University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
FACTORS THAT INFLUENCE MAMMOGRAPHIC DENSITY:
ROLE OF ESTROGEN METABOLISM GENES, BIOMARKERS OF
INFLAMMATION, AND LIFESTYLE

by
Anne Dee

A Dissertation presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
in Partial Fulfillment of the
Requirement for the Degree
DOCTOR OF PHILOSOPHY
(EPIDEMIOLOGY)

August 2013
i

Table of Contents

List of Tables and Figures

Chapter 1
Background
1.1 Mammographic Density
1.2 References

Chapter 2
Mammographic density and CYP17, CYP1B1, COMT, and MTHFR
polymorphisms in postmenopausal breast cancer patients
2.1 Abstract
2.2 Introduction
2.3 Methods
2.4 Results
2.5 Discussion
2.6 References

Chapter 3
Acute-phase proteins (C-reactive protein and Serum Amyloid A) and post-
diagnosis mammographic density in breast cancer survivors
3.1 Abstract
3.2 Introduction
3.3 Methods
3.4 Results
3.5 Discussion
3.6 References

Chapter 4
Association between smoking , alcohol use and mammographic
density in breast cancer survivors
4.1 Abstract
4.2 Introduction
4.3 Methods
4.4 Results
4.5 Discussion
4.6 References

ii

1
23

30
32
34
38
41
68

72
74
76
80
83
107

111
113
116
120
124
139

ii

List of Tables and Figures

Chapter 1

Table 1. Studies of endogenous estrogen levels and
mammographic density among postmenopausal women.
Table 2. Studies of alcohol intake and mammographic density
among postmenopausal women.
Table 3. Studies of mammographic density and cigarette smoking.
Chapter 2

Table 1. Demographic characteristics of study participants at
baseline (pre-diagnosis N=331) and 24-months after baseline
(post-diagnosis N=438).

Table 2. Geometric mean levels of percent mammographic
density among postmenopausal women by CYP17, COMT and
CYP1B1 genotype before and after diagnosis of breast cancer

Table 3. Adjusted mean levels of pre-diagnosis percent
mammographic density among postmenopausal women by
CYP17, COMT and CYP1B1 genotype and stratified by post-
menopausal hormone use .

Table 4. Geometric mean levels of pre-diagnosis percent
mammographic density among postmenopausal women for
CYP17xCOMT and MTHFRxCOMT.
Supplement Table 1a. Adjusted geometric mean levels of percent
mammographic density among postmenopausal women by
CYP17, COMT and CYP1B1 genotype and stratified by race,
before and after breast cancer
diagnosis.
Supplement Table 1b. Adjusted mean levels of percent
mammographic density among postmenopausal women by
CYP17, COMT and CYP1B1 genotype and stratified by race,
before and after breast cancer diagnosis.
Supplement Table 2. Adjusted mean levels of pre-diagnosis
percent mammographic density among postmenopausal women by
CYP17, COMT and CYP1B1 genotype and stratified by post-
menopausal hormone use.

15

17

20

47

50

51

52

53

54

55

iii

Supplement Table 3. Modifying effect of estrogen-related factors
on the association between adjusted mean levels of percent
mammographic % density among postmenopausal women by
CYP17, COMT and CYP1B1 genotypes with pre-diagnosis
mammograms.
Supplement Table 4. Unadjusted geometric mean levels of
percent density and associations with biological and lifestyle
factors among postmenopausal women.
Supplement Table 5a. Associations in odds-ratio between
genotype(continuous) and biological or lifestyle factors (outcome)
in Pre-diagnosis population (n=313).
Supplement Table 5b. Associations in odds-ratio between
genotype(continuous) and biological or lifestyle factors (outcome)
in Post-diagnosis population (n=438).
Supplement Table 6a. Association between mammographic
percent density and genotypes in the pre-diagnosis population
(n=311).
Supplement Table 6b. Association between mammographic
percent density and genotypes in the post-diagnosis population
(n=438).
Figure 1. Participation and data collection of study population.
Chapter 3
Table 1. Descriptive table of characteristics in a study population
(N=479) by median % mammographic density drawn from Health,
Eating, Activity and Lifestyle (HEAL) Study.
Table 1b. Association between characteristics in a study
population (N=479) drawn from Health, Eating, Activity and
Lifestyle (HEAL) Study and geometric means of exposure
(CRP/SAA,) and outcome(mammographic % density).
Table 2a. Linear regression models for log
e
CRP (exposure) on
transformed percent mammographic density (square root) after
adjusting for covariates (n=479).
Table 2b. Linear regression models for log
e
SAA (exposure) on
transformed percent mammographic density (square root) after
adjusting for covariates.

56

58

61

63

65

66

67

90

92

95

95

iv

Table 3a. Linear regression models for log
e
CRP on
mammographic density (continuous) outcome on square root
transformation) stratified on variables related to hormone,
inflammation, or adiposity after adjustment for covariates.
Table 3b. Linear regression models for log
e
(SAA) on
mammographic density (continuous outcome on square root
transformation) stratified on variables related to hormone,
inflammation, or adiposity after adjustment for covariates.
Supplement Table 1. Associations between geometric mean
mammographic density by midpoint of CRP after adjusting for
covariates: (age, BMI(<25,25-29.9,>=30), post-menopausal
hormone use(never, estrogen only, combined estrogen and
progestin).

Supplement Table 2. Correlation coefficients (Pearson) evaluated
between CRP, SAA and measures of mammographic density.

Figure 1. Flowchart of study population from HEAL.
Figure 2a. Scatter plot and predicted regression line of square root
percent dense area on ln(CRP). Predicted values are adjusted for
age, BMI, post-menopausal hormone use.
Figure 2b. Scatter plot and predicted regression line of square
root percent dense area on ln(CRP). Predicted values are adjusted
for age, BMI, post-menopausal hormone use.
Figure 3a. Scatter plot and predicted regression line of square root
percent dense area on ln(CRP). Predicted values are adjusted for
age, race, BMI, post-menopausal hormone use.
Figure 3b. Scatter plot and predicted regression line of square
root percent dense area on ln(SAA). Predicted values are adjusted
for age, race, BMI, post-menopausal hormone use.
Chapter 4
Figure 1. Flowchart of study population from HEAL.
Table 1. Descriptive characteristics of 502 women in HEAL
cohort with alcohol and smoking data.

Table 2. Geometric mean of mammographic density (outcome)
overall and by alcohol use (non user <0.2 grams/day compared to

96

98

100

101

102

103

104

105

106

129

130

132
v

non-users ≥0.2 grams/day), and stratified by smoking status,
reproductive variables, BMI, tamoxifen use and physical activity.

Table 3a. Linear regression models for mammographic density
(continuous outcome on square root transformation) on alcohol
(gm/day) or type of (alcohol servings/day) after adjustment for
covariates among those who drink.

Table 3b. Linear regression models for mammographic density
(continuous outcome on square root transformation) on smoking
variables (exposure) on after adjustment for covariates only
among smokers.

Table 4. Linear regression models for mammographic density
(continuous outcome on square root transformation) on smoking
variables (exposure) on after adjustment for covariates (age, race,
BMI, PMH, physical activity level) and stratified by smoking
status.

Figure 2a. Linear regression coefficient of alcohol(gm/day) and
95% confidence intervals at specific percentile of percent
mammographic density (20%, 40%, 60%, 80%) in the
multivariable (adjusted) quantile regression model. The line
connects the points for the linear regression coefficients and the
blue shading indicates the 95% confidence intervals when
restricting to drinkers (n=229).

Figure 2b. Linear regression coefficient of pack year and 95%
confidence intervals at specific percentile of percent
mammographic density (20%, 40%, 60%, 80%) in the
multivariable (adjusted) quantile regression model. The line
connects the points for the linear regression coefficients and the
blue shading indicates the 95% confidence intervals when
restricting to ever smokers (n=253).

134

135

136

137

138

1

Chapter 1
Background
1.1 Mammographic Density
Introduction
The purpose of this dissertation project is to investigate the associations between estrogen
metabolism genes, biomarkers of inflammation, and lifestyle on mammographic density among
postmenopausal breast cancer survivors. This background aims to give an overview on the
current understanding of these associations in the literature.
Mammographic Density Measurement
Mammographic density (MD) reflects the proportion of radiologically dense tissue in the
breast. Expressed in percentage of dense area of the total breast area, MD measurements are
obtained using computer-assisted method where a human reader who manually determines the
dense areas on the mammogram for calculation for MD. Continuous percent density is
considered to more accurately represent the relation with breast cancer risk than qualitatively
considered classification systems based on parynchymal patterns including Wolfe grade (N
1
,
normal fatty breast; P
1
and P
2
, prominent ducts occupying <25% and 25-75% of the breast,
respectively; and Dy, dysplastic breast with sheets of dense parenchyma) and Breast Imaging
Reporting and Data System (BIRADS) of the American College of Radiology density
classification (fatty, scattered fibroglandular density, heterogeneously dense, and extremely
dense)(1).On the mammogram, the dense area of the breast, consisting of fibroglandular tissues,
appears lucent and the non-dense area, consisting of adipose tissue, appears dark due to different
2

compositions and differences in the radiographic attenuation of the different tissues in the
breast(2).
Biology of Mammographic Density
Stromal Tissue
Stromal, epithelial, and adipose are constitutes elements of the breast. It is held together
by the network of fibrous connective tissue in the form of planar sheets of varying thickness.
This network of interlobular connective, or stromal tissue, is called the “Cooper’s ligaments”,
which runs throughout the breast, forming around the glandular ductal system and providing
structural support. Cooper’s ligaments attach the breast to the skin by retinacula cutis, which is
the superficial extension of Cooper’s elements suspending the breast from the skin. The extent of
this network of fibrous connective tissue is different in every woman.
A more specialized form of connective tissue is contained with and surrounds the
terminal duct lobular unit (TDLU), the most important structure in the breast. It is the functional
unit for milk production and site of carcinogenesis; most of the benign lesions such as cysts and
fibroadenomas of the breast develop at the TDLU. A TLDU is an extra-lobular terminal duct and
its lobule. The lobule consists of collections of glandular acini, the smaller lobules arrayed at the
end and around a terminal duct. This intralobular connective tissue involved is more loosely
arranged than the non-specialized connective interlobular stromal tissue. The two types
connective tissues are fibrous elements in the breast and provide the structure and firmness to the
breast(3).
Stromal tissue is the major tissue compartment in the fibroglandular tissue in the breast
by volume. Thus, stromal fibrosis in mammographically dense breasts plays an important role in
3

carcinogenesis as fibroblasts may sometimes turn malignant (4, 5). In cancerous breasts, high
levels of collagen as well as altered expressions of stromal proteins (6) and prosteoglycan
expression are present(7). These changes in expression can affect the interaction between stromal
and epithelial tissue (8). The stromal-epithelial interactions are important in breast cancer
carcinogenesis as they influence tumorigenesis and progression through direct effects on growth
factor pathway(9); fibroblasts transformed with epithelial cells have been shown in mice model
to accelerate the growth and shorten the latency of human breast epithelial tumors(10).
Epithelial Tissue
The epithelial element in the breast is located in the ductal systems of the breast and the
origin of more than 90% of all breast tumors(11); therefore, studies on the biological basis of
mammographic density have mostly concentrated on its association with epithelial tissue(12, 13).
A few histological studies on breast tissue identified epithelial hyperplasia on normal dense
breasts (5, 12, 14), but other studies have not(4, 15). Furthermore, although more cells in breast
epithelium are present in dense breast stromal tissue, there is no evidence of higher level of
proliferation is present in dense compared to non-dense areas(16).
Structurally, within the breast, each major lactiferous duct extends from the nipple back
into the breast in a branching network of series of ducts. A lobe segment in the breast consists of
a major duct and its tributaries, but the extent and shape of each network is variable within and
between women. The epithelial cells are located on the lining of the ducts, consisting of an inner
layer of epithelial cells, surrounded by a thinner layer of myoetpithelial cells which are important
for milk-production during lactation(3).
4

The phenotype of the epithelial cells that form is determined by extracellular matrix and
its receptors(17). The normal epithelial cell differentiation and proliferation are dependent on
stromal architecture and composition which ensures the correct signaling interactions between
cells and formation of extracellular matrix take place (18, 19). Alteration of the extracellular
matrix of the stroma can result in benign lesions to ductal carcinoma in situ or even invasive
carcinoma. The loss of the homeostasis of the balance of cell growth and apoptosis resulting in
proliferation can lead to development of a tumor microenvironment(17, 20). Thus, mutation of
genes regulating epithelial cell growth can lead to the cancerous cells to proliferate either
immediately off a major duct or anywhere in the breast(3).
Upon the epithelial cells themselves, however, estrogen dependent effects have been
detected (21). Estrogen acts through estrogen receptors on epithelial cells and, on some extent,
the stromal cells as well though the receptors on stromal are less abundant(22).
Adipose Tissue
The adipose element in the breast are located both subcutaneously and at the back of the
breast, the retromammary fascia, near the chest wall. The subcutaneous fat layer in the breast is
perforated by the fibrous elements and the epithelium is found directly beneath the skin in
association with the retinacular cutis(3). In postmenopausal women, adipose tissue in the breast
and the rest of the body are sites of aromatase function. Aromatase overexpression can result in
high estrogen levels which further stimulates epithelial cell growth(23).
In the literature, the association between MD and cancer risk or the prognosis are not
fully understood. Furthermore, dense area alone is not necessary as predictive of MD, which
5

suggests that both non and nondense tissue contribute to risk associated with mammographic
density(24).
Recurrence and Survival
High MD is one of the strongest risk factors associated with breast cancer. It is also
associated with breast cancer risk factors including parity(25), family history(26), genetic
inheritance(27), postmenopausal hormone use(28). Furthermore, studies have consistently shown
higher MD is positively associated with high risk of breast cancer up to four or five times
independent of those risk factors and regardless of reduced mammographic density due to
delayed interaction(29). The association between density and risk is well-established in
literature, being consistent over time and with repeated screenings.
MD is also positively associated with worse prognosis including cancer stage, extent and
size of tumors(30), and lymphatic invasion(31). An retrospective cohort study used diagnostic
mammograms from 28 women diagnosed with DCIS and found that DCIS occurred mostly in
areas of mammographically dense tissue and that the mammographic quandrant with the highest
percentage density consist the majority of the lesions(32).
There have been studies that indicate higher mammographic density is associated with
higher risk of recurrence of breast cancer recurrence (31, 33, 34). A prospective cohort study in
Canada following 504 women diagnosed with previous in-situ cancers measured breast cancer
recurrence and found that women with highly dense breasts had 2.8 (95% CI: 1.3 to 6.1) times
the risk of breast cancer recurrence (DCIS or invasive) and three times the risk of subsequent
invasive breast cancer (95% CI: 1.2 to 7.5) (33). Another prospective cohort study of 355
6

women showed that women with pre-treatment dense breasts are about 5.7 (95% CI: 1.6-2) times
as likely to experience recurrence in 10 years(31).
There is also evidence that mammographic density is associated with survival(35, 36, 37,
38). In prospective cohort study of 607 women, mammographic density was associated with a
reduced risk of dying from breast cancer (HR = 0.77; 95%CI: 0.60-0.99; p = 0.04) in women
who had received radiation, but with an elevated risk (HR = 1.46; 95% CI: 1.00-2.14; p = 0.05)
in patients who had not received radiation(35). Another study of 9232 women found no overall
increased risk of breast cancer death associated with higher mammographic density; however,
risk of death is elevated among women with low density (BI-RADS 1) who were either obese
(HR = 2.02, 95% CI = 1.37 to 2.97) or had tumors of at least 2.0 cm (HR = 1.55, 95% CI = 1.14
to 2.09)(37). In a more recent prospective cohort study that had mammograms of 974
postmenopausal patients at two different times, although the investigators did not found an
overall association between mammographic density and survival, they found that among
tamoxifen users, women with reduction in mammographic density of >20% following cancer
were 50% less likely to die (hazard ratio, 0.50; 95% CI, 0.27 to 0.93)(36).
Estrogen and Mammographic Density
As described previously, from animal and cell line studies, it is believed that estrogen is a
major mitogen in fibroglandular tissue formation, influencing mammographic density. In
epidemiological studies, there is indirect evidence that estrogen is a major determinant of MD.
In the literature, mammographic density decreases with greater parity(39), greater body
weight, menopause (40), and is increased by use of post-menopausal hormones such as estrogen
and progestin combined therapy (41, 42, 43); furthermore, randomized clinical trials have shown
7

that use of postmenopausal hormone therapy increased mammographic density after one year of
administration (41, 44, 45), with the strongest effect seen with combined equine estrogens plus
medroxyprogesterone. Lower mammographic density has been measured among breast cancer
survivors using the anti-estrogenic drug, tamoxifen (46, 47).
While these observations suggest that variations in mammographic density can be
attributed to the variations in endogenous hormones, studies that look at circulating estrogen and
mammographic density are not consistent (Table 1). Bioavailable estradiol is inversely
associated with mammographic density in several studies of postmenopausal women (48, 49, 50)
However, in the Nurses’ Health Study, the inverse association does not reach statistical
significance after adjusting for BMI (50). In the study by Boyd, the inverse association is
significant also without adjusting for BMI(49). Only one study with a comparatively small
sample of women found the inverse association between estrogen and mammographic density
(Table 1).
However, given the majority of the cross-sectional studies shown in Table 1 found no
association between estrogen and mammographic density among postmenopausal women, it is
worthwhile to note that the major source of steroid sex hormones in postmenopausal women is
estrone, produced by aromatase in the adipose tissue(23). Low levels of estradiol may be more
difficult to detect;furthermore, it is possible that a single measurement of circulating estrogen
may not necessarily reflect the estrogen level in the breast or the cumulative exposure to estrogen
that can directly influence epithelial and stromal cell growth.
Particularly significant is then the significant positive associations between circulating
estrogens and mammographic density in the two placebo controlled randomized trials, the
8

Postmenopausal Estrogen and Progestin Interventions (PEPI) trial (42) and the Women’s Health
Initiative (WHI) trial (41). The use of exogenous hormones elevates estrogen levels significantly.
Both studies observed positive association between circulating estrogen and mammographic
density after adjusting for BMI and prior use of hormone therapy. The WHI trial found that
women assigned to the Estrogen+Progestin treatment arm had an average 5%- 6% increase in
mammographic density respectively after 1 year, while there were only minor changes in the
placebo group or the estrogen alone arm. The PEPI trial found higher levels of estrone (β=
0.0013, p= 0.014), estradiol (β= 0.0009, p = 0.009), and bioavailable estradiol (β= 0.0021, p =
0.018) associated with mammographic density. Furthermore, PEPI also had one arm that
combined conjugated equine estrogen with micronized progesterone and the investigators
observed large variation between the mean mammographic density of the different treatments.
These results further suggest that mammographic density is influenced by changes in estrogen
levels (51, 52).
Estrogen Synthesis and Metabolism Pathway and Mammographic Density
Mammographic density has a strong genetic component. Studies of twins suggest that a
large percent of the variance is due to genetic factors (27, 53, 54). Thus, given the impact of
estrogen on mammographic density, focus has been given to the genes involved in the estrogen
metabolism pathway.
Estrogen Synthesis
Several enzymes are involved in the. After progestogen is derived from cholesterol,
cytochrome P450 17A1 (CYP17A1) enzyme metabolized the progestogens, resulting in the
production of androgens (pregnenolone to dehydroe-piandrosterone (DHEA) and progesterone to
9

androstenedione). The androgens are transformed by enzymes hydroxysteroid dehydrogenase
3B1 (HSD3B1), which converts DHEA to androstenedione, and by hydroxysteroid
dehydrogenase 17B1 (HSD17B1), which converts androstenedione to testosterone. The
androgens are then converted to the estrogens estrone (E1) and 17β-estradiol (E2) by the
cytochrome P450 19A1 (CYP19A1) enzyme (55). However, HSD17B1 could further promote
the transformation of E1 to E2, which is more bioactive(56). E1 and E2 can influence breast cell
proliferation either directly as they bind to ER, or indirectly by their further conversion to
estrogens (57)
Estrogen Metabolism
E1 and E2 undergo 2-hydroxylation by CYP1A1 and CYP1A2 to become catechol
estrogens, 2-OH-E1 and 2-OH- E2(55, 58). As an alternative, they also undergo 4-hydroxylation
by CYP1B1 and become CEs 4-OH-E
1
and 4-OH-E2 (55). These 4-OH may provide excessive
mitogenic stimulation because their binding to ER is of longer duration than that of E
2
(59)and
they are believed to generate reactive estrogen intermediates that may damage DNA and induce
tumorigenesis(56, 60). Conversely 2-OH bind to ER with less affinity compared to E
2
, making
them less potent mitogenic agents(59).
This dissertation a candidate gene analyses on the 3 functional SNPs on the estrogen
synthesis and metabolism pathway—CYP17, CYP1B1, COMT—and 1SNP of MTHFR, which
plays a role in COMT activity. Details of their biological function and previous epidemiological
studies are discussed in detail in Chapter 2.

10

A brief overview is as follows:
CYP17 codes for cytochrome p450C17α, which catalyzes the conversion of 17-
hydroxypregnenolone and 17-hydroxyprogestin to dehydroepiandrosterone (DHEA) and
androstenedione (61). The CYP17 (T27C) promoter region SNP (A2 allele) has been associated
with higher circulating estrone (E1) and estradiol (E2), however it has not been associated with
breast cancer risk in genome wide association studies (62).
CYP1B1 codes for an enzyme that catalyzes the addition of 2-hydroxyl (2-OH) and 4-
hydroxyl (4-OH) groups to E1 and E2. The CYP1B1’s Valine (Val) to Leucine (Leu)
substitution at codon 432 results in an enzyme (Leu) that may have 3 times greater activity than
the wildtype (Val)(63, 64).
The which is thought to provide excessive mitogenic stimulation of breast cells because
its binding to ER is of longer duration than that of E
2
(56) expected effect of the Leu (C) variant
on breast cancer risk is unclear since it has been associated with increased breast cell
proliferation (65).
COMT gene is located on chromosome 22q11.21. It codes for catechol-O-
methyltransferase, which catalyzes the transfer of a methyl group from S-adenosylmethione
(SAM) to catecholamines and to both 2-OH and 4-OH catecholestrogens, a process which
inactivates and detoxifies the catecholestrogens. The COMT (Val158Met) variant results in a less
active enzyme and slower 2-OH and 4-OH catecholestrogen clearance than the wildtype (Val)
(66, 67).
Previous epidemiologic studies of MTHFR and risk of breast cancer have focused on the
role of MTHFR on DNA methylation, synthesis, and repair through the folate pathway (68, 69,
11

70, 71).. While not directly in the estrogen metabolism pathway, methylene tetrahydrofolate
reductase gene MTHFR controls the availability of the methyl donor responsible for transforming
catecholestrogens to methoxyestrogens (72) which COMT activity requires. Since Alanine to
Valine variant at codon 222 have reduced enzyme activity, this variant may also reduce COMT
activity (73).
Given the highly heritable nature of breast density, understanding of the impacts of these
genes on mammographic density may lead to more personalized treatment plans for better
prognoses after breast cancer.
Inflammatory Biomarkers and Mammographic Density
Estrogen level in postmenopausal women, as well as influenced by genetic factors, is also
affected by the immune system. As women become postmenopausal and estradiol level drop
there is a corresponding increase in proinflammatory cytokines including IL-1, IL-6, and TNF-
α(74). And as IL-6 rises, C-reactive protein (CRP) and serum amyloid A (SAA) also increase.
CRP and SAA are acute phrase proteins that rise with IL-6 (75). Synthesized in the liver
and adipocytes, CRP is part of the immune system that binds the phosphocoline of certain
pathogens, dead, or dying cells. SAA is an apolipoprotein that binds to high-density lipoprotein
after their synthesis, influencing cholesterol metabolism during inflammatory states, causing
adhesion and chemotaxis of phagocitic cells and lymphocytes(76) .
CRP and SAA display a similar pattern in most inflammatory diseases, reaching a
maximum serum concentration about 24 hours after an acute inflammatory episodes before
decreasing(77). Although SAA concentrations usually parallel those of CRP, an acute episode,
such as an injury or infection, results in SAA to increase earlier and at a higher rate than CRP.
12

(76) . This has led studies some studies to suggest that although SAA as a clinically useful
marker of inflammation in bacterial or viral infection like CRP(78), it may be a more sensitive
marker of inflammatory disease(76). For cancer survivors, elevated levels of CRP and SAA
measured pre- and post-treatment are associated with reduced breast cancer survival(79, 80).
The exact mechanisms by which estrogen and cytokine interact is unknown. However,
studies have shown that cytokines, such as IL-6 and TNF-α, also stimulate aromatase activity,
and catalyzes the conversion of androstenedione to estrone in the estrogen synthesis pathway
(81). Experiments on breast cancer cell lines have shown that IL-6, in combination with estrone
sulfate (E-S), enhance cellular proliferation through their action on aromatase(82).
Another, more indirect association between cytokines and mammographic density
involves COX-2. COX-2 is believed to drive production of estrogen in the breast since the level
of COX-2 and expression of cytochrome P19 (CYP19) is correlated in human breast cancer(83).
Furthermore, non-steroidal anti-inflammatory drugs (NSAIDs) including aspirin, inhibit COX-2,
thus reduce prostaglandin E2 or local estrogen production(49), and hinder tumor cell growth in
vitro and in animal models(47, 50-53). However, if COX-2 is involved in estrogen-influenced
mammographic density, studies of NSAIDs’ association with mammographic density do not
support the hypothesis. One cross-sectional study of 3286 women found a null association(84),
another retrospective cohort study of 1492 women found long-term total NSAID use had non-
significantly higher densities than non-users(85).
In contrast to evidence of cellular proliferation due to increase in cytokines, there is
evidence that cytokines, in some circumstances, may exert an inhibitory effect on breast cell
proliferation. It has been suggested that differential intracellular signaling between mesenchyme
13

and breast cancer epithelium in normal versus cancer cells may explain the varied effect of IL-
6(86). In cancer cell lines, experimental studies have shown that proinflammatory cytokines,
including IL-6, block cell growth signaling following activation of the insulin-like growth factor-
I (IGF-I) tyrosine kinase receptor(87). For more details of the possible opposing affects of
increase of cytokines on cellular proliferation, thus, increasing mammographic density, see
Introduction and Discussion of Chapter 3 where the results of two previous studies on CRP and
mammographic density are also discussed.
Alcohol, Smoking, and Mammographic Density
The introduction to Chapter 4 of this dissertation explores the biological basis of alcohol
intake and cigarette smoking can influence mammographic density by inducing changes on
estrogen level.
The formation of genotoxic metabolites (e.g. acetaldehydes) from alcohol intake may
induce changes in circulating estrogen levels as well as insulin-like growth factor (IGF-1) and
insulin-like growth factor binding proteins (IGF-BP3)(88). Epidemiological studies have shown
alcohol associated with higher levels of estradiol and estrone levels among premenopausal
women(89, 90), or androstenedione, an estrone precursor(91) Among postmenopausal women,
increased alcohol drinking is associated estrone and estradiol(92, 93). The mechanism suggested
is the influence of alcohol on increased expression of the aromatase enzyme, which contributes
to the peripheral production of estrone, thus estradiol(94).
These higher levels elevated levels of estrogen and estrogen metabolites could influence
mammographic density by increasing rates of cell turnover and proliferation(95), increasing the
chance of mutations(60) or influence breast cancer risk by potential DNA adduct formation(60).
14

However, though alcohol and smoking habit sometime occur in tandem, the
antiestrogenic properties in the components of tobacco can reduce estrogen levels (96, 97, 98),
thus potentially mammographic density.
Table 2 shows the current literature on alcohol intake and mammographic density where
majority of the studies show that about half the studies did not observe an association between
alcohol intake and mammographic density. Among the studies that found a positive association ,
it is more consistent among premenopausal women than postmenopausal women. Table 3
presents the current literature on smoking and mammographic density, showing that most studies
found an inverse association though the amount of detail defining smoking varied greatly
between studies as many studies only compared current smoking vs. not current smoking.
Further discussion of these studies is in the Discussion section of Chapter 4.
Nevertheless, the direction of the association between alcohol/tobacco and
mammographic density agree with the association of alcohol/tobacco’s influence on estrogen
levels.
Conclusion
Percent mammographic density has been associated with both breast cancer recurrence
and survival among postmenopausal women. However, its exact biology has not been established
though there is laboratory and epidemiological evidence that estrogen is positively associated
with increasing proportion of fibroglandular element in the breast. By exploring how estrogen
metabolism genes, biomarkers of inflammation, and lifestyle impact mammographic density
through their influence on estrogen levels, this dissertation seeks to understand the underlying
biology of mammographic density in order improve breast cancer prognoses.
15

Table 1. Studies of endogenous estrogen levels and mammographic density among postmenopausal women.
Study Design (Year Published)

Study Population (Cohort) Exposure/Outcome Results
Cross-sectional; 2005(48) 173 postmenopausal women
(Physical Activity for Total Health
study)
Circulating estrone,
estradiol//percent mammographic
density
Among former hormone therapy
users, MD’s inversely associated
with circulating levels of estrone
(P = 0.01), estradiol (P = 0.003),
free estradiol
Cross-sectional; 2008(99) 226 postmenopausal women
participating in a clinical
prevention trial
circulating levels of
estradiol/percent mammographic
density
estradiol was the only biomarker
significantly correlated with
mammographic density
Population-based cross-sectional
study;2007(100)
722 postmenopausal (Tromsø
Mammography and Breast Cancer
Study)
circulating levels of sex hormones
including free and bound estradiol,
estrone/percent MD
estrone is positively associated;
free and bound estradiol are not

Cross-sectional analysis of
baseline clinical trial data;
2005(42)
603 postmenopausal
women(Postmenopausal
Estrogen/Progestin Interventions
(PEPI) Trial)
Circulating estrone and estradiol
and free estradiol/percent
mammographic density
estrone (beta = 0.0013, p =
0.014), estradiol (beta = 0.0009,
p = 0.009), and bioavailable
estradiol is positively statistically
significantly associated
Cross-sectional; 2007 (50) 520 postmenopausal women
(Harvard Nurses’ Health Study)
circulating estrogens including free
and bound estradiol, estrone
/percent mammographic density
Inverse association between
estradiol/estrone and breast
density; no association after
adjusting for BMI

Cross-sectional; 2002(49)
382 subjects, 193 premenopausal
and 189 postmenopausal.
Circulating estradiol, progesterone/
five categories of breast density
from mammography units
Inverse association with free
estradiol , did not adjust for BMI
or hormone replacement therapy
Cross-sectional; 2011 (101) 257 postmenopausal women with
no history of PMH use ( The
Wisconsin Breast Density Study)
Circulating estrogens/quartiles of
mammographic density
No association after adjustment
BMI
Cross-sectional; 2007(102) 775 postmenopausal women
(Prospect-EPIC)
circulating estrogens including free
and bound estradiol, estrone
/percent breast density
No association after adjustment
for BMI

16

Table 1, continued.

Cross-sectional; 2007 (62) 1,413 postmenopausal women from
the European Prospective
Investigation into Cancer and
Nutrition-Norfolk.
Circulating estrogens/Boyd six-
category and Wolfe four-category
scales of mammographic density
No association after adjustment
for BMI
Cross-sectional; 2009(103) 270 postmenopausal UK Caucasian
and Afro- Caribbean women
plasma biomarkers (estradiol,
estrone)/percent mammographic
density
No association after adjustment
for BMI
Cross-sectional; 2005 (104) 239 women ages 70–92 not taking
PMH participating in Study of
Osteoporotic Fractures(1986-1988)
Circulating estrone, total
estradiol(n=237)/percent
mammographic density
No association with estrone or
estradiol

17

Table 2. Studies of alcohol intake and mammographic density among postmenopausal women.
Design (Year Published) Subjects (Cohort) Exposure/Outcome Results
Population-based Cross-
sectional; 2011(105)
3584 women, aged 45-68 years,79% were
postmenopausal.42.2% ever smokers
FFQ portion and frequency of:
red wine (125 ml), white wine
(125 ml), beer (200 ml), sherry
(50 ml), hard cider (125 ml),
spirits (30 ml), and brandy, gin,
rum, whiskey, and vodka
(40 ml)*ethanol= gm/day/percent
mammographic density
Current alcohol
consumption increasing
the odds of high MD by
13% (OR = 1.13; 95% CI
0.99-1.28) and high daily
grams of alcohol being
positively associated with
increased MD (P for
trend = 0.045).
alcohol consumption and
daily grams of alcohol
were positively
associated with higher
MD in postmenopausal
women and in women
who were not currently
smoking, alcohol
consumption had no
effect on MD in
premenopausal women
and current smokers.
Cross-sectional; 2012(106) 2,251 postmenopausal women aged 50-69 years
Norwegian Breast Cancer Screening Program in
2004.
Different cutpoints based on
gm/day or drinks/week/ mean
percent mammographic density
No association

18

Table 2, continued.
Cross-sectiona; 2005 (107) Premenopausal (n = 451) Postmenopausal (n =
1442) (Minnesota Breast Cancer Family study)
Questionnaire regarding
adolescent drinking: age at
initiation, alcohol consumed at
one sitting, frequency of alcohol
before 18/percent mammographic
density
No association
Case-control; 1993(108) Breast cancer cases n=266, age matched controls
n=301
(Breast Cancer Detection and Demonstration
Project.)
Lifetime: Drinkers vs. non-
drinkers/Wolfe categories
No association
Case-control; 1988(109) 90 patients with newly diagnosed breast cancer
who were first treated in Quebec in 1982-1984.
The controls included 645 women who
participated in the Canadian National Breast
Screening Study.
Drink vs Non-drinker/Wolfe’s
categories
No association
Cross-sectional; 2007(110) 1,327 women aged 40-80 of different
ethnicities(60%postmenopausal)
Quartiles of alcohol
intake/mammographic density
assessment
No association
Cross-sectional(111) 200 cases (148postmenopausal)with high-risk
(P2 and DY) mammographic parenchymal
pattern and 200(164postmenopausal) controls
with lowrisk (N1 and P1) patterns
(EPIC-Norfolk)
Week daily food diary/ P20DY
Wolfe’s mammographic
parenchymal pattern

No association

19

Table 2, continued.
Cross-sectional; 2003(112) 296 Hispanic wome:105 premenopausal22 ever
smoker) 191 postmenopausal(22 ever smoker)
Questionnaire(No drink/1
drink/>=1 drink per week /percent
mammographic density
No association
Cross-sectional; 2010(98) 246 premenopausal+1554 post menopausal
women
Alcohol intake(frequency only )
by telephone interview/percent
mammographic density
Positive association
among both
premenopausal and
postmenopausal women

Prospective cohort--
Longitudinal; 2005(113)
2001 Healthy Women
(EPIC-Florence study)
alcohol consumption (0, <0.5,
≥0.5 and <1.0, ≥1.0 and <2.0,
≥2.0 drinks per
day)/mammographic density
Positive association with
increasing consumption
of wine
Cross-sectional analyses of
nested case-control and
intervention study;
2006(114)
217 premenopausal women + 1,250 ((592 cases
and 658 controls))primarily postmenopausal
women (The Breast, Estrogens, and Nutrition
Study; BEAN)
FFQ including questions on draft
beer, light beer, white or pink
wine, red wine and hard liquor.
/mean percent density taken
before diagnoses
postmenopausal women
without hormone
replacement therapy
(HRT), breast density
increased by 2% for each
higher alcohol intake
category.
Cross-sectional; 2000(115) 1508 women in a historical cohort study of breast
cancer families in Minnesota
Never drinkers vs. ≤3.9 g/day vs
>3.9 g/day/percent breast density
Marginal positive
association among
premenopausal c(P
trend
=
0.08) and
postmenopausal women
(P
trend
= 0.09). ;

20

Table 3. Studies of mammographic density and cigarette smoking.
Design (Year Published) Subjects (Cohort) Exposure/Outcome Results
Cross-sectional; 2000(98) 246 premenopausal+1554 post
menopausal
Smoking history((never smoker, former
smoker with <20 pack-years, former
smoker with ≥20 pack-years, current
smoker with <20 pack-years and current
smoker with≥ 20 pack-years)/percent
mammographic density
Inverse association among pre-
but not in postmenopausal women
Cross-sectional; 2009(97) 799 (63% smokers)pre- and early
perimenopausal women in the Study
of Women’s Health Across the
Nation (SWAN).
self-administered questionnaire at
baseline(smoking age, smoking
duration, cigarettes per day, smoking
status)/percent mammographic density
Inverse association with current
active smoking, smoked
≥20 cigarettes/day
Cross-sectional; 2005
(104)
239 women ages 70–92 not taking
PMH participating in Study of
Osteoporotic Fractures(1986-1988)
Current(n=12) vs. Not Current
smoker(n=226)/percent mammographic
density
Inverse association; not current
smokers (26.0versus 17.3% for
smokers, p =0.02).

Cross-sectctional;
2012(116)
907 postmenopausal
participants(65% ever smokers)
(Tromsø Mammography and Breast
Cancer study)
Lifetime smoking history was collected
through interview and
questionnaires(cigarettes smoked daily,
years smoked, smoking
status)/percentage and absolute
mammographic density
Inverse dose-response association:
number of cigarettes (p=.003) and
the number of pack-years smoked
among current smokers (p=.006)
Current and former smokers had
significantly lower adjusted mean
percentage mammographic density
compared with never smokers
(p=.008)
21

Table 3, continued.
Cross-sectional;(117) 528 women(455 postmenopausal+68
premenopausal+10 unknown):
(Indian, Aleut, or Eskimo- 43, 37,
50% smokers respectively)
Current smoking: yes vs. no/(BIRADS)
density patterns and percent
mammographic density
No association
Nested Case-control;
2000(118)
406 (cases are2 03 women with
P2/DY , matched with 203 women
with N1/P1
patterns)women(313postmenopausal)
from European Prospective
Investigation on Cancer in Norfolk
(EPIC-Norfolk)
Smoking status(never/past/current)/(148
cases/165 controls)/Wolfe's high-risk
mammographic parenchymal patterns
No association
Population-based Cross-
sectional; 2011(105)
3584 women, aged 45-68 years,79%
were postmenopausal.42.2% ever
smokers
Questionnaire including current and
lifetime e) current smoking (yes or no);
(f) lifetime tobacco smoking (never,
past, or current smoker); (g) daily
amount of cigarettes; (h) accumulated
amount of cigarettes; and (i) age at
which smoking had been
initiated/categories of mammographic
density namely: A (0%); B (<10%); C
(10–25%); D (25–50%); E (50–75%);
and F (>75%)
Positive association with daily
cigarettes, number of accumulated
lifetime cigarettes (P for trend
0.017 and 0.021)
Cross-sectiona; 2004(96) Premenopausal 628 women questionnaire questions: smoking at
university, smoker vs. nonsmoker/9
category classification (SCC) of breast
density percent.
Positive association with smoking
at university, OR smokers versus
non-smokers: 0.58 (0.36-0.92) –
protective against high risk MD
category
22

Cross-sectional;
2003(112)
296 Hispanic women:105
premenopausal(22 ever smoker) 191
postmenopausal(22 ever smoker)
health and lifestyle
questionnaire(never/past/current)
/percent mammographic density
Positive association: among
premenopausal ever smokers had
higher percentage of density (beta
= 6.23%; P = 0.06) compared with
nonsmokers. No association for
postmenopausal women.

23

1.2 References
1. McCormack VA, dos Santos Silva I. Breast density and parenchymal patterns as markers
of breast cancer risk: a meta-analysis. Cancer Epidemiol Biomarkers Prev
2006;15(6):1159-69.
2. Yaffe M, Boyd N. Mammographic breast density and cancer risk: the radiological view.
Gynecol Endocrinol 2005;21 Suppl 1:6-11.
3. Kopans DB. Breast Imaging. 3 ed. Philadelphia, PA: Lippincott Williams & Wilkins;
2007.
4. Arthur JE, Ellis IO, Flowers C, et al. The relationship of "high risk" mammographic
patterns to histological risk factors for development of cancer in the human breast. Br J
Radiol 1990;63(755):845-9.
5. Boyd NF, Lockwood GA, Byng JW, et al. Mammographic densities and breast cancer
risk. Cancer Epidemiol Biomarkers Prev 1998;7(12):1133-44.
6. Guo YP, Martin LJ, Hanna W, et al. Growth factors and stromal matrix proteins
associated with mammographic densities. Cancer Epidemiol Biomarkers Prev
2001;10(3):243-8.
7. Alowami S, Troup S, Al-Haddad S, et al. Mammographic density is related to stroma and
stromal proteoglycan expression. Breast Cancer Res 2003;5(5):R129-35.
8. Li T, Sun L, Miller N, et al. The association of measured breast tissue characteristics with
mammographic density and other risk factors for breast cancer. Cancer Epidemiol
Biomarkers Prev 2005;14(2):343-9.
9. Warren R, Lakhani SR. Can the stroma provide the clue to the cellular basis for
mammographic density? Breast Cancer Res 2003;5(5):225-7.
10. Camps JL, Chang SM, Hsu TC, et al. Fibroblast-mediated acceleration of human
epithelial tumor growth in vivo. Proc Natl Acad Sci U S A 1990;87(1):75-9.
11. Silberstein GB. Tumour-stromal interactions. Role of the stroma in mammary
development. Breast Cancer Res 2001;3(4):218-23.
12. Boyd NF, Jensen HM, Cooke G, et al. Relationship between mammographic and
histological risk factors for breast cancer. J Natl Cancer Inst 1992;84(15):1170-9.
13. Urbanski S, Jensen HM, Cooke G, et al. The association of histological and radiological
indicators of breast cancer risk. Br J Cancer 1988;58(4):474-9.
14. Bright RA, Morrison AS, Brisson J, et al. Relationship between mammographic and
histologic features of breast tissue in women with benign biopsies. Cancer
1988;61(2):266-71.
15. Moskowitz M, Gartside P, McLaughlin C. Mammographic patterns as markers for high-
risk benign breast disease and incident cancers. Radiology 1980;134(2):293-5.
16. Hawes D, Downey S, Pearce CL, et al. Dense breast stromal tissue shows greatly
increased concentration of breast epithelium but no increase in its proliferative activity.
Breast Cancer Res 2006;8(2):R24.
17. Bissell MJ, Hines WC. Why don't we get more cancer? A proposed role of the
microenvironment in restraining cancer progression. Nat Med 2011;17(3):320-9.
18. Roskelley CD, Bissell MJ. Dynamic reciprocity revisited: a continuous, bidirectional
flow of information between cells and the extracellular matrix regulates mammary
epithelial cell function. Biochem Cell Biol 1995;73(7-8):391-7.
19. Lochter A, Bissell MJ. Involvement of extracellular matrix constituents in breast cancer.
Semin Cancer Biol 1995;6(3):165-73.
24

20. Boyd NF, Guo H, Martin LJ, et al. Mammographic density and the risk and detection of
breast cancer. N Engl J Med 2007;356(3):227-36.
21. Lundholt BK, Briand P, Lykkesfeldt AE. Growth inhibition and growth stimulation by
estradiol of estrogen receptor transfected human breast epithelial cell lines involve
different pathways. Breast Cancer Res Treat 2001;67(3):199-214.
22. Zhang HZ, Bennett JM, Smith KT, et al. Estrogen mediates mammary epithelial cell
proliferation in serum-free culture indirectly via mammary stroma-derived hepatocyte
growth factor. Endocrinology 2002;143(9):3427-34.
23. Bulun SE, Lin Z, Imir G, et al. Regulation of aromatase expression in estrogen-
responsive breast and uterine disease: from bench to treatment. Pharmacol Rev
2005;57(3):359-83.
24. Woolcott CG, Cook LS, Courneya KS, et al. Associations of overall and abdominal
adiposity with area and volumetric mammographic measures among postmenopausal
women. Int J Cancer 2011;129(2):440-8.
25. van Gils CH, Hendriks JH, Otten JD, et al. Parity and mammographic breast density in
relation to breast cancer risk: indication of interaction. Eur J Cancer Prev 2000;9(2):105-
11.
26. Mitchell G, Antoniou AC, Warren R, et al. Mammographic density and breast cancer risk
in BRCA1 and BRCA2 mutation carriers. Cancer Res 2006;66(3):1866-72.
27. Boyd NF, Dite GS, Stone J, et al. Heritability of mammographic density, a risk factor for
breast cancer. N Engl J Med 2002;347(12):886-94.
28. Ursin G, Pike M. Mammographic density, hormone therapy, and risk of breast cancer.
Cancer Epidemiol Biomarkers Prev 2006;15(9):1750; author reply
29. Boyd NF, Martin LJ, Rommens JM, et al. Mammographic density: a heritable risk factor
for breast cancer. Methods Mol Biol 2009;472:343-60.
30. Ghosh K, Brandt KR, Sellers TA, et al. Association of mammographic density with the
pathology of subsequent breast cancer among postmenopausal women. Cancer Epidemiol
Biomarkers Prev 2008;17(4):872-9.
31. Cil T, Fishell E, Hanna W, et al. Mammographic density and the risk of breast cancer
recurrence after breast-conserving surgery. Cancer 2009;115(24):5780-7.
32. Ursin G, Hovanessian-Larsen L, Parisky YR, et al. Greatly increased occurrence of breast
cancers in areas of mammographically dense tissue. Breast Cancer Res 2005;7(5):R605-
8.
33. Habel LA, Dignam JJ, Land SR, et al. Mammographic density and breast cancer after
ductal carcinoma in situ. J Natl Cancer Inst 2004;96(19):1467-72.
34. Cil T, Fishell E, Hanna W, et al. Mammographic density and the risk of breast cancer
recurrence after breast-conserving surgery. Cancer 2009;115(24):5780-7.
35. Maskarinec G, Pagano IS, Little MA, et al. Mammographic density as a predictor of
breast cancer survival: the Multiethnic Cohort. Breast Cancer Res 2013;15(1):R7.
36. Li J, Humphreys K, Eriksson L, et al. Mammographic Density Reduction Is a Prognostic
Marker of Response to Adjuvant Tamoxifen Therapy in Postmenopausal Patients With
Breast Cancer. J Clin Oncol 2013.
37. Gierach GL, Ichikawa L, Kerlikowske K, et al. Relationship between mammographic
density and breast cancer death in the Breast Cancer Surveillance Consortium. J Natl
Cancer Inst 2012;104(16):1218-27.
25

38. Chiu SY, Duffy S, Yen AM, et al. Effect of baseline breast density on breast cancer
incidence, stage, mortality, and screening parameters: 25-year follow-up of a Swedish
mammographic screening. Cancer Epidemiol Biomarkers Prev 2010;19(5):1219-28.
39. Hankinson SE, Colditz GA, Hunter DJ, et al. Reproductive factors and family history of
breast cancer in relation to plasma estrogen and prolactin levels in postmenopausal
women in the Nurses' Health Study (United States). Cancer Causes Control
1995;6(3):217-24.
40. Boyd N, Martin L, Stone J, et al. A longitudinal study of the effects of menopause on
mammographic features. Cancer Epidemiol Biomarkers Prev 2002;11(10 Pt 1):1048-53.
41. McTiernan A, Martin CF, Peck JD, et al. Estrogen-plus-progestin use and
mammographic density in postmenopausal women: Women's Health Initiative
randomized trial. J Natl Cancer Inst 2005;97(18):1366-76.
42. Greendale GA, Palla SL, Ursin G, et al. The association of endogenous sex steroids and
sex steroid binding proteins with mammographic density: results from the
Postmenopausal Estrogen/Progestin Interventions Mammographic Density Study. Am J
Epidemiol 2005;162(9):826-34.
43. Vachon CM, Sellers TA, Vierkant RA, et al. Case-control study of increased
mammographic breast density response to hormone replacement therapy. Cancer
Epidemiol Biomarkers Prev 2002;11(11):1382-8.
44. 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(4 Pt 1):262-9.
45. Greendale GA, Reboussin BA, Slone S, et al. Postmenopausal hormone therapy and
change in mammographic density. J Natl Cancer Inst 2003;95(1):30-7.
46. Boyd NF. Tamoxifen, mammographic density, and breast cancer prevention. J Natl
Cancer Inst 2011;103(9):704-5.
47. Cuzick J, Warwick J, Pinney E, et al. Tamoxifen-induced reduction in mammographic
density and breast cancer risk reduction: a nested case-control study. J Natl Cancer Inst
2011;103(9):744-52.
48. Aiello EJ, Tworoger SS, Yasui Y, et al. Associations among circulating sex hormones,
insulin-like growth factor, lipids, and mammographic density in postmenopausal women.
Cancer Epidemiol Biomarkers Prev 2005;14(6):1411-7.
49. Boyd NF, Stone J, Martin LJ, et al. The association of breast mitogens with
mammographic densities. Br J Cancer 2002;87(8):876-82.
50. Tamimi RM, Hankinson SE, Colditz GA, et al. Endogenous sex hormone levels and
mammographic density among postmenopausal women. Cancer Epidemiol Biomarkers
Prev 2005;14(11 Pt 1):2641-7.
51. Ursin G, Palla SL, Reboussin BA, et al. Post-treatment change in serum estrone predicts
mammographic percent density changes in women who received combination estrogen
and progestin in the Postmenopausal Estrogen/Progestin Interventions (PEPI) Trial. J
Clin Oncol 2004;22(14):2842-8.
52. Crandall CJ, Guan M, Laughlin GA, et al. Increases in serum estrone sulfate level are
associated with increased mammographic density during menopausal hormone therapy.
Cancer Epidemiol Biomarkers Prev 2008;17(7):1674-81.
53. Stone J, Dite GS, Gunasekara A, et al. The heritability of mammographically dense and
nondense breast tissue. Cancer Epidemiol Biomarkers Prev 2006;15(4):612-7.
26

54. Ursin G, Lillie EO, Lee E, et al. The relative importance of genetics and environment on
mammographic density. Cancer Epidemiol Biomarkers Prev 2009;18(1):102-12.
55. Kulendran M, Salhab M, Mokbel K. Oestrogen-synthesising enzymes and breast cancer.
Anticancer Res 2009;29(4):1095-109.
56. Yager JD, Liehr JG. Molecular mechanisms of estrogen carcinogenesis. Annu Rev
Pharmacol Toxicol 1996;36:203-32.
57. Russo J, Ao X, Grill C, et al. Pattern of distribution of cells positive for estrogen receptor
alpha and progesterone receptor in relation to proliferating cells in the mammary gland.
Breast Cancer Res Treat 1999;53(3):217-27.
58. Tsuchiya Y, Nakajima M, Yokoi T. Cytochrome P450-mediated metabolism of estrogens
and its regulation in human. Cancer Lett 2005;227(2):115-24.
59. Barnea ER, MacLusky NJ, Naftolin F. Kinetics of catechol estrogen-estrogen receptor
dissociation: a possible factor underlying differences in catechol estrogen biological
activity. Steroids 1983;41(5):643-56.
60. Cavalieri EL, Stack DE, Devanesan PD, et al. Molecular origin of cancer: catechol
estrogen-3,4-quinones as endogenous tumor initiators. Proc Natl Acad Sci U S A
1997;94(20):10937-42.
61. Hanukoglu I. Steroidogenic enzymes: structure, function, and role in regulation of steroid
hormone biosynthesis. J Steroid Biochem Mol Biol 1992;43(8):779-804.
62. Warren R, Skinner J, Sala E, et al. Associations among mammographic density,
circulating sex hormones, and polymorphisms in sex hormone metabolism genes in
postmenopausal women. Cancer Epidemiol Biomarkers Prev 2006;15(8):1502-8.
63. Hanna IH, Dawling S, Roodi N, et al. Cytochrome P450 1B1 (CYP1B1)
pharmacogenetics: association of polymorphisms with functional differences in estrogen
hydroxylation activity. Cancer Res 2000;60(13):3440-4.
64. Li DN, Seidel A, Pritchard MP, et al. Polymorphisms in P450 CYP1B1 affect the
conversion of estradiol to the potentially carcinogenic metabolite 4-hydroxyestradiol.
Pharmacogenetics 2000;10(4):343-53.
65. Shimada T, Watanabe J, Kawajiri K, et al. Catalytic properties of polymorphic human
cytochrome P450 1B1 variants. Carcinogenesis 1999;20(8):1607-13.
66. Goodman JE, Lavigne JA, Wu K, et al. COMT genotype, micronutrients in the folate
metabolic pathway and breast cancer risk. Carcinogenesis 2001;22(10):1661-5.
67. Dawling S, Roodi N, Mernaugh RL, et al. Catechol-O-methyltransferase (COMT)-
mediated metabolism of catechol estrogens: comparison of wild-type and variant COMT
isoforms. Cancer Res 2001;61(18):6716-22.
68. Chen J, Gammon MD, Chan W, et al. One-carbon metabolism, MTHFR polymorphisms,
and risk of breast cancer. Cancer Res 2005;65(4):1606-14.
69. Shrubsole MJ, Gao YT, Cai Q, et al. MTHFR polymorphisms, dietary folate intake, and
breast cancer risk: results from the Shanghai Breast Cancer Study. Cancer Epidemiol
Biomarkers Prev 2004;13(2):190-6.
70. Gao CM, Kazuo T, Tang JH, et al. [MTHFR polymorphisms, dietary folate intake and
risks to breast cancer]. Zhonghua Yu Fang Yi Xue Za Zhi 2009;43(7):576-80.
71. Alshatwi AA. Breast cancer risk, dietary intake, and methylenetetrahydrofolate reductase
(MTHFR)single nucleotide polymorphisms. Food Chem Toxicol 2010;48(7):1881-5.
27

72. Le Marchand L, Haiman CA, Wilkens LR, et al. MTHFR polymorphisms, diet, HRT, and
breast cancer risk: the multiethnic cohort study. Cancer Epidemiol Biomarkers Prev
2004;13(12):2071-7.
73. Campbell IG, Baxter SW, Eccles DM, et al. Methylenetetrahydrofolate reductase
polymorphism and susceptibility to breast cancer. Breast Cancer Res 2002;4(6):R14.
74. Pfeilschifter J, Koditz R, Pfohl M, et al. Changes in proinflammatory cytokine activity
after menopause. Endocr Rev 2002;23(1):90-119.
75. Yamada T. Serum amyloid A (SAA): a concise review of biology, assay methods and
clinical usefulness. Clin Chem Lab Med 1999;37(4):381-8.
76. Gabay C, Kushner I. Acute-phase proteins and other systemic responses to inflammation.
N Engl J Med 1999;340(6):448-54.
77. Meier-Ewert HK, Ridker PM, Rifai N, et al. Absence of diurnal variation of C-reactive
protein concentrations in healthy human subjects. Clin Chem 2001;47(3):426-30.
78. Hilliquin P. Biological markers in inflammatory rheumatic diseases. Cell Mol Biol
(Noisy-le-grand) 1995;41(8):993-1006.
79. Pierce BL, Ballard-Barbash R, Bernstein L, et al. Elevated Biomarkers of Inflammation
Are Associated With Reduced Survival Among Breast Cancer Patients. J Clin Oncol
2009.
80. Allin KH, Nordestgaard BG, Flyger H, et al. Elevated pre-treatment levels of plasma C-
reactive protein are associated with poor prognosis after breast cancer: a cohort study.
Breast Cancer Res 2011;13(3):R55.
81. Purohit A, Newman SP, Reed MJ. The role of cytokines in regulating estrogen synthesis:
implications for the etiology of breast cancer. Breast Cancer Res 2002;4(2):65-9.
82. Honma S, Shimodaira K, Shimizu Y, et al. The influence of inflammatory cytokines on
estrogen production and cell proliferation in human breast cancer cells. Endocr J
2002;49(3):371-7.
83. Brueggemeier RW, Quinn AL, Parrett ML, et al. Correlation of aromatase and
cyclooxygenase gene expression in human breast cancer specimens. Cancer Lett
1999;140(1-2):27-35.
84. Stone J, Willenberg L, Apicella C, et al. The association between mammographic density
measures and aspirin or other NSAID use. Breast Cancer Res Treat 2012;132(1):259-66.
85. Maskarinec G, Urano Y, Gill J, et al. Nonsteroidal anti-inflammatory drugs (NSAIDs)
and mammographic density. Breast Cancer Res Treat 2008;112(1):133-9.
86. Hutchins D, Steel CM. Regulation of ICAM-1 (CD54) expression in human breast cancer
cell lines by interleukin 6 and fibroblast-derived factors. Int J Cancer 1994;58(1):80-4.
87. Shen WH, Zhou JH, Broussard SR, et al. Proinflammatory cytokines block growth of
breast cancer cells by impairing signals from a growth factor receptor. Cancer Res
2002;62(16):4746-56.
88. Singletary KW, Gapstur SM. Alcohol and breast cancer: review of epidemiologic and
experimental evidence and potential mechanisms. JAMA 2001;286(17):2143-51.
89. Muti P, Trevisan M, Micheli A, et al. Alcohol consumption and total estradiol in
premenopausal women. Cancer Epidemiol Biomarkers Prev 1998;7(3):189-93.
90. Reichman ME, Judd JT, Longcope C, et al. Effects of alcohol consumption on plasma
and urinary hormone concentrations in premenopausal women. J Natl Cancer Inst
1993;85(9):722-7.
28

91. Dorgan JF, Reichman ME, Judd JT, et al. The relation of reported alcohol ingestion to
plasma levels of estrogens and androgens in premenopausal women (Maryland, United
States). Cancer Causes Control 1994;5(1):53-60.
92. Gavaler JS. Alcohol effects on hormone levels in normal postmenopausal women and in
postmenopausal women with alcohol-induced cirrhosis. Recent Dev Alcohol
1995;12:199-208.
93. Onland-Moret NC, Peeters PH, van der Schouw YT, et al. Alcohol and endogenous sex
steroid levels in postmenopausal women: a cross-sectional study. J Clin Endocrinol
Metab 2005;90(3):1414-9.
94. Mendelson CR, Evans CT, Simpson ER. Regulation of aromatase in estrogen-producing
cells. J Steroid Biochem 1987;27(4-6):753-7.
95. Sutherland RL, Prall OW, Watts CK, et al. Estrogen and progestin regulation of cell cycle
progression. J Mammary Gland Biol Neoplasia 1998;3(1):63-72.
96. Jeffreys M, Warren R, Gunnell D, et al. Life course breast cancer risk factors and adult
breast density (United Kingdom). Cancer Causes Control 2004;15(9):947-55.
97. Butler LM, Gold EB, Conroy SM, et al. Active, but not passive cigarette smoking was
inversely associated with mammographic density. Cancer Causes Control
2010;21(2):301-11.
98. Vachon CM, Kuni CC, Anderson K, et al. Association of mammographically defined
percent breast density with epidemiologic risk factors for breast cancer (United States).
Cancer Causes Control 2000;11(7):653-62.
99. Johansson H, Gandini S, Bonanni B, et al. Relationships between circulating hormone
levels, mammographic percent density and breast cancer risk factors in postmenopausal
women. Breast Cancer Res Treat 2008;108(1):57-67.
100. Bremnes Y, Ursin G, Bjurstam N, et al. Endogenous sex hormones, prolactin and
mammographic density in postmenopausal Norwegian women. Int J Cancer
2007;121(11):2506-11.
101. Sprague BL, Trentham-Dietz A, Gangnon RE, et al. Circulating sex hormones and
mammographic breast density among postmenopausal women. Horm Cancer
2011;2(1):62-72.
102. Verheus M, Peeters PH, van Noord PA, et al. No relationship between circulating levels
of sex steroids and mammographic breast density: the Prospect-EPIC cohort. Breast
Cancer Res 2007;9(4):R53.
103. McCormack VA, Dowsett M, Folkerd E, et al. Sex steroids, growth factors and
mammographic density: a cross-sectional study of UK postmenopausal Caucasian and
Afro-Caribbean women. Breast Cancer Res 2009;11(3):R38.
104. Modugno F, Ngo DL, Allen GO, et al. Breast cancer risk factors and mammographic
breast density in women over age 70. Breast Cancer Res Treat 2006;97(2):157-66.
105. Cabanes A, Pastor-Barriuso R, Garcia-Lopez M, et al. Alcohol, tobacco, and
mammographic density: a population-based study. Breast Cancer Res Treat
2011;129(1):135-47.
106. Qureshi SA, Couto E, Hofvind S, et al. Alcohol intake and mammographic density in
postmenopausal Norwegian women. Breast Cancer Res Treat 2011.
107. Vachon CM, Sellers TA, Janney CA, et al. Alcohol intake in adolescence and
mammographic density. Int J Cancer 2005;117(5):837-41.
29

108. Herrinton LJ, Saftlas AF, Stanford JL, et al. Do alcohol intake and mammographic
densities interact in regard to the risk of breast cancer? Cancer 1993;71(10):3029-35.
109. Brisson J, Verreault R, Morrison AS, et al. Diet, mammographic features of breast tissue,
and breast cancer risk. Am J Epidemiol 1989;130(1):14-24.
110. Maskarinec G, Pagano I, Chen Z, et al. Ethnic and geographic differences in
mammographic density and their association with breast cancer incidence. Breast Cancer
Res Treat 2007;104(1):47-56.
111. Sala E, Warren R, Duffy S, et al. High risk mammographic parenchymal patterns and
diet: a case-control study. Br J Cancer 2000;83(1):121-6.
112. Gapstur SM, Lopez P, Colangelo LA, et al. Associations of breast cancer risk factors with
breast density in Hispanic women. Cancer Epidemiol Biomarkers Prev
2003;12(10):1074-80.
113. Masala G, Ambrogetti D, Assedi M, et al. Dietary and lifestyle determinants of
mammographic breast density. A longitudinal study in a Mediterranean population. Int J
Cancer 2006;118(7):1782-9.
114. Maskarinec G, Takata Y, Pagano I, et al. Alcohol consumption and mammographic
density in a multiethnic population. Int J Cancer 2006;118(10):2579-83.
115. Vachon CM, Kushi LH, Cerhan JR, et al. Association of diet and mammographic breast
density in the Minnesota breast cancer family cohort. Cancer Epidemiol Biomarkers Prev
2000;9(2):151-60.
116. Bremnes Y, Ursin G, Bjurstam N, et al. Different measures of smoking exposure and
mammographic density in postmenopausal Norwegian women: a cross-sectional study.
Breast Cancer Res 2007;9(5):R73.
117. Roubidoux MA, Kaur JS, Griffith KA, et al. Relationship of mammographic
parenchymal patterns to breast cancer risk factors and smoking in Alaska Native women.
Cancer Epidemiol Biomarkers Prev 2003;12(10):1081-6.
118. Sala E, Warren R, McCann J, et al. Smoking and high-risk mammographic parenchymal
patterns: a case-control study. Breast Cancer Res 2000;2(1):59-63.

30

Chapter 2
Mammographic density and CYP17, CYP1B1, COMT, and MTHFR polymorphisms in
postmenopausal breast cancer patients
2.1 Abstract
Authors: Anne Dee
1
, Anne McTiernan
2
, Leslie Bernstein,
1,5
Erin Aiello
2
, Richard N.
Baumgartner
3
, Kathy Baumgartner
3
, Rachel Ballard-Barbash
4
, Roberta McKean-Cowdin
1

1
University of Southern California, Los Angeles, CA,
2
Fred Hutchinson Cancer Research Center,
Seattle, WA,
3
University of Louisville, Louisville, KY,
4
National Cancer Institute, Bethesda,
MD,
5
City of Hope, Duarte, CA.
Introduction: Increased mammographic density is strongly associated with increased risk of
breast cancer incidence and worse prognosis. Genetic variants influencing lifetime exposure to
estrogen and progesterone have been considered as factors that influence mammographic density
in postmenopausal women. This analysis examines the association between 3 SNPs
(CYP1B1,Val423Leu and COMT,Val108/158Met) in the estrogen metabolism pathway,
MTHFR(Ala222Val) which influences COMT activity, and percent mammographic density. We
further examined whether factors that influence lifetime exposure to estrogen and progesterone
modify these associations including post-menopausal hormone therapy (PMH; ever and at the
time of diagnosis), body mass index (BMI), and age at menarche.
Methods: The analysis utilized mammograms obtained before and after breast cancer diagnosis
for post-menopausal breast cancer survivors in the Health, Eating, Activity, and Lifestyle Study.
Regression models were used to calculate geometric means and 95% confidence intervals (CIs)
for percent mammographic density by genotype, PMH use(never, ever estrogen only, ever
31

estrogen and progesterone), BMI (<=25 kg/m2,>25 kg/m2), and age at menarche (<12.5
years,>=12.5 years).
Results: We found a dose-response association between COMT and pre-diagnosis percent
mammographic density (p-trend=0.02); women with COMT Val/Val had higher mammographic
density than women with Met/Met (13.1%; 95%CI:10.5-16.2 vs. 9.03%; 95%CI:6.31-12.9). The
association between COMT and percent mammographic density was stronger among women
taking estrogen and progestin (p-trend=0.03) than among non-PMH users; Val/Val carriers had
higher mammographic density (21.1%; 95% CI:15.0-29.6) than Val/Met (16.2%; 95%CI 12.3-
21.4), or Met/Met carriers (12.3%; 95% CI:8.80-17.1). The association between COMT and
mammographic density did not differ by BMI or age at menarche. No association was found with
CYP1B1. Percent mammographic density was lower among women with the MTHFR Val/Val
(9.03%; 95%CI:6.31-12.9), or Val/Ala (10.4%:8.29,13.1), compared to Ala/Ala
(13.1%;95CI:10.5,16.2), p-trend=0.02. Associations are not significant between the SNPS and
post-diagnosis mammographic density.
Conclusion/Implications: The relationship between estrogen metabolism genes and
mammographic density remains an important area of investigation for breast cancer mechanisms
and prognosis.
32

2.2 Introduction
Increased mammographic density is strongly associated with increased breast cancer risk
(20); however, the mechanisms that may explain the association between mammographic density
and breast cancer are not fully understood. Risk factors for mammographic density that also may
influence breast cancer risk include genetic polymorphisms in the estrogen biosynthesis and
metabolism pathway (119). The associations described between genetic variants and
mammographic density in epidemiologic studies have been inconsistent, potentially due to
variability in study populations, small sample sizes, and the limited number of genetic variants
that have been studied (120). Other factors which influence lifetime exposure to steroid
hormones including use of postmenopausal hormones, age at menarche, and reproductive
history, also have been considered important potential predictors of both breast cancer risk and
mammographic density (41, 121, 122, 123, 124).
The evidence for an association between steroid hormones and mammographic density
has varied by the source of hormone measured. Higher mammographic density has been found
among women using exogenous hormones such as estrogen and progestin combined therapy
(41, 42, 43); furthermore, randomized clinical trials have shown that use of postmenopausal
hormone therapy increased mammographic density after one year of administration (41, 44, 45),
with the strongest effect seen with combined equine estrogens plus medroxyprogesterone.
Lower mammographic density has been measured among breast cancer survivors using the anti-
estrogenic drug, tamoxifen (46, 47). The greatest decline in mammographic density during a
woman’s lifetime occurs at the age of menopause (40). However, studies of directly measured
circulating steroid hormone levels and mammographic densities in postmenopausal women
without breast cancer are conflicting, with studies reporting no (48, 49), positive (42), and
33

negative associations (50).
Candidate genes in the estrogen biosynthesis and metabolism pathway have been
evaluated as risk factors for mammographic density in several studies (120, 122, 125, 126, 127,
128). Single nucleotide polymorphisms (SNPs) in 3 estrogen synthesis and metabolism genes
including CYP17 (T27C), CYP1B1 (Val 432 Leu), and COMT (Val 158 Met) have been found to
alter circulating hormones and cellular proliferation (129). A functional SNP in the methylene
tetrahydrofolate reductase gene MTHFR (Ala222Val) may alter MTHFR activity which
influences COMT enzyme activity by controlling the availability of the methyl donor responsible
for transforming catecholestrogens to methoxyestrogens (72). CYP17 codes for cytochrome
p450C17α, which catalyzes the conversion of 17-hydroxypregnenolone and 17-hydroxyprogestin
to dehydroepiandrosterone (DHEA) and androstenedione (61). The CYP17 (T27C) promoter
region SNP (A2 allele) has been associated with higher circulating estrone (E1) and estradiol
(E2), however it has not been associated with breast cancer risk in genome wide association
studies (62). The CYP1B1 gene codes for an enzyme that catalyzes the addition of 2-hydroxyl (2-
OH) and 4-hydroxyl (4-OH) groups to E1 and E2. The CYP1B1 Valine (Val) to Leucine (Leu)
substitution at codon 432 results in an enzyme (Leu) that may have 3 times greater activity than
the wildtype (Val)(63, 64). COMT codes for catechol-O-methyltransferase, which catalyzes the
transfer of a methyl group from S-adenosylmethione (SAM) to catecholamines and to both 2-OH
and 4-OH catecholestrogens, a process which inactivates and detoxifies the catecholestrogens.
The COMT (Val158Met) variant results in a less active enzyme and slower 2-OH and 4-OH
catecholestrogen clearance than the wildtype (Val) (66, 67). Individuals with the MTHFR
Alanine to Valine variant at codon 222 have reduced enzyme activity, and therefore may have
reduced COMT activity (73).
34

We evaluated the associations of 4 selected SNPs in CYP17, CYP1B1, COMT, and
MTHFR with mammographic density in postmenopausal breast cancer survivors participating in
the Health, Eating, and Activity Lifestyle (HEAL) Study. Factors that influence mammographic
density are of importance to breast cancer survivors as data indicate that higher mammographic
density is predictive of greater risk of breast cancer recurrence (33, 34). We focused on
postmenopausal women because the majority of breast cancer survivors in the HEAL Study are
postmenopausal, and any large differences in mammographic density when comparing
premenopausal to postmenopausal women may be due to the large decline in circulating steroid
hormones with the cessation of ovarian function, rather than small changes in enzyme efficiency
due to SNPs. We evaluated mammographic density at two time points: approximately one year
prior to diagnosis and two years after diagnosis. We further explored the association between
each SNP and mammographic density by history within categories of selected reproductive
factors (i.e. parity, age at menarche), body mass index (BMI), and postmenopausal hormone use,
and assessed interactions between the SNPs in CYP17, CYP1B1, COMT and MTHFR.
2.3 Methods
Study population
Data for this analysis were collected for the HEAL Study, a population-based prospective
study of women diagnosed with in-situ to stage IIIa breast cancer from 1996 through 1999. At
initiation, the HEAL Study included 1,183 women, 18 years of age or older, who were identified
through the Surveillance, Epidemiology, and End Results

(SEER) registries in New Mexico
(n=615), Los Angeles County, California (n=366) and western Washington (n=202). Baseline
data were collected within the first year after diagnosis, on average 6 months post diagnosis. A
follow-up interview was conducted about two years later, on average 31 months post-diagnosis.
35

The study was designed to evaluate the independent roles of sex-hormones, diet, weight, physical
activity, genetics, and other factors on breast cancer prognosis and survival. Details of study
design and recruitment procedures have been described previously (130, 131, 132). The
institutional review board eat ach participating center approved the study in accord with
assurances filed with and approved by the US Department of Health and Human Services. We
obtained written informed consent from all study participants.
The current analysis includes postmenopausal women with pre-diagnosis and post-
diagnosis mammograms obtained from medical providers. Of the 1,183 HEAL participants, 825
women had mammograms available. If a woman did not have a post-diagnosis mammogram, she
was still included in the pre-diagnosis analyses and vice-versa. For all analyses, we used the
image from the breast contralateral to where the breast cancer was diagnosed. Women with
bilateral breast cancer were excluded from this analysis.
The pre-diagnosis exam used the mammogram taken closest in time to 12-months before
a woman’s diagnosis date. In this sample, there were 687 pre-diagnosis mammograms; however,
33 mammograms were excluded due to poor image quality. Of the 654 mammograms available,
412 were from postmenopausal women. Of those 412 women, genotyping data were available
for 350 women. An additional 19 women were excluded for missing covariate data. The final
analyses of pre-diagnosis mammograms included 331 women (Figure 1).
The post-diagnosis exam used the mammogram taken closest in time to 24-months after a
woman’s diagnosis date. A total of 756 post-diagnosis mammograms were collected; however,
44 mammograms were excluded due to poor image quality. Of the 712 mammograms available,
551 were from postmenopausal women. Genotyping data were available for 474 women of these
36

551 women. An additional 36 women were excluded for missing covariate data. Therefore, the
final analyses included 438 women with post-diagnosis mammograms (Figure 1).
Data collection
Questionnaire Variables:
Women completed mailed (western Washington) or interview-administered (New
Mexico and Los Angeles) questionnaires at baseline (on average 6 months post-diagnosis) and at
the 24-month follow-up (on average 31 months post-diagnosis). Detailed information was
collected on demographics, reproductive history, postmenopausal hormone therapy use, oral
contraceptive use, and menopausal status. Data on tumor characteristics were obtained from
medical record review, and participants’ SEER reports.
Clinical Variables:
Trained staff measured height and weight at the baseline for New Mexico and western
Washington participants. In Los Angeles, weight and height five years before diagnosis was
self reported. All sites measured weight at the 24 month follow up assessment. BMI was
calculated as weight in kilograms (kg) divided by height in meters squared (m
2
).
Genotype assessment:
The genotyping methods used have been described in detail elsewhere (133). Briefly, in
western Washington and Los Angeles, DNA was extracted from buffy coat fractions using
Phenol/Chloroform methods (134). The CYP17, CYP1B1, COMT, and MTHFR genotyping were
performed at Albany Molecular Research in Bothell, Washington, using the Taqman allelic
discrimination method with the ABI 7700 Sequence Detection System (Applied Biosystems,
Foster City, CA), which has been described (135, 136). CYP17 and CYP1B1 probes were
complementary to their corresponding sense strands, whereas the COMT probes were
37

complementary to the antisense strand. We included 10% replicate samples and genotype
concordance was 100%.
In New Mexico, DNA was isolated from buffy coats using the Puregene DNA isolation
kit from Gentra Systems (Minneapolis, MN) at the Protein Chemistry Laboratory at the
University of New Mexico Health Science Center (Albuquerque, NM) (133). Briefly, the CYP17
T27C SNP was assessed using a PCR/MspA-I restriction enzyme procedure established by
Feigelson et al (137); CYP1B1 was analyzed using a PCR/Eco 571 restriction enzyme procedure
established by Bailey et al (138); and COMT genotype was determined using a PCR/Nla-III
restriction enzyme procedure(139). For quality control, a common set of DNA was genotyped
across centers with concordance of 100%.
Mammogram density assessment:
We collected mammograms using information on provider names and mammogram dates
reported by women on their questionnaires. Mammograms were digitized using an Epson 1680
scanner (Epson America Inc., CA). We measured dense area (per 1,000 pixels) and total breast
area (per 1,000 pixels), and from these, calculated percent mammographic density (dense
area/total breast area * 100) using the craniocaudal view of the breast contralateral to the one
where breast cancer was diagnosed. All density readings were conducted by one of the
investigators (EJAB) using Cumulus, a computer program developed by the University of
Toronto (33); a 5% random sample of films were digitized and read for quality control. The
correlation coefficients comparing the original measures with the repeat measures were 0.86 for
percent density, 0.89 for dense area, and 0.70 for total breast area.

38

Statistical analyses
We analyzed the association between each SNP and percent mammographic density
separately for pre-diagnosis and post-diagnosis mammograms. Potential associations between
genotype and potential confounders evaluated with logistic regression. Geometric means and
their 95% confidence intervals (CI) were calculated for percent mammographic density and
dense area (x1000 pixels) for association between potential confounder and mammographic
density before being calculated for each genotype in the final model (Supp Tables. We tested for
linear trend in geometric means across genotypes and considered p-values <0.05 as statistically
significant. Stratum-specific estimates were calculated to evaluate potential effect modification
by use of postmenopausal hormone therapy at the time of diagnosis (never, current estrogen
only, current estrogen and progestin, and past users of estrogen and progestin), BMI (≤25 vs. >25
kg/m
2
), parity (nulliparous vs. parous), and age at menarche (median split <12.5 vs. ≥12.5 years),
and race/ethnicity (African-American, Hispanic, non-Hispanic white from New Mexico, non-
Hispanic white from Washington, and other). We used ANCOVA to calculate geometric means
and used Scheffe multiple comparison procedures (p<0.05) to compare differences across
groups. We adjusted for age at mammogram (years continuous), race/ethnicity, postmenopausal
hormone therapy, age at menarche (years continuous), and other polymorphisms (categories; 0,
1, 2) as covariates in our models. We modeled BMI, age, and postmenopausal hormone therapy
use at the time of the pre-diagnosis mammogram for all pre-diagnosis analyses; BMI and age at
the time of the post-diagnosis mammogram was used for post-diagnosis analyses; we did not
adjust for postmenopausal hormone therapy use for analyses of the post-diagnosis using SAS 9.2
(SAS Institute, Cary, NC).

39

2.4 Results
The analyses include 331 women with pre-diagnosis mammograms and 438 women with
post-diagnosis mammograms, of whom 293had mammograms available from both time points.
The mean percent density was 19.8% for pre-diagnosis mammograms and 15.6% for post-
diagnosis mammograms (Table 1). Among the 293 women who had mammograms available
from both time points the mean percent density was similar (19.9% for pre-diagnosis
mammograms and14.8% for post-diagnosis mammograms. Approximately 76% of the women
were diagnosed with early-stage breast cancer (in-situ or localized). Of the women with pre-
diagnosis mammograms, approximately 70% of the women self-identified themselves as non-
Hispanic white, 14% as African American, and 12% as Hispanic. Of the women with post-
diagnosis mammograms, 61% self-identified as non-Hispanic white, followed by African-
American (26%), and Hispanic (10%). Genotype frequencies for each of the 4 SNPs are shown
in Table 1.
No consistent pattern was observed between the CYP17 genotypes or the CYP1B1
genotypes and percent mammographic density for pre or post-diagnosis mammograms (Table 2).
For COMT genotypes, using pre-diagnosis mammograms, we found a pattern of decreasing
mean percent mammographic density with increasing copies of the COMT variant coding for
Met in models that adjusted for age, race/ethnicity, BMI, postmenopausal hormone use, age of
menarche, and 2 other genotypes (CYP17, CYP1B1) (p-trend=0.04). Women homozygous for the
high-activity COMT variant (Val) had the highest percent mammographic density and women
homozygote for the low-activity variant (Met) had the lowest density. Post-diagnosis
mammograms of women showed the same pattern, although the test for trend was not
statistically significant (p-trend =0.18) (Table 2 and Supp Table 1a-b). We found a similar
40

pattern with the MTHFR variant (Val), where increasing number of the MTHFR variant was
associated with decreasing percent density for pre-diagnosis (p-trend=0.02) and post-diagnosis
(p-trend=0.12) mammograms.
When we examined the effect of the 4 SNPs on percent mammographic density by use of
postmenopausal hormone therapy we observed the highest breast densities among women who
were current users of estrogen and progesterone postmenopausal hormone therapy at time of
breast cancer diagnosis. We observed no consistent association between CYP17 or CYP1B1
genotype and percent mammographic density for pre-diagnosis mammograms (Table 3) or post-
diagnosis mammograms (data not shown). The association of decreasing percent
mammographic density with increasing copies of the low-activity COMT variant (Met) was
strongest among women who reported current use of estrogen and progestin (Val/Val 25.8%;
Val/Met 19.3%; Met/Met 14.9%; p-trend=0.03) at the time of diagnosis. The pattern was similar
for women reporting current use of estrogen alone (Val/Val 16.1%; Val/Met 13.8%; Met/Met
9.8%; p-trend=0.06) (Table 3 and Supp Table 2) and a similar decreasing trend was observed
among women reporting ever use of estrogen alone, or ever use of estrogen and progestin (data
not shown).
We observed no evidence of heterogeneity of effect between each SNP and percent
mammographic density when we stratified by age at menarche, parity, or BMI for either pre-
diagnosis or post-diagnosis mammograms (Supp Table 3).
In Table 4, we show the mean percent mammographic density by 2-way combinations of
the select SNPs. We observed a statistically significant interaction between COMT and CYP17
(p-interaction = 0.02). Among the low activity COMT carriers (Met/Met) who are less able to
clear 2-OH and 4-OH estrogens, we found a pattern of increasing mammographic density with
41

increasing copies of the CYP17 A2 allele (A2 allele is associated with higher E2 synthesis) (P-
trend=0.01). We also found among the low activity COMT carriers a pattern of decreasing
mammographic density among women with low activity MTHFR (Val) (p trend = 0.03); women
with the low activity COMT and MTHFR variants had the lowest percent mammographic
density.
2.5 Discussion
In this study, we found an association between the COMT Val158Met variant and the
MTHFR Ala222Val variant and percent mammographic density using mammograms of breast
cancer survivors taken approximately 1 year before their diagnosis. The variant coding for low
activity COMT (Met) and the low activity MTHFR (Val) were associated with lower levels of
pre-diagnosis mammographic density. The strongest association between COMT and percent
mammographic density was found among women who reported taking estrogen alone or
estrogen and progestin combined therapy near the time of diagnosis. Further, we also observed
that women who reported never using postmenopausal hormone therapy had lower mean percent
mammographic density than women who reported taking postmenopausal hormone therapy ever
or at the time of diagnosis. We found an interesting pattern between the variants, such that
women with the low activity variants for COMT and MTHFR had lower mean percent
mammographic density (5.6%; 95%CI 2.8-11.1) than women with the high activity COMT and
the high activity MTHFR variant (14.2% 95%CI 10.5-19.2).
Randomized trials have shown that use of postmenopausal hormone therapy increases
percent mammographic density after one year of administration (41, 44, 45), with the strongest
effect seen with combined equine estrogens plus medroxyprogesterone. In our data, the
association between the COMT Val158Met SNP and mammographic density was strongest
42

among ever users of estrogen and progestin who were still using postmenopausal hormone
therapy at the time of diagnosis. Two previous papers found that the associations between
COMT, CYP1B1 and mammographic density were modified by reproductive factors including
parity, menarche, and obesity (122). In contrast to these findings, we did not observe evidence
of effect modification by reproductive factors in our data.
The association between COMT and mammographic density seems biologically
plausible, given the role of COMT in estrogen metabolism (129). Our observation of lower
mammographic density among women with lower COMT activity may be due to the fact that
COMT acts to clear 2-hydroxy estrogens which inhibit breast cell proliferation. In laboratory
studies, 2-hydroxyestrone (2-OHE1) is observed to suppress the growth and proliferation of
hormone-dependent breast cancer cell lines (140, 141). The other 2-catecholestrogen, 2-OHE2,
also inhibits hormone-induced cell proliferation (142). Since 2-OHE1 and 2-OHE2 are cleared
more slowly than 4-OHE1 and 4-OHE2 (67), there may be a buildup of 2-OHE1 and 2-OHE2
among women with the lower activity (Met variant) COMT genotype. Therefore, slower
clearance of 2-OHE1 and 2-OHE2 may result in decreased mammographic density through the
inhibition of breast cell proliferation.
One genome-wide association study (GWAS) considering genetic variations and
mammographic density has been published (143). This study identified a SNP in region 12q24
(rs1265507) associated with mammographic density, located between TBX5 and TBX3,which
encodes transcription factors involved in developmental regulation of embryos including limb
and heart development. The GWAS included data from 1,241 women resulting in the selection
of the top 99 SNPs from approximately 521,571 SNPS(2,510,880 SNPs after imputation) with
the smallest p-values. A separate pooled analysis of genetic data from 7,018 pre and
43

postmenopausal women from seven additional studies; however, 33 mammograms were
excluded due to poor image quality. Of the 654 mammograms available, was used for
replication and found rs1265507 on chromosome 12q24 had the strongest association with
percent mammographic density (p=2.74×10
-5
). The 12q24 SNP reached genome-wide statistical
significance with the inclusion of two additional studies (n=10,377), (p=1.03x10
-8)
(143).
Eight previous studies of candidate genes have examined the association between COMT
and percent mammographic density. In three cross-sectional studies of healthy women, lower
mean percent density was found among women with the COMT variant (Met) compared to
women with the wildtype (Val) among premenopausal women (122, 144, 145). Two
independent studies of postmenopausal women found no association between the COMT variant
and percent mammographic density (62, 146). In both studies, a significant association was
found unadjusted for body size (i.e. BMI); adjustment for BMI attenuated the association. The
Nurses' Health Study found no statistically significant association between COMT genotype and
mammographic density among premenopausal or postmenopausal women (125). However, in a
study of African-American and Caucasian breast cancer cases in Los Angeles, CA, current
postmenopausal hormone therapy users who carried the low activity COMT variant had greater
percent mammographic density. No association was found between COMT and mammographic
density when ignoring history of postmenopausal hormone therapy (146). The European
Prospective Investigation into Cancer and Nutrition (EPIC) study also found no statistically
significant association between COMT Val158Met and mammographic density overall (62).
This is the first study, to our knowledge, to report on the direct association between
MTHFR (Ala222Val) variant and percent mammographic density. Previous epidemiologic
studies of MTHFR and risk of breast cancer have focused on the role of MTHFR on DNA
44

methylation, synthesis, and repair through the folate pathway (68, 69, 70, 71). In a meta-analysis
of 41 studies including 16,480 cases and 22,388 controls, the MTHFR Ala222Val variant was not
associated with elevated breast cancer risk when restricted to studies of postmenopausal women
(147). In the Multiethnic Cohort (MEC), investigators detected an inverse association between
the MTHFR Val/Val genotype and breast cancer risk among women on postmenopausal hormone
therapy (72). A nested case-control study (cases=88, controls=344) of a cancer screening cohort
investigating the MTHFR SNP found that women with the low activity variant (Val) were at
higher risk of breast cancer, but only among women with a greater time interval between age at
menarche and first full-term pregnancy. The authors suggested that the impact of MTHFR on
breast cancer risk may be modified by levels of endogenous estrogen exposure (148). Women
with lower MTHFR activity may have lower clearance of catecholestrogens, due to the reduced
efficiency of 2-OH estrogen clearance via COMT. While catecholestrogens may decrease breast
cell proliferation, they are also suspected of being breast cell carcinogens (149). Women with
longer exposure of undifferentiated breast cells to elevated levels of circulating estrogens or
progestins, as indicated by early age at menarche and first pregnancy at an older age, may be at a
higher risk of breast cancer (150).
We observed no direct association between the CYP17 T27C SNP and percent
mammographic density. While early studies of young, nulliparous women found the women
carrying the A2 allele of the CYP17 variant had higher levels of circulating serum estradiol and
estrone, the functional relevance of the SNP has not been demonstrated in experimental studies
(151). With respect to mammographic density, none of the three studies that examined the
potential association between CYP17 and mammographic density found significant associations
(62, 125, 128)
45

Studies of the CYP1B1 (Val 432 Leu) variant and mammographic density have largely
found no significant associations, although a few case-control studies have reported associations
between the CYP1B1 variant and mammographic density among women using postmenopausal
hormone therapy. A cohort study of 1,260 women who were past or never users of
postmenopausal hormone therapy did not find an association between CYP1B1 and percent
mammographic density (62). A cross-sectional study of 538 women using postmenopausal
hormone therapy also showed no association between CYP1B1 genotype and mammographic
density (6). However, in a clinical trial involving 232 women, investigators found a statistically
significant interaction between the CYP1B1 polymorphism and use of postmenopausal hormone
therapy (yes/no) with mammographic density (P = 0.0004); the largest increase in
mammographic density due to use of postmenopausal hormone therapy was observed among
women homozygous for the variant (Leu/Leu) using estrogen and progestin (127). In contrast,
an analysis of 140 current users of postmenopausal hormone therapy found a non-statistically
significant association between CYP1B1 and mammographic percent density such that women
with Leu/Leu had lower percent mammographic density than carriers of Val/Val. (125).
Both strengths and limitations of the study must be considered in the interpretation of the
results. Strengths of this study include the availability of both pre- and post-diagnosis
mammograms, the estimation of densities by one reader, the inclusion of detailed lifestyle
information (e.g. reproductive history, exogenous hormone use), and dates of mammograms and
diagnosis (i.e. length of time between mammography and diagnosis). Furthermore, our study
focuses on the potential relevance of genetic variation on mammographic density among breast
cancer survivors, a characteristic which is considered important to breast cancer prognosis. A
limitation of our study is the relatively small sample size, and therefore our inability to explore
46

differences in associations by racial/ethnic subgroups. We therefore report our primary findings
for all racial/ethnic groups combined with adjustment for race/ethnicity; results for analyses
stratified by race/ethnicity are available to the reader in supplemental tables. While we controlled
for the elapsed time between the mammogram date and the date of diagnosis in our models, the
variability of this time window for individual participants may increase the variability in
reporting exposure history accuracy. Furthermore, menopausal status data was obtained either 6
months after diagnosis or 30 months after diagnosis and not at the time of when the
mammograms were taken. Data of the date of last menstruation were not available for all
women. This result may skew
In this study of breast cancer survivors, we found an association between the low activity
COMT Val158Met SNP and the low activity MTHFR Ala222Val SNP and percent
mammographic density when considering mammograms taken approximately 1 year before
diagnosis; these associations were strongest among women taking estrogen and progestin near
the time of diagnosis. No association was found with post-diagnosis mammograms, but
mammographic density 2 years post diagnosis may have been impacted by treatment and further
aging related involutional breast changes. The impact of these findings on mammographic
density among breast cancer survivors is important given the published associations between
percent mammographic density and breast cancer recurrence (31, 33).
47

Table 2. Demographic characteristics of study participants at baseline (pre-diagnosis N=331) and 24-
months after baseline (post-diagnosis N=438).

Pre-Diagnosis

Post-Diagnosis
N=331 N=438

Mean (SD)

Mean (SD)
Age at mammogram (years) 60.4 8.58

60.5 9.79
Body Mass Index 26.9 5.79

27.6 6.31
Age at menarche 12.5 1.59

12.5 1.60
Age at menopause
1
47.6 7.79

47.7 7.65
Age at first live birth
2
22.8 4.82

23.1 5.19
Length of progestin use (years)
3
9.8 8.02

2.93 6.95
Length of estrogen use (years)
4
13.6 10.7

7.73 10.8
Time between mammogram and diagnosis (months) 16.5 11.9

20.9 7.44
Total breast area (in 1000 pixels) 394 184 369 160
Dense breast area (in 1000 pixels) 3.87 1.15

4.02 1.01
Percent mammographic density (%) 19.8 14.6

15.6 12.4

n(%)

n(%)

Postmenopausal hormone therapy use

Never 60(18.2)

138(31.8)

Past Use of Estrogen or Progestin 61(18.2)

NA

Current (Use at Diagnosis)
5

Estrogen only 91(27.7)

NA

Combined Estrogen and Progestin 117(35.6)

NA

Ever

Ever Estrogen only Use 112(35.5)

126(28.8)

Ever Estrogen and Progestin Use 159(48.2)

174(39.7)

Former oral contraceptive use (yes) 192 (58.0)

282(64.5)
6

Stage of disease
NA

In Situ

84(19.2)

Localized

248(56.6)

Regional

106(24.2)

Table 1, continued.

1
Age of menopause data missing for 102 women in Post-diagnosis data (n=336). Menopause includes both natural
and menopause resulting from treatment or surgery.
2
Data is not available for 40 women in Pre-diagnosis data (n=291). Data is not available for 56 women in Post-
diagnosis(n=384). The youngest at first live birth is 12 years; the oldest is 40 years (for Pre-dx) and 41(Post-dx).
3
Data of length of progestin use missing for 176 women (Pre-diagnosis n=155, Post-diagnosis=434).
4
Data of length of estrogen use missing for 76 women (Pre-diagnosis n=255, Post-diagnosis n=426).
5
Recency of post-menopausal hormone use missing for 2 women.
6
Data of use oral contraceptives missing for one woman (n=437).
48

ER status
Positive NA

261(59.6)

Negative

69(15.8)

Unknown

108(11.0)

Tamoxifen use for treatment (yes) NA

226(51.6)

with Positive ER status
with Negative ER status
with Unknown ER status

180(41.1)
15(3.4)
31(7.1)

Marital Status

Married 211(63.8)

275(62.8)

Widowed 50(15.1)

60(13.7)

Divorced/Separated 56(16.9)

87(19.8)

Never married 14(4.23)

16(3.7)

Parous (yes) 290(87.6)

285(89.9)
Number of Live Births

None 40(12.1)

54(12.3)
<3 137(41.4)

203(46.4)
≥3 154(46.5)

181(41.3)
Age at First Birth

<18 27(9.3)

42(10.9)

18-25 186(63.9)

235(61.2)

>25 78(26.8)

107(27.9)

Race

Non-Hispanic white – Western Washington 74(22.2)

80(18.3)
Non-Hispanic white – New Mexico 160(48.3)

187(43.0)
African-American – Los Angeles 46(13.9)

112(25.6)
Hispanic – New Mexico 40(12.1)

45(10.3)
Other 11(3.30)

14(3.20)

CYP17 genotype

A1/A1 125(37.8)

165(37.7)
A1/A2 170(51.4)

220(50.2)
A2/A2 36(10.9)

53(12.1)

CYP1B1 genotype

Val/Val 96(29.0)

154(35.2)

Val/Leu 180(54.4)

210(48.)

Leu/Leu 55(16.6)

74(16.9)

49

Table 1, continued.
COMT genotype

Val/Val 97(29.3)

136(31.1)

Val/Met 135(40.8)

193(44.1)

Met/Met 99(29.9)

109(24.9)

MTHFR genotype

Ala/Ala 148(44.7)

227(51.8)

Ala/Val 148(44.7)

167(38.13)

Val/Val 35(10.6)

44(10.1)

*Other include: Asian, Pacific Islander, American-Indian, and Other
50

Table 2. Geometric mean levels* of percent mammographic density among postmenopausal women
by CYP17, COMT and CYP1B1 genotype before and after diagnosis of breast cancer
Pre-Diagnosis Post-Diagnosis
N=331 N=438

N
Mean % density
(95% CI)
P for trend
N
Mean % density
(95% CI)
P for trend

CYP17

0.09

0.96
A1/A1 125 9.22(7.31-11.6) 165 7.97(6.36-9.98)
A1/A2 170 10.7(8.68-13.3) 220 7.09(5.74-8.75)
A2/A2 36 12.1(8.51-17.2) 53 8.89(6.36-12.4)

0.72

0.13 CYP1B1
Val/Val 96 10.9(8.47-14.1) 154 8.90(7.15-11.1)
Val/Leu 180 10.7(8.53-13.4) 210 7.77(6.33-9.55)
Leu/Leu 55 10.3(7.59-14.0)
0.04

74 7.17(5.43-9.45)
0.18
COMT
Val/Val 97 13.4(10.5-17.1) 136 9.31(7.49-11.6)
Val/Met 135 11.2(8.91-14.1)

0.02
193 8.84(7.27-10.8)
0.12 Met/Met 99 9.97(7.79-12.8) 109 7.70(6.11-9.72)

MTHFR
Ala/Ala 148 13.1(10.5-16.2)

227 9.17(7.47-11.2)
Ala/Val 148 10.4(8.29-13.1)

167 8.83(7.06-11.0)
Val/Val 35 9.03(6.31-12.9)

44 6.40(4.46-9.18)

*adjusted for age, race, body mass index, post-menopausal hormone use(never, ever estrogen use
only, ever estrogen and progestin use), age at menarche(years), other polymorphisms; MTHFR and
COMT are not mutually adjusted
51

*adjusted for age, race, body mass index, post-menopausal hormone use(never, ever estrogen use only, ever estrogen and progestin use)
**past users not shown (n=61), recency of post-menopausal hormone missing for 2 women

Table 3. Adjusted* mean levels of pre-diagnosis percent mammographic density among postmenopausal women by CYP17,
COMT and CYP1B1 genotype and stratified by post-menopausal hormone use .
Pre-Diagnosis
Never Current Estrogen Only Current Estrogen and Progestin
N=60 N=91 N=117

N
Mean % density
(95% CI)
P for
N
Mean % density
(95% CI)
P for
N
Mean % density
(95% CI)
P for
trend trend trend
CYP17
0.55

0.23

0.36
A1/A1 23 5.83(2.65-12.9) 37 10.2(5.99-17.3) 38 16.2(10.6-24.7)
A1/A2 31 11.0(4.84-24.8) 46 11.5(6.86-19.1) 62 16.1(10.9-23.6)
A2/A2 6 4.85(1.46-16.2) 8 17.2(7.66-38.5) 17 22.9(12.5-42.0)

CYP1B1

Val/Val 13 11.7(4.01-33.9)

26 9.93(5.34-18.5)

35 17.8(11.9-26.6)
Val/Leu 38 6.11(3.23-11.6)

47 14.8(8.88-24.7)

69 14.8(9.67-22.6)
Leu/Leu 9 4.35(1.46-13.0) 0.15 18 13.6(7.37-25.1) 0.22 13 22.6(12.2-42.1) 0.75

COMT

Val/Val 20 6.55(3.36-12.8)

25 16.1(9.33-27.9)

33 25.8(16.4-40.5)
Val/Met 28 7.68(3.96-14.9)

35 13.8(7.92-24.2)

49 19.3(12.6-29.6)
Met/Met 12 6.79(2.67-17.3) 0.86 31 9.76(5.63-16.9) 0.06 35 14.9(9.23-24.0) 0.03

MTHFR

Ala/Ala 30 7.73(4.05-14.8)

42 15.4(9.35-25.3)

43 23.2(15.2-35.5)
Ala/Val 26 6.50(3.27-12.9)

39 9.58(5.60-16.4)

57 20.9(13.5-32.5)
Val/Val 4 6.15(1.42-26.6) 0.64 10 14.7(6.89-31.4) 0.27 17 13.2(7.37-23.5) 0.09

52

Table 4. Geometric mean levels* of pre-diagnosis percent mammographic density among postmenopausal women for
CYP17xCOMT and MTHFRxCOMT.
Pre-diagnosis Mean % Density (95%CI)

n

p-trend p-interaction**

MTHFR (Ala222Val)

COMT
(Val108/158Met)

Ala/Ala Ala/Val Val/Val
Val/Val 51/34/12 14.2(10.5-19.2) 11.8(8.17-17.0) 14.4(8.17-25.6) 0.89 .11
Val/Met 57/63/15 12.3(9.09-16.7) 10.8(8.15-14.4) 7.76(4.61-13.1) 0.07
Met/Met 40/51/8 12.6(9.04-17.5) 8.80(6.37-12.1) 5.56(2.79-11.1) 0.03
p-trend

0.55 0.24 0.05

CYP17 (T27C)
COMT
(Val108/158Met)
(Val108/158Met)

A1/A1 A1/A2 A2/A2

.02
Val/Val 40/52/5 11.0(7.84-15.4) 12.3(9.2-16.4) 7.97(3.38-18.8) 0.87
Val/Met 50/67/18 10.4(7.67-14.0) 9.19(6.82-12.4) 10.3(7.53-14.0) 0.93
Met/Met 35/51/13 5.82(4.03-8.41) 10.3(7.53-14.0) 14(8.14-24.0) 0.01
p-trend
0.01 0.42 0.63

*adjusted for age, race, body mass index, postmenopausal hormone use (never, current estrogen use only, current estrogen and progestin
use), age at menarche (years), other polymorphisms (COMT, CYP17, CYP1B1,MTHFR)

**interaction p value obtained via formal statistical interaction test between genotypes(continuous) adjusted for covariates including age,
race, body mass index, postmenopausal hormone use (never, current estrogen use only, current estrogen and progestin use), age at menarche
(years), other polymorphisms (COMT, CYP17, CYP1B1,MTHFR)
53

Supplement Table 1a. Adjusted* geometric mean levels of percent mammographic density among postmenopausal women by CYP17,
COMT and CYP1B1 genotype and stratified by race, before and after breast cancer
diagnosis.
Pre-Diagnosis
Non-Hispanic White
(Western Washington)
Non-Hispanic White
(New Mexico)
African-American
(Los Angeles)
Hispanic
N=74 N=160 N=46 N=40

N
Mean
percent
density
(95% CI)
P
for
N
Mean percent
density
(95% CI)
P for
N
Mean percent
density
(95% CI)
P for
N
Mean percent
density
(95% CI)
P for
tren
d
trend trend `trend
CYP17

A1/A1 23 9.07(4.97-16.6)

58 11.2(8.65-14.6)

23 9.98(5.55-17.9)

19 16.9(8.43-34.0)
A1/A2 39 8.81(5.34-14.5)

85 14.3(11.1-18.3)

19 10.1(5.73-18.0)

19 13.5(6.18-29.6)
A2/A2 12 13.7(5.95-31.3) 0.4
8
17 16.0(10.3-24.9) 0.08 4 10.3(3.59-29.5) 0.95 2 3.59(0.572-
22.5)
0.18

CYP1B1

Val/Val 19 13.0(7.05-24.1)

34 17.1(12.2-23.9)

28 10.6(5.88-19.0)

9 8.37(3.19-22.0)
Val/Leu 44 8.07(5.02-13.0)

100 16.4(13.1-20.5)

13 8.78(4.32-17.9)

19 11.4(5.02-25.8)
Leu/Leu 11 12.1(5.66-26.1) 0.7 26 12.8(8.78-18.7) 0.45 5 8.54(3.18-22.9) 0.44 12 12.4(5.35-28.8) 0.49

COMT

Val/Val 20 11.3(5.77-22.0)

34 17.2(12.3-24.0)

22 14.1(8.20-24.3)

16 8.76(3.69-20.8)
Val/Met 30 9.82(5.87-16.4)

73 15.9(12.3-20.5)

15 7.93(4.38-14.4)

14 8.74(3.76-20.3)
Met/Met 24 13.1(7.2-23.8) 0.6
6
53 11.2(8.37-15.0) 0.02 9 12.0(5.65-25.6) 0.37 10 8.82(3.59-21.7) 0.99

MTHFR

Ala/Ala 29 13.6(8.15-22.7)

64 19.5(15.0-25.3)

37 10.5(6.70-16.4)

13 8.33(3.68-18.9)
Ala/Val 38 12.0(7.68-18.7)

76 14.4(11.2-18.4)

9 8.39(3.93-17.9)

21 8.62(4.23-17.6)
Val/Val 7 7.65(3.12-18.8) 0.3
3
20 12.7(8.32-19.4) 0.02 0 NA 0.46 6 16.2(6.03-43.5) 0.58

*adjusted for age, race, body mass index, post-menopausal hormone use(never, ever estrogen use only, ever estrogen and progestin use), age
at menarche(years), other polymorphisms; MTHFR and COMT are not mutually adjusted.
54

Supplement Table 1b. Adjusted* mean levels of percent mammographic density among postmenopausal women by CYP17, COMT and
CYP1B1 genotype and stratified by race, before and after breast cancer diagnosis.
Post-Diagnosis
Non-Hispanic White
(Western Washington)
Non-Hispanic White
(New Mexico)
African-American
(Los Angeles)
Hispanic
N=80 N=187 N=112 N=45

N
Mean percent
density
(95% CI)
P for
N
Mean percent
density
(95% CI)
P for
N
Mean percent
density
(95% CI)
P for
N
Mean percent
density
(95% CI)
P for
trend trend trend trend
CYP17

A1/A1 28 8.55(5.33-13.7)

69 8.84(6.93-11.3)

46 9.98(5.55-17.9)

19 8.66(4.01-18.7)
A1/A2 42 7.75(5.17-11.6)

95 9.50(7.46-12.1)

53 5.85(3.32-10.3)

21 8.03(3.98-16.2)
A2/A2 10 8.77(4.03-19.1) 0.92 23 11.3(7.55-17.0) 0.3 13 9.57(4.15-22.1) 0.08 5 9.34(2.55-34.3) 0.99

CYP1B1

Val/Val 26 13.0 (7.05-24.1)

42 17.1(12.2-23.9)

67 10.6(5.88-19)

10 8.37(3.19-22.0)
Val/Leu 41 8.07(5.02-13.0)

11
4
16.4(13.1-20.5)

36 8.78(4.32-17.9)

15 11.4(5.02-25.8)
Leu/Leu 13 12.1(5.66-26.1) 0.26 31 12.8(8.78-18.7) 0.31 9 8.54(3.18-22.9) 0.73 20 12.4(5.35-28.8) 0.33

COMT

Val/Val 18 8.33(4.55-15.2)

42 11.3(8.24-15.4)

50 9.51(5.66-16.0)

20 12.2(5.97-25.1)
Val/Met 38 11.1(7.37-16.8)

85 11.4(8.94-14.6)

51 6.76(4.20-10.9)

14 9.36(4.54-19.3)
Met/Met 24 8.88(5.4-14.6) 0.93 61 9.70(7.41-12.7) 0.39 11 10.2(4.31-23.9) 0.6 11 5.50(2.46-12.3) 0.12

MTHFR

Ala/Ala 31 13.6(8.15-22.7)

80 10.7(8.46-13.7)

90 8.02(5.37-12)

19 8.33(3.68-18.9)
Ala/Val 42 12.0(7.68-18.7)

80 12.1(9.50-15.4)

22 8.73(4.3-17.7)

19 8.62(4.23-17.6)
Val/Val 7 7.65(3.12-18.8) 0.04 27 7.37(4.99-10.9) 0.28 0 NA 0.81 7 16.2(6.03-43.5) 0.9

*adjusted for age, race, body mass index, post-menopausal hormone use(never, ever estrogen use only, ever estrogen and progestin use), age
at menarche(years), other polymorphisms; MTHFR and COMT are not mutually adjusted.

55

Supplement Table 2. Adjusted* mean levels of pre-diagnosis percent mammographic density among postmenopausal women* by CYP17,
COMT and CYP1B1 genotype and stratified by post-menopausal hormone use .
Pre-Diagnosis
Never Past Current Estrogen Only Current Estrogen and
Progesterone
N=60 N=61 N=91 N=117

N
Mean % density
(95% CI)
P for
N
Mean % density
(95% CI)
P for
N
Mean % density
(95% CI)
P for
N
Mean % density
(95% CI)
P for
trend trend trend trend
CYP17
0.55

0.14

0.23

0.36
A1/A1 23 5.83(2.65-12.9) 26 9.49(5.64-16.0) 37 10.2(5.99-17.3) 38 16.2(10.6-24.7)
A1/A2 31 11.0(4.84-24.8) 30 11.0 (7.08-17.9) 46 11.5(6.86-19.1) 62 16.1(10.9-23.6)
A2/A2 6 4.85(1.46-16.2) 5 17.2(8.18-36.2) 8 17.2(7.66-38.5) 17 22.9(12.5-42.0)

CYP1B1

Val/Val 13 11.7(4.01-33.9)

21 14.1(8.53-23.5)

26 9.93(5.34-18.5)

35 17.8(11.9-26.6)
Val/Leu 38 6.11(3.23-11.6)

26 11.0 (7.8-20.7)

47 14.8(8.88-24.7)

69 14.8(9.67-22.6)
Leu/Leu 9 4.35(1.46-13.0) 0.15 14 10.2(5.65-18.4) 0.72 18 13.6(7.37-25.1) 0.22 13 22.6(12.2-42.1) 0.75

COMT

Val/Val 20 6.55(3.36-12.8)

18 15.5(9.3-25.8)

25 16.1(9.33-27.9)

33 25.8(16.4-40.5)
Val/Met 28 7.68(3.96-14.9)

23 11.0(6.65-17.0)

35 13.8(7.92-24.2)

49 19.3(12.6-29.6)
Met/Met 12 6.79(2.67-17.3) 0.86 20 13.7(8.93-20.9) 0.27 31 9.76(5.63-16.9) 0.06 35 14.9(9.23-24.0) 0.03

MTHFR

Ala/Ala 30 7.73(4.05-14.8)

32 12.3(8.16-18.5)

42 15.4(9.35-25.3)

43 23.2(15.2-35.5)
Ala/Val 26 6.50(3.27-12.9)

25 11.0(10.1-25.5)

39 9.58(5.60-16.4)

57 20.9(13.5-32.5)
Val/Val 4 6.15(1.42-26.6) 0.64 4 9.18(3.96-21.3) 0.27 10 14.7(6.89-31.4) 0.27 17 13.2(7.37-23.5) 0.09
*2 women have missing post-menopausal hormone use data.
**adjusted for age, race, body mass index, post-menopausal hormone use(never, ever estrogen use only, ever estrogen and progestin use),
age at menarche(years), other polymorphisms; MTHFR and COMT are not mutually adjusted.

56

SupplementTable 3. Modifying effect of estrogen-related factors on the association between adjusted* mean levels of percent
mammographic % density among postmenopausal women by CYP17, COMT and CYP1B1 genotypes with pre-diagnosis mammograms.
Gene

Pre-diagnosis Mean % Density (95%CI)

Post-diagnosis Mean % Density (95%CI)

Age at
menarche
n
Early Menarche
<12.5yrs (n=174)
≥12.5 Years
(n=157)
P-
inter
n
Early Menarche
<12.5yrs (n=222)
≥12.5 Years
(n=216)
P-
inter
CYP17

0.86

0.39
A1/A1 63,62 9.90(7.50-13.0) 9.94(7.47-13.2)

69,67 9.50(7.20-12.4) 7.84(5.94-10.3)
A1/A2 94,76 11.0(8.70-13.9) 12.3(9.46-16.00)

98,95 7.35(5.80-9.31) 8.21(6.39-10.6)
A2/A2 17,19 12.0(7.32-19.5) 14.2(8.96-22.5)

55,54 10.6(6.59-17.1) 9.02(5.96-13.7)
CYP1B1

0.26

0.63
Val/Val 51,45 10.3(7.61-14.00) 13.9(10.1-19.3)

80,74 9.10(6.84-12.1) 9.94(7.47-13.2)
Val/Leu 94,86 11.1(8.67-14.3) 11.9(9.15-15.5)

107,1
03
8.91(6.90-11.5) 8.33(6.37-10.9)
Leu/Leu 29,26 12.4(8.46-18.2) 9.80 (6.57-14.6)

35,39 8.73(5.86-13.0) 7.15(4.91-10.4)
COMT

0.64

0.92
Val/Val 53,44 13.1(9.73-17.5) 13.8(9.83-19.3)

69,67 10.1(7.51-13.6) 9.14(6.68-12.5)
Val/Met 71,64 11.3(8.50-14.9) 11.1(8.28-14.9)

97,94 8.60(6.57-11.3) 8.53(6.55-11.1)
Met/Met 50,49 8.88(6.48-12.2) 11.2(8.16-15.3)

55,54 7.79(5.59-10.8) 7.78(5.60-10.8)

Parity

Nulliparous (n=39) Parous (n=292)

Nulliparous (n=53) Parous (n=385)

CYP17

0.70

0.22
A1/A1 13,112 15.0 (8.50-26.5) 9.41(7.52-11.8)

19,14
6
10.6(6.30-18.0) 8.44(6.77-10.5)

A1/A2 17,153 15.9(9.87-25.5) 11.1(9.04-13.6)

23,19
7
6.38(4.00-10.2) 7.93(6.46-9.72)

A2/A2 9,27 13.7(7.07-26.5) 12.7(8.61-18.8)

11,42 14.6(7.43-28.6) 8.71(6.06-12.5)

CYP1B1

0.11

0.44
Val/Val 13,112 11.2(5.79-21.6) 11.9(9.18-15.4)

15,13
9
8.20(4.60-14.6) 9.72(7.65-12.4)

Val/Leu 19,153 16.7(10.5-26.4) 11.0(8.83-13.7)

21,18
9
11.7(7.01-19.4) 8.41(6.74-10.5)

Leu/Leu 9,45 17.1(9.18-31.9) 10.1(7.32-13.8)

17,57 7.81(4.49-13.6) 7.97(5.77-11.0)

COMT

0.36

0.27
Val/Val 10,87 16.4(8.84-30.4) 12.8(9.94-16.6)

13,12
3
14.6(7.81-27.3) 9.23(7.21-11.8)

Val/Met 19,116 12.7(7.97-20.2) 10.8(8.49-13.8)

26,16
7
7.39(4.74-11.5) 8.83(7.02-11.1)

Met/Met 10,89 19.0(10.2-35.2) 9.18(7.11-11.9)

14,95 9.24(5.06-16.9) 7.65(5.85-10.0)

57

Supp Table 3, continued.

Body Mass
Index
(kg/m^2)

BMI≤25 (n=141) BMI>25 (n=190)

BMI≤25 (n=174) BMI>25 (n=264)

CYP17

0.81

0.92
A1/A1 44,81 8.70 (6.10-12.4) 10.9(8.31-14.2)

61,10
4
9.10(6.5-12.8) 8.38(6.46-10.9)

A1/A2 78,92 10.4(7.72-14.0) 12.7(9.76-16.4)

92,12
8
8.50(6.33-11.4) 7.25(5.64-9.32)

A2/A2 19,17 10.5(6.56-16.8) 16.2(9.76-26.8)

21,32 9.90(6.00-16.3) 9.56(6.26-14.6)

CYP1B1

0.38

0.67
Val/Val 36,60 9.46(6.52-13.7) 13.7(10.1-18.6)

55,99 11.1(7.82-15.7) 8.66(6.56-11.4)

Val/Leu 77,103 10.7(7.86-14.6) 11.9(9.18-15.5)

88,12
2
8.89(6.47-12.2) 8.42(6.50-10.9)

Leu/Leu 28,27 8.72(5.72-13.3) 13.9(9.20-20.9)

31,43 7.85(5.02-12.3) 7.83(5.40-11.3)

COMT

0.31

0.87
Val/Val 38,59 9.88(6.78-14.4) 16.3(12.0-22.2)

51,85 10.0 (6.94-14.5) 9.36(6.96-12.6)

Val/Met 62,73 10.0(7.26-13.9) 12.3(9.06-16.8)

76,11
7
9.57(6.94-13.2) 7.98(6.12-10.4)

Met/Met 41,58 9.33(6.49-13.4) 10.1(7.46-13.7)

47,62 8.12(5.6-11.8) 7.67(5.57-10.6)

*adjusted for age, race, BMI, postmenopausal hormone use (never, ever estrogen use, ever estrogen and progesti), age at menarche(years),
other polymorphisms(CYP17,CYP1B1,COMT). Early menarche is not adjusted by age at menarche.
58

Supplement Table 4. Unadjusted geometric mean levels of percent density and associations with biological and lifestyle factors among
postmenopausal women.

Pre-diagnosis (n=331)

Post-diagnosis (n=438)

n
Mean % density
(95% CI)
p-overall
1
n
Mean % density
(95% CI)
p-overall

Postmenopausal hormone therapy use

<0.01
2

0.06
Never 60 8.29(6.36-10.8)

138 9.18(7.49-11.3)

Past use of Estrogen or Progestin 60 13.9(10.7-18.2)

84 NA

Current (Use at Diagnosis)

Estrogen only 91 13.9(11.2-17.3)

95 NA

Combined Estrogen and Progestin 117 16.2(13.4-19.6)

117 NA

Ever Use

0.06
3

<0.01
3
Ever Estrogen only Use 112 13.1(10.8-15.9)

126 8.34(6.74-10.3)

Ever Estrogen and Progestin Use 159 16.3(13.9-19.2)

174 11.5(9.59-13.8)

Former oral contraceptive use
4

0.33

0.06
No 139 12.5(10.5-15.0)

155 8.63(7.11-10.5)

Yes 192 14.1(12.1-16.4)

282 10.4(9.02-12.0)

Tamoxifen use for treatment

0.17
No 175 NA

229 10.6(9.00-12.4)

Yes 156 NA

209 8.97(7.60-10.6)

1
P-value reflects overall model association between covariate and mammographic density.
2
P-value reflects overall association between Never, Past use of Estrogen or Progestin, and Current use of estrogen only, and Current use of combined estrogen
and progestin, and mammographic density.
3
P-value reflects overall association between Never, Ever estrogen only use, and Ever estrogen and progestin use with mammographic density.
4
Data of use oral contraceptives missing for one woman among post-diagnosis population (n=437).
59

Supplement Table 4, continued.

Stage of disease

0.13
In Situ 75 NA

84 10.5(8.08-13.7)

Localized 201 NA

248 9.92(8.52-11.6)

Regional 55 NA

106 8.88(7.03-11.2)

ER status

0.96
Positive 202 NA

261 9.83(8.47-11.4)

Negative 37 NA

69 9.95(7.45-13.3)

Unknown 92 NA

108 9.50(7.54-12.0)

Race

0.02

0.02
Non-Hispanic white – Western Washington 74 10.0(7.88-12.8)

80 8.47(6.48-11.1)

Non-Hispanic white – New Mexico 160 16.1(13.7-19.0)

187 11.5(9.64-13.7)

African-American – Los Angeles 46 11.7(8.66-15.9)

112 8.55(6.82-10.7)

Hispanic – New Mexico 40 13.3(9.57-18.4)

45 10.1(7.05-14.4)

Other 11 11.7(6.27-21.9)

14 6.59(3.48-12.5)

Not Overweight(BMI<25) 141 17.1(14.4-20.3) <0.01 174 14.7(12.3-17.5) <.0001
Overweight (BMI≥25) 190 11.2(9.64-13.0)

264 7.47(6.48-8.62)

Not Obese (BMI<30) 240 15.1(13.2-17.2) <0.01 338 11.4(10.0-12.9) <.0001
Obese (BMI≥30) 91 9.82(7.91-12.2)

100 5.83(4.61-7.37)

Age of Menarche (≥12.5 years)
1
174 12.6(10.7-14.7) 0.50 222 9.40(7.99-11.0) 0.50
Early Menarche (<12.5 years) 157 14.4(12.2-17.0)

216 10.2(8.63-12.0)

1
Using continuous age of menarche, beta coefficient is 0.06 (p=.05) among women with Pre-diagnosis mammograms; beta coefficient is 0.06 (p=.12) among
women with post-diagnosis mammograms.
60

Supplement Table 4, continued.

Parity

<0.01

0.20
Nulliparous 39 22.3(16.0-31.1)

53 12.0(8.59-16.6)

Parous 292 12.5(11.1-14.1)

285 9.50(8.40-10.7)

Number of Live Births

0.13

0.16
None 40 21.9(15.8-30.3)

54 11.90(8.63-16.5)

<3 137 13.6(11.4-16.3)

203 11.2(9.51-13.3)

≥3 154 11.6(9.84-13.7)

181 7.86(6.58-9.38)

Age at First Live Birth
1

0.66

0.11
Under 18 27 13.6(8.98-20.5)

42 6.79(4.70-9.82)

18-25 186 12.9(11.-15.1)

235 9.51(8.14-11.1)

>35 78 11.4(8.97-14.6)

107 10.8(8.57-13.6)

1
Data not available for 40 women among pre-diagnosis population, 39 of whom have never been pregnant. Data not available for 54 women
among post-diagnosis population, 54 have never been pregnant. The unavailable data is for the same woman.
61

Supplement Table 5a. Associations in odds-ratio between genotype(continuous) and biological or lifestyle factors (outcome) in Pre-
diagnosis population (n=313).

CYP17 CYP1B1 COMT MTHFR
Postmenopausal hormone therapy
1

Past use of Estrogen or Progestin 0.85(0.48-1.49) 0.93(0.54-1.59) 1.37(0.86-2.19) 0.92(0.52-1.63)
Current (Use at Diagnosis)

Estrogen only 0.92(0.55-1.53) 0.95(0.58-1.56) 1.41(0.92-2.16) 1.22(0.73-2.03)
Combined Estrogen and Progestin 1.28(0.79-2.00) 0.76(0.47-1.21) 1.29(0.86-1.95) 1.62(1.00-2.64)
Ever Use

Ever Estrogen only 0.92(0.55-1.53) 0.95(0.57-1.58) 1.41(0.92-2.18) 1.22(0.73-2.02)
Ever Estrogen and Progestin 1.28(0.79-2.08) 0.74(0.46-1.21) 1.30(0.86-1.96) 1.62(1.00-2.63)
Former oral contraceptive use
2
1.02(0.73-1.43) 1.05(0.76-1.46) 1.02(0.77-1.35) 1.14(0.82-1.59)

Race
3

Non-Hispanic white – New Mexico 0.77(0.51-1.18) 1.15(0.75-1.75) 1.12(0.78-1.61) 1.05(0.69-1.60)
African-American – Los Angeles 0.52(0.29-0.94) 0.37(0.2-0.68) 0.55(0.34-0.91) 0.19(0.09-0.41)
Hispanic – New Mexico 0.50(0.27-0.94) 1.54(0.85-2.78) 0.70(0.42-1.17) 1.32(0.74-2.35)
Other 1.14(0.44-3.01) 0.42(0.14-1.2) 0.67(0.29-1.55) 1.06(0.41-2.76)

Overweight (BMI≥25)
4
0.68(0.48-0.96) 0.77(0.55-1.07) 0.96(0.72-1.27) 0.87(0.62-1.20)
Obese (BMI≥30)
5
0.88(0.60-1.28) 0.88(0.61-1.27) 0.87(0.63-1.19) 1.00(0.70-1.44)

Early Menarche (<12.5 years) 0.98(0.70-1.37) 1.01(0.73-1.40) 1.09(0.82-1.44) 1.14(0.82-1.58)

1
Reference group is Never users of postmenopausal hormones.
2
Reference group is Non-users of oral contraceptives.
3
Reference group is Non-Hispanic white from Washington.
4
Reference group are Non-Overweight (BMI<25).
5
Reference group is Non-users of oral contraceptives.
62

Supplement Table 5a, continued.

Parity

Parous
1
0.64(0.38-1.07) 0.68(0.41-1.13) 1.01(0.66-1.56) 1.20(0.71-2.02)
Number of Live Births
2

<3 1.33(0.68-2.58) 1.55(0.82-2.95) 1.00(0.59-1.68) 1.49(0.79-2.85)
≥3 1.58(0.77-3.25) 1.90(0.95-3.79) 1.02(0.58-1.80) 1.43(0.71-2.85)
Age at First Live Birth
3

18-25 0.69(0.4-1.19) 0.73(0.43-1.23) 0.969(0.59-1.68) 1.18(0.69-2.04)
>25 0.59(0.34-1.00) 0.64(0.38-1.09) 1.02(0.58-1.80) 1.16(0.68-1.98)

1
Reference group is Nulliparous women.
2
Reference group is 0 Live Births. There is one woman who had a pregnancy but did not have a live birth.

3
Reference group is <18 years at First live birth.

63

Supplement Table 5b. Associations in odds-ratio between genotype(continuous) and biological or lifestyle factors (outcome) in Post-
diagnosis population (n=438).

CYP17 CYP1B1 COMT MTHFR
Postmenopausal hormone therapy
1

Ever Use
Ever Estrogen only 0.91(0.61-1.36) 1.14(0.78-1.68) 1.58(1.1-1.14) 1.56(1.03-2.35)
Ever Estrogen and Progestin 1.19(0.82-1.72) 0.88(0.61-1.27) 1.41(1.01-0.88) 1.99(1.35-2.92)
Former oral contraceptive use
2
1.02(0.75-1.37) 0.93(0.71-1.24) 0.92(0.71-0.93) 0.88(0.66-1.17)
Tamoxifen use for treatment
3

0.87(0.65-1.16) 1.10(0.84-1.44) 0.83(0.65-1.10) 1.14(0.86-1.51)
Stage of disease
4

Localized
1.25(0.86-1.83) 0.93(0.66-1.33) 1.16(0.83-0.93) 0.88(0.61-1.27)
Regional
0.78(0.5-1.22) 0.84(0.56-1.27) 0.82(0.56-0.84) 0.77(0.50-1.19)
ER status
5

Negative
0.84(0.56-1.27) 1.04(0.71-1.52) 0.96(0.67-1.04) 0.72(0.47-1.12)
Unknown
1.01(0.72-1.42) 1.50(1.08-2.06) 1.03(0.76-1.50) 1.41(1.02-1.95)
Race
6

Non-Hispanic white – New Mexico 0.95(0.64-1.42) 1.25(0.85-1.84) 1.04(0.73-1.25) 1.04(0.71-1.52)
African-American – Los Angeles 0.85(0.55-1.32) 0.42(0.27-0.67) 0.44(0.30-0.42) 0.21(0.12-0.36)
Hispanic – New Mexico 0.82(0.47-1.43) 2.29(1.32-3.97) 0.60(0.36-2.29) 1.07(0.63-1.83)
Other 1.41(0.61-3.28) 0.36(0.14-0.96) 0.58(0.26-0.36) 1.03(0.45-2.37)

Overweight (BMI≥25) 0.91(0.68-1.21) 0.86(0.65-1.13) 0.89(0.69-0.86) 0.92(0.69-1.23)
Obese (BMI≥30) 1.05(0.75-1.47) 0.96(0.69-1.32) 1.00(0.74-0.96) 1.12(0.80-1.55)

1
Reference group is Never users of postmenopausal hormones.
2
Reference group is Non-users of oral contraceptives.
3
Reference group is Non-users of tamoxifen.
4
Reference group is In-situ.
5
Reference group is Positive ER status.
6
Reference group is Non-Hispanic white from Washington.
64

Supplement Table 5b, continued.

Early Menarche (<12.5 years) 1.09(0.82-1.45) 1.09(0.83-1.42) 1.01(0.78-1.09) 1.07(0.81-1.42)
Parity
1

Parous 0.76(0.50-1.17) 0.60(0.40-0.91) 0.85(0.58-0.60) 1.28(0.81-2.01)
Number of Live Births
2

<3 1.25(0.74-2.10) 1.12(0.69-1.83) 1.16(0.75-1.12) 1.76(1.02-3.04)
≥3 1.32(0.75-2.32) 1.37(0.81-2.33) 1.11(0.69-1.37) 1.52(0.84-2.73)
Age at First Live Birth
3

18-25 0.82(0.52-1.28) 0.6(0.39-0.93) 0.80(0.53-0.6) 1.19(0.74-1.90)
>35 0.69(0.44-1.09) 0.6(0.39-0.93) 0.84(0.56-0.6) 1.32(0.82-2.12)

1
Reference group is Nulliparous women.
2
Reference group is women with 0 Live Births. There is one woman who had a pregnancy but did not have a live birth.

3
Reference group is <18 years at First live birth.

65

Supplement Table 6a. Association between mammographic percent density and genotypes in the pre-diagnosis population (n=311).

CYP17 CYP1B1 COMT MTHFR

p-value p-value p-value p-value
Pre-diagnosis (n=311)
Age at mammogram (years)

0.45 0.25 0.28 0.45
Body Mass Index

0.40 0.39 0.40 0.92
Age at menarche

0.53 0.99 0.93 0.65
Age at menopause

0.13 0.65 0.77 0.69
Length of progestin use (years)

0.04 0.27 0.65 0.45
Length of estrogen use (years)
2

0.79 0.66 0.34 0.02
Age at First Live Birth

0.78 0.93 0.17 0.63
Time from mammogram to diagnosis
(months)

0.09 0.52 0.93 0.68
Postmenopausal hormone therapy use
3

0.50 0.55 0.35 0.43
Ever Postmenopausal hormone therapy use

0.77 0.19 0.60 0.38
Former oral contraceptive use
4

0.86 0.54 0.62 0.69
Tamoxifen use for treatment

0.57 0.46 0.11 0.56
Stage of disease

0.55 0.94 0.49 0.85
ER status

0.78 0.07 0.84 0.62
Race

0.23 <.0001 0.03 <0.01
Overweight(>25)

0.08 0.28 0.57 0.66
Obesity (≥30)

0.74 0.76 0.32 0.42
Parous (yes)

0.03 0.25 0.56 0.49
Live Birth (0, 1-3, ≥3)

0.24 0.15 0.72 0.99
Early Menarche (≥12.5)

0.56 0.99 0.85 0.64
Data of length of progestin use missing for 4 women (Pre diagnosis n=327, Post-diagnosis=434).
2
Data of length of estrogen use missing for 12 women (Pre-diagnosis n=319, Post-diagnosis n=426).
3
Recency of post-menopausal hormone use missing for 2 women.
4
Data of use oral contraceptives missing for one women (n=437).
66

Supplement Table 6b. Association between mammographic percent density and genotypes in the post-diagnosis population (n=438).

CYP17 GCYP1B1 GCOMT GMTHFR

p-value p-value p-value p-value
Age at mammogram (years)

0.27 <0.01 0.12 0.04
Body Mass Index

0.92 0.18 0.33 0.67
Age at menarche

0.23 0.82 0.84 0.73
Age at menopause

0.55 0.01 0.26 0.01
Length of progestin use (years)
2

0.14 0.01 0.50 0.25
Length of estrogen use (years)
3

0.30 0.61 0.21 0.02
Age at First Live Birth

0.23 0.34 0.01 0.04
Time from mammogram to diagnosis (months)

0.60 0.58 0.98 0.29
Postmenopausal hormone therapy use
4

0.24 0.96 0.06 0.06
Ever Postmenopausal hormone therapy use

0.66 0.27 0.10 0.04
Former oral contraceptive use
5

0.98 0.02 0.63 0.66
Tamoxifen use for treatment

0.58 0.29 0.29 0.66
Stage of disease

0.11 0.94 0.09 0.13
ER status

0.56 0.14 0.84 0.05
Race

0.94 <.0001 <.0001 <.0001
Overweight(>25)

0.63 0.45 0.67 0.82
Obesity (≥30)

0.91 0.82 0.91 0.52
Parous (yes)

0.11 0.01 0.54 0.27
Live Birth (0, 1-3, ≥3)

0.65 0.16 0.59 0.98
Early Menarche (≥12.5)

0.51 0.80 1.00 0.85
Data missing for 12 women. Post-diagnosis data (n=336). Menopause includes both natural and menopause resulting from treatment or surgery.
2
Data of length of progestin use missing for 4 women (Pre diagnosis n=327, Post-diagnosis=434).
3
Data of length of estrogen use missing for 12 women (Pre-diagnosis n=319, Post-diagnosis n=426).
4
Recency of post-menopausal hormone use missing for 2 women.
5
Data of use oral contraceptives missing for one women (n=437).
67

68

1.6 References

1. Boyd NF, Guo H, Martin LJ, et al. Mammographic density and the risk and detection of
breast cancer. N Engl J Med 2007;356:227-36.
2. Huang CS, Chern HD, Chang KJ, et al. Breast cancer risk associated with genotype
polymorphism of the estrogen-metabolizing genes CYP17, CYP1A1, and COMT: a
multigenic study on cancer susceptibility. Cancer Res 1999;59:4870-5.
3. Kelemen LE, Sellers TA, Vachon CM. Can genes for mammographic density inform
cancer aetiology? Nat Rev Cancer 2008;8:812-23.
4. Butler LM, Gold EB, Greendale GA, et al. Menstrual and reproductive factors in relation
to mammographic density: the Study of Women's Health Across the Nation (SWAN).
Breast Cancer Res Treat 2008;112:165-74.
5. Dumas I, Diorio C. Polymorphisms in genes involved in the estrogen pathway and
mammographic density. BMC Cancer 2010;10:636.
6. McTiernan A, Martin CF, Peck JD, et al. Estrogen-plus-progestin use and
mammographic density in postmenopausal women: Women's Health Initiative
randomized trial. J Natl Cancer Inst 2005;97:1366-76.
7. Anderson GL, Chlebowski RT, Rossouw JE, et al. Prior hormone therapy and breast
cancer risk in the Women's Health Initiative randomized trial of estrogen plus progestin.
Maturitas 2006;55:103-15.
8. Chlebowski RT, Anderson GL. Changing concepts: Menopausal hormone therapy and
breast cancer. J Natl Cancer Inst 2012;104:517-27.
9. Greendale GA, Palla SL, Ursin G, et al. The association of endogenous sex steroids and
sex steroid binding proteins with mammographic density: results from the
Postmenopausal Estrogen/Progestin Interventions Mammographic Density Study. Am J
Epidemiol 2005;162:826-34.
10. Vachon CM, Sellers TA, Vierkant RA, et al. Case-control study of increased
mammographic breast density response to hormone replacement therapy. Cancer
Epidemiol Biomarkers Prev 2002;11:1382-8.
11. 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.
12. Greendale GA, Reboussin BA, Slone S, et al. Postmenopausal hormone therapy and
change in mammographic density. J Natl Cancer Inst 2003;95:30-7.
13. Boyd NF. Tamoxifen, mammographic density, and breast cancer prevention. J Natl
Cancer Inst 2011;103:704-5.
14. Cuzick J, Warwick J, Pinney E, et al. Tamoxifen-induced reduction in mammographic
density and breast cancer risk reduction: a nested case-control study. J Natl Cancer Inst
2011;103:744-52.
15. Boyd N, Martin L, Stone J, et al. A longitudinal study of the effects of menopause on
mammographic features. Cancer Epidemiol Biomarkers Prev 2002;11:1048-53.
16. Boyd NF, Stone J, Martin LJ, et al. The association of breast mitogens with
mammographic densities. Br J Cancer 2002;87:876-82.
17. Aiello EJ, Tworoger SS, Yasui Y, et al. Associations among circulating sex hormones,
insulin-like growth factor, lipids, and mammographic density in postmenopausal women.
Cancer Epidemiol Biomarkers Prev 2005;14:1411-7.
69

18. Tamimi RM, Hankinson SE, Colditz GA, et al. Endogenous sex hormone levels and
mammographic density among postmenopausal women. Cancer Epidemiol Biomarkers
Prev 2005;14:2641-7.
19. Haiman CA, Hankinson SE, De Vivo I, et al. Polymorphisms in steroid hormone pathway
genes and mammographic density. Breast Cancer Res Treat 2003;77:27-36.
20. Crandall CJ, Sehl ME, Crawford SL, et al. Sex steroid metabolism polymorphisms and
mammographic density in pre- and early perimenopausal women. Breast Cancer Res
2009;11:R51.
21. Lord SJ, Mack WJ, Van Den Berg D, et al. Polymorphisms in genes involved in estrogen
and progesterone metabolism and mammographic density changes in women randomized
to postmenopausal hormone therapy: results from a pilot study. Breast Cancer Res
2005;7:R336-44.
22. Maskarinec G, Lurie G, Williams AE, et al. An investigation of mammographic density
and gene variants in healthy women. Int J Cancer 2004;112:683-8.
23. Bugano DD, Conforti-Froes N, Yamaguchi NH, et al. Genetic polymorphisms, the
metabolism of estrogens and breast cancer: a review. Eur J Gynaecol Oncol 2008;29:313-
20.
24. Le Marchand L, Haiman CA, Wilkens LR, et al. MTHFR polymorphisms, diet, HRT, and
breast cancer risk: the multiethnic cohort study. Cancer Epidemiol Biomarkers Prev
2004;13:2071-7.
25. Hanukoglu I. Steroidogenic enzymes: structure, function, and role in regulation of steroid
hormone biosynthesis. J Steroid Biochem Mol Biol 1992;43:779-804.
26. Warren R, Skinner J, Sala E, et al. Associations among mammographic density,
circulating sex hormones, and polymorphisms in sex hormone metabolism genes in
postmenopausal women. Cancer Epidemiol Biomarkers Prev 2006;15:1502-8.
27. Hanna IH, Dawling S, Roodi N, et al. Cytochrome P450 1B1 (CYP1B1)
pharmacogenetics: association of polymorphisms with functional differences in estrogen
hydroxylation activity. Cancer Res 2000;60:3440-4.
28. Li DN, Seidel A, Pritchard MP, et al. Polymorphisms in P450 CYP1B1 affect the
conversion of estradiol to the potentially carcinogenic metabolite 4-hydroxyestradiol.
Pharmacogenetics 2000;10:343-53.
29. Goodman JE, Lavigne JA, Wu K, et al. COMT genotype, micronutrients in the folate
metabolic pathway and breast cancer risk. Carcinogenesis 2001;22:1661-5.
30. Dawling S, Roodi N, Mernaugh RL, et al. Catechol-O-methyltransferase (COMT)-
mediated metabolism of catechol estrogens: comparison of wild-type and variant COMT
isoforms. Cancer Res 2001;61:6716-22.
31. Campbell IG, Baxter SW, Eccles DM, et al. Methylenetetrahydrofolate reductase
polymorphism and susceptibility to breast cancer. Breast Cancer Res 2002;4:R14.
32. Habel LA, Dignam JJ, Land SR, et al. Mammographic density and breast cancer after
ductal carcinoma in situ. J Natl Cancer Inst 2004;96:1467-72.
33. Cil T, Fishell E, Hanna W, et al. Mammographic density and the risk of breast cancer
recurrence after breast-conserving surgery. Cancer 2009;115:5780-7.
34. McTiernan A, Rajan KB, Tworoger SS, et al. Adiposity and sex hormones in
postmenopausal breast cancer survivors. J Clin Oncol 2003;21:1961-6.
70

35. Irwin ML, Crumley D, McTiernan A, et al. Physical activity levels before and after a
diagnosis of breast carcinoma: the Health, Eating, Activity, and Lifestyle (HEAL) study.
Cancer 2003;97:1746-57.
36. Irwin ML, McTiernan A, Bernstein L, et al. Physical activity levels among breast cancer
survivors. Med Sci Sports Exerc 2004;36:1484-91.
37. Abrahamson PE, Tworoger SS, Aiello EJ, et al. Associations between the CYP17,
CYPIB1, COMT and SHBG polymorphisms and serum sex hormones in post-
menopausal breast cancer survivors. Breast Cancer Res Treat 2007;105:45-54.
38. Edwards AM, Hunter, S.V., Hankin, R.C. Gene Rearrangement Analysis by Nonorganic
Extraction and Chemiluminescent Detection. Laboratory Medicine 1993;24.
39. Livak KJ. Allelic discrimination using fluorogenic probes and the 5' nuclease assay.
Genet Anal 1999;14:143-9.
40. Haiman CA, Hankinson SE, Spiegelman D, et al. The relationship between a
polymorphism in CYP17 with plasma hormone levels and breast cancer. Cancer Res
1999;59:1015-20.
41. Feigelson HS, Coetzee GA, Kolonel LN, et al. A polymorphism in the CYP17 gene
increases the risk of breast cancer. Cancer Res 1997;57:1063-5.
42. Bailey LR, Roodi N, Dupont WD, et al. Association of cytochrome P450 1B1 (CYP1B1)
polymorphism with steroid receptor status in breast cancer. Cancer Res 1998;58:5038-41.
43. Lavigne JA, Helzlsouer KJ, Huang HY, et al. An association between the allele coding
for a low activity variant of catechol-O-methyltransferase and the risk for breast cancer.
Cancer Res 1997;57:5493-7.
44. Schneider J, Huh MM, Bradlow HL, et al. Antiestrogen action of 2-hydroxyestrone on
MCF-7 human breast cancer cells. J Biol Chem 1984;259:4840-5.
45. Bradlow HL, Telang NT, Sepkovic DW, et al. 2-hydroxyestrone: the 'good' estrogen. J
Endocrinol 1996;150 Suppl:S259-65.
46. Gupta M, McDougal A, Safe S. Estrogenic and antiestrogenic activities of 16alpha- and
2-hydroxy metabolites of 17beta-estradiol in MCF-7 and T47D human breast cancer
cells. J Steroid Biochem Mol Biol 1998;67:413-9.
47. Stevens KN, Lindstrom S, Scott CG, et al. Identification of a novel percent
mammographic density locus at 12q24. Hum Mol Genet 2012.
48. Maskarinec G, Pagano I, Lurie G, et al. Mammographic density and breast cancer risk:
the multiethnic cohort study. Am J Epidemiol 2005;162:743-52.
49. Takata Y, Maskarinec G, Le Marchand L. Breast density and polymorphisms in genes
coding for CYP1A2 and COMT: the Multiethnic Cohort. BMC Cancer 2007;7:30.
50. Haiman CA, Bernstein L, Berg D, et al. Genetic determinants of mammographic density.
Breast Cancer Res 2002;4:R5.
51. Chen J, Gammon MD, Chan W, et al. One-carbon metabolism, MTHFR polymorphisms,
and risk of breast cancer. Cancer Res 2005;65:1606-14.
52. Shrubsole MJ, Gao YT, Cai Q, et al. MTHFR polymorphisms, dietary folate intake, and
breast cancer risk: results from the Shanghai Breast Cancer Study. Cancer Epidemiol
Biomarkers Prev 2004;13:190-6.
53. Gao CM, Kazuo T, Tang JH, et al. [MTHFR polymorphisms, dietary folate intake and
risks to breast cancer]. Zhonghua Yu Fang Yi Xue Za Zhi 2009;43:576-80.
54. Alshatwi AA. Breast cancer risk, dietary intake, and methylenetetrahydrofolate reductase
(MTHFR)single nucleotide polymorphisms. Food Chem Toxicol 2010;48:1881-5.
71

55. Zhang J, Qiu LX, Wang ZH, et al. MTHFR C677T polymorphism associated with breast
cancer susceptibility: a meta-analysis involving 15,260 cases and 20,411 controls. Breast
Cancer Res Treat 2010;123:549-55.
56. Lin WY, Chou YC, Wu MH, et al. The MTHFR C677T polymorphism, estrogen
exposure and breast cancer risk: a nested case-control study in Taiwan. Anticancer Res
2004;24:3863-8.
57. Adjei AA, Weinshilboum RM. Catecholestrogen sulfation: possible role in
carcinogenesis. Biochem Biophys Res Commun 2002;292:402-8.
58. Apter D, Reinila M, Vihko R. Some endocrine characteristics of early menarche, a risk
factor for breast cancer, are preserved into adulthood. Int J Cancer 1989;44:783-7.
59. Feigelson HS, Shames LS, Pike MC, et al. Cytochrome P450c17alpha gene (CYP17)
polymorphism is associated with serum estrogen and progesterone concentrations. Cancer
Res 1998;58:585-7.
60. Cil T, Fishell E, Hanna W, et al. Mammographic density and the risk of breast cancer
recurrence after breast-conserving surgery. Cancer 2009;115:5780-7.

72

Chapter 3
Association between acute-phase proteins (C-reactive protein and Serum Amyloid A) and
post-diagnosis mammographic density in breast cancer survivors
3.1 Abstract

Authors: Anne Dee
1
, Leslie Bernstein,
1,5
Anne McTiernan
2
, Richard N. Baumgartner
3
, Kathy
Baumgartner
3
, Rachel Ballard-Barbash
4
, Roberta McKean-Cowdin
1

1
University of Southern California, Los Angeles, CA,
2
Fred Hutchinson Cancer Research Center,
Seattle, WA,
3
University of Louisville, Louisville, KY,
4
National Cancer Institute, Bethesda,
MD,
5
City of Hope, Duarte, CA.
Introduction: Mammographic density (MD) is positively associated with breast cancer risk and
with breast cancer recurrence. Factors that influence MD include age, circulating endogenous
hormone levels, hormone therapy, menopausal status, parity, adiposity, and genetic variation.
The association between MD and estrogen levels may be regulated to an extent by the immune
system. Cytokines such as IL-6 influence aromatase activity and therefore estrogen synthesis.
Experiments on breast cancer cell lines have shown that IL-6, in combination with estrone
sulfate, enhances cellular proliferation through their action on aromatase. However, other cell
line studies have shown that IL-6 may also inhibit breast cell proliferations. Acute phase proteins
C-Reactive Protein (CRP) and Serum Amyloid A (SAA) are nonspecific inflammatory markers
that increase with systemic inflammation in response to elevated levels of IL-6. The purpose of
this study is to evaluate whether CRP or SAA are associated with MD among postmenopausal
breast cancer survivors.
73

Methods: Circulating levels of CRP, SAA, and percent MD approximately 30 months after
diagnosis were obtained from 479 women participating in the Health, Eating, Activity, and
Lifestyle Study (HEAL). HEAL is a prospective cohort study of 1,183 breast cancer survivors
identified through Surveillance, Epidemiology, and End Results registries in Los Angeles
County, New Mexico, and Western Washington. For this analysis, regression models were used
to estimate associations between CRP/SAA and MD after adjustments for age, race, body mass
index (BMI), post-menopausal hormone use, tamoxifen use, and study center. Potential effect
modification by factors that could influence inflammation or circulating estrogen was evaluated
by stratification and with formal tests of interaction.
Results: We found an inverse association between CRP and MD (β=-0.15, p=0.01) after
adjusting for covariates; no statistically significant association was found between SAA and MD
(β=-0.10, p=0.24). These findings were consistent by age, race, BMI, use of non-steroidal anti-
inflammatory drug, smoking status, and physical activity. Women with CRP levels that were
lower than the median (<2.4 CRP mg/L) had higher geometric mean percent MD than women at
or above the median; this association was consistent across 5-year age groups. The largest
association between CRP and MD was found among women <50 years of age (β=-0.50, p=.01)
although the interaction with age was not statistically significant (p=.95).
Conclusion: The inverse association found between CRP and percent MD may be explained by
the suppression of cell growth in normal breast tissue due to cytokines. These findings should be
confirmed in an independent disease free population that includes direct measures of cytokine
activity such as IL-6. Additional exploration of biological mechanism is needed.

74

3.2 Introduction
Mammographic density is one of the strongest predictors of breast cancer risk and may
predict breast cancer recurrence (20, 28, 31, 33, 152). Percent mammographic density represents
the proportion of a mammogram occupied by radiologically dense tissue, which includes both
dense and adipose tissue. The dense areas of the breast result from increased numbers of
epithelial and stromal cells (153), which change over the course of a woman’s life. Factors that
influence mammographic density include age, menopausal status, parity, adiposity, tamoxifen
therapy, genetic variation, and other factors that influence circulating endogenous hormones (28,
47, 120). Specifically, higher mammographic density has been found among women using
exogenous hormones such as estrogen and progestin combined therapy (41, 42, 43), and lower
mammographic density has been measured among breast cancer survivors using the anti-
estrogenic drug, tamoxifen (46, 47). However, evidence for an association between circulating
levels of sex hormones and increased breast density is more conflicting: studies have indicated
positive (42, 99), inverse (50), or no associations (48, 49).
The association between mammographic density and estrogen levels may be regulated to
an extent, by the immune system (154). Among postmenopausal women, the primary production
of endogenous estrogen takes place peripherally in the adipose tissue (130). Elevated levels of
adipose tissue may contribute to increased production of biologically active estrone, as well as to
higher production of inflammatory proteins, including interleukin-6 (IL-6) and associated acute
phase proteins, C-Reactive Protein (CRP) and Serum Amyloid A (SAA) (130, 155). CRP and
SAA are nonspecific inflammatory markers that increase with systemic inflammation in response
to elevated levels of cytokines such as IL-6 (75), and may increase 10 to 100 fold in response to
the rise of IL-6 (156). This is of particular importance in breast cancer patients, because elevated
75

levels of CRP and SAA measured pre- and post-treatment are associated with reduced breast
cancer survival (79, 80). Cytokines, such as IL-6 and TNF-α, also stimulate aromatase activity,
and catalyzes the conversion of androstenedione to estrone in the estrogen synthesis pathway
(81) . Experiments on breast cancer cell lines have shown that IL-6, in combination with estrone
sulfate (E-S), enhance cellular proliferation through their action on aromatase (82). Further, the
association between IL-6 and aromatase has been found to enhance cellular proliferation in
normal breast tissue (157). A study using the multidrug-resistant breast cancer cell line
MCF7/ADR found high levels of IL-6 expression in these cells; the same investigators found that
pretreatment of the cells with exogenous IL-6 decreased the ability of doxorubicin to induce
apoptosis (i.e. increased resistance to drug and therefore its ability to induce apoptosis) (158).
However, there have also been evidence that cytokines, in some circumstances, may exert
an inhibitory effect on breast cell proliferation. In ER+ carcinoma cell lines, IL-6 has been
shown to inhibit cellular proliferation through the induction of apoptosis; the same study did not
find an antiproliferative effect of IL-6 in ER- breast cancer cells, whereas in normal breast
epithelial cells, investigators observed IL-6 was associated with inhibited growth but not
apoptosis (159). A separate study found an inhibitory effect of IL-6 on ER+ breast cancer cell
lines regardless of the presence of estradiol (160, 161). It has been suggested that differential
intracellular signaling between mesenchyme and breast cancer epithelium in normal versus
cancer cells may explain the varied effect of IL-6 (86). In cancer cell lines, experimental studies
have shown that proinflammatory cytokines, including IL-6, block cell growth signaling
following activation of the insulin-like growth factor-I (IGF-I) tyrosine kinase receptor (87).
The potential association between inflammatory markers and breast density has not been
well described epidemiologically. One case-control study has found that age-adjusted mean CRP
76

levels are higher among women with benign breast disease than women without benign breast
disease ( i.e. normal screening mammograms). Among the women with benign breast disease,
there was a statistically significant inverse association between immune markers (CRP, IL-6, and
TNF-a) and mammographic density that was diminished after additional adjustment for BMI
(162). Another cross-sectional analysis among postmenopausal women evaluated the association
between inflammatory markers (CRP, IL-6, and TNF-a) found a similar result where CRP
inverse association with percent dense area became null after adjustment for BMI and waist
circumference; however, this association persists after adjustment between CRP and percent
dense volume. Furthermore, the association between IL-6 and percent density is significant even
after adjustments (163). To further explore the potential association between inflammatory
markers and mammographic density, we examined the association between circulating levels of
CRP (mg/L) and SAA (mg/L) with mammographic density among post-menopausal breast
cancer survivors in the unaffected breast. Additionally, we explored other biological and
lifestyle variables that may influence the association between acute phase proteins and
mammographic density.
3.3 Methods
Study Population
This cross-sectional analysis used data from the Health, Eating, Activity and
Lifestyle (HEAL) Study, which was a population-based prospective cohort of breast
cancer survivors. The study included women, 18 years of age or older, who were
diagnosed with in-situ to stage IIIa breast cancer from 1996 through 1999. Cases were
identified through the Surveillance, Epidemiology, and End Results

(SEER) registries in
77

New Mexico, Los Angeles County, California and western Washington. The HEAL study
is observational in design and conducted to evaluate the independent roles of sex-
hormones, diet, weight, physical activity, genetics, and other factors on post-diagnostic
breast cancer prognosis and survival . Details of study design and recruitment procedures
have been described previously (17-19).
Briefly, the HEAL study included 1,183 women at baseline, of whom 615 women
were recruited from New Mexico, 202 from Washington and 366 from Los Angeles.
Baseline data were collected within the first year after diagnosis, on average 6 months
post diagnosis and follow-up data were collected approximately 24 months after baseline
(about 31 months after diagnosis). Of the 1,183 participants, 807 women had both CRP
and SAA measurements at the 24-month follow-up interview (CRP and SAA were not
measured at baseline) and 657 women had post-diagnostic mammographic density. For
the current analysis, we restricted to data from 479 women who had post-diagnostic
mammographic density, CRP, and SAA measurements.
Clinical Variables
Anthropometric measures were taken by trained staff at the baseline and follow-up clinic
visits, including measured height and weight. BMI was computed as weight in kilograms (kg)
divided by height in meters squared (m
2
).
Circulating concentrations of CRP and SAA were measured at the University of
Washington by latex-enhanced nephelometry using highly sensitive assays on the
Behring Nephelometer II analyzer (Dade Behring Diagnostics, Deerfield, IL). Fasting
blood samples were collected at the 24 month follow-up assessment and used to complete
78

the tests. Each sample was processed within 3 hours of collection and stored at -70° to -
80°C until analysis. Interassay coefficients of variation were 5% to 9% for CRP and 4%
to 8% for SAA. The lowest detectable value for CRP is 0.2 mg/L, the lowest for SAA is
0.7mg/L. The control materials that were included with assay batches for quality control
purposes came from Bio-Rad Laboratories (Hercules, CA).
Mammogram density assessment
Post-diagnosis mammograms (approximately 2 years post-diagnosis) were collected for
each woman using the information on clinical providers and mammogram dates reported by each
woman; mammograms were digitized using an Epson 1680 scanner (Epson America Inc., CA).
Data from mammograms included continuous percent mammographic density, dense area (in
1,000 pixels), and total breast area (in 1,000 pixels) on the craniocaudal view of the breast
contralateral to the one diagnosed with breast cancer. All of the density readings were conducted
by one technician using Cumulus 108, a computer program developed by the University of
Toronto (33) and included a 5% random sample of repeat films for quality assurance.
Questionnaire Variables
In-person interviews (New Mexico) and self-administered questionnaire forms
(Washington and California) provided information on demographics, dietary intake,
menopausal status, smoking status, disease history (arthritis, chronic lung disease,
diabetes, heart attack, heart failure, hypertension, other cancers) and current use of any
over the counter or prescription non-steroidal anti-inflammatory drugs (NSAIDs) at the
24-month follow-up survey (approximately 30 months after diagnosis on average).
Cancer treatment history, including history of radiation, chemotherapy, and tamoxifen
79

use, were obtained through medical record review, participants’ SEER records, or
responses to the questionnaire.
Statistical analyses
We examined the association between CRP or SAA and percent mammographic density
using linear regression models. In order to fulfill regression assumptions of linearity and
homoscedasticity, CRP and SAA values were logarithmically transformed and square root
transformation was applied to percent mammographic density. Regression coefficients (β) and
95% confidence intervals (CIs) were estimated from Linear regression models to assess the
associations between continuous values of CRP or SAA and percent mammographic density. P-
values <0.05 were considered statistically significant.
We analyzed the association between CRP/SAA and percent mammographic density after
adjusting for covariates among post-menopausal women diagnosed with breast cancer.
Covariates were determined by evaluating for significant association between exposure
(CRP/SAA) and potential confounders and evaluating association between mammographic
density and potential confounders (Table 1b). The final covariates in the model (from the 24-
month interview) include the following: age (continuous), race (New Mexico whites, Eastern
Washington whites, Hispanic, African-American, Other), BMI (normal or underweight <25
kg/m
2
, overweight 25-29.9 kg/m
2
, obese =<30 kg/m
2
), current tamoxifen use (24 month
questionnaire), PMH use (never, estrogen only, estrogen and progestin). The “other”
race/ethnicity category including those identified as Asian, American-Indian, Pacific Islander,
and small numbers of other descriptors. We also explored age (categorical), physical activity
(MET hrs/week), history of heart disease, and BMI (categorical) as potential confounders.
80

Predicted values were estimated by fitting log
e
(CRP) or log
e
(SAA) to the final Linear regression
model.
Formal interaction tests and stratum-specific modeling were performed to evaluate
potential effect modification by ever use of post-menopausal hormones at time of diagnosis
(never, estrogen only, estrogen and progestin), estrogen receptor status(positive, negative, or
unknown), BMI (normal or underweight <25 kg/m
2
, overweight 25-29.9 kg/m
2
, obese =<30
kg/m
2
), current NSAID use (yes/no), smoking history (never, former, current), physical activity
level (sedentary, moderate, vigorous), and age(<49.9, 50-54.9, 55-59.9,60-64.9,65-69.9,>=70
years).
Pearson correlations coefficients were evaluated between CRP, SAA and measures of
mammographic density.
3.4 Results
Among the 479 post-menopausal breast cancer survivors included in the analysis,
the median percent mammographic density of participants’ contralateral breast
approximately 2 years after breast cancer diagnosis was 13.0% (data not shown), with a
mean percent density of 15.5% (Table 1). The mean age of these women at the 24-
month follow-up interview was 61 years and the mean BMI was approximately 28 kg/m
2
.
We found that BMI was significantly lower (p<0.0001) and physical activity was
significantly higher (p<0.01) among women at or above the median level of MD (Table
1). Women with higher ( ≥13.0%) mammographic density also were less likely to have a
number of co-morbidities including: heart failure (p=.006), high blood pressure (p=.002),
diabetes (p<.001) and chronic lung disease (p=0.011). Women above the median percent
81

density were more likely to have a college education (p=.033) (Table 1). No significant
differences were seen by race/ethnicity, previous smoking history, use of tamoxifen,
NSAIDS or statins.
When we regressed square root transformed percent mammographic density on
log transformed CRP, we found a statistically significant inverse association between the
variables in a minimally adjusted model (i.e. age and race) (p<0.001), and after additional
adjustment for BMI and postmenopausal hormone replacement therapy (full model
p=0.01) (Table 2). The same models were fit for log
e
(SAA) (Table 2), revealing an
inverse association between log
e
(SAA) and percent mammographic density; however, the
association was not statistically significant after adjusting for all covariates (p=.24)
(Table 2).
The inverse association described between log
e
(CRP) and percent mammographic
density is also present in the association between log
e
(CRP) and mammographic dense
area, though the association in the models did not reach statistical significance (age and
race adjusted model p<.001; full model p=.06) (Data not shown).
We found no evidence of effect modification of log
e
(CRP) by age, race/ethnicity,
or select factors influencing hormone levels, adiposity, or inflammation (Table 3a); we
also did not find evidence of effect modification of log
e
(SAA) by any of these same
factors (Table 3b). Formal tests of statistical interaction resulted in p-values >0.05.
When we stratified the regression models by age, race/ethnicity, and factors influencing
hormone levels, adiposity, or inflammation (Table 3a), we found a stronger inverse
association between log
e
(CRP) and percent mammographic density among women: who
82

had a history of postmenopausal hormone therapy use (combined estrogen and progestin:
β = -0.23; p=0.02), who reported recently using NSAIDs (β = -0.19; p=0.06), or anti-
inflammatories (β = -0.24; p<0.01), who were overweight (BMI 25-29.9 kg/m
2
; β = -
0.32; p=0.01), who were more physically active (>26 MET hrs/week; β = -0.25;
p=0.004), who were among the lowest tertile of dietary omega 6:3 ratio (<8.05; β = -0.26;
p=0.02). Associations did not differ between estrogen receptor (ER) status (β = -0.20 for
both; p=.08 and p=.16 respectively); however, 122 women had unknown ER status.
The inverse association between percent MD and CRP was observed in each
racial/ethnic group and across all 6 age groups (5 year strata), with the strongest
association observed in the youngest age group at diagnosis (age<49.9: β = -0.35;
p=0.05).
The same analysis for log
e
(SAA) with percent mammographic density revealed
no evidence of effect modification by the stratifying variables and no significant
associations between log
e
(SAA) and percent mammographic density within any strata,
though the pattern of magnitude is similar to that of log
e
(CRP).
In Figure 2a, we show the scatter plot for log
e
(CRP) and square root percent
mammographic density for observed values (open boxes) and predicted values after
adjustment for age, BMI, post-menopausal therapy use, tamoxifen use, and study site
(open circles). The best fit predicted regression line explains approximately 20% of the
variation in the data (adjusted R
2
=.20). The Pearson correlation between predicted and
observed value is 0.46 (.33-.53). For both the unadjusted and adjusted values, there is a
strong and significant correlation between log
e
(CRP) and square root percent density
83

(Correlation= -0.27 p<0.0001, unadjusted; Correlation=- 0.53, p<0.0001, full model).
Further, there is a significant linear trend between quartiles of log
e
(CRP) and percent
mammographic density (p=0.01) after adjusting for covariates.
Figure 1b shows the scatter plot and predicted regression line for log
e
(SAA) and
square root transformed percent mammographic density for data points that are adjusted
(open boxes) and predicted values after adjustment for age, race, BMI, and
postmenopausal hormone use (open circles). The Pearson correlation between predicted
and observed value is 0.46 (.38-.53). The correlation is modest (-0.038) and not
statistically significant (p=0.40). The linear trend between quartiles of log
e
(SAA) and
percent mammographic density is not significant (p=0.48).
3.5 Discussion
In a cross-sectional analysis of data collected approximately 2 years after breast cancer
diagnosis, we found an inverse association between CRP and post-diagnostic mammographic
density in the contralateral breast; this finding held after controlling for body mass index or when
restricting our analysis to women of a similar age, BMI, or racial/ethnic group. This association
was more apparent among women who had taken combined estrogen and progestin therapy prior
to diagnosis, or who were overweight or obese (BMI >25). Recent studies have found a
protective association between NSAID use, breast cancer risk(164), recurrence(165, 166), and
survival(166, 167); the potential relevance of inflammatory proteins, including IL-6 and acute
phaseproteins on breast density, cellular proliferation, and the activity of aromatase is important
toimprove out understanding of biological mechanisms and outcomes for breast cancer patients.
84

To date, two epidemiological studies have directly examined the association between
CRP and mammographic density; no study has looked at the potential association with SAA. In
this case-control study of postmenopausal women, cases were defined as women with benign
breast disease and controls as those without disease (cases n=145, controls n=397). Investigators
examined the association between IL-6, TNF-α, CRP and mammographic density, and found a
statistically significant inverse associations between percent mammographic density with all
three markers before and after adjustment by age; however, the inverse associations became non-
significant after adjustment for BMI (162). The investigators hypothesized that BMI may
confound any effect of inflammation on mammographic density since cytokines, TNF-alpha and
IL-6 are produced in adipocytes and macrophages accumulate in adipose tissue. Another cross-
sectional analysis of 302 post-menopausal women, 25% of whom had benign breast disease, with
BMI 22–40 kg/m
2
and not taking post-menopausal hormones within the last 12 months found a similar
result where statistically inverse associations between CRP and mammographic density becomes
non-significant after adjustment for BMI; however, the inverse association between IL-6 and
mammographic density remains after adjustment for BMI. (163). .; However, it is not clear that
BMI is a confounder of the cytokine-mammographic density association and therefore
controlling for BMI may be an over correction. The association between BMI and cytokines
could either be described as: (1) correlated variables through their association with adipocytes
(BMI may influence MD through its impact on peripheral estrone production while cytokines,
may exhibit a direct impact on breast cell proliferation ; or (2) both factors influencing
production of estrogen, in which case cytokines could be considered an intermediate between
adipose tissue and estrone synthesis, as cytokines are produced in adipose (and the liver) and
influence aromatase activity. In the same paper, theauthors note that the mean age-adjusted CRP
85

levels are statistically significantly higher among women with benign breast disease (cases) than
women without the benign breast disease (controls), which they speculate may reflect a true
effect of inflammation on breast disease.
Studies evaluating the associations between inflammatory markers (Il-6 or CRP) and
breast cancer risk have not been consistent. Three studies reported no association between
inflammatory markers and breast cancer risk (168, 169, 170), one study found higher mean
levels of IL-6 among 20 breast cancer patients before treatment compared to 20 controls without
breast cancer(p<.001); however the analysis did not include an adjustment for age or body mass
index, although both cases and controls had measures of BMI above 25 kg/m
2
(171). In the
Rotterdam study (n=7,017), high levels (> 3 mg/L) of CRP were associated with an increased
risk of incident breast cancer in premenopausal women (hazard ratio, 1.59; 95%CI 1.05 to 2.41)
after adjustment for age, smoking, body mass index, age at menarche, age at menopause,
hormone use, and number of children (172).
In postmenopausal woman, peripheral estrogen production is the primary source of
circulating estrone levels as estradiol levels drop with the cessation of ovarian function at the
time of menopause. Adipocytes are production sites for cytokines (e.g. IL-6) and acute-phase
proteins (173), as well as for the peripheral production of estrone in postmenopausal women. IL-
6 has been shown to stimulate aromatase activity in adipose stromal cells specifically derived
from subcutaneous adipose tissues (174) . Although mammographic density may be impacted
by increasing blood levels of IL-6 found with increasing age (which could result in activities that
increase or decrease cellular proliferation or apoptosis), the significant inverse association we
found between CRP and percent mammographic density may be partly explained by the
substantial drop in estradiol production at the time of menopause (74, 175, 176, 177). The
86

impact of decreasing estradiol levels on mammographic density may be greater than any increase
in mammographic density from the gain in cytokine production that occurs with increasing
age(74). At the time of menopause, the reduction in estrogen may overwhelm any proliferative
effect in the breast tissue due to inflammatory proteins. However, we observed the inverse
association in both premenopausal and postmenopausal women and after adjusting for
circulating estrone and estradiol levels. It is also possible that the anti-proliferative feedback
mechanisms of cytokines and acute phase proteins contribute to the drop in mammographic
density.
One hypothesis is that a positive association may exist between inflammation and
mammographic density due to the impact of cytokines on cellular proliferation, either directly or
by impacting estrogen production. Alternatively, cytokines including IL-6 and by association,
the acute phase proteins CRP and SAA, may have a direct anti-inflammatory effect in breast
tissue. While CRP and SAA are both markers of inflammation and reflect the same pattern of
increase for most diseases, SAA is considered more sensitive for earliest detection of episodes of
acute inflammation than CRP(76). The activation of CRP and SAA are mediated by IL-6 acts
through a signal transducers and activators of transcription (STAT) pathway, STAT3(178). The
inverse association observed between acute phase proteins and mammographic density may be
partly attributed to a negative feedback loop impacting cytokine expression in the breast.
Suppressor of cytokine signaling (SOCS) proteins have been hypothesized to suppress cytokine
associated cell proliferation and tumorigenesis in normal breast cells. SOCS act in a negative
feedback loop in STAT pathways and block cytokine signaling by several mechanisms
including: directly inhibiting JANUS kinase receptor binding in STAT pathway, occupying
binding sites of other signaling proteins, or accelerating destruction of activated cytokine-
87

receptor complexes that disrupt any cytokine-associated signaling including cell differentiation
and growth(179).
Studies have shown SOCS expression results in induction of apoptosis and suppression of
growth. Expression of SOCS occur within mammary epithelial cells and act to regulate IL-6 and
prolactin-mediated signal transduction in the breast. In cell-line studies, overexpression of
SOCS1 and SOCS2 resulted in the suppressed growth of breast cancer cells (180). Additionally,
SOCS2 expression was associated with high differentiation and a low proliferation rate when
tested in 50 archival breast cancer samples (181). It is also of interest that among breast cancer
survivors, investigators found higher expression of SOCS2 in tissue as a significant independent
predictor of longer survival (182). In another study, higher expression levels of SOCS 1, 3, 4, and
7 are associated with better outcome in 93 breast cancer survivors followedf for ten years; the
investigators found higher levels of SOCS protein expression among those who remain disease
free compared those who developed recurrence; both disease free survival and overall survival
were improved (183).
The primary limitation of this study is that despite the relatively large sample size which
has 80% power to detect the associations we found, this study has a limited number of
inflammatory markers to evaluate. Specifically, this study does not include measurements of IL-
6 which is a direct indicator of inflammatory activity that may be associated with mammographic
density. Furthermore, CRP and SAA were measured at a single point near the time the
mammogram was taken, however the single measure may not reflect the woman's levels over
short (diurnal) or long (years) periods of time. One study has shown an absence of diurnal
variation of CRP among healthy subjects (77). While women were asked to fast before blood-
draw, we do not have data on adherence to the recommendation . While both acute phase
88

proteins and mammographic density were collected at the 24-month interview (approximately
31 months after diagnosis), the mammograms were taken on average 9 months before the blood
was drawn for the CRP and SAA laboratory assays. Finally, our population included largely
postmenopausal breast cancer cases, therefore we had limited power to look for differences in
the acute phase protein and mammographic density associations by menopausal status.
A strength of this study is the use of data from a cohort of breast cancer survivors with
detailed clinical and laboratory measures including: CRP, SAA, mammographic density as well
as treatment data. Furthermore, our data was obtained from a multiethnic sample drawn from the
same cohort that provided evidence that CRP and SAA are associated with breast cancer survival
(79). We adjusted for both BMI and measures of central adiposity (waist circumference and
waist-to-hip ratio) and found that the inverse association is present with and without adjustment,
although the magnitude of association is smaller after adjustment for BMI. If the impact of BMI
(adiposity) on mammographic density is due to the peripheral production of estrogens in
adipocytes, and cytokines act to stimulate the production of aromatase (and therefore contribute
to estrogen production) then controlling for BMI would underestimate the impact of cytokines or
inflammatory proteins on mammographic density. If instead the impact of adiposity on
mammographic density is due to the direct impact of IL-6 or inflammatory proteins on cellular
proliferation, than adjusting for measures of adiposity may be controlling for a correlate of the
exposure (over adjusting models). Because adipocytes are a production site for IL-6 and IL-6
may act on breast cells directly or through their influence on estrogen production in post-
menopausal women, it seems that adjustment for BMI (a measure of adiposity) would be an
over correction. However, we have shown the models with and without the adjustment for BMI.
89

While an association between inflammation and carcinogenesis has been hypothesized
(184), and is believed to impact risk through the impact of inflammation on cellular proliferation
(185), the association between inflammation and mammographic density has not been
established. Although we found a statistically significant inverse association between CRP and
mammographic density, the direction of the effect is somewhat unexpected though consistent
with certain in-vitro studies on the inhibitory effect of cytokines on breast cells; furthermore, the
magnitude of the association has a suggestive trend of being greater among those who take anti-
inflammatories or undertake activities that reduce inflammation (e.g. not smoking, >26MET
physical exercise, lower omega 6:3 ratio). The inverse association was not explained by
measures of adiposity (BMI, waist circumference) or age. Additional investigations into the
biological mechanism that may explain this effect are needed and may improve the treatment of
women with breast cancer through targeted therapies or recommendations for post-diagnosis
monitoring or lifestyle.
90

Table 1. Descriptive table of characteristics in a study population (N=479) by median %
mammographic density drawn from Health, Eating, Activity and Lifestyle (HEAL) Study.

All (n=479)

MD <13.00
(n=238)
(n=238)
MD >=13.00
(n=241)

Mean (SD) Mean (SD) Mean (SD) p-val
value CRP (mg/L) 4.80(9.49) 5.76(11.9) 3.85(6.2) 0.03
SAA (mg/L) 11.6 (35.1) 12.6(38.2) 10.7(31.8) 0.55
Mean Dense Area (1000 pixels) 53.0 (48.3) 22.4(18.9) 83.3(49.3) <.0001
Mean % Density 15.5(12.5) 5.57(3.8) 25.2(10.2) <.0001
Age at interview (years) 60.9 (9.7) 61.8(9.7) 60.1(9.59) 0.06
Age at menarche
1
12.6(1.6) 12.4(1.6) 12.7(1.6) 0.10
Time between interview
and Mammogram (months)
2
9.73(6.7)

9.60(6.2) 9.85(7.08)
0.71
Body Mass Index (kg/m
2
) 27.7(6.4) 29.8(6.65) 25.6(5.4) <.0001
Physical Activity (MET h/week) 25.6(26.3) 22.4(23.4) 28.7(28.6) 0.01
Omega 6:3 Ratio
3
9.1(2.9) 9.1(2.8) 9.1(8.7) 0.98

n (%) n (%) n (%)
Race/Ethnicity 0.07
White, Non-Hispanic (WA) 92(19.2) 55(23.1) 37(15.4)
White, Non-Hispanic (NM) 214(44.7) 93(39.1) 121(50.2)
Black/African-American 109(22.8) 57(24.0) 52(21.6)
Hispanic 50(10.4) 24(10.0) 26(10.8)
Other 14(2.92) 9(3.78) 5(2.07)

Highest Level of Education
Less than High School 26(5.43) 19(7.98) 7(2.90) 0.03
High school grad 113(23.6) 64(26.9) 49(20.3)
Some college/tech 164(34.2) 76(31.9) 88(36.5)
College grad 81(16.9) 38(16.0) 43(17.8)
Grad school 95(19.8) 41(17.2) 54(22.4)

Post-Menopausal Therapy Use
(%)

Never 145(30.3) 76(31.9) 69(28.6) 0.03
Estrogen Only 139(29.0) 79(33.2) 60(24.9)
Estrogen and Progestin 195(40.7) 83(34.9) 112(46.5)

Parity (%)
Nulliparous 70(14.61) 32(13.45) 38(15.77) 0.02
One or two live births 213(44.47) 94(39.50) 119(49.38)
Three or more live births 196(40.92) 112(47.06) 84(34.85)

1
One woman missing age at menarche data.
2
81 women do not have this datum.
3
Omega 6:3 ratio is calculated as ratio of Omega 6(linoleic acid,linolenic acid,arachidonic acid,linolelaidic acid) to
Omega 3 consumed according to FFQ. Data for 8 women are unavailable.
91

Table 1, continued.

Age At First Birth (%)
1

<18 40(9.8) 26(12.6) 14(6.9) 0.18
18-25 249(60.9) 125(60.7) 124(61.1)
>=25 120(29.3) 55(26.7) 65(32.0)

Smoking Status (%) 0.26
Never 234(48.9) 116(48.7) 118(49.0)
Past 190(39.7) 100(42.0) 90(37.3)
Current 55(11.5) 22(9.24) 33(13.7)

History of Medication Use (yes,
%)

Tamoxifen 236(49.3) 116(48.7) 120(49.8) 0.82
NSAID 196(40.9) 101(42.4) 95(39.4) 0.50
Chemotherapy
2
130(27.1) 61(25.6) 69(28.6) 0.44
Inflammatory Drugs 286(59.7) 149(62.6) 137(56.9) 0.20

History of Medical Conditions
(yes, %)

High blood pressure or
hypertension
2
187(39.0) 111(46.6) 76(31.5) 0.002
Diagnosed with arthiritis
3
208(43.4) 113(47.5) 95(39.4) 0.20
Chronic lung disease
emphysema/bronchitis
67(14.0) 43(18.1) 24(9.96) 0.01
Diabetes or high blood sugar
4
55(11.5) 41(17.2) 14(5.81) <.001
Heart attack of myocardial infarction 20(4.2) 11(4.62) 9(3.73) 0.63
Heart failure or congestive heart
failure
11(2.30) 10(4.20) 1(0.41) 0.01
Cancer found anywhere else 23(4.80) 13(5.46) 10(4.15) 0.49
Heart disease (heart failure/high bp) 190(39.7) 114(47.9) 76(31.5) <.001

1
70 women are nulliparous and do not have this data.

2
9 women do not have this data
3
2 woman do not have this data.
3
1 woman do not have this datum.

92

Table 1b. Association between characteristics in a study population (N=479) drawn from Health, Eating, Activity and Lifestyle (HEAL)
Study and geometric means of exposure (CRP/SAA,) and outcome(mammographic % density).
Mean CRP p-value Mean SAA p-value
Mean
Mammographic % Density p-value
Age at 24months interview (Years) 0.528 <.001 0.024
<49.9 1.74(1.25-2.42) 4.76(3.85-5.88) 15.7(12.5-19.3)
50-54.9 2.27(1.66-3.10) 6.89(5.64-8.42) 15.2(12.2-18.5)
55-59.9 2.22(1.76-2.80) 6.39(5.50-7.41) 12.4(10.4-14.6)
60-64.9 2.52(1.95-3.27) 6.19(5.23-7.31) 11.7(9.50-14.2)
65-69.9 2.05(1.52-2.76) 6.44(5.32-7.80) 13.9(11.1-16.9)
>=70 2.51(1.92-3.29) 9.36(7.87-11.12) 9.8(7.8-12.2)
Age at Menarche (Years)
0.085

0.175

0.349

<12 (median)
2.12(1.86-2.42)

6.45(5.93-7.03)

13.0(11.8-14.3)

>=12
2.68(2.13-3.37)

7.28(6.26-8.46)

11.8(9.8-14.0)

Race/Ethnicity
0.004 0.228 0.012
White, Non-Hispanic (WA)
1.88(1.46-2.43) 5.58(4.71-6.60) 10.4(8.31-12.6)
White, Non-Hispanic (NM)
1.95(1.65-2.30) 7.10(6.35-7.93) 14.8(13.1-16.5)
Black/African-American
3.28(2.59-4.14) 6.80(5.82-7.94) 11.5(9.49-13.7)
Hispanic
2.55(1.8-3.61) 6.57(5.22-8.26) 12.6(9.53-16.0)
Other
1.91(0.99-3.68) 6.31(4.09-9.74) 9.2(4.73-15.2)
Post-Menopausal Therapy 0.027 0.772 0.011
Never 2.26(1.84-2.77) 6.91(6.04-7.91) 12.2(10.4-14.1)
Estrogen Only 2.78(2.25-3.42) 6.53(5.69-7.50) 10.8(9.06-12.7)
Combined Estrogen and Progesterone 1.91(1.60-2.28) 6.50(5.79-7.31) 14.6(12.9-16.5)
Parity

0.277

0.375

0.002

Nulliparous 2.29(1.70-3.07)

6.80(5.60-8.26)

14.3(11.5-17.3)

One or two live births 2.03(1.71-2.4)

6.26(5.60-7.00)

14.3(12.7-16.1)

Three or more live births 2.47(2.07-2.95)

7.00(6.24-7.87)

10.6(9.1-12.1)

93

Table 1b, continued.

Tamoxifen use 0.001 0.031 .779
No 2.69(2.30-3.15) 7.19(6.48-7.98) 12.9(11.4-14.4)
Yes 1.85(1.58-2.17) 6.11(5.50-6.79) 12.6(11.1-14.1)
Smoking 0.495 0.813 0.111
Never 2.15(1.83-2.53) 6.69(6.01-7.44) 12.7(11.3-14.3)
Past 2.22(1.86-2.66) 6.70(5.95-7.54) 11.9(10.3-13.6)
Current 2.69(1.93-3.76) 6.20(4.98-7.72) 15.8(12.5-19.5)
BMI WHO Categories <.0001 <.0001 <.0001
<25 1.21(1.03-1.43) 5.46(4.85-6.14) 17.9(16.0-19.8)
25-29.9 2.67(2.23-3.21) 7.07(6.22-8.04) 11.9(10.3-13.7)
≥30 4.11(3.40-4.97) 7.98(6.98-9.13) 8.02(6.63-9.55)
Omega 6:3 Categories (tertiles)
0.580

0.269

0.803

<8.05
2.09(1.72-2.55)

6.45(5.67-7.34)

12.32(10.56-14.21)

8.05-9.75
2.25(1.85-2.74)

7.25(6.37-8.25)

13.02(11.2-14.97)

>9.75
2.42(1.99-2.95)

6.29(5.53-7.16)

13.14(11.32-15.09)

Physical Activity 0.002 0.019 0.001
<13 MET hrs/week 2.86(2.39-3.42) 7.57(6.73-8.52) 10.4(8.93-12.0)
13-26 MET hrs/week 2.10(1.68-2.64) 6.17(5.31-7.16) 14.7(12.5-17.1)
>26 MET hrs/week 1.80(1.50-2.17) 6.05(5.35-6.83) 14.1(12.3-15.9)

Anti-Inflammatory Drug Use 0.585 0.053 0.353
No 2.10(1.76-2.51) 6.07(5.40-6.82) 13.3(11.7-15.1)
Yes 2.34(2.02-2.71) 7.04(6.4-7.75) 12.3(11.0-13.7)
Current NSAID Use 0.604 0.585
No 2.30(1.98-2.66) 6.75(6.13-7.43) 12.8(11.4-14.2) 0.873
Yes 2.16(1.81-2.58) 6.47(5.76-7.27) 12.6(11.0-14.3)

94

Table 1b, continued.

Heart Disease
<.0001 <.0001 <.0001
No 1.83(1.58-2.11) 5.96(5.42-6.55) 14.6(13.2-16.1)
Yes 3.05(2.56-3.64) 7.81(6.95-8.78) 10.1(8.70-11.7)
Arthritis
1

0.032 0.030 0.047
No 2.00(1.72-2.32) 6.19(5.60-6.83) 13.7(12.3-15.2)
Yes 2.56(2.16-3.04) 7.30(6.53-8.17) 11.6(10.1-13.2)

1
2 people do not have this data.
95

Table 2a. Linear regression models for log
e
CRP (exposure) on transformed percent mammographic density (square root) after adjusting
for covariates (n=479).
Covariates Β SE 95%CI p-value adj R-sq

Unadjusted -0.36 0.06 (-0.47, -0.24) <.0001 0.07
Age and Race -0.33 0.06 (-0.44, -0.21) <.0001 0.12
Age, Race, PMH
1
-0.32 0.06 (-0.43, -0.20) <.0001 0.13
Age, Race, PMH, BMI
1

-0.15 0.06 (-0.27, -0.03) 0.01 0.20
Age, Race, PMH, Heart Disease, Arthritis
2

-0.28 0.06 (-0.39, -0.16) <.0001 0.14
Age, Race, PMH, Heart Disease, Arthritis
2
, BMI
1

-0.14 0.06 (-0.26, -0.02) 0.02 0.20
1. BMI categorical variable of <25, 25-29.9, >=30. Post-menopausal hormone use (PMH) is categorized as never, ever estrogen only, ever
estrogen and progestin only.
2 n=427 in models including arthritis as two women have missing data on whether they've ever been diagnosed with arthritis.

Table 2b. Linear regression models for log
e
SAA (exposure) on transformed percent mammographic density (square root) after adjusting
for covariates.
Covariates Β SE 95%CI p-value adj R-sq

Unadjusted

-0.31 0.09 (-0.48, -0.13) 0.002 0.02
Age and Race

-0.27 0.09 (-0.44, -0.09) 0.007 0.08
Age, Race, PMH

-0.27 0.09 (-0.45, -0.09) 0.006 0.09
Age, Race, PMH, BMI
1

-0.10 0.09 (-0.27, 0.07) 0.24 0.19
Age, Race, PMH, Heart Disease, Arthritis
2

-0.22 0.09 (-0.39, -0.04) 0.02 0.11
Age, Race, PMH, Heart Disease, Arthritis
2
,BMI
1

-0.09 0.09 (-0.26, 0.09) 0.33 0.19
1. BMI categorical variable of <25, 25-29.9, >=30. Post-menopausal hormone use (PMH) is categorized as never, ever estrogen only, ever
estrogen and progestin only.
2. n=427 in models including arthritis as two women have missing data on whether they've ever been diagnosed with arthritis.
96

Table 3a. Linear regression models for log
e
CRP on mammographic density (continuous) outcome
on square root transformation) stratified on variables related to hormone, inflammation, or adiposity
after adjustment for covariates
1
.

n β SE 95%CI p-value p-inter
Post-Menopausal Therapy .799
Never 145 -0.10 0.12 (-0.33, 0.13) 0.40
Estrogen Only 139 -0.07 0.11 (-0.28, 0.14) 0.53
Combined Estrogen and Progestin 195 -0.23 0.10 (-0.42, -0.03) 0.02

Current Tamoxifen use
2
.836
No 243 -0.13 0.08 (-0.29, 0.02) 0.10
Yes 236 -0.17 0.10 (-0.36, 0.02) 0.08

Current NSAID Use .776
No 283

-0.14 0.08 (-0.29, 0.01) 0.07
Yes 196 -0.19 0.10 (-0.39, 0.01) 0.06

Current use of anti-inflammatories
3
.113
No 193 -0.03 0.09 (-0.21, 0.15) 0.74
Yes 286 -0.24 0.08 (-0.40, -0.07) <.01

Arthritis
4
.557
No 269 -0.11 0.08 (-0.27, 0.05) 0.17
Yes 208 -0.22 0.09 (-0.40, -0.04) 0.02

Heart Disease .678
No 289 -0.13 0.08 (-0.28, 0.02) 0.10
Yes 190 -0.20 0.11 (-0.41, 0.01) 0.06

Smoking history .707
Never 234 -0.16 0.09 (-0.34, 0.03) 0.10
Past 190 -0.24 0.10 (-0.43, -0.05) 0.01
Current 55 0.002 0.18 (-0.37, 0.37) 0.99

Body Mass Index 0.294
<25 kg/m
2
184 -0.02 0.09 (-0.20, 0.16) 0.83
25-29.9 kg/m
2
154 -0.32 0.12 (-0.56, -0.08) 0.01
≥30 kg/m
2
141 -0.16 0.12 (-0.41, 0.09) 0.20

1
Covariates include age 31months after diagnosis, race, BMI (<25,25-29.9,>=30), and PMH use at
diagnosis (never, ever estrogen only use, ever estrogen and progestin use).
2
Tamoxifen use(yes/no) 31months after diagnosis.
3
Anti-inflammatories include beta blockers, ace inhibitors, hyperlipedmia drugs, estrogens/progestins,
antigout, nsaid, antipyretics, corticosteroids-inhalation/nasal drugs.
4
Arthritis data is missing for two people.
97

Table 3a continued.

Physical Activity 0.551
<13 MET hrs/week 187 -0.09 0.11 (-0.31, 0.14) 0.44
13-26 MET hrs/week 116 0.02 0.14 (-0.26, 0.30) 0.88
>26 MET hrs/week 176 -0.25 0.09 (-0.42, -0.08) 0.004

Omega 6:3 Categories (tertiles) 0.632

<8.05 157 -0.26 0.11 (-0.47, -0.05) 0.02
8.05-9.75 156 -0.12 0.10 (-0.32, 0.08) 0.26
>9.75 158 -0.13 0.12 (-0.36, 0.11) 0.28

Age 31months post-dx(Years) .948
<49.9 57 -0.35 0.18 (-0.71, 0.00) 0.05
50-54.9 63 -0.23 0.20 (-0.64, 0.18) 0.26
55-59.9 114 -0.08 0.13 (-0.33, 0.18) 0.55
60-64.9 91 -0.28 0.15 (-0.58, 0.03) 0.08
65-69.9 69 -0.12 0.14 (-0.41, 0.16) 0.38
>=70 85 -0.12 0.17 (-0.46, 0.22) 0.50

Race/Ethnicity .993
White, Non-Hispanic (WA) 92 -0.07 0.15 (-0.36, 0.21) 0.61
White, Non-Hispanic (NM) 214 -0.17 0.10 (-0.36, 0.02) 0.08
Black/African-American 109 -0.17 0.13 (-0.42, 0.08) 0.19
Hispanic 50 -0.13 0.18 (-0.50, 0.24) 0.48
Other 14 -0.12 0.27 (-0.77, 0.53) 0.67

98

Table 3b. Linear regression models for log
e
(SAA) on mammographic density (continuous outcome
on square root transformation) stratified on variables related to hormone, inflammation, or adiposity
after adjustment for covariates
5
.

N β SE 95%CI p-value p-inter
Post-Menopausal Therapy 0.410
0.4101
Never 145 -0.15 0.15 (-0.44, 0.15) 0.33
Estrogen Only 139 -0.02 0.16 (-0.33, 0.29) 0.91
Combined Estrogen and Progestin 195 -0.10 0.15 (-0.40, 0.21) 0.52

Current Tamoxifen use
6
0.665
No 243 -0.03 0.12 (-0.26, 0.20) 0.82
Yes 236 -0.18 0.13 (-0.44, 0.08) 0.18

Current NSAID Use 0.255
No 283

-0.17 0.11 (-0.40, 0.05) 0.13
Yes 196 0.01 0.15 (-0.28, 0.29) 0.96

Current use of anti-inflammatories
7
0.904
No 193 -0.05 0.15 (-0.33, 0.24) 0.75
Yes 286 -0.12 0.11 (-0.34, 0.10) 0.27

Arthritis
8
0.786
No 269 -0.06 0.12 (-0.30, 0.18) 0.60
Yes 208 -0.13 0.13 (-0.38, 0.12) 0.32

Heart Disease 0.197
No 289 0.05 0.12 (-0.18, 0.28) 0.68
Yes 190 -0.31 0.13 (-0.57, -0.05) 0.02

Smoking history 0.836
Never 234 -0.16 0.14 (-0.43, 0.11) 0.25
Past 190 -0.11 0.15 (-0.41, 0.19) 0.46
Current 55 0.05 0.21 (-0.37, 0.47) 0.80

Body Mass Index 0.349
<25 kg/m
2
184 0.13 0.13 (-0.14, 0.39) 0.34
25-29.9 kg/m
2
154 -0.20 0.17 (-0.54, 0.14) 0.25
≥30 kg/m
2
141 -0.27 0.16 (-0.59, 0.04) 0.09

5
Covariates include age 31months after diagnosis, race, BMI (<25,25-29.9,>=30), and PMH use at
diagnosis (never, ever estrogen only use, ever estrogen and progestin use).
6
Tamoxifen use(yes/no) 31months after diagnosis.
7
Anti-inflammatories include beta blockers, ace inhibitors, hyperlipedmia drugs, estrogens/progestins,
antigout, nsaid, antipyretics, corticosteroids-inhalation/nasal drugs.
8
Arthritis data is missing for two people.
99

Table 3a continued.

Physical Activity 0.670
<13 MET hrs/week 187 -0.12 0.14 (-0.39, 0.15) 0.39
13-26 MET hrs/week 116 0.05 0.21 (-0.36, 0.46) 0.80
>26 MET hrs/week 176 -0.18 0.14 (-0.47, 0.10) 0.20

Omega 6:3 Categories (tertiles) 0.101

<8.05 157 -0.28 0.16 (-0.60, 0.05) 0.10
8.05-9.75 156 -0.18 0.13 (-0.44, 0.09) 0.18
>9.75 158 0.13 0.17 (-0.21, 0.46) 0.46

Age 31months post-dx(Years) 0.945
<49.9 57 -0.29 0.27 (-0.84, 0.26) 0.29
50-54.9 63 -0.50 0.38 (-1.25, 0.26) 0.19
55-59.9 114 0.10 0.19 (-0.27, 0.47) 0.60
60-64.9 91 -0.23 0.19 (-0.61, 0.16) 0.24
65-69.9 69 -0.04 0.23 (-0.51, 0.43) 0.86
>=70 85 -0.08 0.20 (-0.47, 0.32) 0.70

Race/Ethnicity 0.808
White, Non-Hispanic (WA) 92 -0.09 0.24 (-0.56, 0.39) 0.72
White, Non-Hispanic (NM) 214 -0.06 0.13 (-0.32, 0.20) 0.66
Black/African-American 109 -0.24 0.18 (-0.59, 0.11) 0.17
Hispanic 50 0.05 0.31 (-0.56, 0.67) 0.86
Other 14 0.04 0.34 (-0.76, 0.84) 0.91
100

Supplement Table 1. Associations between geometric mean mammographic density by midpoint of
CRP after adjusting for covariates: (age, BMI(<25,25-29.9,>=30), post-menopausal hormone
use(never, estrogen only, combined estrogen and progestin).

n CRP<2.4 CRP≥2.4 Overall
Overall 11.5(9.87-13.3) 9.95(8.45-11.6)

Post-Menopausal Therapy
Never 74,71 11.1(8.67-13.8) 10.2 (7.55-13.2) 10.3(8.42-12.3)
Estrogen Only 53,86 13.3(11.0-15.7) 9.43(7.11-12.1) 9.79(7.99-11.8)
Estrogen and Progestin 114,81 9.25(7.19-11.6) 11.2(8.96-13.7) 12.2(10.5-14.1)

Current Tamoxifen use
9

No 113,137 12.0(9.86-14.4) 11.1(9.10-13.3) 11.4(9.82-13.1)
Yes 133,107 10.7(8.88-12.8) 8.98(7.12-11.1) 10.0(8.52-11.7)

Current NSAID Use
No 142,145 10.6(8.71-12.8) 12.7(10.4-15.2) 10.6(9.00-12.2)
Yes 104,99 10.4(8.50-12.4) 9.21(7.29-11.4) 10.9(9.31-12.7)

Smoking history
Never 123,115 11.6(9.52-13.9) 10.9(8.69-13.3) 10.8 (9.28-12.5)
Past 99,98 13.8(9.55-18.7) 10.1(8.18-12.2) 9.99(8.37-11.8)
Current 31,55 9.10(7.15-11.3) 11.5(8.01-15.5) 12.6(9.74-15.8)

Body Mass Index
10

<25 kg/m
2
132,56 16.6(14.2-19.3) 15.4(12.2-19.0) 15.9(13.7-18.2)
25-29.9 kg/m
2
68,90 11.0(8.50-13.8) 9.27(7.23-11.6) 10.1(8.35-12.1)
≥30 kg/m
2
46,98 7.99(5.62-10.8) 6.33(4.76-8.13) 7.08(5.62-8.71)

Physical Activity
<13 MET hrs/week 83,108 10.5(8.32-12.9) 12.5(9.76-15.6) 10.1(8.53-11.8)
13-26 MET hrs/week 63,55 12.7(9.85-14.9) 9.72(7.74-11.9) 12.2(10.0-14.5)
>26 MET hrs/week 100,81 12.0(9.12-15.2) 9.25(7.17-11.6) 10.7(8.96-12.7)

9
Tamoxifen use(yes/no) at 24months interview.
10
BMI is categorized according World Health Organization definition of obesity35. Gallagher D,
Heymsfield SB, Heo M, Jebb SA, Murgatroyd PR, Sakamoto Y. Healthy percentage body fat ranges: an approach
for developing guidelines based on body mass index. Am J Clin Nutr. 2000 Sep;72(3):694-701..
101

Supplement Table 2. Correlation coefficients (Pearson) evaluated between CRP, SAA and measures of mammographic density.

CRP
(mg/L)
SAA
(mg/L)
log
e
(SAA) log
e
(CRP
)
Dense
Area
log
e
Dense
Area
Percent
Density
log
e
Percent
Density
square
root
Percent
Density
square
root
Dense
Area
CRP (mg/L) 0.545 0.615 -0.086 -0.135 -0.109 -0.118 -0.153 -0.137
p-value <.0001 <.0001 0.054 0.003 0.014 0.008 0.001 0.002
SAA (mg/L) 0.619 0.344 0.017 -0.039 -0.005 0.029 -0.038 -0.006
p-value <.0001 <.0001 0.703 0.387 0.906 0.523 0.396 0.899
log
e
(SAA) 0.545 0.619 -0.067 -0.091 -0.081 -0.112 -0.132 -0.130
p-value <.0001 <.0001 0.131 0.041 0.068 0.012 0.003 0.004
log
e
(CRP) 0.615 0.344 -0.165 -0.196 -0.188 -0.257 -0.259 -0.269
p-value <.0001 <.0001 0.000 <.0001 <.0001 <.0001 <.0001 <.0001

102

Figure 1. Flowchart of study population from HEAL.

Women with 24-month
assessment (n=944)
Circulating CRP and SAA
available (n=807)
Post-diagnosis Mammograms
available (n=658)
Women with CRP , SAA, and
mammograms available (n=504)
Women with CRP , SAA, and
mammograms analyzed (n=479)
exclude pre-menopausal (n=112),
unknown menopausal status (n=42)
Exclude women with missing
covariates (n=25)

103

Figure 2a. Scatter plot and predicted regression line of square root percent dense area on
ln(CRP). Predicted values are adjusted for age, BMI, post-menopausal hormone use. Pearson
corr between predicted value according to model and observed value: .463(.390-.541).

104

Figure 2b. Scatter plot and predicted regression line of square root percent dense area on
ln(CRP). Predicted values are adjusted for age, BMI, post-menopausal hormone use. Pearson
corr between predicted value according to model and observed value: .457 (.383-.525).

105

Figure 3a: Scatter plot and predicted regression line of square root percent dense area on
ln(CRP). Predicted values are adjusted for age, race, BMI, post-menopausal hormone use.
Pearson corr between predicted value according to model and observed value: .475 (.296-.450).

106

Figure 3b. Scatter plot and predicted regression line of square root percent dense area on
ln(SAA). Predicted values are adjusted for age, race, BMI, post-menopausal hormone use.
Pearson corr between predicted value according to model and observed value: .378(.298-.452).

107

2.6 References
1. Boyd NF, Guo H, Martin LJ, et al. Mammographic density and the risk and detection of
breast cancer. N Engl J Med 2007;356:227-36.
2. Cil T, Fishell E, Hanna W, et al. Mammographic density and the risk of breast cancer
recurrence after breast-conserving surgery. Cancer 2009;115:5780-7.
3. Habel LA, Dignam JJ, Land SR, et al. Mammographic density and breast cancer after
ductal carcinoma in situ. J Natl Cancer Inst 2004;96:1467-72.
4. Ursin G, Pike M. Mammographic density, hormone therapy, and risk of breast cancer.
Cancer Epidemiol Biomarkers Prev 2006;15:1750; author reply
5. Pike MC. The role of mammographic density in evaluating changes in breast cancer risk.
Gynecol Endocrinol 2005;21 Suppl 1:1-5.
6. Boyd NF, Martin LJ, Yaffe MJ, et al. Mammographic density: a hormonally responsive
risk factor for breast cancer. J Br Menopause Soc 2006;12:186-93.
7. Kelemen LE, Sellers TA, Vachon CM. Can genes for mammographic density inform
cancer aetiology? Nat Rev Cancer 2008;8:812-23.
8. Cuzick J, Warwick J, Pinney E, et al. Tamoxifen-induced reduction in mammographic
density and breast cancer risk reduction: a nested case-control study. J Natl Cancer Inst
2011;103:744-52.
9. McTiernan A, Martin CF, Peck JD, et al. Estrogen-plus-progestin use and
mammographic density in postmenopausal women: Women's Health Initiative
randomized trial. J Natl Cancer Inst 2005;97:1366-76.
10. Greendale GA, Palla SL, Ursin G, et al. The association of endogenous sex steroids and
sex steroid binding proteins with mammographic density: results from the
Postmenopausal Estrogen/Progestin Interventions Mammographic Density Study. Am J
Epidemiol 2005;162:826-34.
11. Vachon CM, Sellers TA, Vierkant RA, et al. Case-control study of increased
mammographic breast density response to hormone replacement therapy. Cancer
Epidemiol Biomarkers Prev 2002;11:1382-8.
12. Boyd NF. Tamoxifen, mammographic density, and breast cancer prevention. J Natl
Cancer Inst 2011;103:704-5.
13. Johansson H, Gandini S, Bonanni B, et al. Relationships between circulating hormone
levels, mammographic percent density and breast cancer risk factors in postmenopausal
women. Breast Cancer Res Treat 2008;108:57-67.
14. Tamimi RM, Hankinson SE, Colditz GA, et al. Endogenous sex hormone levels and
mammographic density among postmenopausal women. Cancer Epidemiol Biomarkers
Prev 2005;14:2641-7.
15. Boyd NF, Stone J, Martin LJ, et al. The association of breast mitogens with
mammographic densities. Br J Cancer 2002;87:876-82.
16. Aiello EJ, Tworoger SS, Yasui Y, et al. Associations among circulating sex hormones,
insulin-like growth factor, lipids, and mammographic density in postmenopausal women.
Cancer Epidemiol Biomarkers Prev 2005;14:1411-7.
17. Reed MJ, Purohit A. Breast cancer and the role of cytokines in regulating estrogen
synthesis: an emerging hypothesis. Endocr Rev 1997;18:701-15.
18. McTiernan A, Rajan KB, Tworoger SS, et al. Adiposity and sex hormones in
postmenopausal breast cancer survivors. J Clin Oncol 2003;21:1961-6.
108

19. Visser M, Bouter LM, McQuillan GM, et al. Elevated C-reactive protein levels in
overweight and obese adults. JAMA 1999;282:2131-5.
20. Yamada T. Serum amyloid A (SAA): a concise review of biology, assay methods and
clinical usefulness. Clin Chem Lab Med 1999;37:381-8.
21. Schultz DR, Arnold PI. Properties of four acute phase proteins: C-reactive protein, serum
amyloid A protein, alpha 1-acid glycoprotein, and fibrinogen. Semin Arthritis Rheum
1990;20:129-47.
22. Pierce BL, Ballard-Barbash R, Bernstein L, et al. Elevated Biomarkers of Inflammation
Are Associated With Reduced Survival Among Breast Cancer Patients. J Clin Oncol
2009.
23. Allin KH, Nordestgaard BG, Flyger H, et al. Elevated pre-treatment levels of plasma C-
reactive protein are associated with poor prognosis after breast cancer: a cohort study.
Breast Cancer Res 2011;13:R55.
24. Purohit A, Newman SP, Reed MJ. The role of cytokines in regulating estrogen synthesis:
implications for the etiology of breast cancer. Breast Cancer Res 2002;4:65-9.
25. Honma S, Shimodaira K, Shimizu Y, et al. The influence of inflammatory cytokines on
estrogen production and cell proliferation in human breast cancer cells. Endocr J
2002;49:371-7.
26. Purohit A, Ghilchik MW, Duncan L, et al. Aromatase activity and interleukin-6
production by normal and malignant breast tissues. J Clin Endocrinol Metab
1995;80:3052-8.
27. Conze D, Weiss L, Regen PS, et al. Autocrine production of interleukin 6 causes
multidrug resistance in breast cancer cells. Cancer Res 2001;61:8851-8.
28. Chiu JJ, Sgagias MK, Cowan KH. Interleukin 6 acts as a paracrine growth factor in
human mammary carcinoma cell lines. Clin Cancer Res 1996;2:215-21.
29. Sansone P, Storci G, Tavolari S, et al. IL-6 triggers malignant features in mammospheres
from human ductal breast carcinoma and normal mammary gland. J Clin Invest
2007;117:3988-4002.
30. Danforth DN, Jr., Sgagias MK. Interleukin-1 alpha and interleukin-6 act additively to
inhibit growth of MCF-7 breast cancer cells in vitro. Cancer Res 1993;53:1538-45.
31. Hutchins D, Steel CM. Regulation of ICAM-1 (CD54) expression in human breast cancer
cell lines by interleukin 6 and fibroblast-derived factors. Int J Cancer 1994;58:80-4.
32. Shen WH, Zhou JH, Broussard SR, et al. Proinflammatory cytokines block growth of
breast cancer cells by impairing signals from a growth factor receptor. Cancer Res
2002;62:4746-56.
33. Reeves KW, Weissfeld JL, Modugno F, et al. Circulating levels of inflammatory markers
and mammographic density among postmenopausal women. Breast Cancer Res Treat
2011;127:555-63.
34. Woolcott CG, Courneya KS, Boyd NF, et al. Association between sex hormones, glucose
homeostasis, adipokines, and inflammatory markers and mammographic density among
postmenopausal women. Breast Cancer Res Treat 2013.
35. Takkouche B, Regueira-Mendez C, Etminan M. Breast cancer and use of nonsteroidal
anti-inflammatory drugs: a meta-analysis. J Natl Cancer Inst 2008;100:1439-47.
36. Kwan ML, Habel LA, Slattery ML, et al. NSAIDs and breast cancer recurrence in a
prospective cohort study. Cancer Causes Control 2007;18:613-20.
109

37. Holmes MD, Chen WY, Li L, et al. Aspirin intake and survival after breast cancer. J Clin
Oncol 2010;28:1467-72.
38. Blair CK, Sweeney C, Anderson KE, et al. NSAID use and survival after breast cancer
diagnosis in post-menopausal women. Breast Cancer Res Treat 2007;101:191-7.
39. Il'yasova D, Colbert LH, Harris TB, et al. Circulating levels of inflammatory markers and
cancer risk in the health aging and body composition cohort. Cancer Epidemiol
Biomarkers Prev 2005;14:2413-8.
40. Heikkila K, Harris R, Lowe G, et al. Associations of circulating C-reactive protein and
interleukin-6 with cancer risk: findings from two prospective cohorts and a meta-analysis.
Cancer Causes Control 2009;20:15-26.
41. Allin KH, Bojesen SE, Nordestgaard BG. Baseline C-reactive protein is associated with
incident cancer and survival in patients with cancer. J Clin Oncol 2009;27:2217-24.
42. Gonullu G, Ersoy C, Ersoy A, et al. Relation between insulin resistance and serum
concentrations of IL-6 and TNF-alpha in overweight or obese women with early stage
breast cancer. Cytokine 2005;31:264-9.
43. Siemes C, Visser LE, Coebergh JW, et al. C-reactive protein levels, variation in the C-
reactive protein gene, and cancer risk: the Rotterdam Study. J Clin Oncol 2006;24:5216-
22.
44. Fain JN. Release of interleukins and other inflammatory cytokines by human adipose
tissue is enhanced in obesity and primarily due to the nonfat cells. Vitam Horm
2006;74:443-77.
45. Zhao Y, Nichols JE, Bulun SE, et al. Aromatase P450 gene expression in human adipose
tissue. Role of a Jak/STAT pathway in regulation of the adipose-specific promoter. J Biol
Chem 1995;270:16449-57.
46. Ershler WB, Keller ET. Age-associated increased interleukin-6 gene expression, late-life
diseases, and frailty. Annu Rev Med 2000;51:245-70.
47. Pacifici R, Brown C, Puscheck E, et al. Effect of surgical menopause and estrogen
replacement on cytokine release from human blood mononuclear cells. Proc Natl Acad
Sci U S A 1991;88:5134-8.
48. Balkwill F, Charles KA, Mantovani A. Smoldering and polarized inflammation in the
initiation and promotion of malignant disease. Cancer Cell 2005;7:211-7.
49. Pfeilschifter J, Koditz R, Pfohl M, et al. Changes in proinflammatory cytokine activity
after menopause. Endocr Rev 2002;23:90-119.
50. Gabay C, Kushner I. Acute-phase proteins and other systemic responses to inflammation.
N Engl J Med 1999;340:448-54.
51. Zhang D, Sun M, Samols D, et al. STAT3 participates in transcriptional activation of the
C-reactive protein gene by interleukin-6. J Biol Chem 1996;271:9503-9.
52. Ram PA, Waxman DJ. SOCS/CIS protein inhibition of growth hormone-stimulated
STAT5 signaling by multiple mechanisms. J Biol Chem 1999;274:35553-61.
53. Sutherland KD, Lindeman GJ, Choong DY, et al. Differential hypermethylation of SOCS
genes in ovarian and breast carcinomas. Oncogene 2004;23:7726-33.
54. Farabegoli F, Ceccarelli C, Santini D, et al. Suppressor of cytokine signalling 2 (SOCS-2)
expression in breast carcinoma. J Clin Pathol 2005;58:1046-50.
55. Haffner MC, Petridou B, Peyrat JP, et al. Favorable prognostic value of SOCS2 and IGF-
I in breast cancer. BMC Cancer 2007;7:136.
110

56. Sasi W, Jiang WG, Sharma A, et al. Higher expression levels of SOCS 1,3,4,7 are
associated with earlier tumour stage and better clinical outcome in human breast cancer.
BMC Cancer 2010;10:178.
57. Meier-Ewert HK, Ridker PM, Rifai N, et al. Absence of diurnal variation of C-reactive
protein concentrations in healthy human subjects. Clin Chem 2001;47:426-30.
58. Allavena P, Garlanda C, Borrello MG, et al. Pathways connecting inflammation and
cancer. Curr Opin Genet Dev 2008;18:3-10.
59. Mantovani A, Allavena P, Sica A, et al. Cancer-related inflammation. Nature
2008;454:436-44.
111

Chapter 4
Association between smoking, alcohol use and mammographic density in breast cancer
survivors

4.1 Abstract
Authors: Anne Dee
1
, Leslie Bernstein,
1,5
Anne McTiernan
2
, Richard N. Baumgartner
3
, Kathy
Baumgartner
3
, Rachel Ballard-Barbash
4
, Roberta McKean-Cowdin
1

1
University of Southern California, Los Angeles, CA,
2
Fred Hutchinson Cancer Research Center,
Seattle, WA,
3
University of Louisville, Louisville, KY,
4
National Cancer Institute, Bethesda,
MD,
5
City of Hope, Duarte, CA.
Introduction: Mammographic density (MD) is positively associated with breast cancer risk and
with breast cancer recurrence. Factors that influence MD include age, circulating endogenous
hormone levels, hormone therapy, menopausal status, parity, adiposity, and genetic variation.
The association between MD and estrogen levels may be influenced by lifestyle factors including
alcohol consumption, which stimulates estrogen production via aromatase, and cigarette
smoking, which has antiestrogenic properties.
Methods: Alcohol, smoking, and mammographic density (MD) data were collected
approximately 31months after breast cancer diagnosis from 502 postmenopausal women
participating in the Health, Eating, Activity, and Lifestyle Study (HEAL). HEAL is a prospective
cohort study of 1,183 breast cancer survivors identified through Surveillance, Epidemiology, and
End Results registries in Los Angeles County, New Mexico, and Western Washington. Data on
cigarette smoking included data on smoking history as well as recency. Alcohol data was
measured from food frequency questionnaire. For this analysis, regression models were used to
112

estimate associations between alcohol, smoking and percent MD after adjustments for age, race,
body mass index (BMI), post-menopausal hormone use, and physical activity level. Potential
effect modification by factors that could influence circulating estrogen were evaluated by
stratification and with formal tests of interaction.
Results: We found no association between alcohol consumption (gm/day) and percent MD after
adjusting for covariates (β=0.011; p=0.26) . Further, no statistically significant association was
found between cigarette smoking and MD, when measured by pack-years (β=-0.0023; p=0.64),
duration (β=0.0120; p=0.11) overall, or by recency, or cigarettes per day (β=-0.0052; p=0.53).
We did not observe any evidence of effect modification by BMI, post-menopausal replacement
therapy, or physical activity.
Conclusion: We found no evidence that moderate alcohol use after breast cancer diagnosis,
including wine, beer, or alcohol, was associated with higher MD among post-menopausal, breast
cancer survivors. Further, cigarette smoking was not associated with MD in our multi-ethnic
sample of postmenopausal breast cancer survivors.
113

4.1 Introduction
Mammographic density is one of the strongest predictors of breast cancer risk and may
predict breast cancer recurrence(1, 2, 3, 4, 5). Percent mammographic density represents the
proportion of a mammogram occupied by radiologically dense tissue, which includes both dense
and adipose tissue. The dense areas of the breast result from increased numbers of epithelial and
stromal cells(6), which change over the course of a woman’s life. Factors that influence
mammographic density include age, menopausal status, parity, adiposity, tamoxifen therapy,
genetic variation, and other anthropometric factors that influence circulating endogenous
hormones(4, 7, 8). Specifically, higher mammographic density has been found among women
using exogenous hormones such as estrogen and progestin combined therapy(9, 10, 11) and use
of tamoxifen has been associated with reduced mammographic density(8, 12). However,
evidence for an association between circulating levels of sex hormones and increased breast
density is more conflicting: studies have indicated positive(10, 13), inverse(14), or no
associations(15, 16).
Alcohol consumption has been established as a risk factor for breast cancer in
epidemiological and animal studies, but the mechanism by which alcohol influence breast cancer
risk remains unclear(17). Experimental and clinical studies indicated that alcohol plays a role in
the formation of genotoxic metabolites (e.g. acetaldehydes) and may induce changes in
circulating estrogen levels as well as insulin-like growth factor (IGF-1) and insulin-like growth
factor binding proteins (IGF-BP3)(18). Among post-menopausal women, higher alcohol
consumption is associated with higher circulating estradiol and estrone levels(19). This may be
due to the influence of alcohol on increased expression of the aromatase enzyme, which
contributes to the peripheral production of estrone(20). Furthermore, in human breast cancer cell
114

lines, the in vitro addition of ethanol catalyzes the conversion of androgen to estrogen causing a
dose-dependent increase of estrogen receptor (ER+) transcription activity and higher aromatase
expression(21). Additionally, aromatase inhibitors, which prevent the conversion of androgens
to estrogens, occur naturally in the grape skin preserved in the making of red wine(22); these
same aromatase inhibitors are not found in white wine(22), suggesting that, alcohol types (e.g.
red or white wine, beer, or liquor) may have variable impacts on mammographic density since
breast density is sensitive to circulating endogenous hormone levels(23, 24).
Most epidemiological studies that have looked at the association between alcohol
consumption and mammographic density do so in context of a larger dietary and lifestyle studies.
The majority of these studies have shown a positive association between current alcohol
consumption and mammographic density in pre and postmenopausal women(25, 26, 27, 28, 29,
30, 31, 32), although four of these studies had suggestive results that did not reach statistical
significance(29, 30, 31, 32). A few studies have shown no association(30, 33, 34). It is
interesting that for many of these studies, both the magnitude of the associations and the
statistical significance tend to be reduced after adjusting for body mass index.
There have been nine studies conducted on the effects of smoking on mammographic
density, most of them with populations consisting of both pre and postmenopausal women(25,
35, 36, 37, 38, 39, 40, 41, 42). Three studies found inverse associations between current cigarette
smoking and MD among premenopausal women and cite the antiestrogenic properties of tobacco
as the potential biological mechanism(25, 37, 40). The association between cigarette smoking
and MD among postmenopausal women are not consistent: four studies among postmenopausal
women found an inverse association(27, 36, 39, 41); other studies of postmenopausal women
have found no associations between cigarette smoking and percent MD when investigated among
115

postmenopausal Minnesotan(25), British(42), Hispanic(38) and Alaska native women (53%
considered postmenopausal) (35). Antiestrogenic components of cigarette smoke include 3-
methylcholanthrene and benzoapyrene which induces cytochrome p450 activity and the
production of metabolites with antiestrogenic properties. Additionally, laboratory studies have
found that smoking inhibits gonadotropin release and aromatase activity, thus limiting estrone
production(43, 44). In epidemiological studies, smoking has been associated with other
reproductive health issues including infertility and early menarche which are examples of
endocrine activity disruption(40) and experimental and epidemiologic studies have established
that cigarette smoke has carcinogenic effects on the body, including increased risk of breast
cancer (45). Tobacco smoke clearly impacts breast tissue and the antiestrogenic effect of
smoking may influence mammographic density.
We are interested in understanding the modifiable factors that may influence
mammographic density in post-menopausal, breast cancer survivors, with the potential to
improve prognosis through dietary recommendations. While increased alcohol intake may result
in increased levels of circulating estrogens associated with higher mammographic density, a diet
that includes red wine also may result in lower circulating estrogens in some women; therefore,
we will evaluate dietary intake of alcohol overall, as well as by type of alcohol (e.g. wine, beer,
liquor) to determine the potential impact of alcohol use on mammographic density after breast
cancer diagnosis. While tobacco smoking would never be recommended, we will also examine
the potential association between tobacco use and percent mammographic density to better
understand the factors that may influence percent mammographic density in breast cancer
survivors. Using a prospective cohort of breast cancer survivors, we examined the association
116

between post-diagnosis smoking and alcohol consumption, with percent mammographic density
in post-menopausal breast cancer survivors.
4.2 Methods
Study Population
The data for this analysis is abstracted from Health, Eating, Activity and Lifestyle
(HEAL) Study. HEAL is an observational population-based prospective cohort study of 1183
breast cancer survivors. All participants in the study are women 18 years of age or older who
were diagnosed with in-situ to stage IIIa breast cancer from 1996 through 1999 identified
through Surveillance, Epidemiology, and End Results (SEER) registries in New Mexico,
California, and Washington. The aim of the HEAL study is to evaluate the independent roles of
sex-hormones, diet, weight, physical activity, genetics, and other factors on post-diagnostic
breast cancer prognosis and survival. Details of study design and recruitment procedures have
been described previously.
In the study of 1183 women, 615 women were from New Mexico, 202 from Washington,
and 366 from Los Angeles. Interview data were collected at baseline and follow-up. Baseline
interview took place within the first year after diagnosis, on average 6 months post diagnosis and
follow-up data were collected approximately 24 months after baseline (about 31 months after
diagnosis) for 944 women. 755 of those women have post-mammograms data available. For the
current analysis, we used data from 502 women who had post-diagnosis mammographic density
data and alcohol use and smoking data available (Figure 1).
Clinical Variables
Anthropometric measures were taken by trained staff at the baseline and follow-up clinic
visits, including measured height and weight. BMI was computed as weight in kilograms (kg)
117

divided by height in meters squared (m2).
Mammogram density assessment
Using the information on clinical providers and mammogram dates reported by each
woman, mammograms (approximately 2 years post-diagnosis) were collected for each woman.
These mammograms were digitized using an Epson 1680 scanner (Epson America Inc., CA).
Data from mammograms included continuous percent mammographic mammographic density,
dense area (in 1,000 pixels), and total breast area (in 1,000 pixels) on the craniocaudal view of
the breast contralateral to the one diagnosed with breast cancer. All of the density readings were
conducted by one technician using Cumulus 108, a computer program developed by the
University of Toronto and included a 5% random sample of repeat films for quality assurance.
Questionnaire Variables
In-person interviews (New Mexico) and self-administered questionnaire forms
(Washington and California) provided information on demographics, dietary consumption,
menopausal status, smoking status, disease history and current use of any over the counter or
prescription NSAIDs at the 24-month follow-up questionnaire. Cancer treatment history,
including history of radiation, chemotherapy, and tamoxifen use, were obtained through medical
record review, participants’ SEER records, or responses to the questionnaire.
Questionnaire variables regarding alcohol include the type(red/white/pink), frequency,
and amount of wine self-reported in their diet in the last year. Questionnaire variables about
smoking history includes questions on dates when smoking started and stopped and average
number of cigarettes smoked per day.
Alcohol consumption (gm/day) was estimated using data from the Women’s Health
Initiative food frequency questionnaire and the nutrient database from the University of
118

Minnesota’s Nutrition Coordinating Center’s Nutrition Data Systems for Research (NDS-R,
version 2005). Women at the New Mexico study site were asked their usual dietary consumption
in the preceding year, while women at the Washington and California sites were asked about
their consumption in the preceding month.
Analysis
Our overall approach explored the association between alcohol or smoking and
mammographic percent density using linear regression models. In order to fulfill regression
assumptions of linearity and homoscedasticity square root transformation was applied to percent
mammographic density. Regression coefficients (β) and 95% confidence intervals (CIs) were
estimated to assess the associations between continuous values of alcohol(gm/day), type (wine,
beer, liquor; in servings/day) as well as smoking (pack years, smoking duration, and
cigarettes/day) and percent mammographic density among those who drink and smokers(past
and current). Stratum-specific modeling was performed to evaluate potential effect modification
due to smoking status(cessation from smoking <20 years, >=20 years, or current smoker) obesity
status, age categories(<50,60-69.9,>70), race, physical activity level, and post-menopausal
hormone use at diagnosis.
We evaluated the association between drinking status (non-drinkers (0.2gm/day) vs
drinkers (≥ 0.2gm/day)) or smoking status (current, past never) and potential confounders first by
using frequency tables; chi-square tests were used to evaluate statistical differences in frequency
for categorical variables and Wald-test statistics for continuous variables. Further, linear
regression was performed to evaluate association between potential confounders and
mammographic density when treated as a continuous variable or logistic regression to evaluate
associations between potential confounders and the exposure (smoking or alcohol) when treated
119

as categorical variables. Variables associated with both exposure and outcome were considered
as potential confounders and those that substantially change the effect estimates were included in
the final model. Variables considered as potential confounders include: age, race (whites from
Washington, whites from New Mexico, African-American, Hispanics, Others: including those
identified as Asian, American-Indian, Pacific Islander, and small numbers of other descriptors),
BMI(<25, 25-29.9, ≥30), BMI(continuous), age of menarche(<12 years, >=12years), tamoxifen
use, time between interview and mammogram(continuous), breast cancer treatment(surgery only,
chemotherapy, radiation, chemotherapy and radiation), education level(less than high school,
high school graduate, some college/technical school, college, graduate/professional school),
stage(in-situ, proximal, distant), estrogen receptor status(positive, negative, unknown),
progesterone receptor status(positive, negative, unknown), post-menopausal hormone therapy
(never, estrogen only, combined estrogen/progestin), smoking(never, past, current), physical
activity level(<13 MET, 13-26 MET, >26 MET hrs/week), parity (nulliparous, <3, >=3), age at
first birth(<18, 18-25, >25). The final model included: age at interview(continuous), race,
BMI(continuous), post-menopausal hormone use at time of breast cancer diagnosis, and physical
activity level.
Analysis of Covariance (ANCOVA) was used to calculate geometric mean of percent
mammographic density by alcohol use comparing non-drinkers (0.2gm/day) to drinkers (≥
0.2gm/day). Statistical interaction were evaluated by stratification and by including
multiplicative interaction terms in the models for drinking status and selected biological and
lifestyle factors including: smoking (never, past, current), cigarettes/day (never, <10, ≥10), pack
year (never, <10, ≥10), post-menopausal therapy (never, estrogen only, ever estrogen and
progestin), physical activity (<13 MET hrs/week, 13-26 MET hrs/week, >26 MET hrs/week),
120

BMI categories (18-24.9 kg/m
2
, 25-29.9 kg/m
2
, ≥30 kg/m
2
), current tamoxifen use (no, yes),
parity (nulliparous, 1-2, 3+ children), and age at first birth (<18 years, 18-25 years, and ≥30
years).
To further evaluate the influence of alcohol consumption and smoking on mammographic
density, quantile regression was applied after adjusting for covariates at the 20%, 40%, 60%,
80% quantiles of mammographic density in this population. These quantiles corresponds to
percent mammographic density of 10%, 16.6%, 25.5%, and 73.4%. All analyses were conducted
using SAS 9.2 (SAS Institute, Cary, NC).
4.3 Results
In our sample of 502 post-menopausal breast cancer survivors, approximately 50% of the
women were either past (39%) or current smokers (11%) (Table 1). The mean pack years of
smoking for past or current smokers was 21.3, with an average smoking duration of 14 years.
Approximately 46% of the sample, were classified as alcohol drinkers, because they reported
consuming 0.02gm/day alcohol per day or more at the 24-month interview, which was conducted
approximately 31 months after diagnosis. Overall, the women in our study had a relatively low
average consumption of alcohol, 4.0 gm/day. Women classified as drinkers had an average daily
intake of 8.72 gm/day.
The characteristics of the population according to drinking status are presented in Table
1. On average, the drinkers tend to be younger (p=.004), leaner (p=.001), more highly educated
(p<.0001), older at the time of giving birth the first time (p<.001), more likely to use post-
menopausal hormones(p=.040), engage in more vigorous levels of physical activity(<.001), and
less likely to have diabetes or high blood sugar(p=.003).
121

Table 2 presents the geometric mean values of mammographic density by alcohol and
smoking related lifestyle factors after adjustment for age, race, BMI, post-menopausal hormone
use at diagnosis, and physical activity level. We observed no overall consistent pattern for
percent MD by smoking status. Women who were current smokers had higher mean percent MD
than never smokers, however the confidence intervals were wide and were not mutually
exclusive. No consistent patterns were observed by cigarettes per day, smoking duration, or
pack years of smoking. Women who reported smoking in the past 20 years compared to quitting
20 years or more in past, had similar percent MD (10.81 compared to 10.97 respectively).
Overall, women who were non-drinkers (<0.2gm per day of alcohol) had higher
mammographic density, with geometric mean of 11.88 (95%CI:10.28-13.59) compared to
women who drink (≥0.2gm/day) (geometric mean is 10.92(95%CI:9.30-12.68); however, the
geometric means are not statistically significantly different (p=.31) (Table 2). When we
examined geometric mean levels of percent MD by alcohol status (non-drinkers compared to
drinkers) and by smoking status, we observed no statistically significant differences by smoking
status (never, past, current) or by smoking duration. A pattern of higher geometric mean MD was
observed in never and past smokers for non-alcohol drinkers compared to alcohol-drinkers,
however this pattern was reversed in current drinkers, so that the highest geometric mean MD
was among current smokers who also drank alcohol.
Statistically significant interactions were found for alcohol use and categories of
cigarettes/day (p=0.01) and categories of pack years (p=0.02)(Table 2) . When examined by
cigarettes per day, women who reported alcohol use had lower geometric mean % MD than
women who did not drink alcohol at every level of cigarette use (never, <10 per day, and 10+ per
day). More specifically, women who drink and smoke more than 10 cigarettes per day have a
122

lower geometric mean percent mammographic density of 9.34(95%CI: 6.80-12.20) compared to
women who neither drink nor smoke 12.48(95%CI: 10.51-14.61). There is a similar pattern for
pack years. Women who drink and smoke more than 10 pack years have a geometric mean
percent mammographic density of 10.59(95%CI: 8.23-13.23) compared to women who neither
drink nor smoke 12.49(95%CI: 10.52-14.63); however, confidence intervals for both of these
geometric means are not exclusive.
When evaluating geometric mean levels of %MD, there was no statistically significant
interaction between drinking status and post-menopausal hormone use at diagnosis(p=.57),
physical activity(p=.13), obesity status(p=.52), tamoxifen use(p=.77), parity, or age at first
birth(p=.62)(Table 2).
We did not find a statistically significant association between alcohol consumption, or
alcohol type and mammographic density. In Table 3a, we see that amount of alcohol consumed
or type of serving was not statistically significantly associated with percent mammographic
density among alcohol drinkers, although there was an increasing pattern of % MD with
increasing level of alcohol when measured in grams per day for any alcohol (β=0.005, p=0.61),
beer in servings per day ( β=0.147, p=0.50), or wine in servings per day (β=0.182, p=0.37).
Further, among wine drinkers, consumption of red wine exclusively or in combination with other
types of wine did not change the direction or magnitude of association between servings/day and
mammographic density compared to other types of alcohol. In contrast, increasing liquor
consumption in servings per day was associated with a decrease in %MD, however this
association was not statistically significant.
In Table 3b, we showed that there is a consistent inverse association between smoking
variables—in pack years, or cigarettes per day—and percent mammographic density among past
123

or current smokers; however, none of the associations are statistically significant. For smoke
duration, there is a non-statistically significant positive association with percent MD.
To further explore the association between smoking status and percent MD, we stratified
by recency of smoking (< 20 years in ago, or smoking 20 years or more in the past) (Table 4 ).
For smoking by duration, we see that the inverse association with percent MD was present for
more recent smokers (stopped smoking <20 years ago) but was positively associated with
percent MD in non-recent smokers (stopped smoking more than 20 years prior). Other stratum
specific modeling of pack-years or alcohol gm/day on mammographic density including by
obesity status, age categories(<50,60-69.9,>70), race, physical activity level, and post-
menopausal hormone use at diagnosis also showed no association (data not shown).
Figure 2a illustrates the plot of regression coefficients for alcohol gm/day on quantiles
(20
th
, 40
th
, 60
th
, and 80
th
) of untransformed mammographic density among women who drink
alcohol. The figure shows a pattern of greater positive association with increasing quantile at the
60
th
and 80
th
quantiles.
Figure 2b is similar to Figure 2a, showing pack years and the 95% confidence intervals at
20
th
, 40
th
, 60
th
, and 80
th
quantile of untransformed mammographic density among past and
current smokers. The 95%CI bands shows that none of the association is statistically significant
and that the association remains slightly inverse across quantiles.
Similar to Figure2a and 2b, we also plotted estimates of quantiles at smaller intervals that
showed the same overall trend of influence of alcohol consumption or pack years on
mammographic density (Figures not shown).

124

4.4 Discussion
We found no association between cigarette smoking, alcohol consumption and
mammographic density among postmenopausal women, overall or by level of smoking, obesity
or by post-menopausal hormone use. Furthermore, we did not observe important differences
between alcohol consumption by types of alcohol (beer, wine, or liquor) and mammographic
density.
While the previous epidemiologic studies of alcohol use and mammographic density have
suggested a positive association between recent use of alcohol and mammographic density in pre
and postmenopausal women (25, 26, 27, 28, 29, 30, 31, 32), all of these did not reach statistical
significance (29, 30, 31, 32). Our results were consistent with previous studies that found no
association between alcohol use and mammographic density (30, 33, 34, 46, 47). Two case-
control studies of mostly pre-menopausal women and average daily alcohol consumption of
7.4 g/day and range <1–5.35 g/day, found no association with MD; the first study included 735
women (breast cases=290 controls=645) and the second included 400 women (cases are those
considered with high breast cancer risk parenchymal pattern(P2+DY)=200, Controls
(N1+P1)=200), respectively (46, 47). A large cross-sectional study of 2,251 Norwegian
postmenopausal women aged 50–69 years found no association between alcohol consumption
and continuous percent MD overall or by age, body mass index, or use of hormone therapy(34).
The authors speculated that the relatively low average alcohol consumption of 6gm/day
contributed to the null finding. In our study, the mean alcohol consumption was 4 gm/day
overall, which is the lowest of all the studies discussed. However, among women classified as
drinkers, the mean alcohol consumption was 8.72gm/day.
125

Since red wine is rich in polyphenols compared to other alcohol types, it has been
hypothesized that the polyphenol resveratrol, found in grapes and red wine, may influence
mammographic density as resveratrol has been observed to possess antiestrogenic (48) and
antiproliferative properties(49). Anti-aromatase inhibitors have been isolated from red wine
which could act to decrease mammographic density via its ability to decrease levels of
circulating estrone (22). A cross-sectional study of 1,508 women (81.2% postmenopausal) in
Minnosota found wine intake (servings per week) among postmenopausal women was
significantly associated with percent mammographic density. In particular, white wine was found
to be positively associated with percent mammographic density while red wine was inversely
associated with percent mammographic density. One analysis of a longitudinal study of 151
women (90.7% premenopausal) considering alcohol consumption over a woman’s lifetime
showed an overall positive association between alcohol consumption and mammographic density
(after adjusting for age, and body mass index); current alcohol consumption was more strongly
associated with mammographic density than average lifetime alcohol consumption. The same
study showed that for red wine, current consumption was inversely associated with
mammographic density (28). However, in a cross-over clinical trial of 36 women, participants
who consumed 8 ounces of red wine had lower circulating estradiol (E2) levels than those who
consumed white wine, but the differences were not statistically significant (50). Our analysis did
not observe any effect of wine intake on mammographic density by type of wine or alcohol
consumed, possibly due to limited intake of red wine or resveratrol.
Studies on the association between tobacco use and mammographic density have found
conflicting results. Current smoking has been inversely associated with percent mammographic
density among premenopausal women, but the direction of association has varied among
126

postmenopausal women. A study of 799 pre- and perimenopausal women found an inverse
association between smoking and mammographic density with a significant interaction between
smoking and parity (40). Similarly, a population-based cross-sectional study of 906
postmenopausal women also found an inverse association between smoking and mammographic
density; when restricting to current smoking, they found an inverse dose-response relationship
between smoking levels and mammographic density (41). An inverse association is found
among the 246 premenopausal women in a cross-sectional study based in Minnesota; the same
study found no association between smoking and mammographic density(25). Three cross-
sectional studies have found inverse association. One study of 239 women only reported the
inverse association based on 239 women aged 70 years or older with only information on current
smoking status. Another study with detailed smoking history of 907 postmenopausal women in
Norway found current smokers and former smokers have lower mammographic density
compared to never smokers(41). Two studies have found null associations: one involving 313
postmenopausal women and another involving 191 postmenopausal Hispanic women (33, 42).
The differential effects of smoking on breast cancer and breast density (smoking may be
associated with decreased breast density but increased breast cancer risk) may be possible
because tobacco smoke (or various components of tobacco smoke) appears to have multiple
biologic effects on breast tissue including disrupting estrogen production by catalyzing 2-
hydroxylation (51) as well as promoting carcinogenesis in the breast via the formation of DNA
adducts(52).
One population-based cross-sectional study in Spain looked at the interaction between
alcohol, tobacco and its effect on mammographic density in 3,584 pre and postmenopausal
women. The study found increased number of daily cigarettes and increased number of
127

accumulated lifetime cigarettes were inversely associated with high mammographic density, but
not with smoking status. The same study also found a weak positive association between current
alcohol consumption and mammographic density that was modified by tobacco smoking and
menopausal status; higher alcohol consumption was associated with higher mammographic
density among postmenopausal women or non-smokers but not among pre-menopausal women
or current smokers; in a regression model not restricted by menopausal or smoking status, the
study found current alcohol consumption increasing the odds of high MD by 13% (OR = 1.13;
95% CI 0.99–1.28) and high daily grams of alcohol intake were positively associated with
increased MD (P for trend = 0.045) (27).
Strength of the current analysis include the availability of detailed clinical measures and
lifestyle factors for a population of breast cancer survivors. We used detailed, multiple measures
of alcohol consumption and smoking, and continuous measures of mammographic density. Only
three previous studies collected information of pack- years among postmenopausal ever smokers
as well as mammographic density(25, 39, 41). Furthermore, we used data on type of alcohol
consumed (red wine, white wine, beer and liquor) as well as servings of alcohol intake. We also
have a high prevalence of ever smokers in the study (50%) compared to other studies of smoking
and mammographic density among postmenopausal women, only one of which exceeded 45% in
smoking prevalence(41).
The primary limitation of this study is its relative small sample size, which limited the
power of stratum specific modeling. Furthermore, the alcohol consumption is based on one
specific time point of diagnosis, however other studies have found the strongest association with
recent alcohol use rather than lifetime history. While breast cancer survivors sometimes make
lifestyle changes after diagnosis, it is not certain that the alcohol consumption is the same as
128

when the mammogram was taken (on average 10 months before the 24 month interview).
Furthermore, it is not known how long it takes for biological changes on breast tissue resulting
from recent alcohol use to impact breast tissue compared to lifetime cumulative exposure from
alcohol or other estrogenic sources. Our smoking data also assumes continuous use of an average
number of cigarettes during the smoking period when actual pack years may vary over time.
Another limitation is that the median mammographic density in our study is low and that
mammographic density has a narrow range compared to other studies which may contribute to
the reasons why we observed no statistically significant associations.
This analysis of alcohol and smoking on mammographic density among postmenopausal
breast cancer survivors suggest that there is no large impact of recent alcohol use or cigarette
smoking on percent mammographic density. These results suggest that alcohol may not increase
risk of breast cancer recurrence in postmenopausal women, at least through a biological impact
on dense breast tissue. However, our study was relatively small, focused on postmenopausal
women, while studies of premenopausal women have suggested that alcohol may impact breast
density. Further exploration into the interaction between alcohol consumption, smoking, and
mammographic density should be undertaken in both pre and postmenopausal women among
studies with larger sample sizes.

129

Figure 1. Flowchart of study population from HEAL.

Women who completed 24-months
assessment (n=944)
Have post-diagnosis mammograms
available (n=755)
Post-menopausal women with post-
diagnosis mammograms available
(n=575)
Post-menopausal women with post-
diagnosis mammograms, smoking,
and alcohol data available (n=524)
Study population (n=502)
exclude:
missing covariates
(n=22)
exclude:
premenopausal (n=131)
missing menopausal status
(n=49)
130

Table 1. Descriptive characteristics of 502 women in HEAL cohort with alcohol and smoking data.

Non-Drinkers
Drinkers>.2
gm/day
All

(n=273)
Mean (SD)
(n=229)
Mean (SD)
(n=502)
Mean(SD)
(SD)
p-value

Age at Interview 61.8(9.5) 59.3(9.5) 60.7(9.5) .004
BMI 28.7(6.7) 26.9(6.0) 27.9(6.5) .001

Mammographic % Density
1
14.9(12) 16.6(13.3) 15.7(12.6) .143
Dense Area (in 1000 pixels) 50.3(45) 58(57.1) 53.8(51.0)
Follow-up to Mammogram Date
(months)
-10.1(6.9) -9.9(6.8) -10(6.8) .719

Physical Activity (MET h/week) 21.7(23.1) 29.1(29.9) 25.1(26.6) .002

Alcohol (gm/day) 0.02(0.03) 8.72(11.18) 4.0(8.7)
Beer (Servings/Day)

0.1(0.5) 0.1(0.3)
Wine (Servings/Day)

0.3(0.5) 0.1(0.3)
Liquor (Servings/Day)

0.2(0.4) 0.1(0.3)

Pack Years
2
23.3(24.1) 19.2(18.6) 21.3(21.6) .130
Smoking Duration (Years) 29.4(14.1) 26.2(14.0) 14.03(17.15) .074

n(%) n(%) n(%) p-value
3

Race/Ethnicity

.014
White, Non-Hispanic (WA) 40(14.65) 47(20.52) 87(17.33)

White, Non-Hispanic (NM) 109(39.93) 107(46.72) 216(43.03)

Black/African-America 88(32.23) 43(18.78) 131(26.1)

Hispanic 29(53.7) 25(46.3) 54(10.76)

Other 7(2.56) 7(3.06) 14(2.79)

Highest Level of Education

<.0001
Less than High School 20(7.33) 8(3.49) 28(5.58)

High school grad 79(28.94) 40(17.47) 119(23.71)

Some college/tech 103(37.73) 69(30.13) 172(34.26)

College grad 34(12.45) 49(21.4) 83(16.53)

Grad school 37(13.55) 63(27.51) 100(19.92)

1
Untransformed percent mammographic density.
2
Means of pack years and smoking duration for those who indicated past or current smoking.
3
P-value compares difference of distribution between drinkers and non-drinkers.
131

Smoking Status (%)
1

0.396
Never 143(52.38) 106(46.29) 249(49.6)

Past 101(37) 96(41.92) 197(39.24)
Current 29(10.62) 27(11.79) 56(11.16)
Cigarettes/Day
1

0.413
No 143(52.38) 106(46.72) 250(49.8)

<10 68(24.91) 67(29.26) 135(26.89)

10 62(22.71) 55(24.02) 117(23.31)

Parity

0.082
Nulliparous 32(11.72) 39(17.03) 71(14.14)
One or Two 120(43.96) 108(47.16) 228(45.42)
Three or More 121(44.32) 82(35.81) 203(40.44)
Age At First Birth (%)

<.0001
<18 years 33(13.69) 10(5.26) 43(9.98)
18-25 years 156(64.73) 104(54.74) 260(60.32)
≥25 years 52(21.58) 76(40) 128(29.7)

Post-Menopausal Therapy Use

0.040
Never 89(32.6) 68(29.69) 157(31.27)
Estrogen Only 89(32.6) 57(24.89) 146(29.08)
Estrogen and Progestin 95(34.8) 104(45.41) 199(39.64)

Tamoxifen use (yes)
133(48.72) 115(50.22) 248(49.4) 0.738

Physical Activity

0.001
<13 MET hr/wk 128(46.89) 74(32.31) 202(40.24)

13-26 MET hr/wk 66(24.18) 57(24.89) 123(24.5)

> 26 MET hr/wk 79(28.94) 98(42.79) 177(35.26)
History of Medical Conditions
(yes) %)

Liver Disease
2
5(0) 5(0) 10(1.99) 0.954
High blood pressure or hypertension 120(60.3) 79(39.7) 199(39.64)
0.059
Arthiritis
1
128(46.89) 98(42.79) 226(45.02) 0.263
Chronic lung disease emphysema or
bronchitis
3

43(15.75) 29(12.66) 72(14.34) 0.326
Diabetes or high blood sugar
2
42(15.38) 14(6.11) 56(11.16) 0.003
Heart attack of myocardial
infarction
2

19(6.96) 4(1.75) 23(4.58) 0.012
Heart failure or congestive heart
2

failure
11(4.03) 1(0.44) 12(2.39) 0.009
Cancer found anywhere else
2
1(100) 0(0) 1(0.2) 0.657

1
1 woman reported past smoking with age started and stopped smoking but not how many cigarettes per day she
smoked on average. She is classified as a past smoker.
2
Missing data from 2 women.
3
Missing datum from 1 woman.
132

Table 2. Geometric mean
1
of mammographic density (outcome) overall and by alcohol use (non user <0.2 grams/day compared to non-
users ≥0.2 grams/day), and stratified by smoking status, reproductive variables, BMI, tamoxifen use and physical activity.

n Alcohol/day<.2 gm A l c oh ol / d ay≥ .2gm Overall Interaction

Mean(95%CI) Mean(95%CI)

11.88(10.28-13.59) 10.92(9.30-12.68)

0.31
2

Smoke

0.21
Never 143/106 12.51(10.53-14.66) 10.51(8.42-12.84) 11.42(9.84-13.13)
Past
3
101/96 13.30(9.47-17.77) 9.98(8.01-12.18) 10.88(9.19-12.70)
Stopped >=20 years
ago
42/31 10.03(8.04-12.24) 11.78(9.10-14.80) 10.97(8.64-13.58)

Stopped <20 yrs ago 59/65 11.63(8.52-15.22) 13.36(9.52-17.85) 10.81(8.81-13.02)

Current 29/27 11.25(9.07-13.66) 13.46(9.45-18.18) 13.41(10.47-16.72)

Cigarettes/Day

0.01
Never 143/106 12.48(10.51-14.61) 10.00(7.62-12.70) 11.52(9.92-13.23)

<10 68/67 12.32(9.63-15.34) 10.07(8.10-12.26) 11.77(9.79-13.93)

≥10 62/55 13.48(10.78-16.50) 9.34(6.86-12.20) 10.90(8.89-13.11)

Smoking Duration

0.17
Never 143/106 12.56(10.58-14.71) 9.10(6.13-12.66) 11.37(9.54-13.36)

<20 Years 33/40 11.87(9.63-14.34) 9.96(7.99-12.16) 9.86(7.43-12.64)

≥20 Years 87/83 10.68(7.71-14.13) 12.22(9.81-14.89) 10.90(3.07-23.55)

Pack Years

0.02
Never 143 / 106 12.49(10.52-14.63) 8.99(6.49-11.91) 11.53(9.93-13.24)

<10 50 / 49 12.46(10.02-15.17) 10.11(8.13-12.30) 11.00(8.87-13.35)

≥10 80/73 13.07(10.04-16.49) 10.59(8.23-13.23) 11.60(9.72-13.65)

1
Adjusted by age, race, BMI, post-menopausal hormone use at diagnosis, and physical activity level when not acting as a stratifying variable.
2
P-value comparing geometric mean of drinkers to non-drinkers.
3
Interaction is also tested between time since smoking stopped (p=.19).
133

Table 2, continued.

Post-Menopausal
Therapy
0.57
Never 86/68 11.88(9.46-14.57) 11.15(8.88-13.68) 10.83(8.97-12.88)

Estrogen Only 89/57 12.78(10.54-15.23) 9.66(7.34-12.29) 11.17(9.24-13.27)

Estrogen and Progestin 95/104 11.37(8.66-14.44) 11.94(9.78-14.32) 12.36(10.66-14.19)

Physical Activity

0.13
<13 MET hrs/week 128/74 10.95(9.02-13.07) 14.27(11.44-17.42) 10.68(9.09-12.39)

13-26 MET hrs/week 66/57 10.90(8.62-13.46) 10.37(8.15-12.87) 12.32(10.20-14.63)

>26 MET hrs/week 79/98 10.10(7.53-13.05) 11.63(9.41-14.09) 11.37(9.59-9.00)

BMI WHO
Categories
4

0.71
18-24.9 85/101 17.33(14.41-20.52) 10.00(7.88-12.37) 16.77(14.59-19.10)

25-29.9 93/67 8.16(6.31-10.24) 16.24(13.61-19.09) 10.09(8.33-12.03)

≥30 95/61 10.15(7.71-12.93) 6.74(4.74-9.10) 7.59(6.09-9.25)

Current Tamoxifen
Use
0.87
No 140/114 12.23(10.22-14.43) 11.52(9.59-13.63) 11.72(10.12-13.44)

Yes 133/115 11.12(9.11-13.34) 10.73(8.70-12.96) 11.16(9.57-12.86)

Parity

0.77
Nulliparous 32/39 13.42(9.76-17.65) 13.07(10.88-15.47) 12.44(9.90-15.27)

One or Two 120/108 10.39(8.47-12.51) 11.54(8.37-15.21) 12.34(10.59-14.23)

Three or More 121/82 11.49(9.32-13.89) 10.15(7.99-12.58) 10.28(8.68-12.01)

Age At First Birth

0.62
<18 33/10 11.29(7.91-15.28) 11.39(9.44-13.53) 10.33(7.42-13.73)

18-25 156/104 12.53(9.60-15.86) 7.66(3.30-13.84) 11.20(9.50-13.03)

≥25 51/76 10.87(8.69-13.29) 10.77(8.39-13.46) 11.50(9.46-13.73)

4
Covariates does not include BMI(continuous).
134

Table 3a. Linear regression models for mammographic density (continuous outcome on square root transformation) on alcohol (gm/day) or
type of (alcohol servings/day) after adjustment for covariates among those who drink.

β SE 95%CI p-value
Alcohol gm/day (n=229)

age, race

0.011 0.01 (-0.01, 0.03) 0.26
age, race, BMI, postmenopausal hormone use(PMH)

0.005 0.01 (-0.01, 0.02) 0.61
age, race, BMI, PMH, physical activity level

0.005 0.01 (-0.01, 0.02) 0.58
age, race, BMI, PMH, physical activity level, education
0.005 0.01 (-0.01, 0.02) 0.60
Beer servings/day (n=99)

age, race

0.209 0.23 (-0.24, 0.66) 0.36
age, race, BMI, PMH

0.147 0.22 (-0.29, 0.58) 0.50
age, race, BMI, PMH, physical activity level

0.147 0.22 (-0.29, 0.59) 0.51
age, race, BMI, PMH, physical activity level, education 0.141 0.22 (-0.30, 0.59) 0.53
Wine servings/day (n=190)

age, race

0.388 0.25 (-0.11, 0.88) 0.12
age, race, BMI, PMH

0.234 0.24 (-0.25, 0.71) 0.34
age, race, BMI, PMH, physical activity level

0.182 0.20 (-0.22, 0.58) 0.37

1
White wine (n=116): age, race, BMI, PMH, physical activity level

0.156 0.31 (-0.45, 0.77) 0.61
No White wine (n=70)

0.523 0.02 (-0.46, 1.51)

1
Pink wine (n=43): age, race, BMI, PMH, physical activity level

0.683 1.03 (-1.41, 2.78) 0.51
No Pink wine (n=143)

0.248 0.00 (-0.28, 0.78)

1
Red wine (n=95): age, race, BMI, PMH, physical activity level

0.247 0.40 (-0.54, 1.03) 0.53
No red wine (n=91)

0.259 0.49 (-0.41, 0.93)

age, race, BMI, PMH, physical activity level, education 0.241 0.24 (-0.24, 0.72) 0.33
Liquor servings/day (n=103)

age, race

-0.194 0.29 (-0.76, 0.37) 0.50
age, race, BMI, PMH

-0.202 0.26 (-0.71, 0.31) 0.43
age, race, BMI, PMH, physical activity level

-0.201 0.26 (-0.72, 0.31) 0.44
age, race, BMI, PMH, physical activity level, education

-0.204 0.26 (-0.72, 0.32) 0.44

1
187 women did not specify what type(s) of wine they consumed. Total exceeds 190 because most women consumed more than one type of wine.
135

Table 3b. Linear regression models for mammographic density (continuous outcome on square root transformation) on smoking variables
(exposure) on after adjustment for covariates only among smokers.

β SE 95%CI p-value
Pack Years

age, race

-0.0042 0.00 (-0.014, 0.000) 0.38
age, race, BMI

-0.0011 0.00 (-0.010, 0.005) 0.80
age, race, BMI, PMH

-0.0013 0.00 (-0.010, 0.008) 0.77
age, race, BMI, PMH, physical activity level -0.0011 0.00 (-0.010, 0.007) 0.80
age, race, BMI, PMH, physical activity level, education

-0.0023 0.00 (-0.012, 0.008) 0.64
Smoke Duration (Years)

age, race

0.0127 0.01 (-0.002, 0.007) 0.10
age, race, BMI

0.0101 0.01 (-0.004, 0.028) 0.15
age, race, BMI, PMH

0.0095 0.01 (-0.004, 0.024) 0.17
age, race, BMI, PMH, physical activity level 0.0101 0.01 (-0.004, 0.023) 0.15
age, race, BMI, PMH, physical activity level, education 0.0120 0.01 (-0.003, 0.024) 0.11
Cigarettes/Day

age, race

-0.0129 0.01 (-0.029, 0.027) 0.12
age, race, BMI

-0.0059 0.01 (-0.021, 0.004) 0.44
age, race, BMI, PMH

-0.0064 0.01 (-0.021, 0.009) 0.40
age, race, BMI, PMH, physical activity level -0.0064 0.01 (-0.021, 0.009) 0.40
age, race, BMI, PMH, physical activity level, education -0.0052 0.01 (-0.022, 0.009) 0.53

136

Table 4. Linear regression models for mammographic density (continuous outcome on square root
transformation) on smoking variables (exposure) on after adjustment for covariates (age, race, BMI,
PMH, physical activity level) and stratified by smoking status.

n β SE 95%CI p-value
Pack Years (Years)
Smoking Status
Smoking Status

Stopped ≥20 years ago 73
0.0008 0.01 (-0.02, 0.02) 0.94
Stopped <20 yrs ago 124
-0.0040 0.03 (-0.01, 0.01) 0.45
Current 56
-0.0094 0.01 (-0.05, 0.03) 0.62

Cigarettes/Day

Smoking Status

Stopped ≥20 years ago 73 -0.0290 0.02 (-0.06, 0.00) 0.07
Stopped <20 yrs ago 124 -0.0026 0.01 (-0.02, 0.02) 0.79
Current 56 -0.0181 0.04 (-0.10, 0.06) 0.64

Smoke Duration (Years)

Smoking Status

Stopped ≥20 years ago 73 0.0145 0.01 (-0.01, 0.04) 0.26
Stopped <20 yrs ago 124 -0.0007 0.01 (-0.02, 0.02) 0.95
Current 56 0.0122 0.03 (-0.05, 0.08) 0.71

137

Figure 2a. Linear regression coefficient of alcohol(gm/day) and 95% confidence intervals at
specific percentile of percent mammographic density (20%, 40%, 60%, 80%) in the
multivariable (adjusted) quantile regression model. The line connects the points for the linear
regression coefficients and the blue shading indicates the 95% confidence intervals when
restricting to drinkers (n=229).

138

Figure 2b. Linear regression coefficient of pack year and 95% confidence intervals at specific
percentile of percent mammographic density (20%, 40%, 60%, 80%) in the multivariable
(adjusted) quantile regression model. The line connects the points for the linear regression
coefficients and the blue shading indicates the 95% confidence intervals when restricting to ever
smokers (n=253).

139

4.6 References
1. Boyd NF, Guo H, Martin LJ, et al. Mammographic density and the risk and detection of
breast cancer. N Engl J Med 2007;356:227-36.
2. Cil T, Fishell E, Hanna W, et al. Mammographic density and the risk of breast cancer
recurrence after breast-conserving surgery. Cancer 2009;115:5780-7.
3. Habel LA, Dignam JJ, Land SR, et al. Mammographic density and breast cancer after
ductal carcinoma in situ. J Natl Cancer Inst 2004;96:1467-72.
4. Ursin G, Pike M. Mammographic density, hormone therapy, and risk of breast cancer.
Cancer Epidemiol Biomarkers Prev 2006;15:1750; author reply
5. Pike MC. The role of mammographic density in evaluating changes in breast cancer risk.
Gynecol Endocrinol 2005;21 Suppl 1:1-5.
6. Boyd NF, Martin LJ, Yaffe MJ, et al. Mammographic density: a hormonally responsive
risk factor for breast cancer. J Br Menopause Soc 2006;12:186-93.
7. Kelemen LE, Sellers TA, Vachon CM. Can genes for mammographic density inform
cancer aetiology? Nat Rev Cancer 2008;8:812-23.
8. Cuzick J, Warwick J, Pinney E, et al. Tamoxifen-induced reduction in mammographic
density and breast cancer risk reduction: a nested case-control study. J Natl Cancer Inst
2011;103:744-52.
9. McTiernan A, Martin CF, Peck JD, et al. Estrogen-plus-progestin use and
mammographic density in postmenopausal women: Women's Health Initiative
randomized trial. J Natl Cancer Inst 2005;97:1366-76.
10. Greendale GA, Palla SL, Ursin G, et al. The association of endogenous sex steroids and
sex steroid binding proteins with mammographic density: results from the
Postmenopausal Estrogen/Progestin Interventions Mammographic Density Study. Am J
Epidemiol 2005;162:826-34.
11. Vachon CM, Sellers TA, Vierkant RA, et al. Case-control study of increased
mammographic breast density response to hormone replacement therapy. Cancer
Epidemiol Biomarkers Prev 2002;11:1382-8.
12. Li J, Humphreys K, Eriksson L, et al. Mammographic Density Reduction Is a Prognostic
Marker of Response to Adjuvant Tamoxifen Therapy in Postmenopausal Patients With
Breast Cancer. J Clin Oncol 2013.
13. Johansson H, Gandini S, Bonanni B, et al. Relationships between circulating hormone
levels, mammographic percent density and breast cancer risk factors in postmenopausal
women. Breast Cancer Res Treat 2008;108:57-67.
14. Tamimi RM, Hankinson SE, Colditz GA, et al. Endogenous sex hormone levels and
mammographic density among postmenopausal women. Cancer Epidemiol Biomarkers
Prev 2005;14:2641-7.
15. Boyd NF, Stone J, Martin LJ, et al. The association of breast mitogens with
mammographic densities. Br J Cancer 2002;87:876-82.
16. Aiello EJ, Tworoger SS, Yasui Y, et al. Associations among circulating sex hormones,
insulin-like growth factor, lipids, and mammographic density in postmenopausal women.
Cancer Epidemiol Biomarkers Prev 2005;14:1411-7.
17. Hong J, Holcomb VB, Dang F, et al. Alcohol consumption, obesity, estrogen treatment
and breast cancer. Anticancer Res 2010;30:1-8.
18. Singletary KW, Gapstur SM. Alcohol and breast cancer: review of epidemiologic and
experimental evidence and potential mechanisms. JAMA 2001;286:2143-51.
140

19. Onland-Moret NC, Peeters PH, van der Schouw YT, et al. Alcohol and endogenous sex
steroid levels in postmenopausal women: a cross-sectional study. J Clin Endocrinol
Metab 2005;90:1414-9.
20. Mendelson CR, Evans CT, Simpson ER. Regulation of aromatase in estrogen-producing
cells. J Steroid Biochem 1987;27:753-7.
21. Etique N, Chardard D, Chesnel A, et al. Analysis of the effects of different alcohols on
MCF-7 human breast cancer cells. Ann N Y Acad Sci 2004;1030:78-85.
22. Eng ET, Williams D, Mandava U, et al. Anti-aromatase chemicals in red wine. Ann N Y
Acad Sci 2002;963:239-46.
23. Longnecker MP, Tseng M. Alcohol, hormones, and postmenopausal women. Alcohol
Health Res World 1998;22:185-9.
24. Purohit V. Can alcohol promote aromatization of androgens to estrogens? A review.
Alcohol 2000;22:123-7.
25. Vachon CM, Kuni CC, Anderson K, et al. Association of mammographically defined
percent breast density with epidemiologic risk factors for breast cancer (United States).
Cancer Causes Control 2000;11:653-62.
26. Masala G, Ambrogetti D, Assedi M, et al. Dietary and lifestyle determinants of
mammographic breast density. A longitudinal study in a Mediterranean population. Int J
Cancer 2006;118:1782-9.
27. Cabanes A, Pastor-Barriuso R, Garcia-Lopez M, et al. Alcohol, tobacco, and
mammographic density: a population-based study. Breast Cancer Res Treat
2011;129:135-47.
28. Flom JD, Ferris JS, Tehranifar P, et al. Alcohol intake over the life course and
mammographic density. Breast Cancer Res Treat 2009;117:643-51.
29. Herrinton LJ, Saftlas AF, Stanford JL, et al. Do alcohol intake and mammographic
densities interact in regard to the risk of breast cancer? Cancer 1993;71:3029-35.
30. Maskarinec G, Takata Y, Pagano I, et al. Alcohol consumption and mammographic
density in a multiethnic population. Int J Cancer 2006;118:2579-83.
31. Vachon CM, Kushi LH, Cerhan JR, et al. Association of diet and mammographic breast
density in the Minnesota breast cancer family cohort. Cancer Epidemiol Biomarkers Prev
2000;9:151-60.
32. Vachon CM, Sellers TA, Janney CA, et al. Alcohol intake in adolescence and
mammographic density. Int J Cancer 2005;117:837-41.
33. Gapstur SM, Potter JD, Drinkard C, et al. Synergistic effect between alcohol and estrogen
replacement therapy on risk of breast cancer differs by estrogen/progesterone receptor
status in the Iowa Women's Health Study. Cancer Epidemiol Biomarkers Prev
1995;4:313-8.
34. Qureshi SA, Couto E, Hofvind S, et al. Alcohol intake and mammographic density in
postmenopausal Norwegian women. Breast Cancer Res Treat 2011.
35. Roubidoux MA, Kaur JS, Griffith KA, et al. Relationship of mammographic
parenchymal patterns to breast cancer risk factors and smoking in Alaska Native women.
Cancer Epidemiol Biomarkers Prev 2003;12:1081-6.
36. Modugno F, Ngo DL, Allen GO, et al. Breast cancer risk factors and mammographic
breast density in women over age 70. Breast Cancer Res Treat 2006;97:157-66.
37. Jeffreys M, Warren R, Gunnell D, et al. Life course breast cancer risk factors and adult
breast density (United Kingdom). Cancer Causes Control 2004;15:947-55.
141

38. Gapstur SM, Lopez P, Colangelo LA, et al. Associations of breast cancer risk factors with
breast density in Hispanic women. Cancer Epidemiol Biomarkers Prev 2003;12:1074-80.
39. Cabanes A, Pastor-Barriuso R, García-López M, et al. Alcohol, tobacco, and
mammographic density: a population-based study. Breast Cancer Res Treat
2011;129:135-47.
40. Butler LM, Gold EB, Conroy SM, et al. Active, but not passive cigarette smoking was
inversely associated with mammographic density. Cancer Causes Control 2010;21:301-
11.
41. Bremnes Y, Ursin G, Bjurstam N, et al. Different measures of smoking exposure and
mammographic density in postmenopausal Norwegian women: a cross-sectional study.
Breast Cancer Res 2007;9:R73.
42. Sala E, Warren R, McCann J, et al. Smoking and high-risk mammographic parenchymal
patterns: a case-control study. Breast Cancer Res 2000;2:59-63.
43. Gavaler JS. Alcohol and nutrition in postmenopausal women. J Am Coll Nutr
1993;12:349-56.
44. Tanko LB, Christiansen C. An update on the antiestrogenic effect of smoking: a literature
review with implications for researchers and practitioners. Menopause 2004;11:104-9.
45. Johnson KC, Miller AB, Collishaw NE, et al. Active smoking and secondhand smoke
increase breast cancer risk: the report of the Canadian Expert Panel on Tobacco Smoke
and Breast Cancer Risk (2009). Tob Control 2011;20:e2.
46. Sala E, Warren R, Duffy S, et al. High risk mammographic parenchymal patterns and
diet: a case-control study. Br J Cancer 2000;83:121-6.
47. Brisson J, Verreault R, Morrison AS, et al. Diet, mammographic features of breast tissue,
and breast cancer risk. Am J Epidemiol 1989;130:14-24.
48. Gehm BD, McAndrews JM, Chien PY, et al. Resveratrol, a polyphenolic compound
found in grapes and wine, is an agonist for the estrogen receptor. Proc Natl Acad Sci U S
A 1997;94:14138-43.
49. Fontecave M, Lepoivre M, Elleingand E, et al. Resveratrol, a remarkable inhibitor of
ribonucleotide reductase. FEBS Lett 1998;421:277-9.
50. Shufelt C, Merz CN, Yang Y, et al. Red Versus White Wine as a Nutritional Aromatase
Inhibitor in Premenopausal Women. J Womens Health (Larchmt) 2011.
51. Band PR, Le ND, Fang R, et al. Carcinogenic and endocrine disrupting effects of
cigarette smoke and risk of breast cancer. Lancet 2002;360:1044-9.
52. Terry PD, Rohan TE. Cigarette smoking and the risk of breast cancer in women: a review
of the literature. Cancer Epidemiol Biomarkers Prev 2002;11:953-71.

Abstract (if available)

Abstract Increased mammographic density (MD) is strongly associated with increased risk of breast cancer incidence and worse prognosis. The purpose of this dissertation project includes three analyses that examine the associations between estrogen metabolism genes, biomarkers of inflammation, and lifestyle on mammographic density among postmenopausal breast cancer survivors from the Health, Eating, Activity, and Lifestyle (HEAL) study. ❧ Genetic variants influencing lifetime exposure to estrogen and progesterone have been considered as factors that influence mammographic density in postmenopausal women. This analysis examines the association between 3 SNPs (CYP1B1,Val423Leu and COMT,Val108/158Met) in the estrogen metabolism pathway, MTHFR(Ala222Val) which influences COMT activity, and percent mammographic density. Using linear regression models, we found a dose-response association between COMT and percent mammographic density (p-trend=0.02)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

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

PDF

Arm lymphedema in a multi-ethnic cohort of female breast cancer survivors

PDF

Identifying genetic, environmental, and lifestyle determinants of ethnic variation in risk of pancreatic cancer

PDF

Associations between physical activities with bone mineral density in postmenopausal women

PDF

Genes and hormonal factors involved in the development or recurrence of breast cancer

PDF

Predictive factors of breast cancer survival: a population-based study

PDF

Lifetime physical activity and its effects on breast cancer survival

PDF

DEXA measures of body fat percentage and acute phase proteins among breast cancer survivors

PDF

The role of inflammation in non-Hodgkin lymphoma etiology

PDF

Hormone therapy timing hypothesis and atherosclerosis

PDF

The effects of hormonal exposures on ovarian and breast cancer risk

PDF

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

PDF

Red and processed meat consumption and colorectal cancer risk: meta-analysis of case-control studies

PDF

The role of pesticide exposure in breast cancer

PDF

Genetic risk factors in multiple myeloma

PDF

Body size and the risk of prostate cancer in the multiethnic cohort

PDF

Genomic risk factors associated with Ewing Sarcoma susceptibility

PDF

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

PDF

The role of steroid hormones in the etiology of urologic diseases

Asset Metadata

Creator Dee, Anne (author)

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

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Epidemiology

Publication Date 07/24/2013

Defense Date 05/28/2013

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Alcohol,breast cancer,Inflammation,mammographic density,OAI-PMH Harvest,postmenopausal women,Smoking

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Mack, Wendy Jean (committee chair), McKean-Cowdin, Roberta (committee chair), Bernstein, Leslie (committee member), Press, Michael F. (committee member), Wu, Anna (committee member)

Creator Email anne.dee@med.usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-299052

Unique identifier UC11294973

Identifier etd-DeeAnne-1841.pdf (filename),usctheses-c3-299052 (legacy record id)

Legacy Identifier etd-DeeAnne-1841.pdf

Dmrecord 299052

Document Type Dissertation

Format application/pdf (imt)

Rights Dee, Anne

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA