Cancer Epidemiology 38 (2014) 376381

Contents lists available at ScienceDirect

Cancer Epidemiology
The International Journal of Cancer Epidemiology, Detection, and Prevention

journal homepage: www.cancerepidemiology.net

Active cigarette smoking and the risk of breast cancer: a cohort study
Chelsea Catsburg a, Victoria A. Kirsh b,c, Colin L. Soskolne d,e, Nancy Kreiger b,c,
Thomas E. Rohan a,*
a
Department of Epidemiology and Population Health, Albert Einstein College of Medicine, Bronx, NY, United States
b
Prevention and Cancer Control, Cancer Care Ontario, Toronto, ON, Canada
c
Dalla Lana School of Public Health, University of Toronto, Toronto, ON, Canada
d
University of Alberta, Edmonton, AB, Canada
e
Faculty of Health, University of Canberra, ACT, Australia

A R T I C L E I N F O A B S T R A C T

Article history: Background: Tobacco use has been implicated in the etiology of a large number of cancers, and there
Received 3 February 2014 exists substantial biological plausibility that it could also be involved in breast carcinogenesis. Despite
Received in revised form 1 April 2014 this, epidemiological evidence to date is inconsistent. The aim of this study was to investigate the role of
Accepted 19 May 2014
active smoking and the risk of incident, invasive breast cancer using a prospective cohort of women from
Available online 12 June 2014
the Canadian Study of Diet, Lifestyle and Health.
Methods: Using a case-cohort design, an age-stratied subcohort of 3314 women was created from
Keywords:
39,532 female participants who returned completed self-administered lifestyle and dietary ques-
Breast cancer
Smoking
tionnaires at baseline. A total of 1096 breast cancer cases were identied in the entire cohort (including
Case-cohort 141 cases from the subcohort) by linkage to the Canadian Cancer Registry. Cox regression models were
Epidemiology used to estimate hazard ratios for the association between the different smoking exposures and the risk
Risk factors of breast cancer, using a modication for the case-cohort design.
Results: After carefully considering early-life exposures and potential confounders, we found no
association between any smoking exposure and risk of breast cancer in this study (Hazard ratio = 1.00,
95% condence interval = 0.871.17 for ever vs never smokers).
Conclusions: Although these results cannot rule out an association between smoking and breast cancer,
they do agree with the current literature suggesting that, if an association does exist, it is relatively weak.
2014 Elsevier Ltd. All rights reserved.

1. Introduction N-nitrosamines [3]. Mammary epithelial cells can subsequently

Tobacco consumption is the single greatest avoidable risk factor
for cancer and is estimated to be causal for 21% of all cancer
mortality worldwide [1]. In addition to the well-established effects
of tobacco smoke on lung carcinogenesis, there is also strong
evidence that cigarette smoking is associated with increased risk of
many other cancers, including laryngeal, bladder, bowel, kidney,
cervix, esophageal and gastric, indicating that it has extensive
systemic effects [2]. Breast cancer has also been implicated
because mammary tissue is capable of uptake of many tobacco
carcinogens routinely found in systemic circulation, including
polycyclic aromatic hydrocarbons, aromatic amines and
* Corresponding author at: Department of Epidemiology and Population Health,
Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY 10461,
United States. Tel.: +1 718 430 3355; fax: +1 718 430 8653.
E-mail address: Thomas.Rohan@einstein.yu.edu (T.E. Rohan).
metabolize and activate these compounds into electrophilic
intermediates capable of DNA damage and adduct formation
[4,5]. Smokers have a higher prevalence of these tobacco-related
DNA adducts and also p53 gene mutations than non-smokers,
and genomic alterations of mammary epithelial cells exposed
to tobacco carcinogens begin to resemble those seen in familial
breast cancer [6,7]. Moreover, substantial evidence exists for
tobacco-induced breast tumorigenesis in animal models [3].
Despite the plausible biological association between tobacco
smoke and breast cancer risk, epidemiological evidence has
remained inconsistent. In 2004, both the International Agency
for Research on Cancer (IARC) [8] and the US Surgeon General [9]
concluded that the evidence suggested no causal relationship
between active smoking and breast cancer, based on studies
published up until 2002. More recently however, evidence seems
to suggest a weak positive association, and in contrast to these
previous reports, the IARC Monographs were updated in 2012 to
state that a positive association has been observed between

http://dx.doi.org/10.1016/j.canep.2014.05.007
1877-7821/ 2014 Elsevier Ltd. All rights reserved.
 C. Catsburg et al. / Cancer Epidemiology 38 (2014) 376381
tobacco smoking and cancer of the female breast, based on
studies published through 2009 [2]. In agreement with this, the
most recent report from the US Surgeon General states that a
history of ever smoking is consistent with an increase of
relative risk for breast cancer of an estimated 10% [10]. Reviews
of the association between smoking and the risk of breast cancer
have identied many methodological issues that may have
contributed to discrepancies between previous published studies,
either masking a small existing association or spuriously creating
one [1113].
Problems in quantifying risk can arise from the misclassica-
tion of exposure. Indeed, previous studies have not been uniform in
their consideration of duration and intensity of smoking. In
addition, as with many risk factors related to breast cancer
etiology, the timing of exposure may be important. Breast tissue
is most susceptible to chemically-induced carcinogenesis prior
to full differentiation, which occurs at the completion of a womans
rst full-term pregnancy [14]. Thus, age at initiation of smoking,
total duration of smoking, and years smoked prior to rst
pregnancy have been implicated in breast cancer risk; however,
these are all highly correlated with birth cohort and attained age.
Moreover, the duration of smoking pre-pregnancy is associated
with age at rst pregnancy, which itself is a risk factor for
breast cancer [15].
The association between smoking and breast cancer can also be
affected by other confounding factors. In particular, an important
potential confounder of this association is alcohol intake [16].
Alcohol is an established risk factor for breast cancer and is also
strongly correlated with smoking [17]. In addition, it has been
shown that smokers are less likely to subject themselves to
mammograms than non-smokers and, in particular, former
smokers, thus creating a potential under-estimation of breast
cancer incidence in current smokers [18].
In the present study, we investigated the role of active
smoking in the risk of breast cancer using a prospective cohort
of Canadian women. We used case-cohort analyses to evaluate
associations and were able to consider estimates of early life
exposure while also examining confounding by a history of
alcohol intake, reproductive variables, and mammography
screening.
2. Materials and methods
2.1. Study population
The Canadian Study of Diet, Lifestyle and Health (CSDLH) is a
prospective cohort study described elsewhere [19]. Briey, the
study recruited 73,909 Canadian male and female participants,
predominantly from alumni of the Universities of Alberta,
Toronto and Western Ontario between 1995 and 1998. A small
contingent was also recruited through the Canadian Cancer
Society, mostly in 1992. At recruitment, participants completed
detailed self-administered dietary and lifestyle questionnaires.
Incident cases of breast cancer (and other cancers) were
ascertained by means of computerized record linkage (using
identifying information provided by the study participants for
this purpose) to the Canadian Cancer Registry (CCR) and to the
Ontario Cancer Registry. The CCR is a collaborative effort
between the thirteen Canadian provincial and territorial cancer
registries and the Health Statistics Division of Statistics Canada.
Because each Canadian province and territory has a legislated
responsibility for cancer collection and control, reporting is
virtually complete (estimated at 97%) [20]. Migration out of
Canada was very low during this period (estimated to be less than
0.005%) [21]. Deaths from all causes were ascertained by means of
record linkage to the National Mortality Database, and this
377
information was used as a censoring variable. Follow-up was
continued through December 31st 2010 for participants who were
resident in Ontario and through December 31st 2005 for
participants residing in all other provinces.
2.2. Study design
The analysis reported here was conducted using a case-cohort
design. The motivation for using a case-cohort approach was that
where the outcomes of interest are relatively rare, as is the case
here, this design represents a very cost-effective approach and
loses little efciency compared to a full cohort analysis [22,23]. The
study included a total of 1096 incident, invasive breast cancer
cases and a subcohort of 3314 women, which was created by
selecting an age-stratied random sample of the entire female
cohort (N = 39,532) at baseline. Given that we were to make
comparisons between this selected subcohort and cancer cases,
the subcohort was sampled with more weight for older partici-
pants. Thus, each 5-year age group had a different sampling
fraction that increased with age, thereby attempting to approxi-
mate the anticipated distribution of age at diagnosis for all
incident cancers. All women in the subcohort were followed
through either date of death, date of diagnosis (for cases that arose
in the subcohort, n = 141), or the censoring administrative date
(December 31st 2010 for residents of Ontario and December 31st
2005 for residents of all other provinces).
2.3. Cigarette smoking and other known confounding variables
The self-administered lifestyle questionnaire solicited infor-
mation on a variety of demographic, lifestyle and social factors
including height, weight, race/ethnicity, education, physical
activity, and medical history including a detailed reproductive
history section. Detailed tobacco use was queried in the lifestyle
questionnaire, including questions relating to current smoking
status, age at starting smoking, and age at cessation, extent of
inhalation, number of cigarettes per day (allowing for changes in
number of cigarettes during different time periods), and cigar
and pipe smoking. Cigarette smoking exposures evaluated in this
study were as follows: smoking status (never, ever, former,
current), smoking duration in years (none, <10, 10<20, 20<30,
30<40, 40+), smoking intensity in cigarettes per day (none, <5, 5
<10, 10<15, 15<20, 20+), total pack-years of smoking (none,
<10, 10<20, 20<30, 30+), age at starting smoking in years
(never, <15, 15<20, 20<25, 25+), and number of years
smoked prior to rst full term pregnancy (calculated among
women with at least one full-term pregnancy by subtracting age
at starting smoking from age at rst full term pregnancy;
categorized as none, <5, 5<10, 10<15, 15+).
2.4. Statistical analysis
Seven participants (one case and six controls) were missing
smoking information and were thus excluded, leaving 1096 cases
and 3314 subcohort members contributing to these analyses.
Hazard ratios for the association between the different smoking
exposures and risk of breast cancer were evaluated via Cox
regression models with time on study as the time scale, using a
modication for case-cohort analysis as described by Langholz
and Jiao [24]. Models were adjusted for known breast cancer risk
factors selected a priori. These included birth cohort (pre-1915,
19151924, 19251934, 19351944, 19451954, 19551964,
post-1964) [25], age (years) at menarche (12, 13, 14), use of
oral contraceptives (never/ever), use of hormone therapy (never,
estrogen only, estrogen plus progestin, missing or unspecied),
number of live births (nulliparous or missing, 12, 3), age (years)
378 C. Catsburg et al. / Cancer Epidemiology 38 (2014) 376381

at rst live birth (<25, 2529, 30, nulliparous or missing), family Table 1
Distribution of selected baseline characteristics in cases and women in the
history of breast cancer (yes/no), menopausal status at baseline,
subcohort.
routine mammography (yes/no), body mass index (BMI) in kg/m2,
physical activity in MET (metabolic equivalent) hours per week, Variable Cases Subcohort (n = 3314)
(n = 1096)
and alcohol intake in grams per day. Never smokers were used as
the reference group for all analyses with the exception of years Lifestyle Variables, median (IQR)
smoked before rst full-term pregnancy, where the reference Age, years 52.8 (44.564.5) 59.9 (48.971.7)
Follow-up, years 6.9 (3.710.5) 12.6 (10.315.7)
group was women with at least one full-term pregnancy and who
BMI, kg/m2 23.8 (21.726.4) 23.8 (21.726.2)
were never smokers prior to the birth of their rst child. Physical Activity, MET hrs/wk 13.0 (5.225.5) 13.3 (3.926.5)
Associations were assessed among the entire subcohort and Alcohol Intake, gpd 4.0 (1.014.1) 4.0 (014.1)
also stratied by menopausal status at baseline. Analyses were Reproductive and Lifestyle Variables, n (%)
Race White 1044 (95.3) 3173 (95.8)
also conducted stratifying by alcohol use (never/ever), routine
Education
mammography screening (yes/no; in the full cohort and restricted High school or less 23 (2.1) 87 (2.6)
to those 50 years), family history (yes/no), and birth cohort (pre Post-secondary/some college 794 (72.5) 2485 (75.0)
1935/post 1935). In addition, we considered years smoked before Graduate school 279 (25.5) 742 (22.4)
rst full-term pregnancy stratied by age (years) at rst History of diabetes 25 (2.3) 110 (3.4)
Family history of breast cancer 151 (13.8) 346 (10.4)
pregnancy (<30/30) and age (years) at starting smoking (<18/
Age at menarche, years
18). We also conducted a sensitivity analysis excluding never 12 479 (43.7) 1431 (43.2)
smokers and setting the lowest exposure of each category as the 13 385 (35.1) 1042 (31.4)
reference group. 14 223 (20.4) 803 (24.2)
Missing 9 (0.8) 38 (1.2)
Menopausal status
3. Results Pre-menopause 555 (50.6) 1107 (33.4)
Post-menopause 541 (49.4) 2207 (66.6)
Table 1 presents lifestyle and reproductive characteristics of all Parity
incident breast cancer cases (n = 1096) and women in the Nulliparous/missing 307 (28.0) 962 (29.0)
12 458 (41.8) 1161 (35.0)
subcohort (n = 3314) included in this analysis. Of the 3314 women
3+ 331 (30.2) 1191 (35.9)
in the original subcohort, a total of 141 (4.3%) participants became Age at rst childbirth, years
breast cancer cases during follow-up. These women were included <25 198 (18.1) 579 (17.5)
in the characteristics of both the subcohort and the cases as they 2529 342 (31.2) 1120 (33.8)
30 259 (23.6) 659 (19.9)
contributed person-years of observation to the subcohort up until
Nulliparous/missing 297 (27.1) 956 (28.9)
the point that they became cases. Overall, cases were younger, Breastfeed ever 678 (61.9) 1994 (60.2)
were followed for an average of 5.7 years less, and were more likely Oral contraceptive use ever 745 (68.0) 1875 (56.6)
to have a family history of breast cancer than members of the Routine mammogram yes 674 (61.5) 1958 (59.1)
subcohort. Cases were also more likely to have used oral Hormone therapy use
Never 736 (67.2) 2067 (62.4)
contraceptives and estrogen and progestin hormone therapy.
Estrogen only 142 (13.0) 607 (18.3)
Regarding reproductive factors, compared to women in the Estrogen + progestin 202 (18.4) 530 (16.0)
subcohort, cases were more likely to be pre-menopausal, have Missing/Unspecied 16 (1.5) 110 (3.3)
an earlier age of menarche, have fewer children, and be older at Smoking status
the birth of their rst child (Table 1). Never 606 (55.3) 1808 (54.6)
Former 412 (37.6) 1283 (38.7)
After adjusting for potential confounders (see Section 2), none Current 78 (7.1) 223 (6.7)
of the smoking variables considered in this study was associated
with risk of incident breast cancer in the CSDLH cohort (Table 2).
Specically, among all participants, for neither current smokers
(HR = 1.04, 95% CI = 0.781.39; unadjusted HR = 1.05, 95%
CI = 0.791.38) nor former smokers (HR = 1.00, 95% CI = 0.86 as the referent group (Supplementary Table 1). There was some
1.17; unadjusted HR = 1.01, 95% CI = 0.871.17) was there a evidence of an increasing risk of breast cancer with smoking
difference in risk compared to that for never smokers. Further- duration and pack-years of smoking among those with a family
more, there were no signicant associations seen when restricting history of breast cancer, although numbers were very small in this
the analyses to either pre-menopausal women only, or to post- subgroup and the results were not statistically signicant
menopausal women only (Table 2). (Supplementary Table 2).
When considered individually, alcohol consumption and
mammography screening did not confound the association 4. Discussion
between smoking variables and breast cancer in this study (data
not shown). Stratication by alcohol consumption and mammog- We report here a null association between cigarette smoking
raphy screening did not substantially change estimates for any of status, duration, and intensity in relation to the risk of incident
the smoking variables, and on formal testing, there was no breast cancer in this prospective cohort of Canadian women. These
statistical evidence of effect-modication by menopausal status ndings are consistent with the results from the majority of
(Table 2), alcohol use (never/ever) (Table 3), routine mammogra- prospective studies of cigarette smoking and breast cancer risk to
phy screening in the full cohort (yes/no) (Table 3) or routine date, which have predominantly observed either a weak or null
mammography screening in those 50 years of age and over (data association with smoking status, and little or no doseresponse
not shown). We also found no substantial differences when relationship for either duration or intensity of smoking
stratifying by birth cohort, BMI, or when considering [2,10,11,2631].
years smoked prior to rst pregnancy stratied by age at rst The most convincing evidence from the recent literature
pregnancy or by age at starting smoking (data not shown). supporting an association between tobacco use and breast cancer
Magnitudes of association were very similar when we excluded risk is the relatively consistent positive association observed
never smokers and used the lowest category of smoking exposure in women who initiate smoking at a younger age, or who smoke
 C. Catsburg et al. / Cancer Epidemiology 38 (2014) 376381 379

Table 2
Cigarette smoking and risk of breast cancer, overall and stratied by menopausal status.

All participants Pre-menopausal Post-menopausal

a a
Cases HR (95% CI) Cases HR (95% CI) Cases HRa (95% CI)
Ref. Ref.
Never smokers 606 1.0 328 1.0 278 1.0Ref.

Smoking status
Ever 490 1.00 (0.871.17) 227 0.99 (0.791.23) 263 1.02 (0.831.24)
Former 412 1.00 (0.861.17) 191 1.00 (0.791.26) 221 0.99 (0.811.22)
Current 78 1.04 (0.781.39) 36 0.94 (0.611.45) 42 1.18 (0.811.74)

Smoking duration (years)

<10 135 0.94 (0.751.17) 77 0.90 (0.661.22) 58 0.97 (0.701.35)
10<20 139 1.16 (0.921.46) 81 1.16 (0.831.61) 58 1.14 (0.821.59)
20<30 90 0.91 (0.701.19) 44 1.03 (0.681.57) 46 0.80 (0.561.16)
30<40 79 1.15 (0.861.53) 21 0.84 (0.481.45) 58 1.36 (0.971.90)
40+ 47 0.87 (0.611.24) 4 0.93 (0.322.71) 43 0.89 (0.611.30)
ptrend 0.98 0.98 0.88

Smoking intensity (cigarettes per day)

<5 129 0.89 (0.711.12) 61 0.76 (0.551.07) 68 1.04 (0.761.42)
5<10 87 1.02 (0.781.34) 47 1.23 (0.821.84) 40 0.82 (0.561.19)
10<15 89 1.12 (0.851.47) 37 1.19 (0.771.85) 52 1.16 (0.811.65)
15<20 68 1.06 (0.781.44) 33 1.02 (0.641.63) 35 1.08 (0.711.64)
20+ 74 0.96 (0.711.29) 32 0.92 (0.571.48) 42 1.00 (0.681.46)
ptrend 0.74 0.80 0.82

Pack-years
<10 238 0.96 (0.801.15) 128 0.93 (0.721.21) 110 0.97 (0.751.26)
10<20 104 1.18 (0.901.53) 51 1.28 (0.851.93) 53 1.08 (0.761.53)
20<30 46 0.87 (0.611.23) 18 0.74 (0.421.32) 28 0.95 (0.601.50)
30+ 59 1.04 (0.751.44) 13 0.95 (0.461.95) 46 1.11 (0.771.61)
ptrend 0.86 0.87 0.60

Age started smoking (years)

<15 61 1.03 (0.751.43) 42 1.08 (0.701.64) 19 0.97 (0.571.66)
15<20 278 1.00 (0.841.19) 142 0.99 (0.761.28) 136 1.00 (0.791.27)
20<25 105 0.94 (0.741.20) 33 0.84 (0.551.30) 72 0.97 (0.721.32)
25+ 46 1.21 (0.841.74) 10 1.17 (0.532.57) 36 1.27 (0.851.91)
ptrend 0.98 0.96 0.99

Years smoked before rst full-term pregnancyb

Non-smokers/0 427/15 1.0Ref. 225/1 1.0Ref. 202/14 1.0Ref.
<5 51 1.17 (0.811.69) 18 1.41 (0.702.85) 33 1.04 (0.671.62)
5<10 132 1.04 (0.821.32) 60 1.11 (0.761.63) 72 1.00 (0.731.38)
10<15 101 0.80 (0.611.05) 49 0.69 (0.461.01) 52 0.95 (0.661.38)
15+ 68 1.00 (0.701.42) 38 0.74 (0.461.20) 30 1.37 (0.812.34)
ptrend 0.72 0.23 0.48
a
Models adjusted for covariates collected at baseline as follows: birth cohort (pre-1915, 19151924, 19251934, 19351944, 19451954, 19551964, post-1964), age at
menarche in years (12, 13, 14, missing), use of oral contraceptives (never/ever), use of hormone therapy (never, estrogen only, estrogen plus progestin, missing or
unspecied), number of live births (nulliparous or missing, 12, 3), age at rst live birth in years (<25, 2529, 30, nulliparous or missing), family history of breast cancer
(yes/no), routine mammogram (yes/no), menopausal status, BMI (kg/m2), physical activity (MET hours/week), and alcohol intake (gpd).
b
Among women with at least one full-term pregnancy.

for a long time prior to their rst pregnancy. These ndings were
most evident in three recent large prospective cohorts: the Nurses
Health Study (NHS) [31], the Womens Health Initiative Observa-
tional Study (WHI-OS) [30] and the Cancer Prevention Study II
Nutrition Cohort (CPS-II) [29]. All three of these studies found years
smoked before pregnancy to be the largest smoking-related
predictor of breast cancer risk with statistically signicant hazard
ratio estimates ranging from 1.25 (NHS) to 1.45 (CPS-II). Although
we had detailed information on these two early life exposures in
this study, we did not nd either of them to be associated with
breast cancer risk.
One explanation for a consistently null or weak association
between smoking and breast cancer despite biological plausibil-
ity could be the potential anti-estrogenic effects of cigarette
smoke. Smoking has been shown to have protective effects for
endometrial cancer, another cancer known to be strongly
associated with estrogen levels [32]. Similarly, with breast
cancer, the anti-estrogenic effects of smoking may mask its
potential carcinogenic effects [33]. If this were true, it would
suggest that evidence of the underlying association may be seen
only among women who are genetically less capable of
detoxifying tobacco carcinogens, for example carriers of the
NAT2 slow acetylator phenotype. This hypothesis is supported
by several meta-analyses indicating that women classied as
slow acetylators may have up to a 50% increased risk of breast
cancer [3436].
One limitation of our study was that we did not have data on
environmental tobacco exposure and thus we were unable to
assess this association, or exclude those who were exposed to
environmental tobacco smoke from the reference group. If
breast cancer risk is elevated in those with environmental tobacco
exposure, bias is inevitable and inclusion of women with this
exposure in the reference group would attenuated the associations
[37]. Moreover, although our study was large, we lacked statistical
power to detect very modest associations (RR < 1.2), especially
where numbers were limited in stratied analyses. Strengths of
this study were the sample size, resulting in a large number of
breast cancer cases, the prospective design, limiting the effects
of recall bias, and the detailed information collected on both the
exposure and potential confounders.
380 C. Catsburg et al. / Cancer Epidemiology 38 (2014) 376381

Table 3
Cigarette smoking and risk of breast cancer stratied by alcohol use and mammography screening.

Alcohol use Routine mammography

Never Ever Yes No

Cases HRa (95% CI) Cases HRa (95% CI) Cases HRa (95% CI) Cases HRa (95% CI)

Never smokers 187 1.0Ref. 419 1.0Ref. 363 1.0Ref. 243 1.0Ref.

Smoking status
Ever 65 0.89 (0.641.24) 425 1.04 (0.891.23) 311 1.01 (0.841.21) 179 1.00 (0.791.27)
Former 55 0.89 (0.631.27) 357 1.03 (0.871.22) 269 1.01 (0.831.22) 143 0.98 (0.761.26)
Current 10 0.87 (0.401.85) 68 1.10 (0.811.50) 42 0.99 (0.671.46) 36 1.10 (0.721.68)

Smoking duration (years)

<10 20 0.88 (0.521.51) 115 0.95 (0.741.22) 77 0.99 (0.741.33) 58 0.87 (0.621.22)
10<20 15 0.77 (0.421.43) 124 1.26 (0.981.61) 86 1.21 (0.901.62) 53 1.08 (0.761.55)
20<30 17 1.22 (0.662.27) 73 0.88 (0.651.18) 61 0.94 (0.681.31) 29 0.85 (0.541.36)
30<40 8 0.90 (0.402.04) 71 1.22 (0.901.65) 59 1.10 (0.791.53) 20 1.37 (0.792.35)
40+ 5 0.65 (0.251.73) 42 0.94 (0.651.36) 28 0.70 (0.451.09) 19 1.37 (0.792.39)
ptrend 0.57 0.64 0.53 0.30

Smoking intensity (cigarettes per day)

<5 19 0.87 (0.501.52) 106 0.90 (0.701.16) 78 0.94 (0.711.25) 47 0.82 (0.571.18)
5<10 9 1.02 (0.482.15) 77 1.04 (0.781.39) 58 1.06 (0.761.47) 28 0.96 (0.601.53)
10<15 13 1.10 (0.562.15) 74 1.15 (0.861.54) 62 1.25 (0.901.73) 25 0.88 (0.551.42)
15<20 7 0.87 (0.362.10) 59 1.11 (0.801.54) 44 0.98 (0.671.44) 22 1.23 (0.732.07)
20+ 14 0.92 (0.481.77) 59 0.99 (0.711.38) 40 0.79 (0.541.17) 33 1.29 (0.812.04)
ptrend 0.62 0.49 0.71 0.30

Pack-years
<10 32 0.84 (0.551.31) 206 0.99 (0.811.21) 147 1.00 (0.791.26) 91 0.89 (0.671.19)
10<20 14 1.28 (0.652.51) 90 1.19 (0.901.57) 75 1.32 (0.961.81) 29 0.94 (0.591.48)
20<30 8 1.03 (0.432.45) 38 0.86 (0.581.26) 30 0.77 (0.501.18) 16 1.16 (0.622.15)
30+ 9 0.95 (0.442.04) 50 1.09 (0.761.55) 36 0.86 (0.571.29) 23 1.60 (0.942.73)
ptrend 0.94 0.71 0.49 0.17

Age started smoking (years)

<15 10 0.87 (0.401.88) 51 1.11 (0.781.58) 33 1.11 (0.721.72) 28 0.95 (0.591.53)
15<20 37 0.95 (0.621.45) 241 1.02 (0.841.24) 178 0.99 (0.801.23) 100 1.02 (0.761.35)
20<25 14 0.91 (0.481.70) 91 0.96 (0.731.25) 77 1.02 (0.761.37) 28 0.78 (0.501.23)
25+ 4 0.55 (0.191.62) 42 1.39 (0.942.04) 23 0.98 (0.601.61) 23 1.59 (0.932.71)
ptrend 0.58 0.67 0.88 0.88

Years smoked before rst full-term pregnancyb

Non-smokers/0 119/1 1.0Ref. 308/14 1.0Ref. 263/9 1.0Ref. 164/6 1.0Ref.
<5 8 1.43 (0.623.30) 43 1.14 (0.761.70) 35 1.11 (0.721.71) 16 1.38 (0.732.64)
5<10 16 1.02 (0.551.91) 116 1.06 (0.821.37) 97 1.15 (0.871.53) 35 0.84 (0.541.30)
10<15 11 0.72 (0.361.46) 90 0.82 (0.621.09) 69 1.02 (0.731.42) 32 0.54 (0.350.84)
15+ 10 1.10 (0.482.53) 58 1.00 (0.691.46) 37 1.12 (0.711.78) 31 0.87 (0.531.43)
ptrend 0.85 0.71 0.76 0.20
a
Models adjusted for covariates collected at baseline as follows: birth cohort (pre-1915, 19151924, 19251934, 19351944, 19451954, 19551964, post-1964), age at
menarche in years (12, 13, 14, missing), use of oral contraceptives (never/ever), use of hormone therapy (never, estrogen only, estrogen plus progestin, missing or
unspecied), number of live births (nulliparous or missing, 12, 3), age at rst live birth in years (<25, 2529, 30, nulliparous or missing), family history of breast cancer
(yes/no), routine mammogram (yes/no), menopausal status, BMI (kg/m2), physical activity (MET hours/week), and alcohol intake (gpd).
b
Among women with at least one full-term pregnancy.

In summary, after carefully considering early life exposures,
and evaluating important confounders and effect-modiers, we
found no association with any of the cigarette smoking exposure
factors considered in this study in relation to the risk of
breast cancer in a prospective cohort of Canadian women.
Although these results cannot rule out an association between
smoking and breast cancer, they do support current literature
suggesting that, if an association does exist, it is relatively weak.
Funding
This manuscript was supported in part by the Breast Cancer
Research Fund.
Conict of interest
The authors declare that they have no conict of interest.
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.canep.2014.05.
007.
References
[1] Ott JJ, Ullrich A, Mascarenhas M, Stevens GA. Global cancer incidence and
mortality caused by behavior and infection. J Public Health (Oxf)
2011;33(2):22333.
[2] International Agency for Research on Cancer. IARC monographs on the evalu-
ation of carcinogenic risks to humans: volume 100E tobacco smoking. Lyon,
France: IARC Press, 2012.
[3] Hecht SS. Tobacco smoke carcinogens and breast cancer. Environ Mol Mutagen
2002;39(23):11926.
[4] MacNicoll AD, Easty GC, Neville AM, Grover PL, Sims P. Metabolism and
activation of carcinogenic polycyclic hydrocarbons by human mammary cells.
Biochem Biophys Res Commun 1980;95(4):1599600.
[5] Li D, Wang M, Firozi PF, Chang P, Zhang W, Baer-Dubowska W, et al. Charac-
terization of a major aromatic DNA adduct detected in human breast tissues.
Environ Mol Mutagen 2002;39(23):193200.
 C. Catsburg et al. / Cancer Epidemiology 38 (2014) 376381
[6] Conway K, Edmiston SN, Cui L, Drouin SS, Pang J, He M, et al. Prevalence and
spectrum of p53 mutations associated with smoking in breast cancer. Cancer
Res 2002;62(7):198790.
[7] Li D, Zhang W, Sahin AA, Hittelman WN. DNA adducts in normal tissue
adjacent to breast cancer: a review. Cancer Detect Prev 1999;23(6):45462.
[8] International Agency for Research on Cancer. IARC monographs on the evalu-
ation of carcinogenic Risks to humans: volume 83 tobacco smoke and
involuntary smoking. Lyon, France: IARC Press, 2004.
[9] Ofce of the Surgeon General. The consequences of smoking: a report of the
surgeon general. Rockville, MD: United States Public Health Service, 2004.
[10] Alberg AJ, Shopland DR, Cummings KM. The 2014 surgeon generals report:
commemorating the 50th anniversary of the 1964 report of the advisory
committee to the US surgeon general and updating the evidence on the health
consequences of cigarette smoking. Am J Epidemiol 2014;179(4):40312.
[11] Reynolds P. Smoking and breast cancer. J Mammary Gland Biol Neoplasia
2013;18(1):1523.
[12] Terry PD, Rohan TE. Cigarette smoking and the risk of breast cancer in women:
a review of the literature. Cancer Epidemiol Biomark Prev 2002;11(10 Pt
1):95371.
[13] Johnson KC, Miller AB, Collishaw NE, Palmer JR, Hammond SK, Salmon AG,
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(1):e2.
[14] Russo J, Mailo D, Hu YF, Balogh G, Sheriff F, Russo IH. Breast differentiation and
its implication in cancer prevention. Clin Cancer Res 2005;11(2 Pt 2):931s6s.
[15] Kobayashi S, Sugiura H, Ando Y, Shiraki N, Yanagi T, Yamashita H, et al.
Reproductive history and breast cancer risk. Breast Cancer 2012;19(4):3028.
[16] Hamajima N, Hirose K, Tajima K, Rohan T, Calle EE, Heath Jr CW, et al. Alcohol,
tobacco and breast cancercollaborative reanalysis of individual data from 53
epidemiological studies, including 58,515 women with breast cancer and
95,067 women without the disease. Br J Cancer 2002;87(11):123440.
[17] Thomson CA. Diet and breast cancer: understanding risks and benets. Nutr
Clin Pract 2012;27(5):63650.
[18] Gross CP, Filardo G, Singh HS, Freedman AN, Farrell MH. The relation between
projected breast cancer risk, perceived cancer risk, and mammography use,
results from the National Health Interview Survey. J Gen Intern Med
2006;21(2):15864.
[19] Rohan TE, Soskolne CL, Carroll KK, Kreiger N. The Canadian study of diet,
lifestyle, and health: design and characteristics of a new cohort study of cancer
risk. Cancer Detect Prev 2007;31(1):127.
[20] Hall S, Schulze K, Groome P, Mackillop W, Holowaty E. Using cancer registry
data for survival studies: the example of the Ontario Cancer Registry. J Clin
Epidemiol 2006;59(1):6776.
[21] Statistics Canada; 2011, http://www23.statcan.gc.ca/imdb/p2SV.pl?Func-
tion=getSurvey&SDDS=4101. http://www12.statcan.ca/census-recensement/
2011.
381
[22] Barlow WE, Ichikawa L, Rosner D, Izumi S. Analysis of case-cohort designs. J
Clin Epidemiol 1999;52(12):116570.
[23] Cai J, Zeng D. Sample size/power calculation for case-cohort studies. Bio-
metrics 2004;60(4):101520.
[24] Langholz B, Jiao J. Computational methods for case-cohort studies. Comput
Stat Data Anal 2007;51(8):373740.
[25] Brown SB, Morrison DS, Cooke TG. Increasing incidence of breast cancer:
distinguishing between the effects of birth cohort and a national breast
screening programme. Breast Cancer Res Treat 2009;116(3):6037.
[26] Al-Delaimy WK, Cho E, Chen WY, Colditz G, Willet WC. A prospective study of
smoking and risk of breast cancer in young adult women. Cancer Epidemiol
Biomark Prev 2004;13(3):398404.
[27] London SJ, Colditz GA, Stampfer MJ, Willett WC, Rosner BA, Speizer FE.
Prospective study of smoking and the risk of breast cancer. J Natl Cancer Inst
1989;81(21):162530.
[28] Rosenberg L, Boggs DA, Bethea TN, Wise LA, Adams-Campbell LL, Palmer JR. A
prospective study of smoking and breast cancer risk among African-American
women. Cancer Causes Control 2013;24(12):220710.
[29] Gaudet MM, Gapstur SM, Sun J, Diver WR, Hannan LM, Thun MJ. Active
smoking and breast cancer risk: original cohort data and meta-analysis. J Natl
Cancer Inst 2013;105(8):51525.
[30] Luo J, Margolis KL, Wactawski-Wende J, Horn K, Messina C, Stefanick ML,
et al. Association of active and passive smoking with risk of breast cancer
among postmenopausal women: a prospective cohort study. BMJ 2011;
342(d1016).
[31] Xue F, Willett WC, Rosner BA, Hankinson SE, Michels KB. Cigarette smoking
and the incidence of breast cancer. Arch Intern Med 2011;171(2):12533.
[32] Cramer DW. The epidemiology of endometrial and ovarian cancer. Hematol
Oncol Clin North Am 2012;26(1):112.
[33] Tanko LB, Christiansen C. An update on the antiestrogenic effect of smoking: a
literature review with implications for researchers and practitioners. Meno-
pause 2004;11(1):1049.
[34] Ambrosone CB, Kropp S, Yang J, Yao S, Shields PG, Chang-Claude J. Cigarette
smoking, N-acetyltransferase 2 genotypes, and breast cancer risk: pooled
analysis and meta-analysis. Cancer Epidemiol Biomark Prev 2008;17(1):15
26.
[35] Terry PD, Goodman M. Is the association between cigarette smoking and
breast cancer modied by genotype? A review of epidemiologic studies and
meta-analysis. Cancer Epidemiol Biomark Prev 2006;15(4):60211.
[36] Zhang J, Qiu LX, Wang ZH, Wang JL, He SS, Hu XC. NAT2 polymorphisms
combining with smoking associated with breast cancer susceptibility: a meta-
analysis. Breast Cancer Res Treat 2010;123(3):87783.
[37] Morabia A. Smoking (active and passive) and breast cancer: epidemiologic
evidence up to June 2001. Environ Mol Mutagen 2002;39(23):8995.
2014 Elsevier