The relationship concentrations postmenopausal Joilo
C Barbosa,
Terry
among adiposity, diet, and hormone in vegetarian and nonvegetarian women13 D Shultz,
Shelby
J Filley,
and David
ABSTRACT The relationships among anthropometric variables, dietary nutrients, and plasma steroid, polypeptide, and binding-protein hormone concentrations were investigated in 24 Seventh-day Adventist postmenopausal women, 12 vegetarian (SV) and 12 nonvegetarian (SNV). Fasting blood and 7-d dietary intake information were collected. SVs consumed significantly more crude and dietary fiber and fewer saturated fatty acids than did SNVs. The thigh and sum of three skinfold-thickness measurements were significantly greater for SNVs than for SVs. Plasma concentrations of estradiol-l7 were significantly lower in SVs than in SNVs. Significant relationships were observed for the combined groups (SV and SNV) between estradiol-l7fl and triceps and suprailiac skinfold thicknesses and body fat. Plasma concentrations of estradioll7fl ofthe combined groups revealed a significant negative relationship between their crude and dietary fiber intakes. Further study delineating the effects of adiposity and dietary nutrients on basal concentrations ofsex hormones is warranted. Am JClin Nutr l990;5l:798-803. KEY WORDS Seventh-day Adventist, diet-hormone relationships, skinfold, adiposity, estradiol-l 7, breast cancer
anthropometryleanness, fiber,
C Nieman excretion of endogenous estrogens. Goldin et al (1 1) reported lower plasma and urinary estrogen concentrations in premenopausal vegetarian women than in nonvegetarians. Goldin et al (1 1) surveyed vegetarian women who had diets lower in fat (30% of calories) and higher in dietary fiber (28 g/d) than the average American omnivorous diet but the proportion of fat was not at the remarkably lower level (20% calories as fat) seen among Asian women ( 13), who have the lowest breast-cancermortality rates in the world. In an effort to address this issue, Goldin et al (1 3) found that Asian premenopausal and postmenopausal
women
possibly
because
12,
14,
Seventh-day
Adventist
(SDA)
women
experience
lower
mor-
from breast cancer than does the general California population (1-3). It has been suggested that this may be due to their dietary habits because 50% are lactoovovegetarians (1-3). Mills et al (3) recently found no significant relationship between the consumption of animal products (ie, meat, milk, cheese, and eggs) and breast-cancer mortality among SDAs. The possible association between dietary fiber and risk of breast cancer was not determined. Epidemiologic studies have demonstrated associations between dietary nutrients, especially fat and fiber, and the mcidence of hormone-dependent cancers, including breast cancer (4-6). Recent studies indicate that diet may modify sex-hormone metabolism in both men and women (7-16). Indeed, women consuming various diets-vegetarian or nonvegetarian and low- or high-fat-had different plasma, urine, and fecal concentrations ofestrogens (9-16). Diet appears to influence the production, metabolism, and 798
Am
J C/in
Nuir
1990;5l:798-803.
Downloaded from https://academic.oup.com/ajcn/article-abstract/51/5/798/4695325 by Denise Hannibal user on 12 April 2018
of adipose
1 5) studies
reduce
tabolism.
tality
ate low-fat
(20%
ofcalories)
diets
had
tissue
loss.
This
suggests
that
the
influence of a low-fat diet may not have been fully considered because adipose tissue is the major source of estrogen formation in postmenopausal women (I 7). These as well as other (9, may
Introduction
who
lower plasma estrogen concentrations than did American omnivorous women. In another study Boyar et al ( 16) found that serum concentrations of estradiol17$ but not of estrone were significantly reduced in postmenopausal breast-cancer patients placed on a low-fat diet (20% of calories) for 5-6 mo. Mean body weight decreased by ‘-2.3 kg during the same time (16),
suggest
that
a low-fat
and/or
high-fiber
diet
breast-cancer Further
risk through effects on estrogen meresearch is needed to define more precisely among adiposity, dietary nutrients, and estro-
the relationships gen metabolism. The objective of this study was to characterize those dietary and hormonal variables that may explain the observations of lower mortality from breast cancer among SDAs. Herein we describe the hormonal status and interrelation ofdietary nutrients with plasma steroid and polypeptide hormone profiles of postmenopausal SDA women habitually following vegetarian (SV) and nonvegetarian (SNV) dietary regimes. In addition, relationships between various anthropometric variables and plasma I
hormone
From
concentrations
the Department
the Departments
of Health
are
of Biochemistry, Science
reported.
School
and Nutrition,
of Medicine, School
and
of Public
Health, Loma Linda University, Loma Linda, CA. 2 Supported in part by a grant from the Callicott Foundation (TDS). 3 Address reprint requests to TD Shultz, Department of Food Science and Human Nutrition, Washington State University, Pullman,
WA 99164-6376. Received April 4, 1989. Accepted for publication Printed
in USA.
June 7, 1989.
© 1990 American
Society
for Clinical
Nutrition
ADIPOSITY-DIET-HORMONE
RELATIONSHIPS
Methods Subject
selection
Elderly
female
subjects
were
initially
recruited
from
mem-
bars
of the Adventist Health Study, a prospective cohort involving 6 y (1977-1982) of follow-up of 34 198 Caucasian SDAs, including 20 341 women living in California. Participants were chosen who resided within a 160.9-km radius of the Loma Linda University campus. Women were contacted by telephone. vidually
Subjects who volunteered sent a questionnaire that
to participate requested detailed
information regarding their medical their use ofdrugs. These questionnaires the
health
of the
to potential consent
The
participants before
Loma Linda A sample from
subjects.
and
acceptance.
University’s of SV and
the volunteers.
study
dietary histories and were used to ascertain
protocol
was
each
subject
signed
The
protocol
was
The
SV group
consisted
explained
an informed
than once a month 25 y. Additional
and
had
by
Studies. was selected
of 12 women
12 women aged 64-83 if they ate meat, fish, followed
this
aged
y. Subjects or poultry dietary
regi-
selection criteria were as follows: normal weight to minimally obese [body mass index (BMI, kg! m2), 20-29.9] (1 8) and no history ofcancer, cardiovascular disease,
hypertension,
group
comprised
ject; obese
the
diabetes,
1 1 normal-weight
SNVs
comprised
subjects.
coids, treated lism.
with
Sampling
and
Subjects
individual
describing
started
or
any had
hormone
nutrient
and saturated
before
food
combinations their
diets
by telephone.
Upon
included.
analysis
(19)
carbohydrate,
fatty acids,
Foods
they
and
oleic and linoleic
as cassewere
given
completion,
each
to ensure
that
coded
for corn-
were
of 10 dietary crude
writ-
estimat-
(such after
by the nutritionist
was
with
intake,
variables:
dietary
all food
fiber,
total
fatty acids,
and cho-
obtained
from
lesterol.
Blood antecubital
collection. vein
Blood ofeach
samples
subject;
samples were drawn into 0700 and 0900. Aliquots
after
heparinized ofplasma
were
an overnight
evacuated were stored
analyzed. Samples assays were frozen
for sex-hormone-binding at -70 #{176}C until assayed.
the morning
the
after
7-d diet
record
fast,
the
20-mL
tubes between at -20 #{176}C until
globulin (SHBG) All blood was drawn
was completed.
Anthropometric measurements. Weight (wt) and height (ht) were determined by the use of a calibrated physician’s scale and a wall-mounted measuring tape. Shoes and outerwear were removed before weighing. BMI was calculated as an estimate of obesity (20, 2 1). The three-site skinfold test was used as a second indicator of body fat. Triceps, suprailiac, and thigh
Downloaded from https://academic.oup.com/ajcn/article-abstract/51/5/798/4695325 by Denise Hannibal user on 12 April 2018
in duplicate
Industries,
with
Analyticalmethodologyfor and binding-protein
polypeptide,
Plasma (DHEA-S), by
steroid,
estradiol1 7/3, growth hormone,
commercially
RIA
dehydroepiandrosterone-sulfate and prolactin were determined
purchased
(RIA) kits CA). All hormones, with the sulfate, were measured by di-
radioimmunoassay
Inc, Carson, and estrone
of25-l00
ethyl
MD), regres-
hormones
(ICN Biomedicals exception ofestrone rect
Lange
Cambridge,
by use ofa quadratic
L
plasma.
Estrone
was
measured
by RIA
(3:2, vol:vol) extraction and punfication on 0.5 cm X 5 cm Sephadex LH-20 (Sigma Chemical Co, St. Louis) columns as previously described by Shultz et al (24). The mean (±SD) recovery of[3Hjestrone added to plasma was 86 ± 9%. Values were corrected for recovery ofthe tritiated internal standard. All assays were performed in duplicate; in each assay series an equal number ofsubjects from each group was included. Individual assays were monitored by quality control samples (ICN Biomedicals Inc) included in each group of unknowns. All assays were performed with the use of highly specific
acetate-hexane
antiserum
tivity.
Kits
from
Farmos
that
for
in
(Beckman
been
measured
with the percent body fat estimated sion equation (22, 23).
counters Estrone
collection
along
Scientific
type
metabo-
sample
booklet
were
(Cambridge
counted
glucocortiif they
for reporting
to record
was reviewed
protein,
weeks
procedures
information
puterized energy,
6 mo
to influence
Two
instruction
record
relevant
known
used
thicknesses
calipers
sub-
4 minimally
had
progesterone,
past
SV
obese
and
ifthey
estrogens,
The
1 minimally
a 7-d food-record
and
ing portions, roles).
excluded
the
drugs
assessment. were mailed
ten instructions
food
(ie, within
other
and
disease.
procedures
Dietary subjects
were
therapy
or thyroid)
or kidney
8 normal-weight
Women
of hormone
or liver
skinfold
after
approved
Committee on Human SNV Caucasian women
66-80 y and the SNV group, were classified as vegetarians no more men for
and
were mdipersonal
799
exhibits
negligible
immunoradioassay
of
Diagnostica Beckman
LS-7500
Instruments (model and
(Oulunsalo, Co,
1 185, Searle, estradiol-17/3
antiserum SHBG
crossreacwere
Finland).
liquid
Fullerton,
scintillation
CA)
obtained
Samples
were
counters
or Searle
Des Plaines, IL). interassay variabilities
gamma were
15%
(n = 4) and 8% (n = 3), respectively; intraassay variabilities were 1 1% (n = 10) and 12% (n = 10), respectively. DHEA-S interassay and intraassay variabilities were 8% (n = 4) and 6% (n = 10), respectively. Prolactin and growth hormone interassay variabilities were 6% (n = 5) and 3% (n = 4), respectively; intraassay variabilities were 5% (n = 10) and 13% (n = 10), respectively. SHBG between and within CVs were 4% (n = 7) and 3% (n = 10), respectively. [6,7-3H]Estrone sulfate ammonium salt (2.22 TBq/mmol) was purchased from NEN Research Products (El DuPont de Nemours,
pure.
Wilmington,
Plasma
as previously tracted
DE).
estrone
sulfate
described
(25,
to remove
The
radiolabeled
steroid
concentrations 26).
the free estrogens,
were
Initially,
1 mL
followed
was
98%
determined
plasma
was
by enzymatic
cxhy-
drolysis of the estrone sulfate with arylsulfatase. After organic extraction further separation and purification of estrone was performed on Sephadex LH-20 columns (24). Estrone was determined by RIA as described above. Estrone sulfate interassay and intraassay variabilities were 10% (n = 4) and 8% (n = 10), respectively. The recovery of added [3Hjestrone sulfate averaged 73 ± 6%. Values were corrected for recovery of the tntiated internal standard. St atistical
analyses
Statistical analyses were performed with the St atistical Packagefor the Social Sciences (27). Anthropometric variables, dietary
constituents,
were transformed Tests ofstatistical values. Correlations
and
hormone
and
SHBG
concentrations
to logarithmic (base 10) values for analysis. significance were based on logarithms of the between individual variables were exam-
BARBOSA
800 TABLE
and anthropometric characteristics Seventh-day Adventist vegetarian (SNV) women5 Variables
SV (n
Age(y)
75.5±
Ageat menarche(y) Age at first pregnancy (y) Age at menopause (y)
13.0 28.2 50.7
Weight
(kg)
57.4 159.6
Height (cm) Skinfold-thickness (mm)
Thigh Sum ofskinfold Body fat (%) mass
± ± ± ± ±
of the (SV) and
thicknesses
index
(kg/m2)
12) 1.2 0.3 2. 1 1.2 2.4 1.7
SNV
(n
71.5± 13.4 27.3 49.3 63.9 160.4
± ± ± ± ±
12)
=
1.6 0.5 2.4 1.3 3.3 2.1
2 1.6 14.8
± ±
2.5 2.0
30.8 67.2 27.8
±
4.4
± ±
7.7 1.2
27.7 ± 2.0 17.4 ± 2.3 39.9 ± 2.5t 85.0 ± 5.6 32.8 ± 1.5
22.5
±
0.8
24.7
±
0.8
different from SV, p < 0.01. different from SV, p < 0.05. from the triceps, suprailiac, and
1: Significantly § Calculated ness measurements
(g)
=
12)
47±4
SNV (n
=
1334
100
±
12)
54±4
13
16
Carbohydrate
(g) (%energy) Crude fiber(g) Dietaryfiber(g) Dietary fat
216±22 61 7 ± I
Total fatty acids (g) (%energy) (g/l000kcal) Saturated fatty acids
166±11 50 4 ± 0.4t
24±2
l3±lt
50±6
54±6
32
(g)
36
35±2
39±2
10±2
l7±3 12 ± 1 1
8± 1 12±5 7 ± 1 72 ± 18
kcal)
13±4 1 22
7 ±
18 1
±
51±SEM.
Results of the 7-d dietary
SV (n
skinfold-thick-
(22, 23).
med by linear-regression analysis (Pearson correlation coefficients). Mean differences in anthropometric measurements, dietary nutrient intakes, hormones, and hormone-binding protein were compared and tested with the unpaired Student’s t test (two-tailed). All statistical tests were considered to be significant at the p < 0.05 level.
Collection
5V and SNV groups
(%energy)
Oleicacid(g) Linoleic acid (g) Cholesterol(mg) thigh
ofthe
1425 ± 148w
Energy(kcal) Protein
(g/l000
5i±SEM. t Significantly
intake averages
Nutrients
measurements
Triceps Suprailiac
Body
=
AL
TABLE 2 Seven-day dietary
1
Demographic postmenopausal nonvegetarian
ET
nutrient intake data was conSubjects from each group
ducted through April and May 1988. were studied throughout that period.
The 12 SV subjects had been vegetarians for an average (±SEM)of54 ± 4 y, ranging from 30-74 y. This group included eight lactoovovegetanans, one lactovegetarian, and three vegans. The SNVs had eaten meat all their lives. Their average consumption of meat, fish, or poultry was three to four times per week. The SV and SNV participants did not differ significantly with respect to age, age at menarche, age at first pregnancy, age at menopause, weight, height, triceps or suprailiac skinfold thicknesses, body fat, or BMI (Table 1). The thigh skinfold thickness of the SNVs however, was significantly greater than that ofthe SVs. After combining individual skinfold-thickness measurements within groups, the SNVs had significantly higher summed skinfold thicknesses than did the SVs (p < 0.05). Significant relationships were observed for the two groups cornbined (SV and SNV) between estradiol-l7fl and triceps and suprailiac skinfold thicknesses (r = 0.40, p < 0.05; r = 0.37, p < 0.05, respectively) and percent body fat (r = 0.39, p < 0.05). For the combined groups linear-regression analysis also revealed significant negative relationships between SHBG and weight, suprailiac skinfold thickness, and BMI (r = -0.44, p < 0.025; r = -0.61, p < 0.001; r = -0.51, p < 0.005, respectively).
Downloaded from https://academic.oup.com/ajcn/article-abstract/51/5/798/4695325 by Denise Hannibal user on 12 April 2018
t Significantly t Significantly § Significantly II Significantly
Table Dietary acids,
different
different different different
2 presents intakes
from SV, p