Moorman, Patricia G. PhD*; Berchuck, Andrew MD†; Calingaert, Brian MS*; Halabi, Susan PhD‡; Schildkraut, Joellen M. PhD*
It has been hypothesized that antidepressants may have a growth-promoting effect on cancer.1 Experiments in rodents have suggested that antidepressants from 2 distinct chemical classes, the tricyclics and the selective serotonin reuptake inhibitors (SSRIs), enhance the growth of various types of tumors, including mammary tumors, fibrosarcoma, and melanoma.1,2 However, the experimental studies are not entirely consistent and some studies have found either no effect on cancer growth or even an antineoplastic effect.3
There are several possible mechanisms by which antidepressants could affect cancer growth, although none is well-established. Brandes et al1 reported that fluoxetine and amitriptyline stimulated DNA synthesis, suggesting these drugs could act as tumor promoters. It has also been suggested that some antidepressants suppress immune function, increasing the risk for cancer.3 A postulated mechanism that is more specific to ovarian cancer is that antidepressants affect levels of neurotransmitters such as serotonin or norepinephrine, leading to increased gonadotropin secretion.5 At this point, however, all mechanisms remain somewhat speculative.
The concern raised by the experimental studies stimulated a number of epidemiologic investigations into the possible association between antidepressants and various types of cancer, including ovarian cancer. The first published study on antidepressants and ovarian cancer4 reported a 2-fold increase in risk associated with antidepressant use, particularly use before age 50 years. In a subsequent publication on the risk of ovarian cancer associated with use of psychotropic medication, it was reported that the increased risk was confined to users of medications that operate through dopaminergic mechanisms or γ-aminobutyric acid (GABA)-ergic pathways as opposed to serotoninergic pathways.5 Two other studies reported no significant associations between antidepressant use and ovarian cancer.6,7
It is notable that past studies were limited in their ability to examine associations between the SSRI antidepressants and ovarian cancer because relatively few women in these studies had the opportunity to use this newer class of medications. However, since their introduction in the late 1980s, the SSRI antidepressants have become some of the most commonly prescribed drugs in the United States, with over 100 million prescriptions dispensed in the year 2003 (Marketos M. The top 200 brand drugs in 2003 [by units]. Drug Topics 2004;148:76). The high prevalence of SSRI use reflects the fact that they have replaced tricyclics as first-line pharmacologic treatment of depression because they generally have fewer adverse effects and are considered safer in overdose situations.8 In addition, they are used for a variety of conditions besides depression, including premenstrual syndrome, panic and anxiety disorders, and chronic headaches.8 Given the high frequency of use of SSRI antidepressants and the suggestions from experimental studies that these types of drugs may act to promote tumor growth, the issue of whether SSRIs are associated with increased cancer risk in humans remains an important unanswered question.
In this report, we examine the association between use of antidepressants and ovarian cancer in a population-based, case-control study conducted in North Carolina between 1999 and 2003. Because of the time frame in which data were collected for this study, there were adequate numbers of women who had used SSRI antidepressants to evaluate whether they were associated with an increased risk of ovarian cancer.
SUBJECTS AND METHODS
Data for these analyses came from phase I of the North Carolina Ovarian Cancer Study, a population-based, case-control study of newly diagnosed epithelial ovarian cancer cases in a 48-county area of North Carolina. These analyses are based on data from 593 cases and 628 controls. Cases were identified by using a rapid case ascertainment system developed by the North Carolina Central Cancer Registry, a statewide population-based tumor registry. Pathology reports for all ovarian cancers diagnosed in the hospitals in the study area were forwarded to the Central Cancer Registry and then to the study office within 2 months of diagnosis. Of 53 hospitals in the study area that treat ovarian cancer cases, 52 hospitals participated in the study. Eligible cases were women aged 20–74 years who were diagnosed with primary epithelial ovarian cancer between January 1, 1999, and March 31, 2003, and resided in the 48-county study area. Both invasive tumors and those with low malignant potential (borderline tumors) were included. Permission to contact the ovarian cancer patients was requested from the physician of record on the pathology report. All participants were English-speaking, mentally competent to complete an interview, and able to give informed consent. All cases underwent standardized pathologic and histologic review to confirm the diagnosis. The overall response rate among cases was 76%. Reasons cases were not interviewed included death (4%), debilitating illness (2%), patient refusal (7%), physician refusal (3%), and inability to locate the woman (8%).
Controls were identified from the same 48-county area that generated the cases by using random-digit dialing methodology. Controls were frequency-matched to cases by race and age (within 5 years). Race was based on self-report by the participant. Health Care Financing Administration tapes were initially used to identify controls aged 65–74 years, but this strategy was abandoned due to the difficulty in contacting potential controls because of the absence of telephone numbers on these tapes. Eligible controls resided in the same 48-county area as the cases, had no history of cancer, and had at least one ovary intact. As was required for the cases, controls had to be English-speaking, mentally competent to complete an interview, and able to give informed consent. Seventy-one percent of controls identified by random-digit dialing who met the eligibility criteria were successfully enrolled into the study. The study protocol was approved by institutional review boards at Duke University Medical Center and all hospitals that participated in the study.
Each case and control was interviewed in person by a trained nurse interviewer in the home of the study participant or at another convenient location. A 90-minute questionnaire was administered to obtain information on known and suspected risk factors for ovarian cancer, including family history of cancer in first- and second-degree relatives, menstrual characteristics, pregnancy and breastfeeding history, hormone and medication use, and lifestyle factors such as smoking history, alcohol consumption, physical activity, and occupational history. A life-event calendar on which major events in the woman's life, such as pregnancies and marriages, were recorded was used to aid recall. In regard to antidepressant use, women were asked if they had taken antidepressants for more than 6 months. If they responded affirmatively, they were asked to report the name of the drug, age at first use, and total duration of use. A card with pictures and names of the most commonly used antidepressants was shown to the participants to aid recall. For these analyses, we disregarded any antidepressant use that began in the year when the case was diagnosed or the control was interviewed.
Cases and controls and users and nonusers of antidepressants were compared in regard to demographic characteristics and ovarian cancer risk factors using the Student t test for continuous variables and the Pearson χ2 test for categorical variables. Unconditional logistic regression analysis was used to calculate odds ratios (ORs) and 95% confidence intervals (CIs) for the association between antidepressant use and ovarian cancer, taking into account potential confounders. Variables considered as potential confounders in the multivariable models included age (continuous), race (black/nonblack), parity (continuous), tubal ligation (yes/no), hysterectomy (yes/no), months of oral contraceptive use, history of breastfeeding (yes/no), history of ovarian cancer in a first-degree relative (yes/no), history of infertility (yes/no), history of endometriosis (yes/no), body mass index (BMI) 1 year before interview (continuous), waist-to-hip ratio (continuous), alcohol use (yes/no), smoking status (never, past, current), and educational level (less than high school, high school graduate, or greater than high school). All analyses were performed using SAS 8.2 (SAS Institute, Inc, Cary, NC).
Selected characteristics of cases and controls are shown in Table 1. Compared with controls, cases were more likely to be nulliparous, less likely to have breastfed, and less likely to have had a tubal ligation. Cases also tended to have a greater BMI and waist-to-hip ratio. We also compared characteristics of antidepressant users and nonusers among the case and control women. Characteristics that were statistically significantly associated with antidepressant use in controls included white race, history of endometriosis, and current or past smoking. Among cases, antidepressant users were more likely to be nulliparous, have a history of infertility, and be current or past smokers (data not shown).
Use of antidepressants for at least 6 months was reported by 18% of cases and 20% of controls. The SSRI class of antidepressants was the most commonly reported type of antidepressant, with approximately 52% of users reporting SSRI use only, 10% reporting tricyclic use only, 19% reporting use of other or unknown type of antidepressant, and 19% reporting use of multiple types of antidepressants. Selective serotonin reuptake inhibitors accounted for approximately 70% of the reported episodes of antidepressant use in this population.
We found no evidence that antidepressant use was associated with increased risk of ovarian cancer (Table 2). The OR for ever use of any type of antidepressant was 0.9 (95% CI 0.7–1.2), adjusting for age and race. No increased risk was observed when considering duration of use. Adjusting for multiple potential confounders including age, race, family history of ovarian cancer, number of full-term pregnancies, months of oral contraceptive use, alcohol use, smoking, BMI, waist-to-hip ratio, and history of infertility, tubal ligation, or endometriosis had a minimal effect on the ORs.
We also found no indication of increased risk associated with subcategories of antidepressants (Table 3). The ORs for ever use of SSRIs, tricyclics, other types, or multiple types of antidepressant were all 1.0 or less, with no statistically significant differences between cases and controls. The SSRI class of antidepressants was the only category that had a sufficient number of users to do meaningful comparisons by duration of use. There were no significant associations with ovarian cancer for any duration of use and no suggestion that longer duration of use increased risk.
We also evaluated the association when stratifying by age (< 50 and ≥ 50 years) and time since first use (≤ 10 and > 10 years), because these factors had previously been reported as factors that modify the relationship between ovarian cancer and antidepressants.4 Similar to the analyses in the entire study population, there were no statistically significant relationships with antidepressant use in any of these strata (data not shown).
In this population-based, case-control study, we found no evidence that antidepressants are associated with an increased risk of ovarian cancer. No association was apparent, whether we considered all antidepressants combined or individual categories of antidepressants. We also found no association with longer duration of use. With our sample size of 593 cases and 628 controls and α = 0.05, we had greater than 80% power to detect an OR of 1.5. Therefore, our study can exclude with some confidence even a modest increase in risk associated with antidepressant use.
An inherent limitation of our study is that information on past use of antidepressants was self-reported. However, in a large population-based study that covers a wide geographic area, self-report is the only feasible method of obtaining this information because it would be logistically impossible to get pharmacy or medical record data from the hundreds of providers who served this patient population. In addition, it must be recognized that medical record and pharmacy data also are imperfect in that they reflect what was prescribed but not necessarily what patients took. Although self-reported information on antidepressant use has its limitations, it was the most practical way to obtain this information in a large study population.
It is also possible that women who suffered from depression were less likely to have agreed to take part in the study, which could have led to an underestimate of the prevalence of antidepressant use in the population. Because participation rates were lower among controls than cases, this underestimate may have been more pronounced in controls than in cases and, in theory, could have led to an overestimate of the association between antidepressant use and ovarian cancer. Our overall findings were null, however, which argues against the possibility that nonresponse could have substantially influenced our results.
In contrast to earlier studies,4–7 SSRIs were the most commonly used class of antidepressant in our study population, reflecting more recent prescribing patterns. The overall lack of association between antidepressants and ovarian cancer is consistent with the findings reported by Coogan et al6 in a hospital-based, case-control study involving 748 ovarian cancer cases and Dublin et al7 in a case-control study conducted within a health maintenance organization with 314 cases. However, few women in these studies reported taking SSRIs (only 1% in the report by Coogan et al6 and “very few” in the report by Dublin et al7). In our study population, approximately 70% of the reported episodes of antidepressant use were SSRIs, allowing us to conclude that the lack of association between antidepressants and ovarian cancer extends to this newer class of drugs. Although Harlow et al5 reported increased risk for ovarian cancer with certain types of psychotropic medications in a population-based, case-control study involving 563 cases, they found no association with agents that worked through serotonin or serotonin/norepinephrine pathways. These pathways include the SSRI antidepressants, as well as some of the most common tricyclic antidepressants such as amitriptyline and imipramine. The majority of antidepressant use in our study population was the SSRI type; therefore, the lack of association that we observed in our study is consistent with the findings reported by Harlow et al5
The hypothesis that antidepressants could promote cancer growth had important implications in terms of the etiology and treatment of cancer and certainly merited further investigation. As reported in our study and others,7,9 antidepressants have been taken by as much as 20% of the female population. Given this high prevalence of use, even a small increase in cancer risk associated with their use could translate into a large number of cancer cases at the population level. A growth-promoting effect of antidepressants could also have important implications in the treatment of cancer patients. Antidepressants are commonly prescribed after the diagnosis of cancer, not only for the treatment of depression but also for pain management.8,10 Of particular relevance for patients who have their ovaries removed as part of their cancer treatment, certain types of antidepressants have been used as treatment for hot flushes in women for whom estrogen is contraindicated.11,12 The majority of the evidence to date is reassuring that antidepressants are not likely to have a significant impact on ovarian cancer, and the decision to prescribe them should not be based on concerns about cancer risk.
Overall, the evidence from our study and others does not suggest that women who take antidepressants are at significantly increased risk for ovarian cancer. The evidence seems particularly consistent in regard to the use of SSRIs, which are currently the most commonly prescribed class of antidepressant. Given the lack of compelling evidence linking antidepressants and ovarian cancer, the demonstrated benefits of antidepressants in the treatment of depression, premenstrual syndrome, pain, and menopausal symptoms appear to outweigh possible cancer risks associated with their use.
1. Brandes LJ, Arron RJ, Bogdanovic RP, Tong J, Zaborniak CL, Hogg GR, et al. Stimulation of malignant growth in rodents by antidepressant drugs at clinically relevant doses. Cancer Res 1992;52:3796–800.
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;54:229–36.
3. Steingart AB, Cotterchio M. Do antidepressants cause, promote, or inhibit cancers? J Clin Epidemiol 1995;48:1407–12.
4. Harlow BL, Cramer DW. Self-reported use of antidepressants or benzodiazepine tranquilizers and risk of epithelial ovarian cancer: evidence from two combined case-control studies (Massachusetts, United States). Cancer Causes Control 1995;6:130–4.
5. Harlow BL, Cramer DW, Baron JA, Titus-Ernstoff L, Greenberg ER. Psychotropic medication use and risk of epithelial ovarian cancer. Cancer Epidemiol Biomarkers Prev 1998;7:697–702.
6. Coogan PF, Rosenberg L, Palmer JR, Strom BL, Stolley PD, Zauber AG, et al. Risk of ovarian cancer according to use of antidepressants, phenothiazines, and benzodiazepines (United States). Cancer Causes Control 2000;11:839–45.
7. Dublin S, Rossing MA, Heckbert SR, Goff BA, Weiss NS. Risk of epithelial ovarian cancer in relation to use of antidepressants, benzodiazepines, and other centrally acting medications. Cancer Causes Control 2002;13:35–45.
8. Drug facts and comparisons 2004. 58th ed. St. Louis (MO): Wolters Kluwer Health; 2004. p. 1024-71.
9. Moorman PG, Grubber JM, Millikan RC, Newman B. Antidepressant medications and their association with invasive breast cancer and carcinoma in situ of the breast. Epidemiology 2003;14:307–14.
10. Lucas LK, Lipman AG. Recent advances in pharmacotherapy for cancer pain management. Cancer Pract 2002;10(supp 1):S14–20.
11. Loprinzi CL, Kugler JW, Sloan JA, Mailliard JA, LaVasseur BI, Barton DL, et al. Venlafaxine in management of hot flashes in survivors of breast cancer: a randomized controlled trial. Lancet 2000;356:2059–63.
12. Pritchard KI. Hormone replacement in women with a history of breast cancer. Oncologist 2001;6:353–62.
Figure. No caption available.