IJC

International Journal of Cancer

Alcohol consumption and cigarette smoking in combination:

A predictor of contralateral breast cancer risk in the
WECARE study
Julia A. Knight 1,2, Jing Fan1, Kathleen E. Malone3, Esther M. John4,5, Charles F. Lynch6, Rikke Langballe7,
Leslie Bernstein8, Roy E. Shore9, Jennifer D. Brooks2, Anne S. Reiner10, Meghan Woods10, Xiaolin Liang10 and
Jonine L. Bernstein10; on behalf of the WECARE Study Collaborative Group
1
Prosserman Centre for Health Research, Lunenfeld-Tanenbaum Research Institute, Sinai Health System, Toronto, ON, Canada
2
Epidemiology Division, Dalla Lana School of Public Health, University of Toronto, Toronto, ON, Canada
3
Public Health Sciences Division, Fred Hutchinson Cancer Research Center, Seattle, WA
4
Division of Epidemiology, Cancer Prevention Institute of California, Fremont, CA
5
Department of Health Research and Policy (Epidemiology) and Stanford Cancer Institute, Stanford University School of Medicine, Stanford, CA
6
Cancer Epidemiology

Departments of Epidemiology and Pathology, University of Iowa, Iowa City, IA

7
Virus, Lifestyle and Genes Unit, Danish Cancer Society Research Center, Copenhagen, Denmark
8
Department of Population Health Sciences, Beckman Research Institute, City of Hope National Medical Center, Duarte, CA
9
Department of Environmental Medicine, New York University, New York, NY
10
Epidemiology and Biostatistics, Memorial Sloan Kettering Cancer Center, New York, NY

Alcohol drinking and, to a lesser extent, cigarette smoking are risk factors for a first primary breast cancer. Information on
these behaviours at diagnosis may contribute to risk prediction of contralateral breast cancer (CBC) and they are potentially
modifiable. The WECARE Study is a large population-based case-control study of women with breast cancer where cases
(N 5 1,521) had asynchronous CBC and controls (N 5 2,212), matched on survival time and other factors, had unilateral breast
cancer (UBC). Using multivariable conditional logistic regression to estimate rate ratios (RR) and 95% confidence intervals (CI),
we examined the risk of CBC in relation to drinking and smoking history at and following first diagnosis. We adjusted for
treatment, disease characteristics and other factors. There was some evidence for an association between CBC risk and cur-
rent drinking or current smoking at the time of first breast cancer diagnosis, but the increased risk occurred primarily among
women exposed to both (RR 5 1.62, 95% CI 1.24–2.11). CBC risk was also elevated in women who both smoked and drank
alcohol after diagnosis (RR 5 1.54, 95% CI 1.18–1.99). In the subset of women with detailed information on amount con-
sumed, smoking an average of 10 cigarettes per day following diagnosis was also associated with increased CBC risk
(RR 5 1.50, 95% CI 1.08–2.08; p-trend 5 0.03). Among women with a diagnosis of breast cancer, information on current drink-
ing and smoking could contribute to the prediction of CBC risk. Women who both drink and smoke may represent a group
who merit targeted lifestyle intervention to modify their risk of CBC.

Breast cancer is the most common cancer in women world-
wide1 with high survival rates in developed countries.2 Thus,
the population of breast cancer survivors at risk of a second
primary cancer is large and increasing. Contralateral breast
cancer (CBC) is the most common second primary in this
Key words: alcohol, smoking, contralateral breast cancer, case-con-
trol study
Grant sponsor: National Institutes of Health; Grant numbers: P30
CA008748, R01 CA114236, R01 CA129639, R01 CA97397, U01
CA83178
DOI: 10.1002/ijc.30791
History: Received 17 Jan 2017; Accepted 26 Apr 2017; Online 19
May 2017
Correspondence to: Julia A. Knight, 60 Murray Street Box 18,
Toronto, Ontario, Canada M6P 2G3, Tel.: 416-586-8701, Fax: 416-
586-8404, E-mail: knight@lunenfeld.ca
group,3 and in these women, the risk is higher than the risk
of developing a first breast cancer in the general population.4
Therefore, predicting and reducing CBC risk is an important
clinical consideration in breast cancer survivors.
Alcohol is a modest, but well established risk factor for first
primary breast cancer, with plausible biological mechanisms.5,6
There has been greater uncertainty regarding the role of ciga-
rette smoking in first primary breast cancer, but the current
weight of evidence suggests that active smoking is a risk factor.7
Few studies have addressed the potential role of these factors in
the risk of CBC and the results have been inconsistent.8–13
However, these studies all had one or more of the following
limitations: small sample size, limited information on the tim-
ing of exposure and limited information on potential con-
founders such as treatment for the first breast cancer.
Information on alcohol and smoking history at the time
of a first diagnosis of breast cancer may be a useful addition

Int. J. Cancer: 141, 916–924 (2017) V

C 2017 UICC
Knight et al.
What’s new?
Do alcohol and smoking contribute to the risk of developing breast cancer . . . a second time? Both are risk factors for first
breast cancer, and this study set out to identify their impact on contralateral breast cancer risk. The authors set up a case-
control study, where cases had developed cancer in both breasts at different times, and controls had developed cancer in
only one breast. They found that alcohol use and smoking after first diagnosis did increase the risk of developing cancer in
the second breast, and recommending lifestyle changes could reduce their risk.
to CBC risk prediction tools, in combination with other fac-
tors. Smoking and alcohol drinking following an initial diag-
nosis of breast cancer may influence the risk of a subsequent
CBC, offering a potential opportunity for risk modification.
Given that these lifestyle factors may change over time, it is
important to account for the timing of exposure, considering
the potential dual role (risk prediction and risk modification)
of these factors.
In our previous analysis of the first phase of the Women’s
Environmental, Cancer and Radiation Epidemiology
(WECARE) Study, we found evidence that any alcohol con-
sumption and longer duration of alcohol consumption up to
the time of CBC diagnosis (or equivalent date in controls
with unilateral breast cancer (UBC)) was associated with an
increased risk of CBC, but the association with smoking was
not statistically significant.12 We now present results from
the expanded WECARE Study with a focus on smoking and
drinking history at the time of a first diagnosis of breast can-
cer as a potential predictor of CBC, and smoking and drink-
ing following diagnosis as potential modifiable risk factors for
CBC.
Methods
Study population
The WECARE Study is a population-based case-control study
of women in which cases have asynchronous CBC and con-
trols have UBC. The study recruitment and data collection
occurred in two phases, WECARE I (2001–2004) and
WECARE II (2009–2012), using similar study procedures in
seven study centres, all of which identified eligible women
through population-based cancer registries (Fig. 1). The five
U.S. study centres recruited women from registries that con-
tribute to the NCI SEER Registry program: Seattle, Washing-
ton; Iowa; Northern California; Los Angeles County,
California and Orange and San Diego Counties, California.
Additional recruitment and data collection were conducted in
Denmark and in Ontario, Canada. In the current analysis, we
combined the data from the two study phases.
WECARE Study participants were diagnosed prior to age
55 years, between 1985 and 2008, with a first primary inva-
sive breast cancer (local or regional). Cases were also diag-
nosed with a second primary CBC (in situ or invasive in
WECARE I, invasive only in WECARE II) at least one year
later in WECARE I or at least two years later in WECARE II
with no intervening cancer diagnosis, other than a non-
Int. J. Cancer: 141, 916–924 (2017) V
C 2017 UICC
917
melanoma skin cancer or cervical carcinoma in situ. Controls
had no history of any other cancer diagnoses except non-
melanoma skin cancer or cervical carcinoma in situ up to the
reference date. The reference date for cases was the CBC
diagnosis date while for controls it was defined by adding the
Cancer Epidemiology
interval between the first breast cancer and the CBC for the
matched case to the date of breast cancer for the control.
Cases and controls must also have been living in the same
study reporting area at the time of their first breast cancer
diagnosis and at the reference date. Additionally, controls
must not have undergone prophylactic mastectomy of the
contralateral breast. All women had to be alive at the time of
contact and able to provide informed consent. Controls were
individually matched to cases (1:2 for WECARE I and 1:1 for
WECARE II) on year of birth (in 5-year strata), year of diag-
nosis (in 4-year strata), cancer registry region and race/eth-
nicity. Matching criteria were selected from potential
confounders available in the cancer registries. In WECARE I,
cases and controls were further counter-matched based on
cancer registry reported radiation treatment such that two
members of each case-control trio had received radiation
treatment for their first breast cancer diagnosis. Counter-
matching was not used in WECARE II; the type of matching
was taken into account in all statistical analyses (see below).
We identified a total of 2,354 CBC cases and 3,599 UBC
controls eligible for inclusion. Of these, 1,521 (65%) cases
(145 with DCIS as the diagnosis in the contralateral breast)
and 2,212 (61%) controls completed an interview and pro-
vided a biospecimen for DNA extraction. All study partici-
pants provided written informed consent and the study was
approved by local Institutional Review or Research Ethics
Boards in North America and by the ethical committee sys-
tem in Denmark.
Data collection
WECARE Study participants were interviewed by telephone
using a structured questionnaire designed to obtain informa-
tion about events occurring both before the diagnosis of the
first primary breast cancer and during the period between
diagnosis and reference date. The questionnaire addressed
currently known and suspected risk factors for breast cancer,
including personal demographics, medical history, cancer
family history, menstrual and reproductive history, hormone
use, body size, cigarette smoking and alcohol intake. Addi-
tionally, medical records including pathology reports and
 918
Cancer Epidemiology
Figure 1. Flow chart for recruitment and data collection in the two phases of the WCARE Study, termed ‘WECARE I Study’ and ‘WECARE II
Study’.
hospital charts were abstracted to collect detailed treatment
information (chemotherapy, hormonal therapy and radiation
therapy) for the first primary breast cancer and any recur-
rences experienced prior to the reference date, and tumour
characteristics of the first primary.
In both WECARE Study phases, we collected information
on whether the women had ever consumed alcohol regularly
(at least one drink per month) or smoked cigarettes regularly
(at least one cigarette per day for 6 months or longer) before
the reference date, as described previously.12 We also asked
at what age women started, whether they had stopped, and,
if so, at what age they had last stopped. In WECARE II, we
asked more detailed questions on any changes in smoking or
Alcohol and cigarettes and CBC risk
drinking amount, when they occurred and amount consumed
after the change was made.
Statistical analysis
We used multivariable conditional logistic regression to esti-
mate rate ratios (RR) and 95% confidence intervals (CI) to
assess the relationship between alcohol consumption and/or
active cigarette smoking and risk of CBC. Models were
adjusted initially only for age at first breast cancer diagnosis
(continuous) and then also for previously identified risk fac-
tors for CBC, including age at menarche (<13 years, 13
years), number of full-term pregnancies (nulliparous, 1–3,
4), first-degree family history of breast cancer (yes, no),
Int. J. Cancer: 141, 916–924 (2017) V
C 2017 UICC
Knight et al.
body mass index (<25, 25–<30, 30 kg/m2), first tumour
histology (lobular, other), first tumour stage (local, regional),
first tumour estrogen receptor (ER) status (positive, negative)
and chemotherapy treatment (yes, no), hormonal treatment
(yes, no) and radiation therapy (yes, no) for the first breast
cancer or any recurrences occurring before the reference
date. Those with missing information were included in a sep-
arate category. In addition, models with alcohol-related varia-
bles were adjusted for smoking status and models with
smoking-related variables were adjusted for drinking status,
with status defined as never, former, current at first diagnosis
and any or none after first diagnosis. A log-weight covariate
was included in the model to account for the sampling prob-
ability of the counter-matching for radiation treatment status
used in WECARE I. WECARE II participants (who were not
counter-matched) were assigned an offset term of 1. Where
relevant, we evaluated trends across all categories considering
the ordinal category as continuous. We examined drinking
and smoking associations in each study phase (WECARE I
and II) separately to confirm that combining the two phases
was a reasonable approach. As detailed data on amount con-
sumed were only collected in WECARE II, these analyses
were restricted to this study phase (813 CBC cases and 813
UBC controls). A two-sided p value <0.05 was considered
statistically significant. Analyses were conducted using R
3.2.4.
We examined drinking and smoking status (current, for-
mer vs. never) at the time of the first diagnosis of breast can-
cer as well as duration of use (<20, 20–<30, 30 years vs.
never) up to the first diagnosis, age at starting (<20, 20
years vs. never) and drinking and smoking alone or in com-
bination (ever vs. never and current vs. not current) in the
combined WECARE Study. In addition, in WECARE II we
considered the average amount consumed (5, <5 drinks
per week or 10, <10 cigarettes per day vs. none) as well as
amount x duration (pack-years and drink-years in categories
based on the median in UBC controls). In the period follow-
ing, the first diagnosis up to the reference date, any drinking
and smoking, separately and in combination, were examined
in the combined WECARE Study and average amount con-
sumed was considered in WECARE II only. Finally, we tested
interactive terms in the model to assess whether drinking
and smoking relationships with CBC differed by ER status or
first-degree family history of breast cancer or whether there
was an interaction between drinking and smoking.
Results
The distribution of age at first breast cancer diagnosis was
similar in study participants and non-participants with 17.5%
of the former and 18.5% of the latter younger than 40 years.
The study population is described in Table 1. CBC cases and
UBC controls were closely matched on age at diagnosis, age
at reference date and time between diagnosis and reference
date. Cases were more likely to have a family history of
breast cancer. The difference in frequency of radiation
Int. J. Cancer: 141, 916–924 (2017) V
C 2017 UICC
919
treatment is due to the counter-matching design used in
WECARE I. The slight differences in distribution of CBC
cases and UBC controls by study centre arise because some
centres (Los Angeles County and Orange and San Diego
Counties) only participated in WECARE I where the control
to case ratio was 2:1 and some centres (Northern California
and Ontario, Canada) only participated in WECARE II
where the ratio was 1:1.
There was limited evidence that alcohol drinking history
at the time of first breast cancer diagnosis is associated with
CBC risk (Table 2). The association of current drinking with
CBC risk was of borderline statistical significance compared
to never drinking after adjustment (RR 1.19, 95% CI 1.00–
1.42). There were fewer former drinkers than current
drinkers and, although the RR for former drinking was simi-
Cancer Epidemiology
lar, the CI was wider (RR 1.16, 85% CI 0.89–1.50). There was
a borderline trend of increasing CBC risk with longer dura-
tion of drinking, but the trend was attenuated after adjust-
ment (p trend 5 0.10). There was no evidence for increased
risk associated with either early initiation of drinking or, in
WECARE II, with increasing amount consumed. Compared
to never smoking, current, but not former, smoking was also
associated with CBC risk (RR 1.23, 95% CI 1.01–1.51) (Table
3). Age at smoking initiation, duration or average amount
smoked were not statistically significantly associated with
CBC risk; however, compared to never smokers, women at or
above the median for pack-years were at a significantly
increased risk of CBC that was only slightly attenuated after
full adjustment (RR 1.28, 95% CI 0.98–1.67). When consid-
ered together, any history of both drinking and smoking at
or before first breast cancer, but not either in isolation, was
associated with a significantly increased risk of CBC com-
pared to never drinking or smoking (RR 1.24, 95% CI 1.01–
1.52) (Table 4). This risk was greater in women who were
drinking and smoking at the time of initial breast cancer
diagnosis, even after adjustment for multiple factors (RR
1.62, 95% CI 1.24–2.11). The interaction between current
drinking and smoking was statistically significant (p 5 0.008)
whereas the interaction between any drinking and smoking
history was not (p-0.76).
Following a first breast cancer diagnosis, there was only
marginal evidence that drinking or smoking during the
period prior to the reference date was associated with an
increased risk of CBC compared to no drinking or no smok-
ing during this period (RR 1.15, 95% CI 0.98–1.34 and RR
1.20, 95% CI 0.99–1.46, respectively) (Tables 2 and 3). There
was no evidence of a dose-response with increasing intake of
alcohol (Table 2), but women who smoked at least 10 ciga-
rettes per day were at statistically significantly greater risk
compared to non-smokers (RR 1.50, 95% CI 1.08–2.08)
(Table 3). When considered in combination a statistically sig-
nificant increase in risk was only observed in women who
both drank and smoked during this period compared to
those who did neither (RR 1.54, 95% CI 1.18–1.99) (Table 4),
similar to current intake at diagnosis.
 920 Alcohol and cigarettes and CBC risk

Table 1. Characteristics of contralateral breast cancer (CBC) cases Table 1. Characteristics of contralateral breast cancer (CBC) cases
and unilateral breast cancer (UBC) controls in the WECARE Study and unilateral breast cancer (UBC) controls in the WECARE Study
(Continued)
CBC Cases UBC Controls
(N 5 1521) (N 5 2212) CBC Cases UBC Controls
Median (range) Median (range) (N 5 1521) (N 5 2212)
Median (range) Median (range)
Age at 1st diagnosis (years) 46 (24–54) 46 (23–54)
4 108 (7) 225 (10)
Age at reference date (years) 53 (27–73) 52 (27–71)
Missing 5 (<1) 5 (<1)
Time between diagnosis 6 (1–20) 6 (1–20)
and reference Body mass index at 1st diagnosis (kg/m2)
date1 (years)
<25 1048 (69) 1495 (68)
N (%) N (%)
25–<30 303 (20) 498 (22)
Study Centre2
30 166 (11) 215 (10)
Northern California 341 (22) 347 (16)
Missing 4 (<1) 4 (<1)
Denmark 279 (19) 457 (21)
Chemotherapy3
Cancer Epidemiology

Seattle, Washington 224 (15) 317 (14)

No 699 (46) 923 (42)
Iowa 201 (13) 314 (14)
Yes 822 (54) 1289 (58)
Los Angeles County, 199 (13) 391 (18)
Radiation therapy3,4
California
No 641 (42) 525 (24)
Ontario, Canada 159 (10) 157 (7)
Yes 880 (58) 1686 (76)
Orange & San Diego 118 (8) 229 (10)
Counties, California Missing 0 (0) 1 (<1)
Year of 1st diagnosis2 Hormone therapy3
1985–1988 238 (16) 467 (21) No 964 (63) 1270 (57)
1989–1992 415 (27) 647 (29) Yes 557 (37) 940 (42)
1993–1996 427 (28) 632 (29) Missing 0 (0) 2 (<1)
1997–2008 441(29) 466 (21) 1
The reference date for cases was the CBC diagnosis date while for
Histology of 1st diagnosis controls it was defined by adding the interval between the first breast
cancer and the CBC for the matched case to the date of breast cancer
Ductal 1205 (79) 1772 (80) for the control.
2
Lobular 179 (12) 223 (10) Note that two centres (Los Angeles County and Orange and San Diego
Counties) only participated in the WECARE I Study where control to
Other 137 (9) 217 (10) case matching was 2:1 and two sites only participated in the WECARE
Stage of 1st diagnosis II Study (Northern California and Ontario, Canada) where the matching
ratio was 1:1 and the range of years of diagnosis in the two phases
Local 1061 (70) 1442 (65) differed, although with considerable overlap.
3
Therapy for first diagnosis or recurrence prior to reference date.
Regional 448 (29) 759 (34) 4
The differences in radiation therapy between groups are due to the
Missing 12 (1) 11 (1) design of the WECARE I Study where there was counter-matching based
on radiation therapy.
Estrogen receptor status of 1st diagnosis
Positive 797 (52) 1254 (57)
Negative 467 (31) 561 (25)
We also examined drinking and smoking status at and
Missing 257 (17) 397 (18)
after first diagnosis, separately and combined, in each
First degree family history WECARE Study phase and results were generally similar
No 1004 (66) 1706 (77) (data not shown). Cases and controls were matched on
Yes 497 (33) 466 (21) reporting registry and subgroup analyses examining heteroge-
Adopted/Missing 20 (1) 40 (2) neity of results by country did not indicate any evidence for
Age at menarche (years)
differences (data not shown). Most study participants (71%
of CBC cases and 72% of UBC controls) were from the
<13 727 (48) 971 (44)
United States. In addition, we considered potential heteroge-
13 791 (52) 1239 (56) neity in findings by the ER status of the first breast cancer
Missing 3 (<1) 2 (<1) diagnosis, but found no evidence for interaction with p-values
Full-term pregnancies ranging from 0.26 to 0.83. There was no evidence for interac-
None 322 (21) 412 (19) tion between smoking history at first breast cancer diagnosis
1–3 1086 (71) 1570 (71)
and first-degree family history of breast cancer (p 5 0.69),
but the interaction between drinking history and family

Int. J. Cancer: 141, 916–924 (2017) V

C 2017 UICC
Knight et al. 921

Table 2. The associations of alcohol drinking with the risk of CBC in CBC cases and UBC controls in the WECARE Study
Cases N Controls N RR1 (95% CI) RR2 (95% CI)
Drinking history at 1st diagnosis
Ever drinking
Never 545 844 1.00 1.00
Former 169 236 1.13 (0.88–1.45) 1.16 (0.89–1.50)
Current 784 1116 1.19 (1.01–1.41) 1.19 (1.00–1.42)
Duration of drinking
Never 545 844 1.00 1.00
<20 years 359 518 1.14 (0.94–1.39) 1.15 (0.94–1.42)
20–<30 years 415 592 1.14 (0.94–1.39) 1.16 (0.94–1.43)
30 years 174 234 1.28 (0.98–1.68) 1.22 (0.92–1.63)
P trend 0.05 0.10

Cancer Epidemiology
Age at starting
Never 545 844 1.00 1.00
20 years 584 828 1.20 (1.01–1.42) 1.21 (1.01–1.46)
<20 years 384 531 1.16 (0.95–1.42) 1.14 (0.92–1.41)
P trend 0.05 0.22
Average amount3
Never 265 283 1.00 1.00
<5 drinks per week 346 336 1.11 (0.87–1.41) 1.09 (0.84–1.42)
5 drinks per week 182 183 1.05 (0.80–1.39) 1.02 (0.75–1.37)
P trend 0.68 0.88
Drink-years3,4
Never 265 283 1.00 1.00
<Median 256 259 1.08 (0.84–1.39) 1.05 (0.81–1.38)
Median 272 260 1.10 (0.85–1.42) 1.09 (0.82–1.44)
P trend 0.88 0.81
Drinking history after 1st diagnosis5
Any drinking
No 675 1032 1.00 1.00
Yes 831 1157 1.17 (1.00–1.36) 1.15 (0.98–1.34)
3
Average amount
Never 327 360 1.00 1.00
<5 drinks per week 315 289 1.22 (0.96–1.54) 1.22 (0.95–1.56)
5 drinks per week 165 159 1.15 (0.87–1.50) 1.12 (0.84–1.50)
P trend 0.23 0.32
1
Adjusted for age at first diagnosis.
2
Adjusted for age at first diagnosis, first degree family history of breast cancer, body mass index at first diagnosis, age at menarche, number of full
term pregnancies, first diagnosis histology, first stage, first diagnosis estrogen receptor status, chemotherapy, radiation and hormone therapy for
breast cancer prior to the reference date, and ever smoking (or ever smoking after 1st diagnosis).
3
In the WECARE II Study only, averaged over periods of drinking.
4
Average number of drinks per week x duration of drinking, median 5 2,818, range 24–107,018.
5
Up to the reference date (for cases this was the CBC diagnosis date while for controls it was defined by adding the interval between the first breast
cancer and the CBC for the matched case to the date of breast cancer for the control).

history did reach statistical significance (p 5 0.04). The without a family history (RR 5 1.41, 95% CI 1.10–1.82 in
increased risk of CBC associated with current drinking at women without a family history vs. RR 5 0.85, 96% CI 0.42–
first breast cancer diagnosis was only observed in women 1.69 in women with a family history).

Int. J. Cancer: 141, 916–924 (2017) V

C 2017 UICC
 922 Alcohol and cigarettes and CBC risk

Table 3. The associations of cigarette smoking with the risk of CBC in CBC cases and UBC controls in the WECARE Study
Cases N Controls N RR1 (95% CI) RR2 (95% CI)
Smoking history at 1st diagnosis
Ever smoking
Never 801 1175 1.00 1.00
Former 413 635 1.01 (0.86–1.19) 0.95 (0.80–1.14)
Current 306 401 1.25 (1.03–1.51) 1.23 (1.01–1.51)
Duration of smoking
Never 801 1175 1.00 1.00
<20 years 317 496 1.01 (0.84–1.21) 0.97 (0.80–1.18)
20–<30 years 240 334 1.15 (0.94–1.42) 1.12 (0.90–1.39)
30 years 154 198 1.25 (0.96–1.64) 1.15 (0.87–1.53)
P trend 0.06 0.22
Cancer Epidemiology

Age at starting
Never 801 1175 1.00 1.00
20 years 210 329 1.04 (0.84–1.29) 1.03 (0.82–1.29)
<20 years 510 707 1.12 (0.96–1.32) 1.07 (0.91–1.27)
P trend 0.48 0.85
Average amount3
Never 450 473 1.00 1.00
<10 cigarettes per day 123 137 0.97 (0.72–1.29) 0.92 (0.67–1.26)
10 cigarettes per day 232 195 1.23 (0.98–1.55) 1.20 (0.93–1.54)
P trend 0.10 0.20
Pack-years3,4
Never 450 473 1.00 1.00
<Median 162 181 0.96 (0.74–1.24) 0.91 (0.68–1.20)
Median 193 151 1.31 (1.02–1.67) 1.28 (0.98–1.67)
P trend 0.06 0.11
Smoking history after 1st diagnosis5
Any smoking
No 1207 1802 1.00 1.00
Yes 311 408 1.22 (1.02–1.47) 1.20 (0.99–1.46)
3
Average amount
Never 665 694 1.00 1.00
<10 cigarettes per day 37 43 0.89 (0.57–1.39) 0.86 (0.54–1.37)
10 cigarettes per day 110 75 1.52 (1.11–2.07) 1.50 (1.08–2.08)
P trend 0.02 0.03
1
Adjusted for age at first diagnosis.
2
Adjusted for age at first diagnosis, first degree family history of breast cancer, body mass index at first diagnosis, age at menarche, number of full
term pregnancies, first diagnosis histology, first diagnosis stage, first diagnosis estrogen receptor status, chemotherapy, radiation and hormone ther-
apy for breast cancer prior to the reference date, and ever drinking (or ever drinking after 1st diagnosis).
3
In the WECARE II Study only, averaged over periods of smoking.
4
Average number of cigarettes per day/20 3 duration of smoking, median 5 3,579, range 5–27,759.
5
Up to the reference date (for cases this was the CBC diagnosis date while for controls it was defined by adding the interval between the first breast
cancer and the CBC for the matched case to the date of breast cancer for the control).

Discussion alcohol and tobacco at the time of first diagnosis had the
At the time of an initial diagnosis of breast cancer, informa- strongest association with increased CBC risk among all the
tion on drinking and active smoking may contribute to the drinking and smoking variables considered and may be a
prediction of risk of developing a CBC. Consumption of both useful addition to risk prediction models, particularly as this

Int. J. Cancer: 141, 916–924 (2017) V

C 2017 UICC
Knight et al. 923

Table 4. The associations of alcohol drinking and cigarette smoking individually or in combination with the risk of CBC in CBC cases and
UBC controls the WECARE Study
Cases N Controls N RR1 (95% CI) RR2 (95% CI)
History at 1st diagnosis
Ever drinking and/or smoking
Neither 368 568 1.00 1.00
Smoking only 177 275 1.09 (0.85–1.41) 1.09 (0.84–1.42)
Drinking only 419 601 1.16 (0.95–1.43) 1.19 (0.96–1.48)
Drinking and smoking 533 751 1.25 (1.03–1.53) 1.24 (1.01–1.52)
Current drinking and/or smoking
Neither 604 912 1.00 1.00
Smoking only 110 167 0.96 (0.72–1.28) 0.92 (0.68–1.24)
Drinking only 591 887 1.06 (0.89–1.25) 1.04 (0.87–1.24)

Cancer Epidemiology
Drinking and smoking 192 229 1.63 (1.26–2.10) 1.62 (1.24–2.11)
History after 1st diagnosis3
Any drinking or smoking
Neither 570 870 1.00 1.00
Smoking only 104 161 0.92 (0.68–1.24) 0.87 (0.64–1.19)
Drinking only 625 917 1.07 (0.90–1.27) 1.05 (0.88–1.25)
Drinking and smoking 204 239 1.55 (1.21–2.00) 1.54 (1.18–1.99)
1
Adjusted for age at first diagnosis.
2
Adjusted for age at first diagnosis, breast cancer family history, body mass index at first diagnosis, age at menarche, number of full term pregnan-
cies, first diagnosis histology, first diagnosis stage, first diagnosis estrogen receptor status, chemotherapy for first diagnosis, radiation for first diag-
nosis, hormone therapy for first diagnosis.
3
Up to the reference date (for cases this was the CBC diagnosis date while for controls it was defined by adding the interval between the first breast
cancer and the CBC for the matched case to the date of breast cancer for the control).

information is relatively easy to collect. The combination of
both drinking and smoking following a first diagnosis was
also associated with increased CBC risk and suggests that
lifestyle interventions may be beneficial in this group.
Alcohol is a well-established modest risk factor for first
primary breast cancer, although the role of timing of expo-
sure remains unclear.5,6 The data on smoking as a risk factor
for first primary breast cancer have been inconsistent and
controversial, but the weight of evidence now suggests that
active smoking is also a modest risk factor.7 Earlier studies
with smaller sample sizes and often limited information on
the exposures of interest did not find associations between
either alcohol consumption or smoking and CBC risk.11 In
our previous analysis of WECARE I, we found evidence of
increased risk of CBC among ever drinkers and also an
increase in risk with increasing lifetime duration of drinking,
but little evidence of an association with smoking.12
To our knowledge only one other study, in older, ER-
positive women only, separately considered smoking and
drinking history specifically at and after an initial breast can-
cer diagnosis.13 Although their variable definitions differed
somewhat from ours, the evidence for some association
between any drinking prior to a first breast cancer diagnosis
appears consistent between the two studies and both studies
observed increased CBC risk with smoking at first diagnosis
and subsequent smoking. In our much larger study, we were
able to estimate the risks relative to not drinking and smok-
ing more precisely. Furthermore, our findings did not differ
by ER status at first diagnosis. Li et al. also evaluated CBC
risk associated with drinking and smoking in combination
and observed greatly elevated risk of CBC among women
who were in the highest alcohol consumption category and
were current smokers either at the time of first diagnosis or
at the reference date.13 Those estimates, although imprecise
and likely somewhat inflated, are consistent with the
increased CBC risk we observed for drinking and smoking in
combination. Our estimates of increased risk are lower, but
more precise. In both studies, the levels of drinking were
quite modest with most women consuming less than one
drink per day.
Both alcohol drinking and cigarette smoking are plausible
risk factors for breast cancer. Several mechanisms have been
proposed to explain the association of alcohol with first pri-
mary breast cancer5 and carcinogens in cigarettes are well
documented.7 The bulk of evidence suggests that both alcohol
drinking and cigarette smoking are relatively weak risk fac-
tors, but do contribute to breast cancer risk.5,7 It is possible
that engaging in both behaviours may also be a marker of
other unhealthy lifestyle factors, such as poor diet and physi-
cal inactivity, that explain or enhance the association of

Int. J. Cancer: 141, 916–924 (2017) V

C 2017 UICC
 924 Alcohol and cigarettes and CBC risk

drinking and smoking in combination with CBC risk. In
either case, women who both drink and smoke at diagnosis
or who continue to engage in both behaviours following a
breast cancer diagnosis may represent a group who would
benefit from enhanced support for lifestyle modification.
However, we note that few women in our study reported
stopping smoking or drinking around the time of a first
breast cancer diagnosis in the WECARE Study and few in
the WECARE II Study reported a reduction in the amount
consumed after first diagnosis. Therefore, we cannot assess
any potential effect of these changes.
Our study has several important strengths. With 1,521
CBC cases, the size of the study is far larger than any other
on this topic. In addition, we could account for many impor-
tant covariates in the analysis through our collection of
Cancer Epidemiology
cannot comment on risks associated with high alcohol con-
sumption. Despite the depth and breadth of our data, there
was some information we did not collect and, therefore, we
could not evaluate or adjust for other health behaviours such
as diet. We also did not collect information on passive smok-
ing. Although the role of second hand smoking is less clear
than active smoking,7 this may have attenuated the estimates
of the effect of active smoking.
In summary, in our large combined WECARE I and
WECARE II Studies we found evidence that current drinking
and smoking in combination at the time of a first breast can-
cer is a marker of increased risk of CBC. In addition, women
who both drink and smoke following diagnosis continue to
be at increased risk. Although the relative increase in CBC

detailed information from medical records and from the
study participants. Nonetheless, there are some study limita-
tions. We only collected detailed information on lifetime
drinking and smoking in the WECARE II Study and thus
had a more restricted sample size for these analyses. The
study population was restricted to women younger than 55
years at the first breast cancer diagnosis and results may dif-
fer in older women. We note, however, that younger women
have a higher risk of developing CBC than older women.
Case-control studies always have the potential for recall bias
in self-reported behaviours, but this is mitigated in our
design as both the CBC cases and UBC controls had a diag-
nosis of breast cancer and matched cases and controls had a
similar recall period. Some non-differential recall error is
likely, leading to attenuation in results, although ever drink-
ing or smoking and continuing to drink and/or smoke after
diagnosis are likely recalled with some accuracy. Few women
in the study reported heavy drinking and, consequently, we
risk is not high, given the background risk of CBC and con-
cern regarding this risk, lifestyle interventions may be war-
ranted in this group.
Acknowledgements
The Women’s Environmental, Cancer and Epidemiology (WECARE) Study
Collaborative Group includes the following Memorial Sloan Kettering Can-
cer Center (Coordinating Center) Investigators and Staff: Jonine L. Bernstein
Ph.D. (WECARE Study P.I.); Marinela Capanu Ph.D.; Xiaolin Liang M.D.;
Irene Orlow Ph.D.; Anne S. Reiner M.P.H.; Mark Robson, M.D.; Meghan
Woods M.P.H. Collaborative Site Investigators: Leslie Bernstein Ph.D.; John
D. Boice Jr. Sc.D.; Jennifer Brooks Ph.D.; Patrick Concannon Ph.D.; Dave V.
Conti Ph.D.; David Duggan Ph.D.; Joanne W. Elena Ph.D., M.P.H.; Robert
W. Haile Dr.P.H.; Esther M. John Ph.D.; Julia A. Knight Ph.D.; Charles F.
Lynch M.D., Ph.D.; Kathleen E. Malone Ph.D.; Lene Mellemkjær Ph.D.;
Jørgen H. Olsen M.D. DMSc.; Daniela Seminara Ph.D. M.P.H.; Roy E. Shore
Ph.D., Dr.P.H.; Marilyn Stovall Ph.D.; Daniel O. Stram Ph.D.; Marc Tisch-
kowitz M.D., Ph.D.; Duncan C. Thomas Ph.D. Collaborative Site Staff: Kris-
tina Blackmore M.Sc.; Anh T. Diep; Judy Goldstein; Irene Harris B.S.,
C.M.D.; Rikke Langballe M.P.H.; Cecilia O’Brien; Susan Smith M.P.H.; Rita
Weathers M.S.; Michele West Ph.D.

References
1. Ferlay J, Soerjomataram I, Dikshit R, et al. Can-
cer incidence and mortality worldwide: sources,
methods and major patterns in GLOBOCAN
2012. Int J Cancer 2015;136:E359–86.
2. Youlden DR, Cramb SM, Dunn NA, et al. The
descriptive epidemiology of female breast cancer:
an international comparison of screening, inci-
dence, survival and mortality. Cancer Epidemiol
2012;36:237–48.
3. Curtis RE, Hankey BF, Hoover RN. New malig-
nancies following breast cancer. In: Curtis RE,
Ron E, Ries LAG, Hacker DG, Edwards BK,
Tucker MA, Fraumeni JF, Jr, eds. New malignan-
cies among cancer survivors: SEER cancer regis-
tries, 1973–2000. Bethesda, MD: National Cancer
Institute, NIH Publication No. 05-5302, 2006.
4. Chen Y, Thompson W, Semenciw R, et al. Epide-
miology of contralateral breast cancer. Cancer
Epidemiol Biomarkers Prev 1999;8:855–61.
5. Seitz HK, Pelucchi C, Bagnardi V, et al. Epidemi-
ology and pathophysiology of alcohol and breast
cancer: update 2012. Alcohol Alcohol 2012;47:
204–12.
6. Scoccianti C, Lauby-Secretan B, Bello P-Y, et al.
Female breast cancer and alcohol consumption. A
review of the literature. Am J Prev Med 2014;46:
S16–25.
7. Reynolds P. Smoking and breast cancer.
J Mammary Gland Biol Neoplasia 2013;18:15–23.
8. Horn PL, Thompson WD. Risk of contralateral
breast cancer: associations with factors related to
initial breast cancer. Am J Epidemiol 1988;128:
309–23.
9. Bernstein JL, Thompson WD, Risch N, et al. Risk
factors predicting the incidence of second pri-
mary breast cancer among women diagnosed
with a first primary breast cancer. Am J Epide-
miol 1992;136:925–36.
10. Li CI, Malone KE, Porter PL, et al. Epidemiologic
and molecular risk factors for contralateral breast
cancer among young women. Br J Cancer 2003;
89:513–8.
11. Trentham-Dietz A, Newcomb PA, Nichols HB,
et al. Breast cancer risk factors and second primary
malignancies among women with breast cancer.
Breast Cancer Res Treat 2007;105:195–207.
12. Knight JA, Bernstein L, Largent J, et al. Alcohol
intake and cigarette smoking and risk of a con-
tralateral breast cancer. The Women’s Environ-
mental Cancer and Radiation Epidemiology
Study. Am J Epidemiol 2009;169:962–8.
13. Li CI, Daling JR, Porter PL, et al. Relationship
between potentially modifiable lifestyle factors
and risk of second primary contralateral breast
cancer among women diagnosed with estrogen
receptor-positive invasive breast cancer. J Clin
Oncol 2009;27:5312–8.

Int. J. Cancer: 141, 916–924 (2017) V

C 2017 UICC