Prospective Analysis of Association between Statin Use and Breast Cancer Risk in the Women's Health Initiative

Preclinical evidence suggests that statins, the most commonly used cholesterol-lowering medications, may influence mammary cancer growth (1, 2), but the clinical evidence is inconsistent (3). Observational studies of statins and breast cancer risk have shown mixed results, with some studies reporting an increase in risk, (4–9), others showing a protective effect (10–14), and others showing no association (15, 16). In a previous analysis of the Women's Health Initiative (WHI) cohort, Cauley and colleagues reported that, after a mean of 6.7 years follow-up, there was an 18% lower risk of invasive breast cancer seen among women who had reported use of lipophilic statins (11). However, in further analyses limited by the small numbers of cases in medication subgroups, no specific statin was associated with a lower risk of breast cancer. We now reexamine the relationship between statins and breast cancer risk in the WHI with approximately four additional years of follow-up and 3,047 additional breast cancer cases.

The population included 154,587 postmenopausal women enrolled in the WHI clinical trial (67,327) and observational study (87,266). Study implementation details have been published previously (17–19). In brief, women aged 50 to 79 were enrolled in one or more of four clinical trials [Hormone therapy which included estrogen alone and estrogen plus progesterone trials; dietary modification; and calcium and vitamin D supplementation] or an observational study cohort in 40 U.S. clinical centers from October 1, 1993 through December 31, 1998. Follow-up continued from study initiation until planned termination on March, 2005 and thereafter for participants providing re-consent, with data collection updated through September, 2010 for an average of 10.8 (SD, 3.3) years of follow-up. For the purposes of this analysis, we excluded 7,217 women who had a self-reported prior history of breast cancer at baseline entry into the WHI, and two women for whom information on statin intake was not available.

Participants in the WHI were asked to bring all current prescription medications to their first screening interview. Reported medications were then matched to the Master Drug Database (First DataBank, Inc.) and duration of use for each medication was recorded. The procedures for collection of data on medication use were repeated in years one, three, six, and nine of follow-up in the clinical trial, and in year three in the observational study. Memory aides were not used for collection of medication use information.

Statin use was defined as any use of 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase inhibitor at baseline entry into the WHI. Updated information collected at subsequent participant visits were used to measure statin use as a time-dependent exposure for a secondary analysis. Statins were classified as lipophilic (lovastatin, simvastatin, fluvastatin) or hydrophilic (pravastatin and atorvastatin; ref. 20) and, by potency, as low potency (fluvastatin and lovastatin), medium potency (pravastatin), and high potency (simvastatin and atorvastatin; refs. 21–23).

Reported invasive breast cancer cases were initially confirmed by medical record review by trained physician adjudicators at the clinical centers. Final adjudication and coding for stage using the Surveillance Epidemiology and End Results (SEER) system was done at the Clinical Coordinating Center. Hormone receptor status and HER2 status was based on local laboratory criteria (24).

Information on potential confounding and modifying variables were collected by baseline questionnaire including baseline characteristics and known risk factors for invasive breast cancer, as well as factors associated with health-care use and cancer screening, which might impact both statin use and breast cancer detection. Information on baseline food habits was determined by the WHI food frequency questionnaire (25), and interviewer-administered questionnaires were used to determine hormone therapy use, which was defined as any use of hormones for three or more months after menopause. Mammograms and breast exams were mandated by study protocol annually in the two hormone therapy trials and biennially in the dietary modification trial. Mammograms and breast exams were not mandated in the observational study and were left to the discretion of the participant's physician. Information on mammography and breast exams was collected annually in the observational study. The covariates used in the analysis are listed in Table 1.

The characteristics of statin users at baseline were compared with those of nonusers by χ² tests. Annualized rates of breast cancer were calculated for statin users and nonusers at baseline. Planned selected subgroup analyses were conducted by statin-use duration as determined at baseline (<1 year, 1–<3 years, and ≥3 years), type, potency, and lipophilic status. Women who reported using two or more statins were included in analyses that compared statin use to none, but were excluded from analyses that examined details of statin use by type, potency or lipophilic status. Separate analyses were conducted for women with ER/PR-positive and ER/PR-negative breast cancer. HRs for breast cancer among statin users versus nonusers, and 95% confidence intervals (CI) were computed from Cox proportional hazards analyses. Tests for the proportional hazards assumptions were conducted by a Cox model that included statin use and the interaction of statin use with follow-up time, and testing for a zero coefficient on the interaction term.

Cox proportional hazards models were used to assess associations between statin use at baseline and risk of invasive breast cancer. An a priori selection of covariates with a 10% backward selection in Cox modeling was used to create a final multivariable-adjusted model. The base model was adjusted for age and baseline hormone therapy use and stratified by trial participation (hormone therapy, dietary modification, or observational study), WHI extension study participation, and age group. The final multivariable model was adjusted for age, body mass index (BMI), ethnicity, smoking status, baseline hormone therapy use and duration of use, family history of breast cancer, education, hysterectomy, bilateral oophorectomy, mammogram within the last two years, age at first birth, parity, age at menarche, alcohol use, percentage energy from fat, nonsteroidal anti-inflammatory drug (NSAID) use, and physical activity. Comparisons of breast cancer tumor characteristics between statin users and nonusers were based on χ² and Fisher exact tests.

To evaluate the effect of change in statin use over time, final models were rerun by entering statin use as a time-dependent exposure and using updated information on statin use gathered at year three in the observational study and years one, three, six, and nine in the clinical trial. We censored breast cancer outcomes three years after the last medication update in the observational study to more closely parallel statin exposure in the clinical trial. In addition, we conducted the analysis including censored outcomes. Moreover, we conducted a sensitivity analysis by excluding breast cancer cases diagnosed within two years from baseline in order to allow for sufficient statin exposure to have an effect on risk of breast cancer.

All statistical tests were two-sided with a significance level of 0.05. Analyses were conducted using the Statistical Analysis Software (SAS) version 9.2.

The baseline characteristics of statin users and nonusers are outlined in Table 1. Most of the variables listed were statistically significant due to large numbers. Statin use was reported at baseline by 11,584 women (7.5%) in the WHI cohort. Statin users were more likely to be older, have higher past use of tobacco, and to have larger BMI and waist circumference. Other variables associated with statin use included: bilateral oophorectomy, hysterectomy, history of benign breast disease, mammography within the past two years, use of aspirin and NSAIDS, as well as history of diabetes, angina, and myocardial infarction. Nonuse of statins was associated with current hormone therapy use, less physical activity, and a diet with more than 30% of body energy from fat. There was no association between statin use and age at menarche, prior pregnancies, age at first full-term pregnancy, and family history of breast cancer.

Table 2 shows the distribution of statin use at baseline by type, duration, potency, and lipophilicity. Simvastatin was the most common statin used, with 29.3% reporting it's use at baseline. The majority of statin users took lipophilic statins (69.5%) and 39.4% of users were on a statin classified as low potency, 38% high potency, and 22.6% medium potency. Among users at baseline, the percentage of participants using statins for <1 year, 1–<3 years, and ≥3 years was 33.25%, 33.94%, and 32.8%, respectively.

Table 3 shows the risk of invasive breast cancer by statin use at baseline. The annualized rate of breast cancer was 0.42% among statin users and 0.42% among nonusers. In the multivariable-adjusted model, there was no significant relationship between statin use and breast cancer risk (HR, 0.94; 95% CI, 0.83–1.07). There was no significant reduction in breast cancer risk for prior use of simvastatin (HR, 0.87; 95% CI, 0.70–1.09) or other types of statins, use of two or more statins, lipophilicity, or statin potency. Statin use for less than one year duration was associated with a trend toward an inverse association (HR, 0.78; 95% CI, 0.63–0.98); however, there was no significant trend by overall duration of use (HR, 1.19; 95% CI, 0.91–1.33 for 1–<3 years and HR, 0.97; 95% CI, 0.79–1.19 for ≥3 years; P-trend, 0.68). There were no significant differences in the results observed with or without censoring breast cancer outcomes three years after last medication update in the observational study or after exclusion of breast cancer cases diagnosed within two years from baseline (data not shown). Table 4 shows the risk of invasive breast cancer by statin use in a multivariable-adjusted time-dependent model. In this model, there was no significant reduction in risk of breast cancer for users of simvastatin (HR, 0.87; 95% CI, 0.71–1.07). Similarly, no significant associations were seen for other statin types, lipophilicity, or statin potency.

Table 5 shows the distribution of breast cancer tumor characteristics by statin use at baseline. There were no significant differences in the distribution of tumor characteristics including stage, grade, ER/PR receptor, and HER2neu status by use of statins. In addition, we looked at the relationship between statin use at baseline and breast cancer risk stratified by other important breast cancer risk factors including hormone therapy, family history of breast cancer, waist circumference, and BMI. There were no significant interactions seen (Table 6).

Our updated results from the WHI cohort showed no significant relationship between statin use and breast cancer risk; however, there was a nonsignificant reduction in risk for simvastatin in the multivariable analyses and no significant relationship was seen for lipophilic statins as a group. These results are contrary to the previous analysis by Cauley and colleagues (11) that showed a statistically significant reduction in risk of breast cancer associated with lipophilic statins. However, our results did not show this same reduction in risk. In addition, we showed a modestly protective effect associated with statin use of less than 1-year duration compared with multiple years, which argues against biological plausibility. In our analysis, we censored breast cancer outcomes after year six in the observational study participants in order to closely parallel exposure information available for clinical trial participants, which was different than the methodology used in the earlier WHI analysis by Cauley and colleagues. Moreover, we analyzed the data without censoring, and no significant differences in the results were observed (data not shown).

A number of preclinical findings support the biologic plausibility of a protective effect for statins in relationship to the development of breast cancer. Statins act by inhibition of HMG-CoA reductase and have pleomorphic properties in the cell. The multiple downstream effects of statins may have antiproliferative, anti-invasive, and apoptotic activities (26–29). Farnesyl diphosphate (FPP) and geranylgeranyl diphosphate (GGPP), downstream products of the mevalonate pathway, are both involved in posttranslational modification of many proteins (30), such as the Ras molecule, which in turn helps transmit downstream signaling from surface receptors (31). Ras is involved in many intracellular pathways and increases gene transcription and proliferation by acting through the MEK and phosphoinositide 3-kinase (PI3K)/Akt pathways (32) and inhibition of this pathway can have antiproliferative effects on cancer cells (33). In addition, statins have been implicated in reducing cell migration, proliferation, and invasion by inhibiting production of GGPP, which is involved in geranylgeranylation of Rho proteins including Rho GTPases (27) that maintain function of Rho kinases involved in various cellular functions including gene expression, actin cytoskeleton migration, adhesion, and contractility of cells (34).

Earlier observational studies of statins and breast cancer risk have shown mixed results. Individual studies have reported a protective effect (10–14); however, two meta-analyses published in 2005 reported no significant association (15, 16). The most recent meta-analyses, which included 13 cohort and 11 case-control studies with 2.4 million participants and 76,759 breast cancer cases identified through January 2012, also showed no significant relationship between statins and breast cancer risk (RR, 0.99; 95% CI, 0.94–1.04; ref. 3).

Class differences in statins and anticancer efficacy have been explored in previous studies. Lipophilic statins (lovastatin, simvastatin, fluvastatin) penetrate the plasma membrane whereas hydrophilic statins (pravastatin and atorvastatin) do not (22, 35) Cellular uptake of lipophilic statins may be related to their inhibition of cell growth, and this has been supported by a cell culture study in which only lipophilic statins were shown to have anticancer activity (36).

The results from epidemiologic studies which have analyzed specific statin preparation or class have been mixed showing either no relationship (37, 38), an increased risk of breast cancer (6, 7, 39), or a reduced breast cancer risk (10–14). In one case–control study, only fluvastatin was associated with a decreased risk of breast cancer (OR, 0.50; 95% CI, 0.30–0.80) with no association seen overall or with lipophilic statins as a group (12) and, in another case–control study, the specific type of statin was not associated with breast cancer risk, although use of statins for more than five years was related to a trend toward decreased risk (OR, 0.70; 95% CI, 0.40–1.0; ref. 13). In a record-linkage cohort study from Finland, a marginal reduction in breast cancer risk was noted for users of simvastatin (HR, 0.97; 95% CI, 0.95–0.99; ref. 14). In the meta-analysis by Bonovas and colleagues, seven randomized clinical trials were included together with nine cohort studies. Although the overall result of the meta-analysis showed no significant relationship (HR using fixed effects model of 1.04; 95% CI, 0.81–1.33), only two randomized trials were undertaken using simvastatin. Of these, the Heart Protection Study (39), which had 38 incident cases of invasive breast cancer in the simvastatin group and 51 breast cancer cases in the nonstatin group, did not show a protective effect (HR, 0.75; 95% CI, 0.49–1.13) and, in the Scandinavian Simvastatin Survival Study (40), there were only seven incident breast cancer cases in the simvastatin arm and five in the nonstatin arm (HR, 1.44; 95% CI, 0.46–4.52).

The strengths of our study include the prospective cohort design, the large diverse population which was well characterized for breast cancer and cardiovascular risk, breast cancer verification by central review, serial update of statin use, and adjustment for mammography frequency as well as the long follow-up period. In addition, the comprehensive data collection in the WHI allows for a detailed adjustment for confounding variables. Limitations include the low prevalence of statin use at baseline and lack of information on medication compliance, interval data collection rather than continuous, recall bias, and bias due to loss of follow-up. Moreover, the study lacked updated information on use of rosuvastatin and the more recently available pitavastatin as they were not widely in use during the time of the study. Our results may be biased toward the null as participants on statins could potentially not be compliant with taking the medication whereas nonusers at baseline did not have this bias. In addition, due to the observational nature of our study, the results may be biased because of unknown variables that may be associated with both breast cancer and statin use. It is unclear why a protective effect was seen associated with short duration of statin use.

In conclusion, we found no overall association between statin use and breast cancer risk; however, our findings raise the possibility of a marginal reduction in risk associated with simvastatin use. Genetic variation in statin metabolism has been linked to the efficacy of statins in the treatment of coronary artery disease (41) and may possibly have an impact on the effect of statins on colorectal cancer risk (42). Future studies should assess the effect of statins on cancer risk in selected populations based on genotype.

L.W. Martin has commercial research grants from Amarin, Amgen, Sanofi, and Novartis, and is a consultant/advisory board member of WHI Publication committee NIH. No potential conflicts of interest were disclosed by the other authors.

Conception and design: P. Desai, J.A. Cauley, C.A. Rosenberg, J. Wactawski-Wende, M.S. Simon

Development of methodology: P. Desai, R.T. Chlebowski, M.S. Simon

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R.T. Chlebowski, J.A. Cauley, J.E. Manson, D.S. Lane, J. Wactawski-Wende, M.S. Simon

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P. Desai, R.T. Chlebowski, C. Wu, A. Jay, U. Peters, J. Luo, H.L. Park, M.S. Simon

Writing, review, and/or revision of the manuscript: P. Desai, R.T. Chlebowski, J.A. Cauley, J.E. Manson, L.W. Martin, A. Jay, C.H. Bock, M.L. Cote, N. Petrucelli, C.A. Rosenberg, U. Peters, I. Agalliu, N. Budrys, D.S. Lane, J. Luo, H.L. Park, F. Thomas, J. Wactawski-Wende, M.S. Simon, M. Abdul-Hussein

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.A. Cauley, J.E. Manson

Study supervision: J. Wactawski-Wende

The authors thank the investigators and staff at the WHI clinical centers, the WHI Clinical Coordinating Center, and the National Heart, Lung, and Blood program office (listing available at http://www.whi.org) for their dedicated efforts. The authors also recognize the WHI participants for their extraordinary commitment to the WHI program. For a list of all the investigators who have contributed to Women's Health Initiative science, please visit: https://cleo.whi.org/researchers/Documents%20%20Write%20a%20Paper/WHI%20Investigator%20Long%20List.pdf.

The WHI program is funded by the National Heart, Lung, and Blood Institute, NIH, U.S. Department of Health and Human Services through contracts N01WH22110, 24152, 32100-2, 32105-6, 32108-9, 32111-13, 32115, 32118-32119, 32122, 42107-26, 42129-32, and 44221, and the Cancer Center Support Grant NIH:NCI P30CA022453.

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.