Ethnic differences in the effect of parity on breast cancer incidence in premenopausal women in singapore

B Objectives This thesis proposes that there are ethnic differences in the effect of multiparity on breast cancer incidence in pre-menopausal women in the three major ethnic groups in Si

YAP
PENG
LENG
KAREN


(MBBS(Singapore),
FRCS(Edinburgh),
FRCS(Glasgow),
FAMS(General
Surgery))



A
THESIS
SUBMITTED
FOR
THE
DEGREE
OF
MASTER
OF
SCIENCE



This
project
and
thesis
would
not
have
come
to
fruition
if
not
for
the
contributions
and
assistance
received
from
the
following:


My
supervisor
Professor
Chia
Kee
Seng
for
the
invaluable
instruction
and


direction
of
the
project,
from
its
conceptualization
through
to
the
completion
of
this
thesis;



Helena
Verkooijen
and
Cheung
Kwok
Hang
who
have
been
instrumental
in
retrieving
and
sorting
the
data
from
the
registries;


Professor
Soo
Khee
Chee
and
Professor
London
Lucien
Ooi
who
inspired
me
towards
pursuing
this
course
in
the
first
place
and


Dr
Ann
Lee
for
guiding
me
through
the
rigors
of
laboratory
research.


In
addition,
I
must
not
fail
to
mention
my
understanding
husband
and
parents
who
took
turns
to
mind
the
little
ones
in
order
to
give
me
uninterrupted
time
to
work.


TABLE
OF
CONTENTS


ACKNOWLEDGEMENTS I
 TABLE
OF
CONTENTS II
 SUMMARY V


A
 INTRODUCTION V


B
 OBJECTIVES V


C
 MATERIALS
AND
METHODS V


D
 RESULTS VI


E
 DISCUSSION VI
 LIST
OF
TABLES VII
 LIST
OF
FIGURES IX
 LIST
OF
PRESENTATIONS
AND
PUBLICATIONS
FROM
THIS
STUDY XI


MAIN
BODY
OF
THESIS 1


1
 
INTRODUCTION 1


2
 
LITERATURE
REVIEW 2


2.1
 
BREAST
CANCER
WORLDWIDE:
DIFFERENCES
IN
INCIDENCE 2


2.2
 
FACTORS
PREDISPOSING
TO
BREAST
CANCER 5


2.2.1
 
BIOLOGICAL
FACTORS 5


2.2.2
 
HORMONAL
FACTORS 14


2.2.3
 
ENVIRONMENTAL
FACTORS 19


2.3
 
ETHNIC
DIFFERENCES
IN
BREAST
CANCER
RISK
FACTORS 25


2.3.1
 ETHNIC
DIFFERENCES
IN
BREAST
CANCER
INCIDENCE 25


2.3.2
 ETHNIC
DIFFERENCES
IN
REPRODUCTIVE
RISK
FACTORS
FOR
BREAST
CANCER 27


2.3.3
 ETHNIC
DIFFERENCES
IN
CLINICAL
FEATURES
OF
BREAST
CANCER 29


2.3.4
 ETHNIC
DIFFERENCES
IN
BREAST
CANCER
MORTALITY 30


2.4
 
EPIDEMIOLOGY
OF
BREAST
CANCER
IN
SINGAPORE 32


2.4.1
 
SINGAPORE
POPULATION
DEMOGRAPHICS
AND
ETHNIC
DIVERSITY 32


2.4.2
 
INCIDENCE
OF
BREAST
CANCER
IN
SINGAPORE 32


2.4.3
 
TRENDS
IN
BREAST
CANCER
INCIDENCE 32


2.4.4
 
ETHNIC
DIFFERENCES
IN
BREAST
CANCER
TRENDS 35


2.4.5
 
ETHNIC
DIFFERENCES
IN
AGE‐SPECIFIC
INCIDENCE 36


2.4.6
 
FERTILITY
RATES 37


2.5
 
CLINICAL
ASPECTS
OF
BREAST
CANCER 41


2.5.1
 
CLINICAL
PRESENTATION 41


2.5.2
 
STAGING
OF
BREAST
CANCER 41


2.5.3
 
BREAST
CANCER
TREATMENT 41


3
 
OBJECTIVES
OF
STUDY 42


4
 
MATERIALS
AND
METHODS 43


4.1
 
DATA
SOURCES 43


4.1.1
 
THE
SINGAPORE
NATIONAL
REGISTRY
OF
BIRTHS
AND
DEATHS
(SNRBD) 43


4.1.2
 
THE
SINGAPORE
CANCER
REGISTRY
(SCR) 43


4.2
 
ETHICS
APPROVAL 44


4.3
 
LINKAGE
OF
DATA 44


4.4
 
STUDY
COHORT 44


4.5
 
DEFINITIONS 45


4.6
 
DATA
ANALYSIS 45


5
 
RESULTS 47


5.1
 THE
SINGAPORE
NATIONAL
REGISTRY
OF
BIRTHS
AND
DEATHS
(SNRBD)
1986‐2002 47


5.2
 DESCRIPTION
OF
STUDY
POPULATION 50


5.2.1
 
ETHNIC
DISTRIBUTION 50


5.2.2
 
PARITY 50


5.2.3
 
AGE
AT
FIRST
BIRTH 51


5.2.4
 

AGE
AT
LAST
BIRTH 52


5.2.5
 
FOLLOW‐UP 53


5.2.6
 
BREAST
CANCERS 54


5.2.7
 
AGE
AT
CANCER
DIAGNOSIS 54


5.2.8
 DURATION
BETWEEN
LAST
BIRTH
BEFORE
CANCER
AND
BREAST
CANCER
DIAGNOSIS 54


5.2.9
 
DURATION
BETWEEN
FIRST
BIRTH
AND
CENSOR
DATE 55


5.2.10
 MOTHERS
WHO
HAD
CHILDREN
AFTER
BREAST
CANCER 56


5.3
 ETHNIC
DIFFERENCES
IN
REPRODUCTIVE
RISK
FACTORS
FOR
BREAST
CANCER 57


5.3.1
 
AGE
AT
FIRST
BIRTH
AND
BREAST
CANCER
RISK 57


5.3.2
 
AGE
AT
LAST
BIRTH
AND
BREAST
CANCER
RISK 59


5.3.3
 
PARITY
AND
BREAST
CANCER
RISK 59


5.3.4
 
EFFECT
OF
ETHNICITY
ON
BREAST
CANCER
RISK 59


5.3.5
 
EFFECT
OF
ETHNICITY
AND
PARITY
ON
BREAST
CANCER
RISK 60


6
 
DISCUSSION 65


6.1
 
SUMMARY
OF
MAIN
FINDINGS 65


6.2
 
WHAT
OTHER
STUDIES
HAVE
FOUND 66


6.3
 
POSSIBLE
EXPLANATIONS
FOR
THE
FINDINGS 67


6.3.1
 
TRANSIENT
POST‐PREGNANCY
RISE
IN
BREAST
CANCER
RISK 67


6.3.2
 
BREAST‐FEEDING
PRACTICES 67


6.3.3
 
BODY
MASS
INDEX 67


6.3.4
 
AGE
AT
FIRST
BIRTH 68


6.3.5
 
HORMONAL
RECEPTOR
SUBTYPE 69


6.3.6
 
ALPHA

FETOPROTEIN 69


6.4
 
STRENGTHS
OF
STUDY 70


6.5
 
LIMITATIONS
OF
STUDY 72


7
 
CONCLUSION 73


8
 
BIBLIOGRAPHY 74


APPENDICES 90


APPENDIX

A
 
DEFINITIONS
USED 90


APPENDIX

B
 
ABBREVIATIONS 91


APPENDIX
C
 
LIST
OF
VARIABLES
IN
DATASET 92


APPENDIX
D
 
COMMANDS
USED
IN
STATA 93


APPENDIX
E
 
ETHICS
APPROVAL 99


APPENDIX
F
 
PUBLICATION 101
 


B
 Objectives


This
thesis
proposes
that
there
are
ethnic
differences
in
the
effect
of
multiparity


on
breast
cancer
incidence
in
pre-menopausal
women
in
the
three
major
ethnic
groups
in
Singapore.


C
 Materials
and
Methods


Through
the
Singapore
National
Registry
of
Births
and
Deaths,
women
who
had


a
first
childbirth
in
the
years1986-2002
were
linked
with
the
Singapore
Cancer
Registry
to
ascertain
if
they
had
breast
cancer.
The
study
dataset
comprised
228,419
mothers,
of
whom
523
had
breast
cancer.
Multivariate
Cox
analysis
was
used.
The
relationship
between
ethnicity,
parity
and
premenopausal
breast
cancer
risk
was
examined,
adjusted
for
age
at
first
and
last
childbirth.


Our
results
show
that
the
effect
of
parity
on
breast
cancer
risk
is
modified
by
ethnicity.
The
risk
in
uniparous
Malay
women
was
higher
than
that
of
uniparous
Chinese
(hazard
ratio[HR]
1.91
relative
to
Chinese,
95%
confidence
interval
[CI]
1.17-3.13),
whereas
Indians
had
a
lower
risk
(HR
0.38,
95%
CI
0.12-1.19).
In
Chinese
and
Indian
women,
multiparity
had
no
effect
on
breast
cancer
risk.
In
contrast,
Malay
women
had
a
significant
risk
reduction
with
increasing
parity
(2
children:
HR
1.82
relative
to
uniparous
Chinese,
95%
CI
1.21-2.73；
≥3


children:
HR
1.16,
95%
CI
0.73-1.85).


E
 Discussion


This
is
the
first
study
to
show
that
the
effect
of
multiparity
on
premenopausal
breast
cancer
risk
is
modified
by
ethnicity
in
three
Asian
ethnic
groups.
Further
studies
are
needed
with
detailed
prospective
collection
of
information
in
order
to
confirm
these
findings
and
explain
the
underlying
mechanisms
for
the
observed
differences

LIST
OF
TABLES


TABLE
1.

A GE - STANDARDIZED
INCIDENCE
RATES
AND
DEATH
RATES
FOR
BREAST
CANCER
IN


SELECTED
COUNTRIES 3


TABLE
2.
B REAST
CANCER
RISK
FACTORS 6


TABLE
3.
A GE - ADJUSTED
BREAST
CANCER
INCIDENCE
RATES
IN
 USA
 BY
ETHNICITY 25


TABLE
4.
T OTAL
FERTILITY
RATES
IN
WOMEN
IN
 S INGAPORE
 1955-2007 .37


TABLE
5.
N UMBER
OF
BIRTHS
IN
 S INGAPORE
BY
CALENDAR
YEAR 47


TABLE
6.
B IRTH
ORDER
OF
OLDEST
CHILD
REGISTERED
IN
THE
 B IRTH
 R EGISTRY
IN
THE
PERIOD
 1986-2002 48


TABLE
7.
P ARITY
AND
ETHNICITY
OF
THE
 228,419
 MOTHERS
IN
THE
 B IRTH
 R EGISTRY 50


TABLE
8.
P ARITY
STATUS
 ( UNIPAROUS
VS
MULTIPAROUS )
 OF
THE
 228,419
 MOTHERS
 ACCORDING
TO
ETHNICITY 51


TABLE
9.
A GE
AT
BIRTH
OF
FIRST
CHILD
ACCORDING
TO
ETHNICITY
OF
THE
 228,328
 WOMEN 51 TABLE 
 10.
A GE
AT
BIRTH
OF
FIRST
CHILD
ACCORDING
TO
PARITY
OF
THE
 228,328
 WOMEN 51


TABLE
11.
C ATEGORIZING
AGE
AT
FIRST
BIRTH
BY
ETHNICITY
IN
THE
 228,328
 WOMEN 52


TABLE
12.
A GE
AT
BIRTH
OF
LAST
CHILD
AND
ETHNICITY
OF
THE
 228,328
 WOMEN 52


TABLE
13.
A GE
AT
BIRTH
OF
THE
LAST
CHILD
ACCORDING
TO
PARITY
OF
THE
 228,328
 WOMEN 52


TABLE
14.
A GE
AT
WHICH
CANCER
WAS
DIAGNOSED
IN
THE
 523
 WOMEN
WHO
DEVELOPED
 BREAST
CANCER 54


TABLE
15.
D URATION
BETWEEN
BIRTH
OF
FIRST
CHILD
AND
DATE
OF
CENSORING
IN
THE
 228,369
 WOMEN 56


TABLE
16.
M ODELLING
AGE
AT
FIRST
BIRTH
AND
BREAST
CANCER
RISK ,
 UNADJUSTED
AND
 ADJUSTED
FOR
AGE
AT
LAST
BIRTH
AND
PARITY ,
 ACCORDING
TO
ETHNIC
GROUP 58


TABLE
17.
B REAST
CANCER
RISK
WITH
ETHNIC
GROUP
INCLUDED
IN
THE
MODEL 60


TABLE
18.
B REAST
CANCER
RISK ,
 PERSON - YEARS
AND
ADJUSTED
HAZARD
RATIOS ,
 STRATIFIED
 BY
ETHNIC
GROUP
AND
PERIOD
OF
DIAGNOSIS 61


TABLE
19.
B REAST
CANCER
RISK ,
 PARITY
AND
ETHNICITY ,
 WITH
 C HINESE
WOMEN
AS
THE
 REFERENCE
GROUP 62


TABLE
20.
B REAST
CANCER
RISK ,
 PARITY
AND
ETHNICITY ,
 WITH
UNIPAROUS
 C HINESE
WOMEN


AS
THE
REFERENCE
GROUP 64



FIGURE
5.
A GE - ADJUSTED
BREAST
CANCER
MORTALITY
ACCORDING
TO
ETHNICITY
IN
 USA


BASED
ON
 SEER
 DATA
 1975-2004 31


FIGURE
6.
A GE - STANDARDIZED
INCIDENCE
RATES
OF
FEMALE
BREAST
CANCER
IN
 S INGAPORE


FIGURE
9.
A GE - STANDARDIZED
INCIDENCE
RATES
OF
FEMALE
BREAST
CANCER
IN
 S INGAPORE
 FROM
 1968-2002,
 STRATIFIED
BY
ETHNIC
GROUP 36


FIGURE
10.
A GE - SPECIFIC
FEMALE
BREAST
CANCER
INCIDENCE
RATES
IN
 S INGAPORE
FROM


1968-2002,
 STRATIFIED
BY
ETHNICITY 37


FIGURE
11.
T OTAL
FERTILITY
RATES
IN
WOMEN
IN
 S INGAPORE
 1957-2001 37


FIGURE
12.
T OTAL
FERTILITY
RATES
IN
WOMEN
IN
 S INGAPORE
 1968-2002,
 STRATIFIED
BY
 ETHNICITY 38


FIGURE
13.
T OTAL
FERTILITY
RATE
 ( PER
WOMAN )
 AND
AGE - STANDARDISED
INCIDENCE
RATE


OF
BREAST
CANCER
 ( PER
 10,000
 WOMEN
PER
YEAR )
 IN
 S INGAPORE ,
 BASED
ON
DATA
IN


T ABLE
 4
 AND
 S INGAPORE
 C ANCER
 R EGISTRY
DATA 39


FIGURE
14 
 S CATTERPLOTS
OF
CUMULATIVE
BREAST
CANCER
INCIDENCE
RATES
AND
TOTAL
 FERTILITY
BY
ETHNICITY 40


FIGURE
15.
D ESIGN
OF
STUDY
BASED
ON
INFORMATION
AVAILABLE
IN
THE
 S INGAPORE


N ATIONAL
 R EGISTRY
OF
 B IRTHS
AND
 D EATHS 44


FIGURE
16.
D ISTRIBUTION
OF
DURATION
BETWEEN
BIRTH
OF
LAST
CHILD
AND
DEVELOPMENT


OF
BREAST
CANCER ,
 STRATIFIED
BY
ETHNIC
GROUP 55


FIGURE
17.
B REAST
CANCER
RISK ,
 PARITY
AND
ETHNICITY ,
 WITH
 C HINESE
WOMEN
AS
THE
 REFERENCE
GROUP 63


FIGURE
18.
B REAST
CANCER
RISK ,
 PARITY
AND
ETHNICITY ,
 WITH
UNIPAROUS
 C HINESE
WOMEN


AS
THE
REFERENCE
GROUP 64



2.
 Verkooijen
HM,
Yap
KP,
Bhalla
V,
Chow
KY,
Chia
KS.


Multiparity
and
the
risk
of
premenopausal
breast
cancer:
different
effects
across
ethnic
groups
in
Singapore.
Breast
Cancer
Res
Treat.
2009;
113(3):
553-8.


Breast
cancer
incidence
is
highest
in
the
developed
countries.
From
the
International
Agency
for
Research
on
Cancer
(IARC)
2002
estimates,
the
more
developed


countries
had
an
overall
world
age-standardized
rate
of
67.8
per
100,000/year
as
compared
to
a
rate
of
23.8
in
the
less
developed
countries
(3).
In
the
United
States
alone,
the
age-adjusted
incidence
rate
was
126.1
per
100,000
women
per
year
from
2001-2005
(4).
The
incidence
and
death
rate
estimates
of
several
countries
obtained
from
the
WHO
estimates
are
listed
in
Table
1
(5):



Breast
Cancer
Worldwide


Age-Standardized
Incidence
Rate
(per
100,000)
(year
of
estimate)


Age-Standardized
Death
Rate
(per
100,000)


FIGURE
1.
Age-adjusted
breast
cancer
incidence
rates
for
selected
countries.
(6)


However,
after
year
2000,
a
slight
decrease
in
breast
cancer
incidence
has
been
observed,
which
will
be
described
further
in
section
2.3.1.


In
Singapore,
breast
cancer
is
the
most
common
incident
cancer
in
females.
In
the
period
1998-2002,
the
world
age-standardised
incidence
rate
was
54.9
per


100,000/year
(7).
Similar
to
trends
worldwide,
breast
cancer
incidence
in
Singapore
has
also
been
steadily
rising.
This
is
discussed
in
greater
detail
in
section
2.4

2.2
 
 Factors
Predisposing
to
Breast
Cancer


Breast
cancer
is
a
multi-factorial
disease.
Table
2
summarizes
some
of
the
factors
that
influence
the
predisposition
to
breast
cancer,
either
acting
individually
or
in
concert
with
other
factors.


years
delay
in
menarche
 Menopause
 Late
menopause
 17%
increase
in
risk
for
every
5


FIGURE
2.

Age-specific
breast
cancer
incidence
rates
in
selected
countries.
(8)


In
Singapore,
based
on
1998-2002
data,
the
age-specific
incidence
rates
in
women
aged
50-54
are
almost
twice
that
of
women
aged
40-44
(7).


The
effect
of
risk
factors
on
breast
cancer
varies
with
age.
The
following
factors
reduced the risk of early onset breast cancer but increased the risk of later onset breast cancers

— nulliparity, obesity and oral contraceptive use
(9;
10;
11).
Similarly,
differences
were
also
noted
in
tumour
characteristics
(12;
13)
and
survival
(13)
between
age
groups.
Tumours
were
classified
as
high
risk
if
they
were
>2cm,
estrogen
receptor
negative
(ER-),
node
positive
and
high
grade.
These
high
risk
tumours
were
found
to
have
an
early
onset
and
were
associated
with
a
worse
actuarial
survival
and
a
peak
in
hazard


at
2
years
after
cancer
diagnosis,
whereas
later
onset
tumours
had
a
better
survival
and
did
not
exhibit
the
hazard
peak
(13).
This
qualitative
age-interaction
effect


suggests
that
breast
cancers
occurring
in
younger
and
older
women
may
be
different
entities
(12).


Genetics


Besides
the
BRCA
genes,
which
have
a
low
population
frequency
but
a
high


penetrance
in
carriers,
studies
are
beginning
to
detect
various
breast
cancer


susceptibility
genes
which
are
more
common
but
exert
a
smaller
effect
on
risk.
The
Breast
Cancer
Association
Consortium,
an
international
collaboration,
studied
16
putative
single
nucleotide
polymorphisms
(SNPs)
previously
reported
in
smaller
studies
to
affect
breast
cancer
risk
(22).
Eighteen
studies
were
pooled,
with
a
total
of
between
12,013
to
31,595
subjects
(cases
and
control)
for
each
SNP
studied.
Small
associations
with
breast
cancer
were
found
for
5
SNPs
(CASP8
D302H,
IGFBP3
−


202
c
>
a
,
PGR
V660L,
SOD2
V16A,
and
TGFB1
L10P).
Further
evaluation
of
4
of
these
SNPs
and
another
5
SNPs
(comprising
11,391–18,290 cases and
14,753–22,670 controls)
showed
significance
of
the
CASP8
D302H
and
TGFB1
L10P
variants,


estimated
to
attribute
0.3%
and
0.2%
towards
familial
breast
cancer
risk
(23).


As
more
large-scaled
studies
are
done
in
this
field,
it
is
likely
that
other
SNPs
will
be
identified
in
future.


Height


Height
has
a
positive
association
with
breast
cancer
risk.
An
earlier
study
in
Norway


of
570,000
women
had
shown
that,
within
each
age
group,
the
risk
was
highest
in
the
tallest
women
(24).
In
a
large
pooled
analysis
of
7
studies
with
a
total
of
337,819
women
and
4,385
incident
invasive
breast
cancers,
height
was
found
to
have
a


of
1.02
was
observed
with
every
5
cm
height
increment;
in
post-menopausal
women
the
relative
risk
was
1.07.



The
age
at
which
maximum
height
is
reached
is
an
indicator
of
the
age
at
which
the
pubertal
growth
spurt
occurs.
Earlier
age
has
been
found
to
be
associated
with
increased
risk,
particularly
of
more
aggressive
tumour
types
(26)
and
of
ductal-

lobular
carcinoma
but
not
ductal
or
lobular
carcinoma
(27).


Birth
weight


The
role
of
intrauterine
factors
in
the
aetiology
of
breast
cancer
was
suggested
by
the
observation
that
the
initially
low
breast
cancer
risk
in
Asian
migrants
gradually


increased
over
several
generations
to
become
at
par
with
the
majority
Caucasian
population
in
USA
(28).
Although
initial
studies
did
not
reveal
any
relationship


between
birth
weight
and
breast
cancer
risk
(29;
30),
these
studies
were
limited
by
small
sample
sizes.


A
large
case-control
study
nested
within
the
two
Nurses
Health
Studies
showed
an
increased
adjusted
odds
of
breast
cancer
in
women
whose
birth
weights
were
higher
(31).
Park
et
al
(32)
showed
in
a
meta-analysis
that
the
odds
of
breast
cancer
was
1.24
with
birth
weights
of
≥4000g
(95%
CI
1.04-1.48)
and
1.15
(95%
CI
1.04-1.26)
for
birth
weights
3500-3599g,
with
respect
to
the
reference
group
of
2500-2999g.



The
underlying
mechanisms
of
this
birth
weight-breast
cancer
association
are


currently
unclear.
A
possible
explanation
is
the
effect
of
intrauterine
exposures
to
factors
with
mammotrophic
and
growth
hormone-like
effects.


Birth
Order


In
 the
 meta-analysis
 of
 the
 relationship
 between
 birth
 order
 and
 breast
 cancer
 risk
(32),
 it
 was
 found
 that
 higher
 birth
 orders
 were
 associated
 with
 a
 reduced
 breast
cancer
 risk,
 although
 the
 relationship
 was
 only
 seen
 in
 higher
 birth
 orders
 or
 ≥5
(odds
 ratio
 [OR]
 0.88,
 95%
 CI
 0.75-1.01).
 However,
 additional
 evidence
 to
 support
this
is
lacking.


Mammographic
Density


Mammographic
density
is
a
reflection
of
the
amount
of
fibroglandular
tissue
in
the
breast,
which
is
radiodense.
A
small
study
found
that
ductal
carcinoma-in-situ
arose


in
pre-existing
areas
of
mammographically
dense
tissue
(33).
This
suggests
that
epithelial
proliferation
occurs
in
radiodense
regions
of
the
breast.


mammographic
density
and
breast
cancer
risk
(34).
In
a
meta-analysis,
increased
mammographic
density
was
associated
with
increased
breast
cancer
risk
(35).
This
association
also
showed
a
dose-response
relationship.
When
compared
with
breast
density
of
<5%,
the
pooled
relative
risk
of
breast
cancer
was
1.78,
2.46,
3.02
and
4.59
with
densities
of
5-24%,
25-49%,
50-74%
and
≥75%
respectively,
after


adjusting
for
age
and
body
size
(total
cases:non-cases
3004:6468).


Mammographic
density
has
been
shown
to
correlate
with
known
risk
factors
for
breast
cancer,
including
menarche,
age
at
first
full-term
birth,
parity
and


premenopausal
status,
supporting
the
theory
that
these
factors
may
be
associated
with
one
another
in
breast
cancer
pathogenesis
(36).


Although
this
effect
of
mammographic
density
was
seen
across
different
ethnic
groups,
the
magnitude
of
risk
differed
with
ethnicity
(37).

Compared
to
whites,
the
association
was
stronger
in
Asian-Americans
and
weaker
in
African-Americans
(38).
The
weaker
association
seen
in
African-American
women
could
be
related
to
their
larger
breast
size,
as
the
relationship
between
mammographic
density
and
breast
cancer
risk
appears
to
be
weaker
in
women
with
larger
breasts
(39).


Premalignant
Pathology


The
relationship
between
benign
pathology
on
breast
biopsies
and
subsequent
risk


of
cancer
development
has
been
extensively
studied
by
Page
and
Dupont
(40;
41).
Lesions
were
classified
as
non-proliferative,
proliferative
without
atypical
hyperplasia
and
atypical
hyperplasia.
Women
with
proliferative
lesions
without
atypical


hyperplasia
had
1.5
to
2
times
the
risk
of
cancer
compared
to
the
general
population
(women
with
non-proliferative
lesions)
(42;
40).
The
risk
was
4-5
times
with
atypical
hyperplasia
(40;
41;
42).
This
risk
was
further
exacerbated
if
these
women
with
atypical
hyperplasia
had
a
family
history
of
breast
cancer in a first degree relative – the risk
was
9.7
for
atypical
ductal
hyperplasia
(95%
CI
4.7-20)
and
8.4
(95%
CI
3.5-20)
(41).


Strategies
to
reduce
the
cancer
risk
in
women
with
atypical
hyperplasia
include
chemoprevention
and
prophylactic
surgery.
Tamoxifen
was
the
first
drug
proven
to
reduce
breast
cancer
risk
in
women
at
high
risk
(43;
44).
A
more
recent
trial
has
shown
that
raloxifene
is
as
effective
as
tamoxifen
for
chemoprevention
when
used
in
post-menopausal
women
(45).
However,
currently
there
is
no
data
as
to
whether
chemoprevention
improves
survival.


Cancer
in
the
Opposite
Breast


Women
with
breast
cancer
have
a
fivefold
increased
incidence
of
developing
cancer


in
the
contralateral
breast
(46).
In
a
meta-analysis
of
55
trials
on
tamoxifen
therapy
for
early
breast
cancer,
the
incidence
rate
of
contralateral
breast
cancer
in
women
not
on
tamoxifen
(the
control
group)
was
5
per
1000
women
per
year
(47).
The
second
cancer
may
be
synchronous
(in
1-2%)
or
more
often
metachronous
(in
5-6%)
(46;
48).
In
a
retrospective
survey
of
4554
patients
treated
at
the
MD
Anderson
Cancer
Centre,
142
patients
(3.1%)
had
metachronous
or
synchronous
cancers
in
the
opposite
breast
(49).


Risk
factors
for
developing
bilateral
breast
cancer
include
multifocal
cancer,
lobular
carcinoma,
lobular
carcinoma-in-situ
and
younger
age
at
diagnosis
of
the
first
cancer
(46;
48;
50).
Although
mammography
increases
the
detection
of
synchronous


cancers,
the
overall
incidence
of
bilateral
cancers
is
unaltered,
supporting
the
role
of
mammography
in
earlier
cancer
detection
without
causing
radiation-related
cancers
(46;
49)

Menarche
and
Menopause


Women
with
an
earlier
age
at
menarche
and/or
later
age
at
menopause
have
an
increased
risk
of
breast
cancer
(52).
This
is
likely
related
to
longer
lifetime
exposure


to
endogenous
hormones
(53).
One
study
found
that
every
2
years
delay
in


menarche
resulted
in
a
10%
reduction
in
breast
cancer
risk
(54).
Another
study
similarly
showed
a
reduction
in
risk
with
later
menarcheal
age
(age
≥15
versus
age
13)
(55).
In
addition,
this
study
also
demonstrated
that
the
reduction
was
more


marked
in
premenopausal
women
(OR
0.72,
95%
CI
0.57-0.91)
compared
to


postmenopausal
women
(OR
0.90,
95%
CI
0.80-1.03).
Determinants
of
age
at


menarche
include
increased
height
(56)
and
body
mass
index
(56;
57)
(these
factors
are
discussed
separately
in
their
respective
sections).


The
protective
effect
of
late
menarche
was
found
to
result
in
a
greater
reduction
of
estrogen
receptor
positive
(ER+)
progesterone
receptor
positive
(PR+)
(relative
risk
[RR]
0.72,
95%
CI
0.64-0.80)
than
ER-PR-
tumours
(RR
0.84,
95%
CI
0.75-0.94)
based
on
a
meta-analysis
of
9
studies
(p=0.006)
(58).


Later
menopause
is
associated
with
a
higher
risk
of
breast
cancer.
The
risk
was
found
to
be
17%
higher
for
every
5-year
increase
of
age
at
menopause
(54).
On
one
extreme,
women
who
have
premature
menopause
from
bilateral
oophorectomy
have


menopause
before
age
40
had
a
significantly
lower
risk
of
breast
cancer
(OR
0.57,
95%
CI
0.47-0.71)
(55).


Parity


Parity
is
a
well-known
risk
factor
for
breast
cancer.
In
1990,
Adami
et
al
showed
that
nulliparous
women
had
a
35%
higher
risk
as
compared
to
parous
women
(9).
A
dose-response
relationship
was
observed
in
that
higher
parity
conferred
more


protection
in
women
whose
first
birth
was
before
age
25
(9;
60).
In
a
case-control
study
in
Singapore
Chinese
women,
the
effect
of
parity
differed
according
to


menopausal
status
(61).
In
premenopausal
women,
parity
did
not
affect
breast


cancer
risk.
However,
in
postmenopausal
women,
nulliparity
was
a
significant
risk
factor
(p=0.003).


Parity
reduced
the
risk
of
ER+PR+
but
not
ER-PR-
tumours.
A
meta-analysis
of
8
studies
computed
an
11%
reduction
in
risk
of
ER+PR+
cancer
per
birth
(58).
A
recent
large
case-control
study
(CARE
Study)
corroborates
the
finding
(60).


Age
at
First
Birth


Older
maternal
age
at
birth
of
the
first
child
increases
breast
cancer
risk
in
the
mother
(9).
In
Singapore
Chinese
women,
this
was
found
to
be
a
significant
risk
factor
in
premenopausal
women
with
cancer
but
not
in
postmenopausal
women
(61).


In
the
meta-analysis
of
9
studies,
age
at
first
birth
influenced
the
risk
of
ER+PR+
but
not
ER-PR-
tumours
(58).
For
ER+PR+
tumours,
women
in
the
oldest
age
category


at
first
birth
had
a
relative
risk
of
1.27
compared
to
women
in
the
youngest
age
category
(p=0.010).
No
difference
was
observed
in
ER-PR-
tumours
(RR
1.01,
95%


CI
0.85-1.20).


Breastfeeding


analysis
of
18
studies
(62).


Not
only
is
breastfeeding
status
(yes/no)
important,
the
duration
of
breastfeeding
also
influences
breast
cancer
risk.
A
collaborative
analysis
of
47
epidemiological
studies
from
30
countries
found
that
there
was
a
4.3%
(95%
CI
2.9-5.8)
reduction
in
RR
for
every
12
months
of
breastfeeding.
This
was
separate
from
the
7%
(95%
CI
5.0-9.0)
decrease
in
risk
seen
with
each
birth
(63).
This
finding
did
not
vary
with
demographic


or
socioeconomic
variables.
Pooled
data
from
7
studies
show
that
breastfeeding
appears
to
be
equally
protective
against
ER+PR+
and
ER-PR-
tumours
(summary


estrogen
plus
progestin
(65)—the
HR
for
breast
cancer
was
1.26
(95%
CI
1.00-1.59).


In
the
UK
Million
Women
Study,
users
of
HRT
had
an
adjusted
RR
of
1.66
(95%
CI
1.58-1.75),
amounting
to
20,000
extra
breast
cancer
cases
over
a
decade
(66).
Subsequently,
with
more
women
aborting
the
use
of
HRT,
breast
cancer
incidence
has
been
on
the
decline,
as
has
been
observed
in
countries
where
HRT
use
was
widespread
(67).
Beyond
5
years
after
stopping
HRT,
the
risk
is
no
longer
elevated
(64).


Oral
Contraceptives


Use
of
oral
contraceptives
(OC)
increases
breast
cancer
risk.
In
a
meta-analysis
of


34
published
studies,
there
was
an
overall
increase
in
breast
cancer
in
users
of
OC


(pooled
OR
1.52,
95%
CI
1.26.1.82).
A
UK
study
of
23,000
users
(with
744,000
years


of
follow
up)
and
23,000
non-users
(with
339,000
years
of
follow-up)
of
OC
showed


no
difference
in
risk
of
breast
cancer
for
women
who
used
OC
≤8
years
but
an
increased
risk
beyond
8
years
of
use
(RR
1.22,
95%
CI
0.97-1.52)
(69).


Fertility
Treatment


Drugs
used
to
treat
infertility,
namely
clomiphene
and
the
gonadotrophins,
stimulate
ovulation
and
increase
plasma
concentrations
of
oestradiol
and
progesterone
(70).
Thus
it
has
been
postulated
that
fertility
medications
would
enhance
breast
cancer
risk
through
this
mechanism.
However,
studies
have,
in
general,
not
shown
any
significant
risk
of
breast
cancer
with
use
of
these
medications
[
(71;
72),
reviewed
in
(73;
74;
75;
76)].
A
large
study
from
Denmark
showed
a
standardised
incidence
ratio


of
1.08
(95%
CI
1.01-1.16)
of
developing
breast
cancer
(77).
Longer
follow-up
will
be
needed
to
ascertain
breast
cancer
risk
as
these
women
approach
the
peak
cancer
age
range.


Furthermore,
in
a
case
control
study
of
BRCA1
and
BRCA2
mutation
carriers,
who
have
an
inherited
susceptibility
to
developing
breast
cancer,
women
who
used
fertility
medications
were
did
not
have
a
significantly
increased
risk
of
breast
cancer
(OR
1.21,
95%
CI
0.81-1.82)
(78).


Though
no
reports
were
found
on
the
risk
of
breast
cancer
in
infertile
women
who
did
not
receive
fertility
medication,
it
would
be
expected
that
their
risk
of
breast
cancer
would
be
increased
due
to
the
effects
of
nulliparity
or
later
age
of
first
pregnancy.



Abortions


Although
early
studies
suggested
that
termination
of
early
pregnancy
might
increase
the
risk
of
breast
cancer,
subsequent
larger
studies
have
not
shown
any
increase
in


demonstrate
any
difference
in
risk
of
breast
cancer
in
women
who
had
spontaneous


or
induced
abortions
(79) More recently, other studies such as the Nurses’ Health Study II cohort with 105
716
participants
(80),
the
European
Prospective
Investigation
into
Cancer
and
Nutrition
(EPIC)
study
with

4,805
women
with
breast
cancer
(81)
and
the
California
Teachers
Study
(CTS)
cohort
with
3324
women
with
breast
cancer
(82)
also
had
the
same
conclusion.


emerge
that
smoking
may
have
a
selective
action
on
breast
cancer,
with
increased
risk
being
seen
only
with
early
age
of
smoking
commencement
or
commencement
before
first
pregnancy
(89;
88).


Genetic
factors
may
also
play
a
role
in
modifying
breast
cancer
risk
with
cigarette
smoke
exposure.
A
meta-analysis
of
the
effect
of
N-acetyltransferase
2
(NAT2)
gene
variants
on
breast
cancer
found
many
studies
showing
an
increased
risk
among
smokers
who
were
slow
acetylators
(90).
Another
study
showed
that,
among
women
who
smoked
1-9
cigarettes
daily,
those
who
carried
the
NAT2
slow
acetylator


genotype
as
well
as
at
least
one
CYP1B1
432Val
allele
had
over
fourfold
the
risk
of
breast
cancer
compared
to
those
with
the
NAT2
rapid
acetylator
and
CYP1B1


Leu/Leu
genotype
(91).
In
BRCA1
and
BRCA2
mutation
carriers,
breast
cancer
risk
was
increased
more
than
twofold
in
current
smokers
compared
with
those
who
never
smoked
(OR
2.02
in
BRCA1
carriers,
OR
2.35
in
BRCA2
carriers)
(92).
In
addition,
a
dose-dependent
relationship
was
found,
with
a
7%
increase
risk
of
breast
cancer
per
pack
year
of
smoking.


Alcohol


In
many
studies,
women
who
consume
alcohol
have
a
slightly
elevated
breast
cancer
risk,
as
reported
in
a
pooled
analysis
of
6
cohort
studies
from
Canada,
the


Netherlands,
Sweden
and
the
United
States
(93).
For
moderate
intakes
of
up
to
60g


of
alcohol
per
day,
a
10g/day
increment
in
alcohol
consumption
was
associated
with


a
9%
increased
risk
of
breast
cancer.
Thus,
women
who
consumed
30-60
g/day
of
alcohol
had
a
41%
greater
risk
than
non-drinkers.
There
was
no
difference
if
the
source
of
alcohol
was
from
beer,
wine
or
liquor.


In
a
case-control
study
in
Massachusetts,
a
trend
of
increasing
alcohol
consumption
with
increased
breast
cancer
risk
was
observed.
Women
who
consumed
14
or
more
alcoholic
drinks
per
week
had
an
adjusted
odds
of
1.43
compared
to
those
who
consumed
<1
drink/week
(94).
The
estimated
population
attributable
risk
from
alcohol
consumption
was
6%.


A
survey
conducted
via
a
self-administered
lifestyle
questionnaire
within
the


Canadian
National
Breast
Screening
Study
cohort
(2491
incident
breast
cancers),
showed
a
weak
association.
Those
consuming
>30g/day
of
alcohol
had
a
HR
of
1.17
relative
to
nondrinkers
(95).
Recent
alcohol
intake
was
also
found
in
the
EPIC
study


to
increase
breast
cancer
risk,
with
an
incidence
rate
ratio
of
1.03
per
10g/day
higher
alcohol
intake
(96).
The Women’s Health Initiative-Observational Study of 88,530 women showed that breast cancer risk related to alcohol was dose-dependent (HR 1.10 for ≤5
g/day,
HR
1.14
for
>5-15
g/day,
HR
1.13
for
>15
g/day)
(97).


However,
some
recent
studies
found
the
opposite,
with
alcohol
seemingly
reducing
breast
cancer
risk.
In
Southern
France,
it
was
found
that
the
risk
associated
with
alcohol
consumption
was
only
seen
above
a
threshold
consumption
(98).
Women


inactive
estrogen
precursors
to
the
active
form.
This
is
a
major
source
of
estrogens
in
post-menopausal
women
(103).
In
obese
women,
this
peripheral
conversion
in
the
subcutaneous
fat
may
be
substantial.


Data
pooled
from
8
prospective
cohort
studies,
comprising
337,819
women
with
3208
incident
cancers
showed
a
RR
of
1.26
(95%
CI
1.09-1.46)
in
overweight
women
(BMI>28
kg/m2)
compared
with
women
of
BMI
<21
(25).
For
breast
cancer,
the
population
attributable
risk
associated
with
weight
gain
in
menopausal
women
was


found
to
be
21.3%
(94).
Those
who
had
a
weight
gain
of
>30
kg
since
age
18
had
an
odds
of
1.67
of
breast
cancer
compared
to
women
without
significant
weight
change.


Soy
intake


The
observation
that
women
from
cultures
with
a
high
soy
intake
had
a
lower
breast
cancer
risk
suggested
that
soy
may
have
a
protective
effect
on
breast
cancer.


Genistein,
an
isoflavone
found
in
soy,
was
found
to
inhibit
tyrosine
kinase
activity
of
the
epidermal
growth
factor
receptor
in
vitro
(104).
Urinary
excretion
of
isoflavonic
phyto-oestrogens,
as
an
indicator
of
phyto-oestrogen
intake,
was
lower
in
breast
cancer
cases
compared
to
controls,
although
specific
assay
for
genistein
was


hampered
by
technical
factors
(105).


A
meta-analysis
of
8
studies
from
Asian
countries
with
high
consumption
of
soy
found
a
reduction
in
breast
cancer
risk
with
increasing
quantity
of
soy
consumed
(106).
Those
who
consumed
≥20
mg
isoflavones
per
day
had
a
0.71
odds
(95%
CI
0.60-0.85)
of
breast
cancer
compared
to
those
consuming
≤5
mg
per
day.
In


contrast,
in
11
studies
from
Western
populations
where
soy
consumption
was
low
(average
0.15-0.8
mg/day),
soy
had
no
effect
on
breast
cancer
risk.
A
case-control
study
conducted
locally
found
that
soy
had
a
protective
effect
(107).
In
premeno-pausal
women,
those
in
the
highest
tertile
of
soy
consumption
had
an
odds
of
breast
cancer
of
0.39
(95%
CI
0.19-0.8)
compared
to
women
in
the
lowest
tertile.
Recent
results
from
the
prospective
Singapore
Chinese
Health
Study
again
confirmed
the
finding
of
risk
reduction
with
high
soy
intake
(108).
In
a
subset
of
this
cohort
who
underwent
mammographic
screening,
higher
soy
intake
was
associated
with
lower
mammographic
density
(109);
there
was
a
5%
difference
in
mammographic
density
seen
between
the
highest
and
lowest
quartiles
of
soy
isoflavone
consumption.


Ionizing
Radiation


From
historical
observations,
an
excess
of
breast
cancer
was
observed
in
persons
exposed
to
ionizing
radiation
from
sources
such
as
radiotherapy
to
the
thymus
or
for
ankylosing
spondylitis,
X-ray
technicians
and
radium
painters
(110).
In
survivors
of
the
atomic
bombs
in
Hiroshima
and
Nagasaki,
breast
cancer
risk
was
most
elevated


in
women
exposed
during
childhood;
women
above
age
40
were
unaffected
(111).
Among
flight
attendants,
the
risk
was
estimated
to
be
40%
higher
(112).
However,
some
studies
have
failed
to
demonstrate
an
association
between
radiation
and
breast
cancer.
In
a
large
cohort
of
43,316
Norwegian
nurses,
no
clear
association
was
found
between
exposure
to
ionizing
radiation
and
development
of
several


cancers,
including
breast
cancer
(113).
A
case-control
study
of
cervical
cancer


patients
undergoing
radiation
treatment
showed
a
RR
of
breast
cancer
of
1.07
in
women
without
ovaries,
although
the
dose-response
trend
did
not
reach
statistical
significance
(110).


Physical
Activity


Women
who
are
more
physically
active
have
an
approximately
25-30%
lower
risk
of
breast
cancer,
as
shown
in
a
review
of
published
studies
on
this
topic
(114).
A
dose-response
relationship
was
demonstrated
in
the
majority
of
studies
(114).
A
very
recent
report
from
the
NIH-AARP
Diet
and
Health
Study
involving
182,862
women
found
that
the
most
active
quintile
of
women
had
a
13%
lower
risk
as
compared
to
the
lowest
quintile
(115).
The
protective
effect
was
more
pronounced
on
ER-
tumours
(RR=0.75,
95%
CI
0.54-1.04)
than
ER+
tumours
(RR=0.97;
95%
CI,
0.84-1.12)

FIGURE
3.
Age-adjusted
incidence
of
breast
cancer
according
to
ethnicity
in
the
USA,
based
on
SEER
data
1975-2004.
(4)


As
seen
in
Figure
3, there has been a gradual decline in breast cancer incidence since year

2001 In the late 1990’s, evidence began to emerge regarding
the
adverse
effects
of
HRT


on
breast
cancer
risk
(64;
118)(as
discussed
on
page
17).
A
drastic
decline
in
use
of
HRT
followed
the
2002
report
from
the
Women
Health
Initiative
when
the
study
was
stopped
early
due
to
the
excess
risk
of
invasive
breast
cancer
(65).
Similar
trends
in
the
decline
in
breast
cancer
incidence
have
also
occurred
in
Australia,
New
Zealand,
Canada,
Germany
and
France
(67).
In
the
USA,
the
decline
in
breast
cancer
rates
was
most
marked
in
whites
(-14.3%
between
2001-2004)
followed
by
Asians/Pacific
Islanders
(-8.5%)
(119).
Among
the
ethnic
groups
in
USA,
use
of
HRT
was
highest
in
white
women
(120;
121;
122).
This
suggests
that
differential
usage
of
HRT
among
ethnic
groups
account
for
the
observed
differences
in
breast
cancer
incidence
in
post-menopausal
women.


In
Asia,
there
are
differences
in
incidence
rates
between
countries
(123).
Breast
cancer
risk
is
low
in
rural
China,
Korea
and
Thailand
but
high
in
Singapore
and
the
Philippines.
In
rural
China,
the
age
standardized
incidence
rate
per
100,000
women


truncated
at
age
≥20
(ASRt20)
was
19.2.
In
Singapore
the
ASRt20
was
82.2
and
in
the
Philippines
82.5.
Over
10
years
from
1993-2002, breast cancer rates have increased, with the greatest increases seen in countries with low incidence—Korea 7.9% increase—compared to countries with higher incidence such as Singapore with a 4.4% increase.
Breast
cancer
rates
in
Asian-Americans
are
1.5
to
4
times
higher
than
Asians
in
their
host
countries,
suggesting
that
lifestyle
factors
may
be
an
important
factor
in
explaining
these
differences.


In
the
Carolina
Breast
Cancer
Study
(124),
a
questionnaire-based
case-control
study,
differences
in
risk
were
found
between
African-American
and
White
women.


Multiparity
(having
3
or
4
children)
was
associated
with
an
increased
premenopausal
breast
cancer
risk
(adjusted
OR
1.5,
95%
CI
0.9-2.6)
in
African-American
but
a
reduced
risk
(adjusted
OR
0.7,
95%
CI
0.4-1.2)
in
White
women.


In
the
Women’s Contraceptive and Reproductive Experiences
(CARE)
Study
(125),
the
trend
of
reduced
risk
with
increasing
parity
was
seen
in
both
African-American
and
White
women,
but
was
less
pronounced
in
the
former
group.
This
study,
as
well
as
the
Carolina
Breast
Cancer
Study,
also
found
that
later
age
at
first
pregnancy


increased
the
breast
cancer
risk
in
White
women
but
not
in
African-American
women
(124;
125).


A dual effect of parity on breast cancer risk was seen in the Black Women’s Health Study
(126).
In
African-American
women
under
the
age
of
45,
higher
parity
was
associated
with
an
elevated
breast
cancer
risk
(incidence
rate
ratio
2.4
for
four
or
more
births,