Am J Cliii Nutr 1994:60:333-40. Printed in USA. © 1994 American Society for Clinical Nutrition 333

Biological effects of a diet of soy protein rich in isoflavoneson the menstrual cycle of premenopausal women13

Aedin Cassidv, Sheila Bingham, and Kenneth DR Setchell

ABSTRACT The influence of a diet containing soy protein

on the hormonal status and regulation of the menstrual cycle was

examined in six premenopausal women with regular ovulatory

cycles. Soy protein (60 g containing 45 mg isoflavones) given

daily for 1 mo significantly (P < 0.01) increased follicular phase

length and/or delayed menstruation. Midcycle surges of lutein-

izing hormone and follicle-stimulating hormone were signifi-

cantly suppressed during dietary intervention with soy protein.

Plasma estradiol concentrations increased in the follicular phase

and cholesterol concentrations decreased 9.6%. Similar re-

sponses occur with tamoxifen, an antiestrogen undergoing din-

ical trial as a prophylactic agent in women at high risk for breast

cancer. These effects are presumed to be due to nonsteroidal es-

trogens of the isoflavone class, which behave as partial estrogen

agonists/antagonists. The responses to soy protein are potentially

beneficial with respect to risk factors for breast cancer and may

in part explain the low incidence of breast cancer and its corre-

lation with a high soy intake in Japanese and Chinese

women. Am J Clin Nutr l994;60:333-40.

I(EY WORDS Soy, hormonal status, menstrual cycle, ta-

moxifen, breast cancer, isoflavones

Introduction

Breast cancer is the second most common cancer in Western

countries but its incidence is significantly less in Third World

and Asian populations. Age-specific rates for breast cancer in

England and Wales are 199.4 per 100 000 in women 60-65 y of

age compared with 52.3 per 100 000 for women of similar age

in Japan (I ). Epidemiological data from migrant studies suggest

that in most cases the susceptibility to breast cancer is the result

of environmental rather than genetic differences between these

populations and that diet is a major contributing factor (2). The

largest and most carefully controlled prospective study of diet

and breast cancer, however, failed to provide conclusive evidence

to support an association between relative risk and a diet high in

fat (3). Furthermore, although elevated free estrogen concentra-

tions have been associated with increased risk (4), there have

been few adequately controlled studies of the influence of fat on

hormonal status.

Apart from fat, there are many other differences between the

typical diets of Far Eastern and Western populations (5). In the

Far East a significant quantity of soybean protein is consumed in

many different forms, including beans, miso, tofu, and soy milk.

Soy is a rich source of nonsteroidal estrogens of the isoflavone

class (6). These compounds, which are structurally similar to

estrogens, bind to the estrogen receptor and behave as partial

estrogen antagonists (7). For this reason we previously suggested

that a diet of soy protein may be beneficial in the protection

against and/or treatment of breast cancer (8). This hypothesis is

strengthened by recent studies, which have shown that a diet of

soy protein leads to a significant dose-dependent reduction in

mammary tumor growth in two animal models of chemically

induced mammary carcinoma (9). In these studies, tumor for-

mation was negatively correlated with total dietary isoflavone

concentration, and in particular with the dietary intake of geni-

stein and the urinary isoflavone excretion (9).

Because little is known about the biological and physiological

effects of dietary estrogens in humans and in view of well-doc-

umented examples of the biological potency of isoflavone inges-

tion by animals (10- 12), we investigated in a controlled dietary-

intervention study the influence of soy-protein intake on hor-

monal status of premenopausal women. We hypothesized that a

constant diet of soy protein containing isoflavones would lead to

significant modifications to the hormonal status of the menstrual

cycle and that these changes would be beneficial with regard to

risk factors for breast cancer.

Subjects, materials, and methods

Study protocol

Six healthy nonvegetarian women (21 -29 y of age) were en-

rolled in the study. The protocol was approved by the Dunn Nu-

trition Unit Ethical Committee. All women had normal menstrual

cycles and had taken no medication for � 6 mo before starting

‘ From the Dunn Clinical Nutrition Centre, Cambridge, UK. and theDivision of Clinical Mass Spectrometry. Department of Pediatrics, Chil-

dren’s Hospital Medical Center, Cincinnati.

2 Supported in part by the Ministry of Agriculture. Fisheries and Foods

and the National Institutes of Health, National Cancer Institute, grant

ROI -CA56302-Ol.

3 Address reprint requests to KDR Setchell, Division of Clinical

Mass Spectrometry. Department of Pediatrics, Children’s Hospital

Medical Center, 3333 Burnet Avenue, Cincinnati, OH 45229. or

A Cassidy, Dunn Nutrition Center. 100 Tennis Court Road,

Cambridge, UK, CB2 IQL.Received October 5, 1992.Accepted for publication January 14. 1994.

at PE

NN

SY

LVA

NIA

ST

AT

E U

NIV

PA

TE

RN

O LIB

RA

RY

on February 23, 2013

ajcn.nutrition.orgD

ownloaded from

334 CASSIDY ET AL

‘�± SD.

to record menses data. For 4 mo before the study began. and for

several months after completion, each subject monitored basal

body temperature at the same time each morning to determine

cycle length, regularity, and ovulatory status. The women lived

in the metabolic suite at the Dunn Clinical Nutrition Centre for

a total of 4-6 mo. During two of these months, diets and activ-

ities were closely monitored and controlled. The time between

the control diet and soy diet was separated by a I -4-mo period

because the soy diet had to commence on the first day of a men-

strual cycle. Basal metabolic rate was estimated from body

weight (13), and the energy intake necessary to maintain a con-

stant body weight throughout the study was calculated, assuming

a ratio of total energy intake to basal metabolic rate of 1.4.

The first complete menstrual cycle served as a control period

during which time each subject consumed a constant daily diet

of nonsoy-containing foods provided by the metabolic kitchen.

During the second month, and starting on the first day of menses

(which was defined as the first day of menstrual bleeding), ap-

propriate modifications were made to the basal diet to maintain

amounts of macronutrients and nonstarch polysaccharide with

the addition of 60 g/d (dry wt) soy protein/d (Protoveg; Direct

Foods Ltd. Manchester, UK) to meals. All meals were prepared

in advance, accurately weighed, and deep frozen until required.

All frozen and canned foods were of the same batch to minimize

interbatch variability, and bread that contained no soy flour was

specially prepared for the study. The same batch of soy protein

was used throughout the study period.

Collection of biological samples

Every 3 d during each diet period, a fasting blood sample was

collected between 0730 and 0830 (10 mL) and a 24-h urine sam-

pIe was obtained. Throughout the two diet periods, all fecal sam-

pIes were collected to measure intestinal transit time. Urine vol-

umes were recorded and aliquots taken and stored at -20 #{176}C.

Completeness of the 24-h urine collection was assessed by using

a previously validated PABA check method (14). Early morning

urine samples were collected on the other 2 d and were used to

define the day of ovulation with the First Response ovulation

prediction test kits (Carter-Wallace Ltd, Folkestone, Kent, UK).

Blood samples were collected into heparinized tubes and im-

mediately centrifuged at 3000 X g for 10 mm at room tempera-

ture. Plasma was separated and stored at -20 #{176}Cbefore analysis.

Transit time and completeness of fecal collections were assessed

continuously throughout the study by using a radiopaque-marker

technique ( 15). The subjects consumed 10 radiopaque plastic pel-

lets with each meal, three times each day. Stools were collected

from the time the first marker was taken until all markers were

recovered. Each stool was weighed and x rayed to determine the

recovery of markers.

Analytical methods

Duplicate diets were collected at the beginning and end of each

diet period for each of the 3 d of the rotation diet. Six diets were

therefore analyzed per subject for each diet period. Samples were

freeze-dried and analyzed for nitrogen content by using the

Kieldahl technique (Tecator Kjeltec System 1002, Bristol, UK).

Plant estrogen concentrations were determined in the diets by

using previously described HPLC-mass spectrometry (MS) tech-

niques (6, 16).

Urinary isollavone concentrations were determined in urine by

using previously published gas chromatography (GC)-MS tech-

niques (8, 17, 18). The following plasma hormone concentrations

were determined by using commercially available radioimmu-

noassays: sex-hormone-binding globulin (SHBG) (Pharmacia,

Milton Keynes, UK), progesterone, testosterone, estradiol, lu-

teinizing hormone (LH), and follicle-stimulating hormone (FSH),

all obtained from Diagnostic Products Ltd (Abingdon, Oxon,

UK). The plasma cholesterol concentration was measured by us-

ing the Cobas-Bio automatic centrifugal analyzer (Roche Diag-

nostics, Welwyn Garden City, UK).

Two sets of internal quality-control samples were used: trilevel

immunoassay control serum (Lyphocheck; Biorad Laboratories,

Hemel Hempstead, UK) and trilevel control serum (Diagnostics

Product Ltd). Interbatch variability was avoided by assaying all

samples from each subject in a single batch and samples were

assayed in duplicate.

The cross-reactivity of the dietary estrogens daidzein and gen-

istein in the estradiol radioimmunoassay was determined from

solutions of the pure precursors of daidzein and genistein in the

concentration range 40 pmolfL-4 mmolfL (10 pg/mL- I mgi

mL). Over this range there was negligible cross-reactivity.

Statistical analysis

Statistical analysis of the data was carried out on an Apple

Macintosh LC computer by using the Svstat 5. 1 program (Systat

Inc. Evanston, IL). All results are expressed as mean ± SD, and

differences were assessed by using paired t tests. Pearson’s cor-

relation coefficients were used to assess associations between

variables. The CV (%) was used to assess the intra- and interassay

variability.

Results

The demographics of the six women participating in the study

are given in Table 1. Body weight was monitored during the two

diet periods and did not change significantly during the study.

Mean body weight was 61.59 ± 5.28 kg at the beginning of the

study and 61.34 ± 5.45 kg at the end of the study. Nutrient

intakes during the control- and soy-diet periods are compared in

Table 2. There were no significant differences in calculated nu-

trient intake or dietary nitrogen content between the two diet

periods. The total dietary isoflavone content of the control diets

TABLE IDemographics of the six premenopausal women studied

Subject Age

Energy

intake Height

Weight

before study

Weight

after study

y Mild in kg kg

I 25 8.5 1.68 62.58 ± 0.1 1’ 61.34 ± 0.23

2 24 9.0 1.75 63.57 ± 0.1 1 64.54 ± 0.25

3 21 10.0 1.64 69.30 ± 0.26 69.54 ± 0.274 21 8.0 1.60 53.03 ± 0.24 51.60 ± 0.29

S 26 8.0 1.68 56.84 ± 0.39 56.03 ± 0.21

6 29 8.5 1.70 64.23 ± 0.19 64.66 ± 0.17

at PE

NN

SY

LVA

NIA

ST

AT

E U

NIV

PA

TE

RN

O LIB

RA

RY

on February 23, 2013

ajcn.nutrition.orgD

ownloaded from

SOY ISOFLAVONES AND THE MENSTRUAL CYCLE 335

TABLE 2Nutrient intake during the two diet periods’

Nutrient Control diet Soy diet

CalculatedEnergy (MI) 8.5 ± 0.1 8.5 ± 0.3

Protein (g) 97.2 ± 0.6 98.7 ± 0.8Totalfat(g) 61.4±0.7 60.0± 1.7Carbohydrate (g) 258.2 ± I .9 260.3 ± I .0Nonstarch polysaccharide (g) 16.0 ± 0.6 16.3 ± 0.7

Starch (g) 108.3 ± 2.2 106.7 ± 2.7

AnalyzedNitrogen (j.tg/g) 2.09 ± 0.03 2.10 ± 0.06Daidzein (mg) 0.76 ± 0.03 25.08 ± 0.312

Genistein (mg) 0.49 ± 0.03 19.85 ± 0.432

‘1± SD.2 Significantly different from control diet, P < 0.0001.

measured from triplicate samples was significantly lower ( I .26

± 0.02 mg/d) than that of the soy diet (44.93 ± 0.21 mg/d).

Menstrual cycle lengths before, during, and for the 3 mo after

the study are summarized in Table 3. Subjects 1, 2, 3, and 4 had

menstrual cycles of relatively constant length before commenc-

ing the dietary study (CV ranged from 0% to 2.2%). In subjects

5 and 6 the menstrual cycle length over the 4-mo period before

the control period was more variable but these women had pre-

viously used oral contraceptives.

There was no significant difference in the average menstrual

cycle length between the control period (27 ± 2 d) and the 4-mo

period before the study (28 ± 3 d). However, during the period

of the soy diet, menstruation was delayed by I -5 d in five of the

six subjects (Table 3). Subject 5 had a shorter cycle on the soy

diet, but nevertheless the length of the follicular phase increased.

The average length of the follicular phase length was 15 ± 0.9

d during the control-diet period and after the introduction of soy,

the mean follicular phase length increased (17.5 ± 2.3 d). The

average change in follicular phase length was 2.5 ± I .6 d, which

was statistically significant (P < 0.01).

TABLE 3

Despite the increase in follicular phase length and total cycle

length, no change in the length of the luteal phase was observed

during soy consumption (Table 3). There was a trend toward a

shorter menstrual cycle in the first month after dietary interven-

tion with soy in four of the six subjects, but this was not statis-

tically significant. Within 3 mo, menstrual cycle length had re-

turned to prestudy values in all six subjects (Table 3).

Table 4 summarizes the urinary excretion of the isoflavones

daidzein, genistein, and equol for the six subjects during the

period of soy intake. Total urinary isoflavone excretion (5.6-

67.3 nmol/d; I .4- 17. 1 j�g/d) was low during the control pe-

nod but after 60 g soy/d over a complete menstrual cycle, a

1000-fold increase in total urinary isoflavone excretion oc-

curred for all subjects, with values ranging from 1400 to

29 400 nmol/d (from 0.35 to 7.49 mg/d) (Table 4). From the

estimated dietary intake of44.9 mg total isoflavones/d, urinary

excretion accounted for I .8- 12.9% of the total intake. Only

subjects 1 and 2 excreted high amounts of equol during the

soy-diet period. Subject 5 excreted considerably lower

amounts of equol, and only traces of equol were detected in

the other three subjects. The highest urinary excretion of daid-

zein was observed in the four subjects excreting low or neg-

ligible concentrations of its specific metabolite equol, and

daidzein excretion was quantitatively more important than

was genistein in these four subjects (Table 4). Conversely, the

two subjects excreting substantial amounts of equol excreted

low concentrations of the precursors daidzein and genistein.

Plasma hormones

The plasma concentrations of progesterone, estradiol, testos-

terone, SHBG, and cholesterol at times before and during soy-

protein intake are summarized in Table 5. Table 6 shows the

mean plasma LH and FSH concentrations at the time of ovula-

tion, determined from the samples obtained on the day of ovu-

lation and I d either side of ovulation as defined by the First

Response assay kit for predicting ovulation from the early morn-

ing and 24-h urine samples.

The midcycle peaks of LH and FSH were significantly sup-

pressed by soy-protein ingestion (P < 0.05 and P < 0.01 , re-

Length of menstrual cycle and follicul ar phase over a 4-mo period before the stu dy, during the study pe riods, a nd 3 mo after the soy diet

Menstrual cycle length

Follicular phase lengthBefore study’ After soy diet2

I ± SD CVSubject I ± SD CV Control diet Soy diet Control diet Soy diet

d % d d d % d d

I2

3456

1± SD

28.25 ± 0.50

26.50 ± 0.58

27.75 ± 0.50

28.00 ± 0.00

35.00 ± 4.83

25.00 ± 1.41

28.42 ± 3.45

1.8

2.2

1.8

0.013.8

5.6-

29 32

25 30

27 29

28 2931 28

25 26

27.50 ± 2.35 29.00 ± 2.00

24.33 ± 2.31

29.00 ± 2.65

26.33 ± 1.53

28.67 ± 1.1633.67 ± 4.62

24.67 ± 1.16

27.78 ± 3.48

9.5

9.1

5.8

4.113.7

4.7

-

16 21

16 19

15 15

14 1615 18

14 lb

15.00 ± 0.90 17.50 ± 2.30�’

‘ a =4 mo.

2 ,� 3 mo.

3 Significantly different from control diet, P < 0.01 (1 =3.73).

at PE

NN

SY

LVA

NIA

ST

AT

E U

NIV

PA

TE

RN

O LIB

RA

RY

on February 23, 2013

ajcn.nutrition.orgD

ownloaded from

336 CASSIDY ET AL

‘ .� :6 SD. SHBG, sex-hormone-binding globulin.2 � Significantly different from control diet: 2 p < 0.02, ‘ P < 0.05.

!.2 Significantly different from control diet: ‘ P < 0.05 U =2.69), 2 p

< 0.02 (i =4.01).

TABLE 4

Urinary excretion of isoflavones during the period of dietary intervention with soy protein’

Subject Daidzein Genistein EquolTotal isoflavone

excretionPercent of isoflavones

excreted2

p�iioL/d /J?flOIld pniolld pniolld e/

I 0.79 ± 0.79 0.74 ± 0.74 12.81 � 4.55 13.33 � 4.71

(0.24-2.76)’ (0. 12-2.96) (6.6 I -23. I 4) (7.45-22.7) 7.6

2 0.79 :6 0.39 0.3 ± 0.15 6.61 ± 3.30 8.2 � 3.14

(0.28- I . I 8) (0. 1 I -0.37) (2.89- 12.0) (3.92- I 2.54) 4.73 15.35 ± 3.93 8.51 ± 2.96 0.01 ± 0.01 22.74 ± 4.70

(11.8-24.0) (5.55-13.33) W.00I-0.41) (17.64-29.41) 12.9

4 2.76 :1: 1.97 0.30 ± 0.15 0.02 ± 0.02 3.14 ± 1.96

( I . I 8-7.09) (0.07-0.70) (0.0 1-0.()4) ( I .57-7.05) I .8

5 10.63 :5 8.27 1.48 ± 1.18 1.24 ± 0.82 12.16 ± 7.06

(2.76-26.77) (0.30-3.33) (0. I 2-2.48) (3. I 4-24.3 I ) 6.9

6 13.0 ± 5.12 2.59 ± 1.48 ND 17.25 ± 1.96

(7.48-24.4 1 ) (0.33-4.44) ( I 6.07- 18.82) 9.8

‘ Urinary isoflavone excretion during the control period ranged from 5.6 to 67.3 nmol/d. ND. not detected.

1 Expressed as % of total daily dietary isoflavone intake (44.93 ± 0.21 mg/d).

‘ .(� ± SD: range in parentheses.

spectively). Figure 1 illustrates mean plasma concentrations of

LH and FSH for the 4 d either side of ovulation, where day 0 is

assigned the day of ovulation. As a group there was a threefold

reduction in peak plasma LH concentration and twofold reduc-

tion in FSH during the period that soy protein was ingested.

TABLES

Plasma hormone concentrations during the control peroid and duringintervention with soy protein’

Control diet Soy diet

ProgesteroneLuteal phase

(nmol/L) 21.4 ± 6.0 18.4 ± 4.8

(ng/mL) 6.72 ± 1.89 5.79 ± ISO

EstradiolFollicular phase

(pmol/L) 246.2 ± 38.4 362.5 ± 82.02

(pg/mL) 66.96 ± 10.44 98.6() ± 22.312

Midcycle(pmol/L) 520.6 ± 153.5 572.0 ± 180.20

(pg/mL) 141.61 ± 41.75 155.72 ± 49.1

Luteal phase(pmol/L) 416.5 ± I 10.6 415.0 ± 170.3

(pg/mL) I 13.30 ± 37.08 1 12.93 ± 46.33

SHBG (nmol/L)

Follicular phase 58.57 ± 13.61 57.54 ± 9.39Luteal phase 58.33 ± 1.25 55.32 ± 9.96

Entire menstrual cycle 58.48 ± 14.6 56.98 ± 9.3

TestosteroneEntire menstrual cycle

(nmol/L) 1.25 ± 0.38 1.46 ± 0.52

(ng/mL) 0.36 ± 0. 1 1 0.42 ± 0.1S

Cholesterol (mmol/L)Entire menstrual cycle 4.27 ± 1.08 3.86 ± l.Ol�

Mean plasma progesterone concentrations during the luteal

phase of the menstrual cycle were within the normal range of the

assay (8.0-89.2 nmolfL; 2.5-28.0 ng/mL) and were not signif-

icantly different between the control and soy-diet periods (Table

5). Compared with the control period, the maximum plasma pro-

gesterone concentration during the soy-diet period occurred later

in the cycle in all six subjects (Fig 2).

There was no change in the concentration of SHBG during the

two dietary periods and no significant difference in SHBG con-

centration between the follicular phase and luteal phase of the

menstrual cycle for either diet period. In addition, the soy diet

had no effect on SHBG concentration (Table 5). Although therewere individual variations in plasma testosterone concentrations

during the two diet periods, mean values for all subjects com-

pared over a complete menstrual cycle for the two diet periods

did not differ significantly (Table 5).

TABLE 6

Midcycle plasma concentrations of luteinizing hormone andfollicle-stimulating hormone on the days surrounding ovulation

(days -I to +1)

Subject

Luteinizing hormone

Follicle-stimulatinghormone

Control diet Soy diet Control diet Soy diet

(IlL

I

2

3

45

6.t.± SD

9.4

4.3

26.0

21.022.5

44.021.2 ± 12.7

2.9

5.4

9.3

10.87.0

7.47.1 ± 2.6’

10.8 4.8

4.3 2.9

21.0 17.3

17.4 7.3

15.3 8.5

18.8 6.014.6 ± 5.6 7.8 ± 4.62

at PE

NN

SY

LVA

NIA

ST

AT

E U

NIV

PA

TE

RN

O LIB

RA

RY

on February 23, 2013

ajcn.nutrition.orgD

ownloaded from

-4’3-2-1 0 1 2 3 4 “-4-3-2.101234

Day of ovulation Day of ovutation

0

EC

a)C

2tI)

a)0)2Q.EI-

a)U)

C’,a)�1

C3

#{149}00

CoCDC/)a)3a)

:�CO

3C

0 7 14 21

Day of cycle28

SOY ISOFLAVONES AND THE MENSTRUAL CYCLE 337

FIG 2. Mean (±SD) progesterone concentration during the control-(0) and soy-diet (U) periods.

FIG I. Mean (±SD) group luteinizing hormone and follicle-stimulat-ing hormone concentrations in the control- (0) and soy-diet (U) periods.

Plasma estradiol concentrations in the follicular phase, at mid-

cycle (the 3 d surrounding ovulation), and in the luteal phase are

compared in Table 5. Concentrations of estradiol were signifi-

cantly higher in the follicular phase of the menstrual cycle (P

< 0.02) after exposure to isoflavones. However, there were no

significant differences in plasma estradiol concentrations mid-

cycle or during the luteal phase between the two diet periods (Ta-

ble 5).

Plasma cholesterol concentrations (Table 5) were 4.3 ± 1.1

mmol/L (1 ± SD) in the control-diet period and decreased 9.6%

during the period of soy intake (3.9 ± I .0 mmoIIL). This change

was statistically significant (P < 0.05).

Transit time

Each subject was given 2100 radiopaque markers over the

course of the dietary-intervention study, and marker recovery

ranged from 99% to 100%, confirming that all of the subjects

had provided complete fecal collections during the study. Soy

had no significant effect on gut transit time and transit time did

not significantly change during the menstrual cycle. Mean transit

time during the follicular phase was 64.2 ± 24.2 h during the

soy-diet period and 62.2 ± 34.4 h during the luteal phase of the

soy-diet period.

cancer and other common Western diseases is low, they are less

frequently ingested by Western populations. It has therefore been

proposed that the presence of dietary estrogens may contribute

to differences in incidence of diseases such as breast cancer be-

tween Eastern and Western populations (8, 21 -23). Evidence

substantiating this contention includes the recent epidemiological

data indicating that a high soy intake is associated with a de-

creased risk for breast cancer in premenopausal women from

China and Singapore (24) and the earlier studies showing that

soy protein will reduce the number of tumors, in a dose-depen-

dent manner, in animal models of chemically induced mammary

carcinoma (9). There also appear to be differences in hormonal

status and menstrual cycle length in Japanese women (25) corn-

pared with Western women (26-29) and it is possible that these

differences could be accounted for by a high dietary estrogen

intake.

To our knowledge this is the first report that demonstrates

physiological effects of dietary estrogen intake in humans. Six

premenopausal women were studied over a 9-mo period, in

which two menstrual cycles were spent on controlled diets. Men-

strual cycle length appeared to be unaffected by the move to the

metabolic suite because there was no significant change in cycle

length before the study began, and during the control-diet period

(Table 3).

The amount of textured vegetable (soy) protein (60 g/d) used

in the study exposed each individual to a daily intake of 45 mg

isoflavones. This is a relatively modest amount of dietary estro-

gen compared with the amounts generally found in soy-protein

foods typically consumed in the Far East (9, 30, 3 1 ). Separate

studies we carried out indicate that all soy-derived foods contain

isoflavones (30), and on the basis of typical intake we estimate

that the Japanese consume 150-200 mg isoflavones/d (9, 30, 31).

The effectiveness of 45 rng soy isoflavones/d raises the issue of

whether a greater effect would occur with higher doses, but the

complexity of the study precluded examination of dose-response

effects. The order of presentation of the control and test diets was

not randomized because of the small number of subjects studied

and because of the uncertainty of the prolonged effect of con-

Discussion

In view of the concerns over the contamination of meat prod-

ucts by synthetic estrogens such as diethylstilbestrol (DES), it is

surprising that there has been relatively little attention given to

the potential effects of ingestion of plant (dietary) estrogens by

humans ( I 8). In the last 50 y, several important examples of the

way in which nonsteroidal dietary estrogens can influence repro-

ductive physiology in animals have been described ( 10- 12).

Many plants consumed by humans contain dietary estrogens and

although the concentrations are generally low, several plants con-

tam high concentrations of isoflavones ( 19, 20), and if ingested

in large quantities may evoke significant biological effects. Soy

protein, for example, is a relatively rich source of isoflavones,

which have partial estrogen agonist-antagonistcharacteristics (7).

Because soy protein and other legumes are consumed in signif-

icant quantities by humans, the acute and chronic effects of ex-

posure to these dietary estrogens may be biologically important.

Although soy-protein products form a major component of the

diet of the Japanese and Chinese, where the incidence of breast

at PE

NN

SY

LVA

NIA

ST

AT

E U

NIV

PA

TE

RN

O LIB

RA

RY

on February 23, 2013

ajcn.nutrition.orgD

ownloaded from

338 CASSIDY ET AL

sumption of soy on menstrual cycle length. In several women,

after the soy diet, three menstrual cycles elapsed before the cycle

length returned to its original value (Table 3). Subjects lived in

the metabolic suite for an average period of 4-6 mo, and for at

least one complete menstrual cycle before initiation of the study.

With the amount of soy protein used in this study there oc-

curred a significant increase in follicular phase length and a

delay in menstruation. The midcycle surges of the gonadotro-

phins LH and FSH were significantly suppressed during the

soy-diet period. In these studies, no special attempt was made

to ascertain the exact day of ovulation and because peak urinary

LH and plasma LH concentrations can be out of phase by a day,

it could be argued that the differences between the soy and

control cycles could be the result of differences in sampling

times. In more recent studies when sampling was carried out

daily for I wk before ovulation was established by urinary LH

concentrations, a I -d delay was found between the peak plasma

and the urinary LH concentrations; however, when an identical

soy protein devoid of isoflavone was used, no significant change

in gonadotrophins was observed. These observations suggest

that the agonist-antagonistic action of dietary estrogens in soy

may be responsible for the hormonal modification to the cycle.

Equol, a specific metabolite of the ingested soy isoflavones

daidzein and genistein ( 17), had the greatest influence; the two

subjects with the highest urinary equol excretion showed the

largest increase in follicular phase length. Interestingly, in our

recent studies Arcon F (Archer Daniels Midland, Decatur IL)-

a soy-protein product devoid of isoflavones-was found to

have no effect on either plasma gonadotrophins or follicular

phase and menstrual cycle lengths (32), whereas miso-a fer-

mented soy protein that has high concentrations of unconju-

gated isoflavones (30)-had a greater effect than did textured

vegetable protein, providing further evidence that isoflavones

may be the biologically active components of soy.

The major influence on menstrual cycle length is variation

in the length of the follicular phase (33). In the present study

follicular phase length was significantly increased by an av-

erage of 2.5 d when soy protein was consumed daily, whereas

no significant change in luteal phase length was observed. One

possible explanation for the relationship between menstrual

cycle length and breast cancer risk may be that shorter men-

strual cycles would lead to a greater life-time exposure to es-

trogen. Furthermore, because the mitotic rate for breast tissue

is almost fourfold greater during the luteal phase than during

the follicular phase (27, 34), women with shorter cycles and

consequently at high risk for breast cancer, will therefore

spend proportionally more of their lifetime in the luteal phase

of the cycle. A significant increase in menstrual cycle length,

particularly follicular phase length, as was observed in our

study, would therefore be potentially beneficial in lowering

risk for breast cancer. The influence of soy protein containing

dietary estrogens on the menstrual cycle may explain the re-

duced risk for breast cancer in premenopausal women con-

suming soy-protein products (24). Interestingly, Olsson et al

(26) made a retrospective assessment of cycle length and

found a significantly shorter cycle length for breast cancer

patients compared with control subjects (26.4 vs 28.6 d). Men-

strual cycle length is also significantly longer in Asian women

than in Western women. Menstrual cycle length of women

from Western populations ranges from 26 to 29 d (26-29),

whereas the average cycle length for Japanese women is

longer (27). It is conceivable that the differences in hormonal

status and characteristics of the menstrual cycle of Japanese

and Chinese women (35) may be in part due to the ingestion

of substantial amounts of nonsteroidal estrogens present in soy

protein. The results of the present study provide evidence that

soy-protein-containing dietary estrogens when ingested daily

will lead to significant biological effects in premenopausal

Western women, which are potentially beneficial with regard

to breast cancer risk.

Some of the biological effects of a soy-protein diet contain-

ing isoflavones are similar to those induced by the potent syn-

thetic antiestrogen tamoxifen (36). Tamoxifen, when used

therapeutically in breast cancer patients, leads to decreases in

circulating concentrations of LH and FSH (37-39) and a re-

duction in total serum cholesterol and LDL-cholesterol con-

centrations; however, increases in SHBG are consistently ob-

served in both pre- and postmenopausal women after pro-

longed antiestrogen therapy (40-46). The magnitude of the

change in LH and FSH concentrations when soy protein is

included in the diet is markedly greater than that observed in

studies of tamoxifen. Jordan et al (40) showed that 10 mg

tamoxifen/d suppressed LH concentrations by 39% in pre-

menopausal women and 17% in postmenopausal women with

breast cancer. With a dose of4O mg/d, Willis et al (37) showed

a 41% decrease in plasma LH and a 29% decrease in plasma

FSH concentrations after 6 wk of therapy. These data compare

with a 300% reduction of LH and 200% reduction of FSH

during a I-mo period in which soy protein is consumed.

The hypocholesterolemic effect of soy protein may relate to

its nonstarch polysaccharide content (47, 48), but isoflavones

may also play a role in cholesterol homeostasis (20), because we

recently showed that soy-milk formulas containing isoflavones

will influence cholesterol fractional synthesis rates in newborn

infants (49). In the present study a mean reduction in serum cho-

lesterol of 9.6% is impressive given the relatively short duration

of soy-protein intake and the fact that these women had normal

serum cholesterol concentrations. No change in plasma SHBG

concentration was observed during soy intake but it is possible

that the study period was too short to observe an effect. A similar

lack of effect was also observed in a separate study of post-

menopausal women consuming soy protein (�s�50 g/d) for 1 mo

(DD Baird, KDR Setchell, unpublished data, 1988).

Recent and controversial trials of tamoxifen as a prophylactic

agent are underway in women at high risk for breast cancer, but

with no evidence of disease (50-52). The earlier demonstration

that soy-protein-containing isoflavones have anticancer actions

in animal models of breast cancer (9) raises the question of

whether dietary intervention with soy protein should be consid-

ered as an alternative approach to drug therapy for breast cancer

prevention. Significantly higher concentrations of estradiol (mea-

sured by immunoassay) were found in the plasma during the

follicular phase when the women were given soy protein. Similar

increases have been found in women given 20 mg tamoxifen/d

(52). Whereas a high estrogen concentration in the follicular

phase may be considered undesirable from the point of view of

breast cancer, this finding did not deter the recent clinical trials

of tamoxifen.

In summary, our observations clearly indicate how dietary

modifications can lead to significant changes in the regulation of

at PE

NN

SY

LVA

NIA

ST

AT

E U

NIV

PA

TE

RN

O LIB

RA

RY

on February 23, 2013

ajcn.nutrition.orgD

ownloaded from

SOY ISOFLAVONES AND THE MENSTRUAL CYCLE 339

the menstrual cycle. Such changes may be beneficial with regard

to risk factors for breast cancer and suggest that a diet rich in

dietary estrogens may be protective against breast cancer. U

We thank the volunteers for their invaluable help. The technical as-sistance of J Carlson, E Collard, and R Reader are acknowledged. Weare grateful to P Raggett (Department of Clinical Biochemistry,

Cambridge. UK) and M Ilbery (Bourn Hall, Cambridge, UK) for advice

on hormone measurements.

References

1. Muir C, Waterhouse I. Mack T, Powell I, Whelan S. Cancer mci-

dence in five continents. Vol 5. Lyon. France: International Agency

for Research on Cancer, 1987. (IARC Scientific publication no 88.)

2. Committee on Diet, Nutrition, and Cancer, Assembly of Life Sci-

ences, National Research Council. Diet, nutrition, and cancer.

Washington, DC: National Academy Press, 1982.

3. Willett WC. Stampfer MI, Colditz GA. Rosner BA, Hennekens CH,

Speizer FE. Dietary fat and the risk of breast cancer. N EngI I Med

I 987:316:22-8.

4. Key TIA. Pike MC. The role of oestrogens and progestagens in the

epidemiology and prevention of breast cancer. Eur I Cancer Clin

Oncol 1988:24:29-43.

5. Armstrong B. Doll R. Environmental factors and cancer incidence

and mortality rates in different countries with special reference to

dietary practices. Int I Cancer 1975:15:617-31.

6. Setchell KDR, Welsh M, Lim CK. High-performance liquid chro-

matographic analysis of phytoestrogens in soy protein preparations

with ultraviolet, electrochemical and thermospray mass spectromet-

ric detection. I Chromatogr 1987:386:315-23.

7. Setchell KDR, Adlercreutz H. Mammalian lignans and phytoestro-

gens: recent studies on their formation, metabolism and biological

role in health and disease. In: Rowland IA, ed. The role of gut mi-

croflora in toxicity and cancer. New York: Academic Press,

1988:315-45.

8. Setchell KDR, Borriello SP. Hulme P. Axelson M. Nonsteroidal

estrogens ofdietary origin: possible roles in hormone-dependent dis-

ease. Am I Clin Nutr 1984:40:569-78.

9. Bames S. Grubbs C, Setchell KDR. Carlson I. Soybeans inhibit

mammary tumor growth in models of breast cancer. In: Pariza MW,

ed. Mutagens and carcinogens in the diet. New York: Wiley-Liss.

I 990:239-53.

10. Bennetts HW, Underwood EJ, Shier FL. A specific breeding prob-

1cm of sheep on subterranean clover pastures in Western AustraliaAust Vet I 1946:22:2-12.

I I . Shutt DA. The effect of plant oestrogens on animal reproduction.

Endeavour 1976:3�:110-3.

12. Setchell KDR, Gosselin SJ, Welsh M, et al. Dietary estrogens-a

probable cause of infertility and liver disease in captive cheetah.

Gastroenterology 1987:93:225-33.

13. Schofield NW. Schofield C, James WPT. Basal metabolic rate. Hum

Nutr Clin Nutr 1985:39C(suppl I ): I -96.

14. Bingham S. Cummings JH. The use of 4-amino benzoic acid as a

marker to validate the completeness of 24 hour urine collections in

man. Clin Sci 1983:64:629-35.

15. Cummings JH, Jenkins DJA, Wiggins HS. Measurement ofthe mean

transit time of dietary residue through the human gut. Gut 1976:

17:210-8.

16. Barbuch RI, Coutant JE, Setchell KDR, Welsh MB. The use of ther-mospray LC/MS/MS for the class identification and structural veri-

fication of phytoestrogens in soy protein preparations. I Biomed

Mass Spectrom 1989:18:973-7.

17. Axelson M, Kirk DN, Farrant RD. Cooley G, Lawson AM, Setchell

KDR. The identification of the weak estrogen equol in human urine.

Biochem I 1982:201:353-7.

18. McLachlan I, ed. Estrogens in the environment: influence Ofl devel-

opment. New York: Elsevier. 1985.19. Price KR, Fenwick GR. Naturally occurring oestrogens in foods-

a review. Food Addit Contam 198�:2:73-106.

20. Setchell KDR. Naturally occurring non-steroidal estrogens of dietary

origin. In: McLachlan I. ed. Estrogens in the environment: influence

on development. New York: Elsevier. 1985:69-85.21. Setchell KDR, Lawson AM, Borriello SP, et al. Lignan formation

in man-microbial involvement and possible roles in relation to

cancer. Lancet 1981:2:4-8.

22. Adlercreutz H. Does fiber-rich food containing animal lignan pre-

cursors protect against both colon and breast cancer? An extension

of “fiber hypothesis”. Gastroenterology 1984:86:761-6.23. Adlercreutz H. Western diet and Western diseases: some hormonal

and biochemical mechanisms and associations. Scand I Clin LabInvest Suppl 1990:201:3-21.

24. Lee HP, Gourley L, Duffy SW. Esteve I. Lee I, Day NE. Dietaryeffects on breast cancer risk in Singapore. Lancet 1991 :337: 1 197-

200.25. Henderson BE, Ross RK. Judd HL, Frailo MD, Pike MC. Do reg-

ulatory ovulatory cycles increase breast cancer risk? Cancer I 985:

56:1206-8.

26. Olsson H, Landin-Olsson M, Gullberg B. Retrospective assessment

of menstrual cycle length in patients with breast cancer, in patientswith benign breast disease, and in women without breast disease. I

Nati Cancer Inst 1983:70:17-20.

27. Treolar AE, Boynton RE, Behn BG, Brown BW. Variation of the

human menstrual cycle through reproductive life. Int I Fertil

1970:12:77-126.

28. World Health Organization. A prospective multicenter trial of the

ovulation method of natural family planning. III Characteristics of

the menstrual cycle and of the fertile phase. Fertil Steril 1983:40:

773-8.

29. Munster K, Schmidt L, Helm P. Length and variation in the men-

strual cycle-a cross sectional study from a Danish county. Br I

Obstet Gynaecol 1992:99:422-9.

30. Coward L, Barnes NC, Setchell KDR, Barnes S. The isoflavones

genistein and daidzein in soy bean foods from American and Asian

diets. I Agric Food Chem I 993:4 1 : I 961-7.

31 . Adlercreutz H, Honjo A, Higashi A. et al. Lignan and phytoestrogen

excretion in Japanese consuming traditional diet. Scand I Clin In-

vest, 1988:48:190 (abstr).

32. Cassidy A, Bingham 5, Carlson I, Setchell KDR. Biological effects

of plant estrogens in premenopausal women. FASEB I 1993:A866(abstr).

33. Aksel S. Hormonal characteristics of long cycles in fertile women.

Fertil Steril 1981:36:521-3.

34. Ferguson DIP, Anderson TI. Morphological evaluation of cell turn-

over in relation to the menstrual cycle in the ‘ ‘resting’ ‘ human breast.

BrJ Cancer 1981:44:177-81.35. Key TJA, Chen DY, Wang DY, Pike MC, Boreham I. Sex hormones

in rural China and Britain. Br I Cancer 1990:62:63 1 -6.

36. Miller WR. Endocrine treatment for breast cancer. Biological ra-

tionale and current progress. I Steroid Biochem 1990:37:467-

80.

37. Willis KI, London DR. Ward HWC. Butt WR, Lynch SS, Rudd BT.

Recurrent breast cancer treated with the antiestrogen tamoxifen: cor-

relation between hormonal changes and clinical course. Br Med I

1977:1:425-8.

38. Jordan VC. Fritz NF, Tormey DC. Endocrine effects of adjuvant

chemotherapy and long-term tamoxifen administration of node-pos-

itive patients with breast cancer. Cancer Res 1987:47:624-30.

39. Golder MP, Phillips EA. Fahmy DR. et al. Plasma hormones in

patients with advanced breast cancer treated with tamoxifen. Eur I

Cancer, 1976:12:719-23.

at PE

NN

SY

LVA

NIA

ST

AT

E U

NIV

PA

TE

RN

O LIB

RA

RY

on February 23, 2013

ajcn.nutrition.orgD

ownloaded from

340 CASSIDY ET AL

40. Jordan VC, Fritz NF. Tormey DC. Long-term adjuvant therapy with

tamoxifen: effects on sex hormone binding globulin and antithrom-

bin III. Cancer Res 1987:47:4517-9.

41 . Love RR. Newcomb PA. Wiebe DA, et al. Effects of tamoxifen onlipid and lipoprotein levels in postmenopausal patients with node-

negative breast cancer. I NatI Cancer Inst 1990:82:1327-32.

42. Baglade ID, Wolter I, Subbaiah PV. Ryan W. Effects of tamoxifenon plasma lipids and lipoprotein lipid composition. I Clin Endocri-nol Metab 1990:70:1132-5.

43. Sakai F, Cheix F. Clavel M, et al. Increases in steroid binding glob-

ulins induced by tamoxifen in patients with carcinoma of the breast.

I Endocrinol 1978:76:219-26.44. Vincze B, Szamel I, Hindy I. Kerpel-Fronius S. Eckhardt S. Deter-

mination of free estradiol concentration and the binding capacity of

SHBG in the serum of patients treated with tamoxifen. In: Gorog,

ed. Advances in steroid analysis. Budapest: Akademia Kiado.

1982:107- 12.

45. Szamel I. Vincze B, Hindy I. Hermann I. Borvendeg I. Eckhardt S.

Hormonal changes during a prolonged tamoxifen treatment in pa-tients with advanced breast cancer. Oncology 1986:43:7- I 1.

46. Bruning PF, Bonfrer JMG, Hart AAM. et al. Tamoxifen, serum Ii-poproteins and cardiovascular risk. Br I Cancer, 1988:58:497-9.

47. Carroll KK, Giovannetti PM, Huff MW, Moase 0, Roberts DCK,

Wolfe BM. Hypocholesterolemic effect of substituting soybean pro-tein for animal protein in the diet of healthy young women. Am IClin Nutr 1978:31:1312-21.

48. Sirtori CR, Gatti E, Mantero 0, et al. Clinical experience with the

soybean protein diet in the treatment of hypercholesterolemia. Am

I Clin Nutr, 1979:32:1645-58.

49. Cruz MLA, Wong WW, Mimouni F, et al. Effects of infant nu-

trition on cholesterol synthesis rates. Pediatr Res 1994:35:135-

46.

50. Powles TI, Hardy JR. Ashley SE. et al. A pilot trial to evaluate the

acute toxicity and feasibility of tamoxifen for the prevention ofbreast cancer. Br J Cancer 1989:60:126-31.

5 1 . Powles TI, Tillyer CR, Jones AL. et al. Prevention of breast cancerwith tamoxifen-an update on the Royal Marsden Hospital Pilot

Programme. Eur I Cancer 1990:26:680-4.52. Davidson NE. Tamoxifen-panacea or Pandora’s box. N EngI J

Med 1992:326:885-6.

at PE

NN

SY

LVA

NIA

ST

AT

E U

NIV

PA

TE

RN

O LIB

RA

RY

on February 23, 2013

ajcn.nutrition.orgD

ownloaded from