It has been hypothesized that antidepressants may have a growth-promoting effect on breast cancer. This hypothesis is based on the similarity of the chemical structures of certain antidepressants and DPPE (N,N-diethyl-2-[4-phenylmethyl] phenoxyl ethanamine), a compound known to stimulate mammary tumor growth and development in animals. 1 Experiments in rodents showed enhanced growth of mammary tumors in animals treated with antidepressants from two distinct chemical categories, the tricyclics and the selective serotonin reuptake inhibitors (SSRIs), at clinically relevant doses. 1,2 Additional support for the hypothesis that antidepressants have effects on the breast comes from the observation that reported side effects (albeit at low frequency) include lactation, breast enlargement, breast pain and increases in prolactin concentrations. 3,4
Epidemiologic studies that have examined the association between antidepressant use and breast cancer have produced inconsistent results. Increased risks were not observed in some studies, 5–8 whereas others found increased risk only in particular subgroups, as defined either by a specific drug or drug category, or by duration of use. 9–11 To further explore these relations, we examined the associations of antidepressant use with the risk of invasive breast cancer and of carcinoma in situ of the breast, using data from a case-control study conducted in North Carolina.
We analyzed data from the Carolina Breast Cancer and Carcinoma in situ Study, a population-based, case-control study conducted in North Carolina between 1993 and 2000. 12 Information on antidepressant use was collected from women interviewed during Phase 2 of the study, between 1996 and 2000. A rapid case ascertainment system identified invasive breast cancer and carcinoma in situ cases age 20 to 74 years with no prior history of breast cancer who resided in a 24-county area of central and eastern North Carolina. Carcinoma in situ cases included ductal carcinoma in situ, lobular carcinoma in situ, mixed ductal and lobular carcinoma in situ and any carcinoma in situ with microinvasion up to 2 mm. Among invasive breast cancer cases, we oversampled African Americans and women under age 50, using a modification of randomized recruitment, to increase the statistical power for examining race and age differences. 13 The intent of the oversampling was to obtain roughly equal numbers of African-American and white women and of women diagnosed before age 50 and at age 50 or older. All in situ cases were eligible, with no oversampling by race or age. We selected control women from lists at the Division of Motor Vehicles (for women age 20 to 64 years) and Health Care Financing Administration (for women age 65 to 74 years); they were approximately frequency matched to cases by age and race. Separate control groups were created for the invasive cases and the carcinoma in situ cases, taking into account the different age and race distributions of the two case groups.
We mailed a description of the study to those cases for whom we received physician consent and all controls; we then contacted them by telephone. Those who agreed to participate were subsequently interviewed in person by registered nurses. The interviewer administered a standardized questionnaire, took body measurements and drew a blood sample. Of the women identified as eligible invasive breast cancer cases, over 98% were located and contacted, and 79% of these agreed to participate. Among women identified as potential controls for the invasive cases, 81% were located and contacted, and 73% of these agreed to participate. Contact rates were 99% for the carcinoma in situ cases, and 83% of contacted women were interviewed. Among women identified as potential controls for the in situ cases, 91% were contacted and 73% were interviewed. The analyses in this report are based on 938 invasive cases and 771 controls, and 507 carcinoma in situ cases and 455 controls. The vast majority of participants were either African American or white, with fewer than 2% of women from other racial groups. For race-specific analyses, participants were categorized as African American or white, with all women other than African Americans included with the white women. Forty-eight percent of the invasive cases and 50% of their controls were African-American, and 21% of the carcinoma in situ cases and 15% of their controls were African-American.
We asked for information on established and suspected breast cancer risk factors, including reproductive and menstrual characteristics, hormone use, family history of cancer, alcohol use and smoking history. Women were also queried about their use of antidepressants over the past 10 years. To aid their memory, women were prompted with indications for antidepressant use (eg, depression, anxiety, insomnia, headaches or bladder problems) and were shown a pictorial display of the most commonly used antidepressants. Women who said that they had taken antidepressants were asked to report the name and dose of the drug, the condition being treated, how frequently they took the drug and total duration of use. Cases were queried about use both before and after diagnosis, although the analyses presented here are limited to use before diagnosis. Controls were asked to report antidepressant use before the date they were selected for inclusion in the study. For purposes of analysis, we considered women to be ever-users of antidepressants if they reported 3 or more months of use. Women were further categorized by type of antidepressants taken: tricyclics (eg, amitriptyline, imipramine), SSRIs (eg, fluoxetine, paroxetine), atypical (eg, trazodone, bupropion), monoamine oxidase inhibitors, lithium or multiple types.
We performed logistic regression analyses with the GENMOD procedure in SAS (Cary, NC), which allows for the incorporation of an offset term to take into account the sampling probabilities used to select cases and controls. Variables evaluated as potential confounders in the models included age, race (African American or white), age at menarche, age at first full-term pregnancy, lactation history, menopausal status, family history of breast cancer in a first-degree relative, oral contraceptive use, hormone replacement therapy, educational level, body mass index, waist-to-hip ratio, alcohol consumption and smoking history. We used multivariate models to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for invasive breast cancer or carcinoma in situ associated with use of antidepressants, controlling for potential confounders.
We examined characteristics that were associated with antidepressant use among control women to identify factors that potentially could act as confounders of the association between antidepressant use and breast cancer. Table 1 presents breast cancer risk factors associated with antidepressant use among control women for the invasive and carcinoma in situ components of the study. Generally, the same factors were predictors of antidepressant use among both control groups; these variables were white race, perimenopausal status, ever-use of hormone replacement therapy, high body mass index, high waist-to-hip ratio and current or former smoking. Early age at menarche, low educational achievement and ever-use of alcohol were associated with antidepressant use only among the carcinoma in situ controls, and early age at first pregnancy only among the invasive controls.
Characteristics of cases and controls for the invasive breast cancer and carcinoma in situ components of the study are presented in Table 2. Among women in the invasive breast cancer study, the following variables were associated with breast cancer: early age at menarche, nulliparity, no history of breastfeeding, family history of breast cancer, past smoking and high waist-to-hip ratio. Nulliparity, family history of breast cancer, higher education, low body mass index, low waist-to-hip ratio and alcohol consumption were associated with carcinoma in situ. There was no association with oral contraceptives or hormone replacement therapy in either group.
Table 3 presents the associations between antidepressant use and invasive breast cancer. Ever-use of antidepressants was reported by approximately one-fifth of both cases and controls. There was no indication of increased risk associated with ever-use of antidepressants (age- and race-adjusted OR = 1.0; CI = 0.7–1.2). There was no consistent relation between duration of antidepressant use and breast cancer. Examination of individual categories of antidepressants showed no increased risk of breast cancer associated with ever-use of any type of antidepressant; ORs ranged from 0.8 to 1.0 among women taking tricyclics, SSRIs, atypical or multiple types of antidepressants. For duration of use for individual categories of antidepressants, there was a suggestion that use of SSRIs for 36 months or longer increased risk (age- and race-adjusted OR = 2.2; CI = 0.8–6.3); however, there were only 13 cases and 5 controls with this level of exposure, and so these results should be interpreted cautiously.
Factors that were associated with antidepressant use (see Table 1) were evaluated as potential confounders of the relation between antidepressants and breast cancer. Adjusting for confounders individually or jointly did not result in ORs that were markedly different from the age- and race-adjusted ORs, with the only suggestion of increased risk being among women with longer duration of SSRI use (multivariate-adjusted OR = 2.7; CI = 0.9–7.9). The associations between any antidepressant use and invasive breast cancer were also evaluated in strata defined by race (African-American or white) and age (<50 or ≥50 years). Results were very similar across all age and race categories, with multivariate-adjusted ORs for ever-use of antidepressants between 0.8 and 1.1 for all strata (data not shown).
The associations between antidepressants and carcinoma in situ of the breast are shown in Table 4. Carcinoma in situ cases were less likely than controls to have reported any use of antidepressants (age- and race-adjusted OR = 0.6; CI = 0.4–0.8). ORs less than 1.0 were also observed for all durations of use, but there was no evidence of a dose-response relation with duration. ORs less than or equal to 1.0 were observed for any use of each category of antidepressant use, and there was no evidence of increased risk associated with long-term use within specific categories of antidepressants. The race-specific, age-adjusted ORs for ever-use of antidepressants and carcinoma in situ were 0.8 (CI = 0.4–1.9) among African-American women and 0.6 (CI = 0.4–0.8) among white women. The small number of African-American women who had taken antidepressants precluded doing race-specific analyses adjusting for the full list of potential confounders.
We observed only slight evidence of an association between antidepressant use and breast cancer. Among women with invasive breast cancer, the use of antidepressants was unrelated to breast cancer overall. There was a modest positive association among women who had taken SSRI antidepressants for 3 years or longer. Although the multivariate OR for this exposure was 2.7, this estimate was based on a small number of cases and controls with this level of exposure and should be interpreted with caution. Nonetheless, these findings suggest that continued evaluation of the relation between SSRIs and breast cancer may be warranted because other published studies have not included women with prolonged exposure to SSRIs.
We found that women with carcinoma in situ of the breast were less likely than control women to report use of antidepressants. This inverse association with carcinoma in situ of the breast was an unexpected finding. Although experimental evidence is more supportive of antidepressants increasing growth of mammary tumors, growth inhibitory effects of various antidepressants have been reported in in vivo studies in rodents and in vitro studies of various cells lines. 1,14,15 A bell-shaped relation between fluoxetine dose and cell growth has been described by Brandes and colleagues, 14 with growth promotion occurring at moderate doses and growth inhibition at higher doses. However, the cytostatic effects in these experiments occurred at doses that were higher than typical therapeutic doses in humans. Despite biological plausibility for a growth inhibitory effect at high doses, we believe the reduced risk of carcinoma in situ observed in this study was most likely a chance finding. The proportion of women reporting ever-use of antidepressants was much higher among the controls for the in situ cases (29%) as compared with the lower (and quite similar) proportions (19% to 22%) of ever-users among invasive breast cancer patients, carcinoma in situ cases and controls for the invasive cases. The lack of a dose-response relation between duration of use and carcinoma in situ also argues against a causal relation.
Our findings are generally consistent with most previous epidemiologic studies of antidepressants and breast cancer, with little evidence to link this class of drugs as a whole to an increased risk of cancer, but some suggestions of increased risk within subgroups. 5–11 Reports of positive associations between antidepressants and breast cancer have not been consistent across studies in regard to type of drug or duration of use. Sharpe et al. 11 reported increased risk associated with heavy exposure to tricyclic antidepressants after a delay of more than 10 years. Kelly et al. 10 found no increased risk associated with tricyclic antidepressant use, but a suggestion of increased risk associated with recent use of SSRIs. In another report, 9 SSRIs were not generally associated with breast cancer, but a strong association was found with one specific SSRI (paroxetine), based on a small number of exposed women. This study also found some evidence for increased risk of breast cancer among women who had taken tricyclic antidepressants for more than 2 years. In contrast to these studies with modest suggestions of increased risk for particular subgroups of users, other investigators have concluded that there is no association between antidepressant use and breast cancer. 5–8 Overall, our findings do not suggest an increased risk associated with most antidepressant use, but our results raise the possibility that long-term use of SSRI antidepressants may increase risk.
The relatively low response rates among control women in our study raise concerns as to whether nonresponse could have affected the findings. We did not have information from nonrespondents regarding antidepressant use, although one might speculate that women taking antidepressants may be less likely to take part in a research study because their depression may make it difficult to have the type of interaction required for an in-person interview. If this scenario were true, we would have underestimated the prevalence of antidepressant use in our underlying population among both cases and controls, but perhaps to a greater extent among the controls because of their lower response rates. This would have led to an overestimate of the association between antidepressant use and breast cancer. If nonresponse bias were present, we expect that we would have found a stronger positive association with antidepressant use. Given that our overall findings were null, with an association between antidepressants and breast cancer in only one small subgroup, we do not believe our findings can be explained by nonresponse bias.
Even though the studies to date have not provided strong and consistent evidence of an association between antidepressant use and breast cancer, continued investigation of this association is warranted for several reasons. Most studies to date have had limited statistical power to examine associations between breast cancer and specific classes of antidepressants. Antidepressant drugs fall into several distinct chemical classes, with tricyclics and SSRIs being the major categories. 16 Although animal experiments suggest an increased risk of mammary cancers for both tricyclic and SSRI antidepressants, 1–2 it is unlikely that their effects on breast cancer are identical and, ideally, different chemical classes of antidepressants should be considered as separate exposures. In addition, SSRI antidepressants were first marketed in the United States in 1993 and the studies to date have had too few users (and no long-term users) for a thorough evaluation of the risk associated with these drugs. Our findings suggesting a positive association with longer-term use of SSRI antidepressants underscore the importance of continuing to evaluate this class of drugs in relation to breast cancer risk. Future studies also should consider the possibility of interactions between antidepressants and other exposures. Animal studies do not suggest that antidepressants have a direct carcinogenic effect. 17 Rather, the experiments show a growth promoting effect of antidepressants on mammary tumors induced by dimethylbenzanthracene (DMBA). 1,2 It is conceivable that antidepressants have an adverse effect on breast tumor growth only in the presence of other carcinogenic factors that are yet to be defined.
The association between antidepressants and breast cancer risk should continue to be monitored for clinical reasons as well. Antidepressants are among the most commonly prescribed drugs in the United States, with approximately 100 million prescriptions dispensed annually. 18 Three antidepressants are among the top 25 most frequently prescribed drugs. 19 The numbers of prescriptions reflect not only that depression is the most common psychiatric disorder among women, 20 but also that these medications are used for a number of other indications, including premenstrual syndrome, chronic pain and insomnia. 16 If these drugs are associated with even a modest increase in breast cancer risk, there could be substantial public health implications given their widespread use.
Concern about a potential detrimental effect of antidepressants on breast cancer also extends to the use of antidepressants by women who have already been diagnosed with breast cancer. Antidepressants are commonly prescribed to cancer patients, not only for depression but also for pain management. Among the cases in this study, approximately 14% reported taking antidepressants after diagnosis. In addition, antidepressants have recently been evaluated in clinical trials as a treatment for hot flushes in women such as breast cancer patients for whom estrogen is contraindicated. 21,22 Although antidepressants seem to be effective in treating these symptoms, the potential for these drugs to increase the likelihood of recurrence of the cancer needs to be evaluated and weighed against the possible benefits.
Overall, the epidemiologic evidence to date does not present a consistent and coherent picture of increased risk associated with antidepressant use, either in general, with specific categories of drugs or with defined patterns of use. However, the potential adverse effects of antidepressants on cancer risk and recurrence need continued monitoring because of their high prevalence of use in the general population.
1. 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: 3796–3800.
2. Hilakivi-Clarke L, Wright A, Lippman ME. DMBA-induced mammary tumor growth in rats exhibiting increased or decreased ability to cope with stress due to early postnatal handling or antidepressant treatment. Physiol Behav 1993; 54: 229–236.
3. Marken PA, Haykal RF, Fisher JN. Management of psychotropic-induced hyperprolactinemia. Clin Pharm 1992; 11: 851–856.
4. Physician’s Desk Reference, 56th Edition
. Montvale, NJ: Medical Economics Company, 2002;1242, 2714.
5. 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: 728–734.
6. Wallace RB, Sherman BM, Bean JA. A case-control study of breast cancer and psychotropic drug use. Oncology 1982; 39: 279–283.
7. Weiss SR, McFarland BH, Burkhart GA, Ho PTC. Cancer recurrences and second primary cancers after use of antihistamines or antidepressants
. Clin Pharmacol Ther 1998; 63: 594–599.
8. Selby JV, Friedman GD, Fireman BH. Screening prescription drugs for possible carcinogenicity: eleven to fifteen years of follow-up. Cancer Res 1989; 49: 5736–5747.
9. Cotterchio M, Kreiger N, Darlington G, Steingart A. Antidepressant medication use and breast cancer risk. Am J Epidemiol 2000; 151: 951–957.
10. 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: 861–868.
11. Sharpe CR, Collet J-P, Belzile E, Hanley JA, Boivin J-F. The effects of tricyclic antidepressants
on breast cancer risk. Br J Cancer 2002; 86: 92–97.
12. Newman B, Moorman PG, Millikan R, et al
. The Carolina Breast Cancer Study: integrating population-based epidemiology and molecular biology. Breast Cancer Res Treat 1995; 35: 51–60.
13. Weinberg CR, Sandler DP. Randomized recruitment in case-control studies
. Am J Epidemiol 1991; 134: 421–432.
14. Brandes LJ, Cheang M. Response to antidepressants
and cancer: cause for concern? [Letter]. J Clin Psychopharmacol 1995; 15: 84–85.
15. Steingart AB, Cotterchio M. Do antidepressants
cause, promote, or inhibit cancers? J Clin Epidemiol 1995; 48: 1407–1412.
16. Drugs Facts and Comparisons, 56th Edition
. St. Louis: Wolters Kluwer Company, 2002;929.
17. Bendele RA, Adams ER, Hoffman WP, Gries CL, Morton DM. Carcinogenicity studies of fluoxetine hydrochloride in rats and mice. Cancer Res 1992; 52: 6931–6935.
19. Drug Topics
. Top 200 brand-name drugs by units in 2001. 4 March 2002. Available at: http://www.drugtopics.com
. Accessed 21 March 2002.
20. Kessler RC, McGonagle KA, Zhao S, et al
. Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States. Arch Gen Psych 1994; 51: 8–19.
21. Loprinzi CL, Kugler JW, Sloan JA, et al
. Venlafaxine in management of hot flashes in survivors of breast cancer: a randomised controlled trial. Lancet 2000; 356: 2059–2063.
22. Pritchard KI. Hormone replacement in women with a history of breast cancer. Oncologist 2001; 6: 353–362.