Modern comprehensive management of breast cancer, including surgery, chemotherapy, and radiotherapy, has resulted in prolonged cancer-related survival (1), but treatment-related cardiotoxicity remains a major concern in this population (2). Anthracyclines are widely used for the treatment of patients with breast cancer (1). The incidence and severity of anthracycline-induced cardiomyopathy is associated with the cumulative dose of the chemotherapeutic agent. Once clinically apparent, anthracycline-induced cardiomyopathy is often irreversible and potentially lethal ((3),(4),(5),6). Prior studies of beta-blockers and/or angiotensin-converting enzyme (ACE) inhibitors have shown potential protection against cardiomyopathy ((7),(8),9). However, chemotherapy patients are susceptible to abrupt fluctuations in volume status due to side effects of chemotherapy (nausea, vomiting, diarrhea), and the effects of these medications on systemic vascular resistance increase the risk for hypotension.

The mechanism of anthracycline cardiotoxicity is thought to be related to oxygen free radicals ((10),(11),12). The pleiotropic effects of statins, including decreased vascular inflammation (13) and oxidative stress (14), may mitigate cardiotoxicity. Animal experiments (15) have suggested that statins can attenuate cardiotoxicity without evidence of compromise in treatment efficacy (16). The clinical role of statins in the prophylaxis of cardiotoxicity remains unknown, although they are known to have protective effects in patients being treated for hypercholesterolemia (17), coronary artery disease (CAD) (18), peripheral artery disease, and cerebrovascular disease ((19),20). We hypothesized that the coincidental use of statins in patients undergoing chemotherapy was protective against cardiotoxicity. Our study aim was to investigate the effect of uninterrupted statin use during anthracycline chemotherapy on risk for incident heart failure (HF) in a cohort of relatively young women who were newly diagnosed with breast cancer.

We analyzed 628 consecutive women (≥18 years of age) who were newly diagnosed with biopsy-confirmed breast cancer (International Classification of Diseases-Ninth Revision diagnosis codes 174.0, 174.1 to 174.9, 233.0, and V10.3) during face-to-face outpatient encounters with Cleveland Clinic Health System (CCHS) physicians and who underwent anthracycline-based chemotherapy at CCHS between January 2005 and December 2010. The exclusion criteria were other neoplasms, kidney or heart transplantation, dialysis, HF, chronic pulmonary disease, hypertrophic cardiomyopathy, valvular heart disease, and/or aortic aneurysm before breast cancer diagnosis. Study subjects were identified from the clinical data repository available in the electronic medical record (EMR). The study was approved by the CCHS institutional review board.

The date of the cancer confirmatory diagnosis visit was considered the study entry date. Baseline characteristics were assessed at this time and during echocardiographic data collection precluding cancer treatment. These included age, socioeconomic characteristics, comorbidities, medications, family history of cardiovascular disease, smoking status, and body measurements. Socioeconomic characteristics of interest were white race, marital status, ZIP code, and county of residence. The latter was used through aggregated census data per “neighborhood” to estimate patient income and education level (21). Comorbidities of interest were hypertension, diabetes, cardiovascular disease, CAD, chronic obstructive pulmonary disease, cardiac dysrhythmias, depression, overweight or obesity, and Charlson score. We identified the use of statins, ACE inhibitors, beta-blockers, and insulin at or before the study entry date. Body measures consisted of body mass index, systolic and diastolic blood pressure, baseline heart rate, ejection fraction, left ventricular systolic and diastolic dysfunction, low-density lipoprotein, and cholesterol. A composite variable of cardiac risk factors for cardiotoxicity was derived from baseline characteristics and considered positive for subjects age <55 years who had ≥1 of the following: hypertension, history of smoking, family history of cardiovascular conditions, CAD, and/or low-density lipoprotein >100 mg/dl.

EMR review was used to obtain information on the estrogen receptor status of the tumor, cumulative anthracycline and radiotherapy doses and administration routes, site or sites of radiation, and additional chemotherapeutic agents used concomitantly, including trastuzumab. Patients with statin use at baseline were carefully monitored for continuous use during anthracycline treatment through the EMR and medication lists provided at each chemotherapy visit. The type and frequency of healthcare obtained by patients were monitored using the total number of oncology, cardiology, and primary care physician office visits, as well the total number of physician encounters.

The primary outcome of interest was new-onset HF requiring hospitalization, after the initiation of anthracycline treatment. It was obtained through EMR review of all patients, using International Classification of Diseases-Ninth Revision diagnosis codes 428.0, 428.1, 428.20 to 428.23, 428.30 to 428.33, 428.40 to 428.43, and 428.9. Loss of follow-up because of non-cardiac-related mortality was verified against the Social Security Death Index. All 628 patients were followed using EMRs through November 2011.

Descriptive statistics including mean ± SD, median, interquartile range, and frequencies for continuous and categorical data are presented for all participants and by statin treatment categories. The Wilcoxon rank sum and Pearson's chi-square test were used to compare the characteristics of patients treated and not treated with statins at any time prior and during anthracycline therapy. Using data from all 628 patients, a nonparsimonious logistic regression model was built to calculate a propensity score reflecting the probability of receiving statin treatment, conditional to patients' covariate scores. Linearity assumptions were relaxed by fitting the models using restricted cubic splines. The development of the model started with a “full” model, which included all potential risk factors for receiving statins on the basis of clinical plausibility and on the differences found between the general characteristics of the 2 groups taking continuous statins and those that were not. To ensure a highly inclusive model while maintaining statistical validity and providing a good fit, variables were dropped only when there was collinearity between the risk factors. The treatment group comprised patients on continuous use of statin therapy, coincidental to anthracycline chemotherapy. Patients not conforming to this criterion were candidates for selection into the control group. This was accomplished using a propensity score for statin therapy, derived by the model, assigned to each patient and used in a case-control matching statistical technique to reduce treatment selection bias. Patients with uninterrupted statin treatment during anthracycline therapy were thereby matched (2:1) with 134 controls by propensity score matching, using a greedy 5-to-1 digit-matching algorithm (22).

Cox proportional hazards regression analyses were used to assess the association between treatment with statins and incident HF. Patients were censored at the time of death. A series of multivariate survival analyses included social determinants of health vulnerability, cancer treatment characteristics, cardiac-related risk factors, coincidental use of other cardioprotective medication (ACE inhibitors, beta-blockers), and frequency of healthcare visits during follow-up time in the study. Those models were further validated using a measure of discrimination proposed by Pencina and D'Agostino (23). Statistical analyses were done using SAS version 9.2 (SAS Institute Inc, Cary, North Carolina). Two-tailed p values <0.05 were used to report statistical significance.

Of the total cohort of 628 patients, 67 (10.7%) received uninterrupted statin therapy through the follow-up period. Analyses of the baseline variables showed that statin users were more likely to be older and to have cardiac risk factors (Table 1). Consistent with these comorbidities, 39% of the statin users were also receiving ACE inhibitors, and 45% were on beta-blockers.

The propensity score reflecting the probability of receiving statins on the basis of comorbidities was calculated using the logistic regression model shown in (Table 2). After matching the 67 patients receiving uninterrupted statin therapy throughout the follow-up period with 134 patients not receiving statins, the general characteristics of the case cohort (Table 3) showed homogeneity between groups, with similar age, race, social determinants of health vulnerability, smoking habits, baseline body measurements, Charlson score, cancer-related treatment, and follow-up time in the study. However, some differences between the 2 groups persisted. Patients taking statins were significantly more likely to be single, to have hypertension, and to use other potentially cardioprotective medications (ACE inhibitors and beta-blockers) but were significantly less likely to have family histories of cardiovascular disease, had fewer follow-up visits with oncologists, and had lower low-density lipoprotein and baseline heart rate values.

Incident HF and cancer-related mortality were significantly lower in the statin group (Figure 67_gr1.), with only 4 cases of HF in patients treated with statins compared with 23 cases in the control group. Additionally, there were 15 cancer-related deaths in the group not receiving statins, compared with no deaths in the statin group (Table 3).

Incident HF was significantly correlated with concordant trastuzumab chemotherapy, number of cardiology visits, total healthcare visits, and presence of “cardiac risk factors” in univariate analyses (Table 4).(Table 5) shows a series of nested Cox proportional hazards models that identify the hazard attributable to statin treatment after adjusting for combinations of: race, baseline cardiac risk factors and ejection fraction <55%, treatment with other cardioprotective medications, concordant trastuzumab chemotherapy, and healthcare utilization. The protective effect of uninterrupted statin treatment on incident HF (hazard ratio [HR]: 0.3; 95% confidence interval [CI]: 0.1 to 0.9; p = 0.03) remained significant in nested models, and there was no interaction between statin treatment and any of the other covariates. The presence of cardiotoxicity risk factors at the time of cancer diagnosis (HR: 5.0; 95% CI: 2.2 to 11.1; p < 0.001), baseline ejection fraction <55% (HR: 2.7; 95% CI: 1.2 to 6.3; p = 0.02), and trastuzumab use (HR: 3.0; 95% CI: 1.3 to 7.2; p = 0.01) were significant predictors of incident HF events.

In this study of relatively young women with breast cancer treated with anthracyclines, we found that uninterrupted statin therapy initiated before and concurrent with chemotherapy reduced the risk for incident HF. Lower risk persisted after adjustment for many of the known risk factors. To the best of our knowledge, this is the first clinical observational study to assess anthracycline-induced cardiomyopathy with concurrent use of statins. It confirms the findings of a small randomized study reporting that the use of atorvastatin before anthracycline chemotherapy prevented reduction of left ventricular ejection fraction (24).

Concurrent treatment with trastuzumab substantially increased the risk for incident HF in our cohort and remained significant in adjusted models. These results are consistent with previous clinical studies (25), suggesting the need for additional cardioprotection research targeting these patients.

The pathophysiological mechanism of this condition has not been fully elucidated. The iron and reactive oxygen species theory proposes that anthracyclines trigger cardiomyopathy via the formation of free radicals that damage the cardiomyocytes (12), a process that is heightened in the presence of higher levels of cellular iron (10). Another possible mechanism is apoptosis or necrosis of cardiomyocytes and disruption of the structure and function of sarcomeres through degeneration of titin, an integral regulator of sarcomere function and turnover (1). Anthracyclines may also target the sarcomere protein dystrophin, leading to dysfunction (26). In addition, anthracyclines may induce a loss of cardiac progenitor cells, leading to permanently altered vascular architecture, including reductions of capillary density and arteriolar branching (27).

Our results are consistent with pre-clinical experiments using cell cultures and animal models to assess the effect of statin use before the administration of anthracycline on resulting cardiotoxicity ((28),(29),30). In cultured cardiac myocytes treated with anthracyclines (28), statin pre-treatment attenuated oxidative stress by inhibiting the production of isoprenoid intermediates synthesis (RAC1 and other small G proteins). This finding is consistent with theories that anthracyclines induce cardiomyopathy through oxidative stress, deoxyribonucleic acid damage, and ataxia telangiectasia mutated activation, with accumulation of tumor suppressor protein p53 ((28),(29),30). Statin pre-treatment in a doxorubicin-induced cardiotoxic mice model showed improved left ventricular function, with reduced oxidative stress, inflammatory response, cytokine release, and myocyte apoptosis (15).

Other direct and adjuvant effects of statins in cancer therapy are under investigation. Statins may increase doxorubicin accumulation in multidrug resistance tumor cells, thereby potentiating deoxyribonucleic acid damage, growth arrest, and apoptosis, while preserving the viability of normal cells ((16),31). Consistent with our findings, significant reduction of cancer-related short-term mortality has been reported in patients using statins, suggesting that further research is warranted to determine whether pre-treatment with statins might constitute a promising prophylactic option in patients with breast cancer. This benefit may extend to other patient populations treated with anthracyclines.

First, although the focus of this study was on the continuous use of statins during all cycles of anthracycline, to define proof of concept of the cardioprotective role of this therapy, we acknowledge that medication status during follow-up is not known by the clinician at baseline. A study to examine the role of statins in “real life” would have to account for cessation of the medication for a variety of reasons, at a variety of times, and a time-varying covariate design would be needed.

Second, although the data were entered prospectively, as with all analyses using EMRs, there was potential for unidentifiable sources of bias. The use of statins may have been affected by patient comorbidities, access to medical care, and individual ability to tolerate treatment. We tried to address this problem by using propensity matching and multivariate risk adjustment, but these methods cannot account for unmeasured variables. Some comorbidities may have been underestimated because of variability in physician documentation and hospital coding practices; however, the use of templates for EMR data entry should have minimized this factor.

Third, because we did not have patient-level data on income and education, we used an established method consisting of neighborhood socioeconomic-based aggregated census data to capture these factors (21), and all patients were screened at baseline with echocardiography.

It is unclear whether we were able to fully adjust for the cardioprotective effect of concomitant treatment with beta-blockers and ACE inhibitors, although we included the use of these medications in our models (32). Additionally, even though the prescription of statins by physicians at each encounter in the EMR was matched with pharmacy records, acquiring the prescription does not necessarily guarantee that all of the patients took their medications. However, failure to adhere to statin therapy would weaken, rather than overestimate, its protective effects against incident HF.

Last, the study population was composed of patients from 1 healthcare system, so it is unclear to what extent our findings are generalizable to other patient populations. Patients were identified on the basis of International Classification of Diseases, Ninth Revision, diagnosis codes, and those without documented biopsy-confirmed breast cancer were excluded. Although it is possible that patients may have sought care outside of the CCHS, it seems unlikely that this would be selectively related to statin use.

In this observational study, statins appear to be associated with markedly reduced risk for HF and cardiac-related mortality over 2.2 ± 1.7 years of follow-up. Our results are consistent with pre-clinical experiments using cell cultures and animal models ((28),(29),30) and with those of a small randomized clinical study (24). Prospective clinical trials are needed to confirm the relationship between statin use, before and during chemotherapy with anthracyclines, and incident HF. Patients at higher risk for anthracycline-induced cardiotoxicity because of hypertension, CAD, and concurrent use of trastuzumab and mediastinal radiotherapy should especially be considered ((4),(33),34). In addition, it may be beneficial to expand our understanding of the effects of statins on cardiac tissue and on cancer cells.