Use of Nonsteroidal Antiinflammatory Drugs and Risk of Breast Cancer: The Case-Control Surveillance Study Revisited

Abstract

Several studies have suggested that use of nonsteroidal antiinflammatory drugs (NSAIDs) may reduce the risk of breast cancer. Reductions in risk may vary according to the hormone receptor status of the tumor or the type of NSAID used. The authors extended a previous US hospital-based case-control study (the Case-Control Surveillance Study) to include 444 additional cases, for a total of 7,006 incident breast cancer cases (1976–2002). They examined the relation between regular NSAID use and breast cancer risk using logistic regression to adjust for confounding. The odds ratio for regular use of NSAIDs was 0.78 (95% confidence interval: 0.63, 0.97), and a trend of decreasing risk with increasing duration of use was statistically significant (p for trend = 0.02). The inverse association with regular use of NSAIDs was stronger among premenopausal women (odds ratio = 0.62). The overall odds ratios for regular use of aspirin and ibuprofen were 0.86 and 0.85, respectively. The effect of NSAID use on breast cancer risk did not vary according to the hormone receptor status of the tumor. In conclusion, long-term regular use of NSAIDs was associated with decreased risk of breast cancer. The type of NSAID used or the hormone receptor status of the tumor did not modify the effect.

Numerous epidemiologic studies have examined the relation of nonsteroidal antiinflammatory drug (NSAID) use to the risk of breast cancer (1–16). While some have suggested that regular NSAID use is associated with a decreased risk of breast cancer (3, 5, 8, 10, 12, 14–16), others have failed to confirm such a relation (1, 2, 4, 6, 7, 11, 13). In a meta-analysis, Khuder and Mutgi (17) calculated an 18 percent decreased risk of breast cancer among regular NSAID users based on six cohort studies and a 13 percent decreased risk based on eight case-control studies. Recently, two studies (16, 18) provided additional evidence that regular use of NSAIDs, especially ibuprofen or aspirin, is inversely associated with the risk of breast cancer. In the study of aspirin, the reduction in risk was strongest for hormone receptor-positive tumors (18).

In an earlier report based on 6,558 breast cancer patients recruited into the Case-Control Surveillance Study from 1976 to 1996, we found equivocal evidence of an association between NSAID use and breast cancer risk (9). There was a decreasing odds ratio with increasing duration of use, but it was found mostly at one study center. In the current analysis, we extended the time of subject recruitment to the year 2002 and the number of cases to 7,006. In addition to assessing the effect of all NSAID use, we also examined the effect of individual NSAIDs and whether the relation of NSAID use to breast cancer risk varied according to the hormone receptor status of the tumor or menopausal status. The use of NSAIDs, especially nonaspirin NSAIDs, has increased in the United States because of the increasing use of over-the-counter drugs; thus, the data that were collected more recently increased our ability to examine the effects of long-term NSAID use and type of NSAID use on the risk of breast cancer.

MATERIALS AND METHODS

Initiated in 1976, the ongoing Case-Control Surveillance Study has been conducted in hospitals in four US cities (Boston, Massachusetts; New York, New York; Philadelphia, Pennsylvania; and Baltimore, Maryland). Subjects were interviewed in the hospital by trained nurse interviewers using a structured questionnaire. The nurse interviewers identified potentially eligible patients through examination of admission lists and ward logs. To guard against potential selection bias from referrals to the hospital, the interviewers enrolled only patients who lived in areas that were within an hour's drive of the hospital; to ensure that this criterion was met, the interviewers were supplied with lists of acceptable ZIP codes. Data included in the present study were collected from 1976 to 2002; all patients interviewed since the publication of our previous report (9) were from the Philadelphia center. From 1976 through 1997, the participation rate among patients targeted for interview was 95 percent. Since 1998, 88 percent of patients approached have participated. The study was approved by the institutional review boards of all participating institutions.

Information was collected on demographic factors, medical and reproductive history, family history of cancer, lifestyle factors, and medication use. Histories of medication use were elicited through questions about 42 indications, including those for which NSAIDs are used (e.g., pain, headache, and arthritis). For each episode of use, the name of the medication and the duration, timing, and frequency of use were recorded.

After discharge, the diagnosis that led to the patient's hospital admission was abstracted from the hospital record; discharge summaries were obtained for all patients, and pathology reports were obtained for patients with cancer. Information on hormone receptor status (estrogen receptor and progesterone receptor) was obtained for 1,294 breast cancer cases. These cases tended to be patients interviewed in the later years of the study, because hormone receptor status was not usually reported in the earlier years.

NSAID use

NSAID use was defined as use of any drug in the following classes: salicylates (e.g., aspirin), indoles (e.g., indomethacin), propionic acids (e.g., ibuprofen), fenamates (e.g., mefenamic acid), pyrazolines (e.g., phenylbutazone), oxicams (e.g., piroxicam), and cyclooxygenase-2 inhibitors (e.g., celecoxib). Most NSAID use was sporadic. We theorized that if NSAID use had a preventive role, it would most likely be regular use of long duration. We defined regular NSAID use as use of any NSAID at least four times per week for 3 or more continuous months. All other use was considered nonregular. Regular use was further subdivided according to when NSAIDs were first and last used. Only use that began a year or more before hospital admission was considered etiologically relevant.

Cases

Eligible cases were 7,006 women aged 30–79 years who met the following criteria: 1) a diagnosis of breast cancer recorded in the discharge summary or pathology report that had been made within the year before the current admission and 2) no other primary cancer or history of cancer.

Controls

Controls were selected from a pool of 3,906 women aged 30–79 years with no history of cancer who had been admitted to the hospital for nonmalignant diseases that we considered unrelated to NSAID use. Eligible diagnoses included appendicitis, hernia of the abdominal cavity, and traumatic injury. Subjects with missing information on NSAID use were not eligible as potential controls. The potential controls were frequency matched to the cases at a ratio of up to 4:1 on 5-year age group, study center (New York, Boston, Philadelphia, or Baltimore), and year of interview (in 5-year categories). The final control series comprised 3,622 women. In our previous study (9), we also used a control series of women with cancer. However, because cancers at various sites have been linked to NSAID use, we did not use cancer controls in the present study.

The prevalences of regular NSAID use that began at least 1 year before admission, standardized by age and geographic region, were similar among controls admitted for appendicitis or hernia of the abdominal cavity and controls admitted for traumatic injury. The prevalences of never use of NSAIDs, nonregular use of NSAIDs, and regular use of NSAIDs were 37.0 percent, 51.5 percent, and 11.5 percent, respectively, in the first group and 37.8 percent, 50.6 percent, and 11.5 percent, respectively, in the second group. The two subgroups of controls were also similar in terms of mean number of years of regular NSAID use (5.2 years vs. 6.1 years).

Data analysis

We assessed the relation of NSAID use to the risk of breast cancer using multiple logistic regression models. In a multivariable model, we adjusted for age, year of interview, study center, race, years of education, benign breast disease, number of physician visits 2 years before hospitalization, duration of female hormone supplement use, duration of oral contraceptive use, age at menarche, age at menopause, age at first birth, parity, alcohol consumption, family history of breast cancer (breast cancer in a mother or sister), practice of breast self-examination, and body mass index. We included a term for duration of regular NSAID use in the model to test for a trend across duration of use. We assessed regular use of all NSAIDs and use of ibuprofen and aspirin specifically. In all analyses, the reference group comprised women who had never used any NSAIDs.

We assessed the effect of regular NSAID use according to the hormone receptor status of the tumor and menopausal status. As in the previous study of NSAIDs and tumor receptor status (18), a tumor was classified as hormone receptor positive if it was positive for either estrogen receptor or progesterone receptor; otherwise, it was classified as hormone receptor negative.

RESULTS

Data on the relation of NSAID use to the risk of breast cancer are presented in table 1. For the most etiologically relevant category—regular use of NSAIDs that began at least 1 year before hospital admission—the odds ratio was 0.78 (95 percent confidence interval (CI): 0.63, 0.97). The odds ratio did not vary according to whether use continued into the year before admission or was discontinued before that. As table 2 shows, a longer duration of regular NSAID use was associated with a lower risk of breast cancer (p for trend = 0.02), and the odds ratio for regular use was smallest for persons with the longest duration of regular NSAID use (odds ratio = 0.62, 95 percent CI: 0.28, 1.35).

Data on the association of regular use of aspirin and ibuprofen with the risk of breast cancer are presented in table 3. Compared with persons who had never used any NSAIDs, the overall odds ratio was 0.86 (95 percent CI: 0.64, 1.16) for women who used aspirin regularly and 0.59 (95 percent CI: 0.25, 1.36) for women whose duration of use totaled 20 or more years (p for trend = 0.32). The odds ratio for regular use of ibuprofen was 0.85 (95 percent CI: 0.53, 1.37). The odds ratio for the longest duration category for which numbers were sufficient—5 or more years of use—was 0.78 (95 percent CI: 0.29, 2.08).

The relations of regular use of NSAIDs, aspirin, and ibuprofen to breast cancer risk according to the hormone receptor status of the tumor are shown in table 4. The odds ratios for regular use of NSAIDs overall and aspirin separately did not vary by hormone receptor status. The odds ratio for regular use of ibuprofen was 0.94 for receptor-positive tumors and 0.45 for receptor-negative tumors, but the estimates were compatible with a uniform value.

As table 5 shows, the relations of regular use of NSAIDs, aspirin, and ibuprofen to the risk of breast cancer appeared to be modified by menopausal status. Among premenopausal women, the odds ratios for regular use of NSAIDs, aspirin, and ibuprofen were 0.62, 0.67, and 0.61, respectively. The corresponding odds ratios among postmenopausal women were 0.90, 0.98, and 0.94. However, none of the interaction terms were statistically significant (all p's > 0.21).

In the previous study, the reduction in odds ratio for regular NSAID use was present only at the Boston center, not in the data from Philadelphia, New York, or Baltimore. All of the newly added cases and controls were from the Philadelphia center, where a significant inverse association between regular use of NSAIDs and risk of breast cancer became evident (odds ratio = 0.67, 95 percent CI: 0.47, 0.95).

DISCUSSION

Our results suggest a modest decrease in the risk of breast cancer associated with regular NSAID use. The inverse association was more evident among premenopausal women. We did not confirm findings from the Women's Health Initiative Observational Study which suggested that long-term use of ibuprofen reduced breast cancer risk to a greater extent than did use of aspirin (16). Likewise, we did not confirm findings from the Long Island Breast Cancer Study Project that a reduction in breast cancer risk among regular aspirin users was confined to hormone receptor-positive tumors (18).

To date, epidemiologic findings regarding the effect of NSAID use on breast cancer risk have been equivocal. While the majority of case-control studies have shown that regular use of NSAIDs is associated with decreased risk of breast cancer, results from cohort studies are conflicting. Of seven cohort studies (1–3, 6, 8, 16, 19) that evaluated the association, only two had more than 100 incident breast cancer cases (6, 16). In the Nurses' Health Study (2,414 incident breast cancer cases), no relation was found between aspirin use and the risk of breast cancer (odds ratio = 1.01, 95 percent CI: 0.80, 1.27) (6). Investigators from the Women's Health Initiative Observational Study (1,252 incident breast cancer cases) reported relative risks of 0.79 (95 percent CI: 0.60, 1.04) for women who used NSAIDs regularly for 5–9 years and 0.72 (95 percent CI: 0.56, 0.91) for women who used NSAIDs regularly for 10 years or more (16). The effect appeared to be stronger for ibuprofen use (relative risk = 0.51, 95 percent CI: 0.28, 0.96) than for aspirin use (relative risk = 0.79, 95 percent CI: 0.60, 1.03) (16).

To our knowledge, no epidemiologic study has assessed whether the relation between regular NSAID use and breast cancer risk varies according to menopausal status. In the Nurses' Health Study, the relative risk of breast cancer for regular aspirin use among women aged 34–49 years was 0.79 (95 percent CI: 0.43, 1.45), whereas the corresponding effect estimate among women aged 50 years or more was 1.06 (95 percent CI: 0.83, 1.35) (6). These results suggest that regular use of aspirin is associated with lower risk of breast cancer among younger women, and probably most of them were premenopausal. Nevertheless, the number of cases was small (99 cases), and the confidence interval was wide. In our study, the breast cancer odds ratios for regular use of NSAIDs were 0.69 and 0.90 among pre- and postmenopausal women, respectively. It is unclear why an effect of regular NSAID use would vary by menopausal status. The effect of NSAID use on the risk of breast cancer is relatively small, and it is possible that a weak association was simply more likely to be detected among low-risk women, that is, premenopausal women (20).

The inconsistent findings among the different studies may reflect differences in the definitions of regular NSAID use and of nonuse. For example, in the Nurses' Health Study, regular use of aspirin was defined as two or more aspirin taken per week, and the reference group included both nonregular users and never users. The Women's Health Initiative used the same definition of regular NSAID use as the Nurses' Health Study, but the referent group included women who had used NSAIDs regularly for less than 1 year in addition to nonregular and never users (16). The definition of regular use of NSAIDs in the Long Island Breast Cancer Study Project (18) was similar to that in our study. However, it is unclear what the reference group was when the effect of each individual NSAID was assessed. In addition, the studies that reported an inverse association between NSAID use and breast cancer risk (3, 5, 8, 10, 12, 14–16) assessed the dose-response relations of frequency of use and duration of use separately. None addressed this issue with a variable that combines both frequency of use and duration of use. Thus, there is a lack of informative evidence on a possible dose-response relation between NSAID use and risk of breast cancer.

Several biologic mechanisms have been proposed to explain why NSAID use may reduce the risk of breast cancer. Studies have demonstrated that NSAIDs inhibit the synthesis of prostaglandins through interference with cyclooxygenase-2. Prostaglandins play a key role in the proliferation of tumor tissue (21). Other studies indicate that NSAIDs may have the ability to induce apoptosis and inhibit angiogenesis (22). Whether these proposed mechanisms influence breast carcinogenesis is still not clear; however, cyclooxygenase-2 has been found to be overexpressed in breast tumors but not in normal breast tissues (23).

Several characteristics of our study are noteworthy. Controls were selected from women whose diagnoses were unrelated to NSAID use. The distributions of NSAID use among subgroups of the controls were similar, suggesting the absence of selection bias. Important breast cancer risk factors were controlled simultaneously in multivariable analysis. The data on medication use were collected in the context of questions about many drug indications, and the participants and interviewers were unaware of the hypotheses at issue; thus, recall bias is unlikely to have played a role.

Information on the dosage of the NSAID taken was not collected in our study, nor was information on the number of pills taken per day. We used frequency of use as a surrogate for intensity of exposure. Misclassification could have influenced the relation between regular NSAID use and breast cancer risk in either direction. Nevertheless, there is evidence of the validity of the data: With the same database, we found a significant reduction in the risk of colon cancer for regular NSAID use—a finding confirmed in many other studies (24–26).

Data on the hormone receptor status of the tumor were obtained from only 26 percent of the cases in our study. However, the distributions of data on important risk factors for breast cancer, such as duration of female hormone supplement use, duration of oral contraceptive use, age at menarche, age at menopause, age at first birth, parity, alcohol consumption, and body mass index, were similar among cases with data on hormone receptor status and cases without such data, suggesting the absence of selection bias.

In summary, the present study found that regular use of NSAIDs is associated with a decreased risk of breast cancer, but the effect is relatively weak. Given the conflicting results from large cohort studies and the varying definitions of NSAID use, it is still too early to suggest that regular use of NSAIDs could help prevent breast cancer.

This work was supported by grant CA45762 from the National Cancer Institute. Additional support was provided by grant FD-U-00082 from the Food and Drug Administration.

Dr. Brian Strom has been a consultant for most of the manufacturers of nonsteroidal antiinflammatory drugs and has received funding from many for research, though not on this topic.

References