Abstract
Background: Breast cancer incidence has been associated with hypertension, which might worsen disease prognosis, but few nationwide studies have investigated the association between antihypertensive drug use and breast cancer prognosis.
Methods: A cohort of 73,170 women diagnosed with breast cancer during 1995–2013 identified from the Finnish Cancer Registry was combined with information on antihypertensive drug use during the same time period from a national prescription database. Antihypertensive drugs were analyzed in groups categorized by mechanism of action. Usage of antihypertensive drugs, statins, antidiabetic, and anticoagulative drugs was analyzed as time-dependent exposure to model for simultaneous use of multiple drug groups. Influence of protopathic bias was evaluated in lag-time analyses.
Results: In prediagnostic use, only angiotensin receptor (ATR)-blockers were associated with decreased risk of breast cancer death as compared with nonusers (HR: 0.76, 95% confidence interval, CI: 0.69–0.82), and there was an inverse association with cumulative dose of use. Postdiagnostic use of ATR-blockers, angiotensin-converting enzyme (ACE)-inhibitors, beta-blockers, and calcium-channel blockers was dose dependently associated with better breast cancer survival compared with nonusers. The risk decrease was strongest for ATR-blockers (HR: 0.69, 95% CI: 0.63–0.75) and remained for exposures occurring up to 3 years earlier.
Conclusions: Only ATR-blockers were associated with improved breast cancer survival in both prediagnostic and postdiagnostic use. The association was dose dependent and supported by a biological rationale as a causal explanation. In postdiagnostic use, similar reduction was found also for other antihypertensives, supporting a prognostic role of hypertension control.
Impact: Inhibition of angiotensin receptor subtype 1 (AT1) could be a promising novel way to affect breast cancer progression.
Introduction
Breast cancer is the most common cancer among women worldwide causing losses in life expectancy. Breast cancer is mainly a women's disease, being rare in men (1).
Many modifiable risk factors for breast cancer have been observed such as smoking and use of oral contraceptives (2–4). Metabolic factors, for example, hypertension, high blood glucose, and abdominal obesity, have also been linked with a higher breast cancer risk and poorer prognosis (5, 6), whereas physical exercise may improve survival (7). Hypertension is very common among patients with breast cancer (8). However, it is unclear whether the use of antihypertensive medication can decrease the risk of breast cancer diagnosis or improve disease prognosis. One study has reported a higher breast cancer incidence among women using calcium-channel blockers (9) and another study stated that this extended to antihypertensive drug use in general (10). Angiotensin receptor blockers have been associated with more advanced disease at diagnosis (11). Several studies have observed no association between antihypertensive drug use and breast cancer incidence (12–14).
Some studies have explored antihypertensive drug use as a risk factor for breast cancer death. Beta-blocker therapy has been associated with better breast cancer survival as compared with nonusers (15–18) but not all studies agree with this claim (19–21). For example, beta-blockers and diuretics have also been associated with poorer breast cancer prognosis (22). No clear survival associations have been found for other antihypertensive drugs (23–25). One challenge in epidemiologic studies is to take into account the commonly occurring simultaneous use of several antihypertensive drugs and the presence of comorbidities. It is also important to try to separate the direct influence of a drug from the indirect influence of underlying conditions such as obesity among hypertensive participants.
Here we analyze the association between antihypertensive drug use and death from breast cancer among Finnish women in a large nationwide cohort while taking into account these challenges.
Materials and Methods
Study cohort
The study cohort was identified from the Finnish Cancer Registry which registers new cancer diagnoses in Finland by obligatory reports from all health care units (26).
A total of 73,170 new breast cancer cases among women were identified from the database. Cases were diagnosed between 1995 and 2013. Data contained information on date and method of diagnosis, tumor extent at diagnosis (recorded in the registry as local, advanced into regional lymph nodes, advanced to distant organs, no information), information on participation in national mammography screening program, tumor histology (ductal, lobular, other, unknown), and primary breast cancer treatment (surgery, other). The data also included dates and causes of cancer death as well as all-cause deaths until the end of 2015. In the Finnish national mammography screening program, every 50- to 69-year-old woman is invited to a free mammography examination every second year to screen for breast cancer.
Information on antihypertensive medication use
The cohort was linked to the national prescription database maintained by the Finnish Social Insurance Institution (SII) for information on antihypertensive drug use during 1995–2013. SII provides reimbursements for physician-prescribed drug purchases in the outpatient setting for every Finnish citizen. Drugs used during the inpatient period or bought over the counter are not recorded but all antihypertensive drugs are available only by prescription and thus are comprehensively recorded. The information on each purchase includes the date, package size, number of packages, and dose for each purchase.
Antihypertensive drugs were identified from the registry data using drug-specific ATC-codes (Supplementary Table S1). They were divided into six different groups based on their mechanism of action: angiotensin-converting enzyme (ACE) inhibitors (inhibiting ACE), angiotensin receptor (ATR) blockers (blocking AT1-receptor), furosemide (increasing diuresis), other diuretics (increasing diuresis), beta-blockers (blocking only B1- or B1- and B2-receptors) and calcium-channel blockers (decrease Ca2+-ion influx in smooth muscle cells). Furosemide was analyzed separately as it is more often used for the management of oedema rather than hypertension.
Information on comorbidities
The cohort was linked to the nationwide Care Registry (HILMO) which is maintained by the Finnish Institute for Health and Welfare (THL) and houses information on procedures and diagnoses in the cohort population during 1995–2013. It contains all diagnoses and medical procedures from both inpatient and outpatient hospital visits in Finland. Diagnoses recorded in Care Registry were used to calculate the Charlson comorbidity index (CCI) for every participant (27). Conditions used in calculation are listed in Supplementary Table S2.
Information on receptor status
Information on tumor pathologic characteristics was supplemented from the archives of the pathology departments of university hospitals in Tampere and Turku, two of the largest cities in Finland. Information on estrogen receptor (ER), progesterone receptor (PR), and HER2 expression was obtained from the databases. Information on HER2 status was available for 8617, ER for 7,283, and PR for 7,288 participants. Receptor status was used in subgroup analyses to evaluate whether there was a possible effect modification by tumor receptor status.
Statistical analyses
Analyses were run separately for drug use before and after breast cancer diagnosis. Use before diagnosis included all usage between January 1, 1995 and year of breast cancer diagnosis. Use after diagnosis included all use from the year of breast cancer diagnosis until death from breast cancer, common closing date December 31, 2013 or emigration, whichever came first. The risk of breast cancer death was compared between antihypertensive drugs users and nonusers in a model including all antihypertensive drug groups simultaneously to model usage of multiple drugs.
The total yearly mg amount of each antihypertensive drug was calculated for every participant based on the dosing, package size and number of packages from each purchase. Total purchased yearly mg amount was divided by the drug-specific defined daily dose (DDD) to obtain the total number of DDDs purchased per year (28). Each year with any recorded purchase was considered as a year of usage regardless of the purchased amount. Antihypertensive drugs are commonly used in combination and therefore usage categories were not mutually exclusive; we used separate time-dependent variables for all antihypertensive drug groups and participants were categorized as users for all antihypertensive drug groups in which they had recorded purchases during one follow-up year. For example, ACE-inhibitors and ATR-blockers are often used in combination with diuretics so that participants were categorized as users of ACE-inhibitors/ATR-blockers and also diuretics during each follow-up year with any use.
Cumulative years of usage and DDDs were calculated separately before and after breast cancer diagnosis. The amount of use before the diagnosis was calculated by adding together all usage from 1995 until the year of breast cancer diagnosis. Intensity (DDDs/year of use) was evaluated by dividing cumulative amount of DDDs with cumulative number of usage years.
Postdiagnostic antihypertensive drug use was analyzed as a time-dependent variable to control for immortal time bias. Time-dependent variables were calculated by updating user status as well as cumulative amount, duration, and intensity of use separately for every follow-up year after breast cancer diagnosis. Dose dependence was evaluated by stratifying medication users into tertiles of cumulative duration, DDD amount and intensity of use based on the level reached on each follow-up year. It is important to notice that after discontinuation of usage, participants remained in the user category to minimize for an error based on selective discontinuation of medication use, for example, if they were at an advanced stage of breast cancer.
Cox regression was used to calculate HRs and 95% confidence intervals (CI) for the risk of breast cancer –specific death. The time variable was years since the breast cancer diagnosis. Follow-up continued until death, participant emigration or the closing date December 31, 2013. Cox regression analyses were adjusted for age at diagnosis, tumor extent at diagnosis, primary treatment of breast cancer (surgery, other), obesity, CCI, participation in national breast cancer screening program, and use of hormone-receptor antagonist therapy after breast cancer diagnosis.
Simultaneous use of multiple antihypertensive drug groups was modeled, that is, forming separate time-dependent variables for the usage of each antihypertensive drug group and also for usage of drugs which are commonly prescribed simultaneously: statins, antidiabetic drugs, and anticoagulant drugs.
Antihypertensive drugs are subject to protopathic bias. This bias favours antihypertensive drug users in cox regression analysis because exposure and outcome occur at the same time. Latency of the risk association and the protopathic bias between antihypertensive drug use and breast cancer death were evaluated in lag-time analyses where the exposure was lagged in the follow-up time by analyzing medication use that occurred 1, 3, or 5 years earlier than the exposure. In terminal stage of cancer, there is a tendency to minimize all nonpalliative drugs including antihypertensive drugs.
The data were analyzed using the IBM SPSS statistics 25 program.
Results
Population characteristics
Antihypertensive drug use was very common in our study cohort; 36,427 (49.8%) women had used at least one group of antihypertensive medication during the study period (Table 1). A total of 11,258 (15.4%) breast cancer cases were screen-detected. When compared with nonusers, antihypertensive drug users were older at diagnosis and at breast cancer death and were more often using also statins and antidiabetic drugs. A total of 10,900 women died of breast cancer during the follow-up, of which 4,542 (124/1,000) were nonusers and 6,358 (175/1,000) were users of an antihypertensive drug.
Population characteristics.
Antihypertensive drug use before breast cancer diagnosis
In prediagnostic age-adjusted analysis, the use of ACE-inhibitors, ATR-blockers, and beta-blockers was associated with a statistically significant reduction in the risk of breast cancer death (Table 2). However, in multivariable adjusted analysis, only the use of ATR-blockers remained associated with a reduced risk (HR: 0.76, 95% CI: 0.69–0.82). The risk decreased among users of ATR-blockers with an inverse association with DDD amount, duration, and intensity of usage (Table 2). No dose dependence was observed for any of the other antihypertensive drug groups. Furosemide and other diuretics were associated with an increased risk of breast cancer death if there was a high intensity of use.
Risk of breast cancer death by antihypertensive drug use before breast cancer diagnosis.
Antihypertensive drug use after breast cancer diagnosis
Postdiagnostic use of all antihypertensive drugs except furosemide and other diuretics was associated with a decreased risk of breast cancer death as compared with nonusers in age-adjusted analyses. In the multivariable adjusted analysis, use of ATR-blockers, beta-blockers, and calcium-channel blockers remained associated with improved breast cancer survival in comparison with nonusers (Table 3). The strongest risk decrease was evident among ATR-blocker users (HR: 0.77, 95% CI: 0.71–0.84). Furosemide associated with a statistically significant increased risk for breast cancer death. When analyzing risk trends according to the cumulative DDD dose and intensity of usage ACE-inhibitors, ATR-blockers, beta-blockers, and calcium-channel blockers all demonstrated a decreasing risk trend, with a lower risk of breast cancer death when compared with nonusers in high-intensity use (Table 4). Furosemide associated with an increased risk of breast cancer death, with the highest risk estimates for the lowest dose and intensity and shortest duration of use; the risk decreased in association with dose, intensity, and duration also for this drug group, but remained elevated as compared with nonusers even with high-intensity use (Table 4). These findings were similar also for other diuretics.
Risk of breast cancer death by antihypertensive drug use after breast cancer diagnosis.
Risk of breast cancer death by antihypertensive drug use after breast cancer diagnosis. Risk estimates by cumulative dose, duration, and intensity of antihypertensive drug use.
In lag-time analyses, the improvement in breast cancer survival among ATR-blocker users persisted after both a 1-year and 3-year lag time, but not longer (Table 3). Calcium-channel blockers associated with a lowered risk of breast cancer death risk after 3 and 5 years' lag time. The risk increase among furosemide users persisted with a 1-year lag time, but vanished thereafter. Other antihypertensive drugs did not associate with breast cancer survival in lag-time analyses.
Subgroup analyses
We evaluated the association between antihypertensive drug use after breast cancer diagnosis and the risk of breast cancer death stratified by hormone receptor status (Table 5). ACE-inhibitors and calcium-channel blockers were associated with a decreased risk of breast cancer death among HER-negative (cancer does not express HER2) women and ACE-inhibitors also among triple-negative (cancer does not express ER, PR, or HER2) women. Other diuretics were associated with a reduced risk among HER-positive (cancer expresses only HER2) women.
Risk of breast cancer death among different postdiagnostic antihypertensive drug use based on tumor receptors.
Age at diagnosis or the year of breast cancer diagnosis did not clearly modify the risk associations. ATR-blockers were associated with improved breast cancer survival both in women with screening-detected breast cancer and women diagnosed outside of the national screening program. The CCI nonsignificantly modified the risk association with ATR-blockers; the risk decrease was observed in women with no comorbidities but not among women with one or more comorbidity. When only participants using one antihypertensive drug group after diagnosis were included, then ATR-blockers and furosemide were associated with improved breast cancer survival (Supplementary Table S3). Results for prediagnostic and postdiagnostic antihypertensive drug use were not changed when primary end point was all-cause death (Supplementary Tables S4 and S5). The risk is lowered for ATR-blockers in both analysis and thus competing risks do not explain differing risk estimates for ATR-blockers.
In many studies, antihypertensive drug use only at time of breast cancer diagnosis has been evaluated. To assess comparability between our study and previous studies, we also analyzed the association between antihypertensive use at time of breast cancer diagnosis (use during year of breast cancer diagnosis) and risk of breast cancer death. In this analysis, the results were similar to main analysis as ATR-blockers were associated with better breast cancer survival compared with nonusers (Supplementary Table S6). In this analysis, nonselective (block both B1- and B2-receptors such as propranolol and pindolol) and selective beta-blockers (block only B1-receptor such as bisoprolol and karvedilol) were analyzed as separate drug groups. Beta-blocker selectivity did not modify risk association as no association was found between either group and risk of breast cancer death. To minimize selection bias between antihypertensive drug users and nonusers, we run analysis again with limiting it to only participants using at least one group of antihypertensive drugs. This did not change the results (Supplementary Table S7). We also estimated whether risk association was modified by tumor receptors and run prediagnostic and postdiagnostic analysis in different tumor receptor subgroups separately. No risk modification was observed by tumor receptor status. However, this analysis was limited because of low statistical power (Supplementary Table S8).
Discussion
ATR-blockers differ from the other antihypertensive drugs as they were associated with a dose-dependent improvement in breast cancer survival as compared with nonusers and this was consistent for both prediagnostic and postdiagnostic use and the association was seen also after 3 years′ lag time. These findings point to a causal risk association, supporting a prognostic role of angiotensin receptor inhibition in breast cancer.
ACE-inhibitors and ATR-blockers both inhibit the renin–angiotensin–aldosterone system (RAA system). However, we observed differing risk associations between ACE-inhibitors and ATR-blockers. ATR-blockers block selectively only the AT1-receptor (angiotensin receptor subtype 1) and leave free other AT-receptor subtypes [e.g., angiotensin receptor subtype 2 (AT2)] while ACE-inhibitors block the whole pathway by inhibiting the formation of angiotensin. The mechanism of action of ATR-blockers thus allows angiotensin to activate AT2-receptors which in turn have been proved to induce apoptosis in heart endocardial endothelial cells and to reduce the growth of lung adenocarcinoma cells in vitro (29, 30). The role of the RAA signaling pathway in breast cancer cells is mainly unexplored. Our results suggest that AT2-receptor activity may have an important role also in breast cancer progression.
In analyses of postdiagnostic use, also beta-blockers and calcium-channel blockers were associated with better breast cancer survival as compared with nonusers. Results in the same direction were seen also among ACE-inhibitors. In previous studies, beta-blockers have been associated with decreased breast cancer death risk (15–18). Our results differ from those even if our statistics was nationwide and exceptionally detailed. However, among women suffering from triple-negative breast cancer HR for beta-blocker use was 0.79 (Supplementary Table S8) which may indicate that beta-blockers have breast cancer progression inhibiting effects at least in some tumor subtype groups. More research with higher statistical power is needed.
The risk decrease was dose dependent for several antihypertensive drugs with differing mechanisms of action. This suggests that the underlying condition serving as a common indication for usage, this is, hypertension and its treatment, may explain the risk association; good control of hypertension after breast cancer diagnosis might improve breast cancer prognosis. In lag-time analyses, the risk decrease among ATR-blockers was observed for 1 to 3 years after use, although not for 5 years. Thus, our results are not likely to be explained by a protopathic bias, that is, cancer or its treatment affecting antihypertensive medication use rather than vice versa. In that case, the risk association should vanish already with short lag times, as was observed for furosemide.
Nevertheless, our results may be affected by a healthy user bias; medication users may have a healthier lifestyle in general as compared with nonusers, thus creating a bias favouring the users. However, this kind of bias should not be dose dependent; therefore, the dose-dependent risk associations support causality.
The use of furosemide and other diuretics was associated with an elevated risk of breast cancer death. The risk of breast cancer death decreased with increasing intensity of use of furosemide and other diuretics. Furosemide is not primarily used to control hypertension, instead it is prescribed to treat oedema and heart failure. Conditions like oedema are frequently present in terminal breast cancer which could cause a protopathic bias. Because the risk increase was observed when furosemide alone was used after diagnosis, this does seem to indicate that it is the underlying conditions which are responsible for the risk increase among furosemide users, that is, furosemide is not used alone, but in combination with ACE-inhibitors or ATR-blockers in the management of terminal stage heart failure. Other diuretics alone associated with an increased risk of breast cancer death which is probably explained by their administration in the treatment of oedema in advanced cancer. The risk increase mitigated along with increasing cumulative usage; when furosemide is used to treat oedema caused by advanced cancer, the patient usually has a limited life expectancy; this explains why a protopathic bias mostly affects the results seen for the subgroup with low cumulative usage.
There are some previous studies investigating the association between different antihypertensive drugs and the risk of breast cancer death. Beta-blockers have been associated with better survival in previous studies and our results are concordant (16–18). However, these studies have only evaluated beta-blocker use before and at the time of diagnosis, not taking into account simultaneous use of other antihypertensive drugs. There are also studies which have evaluated the risk for breast cancer recurrence which cannot be directly compared with our primary endpoint risk of breast cancer death (20, 22). One study has reported that ever-use of ATR-blockers was not associated with breast cancer survival (25). Another study observed improved breast cancer survival associated with the postdiagnostic use of ACE-inhibitors and/or ATR-blockers (ACE-inhibitors and ATR-blockers analysed together; ref. 31); in that situation, the possible effects of ATR-blockers could not be separated from the influence of ACE-inhibitors.
The strength of our study was our reliable registry-based data on drug use which was detailed and free of recall bias (26, 32). We also had a long follow-up time and a large national cohort consisting of all breast cancers diagnosed in Finland in the period 1995–2013. We compared multiple antihypertensive drug groups with different mechanisms of action and were able to evaluate if a single drug group had anticancer suggesting effects. We were able to take into account the simultaneous use of different antihypertensive drugs, statins, anticoagulant drugs, and antidiabetic drugs, thus controlling for possible confounding. We were also able to adjust the analysis for primary treatment as antihypertensive drug users likely had more comorbidities limiting their possibilities for curative surgery. However, we adjusted the analysis for comorbidities also using CCI.
We did not have information on blood pressure levels which may affect the results if hypertension is a prognostic factor. However, we were able to evaluate this indirectly by taking into account the simultaneous use of different antihypertensive drugs. The possible prognostic effect of hypertension should be observed for all drug groups regardless of mechanism of action, as was the case seen with postdiagnostic use. We did not have information on whether purchased drugs had actually been consumed, which may dilute the observed risk estimates toward the null. We do not have information on socioeconomic or lifestyle factors such as physical activity, smoking habits, body mass index, or nutrition which could have served as confounders depending on their association with medication use and breast cancer survival. The prescribing of ATR-blockers became more common in the late 1990s whereas other antihypertensive drugs were on the market already before that time. This may have caused a selection bias if newer drugs have been prescribed selectively to those individuals with a better prognosis or fewer comorbidities. It is also possible that participants using antihypertensive drugs have higher other-cause mortality, for example, coronary artery disease compared with nonusers which could explain their lower risk of death from breast cancer. However, we were able to use CCI and to standardize the consumption of several drug groups to control for this possible bias. Again, such biases should not explain the dose-dependent risk trends and should be seen in same direction in all drug groups.
In conclusion, the prediagnostic use of only ATR-blockers associated with better breast cancer survival in comparison with nonusers. With respect to postdiagnostic use, multiple antihypertensive drug groups associated with better breast cancer survival. Here too, the association found was strongest for ATR-blockers, which may point to a molecular mechanism behind the effect. However, control of hypertension may also have a prognostic influence after the breast cancer diagnosis. More research into the role of the AT2-receptor in breast cancer is warranted.
Disclosure of Potential Conflicts of Interest
E.E.E. Santala reports grants from Orion Research Foundation and the Cancer Foundation Finland outside the submitted work. T.J. Murtola reports grants from Expert Responsibility Area of the Pirkanmaa Hospital District during the conduct of the study; personal fees from Astellas (consultation fee and lecture fee), Janssen-Cilag (consultation fee and lecture fee), and Ferring (consultation fee) outside the submitted work; as well as a patent for TK1 predicting cancer patient survival pending to AroCell AB. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
E.E.E. Santala: Resources, data curation, formal analysis, funding acquisition, investigation. M.O. Murto: Writing–review and editing. M. Artama: Writing–review and editing. E. Pukkala: Writing–review and editing. K. Visvanathan: Writing–review and editing. T.J. Murtola: Conceptualization, supervision.
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.
Footnotes
Note: Supplementary data for this article are available at Cancer Epidemiology, Biomarkers & Prevention Online (http://cebp.aacrjournals.org/).
Cancer Epidemiol Biomarkers Prev 2020;29:2376–82
- Received May 11, 2020.
- Revision received July 3, 2020.
- Accepted August 21, 2020.
- Published first September 11, 2020.
- ©2020 American Association for Cancer Research.
References
Volume 29, Issue 11
