/

University of Southern California Dissertations and Theses

/

Associations between isoflavone soy protein (ISP) supplementation and breast cancer in postmenopausal women in the Women’s Isoflavone Soy Health (WISH) clinical trial

(USC Thesis Other)

Associations between isoflavone soy protein (ISP) supplementation and breast cancer in postmenopausal women in the Women’s Isoflavone Soy Health (WISH) clinical trial

PDF

Transcript (if available)

Content
i

Associations
Between
Isoflavone
soy
protein
(ISP)
Supplementation
and

Breast
Cancer
in
Postmenopausal
Women
in
the
Women’s
Isoflavone
Soy

Health
(WISH)
Clinical
Trial

by

Ceng
Qian

A
Thesis
Presented
to
the

FACULTY
OF
THE
GRADUATE
SCHOOL

UNIVERSITY
OF
SOUTHERN
CALIFORNIA

In
Partial
Fulfillment
of
the

Requirements
for
the
Degree

MASTER
OF
SCIENCE

(APPLIED
BIOSTATISTICS
AND
EPIDEMIOLOGY)

December
2015

ii

DEDICATION

This
work
is
dedicated
to
my
beloved
parents
for
their
endless
support,
and

to
Wayne
for
always
being
there
for
me.

iii

ACKNOWLEGEMENTS

First
of
all,
I
would
like
to
express
my
deepest
appreciation
to
my
advisor
Dr.

Wendy
Mack,
who
supported
and
helped
me
each
step
of
the
completion
of
this

research
project,
with
her
kindness
and
patience.

In
addition,
I
would
like
to
thank
Dr.
Anna
Wu,
who
helped
me
with
her

expertise
and
guidance;
also,
to
Dr.
Howard
Hodis,
who
gave
his
precious

suggestions
through
this
project.

Finally,
I
would
like
to
thank
all
my
friends
and
the
entire
faculty
from
the

preventive
medicine
department,
for
helping
me
and
accompanying
with
me
all
the

way
through
my
2-‐year
masters
program.

iv

TABLE
OF
CONTENTS

ABSTRACT
1

INTRODUCTION
2

METHODS
3

RESULTS
10

DISCUSSION
14

CONCLUSION
18

REFERENCES
19

FIGURES
23

TABLES
24

1

ABSTRACT

Background:
Although
epidemiological
studies
have
suggested
that
soy
food

consumption
may
be
related
to
the
incidence
of
breast
cancer,
the
effectiveness
of

isoflavone
soy
protein
(ISP)
supplementation
in
reducing
the
risk
of
breast
cancer

remains
unclear.
We
assessed
the
effects
of
ISP
supplementation
on
a
biomarker
of

breast
cancer
risk,
mammographic
density.

Methods:
In
a
double
blind,
placebo
controlled
trial,
350
healthy

postmenopausal
women
aged
45
to
92
years
were
randomly
assigned
to
2
evenly

divided
groups:
daily
25
g
of
soy
protein
(91
mg
aglycone
isoflavone
equivalents)
or

placebo,
for
a
2.5-‐3
year
treatment
period.
The
primary
outcome
was
percent

mammographic
density
that
was
measured
at
baseline
and
annually.

Results:
307
women
with
baseline
and
follow-‐up
mammograms
were

included
in
the
intent-‐to-‐treat
analyses.

After
adjusting
for
baseline
carotid
artery

intima-‐media
thickness
(CIMT)
strata
and
baseline
percent
mammographic
density,

there
was
no
significant
difference
in
the
mean
square-‐root
transformed
percent

mammographic
density
measured
over
the
trial
follow-‐up
among
the
treatment

groups
(p=0.83).

ISP
soy
supplementation
also
did
not
differ
at
the
annual

measures
over
the
trial
follow-‐up.
Analyses
of
total
urinary
isoflavone
levels
showed

no
associations
with
breast
density
over
visit,
except
for
a
marginally
significant

positive
association
of
glycitein
with
percent
mammographic
density(p=0.06).

Conclusion:
In
healthy
postmenopausal
women,
ISP
supplementation
did

not
significantly
affect
mammographic
density.
Effects
of
isoflavones,
such
as

glycitein,
on
mammographic
density
should
be
investigated
by
further
studies.

2

INTRODUCTION

It
has
been
hypothesized
that
soy
food
consumption
in
the
Asian
diet
may

contribute,
in
part,
to
the
historically
lower
breast
cancer
incidence
in
Asian

compared
to
European
and
American
women

1
.

Observational
epidemiological

studies
have
suggested
a
role
of
soy
isoflavones
both
in
reducing
the
risk
of
breast

cancer
development

1

and
recurrence
in
breast
cancer
patients

2
.

The
major
isoflavones
in
soybeans
are
daidzein,
genistein
and
glycitein,

which
have
been
found
to
have
many
biological
effects
in
humans
and
may
act
as
a

weak
estrogen
agonists
in
post-‐menopausal
women
3
.

Isoflavones
also
affect
breast

cell
tumor
metabolism;
for
example,
in
studies
of
the
mammary
glands
of
immature

rats,
genistein
up-‐regulated
the
cell
cycle
of
P21
protein
and
the
expression
of
the

epidermal
growth
factor
receptor
(EGFR),
thus
enhancing
breast
cell
development

and
differentiation,
and
suppressing
the
development
of
breast
cancer

4
.

Mammographic
density
measures
the
amount
of
fat
and
stromal/epithelial

tissues
(dark
and
white
areas
on
mammograms)
in
breast

5
;
higher
mammographic

density
is
a
major
risk
factor
for
breast
cancer

6
.

Mammographic
density
is
in
fact
a

stronger
breast
cancer
risk
factor
than
many
other
documented
risk
factors,
such
as

family
history

7
.
The
association
between
breast
density
and
breast
cancer
risk
is

apparent
across
age
groups
and
race/ethnicities

8
.

Breast
density
has
also
been

used
as
an
indicator
of
treatment
response
and
therapeutic
effect.

For
example,
in
a

primary
breast
cancer
prevention
trial,
high-‐risk
women
randomized
to
tamoxifen

treatment
showed
a
significant
reduction
in
breast
density
compared
to
placebo-‐
treated
women

9
.

3

Since
ISP
has
a
similar
chemical
structure
with
tamoxifen

3
,
there
may
also
be

a
beneficial
association
of
soy
isoflavone
supplementation
on
breast
density.

However,
randomized
clinical
trials
have
not
provided
support
that
soy
isoflavone

supplementation
have
significant
beneficial
effects
on

breast
density
in
post-‐
menopausal
women

10
.

We
tested
the
effects
of
a
3-‐year
intervention
with
daily
soy
isoflavone

supplementation
on
mammographic
density
among
350
healthy
postmenopausal

women
in
the
Women’s
Isoflavone
Soy
Health
(WISH)
trial,
a
double-‐blind,

randomized
placebo-‐controlled
trial.

We
used
WISH
trial
data
to:
(1)
test
the
effect

of
soy
isoflavone
treatment
in
altering
breast
density
and
further,
(2)
test
whether

isoflavone
levels
measured
in
urine
or
blood
are
correlated
with
breast
density

levels
and
change
in
breast
density
over
the
trial.

METHODS

WISH
study
design

The
Women’s
Isoflavone
Soy
Health
(WISH)
trial
was
a
single-‐center

randomized,
double
blind,
placebo-‐controlled
trial,
conducted
between
June
2004

and
March
2009
11
.

Participants
were
350
healthy
postmenopausal
women,

randomly
assigned
to
receive
daily
soy
isoflavone
supplementation
or
placebo.

WISH
was
designed
to
assess
a
number
of
health
outcomes,
including

atherosclerosis
(primary
trial
outcome)
12
,and
secondary
trial
outcomes
of

cognition
11
,
bone
density,
and
breast
density.

The
treatment
period
was
initially

4

planned
to
be
2.5
years,
and
was
increased
to
3
years
to
increase
the
probability
of

detecting
a
difference
in
the
primary
outcomes.
The
study
was
approved
by
the

institutional
review
board
of
the
University
of
Southern
California.

Study
population

Eligible
participants
were
healthy
postmenopausal
women
(absence
of

menses
for
at
least
12
months),
with
a
serum
level
of
total
estradiol
less
than
20

pg/mL.
Women
were
excluded
if
they
had
personal
history
of
cardiovascular

diseases,
diabetes
mellitus
or
fasting
serum
glucose
>126
mg/dL,

triglycerides
>500mg/dL,
uncontrolled
hypertension,
untreated
thyroid
disease,

life-‐threatening
disease,
alcohol
or
substance
use,
taking
menopausal
hormone

therapy
or
isoflavone
dietary
supplements,
or
allergy
to
soy
products.
Participants

were
mainly
recruited
from
the
general
population
in
the
Greater
Los
Angeles
area,

and
provided
written
informed
consent
before
they
participated
in
the
study.

Randomization,
masking,
interventions
and
follow-‐up

Participants
were
randomized
in
a
1:1
ratio
to
either
the
treatment
or

placebo
group.
Blocked
stratified
randomization
was
used;
stratification
used
two

strata
of
carotid
artery
intima-‐media
thickness
(CIMT;
<0.75
mm,
>=0.75
mm).
All

participants,
investigators,
staff,
and
data
monitors
were
blinded
to
treatment

assignment;
only
one
statistician
was
unmasked
to
the
treatment
assignment
for

safety
purposes.

5

Participants
were
randomly
assigned
to
receive
daily
25
g
soy
protein

containing
91
mg
aglycone
equivalent
weight
of
naturally
occurring
isoflavones

(154
mg
total
isoflavones)
or
total
milk
protein-‐matched
placebo
(0
mg
isoflavones).

The
91
mg
aglycone
equivalents

contained
genistein
(52
mg
aglycone
equivalents),

daidzein
(36
mg
aglycone
equivalents)
and
glycitein
(3
mg
aglycone
equivalents).

Both
treatment
and
placebo
groups
were
given
2
doses
daily,
which
were
offered
in

either
beverage
powder-‐food
packs
or
food
bars,
in
identical
taste
and
appearance.

Following
randomization,
participants
were
followed
with
clinic
visits
at

every
month
for
the
first
6
months,
and
then
every
other
month
thereafter
for
the
3-‐
year
trial
follow-‐up.
At
each
visit,
number
of
consumed
soy
protein/
placebo
packs

or
bars
were
counted,
and
data
about
dietary
intake,
intervention
compliance,

nonstudy
and
nutritional
medications,
and
clinical
adverse
events
or
symptoms,

were
ascertained.
Blood
and
urine
samples
were
collected
at
the
baseline
and
every

6-‐months
during
the
trial
follow-‐up
13
.

Mammograms
and
other
examinations
and

laboratory
determinations
such
as
X-‐ray
absorptiometry
bone
scans,
pelvic

examinations
with
Papanicolaou
smears,
complete
blood
counts,
etc,
were

performed
at
baseline
and
annually
12
.

Breast
density
measurement

Mammography
was
performed
at
baseline
and
at
three
annual
visits

following
randomization.

Because
the
trial
intervention
period
was
originally

designed
to
last
2.5
years,
breast
density
measures
were
planned
to
occur
at
0,
1,

and
2
years;
a
smaller
number
of
participants
provided
measures
at
3
years.

6

Mammographic
density
was
measured
in
digitized
mammographic
images
14
.

Mammographic
density
measurements
were
assessed
in
the
scanned
images
by
Dr.

G.
Ursin
(GU)
using
the
University
of
Southern
California
Madena
software,
a

validated
computer-‐assisted,
quantitative
technique
15
.
The
percent
mammographic

density
was
defined
as
the
absolute
density
divided
by
the
total
area
of
the
breast.

The
total
area
of
the
breast
was
first
assessed
by
a
research
assistant
trained
by
Dr.

Ursin,
using
a
computerized
outlining
tool
16
.
The
absolute
density
was
assessed
by

outlining
all
dense
areas
within
the
breast,
except
for
white
artifacts,
prominent

fibrous
strands,
vasculature
or
the
pectoralis
muscle
17
.
All
readers
were
masked
to

the
treatment
group
and
the
order
of
the
mammograms.

Blood
and
urine
isoflavone
measurement

Daidzein
(DE),
genistein
(DHGE),
glycitein
(GLYE),
equol
(EQ),

dihydrodaidzein
(DHDE),
dihydrogenistein
(DHGE),
and
O-‐desmethylangolensin

(DMA)
were
quantified
from
plasma
and
urine
by
HPLC
with
isotope
dilution

electrospray
ionization
(negative
mode)
tandem
mass
spectrometry.
Overnight

urine
samples
were
collected
at
the
baseline
and
every
6-‐month
clinic
visits.

Overnight
urine
collection
began
in
the
previous
night
of
each
visit
and
continued

for
the
entire
night;
subjects
recorded
the
time
of
emptying
their
bladders
before

retiring,
without
storing
this
void.

The
first-‐morning
urine
was
collected,
and
the

time
was
also
recorded.
All
specimens
were
obtained
during
a
fasting
state,
and

immediately
processed
and
stored
at
−80°C
13
.

Spot
urine
was
the
next
urinary
void

7

after
the
first
morning
urine
of
the
overnight
urine
collection,
and
fasting
blood

samples
were
collected
at
all
clinic
visits
18
.

Statistical
analysis

Treatment
group
comparisons
on
percent
breast
density
(PBD)
were

performed
using
the
intent-‐to-‐treat
approach;
analyses
were
conducted
according

to
treatment
assignment,
regardless
of
compliance.
Baseline
characteristics
of

participants
were
compared
between
ISP
and
placebo
groups;
continuous
variables

were
compared
by
independent-‐sample
t
tests
and
categorical
variables
were

compared
by
chi-‐square
tests.
Within
each
treatment
groups,
baseline

characteristics
were
also
compared
between
participants
with
at
least
one

mammographic
follow-‐up
and
without
follow
up,
using
the
same
methods
as
above.

After
excluding
the
multiple
measurements,
normality
was
evaluated
for
the

percent
breast
density
outcome
variable,
the
percent
breast
density
change
from
the

baseline
(on-‐trial
percent
breast
density
for
each
year
–
baseline
percent
breast

density),
area
of
dense
tissue
and
total
breast
area.
The
percent
breast
density,
area

of
dense
tissue
and
total
breast
area
were
not
normally
distributed
(Figure1-‐4,

appendix);
the
square-‐root
transformed
breast
density
best
approximated
a
normal

distribution,
as
did
the
square-‐root
transformed
area
of
dense
tissue
and
square-‐
root
transformed
total
breast
area
(Figure5-‐8,
12-‐13,
appendix).

The

untransformed
percent
breast
density
change
from
the
baseline
was
normally

distributed
(Figure
9-‐11,
appendix).

Subsequent
treatment
group
comparisons

8

therefore
used
the
square
root-‐transformed
values
of
the
percent
breast
density
and

untransformed
values
of
the
percent
breast
density
change.

To
examine
other
known
associations
with
breast
density,
and
given
that

there
were
no
baseline
differences
other
than
parity
between
the
treatment
groups,

and
that
treatment
interventions
had
not
yet
been
initiated,
the
relationships

between
age,
BMI,
race,
parity
status,
and
percent
density
were
tested
at
the

baseline
measurement
in
all
participants
combined,
using
linear
regression
models.

We
also
built
a
multivariate
model
for
baseline
percent
breast
density
and
evaluated

the
normality
of
residuals.

As
an
initial
analysis,
mammographic
density
was
compared
between

treatment
groups
at
each
annual
measurement
using
independent-‐sample
t
tests.

We
also
compared
the
mean
change
from
baseline
in
breast
density
at
each
annual

visit
using
independent
t
tests.
Some
subjects
had
multiple
measurements
of
breast

density
at
the
same
visit
(Table
1,
Appendix),
and
for
those
observations,
we

analyzed
in
two
ways:
(1)
using
the
last
measurement
obtained
at
each
visit

(assuming
this
was
the
“best”
quality
measurement
obtained),
and
(2)
averaging
the

multiple
measurements.

In
a
full
analysis
using
all
trial
data,
linear
mixed
effects
models
were
used
to

compare
treatment
groups
by
on-‐trial
level
of
square-‐root
percent
breast
density

and
change
of
percent
breast
density
from
the
baseline
(using
the
last
measurement

for
multiple
measurements).
At
a
given
visit,
on-‐trial
square-‐root
percent
breast

density
at
each
post-‐randomization
annual
visit
was
regressed
on
randomized

treatment,
adjusting
for
follow-‐up
time
(as
a
class
variable;
1-‐year,
2-‐years,
3-‐years

9

since
randomization),
the
randomization
stratification
factor
(dichotomous
baseline

CIMT)
and
baseline
square-‐root
percent
breast
density.
A
random
effect
was

specified
for
the
subject-‐specific
intercept.
An
interaction
term,
treatment*follow-‐up

time,
was
evaluated
to
determine
whether
the
treatment
effects
on
on-‐trial
percent

density
differed
by
year
since
randomization
(1-‐year,
2-‐years,
3-‐years).
Models
for

the
change
in
percent
breast
density
from
the
baseline,
square-‐root
transformed

area
of
dense
tissue
and
square-‐root
transformed
total
breast
area
as
outcome

variables
were
built
in
the
same
way.

As
a
metabolite
of
daidzen,
equol
is
not
produced
by
all
individuals
and

equol-‐producing
status
may
have
an
effect
on
breast
density.

Linear
mixed
effects

models
were
also
used
to
test
for
a
possible
ISP
effect
on
breast
density
according
to

equol
production.
On-‐trial
square-‐root
percent
breast
density
at
each
post-‐
randomization
annual
visit
was
regressed
on
equol
producer
status,
adjusting
for

follow-‐up
time,
dichotomous
baseline
CIMT
and
baseline
percent
breast
density.

The
change
in
percent
breast
density
from
the
baseline
was
analyzed
in
the
same

way.
Among
ISP-‐treated
subjects,
equol
producer
status
was
defined
based
on

previous
suggestions
in
the
literature
13
,
using
plasma
equol
levels:

(1) equol
nonproducer,
defined
as
equol
level
less
than

20
nmol/L
at
each
of
the

trial
follow-‐up
measurements;

(2) equol
intermittent
producer,
defined
as
equol
level
>20
nmol/L
at
baseline

and/or
at
one
of
the
follow-‐up
measurements;

(3) equol
consistent
producer,
defined
as
equol
level
>20
nmol/L
at
each
of
the

follow-‐up
measurements.

10

Total
isoflavone
levels,
including
daidzein,
genistein,
glycitein,

and
equol

(EQ),
were
analyzed
in
overnight
urine.
Mixed
effects
models
evaluated
the

association
of
on-‐trial
percent
breast
density
with
on-‐trial
urine
isoflavone
and

equol
levels
(all
modeled
as
continuous
variables)
over
visits.
Since
the
regression

coefficient
for
glycitein
was
marginally
statistically
significant,
glycitein
was

categorized
into
quartiles,
and
modeled
as
a
categorical
variable,
using
the
same

method
as
above.
The
association
of
change
in
percent
breast
density
from
the

baseline
with
on-‐trial
urine
isoflavone
and
equol
levels
over
visits
were
examined
in

the
same
way.

Data
were
analyzed
using
SAS
9.0
statistical
software
(SAS
Institute
Inc.,
Cary,

NC).
Two-‐sided
P-‐values
and
significance
level
of
0.05
were
used
for
all
analyses.

RESULTS

Baseline
Characteristics

Of
1,063
women
screened
for
eligibility,
350
women
were
enrolled
and

randomly
assigned
to
a
treatment
group
(175
ISP
and
175
placebo);
307
women

(151
ISP,
156
placebo)
with
baseline
or
follow-‐up
mammographic
data
were

included
in
these
analyses
(4
of
these
subjects
had
no
baseline
mammographic

measurement
but
had
at
least
one
post-‐randomization
mammographic
value).

Most
of
the
demographic
and
clinical
characteristics
of
the
participants
in
the

ISP
treatment
group
were
not
statistically
different
from
the
placebo
group
(Table

1),
except
for
the
number
of
births
among
women
who
were
parous
(p=0.004).
The

11

average
age
was
60.6
years.
Median
percent
breast
density
was
16.9%.
Participants

were
primarily
non-‐Hispanic
whites
(61.9%).
Most
participants
did
not
smoke

(97.7%),
were
high
school
graduates
(94.1%),
and
were
on
average
mildly

overweight
(mean
BMI
26.4
kg/m
2
).
More
than
half
of
the
women
had
taken

hormone
therapy
before
the
trial
(70.0%).

Within
each
treatment
group,
baseline
characteristics
were
not
significantly

different
between
participants
with
at
least
one
follow
up
and
those
without
follow

up,
except
for
total
breast
area.
(p=0.003
for
ISP
group,
p=0.0002
for
placebo
group

(Table
1b)

Associations
between
Percent
Breast
Density
and
Age,
Race
and
BMI
at

baseline

In
prior
studies,
percent
breast
density
was
negatively
associated
with
age,

BMI
and
parity
status
19,20
,
and
also
significantly
differed
by
race
21
.
We
tested
these

associations
in
the
WISH
baseline
data.

In
univariate
analysis
(Table
2a),
age

(p=0.03),
BMI
(p<0.0001),

and
parity
status
(p<0.0001)
were
significantly

negatively
associated
with
baseline
percent
breast
density.

The
baseline
percent

breast
density
also
differed
by
race
groups
(p=0.0003).

In
the
multivariate
model

when
these
four
variables
were
mutually
adjusted,
age
was
no
longer
associated

with
percent
breast
density
after
adjusted
for
BMI,
parity
and
race
(p=0.07).

However,

BMI,
parity,
and
race
remained
significantly
associated
with
percent

breast
density
(Table
2b).

12

Effects
of
Soy
Treatment
on
Percent
Breast
Density

In
the
total
sample,
on-‐trial
percent
breast
density
differed
significantly
over

visits
(p=0.001).
Percent
breast
density
increased
in
the
second
year
of
follow-‐up,

but
decreased
in
the
third
year
(Table
3).
The
difference
of
change
in
percent
breast

density
over
the
three
follow-‐up
visits
was
similarly
statistically
significant

(p=0.0048)
(Table
4).

Means
of
square
root
transformated
percent
breast
density
and
percent

breast
density
change
from
the
baseline
were
compared
between
treatment
and

placebo
at
each
annual
visit
(Tables,
5
and
6).
There
was
no
statistically
significant

difference
in
either
of
these
outcomes
between
the
two
groups.

In
linear
mixed
effects
models
adjusting
for
baseline
CIMT
strata
and

baseline
percent
breast
density,
the
mean
(SE)
square-‐root
transformed
percent

breast
density
measured
over
the
trial
follow-‐up
for
the
ISP
group
was
0.02(0.11)

higher
than
placebo
(p=0.83)
(Table
3).
Mean
(95%
CI)
square-‐root
transformed

percent
breast
density
for
the
ISP
group
was
4.07
(3.90-‐4.23),
and
for
the
placebo

group
was
4.04
(3.88-‐4.20).
Effects
of
soy
treatment
on
breast
density
did
not
differ

over
the
trial
follow-‐up
(Table
7;
p-‐value
for
treatment*measurement
year
=
0.74).

Mean
square-‐root
transformed
percent
breast
density
(95%CI)
for
the
ISP
group
at

each
annual
measurement
was
4.01
(3.84-‐4.18),
4.19
(4.01-‐4.27),
and
4.02
(3.80-‐
4.23),
respectively;
for
the
placebo
group
was
4.02
(3.85-‐4.20),
4.14
(3.97-‐4.32),
and

3.94
(3.74-‐4.15).
Similarly,
change
in
percent
breast
density
from
the
baseline
did

not
significantly
differ
between
the
treatment
groups
(p=0.77)
(Table
4).
Effects
of

13

soy
treatment
on
the
change
of
breast
density
also
did
not
differ
over
the
trial

follow-‐up
(Table
8;
p-‐value
for
treatment*measurement
year
=
0.55).

Square-‐root
transformed
dense
tissue
area
and
square-‐root
transformed

total
breast
area
did
not
differ
between
the
treatment
groups;
both
variables

differed
significantly
over
visits
(p=0.03,
p=0.004,
respectively)
(Table
9-‐10).

Effects
of
Soy
Treatment
on
Percent
Breast
Density:
Stratification
on
Equol

Production

Of
the
154
participants
randomized
to
the
ISP
treatment
group,
33
(21.43%)

were
classified
as
equol
producers,
33
(21.43%)
were
classified
as

intermittent

producers,
and
69
(44.81%)
were
classified
as

non-‐equol
producers.
19
of
154

(12.34%)
of
the
ISP
subjects
had
no
equol
measurements
so
that
equol
producer

status
could
not
be
determined.
Breast
density
changes
were
compared
among
ISP-‐
treated
equol
producer
groups
with
women
receiving
placebo
in
linear
mixed
effects

models.

There
were
no
significant
group
differences
in
on-‐trial
percent
breast

density
(p=0.73)
or
in
percent
breast
density
change
(p=0.82),
after
adjusting
for

baseline
percent
breast
density
and
the
randomization
stratification
factor
(CIMT)

(Table
11,
12).

Association
of
percent
breast
density
with
isoflavone
levels
over
the
trial

In
mixed
effects
models
that
evaluated
the
association
of
on-‐trial
urine

isoflavone
(daidzein,
genistein,
glycitein,
total
IFL)
and
equol
with
breast
density

over
visits,
only
glycitein
had
a
marginally
significant
relationship
with
square-‐root

14

transformed
percent
breast
density
(p=0.06).
When
glycitein
was
categorized
into

quartiles
and
modeled
as
a
categorical
variable;
its
association
with
square-‐root

transformed
percent
breast
density
was
not
significant
(Table
13),
but
there
was
a

marginally
significant
trend
in
percent
breast
density
among
glycitein
groups
(trend

test
p=0.07).
On-‐trial
urine
isoflavones
(daidzein,
genistein,
glycitein,
total
IFL)
and

equol
were
not
associated
with
change
in
breast
density
over
visits
(Table
14).

DISCUSSION

In
this
trial
conducted
among
postmenopausal
women,
randomized
ISP

treatment
did
not
significantly
alter
percent
breast
density,
compared
to
placebo.

In

addition,
most
of
the
isoflavone
levels
(daidzein,
genistein,
equol
and
total
IFL)
did

not
have
significant
associations
with
percent
breast
density.
However,
glycitein

was
marginally
significantly
positively
associated
with
increased
percent
breast

density.

Previous
studies
have
also
shown
that
soy
treatment
did
not
have
a

beneficial
or
a
deleterious
effect
on
breast
density,
consistent
with
our
results
in

WISH.

For
example,
a
randomized,
double
blinded,
placebo
controlled
clinical
trial

(OPUS
trial)
assigned
406
postmenopausal
women
to
three
groups,
consuming
80

or
120
mg/d
of
isoflavones
or
a
daily
placebo
for
2
years.

Treated
and
placebo

subjects
did
not
significantly
differ
on
breast
density,
even
though
breast
density

reduced
significantly
over
the
2-‐year
period
22
.

Although
WISH
had
a
higher
total

dose
of
isoflavones

(154
mg
total
ISP
twice
daily),
and
longer
follow-‐up,
we
also

15

failed
to
detect
a
significant
difference
of
breast
density
or
change
in
breast
density

between
treatment
groups.
We
did
find
a
significant
difference
in
breast
density

over
visits,
but
unlike
the
OPUS
study,

percent
breast
density
in
the
WISH
trial

increased
in
the
second
year
and
decreased
in
the
third
year.
This
might
be
due
to

the
fact
that
fewer
subjects
were
in
our
cohort
at
the
second
and
third
years
of
the

study.
We
also
analyzed
the
same
mixed
model
among
participants
who
completed

all
three
visits,
and
found
that
percent
breast
density
also
increased
significantly

with
over
visits
in
this
sub-‐sample
(p=0.04)
(Appendix
Table
5).

Evidence
from
human
and
animal
studies
has
also
supported
the
finding
that

major
soy
isoflavones
have
no
beneficial
effect
on
breast
cell
proliferation.
In
one

randomized,
double-‐blinded,
placebo-‐controlled
study,
60
healthy
postmenopausal

women
were
assigned
to
60
mg
isoflavones
or
placebo
daily
for
3
months;

comparisons
of
breast
biopsy
specimens
revealed
that
isoflavones
did
not
affect

expression
levels
of
steroid
receptors,
estrogen
receptors
or
progesterone

receptors
23
.
Some
animal
studies
even
showed
an
adverse
effect
on
breast
cell

proliferation;
for
example,
dietary
genistein
stimulated
mammary
gland
growth
and

estrogen-‐dependent
MCF-‐7
tumor
growth
in
ovariectomized
athymic
mice,

expressing
estradiol
(E2)
blood
levels
similar
to
those
found
in
menopausal

women
24
.
Findings
from
WISH,
pertaining
to
associations
of
breast
density
with

urine
levels
of
isoflavones

are
consistent
with
these
studies.
We
detected
a

marginally
significant
association
of
glycitein
on
breast
density,
with
an
increase
of

glycitein
levels
in
urine
associated
with
higher
mean
breast
density.

Contrary
to
our

findings,
a
case-‐control
study
comparing
glycitein
levels
in
breast
cancer
cases
with

16

control
women
25

reported
that

glycitein

had
a
stronger
protective
effect
on
breast

cancer
risk
than
did
other
isoflavone
components,
such
as
genistein.

Since
the

evidence
from
case-‐control
studies
are
weaker
than
randomized
controlled
trials

and
the
glycitein
level
in
their
study
was
much
lower
than
daidzen
and
genisten

levels,
the
significance
of

findings
in
glycitein
is
unclear
but
warrants
further

investigation.

Equol,
a
metabolite
of
daidzein
produced
by
human
intestinal
microflora,
is

not
produced
by
all
individuals
26
.
It
has
been
suggested
that
equol-‐producing
status

may
effect
breast
density
and
those
effects
depend
on
the
ability
to
bio-‐transform

daidzein
to
the
metabolite
equol
26
.
Thus,
we
examined
the
percent
breast
density

change
in
the
placebo
group
to
ISP-‐treated
subjects
by
equol
production
(equol

nonproducers
and
equol
producers).

Comparing
placebo
to
the
treatment
groups

defined
by
equol-‐producing
status,
we
found
no
significant
difference
in
percent

breast
density.

The
WISH
trial,
a
3-‐year
randomized,
double
blinded,
placebo
controlled
trial,

was
a
large
and
long
ISP
supplement
trial.
The
soy
isoflavone
was
provided
to

healthy
postmenopausal
women
in
quantities
within
the
upper
range
of
those
of

traditional
Asian
diets
27
,
so
the
dose
was
high,
tolerable,
and
without
serious

adverse
events.

Participants
were
tested
for
plasma
concentrations
and
the

consumed
study
products
were
counted;

compliance
with
ISP
was
confirmed
to
be

excellent
(approximately
90%)
13
.
Our
results
also
confirmed
the
inverse

relationships
of
percent
breast
density
with
age
19
,
parity,
and
BMI
20
.
A
prospective

review
of
data,
including
15,292
women
presenting
for
mammography,
showed
that

17

breast
density
might
be
greater
in
Asian
women
and
least
in
African
American

women
21
.
In
our
data,
Asian
women
also
had
higher
percent
breast
density

compared
to
other
races;
the
differences
were
significant
between
Asian
women

compared
to
White
and
Hispanic
women.
Due
to
the
small
sample
size
of
African

American
women
(n=19),
their
percent
breast
density
was
not
significantly
less

than
the
breast
density
for
women
of
other
races.

As
the
primary
objectives
of
the
WISH
trial
were
designed
to
test
the
efficacy

of
ISP
supplementation
among
postmenopausal
women,
our
results
are
limited
to

effects
in
postmenopausal
women
and
are
not
generalizable
to
younger
pre-‐
and

peri-‐menopausal
women.

Other
limitations
on
study
design
and
implementation

also
may
have
impacted
our
results.
Many
subjects
did
not
contribute
to
the

mammographic
measurement,
especially
at
the
third
year
point
(n=95).
Moreover,

effects
of
soy
consumption
on
breast
density
may
need
a
longer
time
period
to

appear.
Because
of
the
small
sample
size
and
limited
duration
of
intervention,
we

might
not
have
been
able
to
detect
a
significant
result
with
small
percent
breast

density
change.

Because
of
the
complexity
of
diet
in
different
populations
of
women,
the

protective
effects
of
ISP
observed
in
epidemiological
studies
might
result
from

dietary
components
other
than
soy
isoflavones,
or
interactions
of
soy
isoflavones

with
other
dietary
components.
Therefore,
the
effects
of
isoflavones
warrant
further

mechanism-‐based
studies
or
larger
and
longer
trials.

18

CONCLUSION

ISP
supplementation
over
a
3-‐year
period
did
not
significantly
reduce
the

percent
breast
density
in
postmenopausal
women.
ISP
supplementation
is
safe
and

can
have
high
compliance
over
a
long
time
period.
Analysis
of
isoflavones
suggested

a
positive
association
between
glycitein
and
breast
density,
which
requires
further

animal
or
human
studies.
These
results
should
be
verified
by
future
studies.

19

REFERENCES:

1.
Wu
AH,
Yu
MC,
Tseng
CC,
Pike
MC.
Epidemiology
of
soy
exposures
and
breast

cancer
risk.
Br
J
Cancer
2008;98:9-‐14.

2.
Nechuta
SJ,
Caan
BJ,
Chen
WY,
et
al.
Soy
food
intake
after
diagnosis
of
breast

cancer
and
survival:
an
in-‐depth
analysis
of
combined
evidence
from
cohort
studies

of
US
and
Chinese
women.
Am
J
Clin
Nutr
2012;96:123-‐32.

3.
Kevin
McLaughlin
MS,
D.C.
Soy
Isoflavones
and
Breast
Cancer.
The
Journal
of

Orthomolecular
Medicine
2002;17:10.

4.
Lamartiniere
CA.
Protection
against
breast
cancer
with
genistein:
a

component
of
soy.
Am
J
Clin
Nutr
2000;71:1705S-‐7S;
discussion
8S-‐9S.

5.
Sickles
E,
D’Orsi
CJ,
Bassett
LW,
et
al.
ACR
BI-‐RADS®
Mammography.
In:
ACR

BI-‐RADS®
Atlas,
Breast
Imaging
Reporting
and
Data
System.
Reston,
VA:
American

College
of
Radiology;
2013.

6.
Boyd
NF,
Martin
LJ,
Yaffe
MJ,
Minkin
S.
Mammographic
density
and
breast

cancer
risk:
current
understanding
and
future
prospects.
Breast
Cancer
Res

2011;13:223.

7.
Santen
RJ,
Boyd
NF,
Chlebowski
RT,
et
al.
Critical
assessment
of
new
risk

factors
for
breast
cancer:
considerations
for
development
of
an
improved
risk

prediction
model.
Endocr
Relat
Cancer
2007;14:169-‐87.

8.
Maskarinec
G,
Pagano
I,
Chen
Z,
Nagata
C,
Gram
IT.
Ethnic
and
geographic

differences
in
mammographic
density
and
their
association
with
breast
cancer

incidence.
Breast
Cancer
Res
Treat
2007;104:47-‐56.

20

9.
Cuzick
J,
Warwick
J,
Pinney
E,
Warren
RM,
Duffy
SW.
Tamoxifen
and
breast

density
in
women
at
increased
risk
of
breast
cancer.
J
Natl
Cancer
Inst
2004;96:621-‐
8.

10.
Hooper
L,
Madhavan
G,
Tice
JA,
Leinster
SJ,
Cassidy
A.
Effects
of
isoflavones

on
breast
density
in
pre-‐
and
post-‐menopausal
women:
a
systematic
review
and

meta-‐analysis
of
randomized
controlled
trials.
Hum
Reprod
Update
2010;16:745-‐60.

11.
Henderson
VW,
St
John
JA,
Hodis
HN,
et
al.
Long-‐term
soy
isoflavone

supplementation
and
cognition
in
women:
a
randomized,
controlled
trial.
Neurology

2012;78:1841-‐8.

12.
Hodis
HN,
Mack
WJ,
Kono
N,
et
al.
Isoflavone
soy
protein
supplementation

and
atherosclerosis
progression
in
healthy
postmenopausal
women:
a
randomized

controlled
trial.
Stroke
2011;42:3168-‐75.

13.
Franke
AA,
Hebshi
SM,
Pagano
I,
Kono
N,
Mack
WJ,
Hodis
HN.
Urine

accurately
reflects
circulating
isoflavonoids
and
ascertains
compliance
during
soy

intervention.
Cancer
Epidemiol
Biomarkers
Prev
2010;19:1775-‐83.

14.
Lee
E,
Luo
J,
Su
YC,
et
al.
Hormone
metabolism
pathway
genes
and

mammographic
density
change
after
quitting
estrogen
and
progestin
combined

hormone
therapy
in
the
California
Teachers
Study.
Breast
Cancer
Res
2014;16:477.

15.
Greendale
GA,
Reboussin
BA,
Sie
A,
et
al.
Effects
of
estrogen
and
estrogen-‐
progestin
on
mammographic
parenchymal
density.
Postmenopausal

Estrogen/Progestin
Interventions
(PEPI)
Investigators.
Ann
Intern
Med

1999;130:262-‐9.

21

16.
Wu
AH,
Spicer
D,
Garcia
A,
et
al.
Double-‐blind
randomized
12-‐month
soy

intervention
had
no
effects
on
breast
MRI
fibroglandular
tissue
density
or

mammographic
density.
Cancer
Prev
Res
(Phila)
2015.

17.
Ozhand
A,
Lee
E,
Wu
AH,
et
al.
Variation
in
inflammatory
cytokine/growth-‐
factor
genes
and
mammographic
density
in
premenopausal
women
aged
50-‐55.

PLoS
One
2013;8:e65313.

18.
Franke
AA,
Lai
JF,
Halm
BM,
et
al.
Equol
production
changes
over
time
in

postmenopausal
women.
J
Nutr
Biochem
2012;23:573-‐9.

19.
Checka
CM,
Chun
JE,
Schnabel
FR,
Lee
J,
Toth
H.
The
relationship
of

mammographic
density
and
age:
implications
for
breast
cancer
screening.
AJR
Am
J

Roentgenol
2012;198:W292-‐5.

20.
Buist
DS,
Anderson
ML,
Taplin
SH,
LaCroix
AZ.
The
relationship
between

breast
density
and
bone
mineral
density
in
postmenopausal
women.
Cancer

2004;101:1968-‐76.

21.
del
Carmen
MG,
Halpern
EF,
Kopans
DB,
et
al.
Mammographic
breast
density

and
race.
AJR
Am
J
Roentgenol
2007;188:1147-‐50.

22.
Maskarinec
G,
Verheus
M,
Steinberg
FM,
et
al.
Various
doses
of
soy

isoflavones
do
not
modify
mammographic
density
in
postmenopausal
women.
The

Journal
of
nutrition
2009;139:981-‐6.

23.
Cheng
G,
Wilczek
B,
Warner
M,
Gustafsson
JA,
Landgren
BM.
Isoflavone

treatment
for
acute
menopausal
symptoms.
Menopause
2007;14:468-‐73.

22

24.
Rietjens
IM,
Sotoca
AM,
Vervoort
J,
Louisse
J.
Mechanisms
underlying
the

dualistic
mode
of
action
of
major
soy
isoflavones
in
relation
to
cell
proliferation
and

cancer
risks.
Mol
Nutr
Food
Res
2013;57:100-‐13.

25.
Dai
Q,
Franke
AA,
Jin
F,
et
al.
Urinary
excretion
of
phytoestrogens
and
risk
of

breast
cancer
among
Chinese
women
in
Shanghai.
Cancer
Epidemiol
Biomarkers

Prev
2002;11:815-‐21.

26.
Tseng
M,
Byrne
C,
Kurzer
MS,
Fang
CY.
Equol-‐producing
status,
isoflavone

intake,
and
breast
density
in
a
sample
of
US
Chinese
women.
Cancer
Epidemiology

Biomarkers
&
Prevention
2013;22:1975-‐83.

27.
Messina
M,
Nagata
C,
Wu
AH.
Estimated
Asian
adult
soy
protein
and

isoflavone
intakes.
Nutr
Cancer
2006;55:1-‐12.

23

FIGURES

Figure
1.
Flow
diagram

24

TABLES

Table
1.
WISH
participant
characteristics
at
baseline

a

ISP(n=156)
Placebo(n=151)
p

Age
at
randomization
60.7
(7.1)
60.5
(6.7)
0.84

Race

White
non-‐Hispanic
91(58.3)
91
(65.6)
0.65

Black
non-‐Hispanic
11
(7.1)

8(5.3)

Hispanic
29
(18.6)
20
(13.2)

Asian
or
Pacific
Islander
19
(12.2)
19
(12.6)

Other
6
(3.9)
5
(3.3)

Smoking
status

Current
3
(1.9)
4
(2.6)
0.61

Former
64
(41.0)
54
(35.8)

Never
89
(57.1)
93
(61.6)

Highest
education
level

High
school
or
less
14
(9.0)
4
(2.7)
0.02

Above
high
school
142
(91.0)
147
(97.3)

BMI
(kg/m
2
)
26.4
(4.9)
26.4
(5.1)
0.99

Ever
took
menopausal

hormones

Yes
115
(73.7)
110
(66.2)
0.15

No
41
(26.3)
51
(33.8)

Type
of
menopause

Natural
146
(93.6)
135
(90.0)
0.25

Surgical
10
(6.4)
15
(10.0)

Age
at
menopause
49.2
(4.7)
49.5
(4.5)
0.55

Years
since
menopause
11.4(8.1)
10.8
(7.7)
0.53

Parity
status

Nulliparous
31(21.5)
21(14.4)
0.11

Parous
113
(78.5)
125
(85.6)

Number
of
birth
among

parous
women

2.59
(1.19)
2.19
(0.96)
0.004

Percent
breast
density
b

17.9
(5.78-‐32.72)
c

15.9
(5.41-‐33.12)
0.85

Dense
area

54632.0
(17154.0-‐
106024.0)

48902.0
(10986.0-‐
127050.0)

0.92

Total
breast
area

359131.5

(212345.0-‐
635840.0)

382989.0

(214907.0-‐
632153.0)

0.75

a. Mean
(SD)
for
continuous
variables
and
no.
(%)
for
categorical
variables.

b. If
subjects
had
multiple
measurements
of
breast
density
at
baseline,
we
used
the

last
measurement
(assuming
this
was
the
“best”
quality
measurement
obtained).

25

c. Median
(IQR)
for
dense
area,
total
breast
area
and
baseline
percent
density,

compared
between
treatment
groups
using
Wilcoxon
rank
sum
test.

26

Table
1b.
Comparisons
of
baseline
characteristics
among
participants
with

and
without
follow
up
mammographic
measurement
a

a. Mean
(SD)
for
continuous
variables
and
no.
(%)
for
categorical
variables.

b. If
subjects
had
multiple
measurements
of
breast
density
at
baseline,
we
used

the
last
measurement
(assuming
this
was
the
“best”
quality
measurement

obtained).

c. Median
(IQR)
for
baseline
percent
density,
compared
between
treatment

groups
using
Wilcoxon
rank
sum
test.

ISP

with
follow
up

(n=130)

ISP

without
follow

up

(n=26)

P
Placebo

with
follow
up

(n=125)

Placebo

without
follow

up

(n=26)

P

Age
at
randomization
60.8
(6.9)
60.2
(8.0)
0.71
60.5
(6.8)
60.7
(6.3)
0.88

Race

White
non-‐Hispanic
74
(56.9)
17
(65.4)
0.89
83
(66.4)
16
(66.5)
0.76

Black
non-‐Hispanic
9
(6.9)
2
(7.7)
6
(4.8)
2
(7.7)

Hispanic
26
(20.0)
3
(11.5)
15
(12.0)
5
(19.2)

Asian
or
Pacific
Islander
16
(12.3)
3
(11.5)
17
(13.6)
2
(7.7)

Other
5
(3.9)
1
(3.9)
4
(3.2)
1
(3.9)

Smoking
status

Current
2
(1.5)
1
(3.9)
0.39
3
(2.4)
1(3.9)
0.56

Former
51
(39.2)
13
(50.0)
47(37.6)
7(26.9)

Never
77
(59.2)
12
(46.2)
75
(60.0)
18
(69.2)

Highest
education
level

High
school
or
less
11
(8.5)
3
(11.5)
0.62
4
(3.2)
0
(0)
0.35

Above
high
school
23
(88.5)
119
(91.5)
121
(96.8)
26
(100)

BMI
(kg/m
2
)
26.0
(4.6)
28.3
(6.0)
0.03
26.3
(5.0)
26.9
(5.4)
0.61

Ever
took
menopausal

hormones

Yes
98
(75.4)
17
(65.4)
0.29
85(68.0)
15(57.7)
0.31

No
32
(24.6)
9
(34.6)
40
(32.0)
11
(42.3)

Type
of
menopause

Natural
123
(94.6)
23
(88.5)
0.24
110
(88.7)
25
(96.2)
0.25

Surgical
7
(5.4)
3
(11.6)
14
(11.3)
1
(3.9)

Age
at
menopause
49.3
(4.7)
48.6
(4.8)
0.48
49.6
(4.6)
49.3
(3.5)
0.76

Years
since
menopause
11.4
(8.0)
11.4
(8.6)
0.98
10.8
(7.9)
10.9
(6.5)
0.95

Parous
status

nulliparous
18
(14.9)
3
(12.0)
0.71
24
(20.3)
7
(26.9)
0.46

parous
103
(85.1)
22
(88.0)
94
(79.7)
19
(73.1)

Number
of
birth
among

parous
women

2.62
(1.25)
2.45
(0.80)
0.43
2.16
(0.98)
2.32
(0.89)
0.52

Dense
area
60282.5

(17324.5-‐
122086.3)

45349.0

(16348.0
-‐
75205.9)

0.40
61484.9

(9268.0-‐
142537.0)

25702.5

(17243.0-‐
67577.0)

0.14

Total
breast
area
426720.0

(250141.5-‐
683807.0)

236159.5

(176271.0-‐
343891.0)

0.003
452446.0

(252146.0-‐
684823.0)

240111.0

(200100.0-‐
319011.0)

0.0002

Percent
breast
density
b

18.2
(5.7-‐

31.0)
c

18.8
(7.83-‐
39.5)

0.41
13.0
(5.6-‐
31.2)

17.7
(5.3-‐33.3)
0.52

27

Table
2a.
Correlation
between
percent
breast
density
and
age,
race
and
BMI

measured
at
baseline
a

Characteristic
Beta
(SE)

P

Age
-‐0.04
(0.02)
0.03

BMI
-‐0.16
(0.02)
<0.0001

Parity
status

Nulliparous
-‐
<0.0001

Parous
-‐1.24
(0.31)

Race

White
non-‐Hispanic
-‐
0.0003

Black
non-‐Hispanic
0.07
(0.48)

Hispanic
-‐0.67
(0.32)

Asian
or
Pacific
Islander
1.33
(0.36)

Other
0.28
(0.62)

a.
If
subjects
had
multiple
measurements
of
breast
density
at
the
same
visit,
we
used

the
last
measurement
obtained
at
each
visit
(assuming
this
was
the
“best”
quality

measurement
obtained).

b.
Estimates
of
linear
regressions
on
square-‐root
transformed
percent
breast

density
with
each
of
the
predictors:
age,
BMI,
Parity
and
Race.

28

Table
2b.
Multiple
regression
of
square-‐root
transformed
percent
breast

density
on
age,
BMI,
parity
and
race

a,b

Characteristic

Beta
(SE)
P

Age
-‐0.03
(0.02)
0.07

BMI
-‐0.14
(0.02)
<0.0001

Parity
status

Nulliparous
-‐

Parous
-‐0.18
(0.28)
0.0002

Race

White
non-‐Hispanic
-‐
0.058

Black
non-‐Hispanic
0.69
(0.47)

Hispanic
-‐0.13
(0.31)

Asian
or
Pacific
Islander
0.88
(0.35)

Other
0.45
(0.57)

a.
If
subjects
had
multiple
measurements
of
breast
density
at
the
same
visit,
we
used

the
last
measurement
obtained
at
each
visit
(assuming
this
was
the
“best”
quality

measurement
obtained).

b.
Multivariate
model
of
square-‐root
transformed
percent
breast
density
on
age,

BMI,
parity
and
race

29

Table
3.
Treatment
effects
on
on-‐trial
level
of
square
root
transformed
breast

density:
mixed
effects
models

a,b,
c

Effect
Beta
(SE)

Mean
square-‐root

Percent
Breast
density

(95%
CL)

P

Treatment

Placebo
reference
4.04
(3.88-‐4.20)
-‐

ISP
0.02
(0.11)
4.07
(3.90-‐4.23)
0.83

Measurement
year

(post-‐randomization)

Year
1
reference
4.02
(3.89-‐4.15)
-‐

Year
2
0.15
(0.05)
4.17
(4.04-‐4.30)
0.001

Year
3
-‐0.04
(0.06)
3.98
(3.82-‐4.13)

a. Mixed
effects
model
adjusted
for
baseline
value
of
mammographic
density
and

CIMT
randomization
stratum;
4
subjects
(9
observations)
with
no
baseline

percent
breast
density
value
were
excluded.

b. Dependent
variable
was
square
root
transformed
percent
breast
density.

If
subjects
had
multiple
measurements
of
breast
density
at
the
same
visit,
we

used
the
last
measurement
obtained
at
each
visit
(assuming
this
was
the
“best”

quality
measurement
obtained).

c.

Sample
size:
n=252,
total
subjects;

30

Table
4.
Treatment
effects
on
change
of
breast
density
from
the
baseline:

mixed
effects
models

a,b,c

Effect
Beta
(SE)
P

Treatment

Placebo
reference
-‐

ISP
-‐0.24
(0.81)
0.77

Measurement
year

(post-‐randomization)

Year
1
reference
-‐

Year
2
1.25
(0.44)
0.0048

Year
3
-‐0.30
(0.59)

a. Mixed
effects
model
adjusted
for
baseline
value
of
mammographic
density
and

CIMT
randomization
stratum;
4
subjects
(9
observations)
with
no
baseline

percent
breast
density
value
were
excluded.

b. Dependent
variable
was
change
in
percent
breast
density
from
the
baseline.

If
subjects
had
multiple
measurements
of
breast
density
at
the
same
visit,
we

used
the
last
measurement
obtained
at
each
visit
(assuming
this
was
the
“best”

quality
measurement
obtained).

31

Table
5.
Treatment
groups
comparison
on
square
root
transformed
percent

breast
density

a,b

No.

(ISP/Placebo)

ISP

c

Placebo
P

d

Baseline
303
(154/149)
4.1
(2.0)
4.1(2.2)
0.92

Year
1
219
(113/106)
3.9(2.0)
4.3(2.1)
0.19

Year
2
203
(105/98)
4.0
(2.1)
4.0
(2.1)
0.82

Year
3
95
(44/51)
3.9(2.2)
4.3
(2.2)
0.30

a. If
subjects
had
multiple
measurements
of
breast
density
at
the
same
visit,
we

used
the
last
measurement
obtained
at
each
visit
(assuming
this
was
the
“best”

quality
measurement
obtained).

b. Results
for
analyses
that
averaged
the
multiple
measurements
were
similar

(Table
2,
appendix)

c. ISP
indicates
isoflavone
soy
protein
treatment
group.

Comparisons
were
on
square
root
transformation
of
percent
breast
density

between
treatment
groups
by
independent
t-‐test.
Mean
(SD)
are
means
of
square-‐
root
transformed
percent
breast
density

d. Square
root
transformed
breast
density
was
compared
between
treatment

groups
at
each
annual
measurement
using
independent-‐sample
t
tests

Table
6.
Treatment
group
comparisons
on
change
in
percent
breast
density

from
baseline

a,b

No.
(ISP/Placebo)
ISP
c

Placebo
P

d

Year
1
216
(112/104)
-‐1.8(6.1)
-‐0.8
(8.9)
0.33

Year
2
199
(103/96)
-‐0.2
(5.4)
0.1
(8.4)
0.77

Year
3
93
(43/50)
-‐0.7
(5.5)
-‐2.1
(10.7)
0.45

a. If
subjects
had
multiple
measurements
of
breast
density
at
the
same
visit,
we

used
the
last
measurement
obtained
at
each
visit
(assuming
this
was
the
“best”

quality
measurement
obtained).

b. Results
for
analyses
that
averaged
the
multiple
measurements
were
similar

(Table
3,
appendix)

c. Mean
(SD)
percent
breast
density
change
from
the
baseline

d. Breast
density
change
was
compared
between
treatment
groups
at
each
annual

measurement
using
independent-‐sample
t
tests

32

Table
7.
Treatment
effects
on
on-‐trial
level
of
square
root
transformed
breast

density:
mixed
effect
models
with
test
for
equivalence
of
treatment
effect
over

visits
a,b

Effect
Beta
(SE)
P

Treatment

Placebo
reference
-‐

ISP
-‐0.01
(0.12)
0.92

Treatment

*
years

ISP
*
Year
1
reference
-‐

ISP
*
Year
2
0.06
(0.09)
0.74

ISP
*
Year
3
0.09
(0.13)

Measurement
year

Year
1
reference
-‐

Year
2
0.12
(0.07)
0.0013

Year
3
-‐0.08
(0.09)

a.
Dependent
variable
was
square
root
transformed
percent
breast
density.

If
subjects
had
multiple
measurements
of
breast
density
at
the
same
visit,
we
used

the
last
measurement
obtained
at
each
visit
(assuming
this
was
the
“best”
quality

measurement
obtained).

b.
Sample
size:
n=303,
total
subjects;
n=216,
year1;
n=199,
year2;
n=93,
year3.

33

Table
8.
Treatment
effects
on
change
of
breast
density
from
the
baseline:

mixed
effect
models
with
test
for
equivalence
of
treatment
effect
over
visits
a,b

Effect
Beta
(SE)
P

Treatment

Placebo
reference
-‐

ISP
-‐0.63
(0.93)
0.50

Treatment

*
years

ISP
*
Year
1
reference
-‐

ISP
*
Year
2
0.47
(0.89)
0.55

ISP
*
Year
3
1.30
(1.19)

Measurement
year

Year
1
reference
-‐

Year
2
1.02
(0.64)
0.0057

Year
3
-‐0.92
(0.82)

a.
Dependent
variable
was
change
in
percent
breast
density
from
the
baseline.

If
subjects
had
multiple
measurements
of
breast
density
at
the
same
visit,
we
used

the
last
measurement
obtained
at
each
visit
(assuming
this
was
the
“best”
quality

measurement
obtained).

34

Table
9.
Treatment
effects
on
square-‐root
transformed
area
of
dense
tissue:

mixed
effects
models

a,b,c

Effect
Beta
(SE)
Mean
P

Treatment

Placebo

193.51
(169.45-‐217.58)

ISP
6.65
(16.26)
200.16
(176.07-‐224.25)
0.68

Measurement
year

(post-‐randomization)

Year
1

Year
2
11.33
(4.41)

0.03

Year
3
11.26
(6.47)

a.
Mixed
effects
model
adjusted
for
baseline
area
of
dense
tissue
and
CIMT

randomization
stratum;
51
subjects
(166
observations)
with
no
baseline
dense
area

measurement
value
were
excluded.

b.
Dependent
variable
was
square
root
transformed
percent
breast
density.

If
subjects
had
multiple
measurements
of
breast
density
at
the
same
visit,
we

used
the
last
measurement
obtained
at
each
visit
(assuming
this
was
the
“best”

quality
measurement
obtained).

c.
Sample
size:
n=205,
total
subjects;

35

Table
10.
Treatment
effects
on
square-‐root
transformed
total
breast
area:

mixed
effects
models

a,b,c

Effect
Beta
(SE)
mean
P

Treatment

Placebo

545.32(498.28-‐592.36)

ISP
-‐12.06
(31.68)
533.26(486.08-‐580.44)
0.70

Measurement
year

(post-‐randomization)

Year
1
-‐

Year
2
8.77
(11.08)

0.004

Year
3
53.76
(16.24)

a.
Mixed
effects
model
adjusted
for
baseline
value
of
total
breast
area
and
CIMT

randomization
stratum;
51
subjects
(166
observations)
with
no
baseline
dense
area

measurement
value
were
excluded.

b.
Dependent
variable
was
square
root
transformed
percent
breast
density.

If
subjects
had
multiple
measurements
of
breast
density
at
the
same
visit,
we
used

the
last
measurement
obtained
at
each
visit
(assuming
this
was
the
“best”
quality

measurement
obtained).

c.
Sample
size:
n=205,
total
subjects;

36

Table
11.
Treatment
group
comparisons
on
on-‐trial
square
root
transformed

percent
breast
density
by
equol
producer
status
a

Effect
Beta
(SE)
P
b

P

Treatment

Placebo
reference
-‐

ISP,
nonproducer
-‐0.05(0.13)
0.71
0.73
c

ISP,
intermittent
producer
0.16(0.18)
0.37

ISP,
consistent
producer
0.06(0.17)
0.72

Measurement
year

(post-‐randomization)

Year
1
reference

-‐

Year
2
0.15
(0.04)

0.0009

Year
3
-‐0.04
(0.06)

a. Square
root
transformed
on-‐trial
percent
breast
density
was
used
as

independent
variable
in
the
linear
mixed
effects
model.

If
subjects
had
multiple
measurements
of
breast
density
at
the
same
visit,
we

used
the
last
measurement
obtained
at
each
visit
(assuming
this
was
the
“best”

quality
measurement
obtained).

b. P
value
for
difference
from
referent
group.

c. Overall
p
values
across
ISP
treated
equol
producer
groups
compared
to
the

placebo.

37

Table
12.
Treatment
group
comparisons
on
change
in
breast
density
by
equol

producer
status
a

Effect
Beta
(SE)
P
b

P

Treatment

Placebo
reference

-‐

ISP,
nonproducer
-‐0.78(0.96)
0.42
0.82
c

ISP,
intermittent
producer
0.41(1.34)
0.76

ISP,
consistent
producer
-‐0.16(1.28)
0.90

Measurement
year

(post-‐randomization)

Year
1
reference

-‐

Year
2
1.24
(0.45)

0.0049

Year
3
-‐0.36
(0.60)

a. If
subjects
had
multiple
measurements
of
breast
density
at
the
same
visit,
we

used
the
last
measurement
obtained
at
each
visit
(assuming
this
was
the
“best”

quality
measurement
obtained).

b.P
value
for
difference
from
referent
group.

c.
Overall
p
values
across
ISP
treated
equol
producer
groups
compared
to
the

placebo.

38

Table
13.
Associations
of
urine
isoflavone
levels
with
on-‐trial
square
root

transformed
percent
breast
density.
a

Isoflavones
Beta
(SE)
P
d

P

Daidzen
0.01
(0.01)

0.26

Genistein
0.02
(0.01)

0.17

Glycitein
b

0.14
(0.08)

0.06

Glycitein
Quartiles

<0.14
c

reference
-‐
0.07
e

0.14-‐0.68
0.17
(0.36)
0.63

0.68-‐2.25
0.46
(0.36)
0.20

>2.25
0.58
(0.36)
0.11

Total
isoflavones
0.003(0.003)

0.30

Equol
-‐0.001
(0.01)

0.90

a.
In
all
linear
mixed
effects
models,
dependent
variable
was
square-‐root

transformed
percent
breast
density.

If
subjects
had
multiple
measurements
of
breast
density
at
the
same
visit,
we

used
the
last
measurement
obtained
at
each
visit
(assuming
this
was
the
“best”

quality
measurement
obtained).

b.
Glycitein
was
treated
as
continuous
variable

c.
Glycitein
was
treated
as
categorical
variable,
categorized
in
quartiles.

d.
P
value
for
difference
from
referent
group.

e.
P
value
for
trend
test,
across
glycitein
groups.

39

Table
14.
Associations
of
urine
isoflavone
levels
with
change
of
percent
breast

density
from
the
baseline.
a

Isoflavones
Beta
(SE)
P

Daidzen
-‐0.004
(0.02)
0.86

Genistein
0.006
(0.04)
0.87

Glycitein
-‐0.08
(0.25)
0.74

Total
isoflavones
-‐0.002(0.008)
0.82

Equol
-‐0.01
(0.03)
0.80

a. In
all
linear
mixed
effects
models,
dependent
variable
was
change
of
percent

breast
density
from
the
baseline.

If
subjects
had
multiple
measurements
of
breast
density
at
the
same
visit,
we

used
the
last
measurement
obtained
at
each
visit
(assuming
this
was
the
“best”

quality
measurement
obtained).

Abstract (if available)

Abstract Background: Although epidemiological studies have suggested that soy food consumption may be related to the incidence of breast cancer, the effectiveness of isoflavone soy protein (ISP) supplementation in reducing the risk of breast cancer remains unclear. We assessed the effects of ISP supplementation on a biomarker of breast cancer risk, mammographic density. ❧ Methods: In a double blind, placebo controlled trial, 350 healthy postmenopausal women aged 45 to 92 years were randomly assigned to 2 evenly divided groups: daily 25 g of soy protein (91 mg aglycone isoflavone equivalents) or placebo, for a 2.5-3 year treatment period. The primary outcome was percent mammographic density that was measured at baseline and annually. ❧ Results: 307 women with baseline and follow-up mammograms were included in the intent-to-treat analyses. After adjusting for baseline carotid artery intima-media thickness (CIMT) strata and baseline percent mammographic density, there was no significant difference in the mean square-root transformed percent mammographic density measured over the trial follow-up among the treatment groups (p=0.83). ISP soy supplementation also did not differ at the annual measures over the trial follow-up. Analyses of total urinary isoflavone levels showed no associations with breast density over visit, except for a marginally significant positive association of glycitein with percent mammographic density (p=0.06). ❧ Conclusion: In healthy postmenopausal women, ISP supplementation did not significantly affect mammographic density. Effects of isoflavones, such as glycitein, on mammographic density should be investigated by further studies.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Associations between physical activities with bone mineral density in postmenopausal women

PDF

Association between endogenous sex hormone levels and cognition: the Women’s Isoflavone Soy Health (WISH) trial

PDF

Association of bone density and bone metabolism markers: the Women’s Isoflavone Soy Health (WISH) trial

PDF

Effect of soy isoflavones on anthropometric and metabolic measurements in postmenopausal women

PDF

Relationship of blood pressure and antihypertensive medications to cognitive change in the BVAIT, WISH, and ELITE clinical trials

PDF

The role of heritability and genetic variation in cancer and cancer survival

PDF

Factors that influence mammographic density: role of estrogen metabolism genes, biomarkers of inflammation, and lifestyle

PDF

Effects of post-menopausal hormone therapy on arterial stiffness in the ELITE trial

PDF

A randomized controlled trial of hormone therapy on postmenopausal women’s quality of life

PDF

Soy isoflavone supplements for the treatment of menopausal hot flashes: the Women’s Isoflavone Soy Health (WISH) trial

PDF

Air pollution and breast cancer survival in California teachers: using address histories and individual-level data

PDF

The effects of hormone therapy on carotid artery intima-media thickness and serum lipids by ApoE4 genotype

PDF

An analysis of disease-free survival and overall survival in inflammatory breast cancer

PDF

Pathogenic variants in cancer predisposition genes and risk of non-breast multiple primary cancers in breast cancer patients

PDF

Association of carotid artery stiffness with measures of cognition in older adults

PDF

The effects of tobacco exposure on hormone levels and breast cancer risk among young women

PDF

Diet quality and pancreatic cancer incidence in the multiethnic cohort

PDF

Hormone therapy timing hypothesis and atherosclerosis

PDF

The role of pesticide exposure in breast cancer

PDF

Analysis of factors associated with breast cancer using machine learning techniques