Long-Term Statin Use and Risk of Ductal and Lobular Breast Cancer among Women 55 to 74 Years of Age

3-Hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) inhibitors, or statins, are a highly effective therapeutic class of drugs used to lower lipids for both primary and secondary prevention of coronary heart disease. The use of statins has increased considerably since their introduction in the United States in 1987. Approximately, a quarter of all United States women over the age of 45 report current use of one of the 7 different statin drugs on the market today (1).

Evidence from previous epidemiologic studies investigating the use of statins and risk of developing breast cancer is inconsistent, with some studies reporting an increased risk (2–11), some reporting a decreased risk (12–16), and still others reporting no association (17–23). Early studies in rodents showed a carcinogenic potential of statins (24). However, in in vitro breast cancer models, statins have been shown to inhibit cell proliferation and induce apoptosis (25–27). Meta-analyses have evaluated the conflicting evidence and provided summary estimates of the association between statin use and breast cancer. The meta-analysis by Bonovas and colleagues included 7 randomized trials and 9 observational studies published before 2005, and it reported a relative risk (RR) of 1.02 (95% CI: 0.89–1.18) for the association between statin use and breast cancer (28). Published in 2012, Undela and colleagues included 24 observational studies and similarly found no evidence of an association between statin use and breast cancer risk (RR: 0.99, 95% CI: 0.94–1.04; ref. 29).

Despite extensive prior research on statin use and breast cancer risk, some important questions remain unanswered. Given the increase in statin use over the past few decades, and that these medications are commonly prescribed for chronic, essentially lifetime use, the studies published to date have had limited ability to evaluate the impact of long durations of current statin use. Of the studies published since 2005, in the NHS cohort only 1.0% of women had currently used statins for 4 years or longer (15), in the WHI cohort only 2.5% had used statins for 3 years or longer (14), and in a large case–control study only 2.7% of controls were current statin users for 5 years or longer (20). Rates of use were highest in the most recently published study based on the Cancer Prevention Study II Nutrition Cohort, where 6.8% of women were current statin users for 5 years or longer (16). Given the frequency with which statins are used and the extended duration women are prescribed them, characterizing the relationship between long-term statin use and breast cancer risk is of clinical and public health importance. In addition, further investigation is needed regarding the impact of statin use on different types of breast cancer. We evaluated these questions using data from a recently completed large-scale population-based case–control study of postmenopausal breast cancer where use of statins was more frequent compared with populations included in prior published studies.

We conducted a large population-based case–control study of the 2 most common histologic subtypes of breast cancer, invasive ductal carcinoma (IDC) and invasive lobular carcinoma (ILC), among women 55 to 74 years of age living in the three-county Seattle-Puget Sound metropolitan area (King, Pierce, and Snohomish counties). In the United States, an estimated 70% of all invasive breast cancers diagnosed among postmenopausal women are ductal and approximately 20% are lobular (30). This study was funded in 2 continuous phases, and data from the first phase based on cases enrolled from January 2000 to March 2004 were published previously (31).

Breast cancer cases with no prior history of in situ or invasive breast cancer, diagnosed between January 2000 and December 2008 while residing in King, Pierce, or Snohomish county were identified through the Cancer Surveillance System (CSS), the population-based tumor registry that serves western Washington state and participates in the Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute. The 3 counties included in this analysis are among the 13 counties that comprise the Seattle-Puget Sound SEER site. All women diagnosed with an invasive breast cancer with a lobular component based on ICD-O codes 8520, 8522, and 8524 assigned by CSS were potentially eligible as lobular cases. The pathology reports of all of these cases were then centrally reviewed to confirm eligibility and re-categorize histology groupings as necessary. Given the greater frequency of IDC, an age-matched sample of approximately 25% IDC cases was selected for recruitment. IDC cases were frequency matched to the ILC case group by 5-year age group. As controls were ascertained via random digit dialing of landline home telephone numbers, to be eligible all cases were also required to have a landline home telephone. Of the 2,495 eligible cases identified, 1,984 (80%) were interviewed, including 1,068 ILC and 916 IDC cases, 424 (17%) refused and 75 (3%) were deceased.

We used the Mitofsky–Waksberg (32) method of random digit dialing to identify potential controls from the general population of female residents of King, Pierce, and Snohomish counties. Controls were frequency matched within 5-year age groups and county of residence to the cases using one-step recruitment. Of the 1,313 eligible controls identified, 902 (69%) were interviewed and 411(31%) refused.

The reference date used for each woman with breast cancer was her diagnosis date. Control reference dates were assigned to reflect the expected distribution of reference dates among the cases. Data collection was limited to exposures that occurred before each participant's reference date.

The study protocol was approved by the Fred Hutchinson Cancer Research Center Institutional Review Board, and written informed consent was obtained from all study subjects. In addition to basic information on breast cancer diagnosis, we obtained information on tumor characteristics from CSS and from a centralized review of pathology reports. This includes data on estrogen receptor (ER), progesterone receptor (PR), and HER2-neu (HER2) status, and tumor stage, size, and nodal status.

Cases and controls were interviewed in-person. Through a series of structured questions, detailed histories of hypercholesterolemia and all episodes of lipid lowering medication use, including beginning and ending dates, drug name, dose, route of administration, and pattern of use (number of days per month) were obtained. Study participants were also asked to show bottles of all prescription medications they were currently taking, including lipid lowering drugs, to their study interviewer who then directly transcribed data from these bottles onto our data collection form. To enhance recall of past use, a photo book containing pictures of pills and packages of commonly used lipid lowering medications was used along with a show card listing the brand and generic names of each of these medications. In addition, all participants were queried about various known or suspected breast cancer risk factors including pertinent aspects of their reproductive, medical, breast cancer screening, and family histories, as well as information about their body size and lifestyle habits. Our questioning was limited to exposures that occurred before each participant's reference date. Women were asked if they had ever smoked, drank alcoholic beverages, used a specific medication, been diagnosed with a specific condition, or ever had a mammogram; for cumulative exposure variables patterns of use from the time of first exposure through the reference date were ascertained in the interview.

Women who never used any type of lipid lowering medication served as the reference category. Our main analysis focused on duration and recency of statin use where current users were those who ever used these medications for 6 months or longer and were currently using them within 6 months of their reference date. Former users were ever users of these medications for 6 months or longer who last used them more than 6 months before their reference date. Short-term users were women who used these medications for less than 6 months regardless of their recency of use. Analyses also considered risk associated with each individual statin as well as when grouped as lipophilic (atorvastatin, simvastatin, lovastatin, fluvastatin, and cerivastatin) or hydrophilic (pravastatin, and rosuvastatin) statins. Finally, an analysis restricting the sample to the 273 controls, 286 IDC, and 320 ILC cases with a history of high cholesterol was conducted to investigate the existence and magnitude of potential biases due to the selection of healthy controls and to confounding by indication.

We used polytomous logistic regression to calculate odds ratios (OR) and their associated 95% confidence intervals (CI) to compare IDC and ILC cases to controls. Given the design of this study and its random sampling of IDC cases and exclusion of other histologic subtypes of breast cancer, risk estimates for breast cancer overall could not be calculated. Pearson χ² statistic was calculated modeling categories of years of statin use as a grouped linear variable to test for a trend in duration. All analyses were conducted using Stata/SE version 11.2 (StataCorp LP). All models were adjusted for age (5-year categories), reference year (continuous), and county, as controls were frequency matched to cases on these factors. Several potential confounders and effect modifiers of the relationship between statin use and breast cancer were assessed, including: education, household income, race/ethnicity, use of menopausal hormone therapy, mammography screening history, type of menopause, first-degree family history of breast cancer, body mass index 1 year before reference date, presence of diabetes, parity, alcohol consumption, and smoking history (as grouped in Table 1). Only use of menopausal hormone therapy changed our risk estimates by more than 10% when added to the model, and so, only it was added as a covariate to our final statistical models. Excluded from all analyses were the 25 controls, 25 IDC cases, and 32 ILC cases missing data on either use of lipid lowering medications and/or menopausal hormone therapy use leaving a final analytic sample size of 877 controls, 891 IDC cases, and 1,036 ILC cases. In addition, none of the covariates were found to be a statistically significant effect modifier based on likelihood ratio testing including body mass index (all P_interaction >0.05). To quantify differences in risk according to use of menopausal hormone therapy and mammography screening history, we also assessed risks stratified by these 2 factors. In addition, we assessed whether or not risk estimates differed among women with invasive lobular (ICD-O codes 8520 and 8524) versus invasive ductal-lobular (ICD-O code 8522) carcinomas. No appreciable differences in the magnitudes of risk were observed when the analysis was stratified in this way and none of the P-values comparing lobular versus ductal-lobular risk estimates were statistically significant. Thus lobular and ductal-lobular tumors were grouped together in all analyses. Finally, we conducted analyses stratified according to estrogen receptor status in 3 groupings: ER+ IDC, ER− IDC, and ER+ ILC. Given that fewer than 4% of our ILC cases were ER−, we could not evaluate risk among ER− ILC cases.

Controls, IDC cases, and ILC cases had similar distributions of age, household income, history of hypercholesterolemia, and history of heart disease (Table 1). Compared with controls and IDC cases, ILC cases were somewhat more likely to be non-Hispanic white, to have completed college, to consume greater than equal to 1 alcoholic beverage per day, to have a body mass index less than 25 kg/m², and to be current users of combined estrogen and progestin menopausal hormone therapy. IDC and ILC cases were more likely than controls to have a first-degree family history of breast cancer; however, mammography screening history was similar across all groups.

Compared with never users, women who ever used statins did not have an increased risk of IDC (OR: 1.16, 95% CI: 0.92–1.47; Table 2). There was a modest suggestion of an increased risk of ILC (OR: 1.22, 95% CI: 0.98–1.53) comparing women who ever used statins to never users. However, women using statins for greater than equal to 10 years had a 72% higher risk of IDC (95% CI: 1.10–2.71) and an 82% higher risk of ILC (95% CI: 1.17–2.82), though neither P-value for linear trend was statistically significant. When evaluated by recency, current long-term (≥10 years) statin users had an 83% higher risk of IDC (95% CI: 1.14–2.93, P for trend = 0.043) and a 97% higher risk of ILC (OR: 1.97, 95% CI: 1.25–3.12, P for trend = 0.021) than never users. When analyses were restricted to the 273 controls, 286 IDC cases, and 320 ILC cases with a history of high cholesterol, the associations observed were slightly greater in magnitude to those observed in our overall analyses. Current use of any statin for at least 6 months was associated with a 41% increased risk of IDC (95% CI: 0.95–2.08) and a 63% increased risk of ILC (95% CI: 1.11–2.41). Again, this increased risk was strongest among long-term statin users who had more than double the risk of both IDC (OR: 2.04, 95% CI: 1.17–3.57, P for trend = 0.025) and ILC (OR: 2.43, 95% CI: 1.40–4.21, P for trend = 0.006).

The vast majority (88%) of statin users in this population used lipophilic statins. Current use of lipophilic statins was not associated with either IDC (OR: 1.13, 95% CI: 0.88–1.45) or ILC (OR: 1.21, 95% CI: 0.95–1.54; Table 3). However, current users of lipophilic statins for greater than equal to 10 years had a 74% increased risk of IDC (95% CI: 1.05–2.86) and a 68% increased risk of ILC (95% CI: 1.02–2.76) compared with never users. The 2 most commonly used statins in this population were atorvastatin and simvastatin. Current use of atorvastatin for greater than equal to 10 years was associated with elevated risks of both IDC and ILC but while both of these risk estimates were within the limits of chance, the trend with respect to IDC was statistically significant (P = 0.048). There were no clear associations between simvastatin use and IDC or ILC risk.

Stratifying by estrogen receptor status, there was evidence that while current statin use for 10 or more years was associated with 2-fold increases in risk of ER+ IDC (OR: 1.97, 95% CI: 1.21–3.20, P for trend = 0.035) and ER+ ILC (OR: 2.00, 95% CI: 1.26–3.17, P for trend = 0.023), it was associated with a more modest nonstatistically significant elevated risk of ER− IDC (OR: 1.33; 95% CI: 0.53–3.31; Table 4).

In this population-based case–control study, greater than equal to 10 years of statin use was associated with an increased risk of both IDC and ILC. Our estimate of a 26% increased risk of IDC among women using statins for 5–10 years is consistent with estimates from three previous studies reporting relative effects between 10% and 30% associated with greater than equal to 5 years of statin use (7, 16, 20). However, our estimate differs from studies reporting a decreased risk of breast cancer associated with greater than equal to 4 years of statin use (3, 22). Consistent with previous studies we did not observe evidence of associations between shorter durations of statin use and breast cancer risk. Only one prior study has reported data for statin use for 10 or more years and found no association between long-term statin use and breast cancer risk (OR: 0.8, 95% CI: 0.5–1.4; 20). We observed that risks were highest among long-term current users suggesting that statins may act as promoters of breast carcinogenesis.

This study differed from previous studies in its ability to investigate the association between statin use and breast cancer separately by histologic subtype and estrogen receptor status. Among long-term statin users, we saw substantially increased risks of both ER+ IDC and ER+ ILC, but no increased risk of ER− IDC. Given the relatively small number of ER− cases in our study and the resulting uncertainty in our estimates, it is not clear whether the association between long-term statin use and breast cancer risk differs by estrogen receptor status. While studies of the WHI and NHS cohorts found null associations for use of statins and breast cancer for both ER+ and ER− tumors, neither study had data on long-term statin users (14, 15). If in fact statins are only associated with ER+ breast cancer, this would suggest that statins may be exerting a carcinogenic effect through a hormonally driven pathway. However, the biology underlying an association between statin use and breast cancer risk is uncertain. All statins inhibit HMG-CoA reductase at the rate-limiting step of the mevalonate pathway, an intricate biochemical pathway required for the production of cholesterol, isoprenoids, dolichol, ubiquinone, and isopentenyladine (33). Laboratory studies have investigated how disrupting the melavonate pathway may lead to carcinogenesis and have discovered both pro and anticancer effects of statins (34). However, our results call into question whether the anticarcinogenic properties of statins observed in animal models apply to the long-term effects of disrupting the mevalonate pathway in humans. Our finding of an increased risk only among current long-term statin users suggests that the chronic dysregulation of the mevalonate pathway and/or long-term lowering of serum cholesterol may contribute to breast carcinogenesis. This finding does not rule out the possibility that shorter-term statin use may have no effect or possibly a transitory protective effect on the development of breast cancer.

Prior studies examining individual statins have not found increased risks of breast cancer associated with use of any individual statin, with the exception of the randomized trial of pravastatin, a hydrophilic statin, that reported a 12-fold increased risk of breast cancer (12 cases in the pravastatin arm compared with 1 case in the control arm; 35). However, other trials of pravastatin found no increase in breast cancer in the treatment arm (36, 37). In the WHI cohort while use of hydrophilic statins was not related to breast cancer risk, an 18% (P = 0.02) lower breast cancer incidence was observed among users of lipophilic statins (simvastatin, lovastatin, or fluvastatin) compared with statin nonusers (14). However, it should be noted that two thirds of statin users in the WHI cohort had used statins for less than 3 years. In contrast, a recent large case–control study found no evidence of an association between lipophilic statins and breast cancer (23). Our finding of an increased risk of breast cancer among long-term, primarily lipophilic, statin users does not support the use of statins for breast cancer prevention and suggests that the risk of breast cancer may be elevated for both users of lipophilic and hydrophilic statins.

It is also important to acknowledge the potential limitations of this study. Given its design, recall bias is a potential concern. However, our study interviewers used showcards with pictures of medications to enhance recall and had all participants bring all of their current medications to the interview to minimize misclassification. Furthermore, a validation study comparing self-reported statin use to data from pharmacy records conducted among women aged 65 to 79 years from Washington State found no significant differences in recall of recent statin use between breast cancer cases (sensitivity 83%, 95% CI: 64, 93) and controls (sensitivity 93%, 95% CI: 69, 99) and suggested that any differential misclassification would likely result in a spuriously low proportion of statin users among breast cancer cases biasing the OR for the association between statin use and breast cancer toward the null, given that cases were less likely to be correctly classified as statin users (38). Selection bias is also a possible explanation for the observation that IDC and ILC cases were more likely to be long-term statin users than controls. It is possible that the controls available through RDD who agreed to participate in our study were healthier than the general population, resulting in a spuriously low proportion of statin users in our controls. Our finding that controls, IDC cases, and ILC cases all had similar proportions of women with a history of high cholesterol and heart disease as well as similar mammography screening history does not suggest that the controls were exceptionally healthy. We also did not observe differences in the magnitude of the association between statin use and breast cancer when analyses were stratified by mammography screening history, suggesting that the observed results are not explained by a higher frequency of screening in among statin users. It is possible that our finding of an increased risk of breast cancer among long-term statin users is attributable to differences in characteristics of long-term statin users rather than to statin use itself. However, our finding of a greater than 2-fold increased risk of breast cancer among current long-term statin users in analyses restricted to study participants reporting a history of high cholesterol is evidence that confounding by indication does not account for the excess of breast cancer cases among statin users.

The finding that current, long-term use of statins may double the risk of both ductal and lobular breast cancer warrants further study. This is the first study with the power to examine long-term effects of statins on breast cancer risk. In contrast to the NHS and WHI studies where the prevalence of statin use was 2.6% and 7.5%, respectively, 25% of controls in our study population were current statin users and 18% of those had used statins for 10 or more years. As more women are taking statins and for longer durations than were previously available for study, it is possible that we will observe effects of long-term statin use that prior studies could not detect. Confirmation of these results in other studies is necessary before any changes in clinical practice would be warranted.

No potential conflicts of interest were disclosed.

This article and its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NCI, NIH.

Conception and design: J.A. McDougall, K.E. Malone, J.R. Daling, P.L. Porter, C.I. Li

Development of methodology: J.R. Daling, P.L. Porter, C.I. Li

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.R. Daling, P.L. Porter, C.I. Li

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.A. McDougall, K.L. Cushing-Haugen, P.L. Porter, C.I. Li

Writing, review, and/or revision of the manuscript: J.A. McDougall, K.E. Malone, J.R. Daling, K.L. Cushing-Haugen, P.L. Porter

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K.L. Cushing-Haugen

Study supervision: C.I. Li

This research was funded by a grant from the National Cancer Institute R01 CA 85913. This publication was supported by grant number T32 CA09168 from the NIH.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.