Introduction

Cardiac disease and hypertension have been the third and eighth leading causes of death in Taiwan since 2000 [1]. According to a recent study, the percentage of the population with a prescription for antihypertensive drugs in Taiwan has increased from 2001 to 2006 [2]. The authors of this study report that during this period, the average annual increase in prescriptions for calcium channel blockers (CCBs), angiotensin II receptor blockers (ARBs) and angiotensin-converting-enzyme inhibitors (ACEIs) were 10.7, 22.1 and 4.5 %, respectively [2]. In 2013, the sale volume of the three leading antihypertensive drugs in Taiwan amounted to about US$ 5 million; in comparison, in the USA the value of prescriptions filled for antihypertensive drugs in 2013 totaled about US$ 678.2 million [3].

The use of antihypertensive agents (AHs) has grown globally over the last decade. However, available data on a potential association between the use of AHs and risk of breast cancer are conflicting. Recent epidemiological studies suggest that beta-blockers prevent breast cancer progression or reduce recurrence and then improve survival [4–6]. In contrast, other studies have reported an increased risk or no association at all between the use of beta-blockers/CCBs and breast cancer risk [7–9]. In addition, evidence for any association between the use of ACEIs/ARBs and breast cancer is also inconsistent, with some studies suggesting that ACEIs/ARBs are not associated with cancer risk [10, 11], and others reporting an increased or reduced risk [12].

To address the conflicting evidence from previous studies, the aim of the study reported here was to evaluate the risk of breast cancer associated with long-term use of AHs in hypertensive women.

Methods

Data Source

Data were retrieved from the National Health Insurance Research Database (NHIRD) and Registry for Catastrophic Illness Patient dataset (HV dataset) between January 1, 1998 and December 31, 2011 in Taiwan. The NHIRD contains comprehensive information on demographic characteristics, pharmacy records and medical services from inpatient, outpatient and emergency care under a national health insurance program in which over 99% of the 23 million inhabitants of Taiwan are enrolled. The HV dataset comprises specific data subsets of the NHIRD for research purposes and contains registration files and original claim data on patients registered in the NHIRD who have/had a catastrophic illness. All patients records/information were de-identified and analyzed anonymously. Therefore, this study was exempt from the approval by the Ethics Review Board at our institution.

Study Group

From the HV dataset, we identified 330,699 women with newly diagnosed hypertension [International Classification of Disease, Ninth Revision (ICD-9 CM) codes 401–405] who had been treated with any AHs continuously for at least 6 months between January 1, 1998 and December 31, 2011. Among these, we further identified women with a first diagnosis of breast cancer (ICD-9 CM codes 174.xx and 175.xx); these women were the cases in our study (Fig. 1). The date of diagnosis was the index date.

Fig. 1
figure 1

Study flow diagram. AHT Antihypertensive, H/T hypertension, HV Registry for Catastrophic Illness Patient dataset, NHIRD National Health Insurance Research Database

Full size image

We excluded patients who had a history of breast cancer or any cancer recorded in the HV dataset any time before the initiation of antihypertensive treatment and patients without continuous enrolment in a NHI program. Patients were followed from the date of diagnosis of hypertension in 1998 up to December 31, 2011 (median duration 13 years) or death, whichever came first (Fig. 1).

We randomly selected hypertensive women registered in the NHIRD without any diagnosis of breast cancer who were receiving treatment for hypertension in the same period as the cases. These were matched (1:4) for age (5-year categories), index date and year of hypertension diagnosis with the cases to establish the control group (Fig. 1).

Exposure Variables

The main exposure of interest was that to beta-blocker, CCB, ACEI and ARB therapy. We collected information on prescribed drug types according to Anatomical Therapeutic Chemical Classification System codes (C07 for beta-blockers; C02D, C08C, C08D, C08DA51 for CCBs; C02E, C02L, C09A, C09BA for ACEIs; C09CA for ARBs), dosage, date of prescription, supply days and total number of prescriptions from the outpatient and inpatient records [13]. The cumulative defined daily dose (cDDD) of each AH was calculated as recommended by the World Health Association [14]. Beta-blockers were further classified as nonselective and beta-1 selective beta-blockers, and as selective and nonselective alpha-blockers.

Potential Covariates

Several potential covariates, including age and comorbidities at cancer diagnosis, were also measured in the year preceding the index date. Other covariates tested included the use of statins and hormone replacement therapy.

Sensitivity Analysis

We evaluated the sensitivity effects by changing the inclusion criteria of drug prescription for three types of AH beginning at least from 6–9 months before the index date.

Statistical Analysis

Logistic regression was used to estimate the crude and adjusted odds ratio (OR) and 95% confidence interval (CI) for breast cancer risk. We calculated a running sum of the duration and DDD of each drug from the date of the initial AH prescription to the index date. We categorized the cumulative use for each patient as follows: ≤1, 1–2, 2–3 and ≥3 years of duration. Cumulative DDD of each AH was classified by quartile. Multivariable logistical regression was used to adjust the covariates. We also estimated the trend of the duration and cDDD of each drug use. Data were analyzed using the SAS Statistical Package, version 9.3 (SAS Institute, Cary, NC). The significance level was set at P < 0.05 (two-tailed test).

Results

We identified 6,463 hypertensive women with breast cancer as cases and 18,987 hypertensive women without breast cancer as controls. Among the 6,463 cases, the most commonly prescribed AHs was CCBs (52.8%), followed by ACEIs (45.5%) and beta-blockers (41.1%) (Table 1). No significant differences in age and mean Charlson comorbidity score (P > 0.05) were found between cases and controls. Ever-use of CCBs and beta-blockers for longer than 10 years was significantly associated with breast cancer (OR 1.09; 95% CI 1.03–1.16) in an adjusted model. The risk of breast cancer was even higher in patients receiving hormone replacement therapy (OR 1.28, 95% CI 1.18–1.39) and statins (OR 1.68, 95 % CI 1.50–1.83) (Table 1).

Table 1 Characteristics of hypertensive patients with breast cancer and non-breast cancer during the study period (1998–2011)
Full size table

When we stratified the risk of breast cancer associated with different sub-types of beta-blockers, we found a statistically significant risk of breast cancer with most beta-1 selective beta-blockers, such as atenolol (OR 1.14; 95% CI 1.05–1.25) acebutolol (OR 1.29; 1.00–1.66) and bisoprolol (OR 1.08; 1.01–1.16) (Fig. 2). The non-selective beta-blockers, alpha-selective and beta-non selective showed no significant association with breast cancer (Fig. 2).

Fig. 2
figure 2

Forest plot of breast cancer risk associated with use of beta-blockers, 1998–2011. OR Odds ratio, CI confidence interval

Full size image

We then stratified beta-blocker, ARB and CCB users by exposure duration and the cumulative DDD. The results show that the risk of breast cancer was significantly increased in beta-blocker and CCB users with increasing exposure duration and increasing cDDD compared to the controls [trend test for beta-blocker users: P = 0.003 (exposure duration), P = 0.0003 (cDDD); trend test for CCB users: P = 0.006 (exposure duration), P = 0.002 (cDDD)] (Table 2).

Table 2 Odds risk and 95% confidence intervals for risk of breast cancer associated with exposure to different types of antihypertensives, duration of exposure and dosage
Full size table

The risk of breast cancer increased with ever-use of atenolol or acebutolol (Table 3). This risk increased with increasing exposure,duration of use (trend test: P = 0.0003 for atenolol; P = 0.01 for acebutolol) and cDDD (trend test: P = 0.002 for atenolol; P = 0.02 for acebutolol).

Table 3 Breast cancer risk associated with exposure duration and dosage of specific beta-blockers during the study period (1998–2011)
Full size table

In the sensitivity analysis for exposure duration of AHs, the results were unchanged when the inclusion criteria of AH prescription was changed from <6 to >9 months (Table 4).

Table 4 Sensitivity analysis for criteria of antihypertensive use
Full size table

Discussion

The results of this study suggest that the use of ACEi, ARBs, and nonselective beta-adrenergic receptor antagonists (propranolol or carteolol) is not associated with breast cancer. These results are consistent with those of most observational studies [10, 11].

We also found that CCBs and the beta-1 selective beta-blockers acebutolol, atenolol and bisoprolol may increase the risk of breast cancer. This finding seems to differ from those of previous studies which reported that beta-1 selective beta-blockers and CCBs had marked protective effects [15, 16]. However, the authors of a recently published study reported observing a weak inverse association between cardio nonselective beta-blockers and breast cancer risk [9]. However, since the association did not reach statistical significance, the results did not support the hypothesis of beta-blocker usage protecting against breast cancer progression [9]. The results of a recently published network analysis indicated a lack of consistency in the effect of CCBs on breast cancer; this was attributed to the short duration of the follow-up in the trials included in the network meta-analysis [7].

The results of previous preclinical studies are inconclusive in terms of whether beta-blockers have agonist activity in breast cancer growth. Some studies has demonstrated that beta-2 adrenergic signaling plays a role in several pathways involved in breast tumor progression and metastasis [17, 18], but others have found that beta-adrenergic receptor (AR) stimulation may both inhibit and promote breast tumor growth [19–23]. A recently published study adds further to the body of evidence on the effect of agonist type, indicating that the beta 2-AR antagonist in particular seems to be the most cytotoxic beta-blocker in non-stimulated cancer cells [24]. However, the majority of clinical observational studies carried out to date have focused on comparing the association between the use of propranolol or atenolol and breast cancer risk or mortality and have not explored the relationship between the subtype of beta-AR expression and breast cancer risk [25, 26]. Our study is the first from Asia to report that treatment with the beta-1 selective blocker—but not the nonselective β1/β blocker—may increase the risk of breast cancer (Fig. 2). These results appear to be consistent with those of preclinical studies suggesting that the effects of beta-adrenergic signaling on tumor progression and metastasis are inhibited by the β2-receptor antagonists but not by β1 antagonists [18–24]. Consequently, better designed observational studies or randomized controlled trials are required before this type of beta-blocker can be considered as a therapeutic option for patients with breast cancer.

We also observed that CCBs are likely to be associated with breast cancer risk. This finding is consistent with those from a recently published study performed by Li et al. [3]. Both studies seem to revive an earlier previous hypothesis and focus on the long-term use of CCBs among current or ever-users (10 years if the study of Li et al. [3]; 13 years in our study). However, other previously published studies found no increased risk of breast cancer associated with CCB use [25, 26]. Therefore, to date, the results on the effect of CCBs on breast cancer risk are inconsistent. Again, larger and more comprehensive studies are needed to confirm the effects of long-term use of CCBs on breast cancer.

A major advantage of our study was that we collected information prospectively on healthcare beneficiaries registered in a large population-based database for whom complete data on drug prescriptions and cancer diagnoses were available. Thus, the possibility of selection and information biases was minimized. However, there were still some limitations to our study. First, the health insurance database that we used was developed for administrative purposes and contained de-identified records of each individual registered. Second, the database only provided information on the frequency and classes of prescribed medications and did not provide any clinical laboratory data or clinical information; therefore, we could not estimate patient’s responses to drug therapy. Finally, the database did not contain information on various lifestyle risk factors for cancer, such as physical activity, alcohol consumption, smoking, body mass index, socioeconomic status and diet; therefore, these were not included in the analysis. Although we adjusted the potential covariates, such as co-morbidities and the use of other medications, the misclassification of these covariates may have some impact on our results.

Conclusion

Our findings indicate that the long-term use of CCBs or beta-1 selective blockers are likely to be associated with breast cancer risk. Further comprehensive and large population-based studies are needed to confirm these findings before any definitive conclusion can be drawn.