IJE vol.32 no.6 International Epidemiological Association 2003; all rights reserved.

International Journal of Epidemiology 2003;32:961966

DOI: 10.1093/ije/dyg155

Antidepressant medication use and breast

cancer risk: a case-control study
Allan Steingart,1 Michelle Cotterchio,2,3 Nancy Kreiger2,3,4 and Margaret Sloan2

Accepted

18 February 2003

Methods

The Ontario Cancer Registry (OCR) identified women diagnosed with primary
breast cancer. Controls, randomly sampled from the female population of Ontario,
were frequency matched by 5-year age groups. A mailed self-administered
questionnaire included questions about lifetime use of AD and potential confounders. Multivariate logistic regression yielded odds ratio estimates.

Results

Ever use of AD was reported by 14% (441/3077) cases versus 12% (372/2994)
controls. The age-adjusted odds ratio (AOR) for ever use was 1.17, (95%
CI: 1.01, 1.36). An increased risk was also observed for selective serotonin
reuptake inhibitors = 1.33 (95% CI: 1.07, 1.66), Sertraline = 1.58 (95% CI: 1.03,
2.41), and Paroxetine = 1.55 (95% CI: 1.00, 2.40). None of the 30 variables
assessed for confounding altered the risk estimate by more than 10%.
Multivariate adjustment including all possible breast cancer risk factors yielded an
unchanged, but not significant, point estimate (MVOR = 1.2, 95% CI: 0.96, 1.51).
No relationship was observed for duration or timing of AD use.

Conclusions A modest association between ever use of AD and breast cancer was found using
the most parsimonious multivariate model. OR estimates did not change, but CI
were widened and statistical significance lost, after adjustment for factors
associated with breast cancer risk.
Keywords

Antidepressant medication, breast cancer

There is epidemiological evidence that antidepressant (AD)

medication use may be associated with an increased risk of
breast cancer in women,15 although findings have been
inconsistent, with three studies reporting no association68 and
five other studies finding an increased risk of breast cancer in
women taking tricylic antidepressants (TCA) or selective
serotonin reuptake inhibitors (SSRI).15 The preponderance of
animal studies also indicate that certain antidepressants may
1 Department of Psychiatry, University of Toronto, Toronto, Ontario, Canada.
2 Division of Preventive Oncology, Research Unit, Cancer Care Ontario,

Toronto, Ontario, Canada.

3 Department of Public Health Sciences, University of Toronto, Toronto,

Ontario, Canada.
4 Department of Nutritional Sciences, University of Toronto, Toronto,

Ontario, Canada.
Correspondence: Nancy Kreiger, Research Unit, Division of Preventive
Oncology, Cancer Care Ontario, 620 University Avenue, Toronto, Ontario
M5G 2L7, Canada. E-mail: [email protected]

promote tumours in experimental models.914 The widespread

use of antidepressants1517 and the rise in incidence of breast
cancer18,19 have focused interest on whether AD use, a
potentially modifiable factor, may be associated with breast
cancer risk. We evaluated the association between the use of AD
and risk of breast cancer using a population-based case-control
study design.

Methods
Selection of cases and controls
Cases were selected based on the following criteria: (1) women
diagnosed with breast cancer between June 1996 and May
1998, and reported, via pathology report, to the OCR as a first
primary breast cancer, (2) aged 2574 years at diagnosis, and
(3) alive and resident of Ontario. The OCR is a population-based
cancer registry and includes all cases of invasive cancer
diagnosed among residents of Ontario. Physicians identified in

961

Downloaded from http://ije.oxfordjournals.org/ at Pennsylvania State University on March 3, 2014

Background Animal and human studies have reported an association between antidepressant
(AD) medication use and breast cancer risk. A population-based case-control
study was designed specifically to examine this association among women in
Ontario, Canada.

962

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

the pathology reports were sent a letter describing the study and
were asked for permission to contact the patient, and to provide
the patient address, telephone number, and vital status.
The random selection of controls was based on the following
criteria: (1) females never diagnosed with cancer of the breast,
(2) aged 2574 years, and (3) alive and resident of Ontario.
The population-based assessment rolls of the Ontario Ministry
of Finance were used to identify eligible controls. These
records are organized by geographical region, include all home
owners and tenants, and list age, sex, telephone number, and
address. There is evidence that 95% of cases can be located in
these rolls, indicating a high degree of completeness (Holowaty
EJ, personal communication). Controls were frequencymatched to cases within 5-year age groups.

(menopausal status, age at last menstruation, age at menarche,

number of pregnancies, number of live births, age at first live
birth, history of breastfeeding, months of breastfeeding), anthropometric (height, weight, body mass index), medical (family
history of breast cancer, diagnosis of benign breast disease),
psychiatric (history of symptoms of depression, anxiety, posttraumatic stress, obsessions and compulsions, anorexia or bulimia,
diagnosis of clinical depression), lifestyle (alcohol consumption,
physical activity, dietary fat consumption, smoking status and
exposure to smoke), and use of other medications (hormone
replacement therapy, oral contraceptives, nonsteroidal antiinflammatories, and benzodiazepines).

Data collection

The response rate was 73% (3133/4289) for cases and 61%
(3062/5001) for controls. Table 1 describes selected characteristics of cases and controls. Cases were significantly more likely
to have breast cancer in a first-degree relative, a history of
benign breast cysts, later age at menopause, and fewer than four
pregnancies.
Table 2 shows the frequency distribution of cases and
controls, AOR (age-adjusted odds ratio) and MVOR estimates
and 95% CI for any regular AD use, duration of AD use, and
time since first and last use. There was no confounding by any
of the variables previously mentioned when using the 10%
change in estimate criterion, so the most parsimonious model,
adjustment for age only, is also presented. Ever use of AD was
associated with a statistically significant increased breast cancer
risk, though the magnitude of association was modest (AOR =
1.17, 95% CI: 1.01, 1.36) and was not statistically significant in
the multivariate analysis (MVOR = 1.2, 95% CI: 0.96, 1.51). No
increased risk was observed with trends in duration of use, or
time since first or last use. AD use prior to menopause was
associated with a statistically significant increased breast cancer
risk (AOR = 1.28, 95% CI: 1.07, 1.54), but this was not observed
among women who used AD only after menopause. MVOR
point estimates did not differ significantly in any of the analyses,
but did result in increased CI width and loss of statistical significance, due to increased standard error.
Table 3 shows the frequency distribution of cases and
controls, AOR, and MVOR for the four main classes of AD and
for specific AD. The observed increase in risk was significant for
SSRI in the AOR but not MVOR. In particular, sertraline and
paroxetine were both associated with an increase in breast
cancer risk (AOR = 1.58, 95% CI: 1.03, 2.41; MVOR = 1.45,
95% CI: 0.88, 2.40 and AOR = 1.55, 95% CI: 1.00, 2.40;
MVOR = 1.6, 95% CI: 0.93, 2.77, respectively). In addition,
SSRI used in the pre-menopausal period were associated with
increased breast cancer risk (AOR = 1.32, 95% CI: 1.05, 1.66)
(data not shown). No increased risk was observed for duration
of use, or time since first or last use for any of the classes of AD;
however, power was somewhat limited.

Data analysis
Multivariate logistic regression analysis was performed using
EGRET software.21 Exposure variables included ever/never use
of any AD, duration of use, age at first and last use, and
menopausal status at time of use. AD were also evaluated by
class: SSRI, TCA, monoamine oxidase inhibitor (MAOI), and
atypical. In addition, use of individual medications was
examined when the numbers were sufficient. Reported use of
AD in the one-year period prior to diagnosis date for cases, or a
referent date (based on the mid-point of case diagnoses) for
controls, was excluded from the analysis. In a secondary
analysis, risk factors described as having a definite or probable
association with increased breast cancer risk in the Harvard
Cancer Risk Index22 (age, height, body mass index, age at
menarche, parity, age at menopause, oral contraceptive use,
alcohol consumption, family history of breast cancer, history of
benign breast disease) as well as indication for use of AD
(depression and anxiety) were all included in a multivariate
model to determine the multivariate adjusted odds ratio
(MVOR).
The 10% change-in-estimate method was used to assess potential confounders.23 Potential confounding variables included
demographic (marital status, education, income), reproductive

Discussion
The interpretation of the results of this study depend, in part, on
the choice of statistical methods used to adjust for confounding.
The MVOR results and the absence of a relationship with
duration of exposure do not support the hypothesis that AD

Downloaded from http://ije.oxfordjournals.org/ at Pennsylvania State University on March 3, 2014

A mailed self-administered questionnaire included questions

about past use of AD and other medications as well as information on many other known or potential risk factors for
breast cancer. The questionnaire was 21 pages in length and
extensively pre-tested and contained colour pill photographs of
commonly prescribed AD, as well as non-steroidal antiinflammatory drugs, antihistamines, benzodiazapines, hormones,
and oral contraceptive preparations. In addition, subjects listed
others used that were not shown as a photograph. Subjects
were asked to report details regarding use, such as the start and
stop age for each medication taken. Subjects were also asked
whether they had been diagnosed with various mental illnesses.
Follow-up procedures involved a reminder postcard sent to
all women within 2 weeks of questionnaire mailing, and a
telephone follow-up for non-responders. Telephone calls were
also made after receipt of questionnaires to clarify information
provided. Strategies for increasing response rate included a
personalized cover letter and, for the controls, a $5-payment
which was demonstrated to enhance the cumulative response
rate.20

Results

ANTIDEPRESSANTS AND BREAST CANCER

963

Table 1 Frequency distribution of cases and controls and age-adjusted odds ratio (AOR) estimates for selected variables
Cases (n = 3133)

Controls (n = 3062)

No.a

(%)

No.a

(%)

AOR (95% CI)

Highest level of education

Elementary
High School
Post secondary

399
1469
1246

(13)
(47)
(40)

426
1386
1221

(14)
(46)
(40)

1.0
1.18 (1.01, 1.38)
1.19 (1.00, 1.40)

Cigarette smoking
Never-smoker
Ex-smoker
Current smoker

1522
1044
540

(49)
(34)
(17)

1578
854
594

(52)
(28)
(20)

1.0
1.26 (1.13, 1.42)
0.96 (0.84, 1.10)

Ever diagnosed with depression

No
Yes

2596
417

(86)
(14)

2577
382

(87)
(13)

1.0
1.08 (0.93, 1.26)

Breast cancer in first-degree relativeb

No
Yes

2433
521

(82)
(18)

2527
308

(89)
(11)

1.0
1.72 (1.48, 2.00)

History benign breast cysts

No
Yes

1312
1621

(45)
(55)

2185
720

(75)
(25)

1.0
3.71 (3.32, 4.15)

588
788
907
806

(19)
(26)
(29)
(26)

540
733
880
868

(18)
(24)
(29)
(29)

1.0
0.99 (0.85, 1.16)
0.94 (0.81, 1.09)
0.83 (0.71, 0.97)

461
398
1725
520

(15)
(13)
(56)
(17)

377
357
1673
629

(12)
(12)
(55)
(21)

1.0
0.94 (0.77, 1.16)
0.85 (0.72, 1.01)
0.64 (0.52, 0.77)

Age at menopaused
45
4549
50
Pre-menopausal

612
550
903
859

(21)
(19)
(31)
(29)

647
531
736
949

(23)
(19)
(26)
(33)

1.0
1.09 (0.93, 1.28)
1.27 (1.09, 1.48)

Hormone treatment
Pre-menopausal
No
Yes

859
1235
996

(28)
(40)
(32)

949
1128
946

(31)
(37)
(31)

n/a
1.0
0.96 (0.85, 1.09)

Age at menarche

11
12
13
14
Parityc
Nulliparous
1
23
4

a Numbers may not add to total due to missing values.

b Mother, sister, or daughter.
c Live births.
d Last menstrual period.

may increase the risk of breast cancer. Significant increased

breast cancer risks were observed, in the age-adjusted
parsimonious model, for ever use and with SSRI, in particular
Paroxetine and Sertraline, although these were not statistically
significant in the multivariate analyses. The point estimates for
the two analyses, AOR and MVOR, however, are essentially the
same, suggesting the absence of confounding.
Consistent with our findings, an association between certain
AD and breast cancer risk was also reported in four other casecontrol studies. Kelly et al.,3 using a hospital-based case-control
design, reported a borderline increased risk of breast cancer
associated with regular SSRI use (OR = 1.8, 95% CI: 1.0, 3.3)
but did not find an overall association between ever AD use
and breast cancer. An elevated risk of breast cancer associated
with the use of paroxetine and use of tricyclic medications for
greater than 2 years was reported by Cotterchio et al.,1 although
this case-control study had a limited sample size and CI were
wide. To date, no other breast cancer study has evaluated

paroxetine, a relatively new SSRI. Currently, more than 20%

of women have ever used an SSRI; however, at the time this
study was conducted, only 5% of controls reported using SSRI,
which had been on the market for about 7 years. The possible
association between the use of paroxetine and breast cancer risk
observed in both the current study and one previous study
warrants further investigation. Future studies should have
adequate power to evaluate the individual and commonly used
SSRI. Finally, Sharpe et al.5 reported that heavy exposure to
TCA for greater than 10 years was associated with elevated risk
of breast cancer.
No association between use of AD and breast cancer was
found by Wang et al.13 in a retrospective cohort study. AD use
prior to or after the 2-year time frame of the study (19891991)
was not assessed and this could have resulted in misclassification of subjects exposed to AD before or after the exposure
period. In addition, no subjects were prescribed more recent
SSRI as these were only recently available, so these medications

Downloaded from http://ije.oxfordjournals.org/ at Pennsylvania State University on March 3, 2014

Variable

964

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

Table 2 Frequency distribution of cases and controls, age-adjusted odds ratio (AOR) and multivariate-adjusted odds ratio (MVOR) for
antidepressant (AD) use and duration of use, time since first and last use
Cases

Controls

MVORa

AOR

No.b

(95% CI)

(95% CI)

2636
441

(86)
(14)

2622
372

(88)
(12)

1.0
1.17 (1.01, 1.36)

1.0
1.20 (0.96, 1.51)

70
75
50
58

(2)
(3)
(2)
(2)

62
67
59
49

(2)
(2)
(2)
(2)

1.13
1.10
0.84
1.15

1.03
0.93
0.78
1.14

Time since last AD use (years)

1
27
8+

185
44
48

(6)
(2)
(2)

172
40
42

(6)
(1)
(2)

1.06 (0.85, 1.31)

1.12 (0.73, 1.72)
1.11 (0.73, 1.69)

1.00 (0.74, 1.35)

1.17 (0.70, 1.95)
0.95 (0.57, 1.59)

Time since first AD use (years)

2
36
715
16+

63
79
65
70

(2)
(3)
(2)
(2)

68
62
60
65

(2)
(2)
(2)
(2)

0.93
1.27
1.07
1.04

0.97
1.15
0.99
0.94

Any regular AD use

Neverc
Everd
Duration of AD use (years)
1
13
3.58.5
9+

No.b

(0.80,
(0.79,
(0.57,
(0.78,

1.32)
1.78)
1.52)
1.46)

(0.66,
(0.61,
(0.49,
(0.71,

(0.62,
(0.75,
(0.64,
(0.61,

1.60)
1.43)
1.24)
1.83)

1.50)
1.75)
1.54)
1.46)

a Adjusted for age, height, body mass index, age at menarche, parity, age at menopause, oral contraceptive use, alcohol consumption, family history of breast

cancer, history of benign breast disease, clinical depression, anxiety.

b Numbers may not add to total due to missing values.
c Referent category for each variable.
d Defined as taken daily for at least 2 months started at least 12 months prior to diagnosis date for cases and referent date for controls.

Table 3 Frequency distribution of cases and controls, and age-adjusted odds ratios (AOR) and multivariate-adjusted odds ratio (MVOR) with 95%
CI for ever use of the four main classes of antidepressants (AD) and specific AD
Cases
ADa/Classb

Controls

MVORc

AOR

No.

No.

(95% CI)

(95% CI)

2636

(88)

2622

(89)

1.00

1.00

Sertraline
Paroxetine
Fluoxetine

201
56
51
90

(7)
(2)
(2)
(3)

153
36
34
82

(5)
(1)
(1)
(3)

1.33
1.58
1.55
1.09

(1.07,
(1.03,
(1.00,
(0.81,

1.66)
2.41)
2.40)
1.49)

1.32
1.45
1.60
1.05

(0.97,
(0.88,
(0.93,
(0.70,

1.80)
2.40)
2.77)
1.57)

TCAf
Amitriptyline
Imipramine
Doxepin

208
133
25
29

(7)
(4)
(1)
(1)

182
118
28
23

(6)
(4)
(1)
(1)

1.12
1.10
0.88
1.21

(0.91,
(0.85,
(0.51,
(0.70,

1.37)
1.42)
1.51)
2.10)

1.10
1.19
0.52
1.27

(0.83,
(0.85,
(0.26,
(0.65,

1.45)
1.65)
1.03)
2.46)

Never used any ADd

SSRIe

MAOIg
Atypical

(0.3)

10

(0.3)

0.87 (0.35, 2.14)

0.80 (0.27, 2.40)

19

(0.6)

20

(0.7)

0.94 (0.5, 1.77)

1.04 (0.50, 2.16)

a Defined as taken daily for at least 2 months and started at least 12 months prior to diagnosis date.
b Number within each class do not add up to total, because only the most prevalent antidepressants are listed as individual drugs.
c Adjusted for age, height, body mass index, age at menarche, parity, age at menopause, oral contraceptive use, alcohol consumption, family history of breast

cancer, history of benign breast disease, clinical depression, anxiety.

d Referent category for each variable.
e Selective serotonin reuptake inhibitors.
f Tricylic antidepressants.
g Monoamine oxidase inhibitor.

were not assessed. Friedman et al.7 and Selby et al.8 also reported
no association between use of AD and breast cancer risk but
these studies were also completed prior to the availability of
SSRI.
Our study involved a large sample of cases and controls
selected from population-based sampling frames, and used a
questionnaire specifically designed to obtain information about
use of AD and other potential risk factors for breast cancer.
Nevertheless, potential bias and confounding are concerns of all
case-control studies, in light of which our results must be

considered. Since both cases and controls were selected from

population-based sampling frames, selection bias may have
been limited. Reponse bias is a significant concern as the
response rate for cases was 73% and for controls was 61%. Our
dataset replicates the risk factor associations known to be
associated with breast cancer risk, suggesting that the casecontrol comparison is not on the face of it a biased one.
The potential for recall bias is a significant shortcoming of our
study although efforts were made to reduce this; the research
hypothesis was not stated in the questionnaire, the questionnaire

Downloaded from http://ije.oxfordjournals.org/ at Pennsylvania State University on March 3, 2014

(0.66,
(0.91,
(0.75,
(0.74,

1.60)
1.54)
1.22)
1.69)

ANTIDEPRESSANTS AND BREAST CANCER

susceptibilities. Case-control studies may be too crude a tool to

detect subtle effects unless guided by improved understanding
of potential biological mechanisms. Further research might be
directed towards neuroendocrine consequences of AD, the effects
on intermediate breast cancer endpoints such as benign proliferative breast disease or breast tissue density, and the completion of improved postmarketing surveillance studies and
well-designed cohort studies.

Acknowledgements
We would like to thank the study coordinators, Erin Smith and
Gabrielle Rabbat. As well, we would like to thank Giao Buchan,
Lori-Ann Larmand, Gayle Morrison, and Elana Steingart for
their assistance with data management, and Gerarda Darlington
for her contribution to the development of this research. This
research was funded by the Canadian Breast Cancer Research
Initiative (Grant No. 007235).

References
1 Cotterchio M, Kreiger N, Darlington G, Steingart A. Antidepressant

medication use and breast cancer risk. Am J Epidemiol 2000;

151:95157.
2 Halbreich U, Shen J, Panaro V. Are chronic psychiatric patients at

increased risk for developing breast cancer? [see comments]. Am J

Psychiatry 1996;153:55960.
3 Kelly JP, Rosenberg L, Palmer JR et al. Risk of breast cancer according

to use of antidepressants, phenothiazines, and antihistamines. Am J

Epidemiol 1999;150:86168.
4 Wallace RB, Sherman BM, Bean JA. A case-control study of breast

cancer and psychotropic drug use. Oncol 1982;39:27983.

5 Sharpe CR, Collet JP, Belzile E, Hanley JA, Boivin JF. The effects of

tricyclic antidepressants on breast cancer risk. Br J Cancer

2002;86:9297.
6 Wang PS, Walker AM, Tsuang MT, Orav EJ, Levin R, Avorn J.

Antidepressant use and the risk of breast cancer: a non-association.

J Clin Epidemiol 2001;54:72834.
7 Friedman GD, Ury HK. Screening for possible drug carcinogenicity:

second report of findings. J Natl Cancer Inst 1983;71:116575.

8 Selby JV, Friedman GD, Fireman BH. Screening prescription drugs for

possible carcinogenicity: eleven to fifteen years of follow-up. Cancer

Res 1989;49:573647.
9 Hilakivi-Clarke LWA, Lippman ME. Neonatal antidepressant treat-

ment promotes DMBA-induced mammary tumor growth. Proc Am

Assoc Cancer Res 1993;34:184.
10 Brandes LJ, Arron RJ, Bogdanovic RP et al. Stimulation of malignant

growth in rodents by antidepressant drugs at clinically relevant doses.

Cancer Res 1992;52:3796800.
11 Iishi H, Tatsuta M, Baba M, Taniguchi H. Enhancement by the tricyclic

antidepressant, desipramine, of experimental carcinogenesis in rat

colon induced by azoxymethane. Carcinogenesis 1993;14:183740.
12 Bendele RA, Adams ER, Hoffman WP, Gries CL, Morton DM.

Carcinogenicity studies of fluoxetine hydrochloride in rats and mice.

Cancer Res 1992;52:693135.
13 Tucker WE. Preclinical toxicology of bupropion: an overview. J Clin

Psychiatry 1983;44:6062.
14 Steingart AB, Cotterchio M. Do antidepressants cause, promote, or

inhibit cancers? J Clin Epidemiol 1995;48:140712.

15 Hume AL, Barbour MM, Lapane KL, Carleton RA. Is antidepressant

use changing? Prevalence and clinical correlates in two New England

communities. Pharmacotherapy 1995;15:7884.

Downloaded from http://ije.oxfordjournals.org/ at Pennsylvania State University on March 3, 2014

listed multiple medications of different types, and included

photographs of anti-inflammatories, hormones and antihistamines, substantially de-emphasizing the focus on AD. As
well, AD are currently not a widely suspected cause of breast
cancer. Even in the absence of an obvious research hypothesis,
cases may have made greater efforts to recall previous medication use or be more willing to disclose confidential information and this would tend to bias the results in favour of an
association.
It is unlikely that general over-reporting of medication use by
the cases occurred in our study, since we found that NSAID
medication use was associated with a decreased breast cancer
risk.24 In addition, in a study specifically designed to examine
the issue of recall and misclassification, Cotterchio et al.25
compared self-reported AD use with AD use recorded in
physicians records and found good agreement with physician
records (kappa = 0.60, agreement = 80%; for use of specific AD
agreement ranged from 82% to 100%) and more importantly
no differential reporting of AD use by cases and controls.
Clinical depression or other factors could have resulted in
confounding of the risk estimate. However, we found no
confounding by any of the 30 variables assessed, including
psychiatric history, such as depression or indication for use of
AD. Furthermore, epidemiological evidence does not really
support an association between depression and breast cancer
risk.2632 Multivariate adjustment of the OR for depression and
their possible risk factors did not significantly change any of the
point estimates when compared with the age-adjusted estimates,
also consistent with a lack of confounding. The inclusion of
variables that may not confound in the multivariate model
increased the width of the CI but should not rule out attention
to the more parsimonious age-adjusted model.
The absence of an association between trends in duration of
AD use and increased risk of breast cancer is not consistent with
a causal interpretation of the results. However, recall error due
to the limitations of memory, particularly for dosage and AD use
many years prior to participation in the study, rather than bias,
may reduce the likelihood of detecting an association. It is also
possible that any risk association is not linear.
Plausible biological mechanisms to explain the increased
breast cancer risk associated with AD use may involve:
increased prolactin levels, altered oestrogen metabolism, altered
metabolism of carcinogens, and other modifications of cellular
proliferation. The mechanism by which AD produce elevations in
prolactin is uncertain, but there is evidence that serotonergic
agents stimulate prolactin release directly via postsynaptic
5-HT receptors in the hypothalamus33 or by inhibition of
tubuloinfundibular dopaminergic neurons.34 Furthermore,
the interaction between AD and cytochrome P450 enzymes
may alter oestrogen metabolism or the bioactivation of
carcinogens.9,10,12,14,3538
With the increasing prevalence of use of AD and the public
health importance of breast cancer, even a small increased risk
may be of significant public health concern as this is a
potentially modifiable factor. The aetiology of breast cancer risk
is not yet understood and the majority of the risk factors
identified to date show only a modest increase in breast cancer
risk. Since breast carcinogenesis is a complex multistage process
it is likely that the effects of exposure to AD, if any, would be
difficult to detect above the background of other exposures and

965

966

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

16 Rosholm JU, Gram LF, Isacsson G, Hallas J, Bergman U. Changes in

28 Hahn RC, Petitti DB. Minnesota Multiphasic Personality Inventory-

the pattern of antidepressant use upon the introduction of the new

antidepressants: a prescription database study. Eur J Clin Pharmacol
1997;52:20509.

rated depression and the incidence of breast cancer. Cancer 1988;

61:84548.

17 Ohayon MM, Caulet M, Priest RG, Guilleminault C. Psychotropic

29 Butow PN, Hiller JE, Price MA, Thackway SV, Kricker A, Tennant CC.

medication consumption patterns in the UK general population. J Clin

Epidemiol 1998;51:27383.

Epidemiological evidence for a relationship between life events,

coping style, and personality factors in the development of breast
cancer. J Psychosom Res 2000;49:16981.

18 Feuer EJ, Wun LM, Boring CC, Flanders WD, Timmel MJ, Tong T. The

30 Price MA, Tennant CC, Butow PN et al. The role of psychosocial factors

lifetime risk of developing breast cancer. J Natl Cancer Inst 1993;

85:89297.

in the development of breast carcinoma: Part II. Life event stressors,

social support, defense style, and emotional control and their
interactions. Cancer 2001;91:68697.

19 Madigan MP, Ziegler RG, Benichou J, Byrne C, Hoover RN.

Proportion of breast cancer cases in the United States explained by

well-established risk factors. J Natl Cancer Inst 1995;87:168185.
20 Gilbart E, Kreiger N. Improvement in cumulative response rates

31 Price MA, Tennant CC, Smith RC et al. The role of psychosocial factors

in the development of breast carcinoma: Part I. The cancer prone

personality. Cancer 2001;91:67985.

following implementation of a financial incentive. Am J Epidemiol

1998;148:9799.

32 Friedman GD. Psychiatrically-diagnosed depression and subsequent

21 CYTEL Software Corporation. Egret for Windows. Cambridge, MA: 1999.

22 Colditz GA, Atwood KA, Emmons K et al. Harvard Report on Cancer

33 Nicholas L, Dawkins K, Golden RN. Psychoneuroendocrinology of

Prevention Volume 4: Harvard Cancer Risk Index. Cancer Causes

Control 2000;11:47788.

34 Arya DK. Extrapyramidal symptoms with selective serotonin reuptake

23 Maldonado G, Greenland S. Simulation study of confounder-selection

35 Shou M, Korzekwa KR, Brooks EN, Krausz KW, Gonzalez FJ, Gelboin

inflammatory drug use and breast cancer risk. Cancer Epidemiol

Biomark Prev 2001;10:121317.
25 Cotterchio M, Kreiger N, Darlington G, Steingart A. Comparison of

self-reported and physician-reported antidepressant medication use.

Ann Epidemiol 1999;9:28389.
26 Kaplan GA, Reynolds P. Depression and cancer mortality and

morbidity: prospective evidence from the Alameda County study.

J Behav Med 1988;11:113.
27 Knekt P, Raitasalo R, Heliovaara M et al. Elevated lung cancer risk among

persons with depressed mood. Am J Epidemiol 1996;144:1096103.

depression. Prolactin. Psychiatr Clin North Am 1998;21:34158.

inhibitors. Br J Psychiatry 1994;165:72833.
HV. Role of human hepatic cytochrome P450 1A2 and 3A4 in the
metabolic activation of estrone. Carcinogenesis 1997;18:20714.
36 Guengerich FP, Shimada T. Oxidation of toxic and carcinogenic

chemicals by human cytochrome P-450 enzymes. Chem Res Toxicol

1991;4:391407.
37 Jeppesen U, Gram LF, Vistisen K, Loft S, Poulsen HE, Brosen K.

Dose-dependent inhibition of CYP1A2, CYP2C19 and CYP2D6 by

citalopram, fluoxetine, fluvoxamine and paroxetine. Eur J Clin
Pharmacol 1996;51:7378.
38 Nordenberg J, Fenig E, Landau M, Weizman R, Weizman A. Effects of

psychotropic drugs on cell proliferation and differentiation. Biochem

Pharmacol 1999;58:122936.

IJE vol.32 no.6 International Epidemiological Association 2003; all rights reserved.

International Journal of Epidemiology 2003;32:966967

DOI: 10.1093/ije/dyg243

Commentary: Antidepressants and breast

cancer risk
K McPherson

Animal studies indicate that neonatal exposure to antidepressants may induce or stimulate mammary tumour growth,
and may enhance carcinogenesis in the colons of rats, and therefore might generally represent hazards for cancer in man.1 Hence
the search for a role for antidepressants in breast cancer in
human studies, for example, will no doubt continue. Meanwhile these studies may cause alarm.2 A paper from Ontario in
this issue of the International Journal of Epidemiology continues

Department of Social Medicine, University of Bristol, Canynge Hall,

Whiteladies Road, Bristol BS8 2PR, UK.

this searcharguing for a small increase in riskbased on a

large case-control study.3 The authors state that since
breast carcinogenesis is a complex multistage process it is
likely that the effects of exposure to antidepressants, if any,
would be difficult to detect above the background of other
exposure and susceptibilities.
Possibly, but maybe there really is no general causative
association.4
Clearly it is not possible to be at all certain about this relationship, and certainly not to recommend any change in the use

Downloaded from http://ije.oxfordjournals.org/ at Pennsylvania State University on March 3, 2014

strategies. Am J Epidemiol 1993;138:92336.

24 Cotterchio M, Kreiger N, Sloan M, Steingart A. Non-steroidal anti-

cancer. Cancer Epidemiol Biomarkers Prev 1994;3:1113.