoset, and report in this paper the effects observed during and at the end of the feeding period which encompassed the period of the neonatal testosterone rise.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Animals and treatments
Animals were captive-bred common marmoset monkeys (Callithrix jacchus), maintained in a colony that has been self-sustaining since 1973. For the present studies, a total of 30 newborn male marmosets were used, of which 26 were 13 pairs of co-twins. Marmosets show considerable between-animal variability that would normally necessitate the use of larger numbers of animals, based on power calculations. To minimize between-animal variability, the present study used predominantly co-twins (one SMA-fed, one SFM-fed), which tend to be highly comparable, and this enabled pair-wise evaluation of data and thus minimized the use of animals. Commencing at 4–5 days of age, infant marmosets were treated as follows. During the daytime (~8 h on weekdays, ~2 h at weekends), infant animals were separated from their mothers and left with their fathers in the family cage, though, by using a wire mesh divider, they remained within sight of their mothers. As our marmosets are maintained in family groups and it is normally the father or older siblings who look after infant marmosets (apart from feeding), this daytime separation caused no obvious distress or problems for either mothers or infants. During the period when the mother was absent from the family cage, the infant animals were removed at set intervals and hand-fed with either standard (cow's milk-based) formula milk (SMA Gold; SMA Nutrition, Taplow, Berkshire, UK) or with SFM (Wysoy; SMA Nutrition) from day 4/5 until day 35–45; both formulae milks were purchased from a supermarket. The formula powders were added to tapwater and heated and diluted as per the manufacturer's instructions. The suspension was taken up into a 1 ml syringe with a soft silicon rubber tube fitted to the end and offered to the infant marmoset, which was allowed to drink as much of the milk as it wanted from the tube. The syringe was refilled as appropriate and offered to the marmoset until it decided to stop drinking. The amount of formula taken was recorded and animals were weighed each day prior to feeding. Infants were fed three or four times (weekdays) or once or twice daily (weekends) and habituated well to the routine. After the last formula feed of the day, the infant's mother was returned to the family cage and she was allowed to breastfeed her infants until the following morning. Both SMA-fed (controls) and SFM-fed animals showed normal rates of weight gain during the study period. The isoflavone content of the SFM used in the present studies was not measured, but based on analyses of various soy-based infant formulae bought in the UK, levels in the range 18–41 mg aglycone/l would be predicted and mean levels of 25.5 mg aglycone/l were reported for the particular brand of SFM that we used (Ministry of Agriculture, Food and Fisheries, 1998Go; Irvine et al., 1998aGo; Knight et al., 1998Go; Setchell et al., 1998Go); isoflavones were not detected in any of the brands of cow's formula milk that were analysed (Ministry of Agriculture, Food and Fisheries, 1998Go). In all other respects, SFM has been constituted to provide comparable nutritional/calorific intake with cow's milk-based infant formulae.

Approximately half-way through the study (day 18–20, 14:30–16:30), a blood sample (~0.2 ml) was obtained from the femoral vein of each infant using a 1 ml syringe fitted with a 27G needle. It was not necessary to sedate the animals for this procedure as it caused no obvious discomfort. On day 35–45, seven sets of co-twins were killed via the i.p. injection of an overdose of sodium pentobarbitone (Euthatal; Rhone Merieux Ltd, Harlow, Essex, UK). Immediately after death (10:00–11:30) a terminal blood sample was obtained by cardiac puncture. Each blood sample was collected into a heparinized syringe and the plasma then separated by centrifugation and stored at –20°C until used for assay. The remaining 16 animals (eight SMA-fed, eight SFM-fed), including six sets of co-twins, had a blood sample collected at day 35–45 (14:30–16:30) but were then returned to their cages and treatment discontinued. These animals are currently growing to adulthood when the possibility of any long-term effects of the SFM exposure will be evaluated.

These studies were approved by the local ethical committee for studies in primates and were performed according to the Animal Scientific Procedures (UK) Act (1986) under Project Licence approval by the UK Home Office.

Tissue collection and processing
Testes with epididymides attached, as well as the pituitary gland, were dissected free of connective tissue and immersion-fixed for either 5.5 h (left testis, pituitary gland) or 24 h (right testis) in Bouin's fluid, after which each testis was dissected away from the epididymis and weighed. The left testis and pituitary gland were then processed for 17.5 h in an automated Leica TP-1050 processor and embedded in paraffin wax. Sections of 5 µm thickness were cut and floated onto slides coated with 2% 3-aminopropyltriethoxy-silane (Sigma, Poole, Dorset) and dried at 50°C overnight before being used for morphological evaluation and for cell quantification studies as described below. After fixation, the right testis was sampled in a random systematic manner, i.e. of four transverse slices, either slices, 1 and 3 or slices 2 and 4 were sampled. These were processed through graded ethanols before infiltration with JB4 resin (TAAB, Aldermaston, Berkshire, UK) and used subsequently for enumeration of Sertoli and germ cells using the optical disector method (see below).

Determination of Sertoli and germ cell numbers per testis
After polymerization of the JB4 resin, 20 µm sections were cut on a Reichart 2050 microtome using a Diatome Histoknife, mounted onto glass slides and stained with Harris' haematoxylin. Sertoli and germ cells were then counted using the optical disector method (Wreford, 1995Go) as reported previously by us for the marmoset (Sharpe et al., 2000Go).

Determination of Leydig cell number and immunoexpression of 17{alpha}-hydroxylase/C17–20 lyase
Leydig cell volume per testis was determined using paraffin sections of testis that had been immunostained for 3ß-hydroxysteroid dehydrogenase (3ß-HSD) as described previously (Sharpe et al., 2000Go). The method used the Area Fraction Probe in the Stereologer software programme (Systems Planning and Analysis Inc., Alexandria, VA, USA) and utilized an Olympus BHS microscope fitted with an automatic stage (Applied Scientific Instrumentation Inc., Eugene, OR, USA). The area fraction probe places a grid in the frame and the Object Area fraction is determined by clicking each ‘x’ that touches the object of interest (in this case, the cytoplasm or nucleus of 3ß-HSD positive cells). The number of frames per slide counted was 5–13 and was determined by the program after taking into account the spacing between points, objects counted per frame and the size of the section. Frames for evaluation were selected automatically. A coefficient of error (CE) was calculated at intervals in the counting procedure and, if this exceeded the maximum acceptable value (0.15), then different settings recommended by the program were utilized in order to obtain the optimum CE. Once completed, Stereologer automatically displays the results including area fraction and CE. The values for area fraction were then converted to absolute volumes per testis by reference to testis volume (= weight). Average nuclear diameter was then determined by measuring the diameter of 100 random nuclei of 3ß-HSD positive cells for each animal and then converting Leydig cell nuclear volume per testis to Leydig cell number per testis (Wreford, 1995Go).

Immunoexpression of P450 17{alpha}-hydroxylase/C17–20 lyase utilized an antiserum provided kindly by Professor Ian Mason (Edinburgh). The methods used have been detailed previously (Majdic et al., 1996Go), and were broadly similar to those described above for 3ß-HSD.

Immunostained sections were examined and photographed using a Provis microscope (Olympus Optical, London, UK) fitted with a digital camera (DCS330; Eastman Kodak, Rochester, NY, USA). Captured images were transferred to a computer (G4; Apple Computer Inc., Cupertino, CA, USA) and compiled using Photoshop 5.0 (Adobe Systems Inc., Mountain View, CA, USA) before being printed using an Epson Stylus 870 colour printer (Seiko Epson Corp., Nagano, Japan).

Plasma levels of testosterone
Levels of testosterone in plasma were measured using an enzyme-linked immunosorbent assay (ELISA) adapted from an earlier radioimmunoassay method (Corker and Davidson, 1981Go). Plasma, to which was added trace amounts of [3H]testosterone (Amersham International, Little Chalfont, Bucks, UK), was extracted twice with 10 vols hexane:ether (4:1, v/v) and the organic phase dried down under N2 at 55°C. The efficiency of extraction averaged 75%. The second antibody was immobilized to an ELISA plate by addition of 100 µl acid-purified donkey anti-goat/sheep IgG (250–350 mg/ml) diluted in 0.1 mol/l sodium carbonate buffer, pH 9.6. The plate was sealed and incubated overnight at 4°C. The wells were then washed twice with 0.1% Tween-20 and incubated for 10 min at room temperature with 0.2 ml of the same solution to block non-specific binding sites. Samples in duplicate (50 µl) were assayed after dilution in 0.1 mol/l PBS, pH 7.4 containing 0.1% gelatin (Sigma) and incubated overnight at 4°C with 50 µl sheep anti-testosterone-3-cmo–bovine serum albumin (BSA) diluted 1:100 000 plus 50 µl testosterone-3-cmo labelled with 1:20 000 horse-radish peroxidase (Amdex; Amersham Pharmacia Biotech, St Albans, Herts, UK). The plate was then washed several times with 0.1% Tween-20 before addition of 0.2 ml substrate (5 mmol/l O-phenylenediamine; Sigma) and 0.03% hydrogen peroxide diluted in 0.1 mol/l citrate/phosphate, pH 5.0 to each well. The plate was then incubated in the dark for 10–30 min until the colour reaction was optimal. The reaction was stopped by addition of 50 µl 2 mol/l sulphuric acid to each well and the optical density then read at 492 nm in a plate reader. The limit of detection was 12 pg/ml and inter- and intra-assay coefficients of variation were <15%. Samples from first and second bleeds from each set of twins were always extracted and assayed at the same time.

Enumeration of LHß- and FSHß-immunopositive cells in the pituitary gland
Paraffin-embedded pituitary glands were sectioned transversely until approximately the midline and consecutive sections from the midline region were then immunostained for LHß and FSHß using specific antibodies as reported previously (Brown et al., 2001Go). FSHß was detected using a polyclonal rabbit antibody raised against human FSHß (M94, gift from Dr S.Lynch, Birmingham, UK) and LHß using a monoclonal mouse antibody raised against bovine LHß (518B7, gift from Dr J.F.Roser, University of California–Davis, USA).

Unless otherwise stated, all incubations were performed at room temperature. Sections were cut at 5 µm and floated onto slides coated with either 2% 3-aminopropyltriethoxy-silane (Sigma) or poly-lysine (BDH Chemicals, Poole, Dorset, UK) and dried overnight at 50°C. The slides were dewaxed, rehydrated and endogenous peroxidase was blocked using 3% (v/v) hydrogen peroxide in methanol. After washing in water, the sections were washed twice (5 min each) in Tris-buffered saline (0.05 mol/l TBS, pH 7.4 and 0.85% NaCl). For FSHß, slides were then blocked for 30 min with normal swine serum (NSS; Diagnostics Scotland, Carluke, UK) diluted 1:5 in TBS containing 5% BSA; for LHß, slides were blocked with normal rabbit serum (NRS) diluted 1:5 in TBS with 5% BSA. The primary antibodies were diluted in the appropriate blocking solution (FSHß 1:1000; LHß 1:5000) and 100 µl was added to each slide before incubation at 4°C overnight in a light-proof humidity chamber. The slides were then washed in TBS (2x5 min) before incubation for 30 min with a biotinylated second antibody, namely swine anti-rabbit (Dako, Ely, UK) for FSHß or rabbit anti-mouse (Dako) for LHß, diluted 1:500 in the appropriate blocking serum. After two washes in TBS (2x5 min), avidin–biotin conjugated horse-radish peroxidase (ABC-HRP; Dako) was applied for 30 min. After two final washes (2x5 min) in TBS a solution of diaminobenzidine (DAB) was applied (liquid DAB; Dako). The slides were developed until the colour reached the required intensity in the control sections, and the reaction was then stopped by immersing the slides in distilled water. The slides were counterstained with haematoxylin before being dehydrated by immersion in a graded series of ethanols and then being cleared in xylene. A coverslip was fixed over the sections using Pertex mounting medium (Cell Path, Hemel Hempstead, UK).

Immunolocalization studies for FSHß and LHß were repeated on two separate occasions for each of four sets of co-twin marmosets and showed good agreement between runs. Sections from control and treated animals were run in parallel at all times. Immunostained sections were examined on a Zeiss Axioskop light microscope using a x40 objective. The number of immunopositive cells was determined by placing a 21 mm diameter graticule over a randomly chosen area of the anterior pituitary. Within that graticule, the number of immunopositive (nucleated) cells (brown) was counted and the total number of non-immunopositive cells (blue) was also counted and the percentage immunopositive cells then calculated. This was done for four randomly chosen areas on each slide and repeated once. Final data for each animal were then computed as the average of the two runs.

Statistics
As predominantly co-twins were used for this study, paired t-test comparison of bodyweight, testis weight and cell number/volume data for SMA-fed (controls) and SFM-fed animals was utilized for all such comparisons. In most instances, data were log-transformed prior to analysis because of the considerable variation between sets of twins and because of heterogeneity of variance (especially for testosterone levels). Frequency of testosterone levels in a particular range utilized Fisher's exact test. Correlation analysis between formula intake and plasma testosterone levels utilized simple linear regression.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Bodyweight, testis weight and formula intake
Bodyweights in SMA-fed and SFM-fed infants were comparable at the start and end of the period of treatment and bodyweight increment during this period was also comparable (Table IGo) and was within normal limits for our colony of marmosets. Values were similar for co-twins (n = 13) and for the total study set that included two sets of singleton infants per treatment group (Table IGo). All animals appeared healthy and there were no unexpected problems in running the study. There was considerable variation between twins in the level of formula intake and in most cases this variation applied equally to both co-twins. In most infants (SMA 11/15, SFM 13/15), formula intake per gram bodyweight was higher in the second half of the study (i.e. after the first bleed) than in the first half (i.e. prior to the first bleed) and comparison of formula intake in the 7 days immediately preceding the day of first and second blood sampling revealed a significantly higher intake prior to the second bleed, which was especially marked for SFM-fed infants (Table IGo). Typical formula intake patterns for two sets of co-twins are shown in Figure 1Go. Three males fed with SFM were slightly more reticent in taking the feed at the start of the study whereas others clearly enjoyed taking the feed from the start. These and the between-twin differences account for the considerable variation in overall intake of formula per day between animals (Figure 2Go, top panel). Testis weight in the seven sets of co-twins that were killed at 35–45 days were not significantly different for SMA- and SFM-fed animals (Table IGo and Figure 2Go).


View this table:
[in this window]
[in a new window]
 
Table I. Mean (± SD) bodyweight parameters, formula milk intake and plasma testosterone levels during the study and testis weight at day 35–45 in marmosets fed with either standard (cow) formula milk (SMA, controls) or soy formula milk (SFM)
 


View larger version (41K):
[in this window]
[in a new window]
 
Figure 1. Feeding records of two sets of marmoset co-twins fed with either standard (cow) formula milk (SMA) or soy formula milk (SFM). The upper feeding record is typical of the majority of animals in the present study in that it shows a clearly higher intake of formula in both twins in the second half of the study period (dashed line shows the overall mean intake/g bodyweight). In contrast, the lower feeding record is from twins in which there was little change in formula intake during the study, a pattern shown in only a minority of animals. The times of blood sampling (arrows) and the actual level of testosterone measured are also indicated. Small hatched bars indicate weekends.

 


View larger version (53K):
[in this window]
[in a new window]
 
Figure 2. Plasma levels of testosterone at the end of the feeding period (35–40 days of age) (middle panel) in co-twin marmosets, fed with either standard (cow) formula milk (SMA) or soy formula milk (SFM), in relation to the mean intake of SMA or SFM/g bodyweight during the study period (top panel) and testis weight (lower panel). Each bar in the main panels represents one co-twin whilst overall mean levels for each treatment group are shown to the right. Note that results are plotted in order of descending intake of SFM/g bodyweight by the SFM-fed co-twin (top). Note also that testis weights are shown only for the seven sets of co-twins killed at 35–40 days of age, and cell count data for these organs are illustrated in Figure 3Go. P-Values shown to the right are based on co-twin comparison using the paired t-test after logarithmic transformation of the data. NS = not significant.

 
Plasma testosterone levels and the relationship to soy formula intake
At both day 18–20 and day 35–45, plasma levels of testosterone in control marmosets fed with SMA exhibited typical mean levels for neonatal marmosets of ~2.8–3.1 ng/ml but with a wide range (high SD), reflecting the episodic nature of testosterone secretion (Tables I and IIGoGo, Figure 2Go). Mean testosterone levels in animals fed SFM were 30% lower at day 18–20 and 55% lower at day 35–45 when compared with the SMA group (Table IGo). Paired comparison of (log-transformed) testosterone levels in co-twins at day 35–45 revealed a statistically significant (P = 0.004) difference (Figure 2Go), but similar comparison at day 18–20 was not significant (P = 0.2). This age difference in effect could not be explained by difference in time of sample collection (day 18–20 all collected after noon; day 35–45, seven twin samples collected in the morning and 8 samples after noon), but might be related to the 49% average increase in SFM intake per gram bodyweight prior to the second blood sample, when compared with intake prior to the first blood sample (Table IGo). However, using data for both time- points, no consistent or significant relationship between SFM intake/g/day and testosterone levels was found (not shown). Although testosterone levels varied considerably between males, as shown in Figure 2Go, it was apparent that males fed with SFM had more low values (<0.5 ng/ml) than did SMA-fed males (Figure 2Go, Table IIGo). Testosterone levels <0.5 ng/ml did occur sporadically because of episodicity, but they did not occur with such high frequency as in the SFM-fed group in which 80% of testosterone values were <0.5 ng/ml at day 35–45 (Table IIGo). In contrast, in both the SMA-fed animals and in an historical cohort of 22 measurements made in the same age range in control animals, only 8 and 7% respectively of testosterone values were <0.5 ng/ml. The frequency of testosterone values <0.5 ng/ml in SFM-fed males was significantly different (P < 0.001) from that in both SMA-fed males and the historical cohort, whereas the latter two groups were comparable (Table IIGo). The historical cohort also showed mean ± SD levels of testosterone that were comparable with the SMA-fed animals at day 35–45 (Table IIGo). A similar comparison of testosterone values <0.5 ng/ml in SMA- and SFM-fed animals at day 18–20 did not show any significant difference (data not shown).


View this table:
[in this window]
[in a new window]
 
Table II. Comparative frequency of testosterone values <0.5 ng/ml and >0.5 ng/ml in historical controls compared with standard (cow) formula milk (SMA)- and soy formula milk (SFM)-fed males at age 35–45 days
 
Testicular cell counts and immunoexpression of 17{alpha}-hydroxylase/C17–20 lyase
Sertoli cell number per testis did not show any consistent change between SMA- and SFM-fed co-twins, nor did germ cell numbers per testis, although SFM-fed twins showed lower values than their SMA-fed brothers in six out of seven cases (Figure 3Go). The most marked and consistent change (P = 0.006) in testicular cell composition in SFM-fed co-twins, when compared with SMA-fed co-twins at day 35–45, was in Leydig (3ß-HSD immunopositive) cell numbers, which were increased by an average of 74% (Figures 3 and 4GoGo). As there are several examples in the literature of reduced testosterone levels being associated with low expression of 17{alpha}-hydroxylase/C17–20 lyase (Cigorraga et al., 1980Go; Majdic et al., 1996Go), immunoexpression of this enzyme was assessed in four sets of twins. However, no evidence for a consistent reduction in intensity of immunoexpression of this enzyme was detected (data not shown).



View larger version (47K):
[in this window]
[in a new window]
 
Figure 3. Testicular cell counts at the end of the feeding period (35–40 days of age) in co-twin marmosets fed with either standard (cow) formula milk (SMA) or soy formula milk (SFM). Each bar in the main panels represents one co-twin whilst overall mean levels for each treatment group are shown to the right. Note the consistent increase in Leydig cell numbers per testis in the SFM-fed, compared with the SMA-fed, co-twins (top). P-Values are based on co-twin comparison using the paired t-test after logarithmic transformation of the data. N/A = not available due to processing error (no tissue available for immunohistochemistry).

 


View larger version (166K):
[in this window]
[in a new window]
 
Figure 4. Representative sections of testis immunostained for 3ß-hydroxysteroid dehydrogenase (HSD) (= Leydig cells) at age 35–40 days in two sets of marmoset co-twins fed with either standard (cow) formula milk (SMA) or with soy formula milk (SFM). Note the higher level of cells expressing 3ß-HSD in each SFM-fed co-twin compared with their SMA-fed brothers. Scale bar = 100 µm.

 
Number of LHß and FSHß immunopositive cells in the pituitary gland
There are currently no workable assays for measuring blood gonadotrophins in the marmoset. To obtain some insight as to whether feeding with SFM might have affected the gonadotrophin axes, the numbers of LHß- and FSHß-immunopositive cells in the pituitary gland were determined in all sets of co-twins (Figure 5Go). No statistically significant differences in numbers of LHß- or FSHß-immunopositive cells, or their ratio, was observed though mean values for each parameter were lower for SFM-fed twins (Figure 5Go).



View larger version (44K):
[in this window]
[in a new window]
 
Figure 5. Pituitary cell counts for LHß (top) and FSHß (middle) and their ratio (bottom) at the end of the feeding period (35–40 days of age) in co-twin marmosets fed with either standard (cow) formula milk (SMA) or soy formula milk (SFM). Each bar in the main panels represents one co-twin whilst overall mean levels for each treatment group are shown to the right. P-Values are based on co-twin comparison using the paired t-test after logarithmic transformation of the data.

 

    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
The purpose of this study was to examine whether the feeding of SFM to infant marmosets, using similar ad-libitum feeding (controlled by the infants) as occurs in human infants fed with SFM, resulted in any detectable effects on the developing testis or on hormone levels. Our choice of the marmoset for these studies was based on its comparability with the human in terms of exhibiting a period of neonatal hormone activity (‘testosterone surge’) (Dixson, 1986Go; Lunn et al., 1994Go; Mann and Fraser, 1996Go; McKinnell et al., 2001Go) and Sertoli cell proliferation (Sharpe et al., 2000Go), and the high prevalence (80%) of dizygotic twins which enabled us to use a co-twin comparison that minimized the number of animals required to be used. Feeding with SFM resulted in a significant suppression of the ‘testosterone surge’, at least during the second half of the feeding study. The reason why testosterone levels were suppressed at days 35–40 but not at days 18–20 in SFM-fed males is not clear but may be related to longer duration of exposure to SFM or to the 49% higher intake of SFM prior to the second, compared with prior to the first, blood sample. Paradoxically, there was a commensurate increase in Leydig cell numbers per testis when SFM-fed twins were compared with their SMA (standard formula)-fed co-twin brothers, whereas Sertoli and germ cell numbers and overall testis weight were not significantly different in both feeding groups. These findings raise the likelihood of similar, and perhaps more marked, effects in human infant boys fed with SFM, though it is unclear whether such effects would have any major or untoward consequences.

There are two major considerations that arise from this study. First, to what extent does the present study reproduce the feeding pattern and level of intake of SFM and its constituent isoflavenoid phytoestrogens in human infants fed with SFM? Though the discussions below are focused on intake of phytoestrogen isoflavones, it is recognized that other constituents of the SFM may be responsible for the biological effects in our studies. Second, what are the likely consequences of altered neonatal testosterone levels and Leydig cell numbers following feeding with SFM?

Human infants aged 4 months and fed solely with SFM ingest an average of ~4 mg/kg/day of isoflavones (Setchell et al., 1997Go) and are likely to have up to six feeds per day depending on age. Due to manpower and animal welfare considerations, we were unable to consider around the clock feeding in our marmoset study, with the result that 8 h of hand-feeding with SMA or SFM (only 2 h at weekends) and three or four feeds per day (one or two at weekends) was the closest that we could come to matching the pattern and frequency of human infant consumption of SFM. Based on the measured intake of SFM/kg/day and an average measured level of isoflavones in this brand of SFM of 25.5 mg aglucone/l (Ministry of Agriculture, Food and Fisheries, 1998Go), intake of isoflavone per day in our study was probably in the range 1.6–3.5 mg/kg/day per marmoset over the 5–6 weeks of the study period. Based on actual ingestion of isoflavones, our study therefore averages ~40–87% of the intake reported in 4 month human infants fed solely on SFM (Setchell et al., 1997Go). As we have not measured plasma isoflavone levels (such measurements are planned) it is not clear whether this difference in intake is matched by a difference in blood levels but, assuming similar rates of absorption and metabolism of isoflavones in the human and marmoset, our findings would suggest that similar and possibly more marked effects are likely to occur in human infants fed solely with SFM; effects are also likely in infants fed only partially on SFM. This conclusion takes no account of possible differences in responsiveness or sensitivity of marmoset versus human infants to isoflavenoid phytoestrogens, or other components of SFM, though it is generally accepted that marmosets may be relatively steroid-resistant when compared with humans (Lipsett et al., 1985Go).

A prolonged neonatal testosterone rise is a feature unique to male primates; no such event occurs in the female (Forest, 1990Go; Mann and Fraser, 1996Go; Andersson et al., 1998Go). The role of this rise in the male primate remains unclear. Unlike in rodents it is not believed to play a critically important role in masculinization of sexual behaviour, as this is determined during fetal life rather than neonatally as in rodents (Meisel and Sachs, 1994Go). Thus, studies in the rhesus monkey and marmoset in which the neonatal testosterone rise has been blocked by the administration of GnRH analogues have reported relatively minor (Eisler et al., 1993Go) or no (Lunn et al., 1994Go) effects respectively, in adulthood on sexual behaviour, although neonatal castration of the tamarin was reported to have major effects on adult sexual behaviour that could not be rescued by androgen administration (Epple et al., 1990Go). In both the rhesus and marmoset, suppression of the neonatal testosterone rise has no major effect in adulthood on sperm counts and fertility (Mann et al., 1993Go; Lunn et al., 1994Go, 1997Go; Mann and Fraser, 1996Go) though there is a delay or prolongation of the pubertal rise in testosterone levels (Mann et al., 1989Go, 1993Go; Lunn et al., 1994Go; Mann and Fraser, 1996Go) and there are significant changes in testicular cell composition in adulthood, including an increase in Leydig cell numbers (Sharpe et al., 2000Go); perhaps because of these changes there may also be significant decrements in skeletal growth and in bone mineral density (Mann et al., 1993Go). Other effects of neonatal GnRH analogue treatment in male non-human primates have been reported, including subtle effects on the immune system and thymus weight (Mann and Fraser, 1996Go; Lunn et al., 1997Go), and retardation of penis growth (Liu et al., 1991Go; Wallen et al., 1995Go; Brown et al., 1999Go), though the latter may subsequently normalize during puberty (Wallen et al., 1995Go; Brown et al., 1999Go). We did not measure penis length in all marmosets from the present study but, based on measurements in three sets of co-twins, there was no major difference in penis length at 40–45 days of age in SMA- and SFM-fed males (unpublished data). It seems likely that there are other roles for the neonatal testosterone rise, for example on development of the reproductive tract and prostate (Mann and Fraser, 1996Go; McKinnell et al., 2001Go). We are unaware of any published studies that address this possibility but we have such studies in progress.

With regard to the human relevance of these studies in monkeys, the male human infant exhibits a pronounced neonatal testosterone rise between the ages of 1 and 5 months (Forest, 1990Go). Although feeding with SFM may start at any time up to 5–6 months of age when intolerance of cow formula milk is suspected, in many infants SFM is the chosen breastmilk substitute and is fed within the first 4–5 months, and frequently from birth (Polack et al., 1999Go). For human male infants fed with SFM within the first 3–4 months, significant attenuation of the neonatal testosterone rise is likely to occur, based on the present findings. Penile growth in the human male is reported to occur at a higher rate during the neonatal period than at any other phase of life (Danish et al., 1980Go), as is the case in the rhesus monkey (Wallen et al., 1995Go). Disorders of androgen production or action are associated with under-development of the penis in the human male (Danish et al., 1980Go; George and Wilson, 1994Go). Conversely, perinatal exposure to abnormally high androgen levels, as occurs in congenital adrenal hyperplasia, is associated with accelerated development of the penis (Levy and Husmann, 1996Go), though the same authors have questioned whether the growth of the penis neonatally is under testosterone control (Levy et al., 1996Go). There are conflicting views on the bioavailability of testosterone neonatally in the human male as levels of sex hormone-binding globulin (SHBG), to which most testosterone will be bound, are also high at this age (Huhtaniemi et al., 1986Go; Forest, 1990Go); SHBG is also present in the plasma of neonatal marmosets (unpublished data). Isoflavones have been shown to stimulate SHBG production in vitro by (human) HepG2 liver cells (Mousavi and Adlercreutz, 1993Go) and to induce small increases in SHBG in vivo in adult men (Habito et al., 2000Go), changes that would be expected to decrease bioavailability of testosterone. However, a number of studies have also shown that soy isoflavones can themselves bind to SHBG and compete with or displace testosterone, thus increasing bioavailable (= free) testosterone (Martin et al., 1995Go; Dechaud et al., 1999Go). The net effect of these various changes on testosterone bioavailability is difficult to predict without direct measurement in infants. Irrespective of the foregoing uncertainties, the conservation of the neonatal testosterone rise in all primate species studied (Mann and Fraser, 1996Go; Brown et al., 1999Go) suggests strongly that it must play some biological role(s). However, in view of uncertainty about its precise roles, it is difficult to predict whether potential effects of feeding human infants with SFM on the neonatal testosterone rise are likely to result in biologically important changes. Such changes, if they occur, may not necessarily be adverse.

In the absence of data for blood levels of LH, it is difficult to offer a conclusive explanation for the mechanism via which SFM feeding was able to suppress testosterone levels in the infant marmosets. A deficiency in Leydig cells is ruled out (see below) but a reduction in blood LH levels cannot be excluded. Though our studies of immunoexpression of LHß (and FSHß) in the pituitary gland produced no consistent evidence for suppression in SFM-fed co-twins (when compared with their brothers), this finding cannot be interpreted unequivocally as reflecting no change in LH (FSH) secretion. In this regard, s.c. administration of genistein at 2.5 mg/kg/day to adult mice has been shown to reduce both blood testosterone levels and pituitary LH content (Strauss et al., 1998Go), and other studies in adult rats, using a much higher dose of genistein (100 mg/kg/day), have shown almost complete suppression of testosterone levels (Schleicher et al., 1999Go). A recent study in which adult rats were fed for 5 weeks with a diet containing levels of soy phytoestrogens equivalent to those in a traditional Asian diet showed a 50% reduction in mean testosterone levels with no change in LH levels (Weber et al., 2001Go); however, other studies using relatively low levels of isoflavone exposure via the diet had no effect on testosterone levels (Weber et al., 1999Go). Our present studies also suggest that, neonatally, testosterone levels in the marmoset are sensitive to isoflavones in soy or to some other constituent of the SFM. With regard to the mechanism of SFM-induced testosterone suppression, it is relevant that administration of potent estrogens to adult rats also suppresses testosterone production, and this is thought to occur at least partly via suppression of activity of the key steroidogenic enzyme 17{alpha}-hydroxylase/C17–20 lyase in Leydig cells (Cigorraga et al., 1980Go). Suppression of this enzyme in neonatal marmosets exposed to the phytoestrogens in SFM was therefore considered but, based on intensity of immunoexpression, we failed to demonstrate a consistent reduction. In considering such possibilities it should be kept in mind that fetal Leydig cells (to which neonatal Leydig cells may be more akin) can respond differently to hormonal stimuli than do the adult generation of Leydig cells (Huhtaniemi, 1994Go), although the available evidence suggests that suppression of 17{alpha}-hydroxylase/C17–20 lyase in fetal Leydig cells still occurs in the rat after oestrogen administration to the pregnant mother (Majdic et al., 1996Go). Even if this is the mechanism behind the suppression of testosterone levels in SFM-fed infant marmosets, it would fail to explain the increase in Leydig cell numbers (see below) which, in the adult rat, are negatively regulated by potent estrogens (Abney, 1999Go).

The present finding that feeding infant marmosets with SFM resulted in a consistent increase in Leydig cell numbers was unexpected, and even more so when considering that testosterone levels in the same animals were suppressed. We are unable as yet to offer any mechanistic explanation for this finding but similar Leydig cell hyperplasia associated with gross suppression of testosterone levels, and under-masculinization of the fetus, has been reported for the fetal rat testis after exposure in utero to a phthalate ester (Parks et al., 2000Go). It is unclear what the longer-term consequences of this fetal Leydig cell hyperplasia might be. In marmosets in which the neonatal testosterone rise is blocked by GnRH antagonist treatment, Leydig cells atrophy, although their numbers probably do not decrease greatly (Prince et al., 1998Go). These animals show a delayed rise in blood testosterone levels at puberty (Lunn et al., 1994Go) but by adulthood testosterone levels have normalized (Lunn et al., 1994Go; Sharpe et al., 2000Go) and there is an increase in Leydig cell volume per testis (Sharpe et al., 2000Go). It therefore remains debatable whether an increase in Leydig cell numbers neonatally will advance the onset of the pubertal rise in testosterone levels or alter Leydig cell numbers or function in adulthood. Hopefully, our ongoing studies of SFM-fed infant marmosets as they progress through puberty into adulthood will resolve this uncertainty.

The present findings do not show any effect of feeding with SFM on Sertoli and germ cell numbers, so, based on our present understanding of Sertoli cell development in the marmoset and its similarity to the human (Sharpe et al., 2000Go), these findings are reassuring in suggesting that infant feeding with SFM is unlikely to affect sperm counts in adulthood. Nevertheless, it would seem worthwhile to consider comparative studies of sperm counts in adult men who were fed with SFM in infancy in order to confirm that no adverse effects have occurred.

Although our findings demonstrate significant effects of feeding SFM to infant marmosets, we are unable at present to fully interpret these findings or to predict whether the changes observed will lead to long-term consequences. Once our ongoing cohort of SFM-fed animals reach adulthood (in late 2002–2003), some of this uncertainty should be resolved. Clearly, longitudinal prospective studies and retrospective studies in human infants fed with SFM will also be essential to such an evaluation. However, one obstacle to this evaluation is our current lack of understanding of the roles of the neonatal testosterone rise. It will therefore be important to gain better insight into this event in order that longer-term ‘safety’ evaluation can focus on target tissues that are normally influenced by the neonatal rise in testosterone levels. Our present plans are to focus on the reproductive tract, including the developing prostate, and the immune system. In the absence of the foregoing information, it is premature to make recommendations regarding the risk/benefits of feeding SFM to children. Nevertheless, the most cautious interpretation of our findings is that feeding human infant males with SFM in the first 3–5 months of life will exert effects on the neonatal testosterone rise and consequently on tissues/processes that are affected or regulated by this increase. Until we know what these tissues/processes are, it would seem prudent to avoid feeding infants with SFM whenever alternatives are possible.


    Acknowledgements
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
We are grateful to Professor I.Mason for the gift of antisera to 3ß-HSD and 17{alpha}-hydroxylase/C17–20 lyase and to Drs S.Lynch and J.F.Roser for antisera to FSHß and LHß, respectively.


    Notes
 
1 To whom correspondence should be addressed. E-mail: r.sharpe{at}hrsu.mrc.ac.uk Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Abney, T.O. (1999) The potential roles of estrogens in regulating Leydig cell development and function: a review. Steroids, 64, 610–617.[ISI][Medline]

Adlercreutz, H. (1990) Western diet and western diseases: some hormonal and biochemical mechanisms and associations. Scand. J. Clin. Lab. Invest., 50, 3–23.

Adlercreutz, H. and Mazur, W. (1997) Phytoestrogens and Western diseases. Ann. Med., 29, 95–120.[ISI][Medline]

American Academy of Pediatrics Committee on Nutrition (1998) Soy protein-based formulas: recommendations for use in infant feeding (RE9806). Pediatrics, 101, 148–153.[Abstract/Free Full Text]

Andersson, A-M., Toppari, J., Haavisto, A-M., Petersen, J.H., Simell, T. and Skakkebaek, N.E. (1998) Longitudinal reproductive hormone profiles in infants: peak of inhibin B levels in infant boys exceeds levels in adult men. J. Clin. Endocrinol. Metab., 83, 675–681.[Abstract/Free Full Text]

Atanassova, N., McKinnell, C., Walker, M., Turner, K.J., Fisher, J.S., Morley, M., Millar, M.R., Groome, N.P. and Sharpe, R.M. (1999) Permanent effects of neonatal estrogen exposure in rats on reproductive hormone levels, Sertoli cell number and the efficiency of spermatogenesis in adulthood. Endocrinology, 140, 5364–5373.[Abstract/Free Full Text]

Australian College of Paediatrics (1998) Position statement: soy protein formula. J. Paediatr. Child Health, 34, 318–319.[ISI][Medline]

Brown, G.R., Nevsion, C.M., Fraser, H.M. and Dixson, A.F. (1999) Manipulation of postnatal testosterone levels affects phallic and clitoral development in infant rhesus monkeys. Int. J. Androl., 22, 119–128.[ISI][Medline]

Brown, P., McNeilly, J.R., Evans, J.G., Crawford, J.M., Walker, M., Christian, H.C. and McNeilly, A.S. (2001) Manipulating the in vivo mRNA expression profile for FSH beta to resemble that of LH beta does not promote a concomitant increase in intracellular storage of follicle-stimulating hormone. J. Neuroendocrinol., 13, 50–62.[ISI][Medline]

Canadian Paediatric Society, Dieticians of Canada, Health Canada (1998) Statement of the Joint Working Group. Nutrition for healthy term infants.

Cigorraga, S.B., Sorrel, S., Bator, J., Catt, K.J., and Dufau, M.L. (1980) Estrogen dependence of a gonadotropin-induced steroidogenic lesion in rat testicular Leydig cells. J. Clin. Invest., 65, 699–705.[ISI][Medline]

Corker, C.S. and Davidson, D.W. (1981) Radioimmunoassay of testosterone in various biological fluids without chromatography J. Steroid Biochem., 9, 319–323.

Cortes, D., Muller, J. and Skakkebaek N.E. (1987) Proliferation of Sertoli cells during development of the human testis assessed by stereological methods. Int. J. Androl., 10, 589–596.[ISI][Medline]

Danish, R.K., Lee, P.A., Mazur, T., Amrhein, J.A. and Migeon, C.J. (1980) Micropenis. II. Hypogonadotrophic hypogonadism. John Hopkins Med. J., 146, 177–184.[ISI]

Dechaud, H., Ravard, C., Claustrat, F., Brac de la Perriere, A. and Pugeat, M. (1999) Xenoestrogen interaction with human sex hormone-binding globulin (hSHBG). Steroids, 64, 328–334.[ISI][Medline]

Dixson, A.F. (1986) Plasma testosterone concentrations during postnatal development in the male common marmoset. Folia Primatol., 47, 166–170.[ISI][Medline]

Eisler, J.A., Tannenbaum, P.L., Mann, D.R. and Wallen, K. (1993) Neonatal testicular suppression with a GnRH agonist in rhesus monkeys: effects on adult endocrine function and behavior. Horm. Behav., 27, 551–567.[ISI][Medline]

Epple, G., Alveario, M.C. and Belcher, A.M. (1990) Copulatory behavior of adult tamarins (Saguinus fuscicollis) castrated as neonates or juveniles: effect of testosterone treatment. Horm. Behav., 24, 470–483.[ISI][Medline]

Forest, M.G. (1990) Pituitary gonadotropin and sex steroid secretion during the first two years of life. In Grumabch, M.M., Sizonenko, P.C. and Auberrt, M.L. (eds), Control of the Onset of Puberty. Williams & Wilkins, Baltimore, pp. 451–477.

George, F.W. and Wilson, J.D. (1994) Sex determination and differentiation. In Knobil, E. and Neill, J.D. (eds), The Physiology of Reproduction, 2nd edn. Raven Press, New York, pp. 3–28.

Guy, R.A. and Yeh, K.S. (1938) Soybean milk as a food for young infants. Chinese Med. J., 54, 1–30.

Habito, R.C., Montalto, J., Leslie, E. and Ball, M.J. (2000) Effects of replacing meat with soyabean in the diet on sex hormone concentrations in healthy adult males. Br. J. Nutr., 84, 557–563.[ISI][Medline]

Huhtaniemi, I.P. (1994) Fetal testis—a very special endocrine organ. Eur. J. Endocrinol., 130, 25–31.[ISI][Medline]

Huhtaniemi, I.P., Dunkel, L. and Perheentupa, J. (1986) Transient increase in postnatal testicular activity is not revealed by longitudinal measurements of salivary testosterone. Pediatr. Res., 20, 1324–1327.[Abstract]

Irvine, C.H.G., Fitzpatrick, M., Robertson, I. and Woodhams, D. (1995) The potential adverse effects of soybean isofalvones in infant feeding. NZ Med. J., 108, 218.

Irvine, C.H.G., Fitzpatrick, M.G. and Alexander, S.L. (1998a) Phytoestrogens in soy-based infant foods: concentrations, daily intake and possible biological effects. Proc. Soc. Exp. Biol. Med., 217, 247–253.[Abstract]

Irvine, C.H.G, Shand, N., Fitzpatrick, M.G. and Alexander, S.L. (1998b) Daily intake and urinary excretion of genistein and daidzein by infants fed soy- or dairy-based infant formulas. Am. J. Clin. Nutr., 68 (Suppl. 6), 1462S–1465S.[Abstract]

Jung, A.J. and Carr, S.L. (1977) Infant feeding: a soy protein formula and a milk-based formula. Clin. Pediatr., 16, 982–986.[ISI][Medline]

Klein, K.O. (1998) Isoflavones, soy-based infant formulas and relevance to endocrine function. Nutr. Rev., 56, 193–204.[ISI][Medline]

Knight, D.C., Eden, J.A., Huang, J.L. and Waring, M.A. (1998) Isoflavone content of infant foods and formulas. J. Paediatr. Child Health, 34, 135–138.[ISI][Medline]

Kohler, L., Meeuwise, G. and Mortensson, W. (1984) Food intake and growth of infants between six and twenty-six weeks of age on breast milk, cow's milk formula or soy formula. Acta Paediatr. Scand., 73, 40–48.[ISI][Medline]

Lasekan, J.B., Ostrom, K.M. Jacobs, J.R., Blatter, M.M., Ndife, L.I., Gooch, W.M., III and Cho, S. (1999) Growth of newborn, term infants fed soy formulas for 1 year. Clin. Pediatr., 38, 563–571.[ISI][Medline]

Levy, J.B. and Husmann, D.A. (1996) Congenital adrenal hyperplasia: is there an effect on penile growth? J. Urol., 156, 780–782.[ISI][Medline]

Levy, J.B., Seay, T.M., Tindall, D.J. and Husmann, D.A. (1996) The effects of androgen administration on phallic androgen receptor expression. J. Urol., 156, 775–779.[ISI][Medline]

Lipsett, M.B., Chrousos, G.P., Tomita, M., Brandon, D.D. and Loriaux, D.L. (1985) The defective glucocorticoid receptor in man and non-human primates. Recent Progr. Horm. Res., 41, 199–241.[Medline]

Liu, L., Cristiano, A.M., Southers, J.L., Reynolds, J.C., Bacher, J., Brown, G., Gilley, R.M., Tice, T.R., Banks, S.M., Loriaux, L.D. and Cassorla, F. (1991) Effects of pituitary–testicular axis suppression in utero and during the neonatal period with a long-acting luteinizing hormone-releasing hormone analog on genital development, somatic growth and bone density in male cynomolgus monkeys in the first 6 months of life. J. Clin. Endocrinol. Metab., 73, 1038–1043.[Abstract]

Lunn, S.F., Recio, R., Morris, K. and Fraser, H.M. (1994) Blockade of the neonatal rise in testosterone by a gonadotrophin-releasing hormone antagonist: effects on timing of puberty and sexual behaviour in the male marmoset monkey. J. Endocrinol., 141, 439–447.[Abstract]

Lunn, S.F., Cowen, G.M. and Fraser, H.M. (1997) Blockade of the neonatal rise in testosterone by a GnRH antagonist: the free androgen index, reproductive capacity and postmortem findings in the male marmoset monkey. J. Endocrinol., 154, 125–131.[Abstract]

Ministry of Agriculture, Food and Fisheries (1998) Food surveillance information sheet 167—plant oestrogens in soya-based infant formulae.

Majdic, G., Sharpe, R.M., O'Shaughnessy, P.J. and Saunders, P.T.K. (1996) Expression of cytochrome P450 17{alpha}-hydroxylase/C17–20 lyase (P450c17) in the fetal rat testis is reduced by maternal exposure to exogenous estrogens. Endocrinology 137, 1063–1070.

Mann, D.R. and Fraser, H.M. (1996) The neonatal period: a critical interval in primate development. J. Endocrinol. 149, 191–197.

Mann, D.R., Gould, K.G., Collins, D.C. and Wallen, K. (1989) Blockade of neonatal activation of the pituitary–testicular axis: effect on peripubertal luteinizing hormone and testosterone secretion and on testicular development in male monkeys. J. Clin. Endocrinol. Metab. 68, 600–607.

Mann, D.R., Akinbami, M.A., Gould, K.G., Tanner, J.M. and Wallen, K. (1993) Neonatal treatment of male monkeys with a gonadotropin-releasing hormone agonist alters differentiation of central nervous system centers that regulate sexual and skeletal development. J. Clin. Endocrinol. Metab. 76, 1319–1324.

Martin, M.E., Haourigui, M., Pelissero, C., Benassayag, C. and Nunez, E.A. (1995) Interactions between phytoestrogens and human sex steroid binding protein. Life Sci., 58, 429–436.[ISI]

McKinnell, C., Saunders, P.T.K., Fraser, H.M., Kelnar, C.J.H., Kivlin, C., Morris, K.D. and Sharpe, R.M. (2001) Comparison of androgen receptor (AR) and oestrogen receptor-ß immunoexpression in the testes of marmosets from birth to adulthood: low AR immunoexpression in Sertoli cells during the neonatal testosterone rise. Reproduction, 122, 419–429.[Abstract/Free Full Text]

Meisel, R.L. and Sachs, B.D. (1994) The physiology of male sexual behavior. In Knobil, E. and Neill, J.D. (eds), The Physiology of Reproduction, 2nd edn. Raven Press, New York, pp. 3–106.

Mimouni, F., Campaigne, B., Neylan, M. and Tsang, R.C. (1993) Bone mineralization in the first year of life in infants fed human milk, cow-milk formula or soy-based formula. J. Pediatr., 122, 348–354.[ISI][Medline]

Mousavi, Y. and Adlercreutz, H. (1993) Genistein is an effective stimulator of sex hormone-binding globulin production in hepatocarcinoma human liver cancer cells and suppresses proliferation of these cells in culture. Steroids, 58, 301–304.[ISI][Medline]

Parks, L.G., Ostby, J.S., Lambright, C.R., Abbott, B.D., Klinefelter, G.R., Barlow, N.J. and Gray, L.E., Jr (2000) The plasticizer diethylhexyl phthalate induces malformations by decreasing fetal testosterone synthesis during sexual differentiation in the male rat. Toxicol. Sci., 58, 339–349.[Abstract/Free Full Text]

Polack, F.P., Khan, N. and Maisels, M.J. (1999) Changing partners: the dance of infant formula changes. Clin. Pediatr., 38, 703–708.[ISI][Medline]

Prince, F.P., Mann, D.R. and Fraser, H.M. (1998) Blockade of the hypothalamic–pituitary–testicular axis with a GnRH antagonist in the neonatal marmoset monkey: changes in Leydig cell ultrastructure. Tiss. Cell, 30, 651–661.[ISI]

Schleicher, R.L., Lamartiniere, C.A., Zheng, M. and Zhang, M. (1999) The inhibitory effect of genistein on the growth and metastasis of a transplantable rat accessory sex gland carcinoma. Cancer Lett., 136, 195–201.[ISI][Medline]

Setchell, K.D.R., Zimmer-Nechemias, L., Cai, J. and Heubi, J.E. (1997) Exposure of infants to phytoestrogens from soy-based infant formula. Lancet, 350, 23–27.[ISI][Medline]

Setchell, K.D.R., Zimmer-Nechemias, L., Cai, J. and Heubi, J.E. (1998) Isoflavone content of infant formulas and the metabolic fate of these phyotestrogens in early life. Am. J. Clin. Nutr., 68 Suppl 6, 1453S–1461S.[Abstract]

Sharpe, R.M. (1999) Fetal/neonatal hormones and reproductive function of the male in adulthood. In O'Brien, P.M.S., Wheeler, T. and Barker, D.J.P. (eds), Fetal Programming: Influences on Development and Disease in Later Life. Royal College of Obstetricians and Gynaecologists Press, London, pp. 187–194.

Sharpe, R.M., Walker, M., Millar, M.R., Morris, K., McKinnell, C., Saunders, P.T.K. and Fraser, H.M. (2000) Effect of neonatal GnRH antagonist administration on Sertoli cell number and testicular development in the marmoset: comparison with the rat. Biol. Reprod., 62, 1685–1693.[Abstract/Free Full Text]

Strauss, L., Makela, S., Joshi, S., Huhtaniemi, I.P. and Santti, R. (1998) Genistein exerts estrogen-like effects in male mouse reproductive tract. Mol. Cell. Endocrinol., 144, 83–93.[ISI][Medline]

Strom, B.L., Schinnar, R., Ziegler, E.E., Barnhart, K.T., Sammell, M.D., Macones, G.A., Stallings, V.A., Drulis, J.M., Nelson, S.E. and Hanson, S.A. (2001) Exposure to soy-based formula in infancy and endocrinological and reproductive outcomes in young adulthood. J. Am. Med. Assoc., 286, 807–814.[Abstract/Free Full Text]

Toppari, J., Larsen, J.C., Christiansen, P., Giwercman, A., Grandjean, P., Guillette, L.J. Jr., Jégou, B., Jensen, T.K., Jouannet, P., Keiding, N. et al. (1996) Male reproductive health and environmental xenoestrogens. Environ. Health Perspect., 104 (Suppl. 4), 741–803.[ISI][Medline]

UK Department of Health (1996) Soy based formula (96/244).

Wallen, K., Maestripieri, D. and Mann, D.R. (1995) Effects of neonatal testicular suppression with a GnRH antagonist on social behavior in group-living juvenile rhesus monkeys. Horm. Behav., 29, 322–337.[ISI][Medline]

Weber, K.S., Jacobson, N.A., Setchell, K.D. and Lephart, E.D. (1999) Brain aromatase and 5alpha-reductase, regulatory behaviors and testosterone levels in adult rats on phytoestrogen diets. Proc. Soc. Exp. Biol. Med., 221, 131–135.[Abstract]

Weber, K.S., Setchell, K.D.R., Stocco, D.M. and Lephart, E.D. (2001) Dietary phytoestrogens decrease testosterone levels and prostate weight without altering LH, prostate 5{alpha}-reductase or testicular steroidogenic acute regulatory peptide levels in adult male Sprague–Dawley rats. J. Endocrinol., 170, 591–599.[Abstract/Free Full Text]

Whitten, P.L. and Naftolin, F. (1999) Reproductive actions of phytoestrogens. Baillières Clin. Endocrinol. Metab., 12, 667–690.[ISI]

Whitten, P.L., Lewis, C., Russell, E. and Naftolin, F. (1995) Potential adverse effects of phytoestrogens. J. Nutr., 125, 771S–776S.[Medline]

Williams, K., McKinnell, C., Saunders, P.T.K., Walker, M., Fisher, J.S., Turner, K.J., Atanassova, N. and Sharpe, R.M. (2001) Neonatal exposure to potent and environmental oestrogens and abnormalities of the male reproductive system in the rat: evidence for importance of the androgen:oestrogen balance and assessment of the relevance to man. Hum. Reprod. Update, 7, 236–247.[Abstract/Free Full Text]

Winters, J.S.D., Faiman, C., Hobson, W.C., Prasad, A.V. and Reyes, F.I. (1975) Pituitary–gonadal relations in infancy. I. Patterns of serum gonadotropin concentrations from birth to four years of age in man and chimpanzee. J. Clin. Endocrinol. Metab., 40, 545–551.[Abstract]

Wreford, N.G. (1995) Theory and practice of stereological techniques applied to the estimation of cell number and nuclear volume in the testis. Microsc. Res. Tech., 32, 423–436.[ISI][Medline]

Submitted on January 9, 2002; accepted on March 20, 2002.