Long term antihypertensive drug use and prostate cancer risk: A 9-year population-based cohort analysis

International Journal of Cardiology 193 (2015) 1–7

Contents lists available at ScienceDirect

International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard

Long term antihypertensive drug use and prostate cancer risk: A 9-year population-based cohort analysis Pei-Ying Pai a,b, Vivian Chia-Rong Hsieh c,1, Chang-Bi Wang d, Hsi-Chin Wu e,f, Wen-Miin Liang g,h, Yu-Jun Chang d,i, Trong-Neng Wu j,⁎ a

Division of General Medicine, Department of Internal Medicine, China Medical University Hospital, Taichung, Taiwan Division of Cardiovascular Medicine, Department of Internal Medicine, China Medical University Hospital, Taichung, Taiwan Department of Health Services Administration, China Medical University, Taichung, Taiwan d Graduate Institute of Public Health, China Medical University, Taichung, Taiwan e Department of Urology, China Medical University Hospital, Taichung, Taiwan f School of Medicine, China Medical University, Taichung, Taiwan g Graduate Institute of Biostatistics, China Medical University, Taichung, Taiwan h Biostatistics Center, China Medical University, Taichung, Taiwan i Epidemiology and Biostatistics Center, Changhua Christian Hospital, Changhua, Taiwan j Department of Nursing, College of Medicine and Nursing, Hungkuang University, Taichung, Taiwan b c

a r t i c l e

i n f o

Article history: Received 17 May 2014 Received in revised form 21 April 2015 Accepted 7 May 2015 Available online 9 May 2015 Keywords: Antihypertensive drugs Hypertension Prostate cancer

a b s t r a c t Background: Recent findings from clinical trials have indicated inconsistent associations between angiotensin II receptor blockers and the risk of cancer incidence. Furthermore, the relationship between antihypertensive drugs and prostate cancer in hypertensive patients remains unclear. Methods: From Taiwan's national health insurance database, we identified 80,299 patients diagnosed with hypertension in 2001 and matched with 321,916 subjects without hypertension by age, income, urbanization level, and index day. A total of 684 hypertensive patients without antihypertensive drug use (drug non-user subcohort) were also matched (1:4) with 2736 patients on antihypertensive medication (drug subcohort) using the same criteria. Each subject in the two study groups was followed up for a maximum of nine years, during which death was considered a competing event when performing the stratified Fine and Gray regression hazards model for the estimation of prostate cancer risk for the cohorts. Uptake of antihypertensive prescription was considered a time-dependent variable. Results: Our findings indicate that patients with hypertension are at significantly increased risk for prostate cancer incidence when compared to their matched non-hypertensive counterparts (sHR = 6.80, 95% CI = 1.97–23.44, p = 0.0024). Among hypertensive patients, those with long term antihypertensive drug use are not at elevated risk of developing prostate cancer relative to non-users of antihypertensive drugs (1–5 year vs. non-user sHR = 0.99, 95% CI = 0.32–3.05; N 5 year vs. non-user sHR = 0.88, 95% CI = 0.34–2.26). Conclusions: Hypertension is considered a risk factor for prostate cancer. However, long term uptake of antihypertensive medication in male hypertensive patients should not be a concern for the development of prostate cancer. © 2015 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Hypertension is a highly prevalent morbidity in older adults and acts as a major risk factor for cardiovascular diseases, congestive heart failure, and coronary heart disease [1,2]. The beneficial therapeutic effect of both medical and non-pharmacological interventions on the aggressive

⁎ Corresponding author at: Department of Nursing, College of Medicine and Nursing, Hungkuang University, 1018 Sec. 6 Taiwan Boulevard, Taichung 433, Taiwan. E-mail address: [email protected] (T.-N. Wu). 1 Equal contribution as the first author.

http://dx.doi.org/10.1016/j.ijcard.2015.05.042 0167-5273/© 2015 Elsevier Ireland Ltd. All rights reserved.

and effective control of blood pressure has been well-established in the scientific literature. Several classes of antihypertensive drugs are currently available for initial therapy, including diuretics, alpha- and beta-blockers, aldosterone antagonists, calcium-channel blockers (CCB), angiotensinconverting enzyme inhibitors (ACEI), angiotensin II receptor blockers (ARBs), and direct renin inhibitors (DRIs). Significant reductions in cardiovascular and renal morbidity and mortality through systemic and intensive use of these blood pressure-lowering medications have been previously demonstrated [3]. Patients typically require two or more drugs to effectively control their blood pressure. Nevertheless, ingestion of antihypertensive drugs over a cumulative period of time can potentially lead to adverse effects in older individuals, who tend to be accompanied

2

P.-Y. Pai et al. / International Journal of Cardiology 193 (2015) 1–7

by comorbidities such as hypercholesterolemia, diabetes mellitus, chronic renal disease, and other cardiovascular diseases [3–5]. A population-based case–control study in Finland by Kemppainen et al. [6] describes a marginally increased risk for prostate cancer (PC) in antihypertensive drug users. This positive association, however, does not significantly differ with the cumulative dose or duration of use. Concurrent health conditions that hypertensive patients already possess, in addition to the decreased blood circulation that antihypertensive agents induce, are potential promoters of carcinogenesis. Conversely, in another recent epidemiological study, Poch et al. [7] indicate a lack of association between calcium-channel blockers and PC. The association between the use of ARBs and increased cancer risk was first identified in the Candesartan in Heart failure Assessment of Reduction in Mortality and morbidity (CHARM) trial of candesartan in 2003 [8]. It was proposed that angiotensin type-1 (AT1) and type-2 (AT2) receptors were involved with the regulatory process of cellular proliferation, angiogenesis, and tumor progression [9]. Trials conducted investigating telmisartan, ONTARGET, and TRANSCEND also noted higher hazard for malignancies [10]. These trials commonly had an average follow-up period exceeding three years. A meta-analysis performed by Sipahi et al. [11], also revealed a modest increase in risk of cancer incidence. Conversely, the ARB Trialists Collaboration found no significant increase in overall cancer incidence with long term ARB use in the 15 clinical trials examined [12]. Overall, only a few existing studies were conducted in a cohort design, with a long term follow-up period, and a sufficiently large sample, to examine the cumulative dose effect of antihypertensive agents on the development of PC and the effect of non-drug use in the presence of hypertension. Furthermore, most studies did not study the drug switching behavior commonly seen among hypertension patients and only a few investigated the use of all drug classes which resemble more closely to pragmatic settings. In realization of these shortcomings, we evaluate here the risk of developing PC in hypertensive patients with up to nine years of follow-up in a nationwide study. Among hypertensive individuals, we aim to also compare hazard ratios between the users and non-users of antihypertensive drugs with consideration of cumulative dose effect in a timedependent manner. 2. Material and methods 2.1. Data source Data used in this study is based on claims data from the compulsory National Health Insurance (NHI), which is managed and provided by The Collaboration Center of Health Information Application (CCHIA), Ministry of Health and Welfare of Taiwan. This comprehensive database contains health care data of over 99% of the nation's population including demographics, dates of clinical visits, diagnostic codes, prescription orders, and associated expenditures for which contracted facilities file their medical claims inclusive of services from laboratory tests to treatment and prescription orders. Numerous cardiovascular and cancer epidemiology studies using this database can be found in the published literature [13,14]. 2.2. Ethics statement Because the CCHIA-NHI database entirely consists of anonymous and encrypted secondary data released to the public for research purposes, this study was exempted from a full review by the ethics review committee at the China Medical University Hospital. 2.3. Subject selection Subjects were selected from the CCHIA-NHI database: they included male patients aged 50 or above with newly diagnosed hypertension (International Classification of Disease, 9th Revision, Clinical Modification

(ICD-9-CM) diagnostic codes 401 to 405) in 2001. At least one of the following enrollment criteria had to be met for inclusion in the study: (1) one or more inpatient admissions with diagnosis of hypertension, or (2) three or more outpatient visits within a 6-month period, each with a diagnosis of hypertension. Subjects from the control cohort were selected in the same year from the CCHIA-NHI database on the basis that they were not diagnosed with hypertension nor had any use of antihypertensive drugs. They were individually matched at a ratio of 4:1 with the hypertension cohort by age, urbanization level, income, and index day (i.e., date of newly diagnosed hypertension) (Fig. 1). Prescribed use of antihypertensive drugs in the follow-up period was considered: prescription records contained dates of order, dosage, route of every prescription, and number of days prescribed for each dispensed drug. Furthermore, two subcohorts were then categorized from the hypertension cohort: one with regular use of antihypertensive drugs (drug subcohort), and the other without any antihypertensive drug use (drug non-user subcohort) during the follow-up period. Regular drug use was defined as uninterrupted use for a minimum of one year. Drug non-users were matched (1:4) with drug users according to age, urbanization level, income, and index day (Fig. 1). Index day for the drug non-users was assigned as the date of incident hypertension diagnosis, and index day for the drug users was assigned as the date of new diagnosis which coincided with the prescription of antihypertensive drugs. Patients with diagnosis of cancer prior to or within the first year following index day were excluded from the study to preclude immortal time bias. Comorbidities were classified as those existing prior to the index day and included congestive heart failure, diabetes mellitus, cardiovascular disease (CVD), and hyperlipidemia. The end of follow-up period for the two-part analysis (hypertension vs. non-hypertension cohorts, and hypertensive drug subcohort vs. hypertensive drug non-user subcohort) was marked on the day of PC diagnosis, terminated enrollment from the NHI, death, or until the end of this study [11]. Follow-up data was available up to a maximum of nine years for study subjects. 2.4. Identification of prostate cancer cases Taiwan's NHI has established a well-utilized program to alleviate the financial burden of cancer patients following their cancer diagnoses. But because costs involved in cancer treatment can be substantial, strict guidelines are in place requiring careful examination of medical records before patients can qualify for coverage. In order to be covered by the aid program, patients have to apply for a cancer catastrophic illness certificate. Successful candidates must possess adequate evidence supporting their diagnosis of cancer such as histology or pathology reports, additional laboratory evidence, and clinical images which can include tumor marker surveys, X-ray, bone scan, computed tomography scan, or magnetic resonance imaging. Detailed inspection of patient medical records and laboratory information (including images) by a minimum of two oncologists is also required. Such medical information and data on the cancer patients are compiled in the Registry of Catastrophic Illness Patient Database. Since inclusion in the database requires close scrutiny of cancer diagnoses, we believe to have retrieved reliable information from this database for our identification of PC diagnoses (ICD-9-CM = 185x). 2.5. Antihypertensive drugs We calculated the daily dosage for the drugs and evaluated amount of drug use with defined daily dose (DDD). DDD is the recommended daily average maintenance dose for a drug, which is needed to provide its main indication in adults. Moreover, we constructed time-varying covariates for antihypertensive drug use in the analysis. Seven major groups of antihypertensive agents were included and their Anatomical Therapeutic Chemical (ATC) classifications are listed in Supplementary Table 1.

P.-Y. Pai et al. / International Journal of Cardiology 193 (2015) 1–7

3

The Collaboration Center of Health Information Application (CCHIA) National Health Insurance (NHI) Database containing insurance claims data for more than 23-million residents and enrollees in Taiwan Exclusion: History of prostate cancer Matched: Age; Urbanization level; Income level; Index date Hypertension Cohort (N=80,299): Subjects were selected from the CCHIA-NHI

Control Cohort (N=321,196): Subjects were selected from the CCHIA-NHI

database and included patients aged 50 or above with hypertension diagnostic codes (ICD9-CM 401 to 405) in 2001.

database, excluding those with diagnosis of hypertension or any use of antihypertensive drugs in 2001.

Matched: Age; Urbanization level; Income level; Index date Drug Nonuser Subcohort (n=684): Without use of any antihypertensive drug

Event: 1. PC occurrence 2. Terminated insurance 3. Death

Drug User Subcohort (n=79,615 n=2,736): Regular use of antihypertensive drugs

Event: 1. PC occurrence 2. Terminated insurance 3. Death

Fig. 1. Flow diagram of subject selection and event definition.

2.6. Statistical analysis Baseline characteristics and comorbidities for all cohorts were first analyzed. Frequency of antihypertensive drug use (days per year) was calculated for the follow-up period. Traditional survival analysis typically only considers one event at a time (e.g., death or PC), possibly causing other events to be overlooked and the resulting risk estimates overestimated. Thus, these results should not be directly interpreted and applied in clinical settings. To overcome this issue, our study considered the competing risk of death using the Fine and Gray regression hazards model in our calculation of subdistribution hazards (sHRs) — a method found to be adopted by previous studies [15,16]. We compared both the drug use and non-use subcohorts to the control cohort in order to evaluate the

varying risks of developing PC — with the stratified Fine and Gray regression hazards model, adjusting for age, urbanization level, income, comorbidities, and antihypertensive drug (as time-dependent variable). This time-varying approach ascertained inclusion of the temporal effect and cumulative dose of drug uptake. The Kaplan–Meier method was adopted to determine the cumulative incidence of PC in both cohorts, and differences between cohorts were tested using log-rank test. To examine whether the main findings met different assumptions, sensitivity analyses were performed also with the Fine and Gray competing risks regression model on subgroups classified by age. All data management and sHR calculations were done using the Statistical Analysis System (SAS) software for Windows (version 9.3; SAS Institute, Cary, NC). The Fine and Gray regression hazards model was performed using the PHREG package.

4

P.-Y. Pai et al. / International Journal of Cardiology 193 (2015) 1–7

3. Results The hypertension cohort consisted of 80,299 patients identified from 1 January 2001 to 31 December 2001. Among all subjects without any diagnosis of hypertension or use of hypertensive medication at baseline, we randomly selected 321,196 people for the control cohort to be matched with the hypertension cohort by age, income, urbanization level, and index day after exclusion of ineligible subjects (Fig. 1). After matching, distributions of age, urbanization level, and income were similar between the hypertension and control cohorts (Table 1). Most subjects were found in the aged 60–69 and 70–79 groups, which agreed with the characteristic of PC. A substantially higher proportion of hypertensive patients had comorbid conditions including congestive heart failure, diabetes, hyperlipidemia, and cardiovascular disease (CVD) compared to the controls. Cumulative incidence of PC was significantly higher in the hypertension cohort compared with the control as shown in the Kaplan–Meier curves (Fig. 2). Based on the stratified Fine and Gray model (Table 2), after adjusting for age, urbanization level, income, and comorbidities, the risk of developing PC in the hypertension cohort compared to its matching control had an adjusted sHR of 6.80 (95% confidence interval (CI) = 1.97–23.44, p = 0.0024). A subcohort of 2736 hypertensive patients with antihypertensive treatment was subsequently identified (drug subcohort), while 684 patients not on any antihypertensive medication were also selected (nondrug subcohort). We postulated that these latter individuals were not on medication because they were seeking complementary medicine or self-healing therapies which are relatively more common in an Asian context. Subsequent to matching, baseline distributions of the subcohorts were similar for the matched variables (Table 3). For the prevalence of comorbidities, however, a higher proportion of drug users had diabetes (27.60% vs. 22.37%) and hyperlipidemia (14.73% vs. 6.58%) relative to the non-users. Conversely, congestive heart failure and CVD were more prevalent in the non-drug patients. As shown in Fig. 2 (bottom curves), cumulative incidence for drug users and non-users did not differ significantly, although that of the latter group appeared to be slightly higher during the initial period of observation. In addition, duration of cumulative drug use did not appear

Table 1 Baseline characteristics of hypertension and control cohorts.

Age (year) 50–59 60–69 70–79 ≥80 Drug (days/year)a Prostate cancer Urbanization level Highest High Middle Low Lowest Income level (NTD) b10,000 10,000–20,000 20,001–30,000 30,001–40,000 40,001–50,000 N50,000 Comorbidities Congestive heart failure Diabetes Hyperlipidemia CVD

Control (%)

Hypertension (%)

N = 321,196

N = 80,299

69.28 ± 9.34 59,850 (18.63) 98,088 (30.54) 125,116 (38.95) 38,142 (11.87) – 809 (0.25)

69.31 ± 9.31 14,969 (18.64) 24,186 (30.12) 31,588 (39.34) 9556 (11.90) 275.72 ± 124.39 1162 (1.45)

97,712 (30.42) 73,392 (22.85) 49,156 (15.30) 52,613 (16.38) 48,323 (15.04)

24,428 (30.42) 18,348 (22.85) 12,289 (15.30) 13,777 (17.16) 11,457 (14.27)

169,448 (52.76) 114,848 (35.76) 12,186 (3.79) 8438 (2.63) 7832 (2.44) 8444 (2.63)

44,139 (54.97) 26,935 (33.54) 3216 (4.01) 1940 (2.42) 2177 (2.71) 1892 (2.36)

148 (0.05) 4646 (1.45) 1598 (0.50) 1153 (0.36)

19,467 (24.24) 23,132 (28.81) 12,350 (15.38) 13,210 (16.45)

NTD: new Taiwan dollars; CVD: cardiovascular disease. a All drug types.

Hypertension

Control

Non-drug

Drug

Fig. 2. Cumulative incidence of prostate cancer using Kaplan–Meier survival curves (top: hypertension vs. control; bottom: drug vs. non-drug subcohort).

to exert an influence on PC development. Using the stratified Fine and Gray survival model for the subcohort analysis (Table 4), the PC risk in the ‘1–5 year drug use’ group compared to the ‘non-user’ group had a crude sHR of 1.04 (95% CI = 0.36–2.97, p = 0.945). Adjusted sHRs of 0.99 (95% CI = 0.32–3.05, p = 0.991) and 0.88 (95% CI = 0.34–2.26, p = 0.793) were found in the ‘1–5 year’ and ‘N5 year’ subgroups when compared to its ‘non-user’ group, respectively. In the sensitivity analyses, we further examined the risk of PC incidence for each of the age groups: 50–59 years, 60–69 years, 70–79 years, and ≥80 years. Although we observed wider ranges for the 95% CI due to smaller sample sizes, the lack of statistical significance of sHRs for PC incidence was consistent across different age groups. Thus, our results from sensitivity analyses were robust (Table 5). We also considered effect of different drug classes. Even though drug classes were not included in the final analyses after stratification because the sample sizes were inadequate for adjusted models (Supplementary Table 2), Kaplan–Meier survival

Table 2 The stratified Fine and Gray survival model for the prediction of prostate cancer development (hypertension versus control cohort). Crude sHR Hypertension vs. control Age (year) Comorbidities Congestive heart failure Diabetes Hyperlipidemia CVD

Adjusted p-Value sHR

p-Value

7.20 (6.53–7.94) b0.001

6.80 (1.97–23.44) 0.002

1.10 (1.02–1.19)

0.010

0.99 (0.92–1.08)

0.893

7.00 (5.78–8.49) b0.001

0.96 (0.77–1.21)

0.744

4.90 (4.12–5.84) b0.001 7.27 (5.78–9.15) b0.001 4.47 (3.44–5.81) b0.001

0.98 (0.79–1.22) 1.41 (1.08–1.84) 0.72 (0.54–0.97)

0.861 0.012 0.029

sHR: subdistribution hazard ratio; CVD: cardiovascular disease.

P.-Y. Pai et al. / International Journal of Cardiology 193 (2015) 1–7 Table 3 Baseline characteristics of hypertensive patients in drug user and non-user subcohorts.

Age (year) 50–59 60–69 70–79 ≥80 Drug (days/year)a Prostate cancer Urbanization level Highest High Middle Low Lowest Income (NTD) b10,000 10,000–20,000 20,001–30,000 30,001–40,000 40,001–50,000 N50,000 Comorbidities Congestive heart failure Diabetes Hyperlipidemia CVD

Non-drug (%)

Drug (%)

N = 684

N = 2736

69.31 ± 10.75 165 (24.12) 185 (27.05) 214 (31.29) 120 (17.54) – 7 (1.02)

69.27 ± 10.71 660 (24.12) 741 (27.08) 860 (31.43) 475 (17.36) 274.60 ± 123.69 41 (1.50)

207 (30.26) 154 (22.51) 103 (15.06) 117 (17.11) 103 (15.06)

828 (30.26) 616 (22.51) 412 (15.06) 467 (17.07) 413 (15.10)

385 (56.29) 228 (33.33) 30 (4.39) 15 (2.19) 15 (2.19) 11 (1.61)

1454 (53.14) 998 (36.48) 107 (3.91) 73 (2.67) 59 (2.16) 45 (1.64)

185 (27.05) 153 (22.37) 45 (6.58) 129 (18.86)

643 (23.50) 755 (27.60) 403 (14.73) 436 (15.94)

NTD: new Taiwan dollars; CVD: cardiovascular disease. a All drugs.

curves for cumulative PC risk among different drug groups are presented in Supplementary Fig. 1. 4. Discussion Renin–angiotensin–aldosterone system (RAAS) inhibitors have been introduced for the management of hypertension since the early 1980s and used extensively on hypertensive patients, as well as for conditions related to heart failure and diabetic nephropathy. Inhibition of the RAAS has a major role in controlling blood pressure and hypertension-related complications: ACEIs were the first class of drugs used to block the RAAS. Clinical trials have also documented the positive effect of RAAS inhibitors on improving cardiovascular outcomes. The ARBs, which belong to another class of RAASantagonistic antihypertensive drugs, are also potent on cardiovascular protection [3,17–20]. Although these drugs hold a central role in the control and maintenance of blood pressure, cardiovascular events, and

Table 4 The stratified Fine and Gray survival model for the prediction of prostate cancer development among hypertensive patients (non-user subcohort as reference). Crude sHR Age (year) Drug use Non-use 1–5 year N5 year Comorbidities Congestive heart failure Diabetes Hyperlipidemia CVD

Adjusted p-Value sHR

p-Value

1.02 (0.57–1.81) 0.950

1.05 (0.56–1.95) 0.886

1.00 (reference) 1.04 (0.36–2.97) 0.945 1.13 (0.47–2.72) 0.794

1.00 (reference) 0.99 (0.32–3.05) 0.991 0.88 (0.34–2.26) 0.793

0.71 (0.30–1.68) 0.431

0.63 (0.25–1.58) 0.325

0.92 (0.40–2.10) 0.840 2.31 (0.91–5.89) 0.080 0.94 (0.35–2.49) 0.896

0.74 (0.30–1.78) 0.495 2.63 (0.96–7.16) 0.059 0.79 (0.28–2.20) 0.651

sHR: subdistribution hazard ratio; CVD: cardiovascular disease.

5

renal hemodynamics, accumulating evidence is suggesting that RAAS blockade can participate in the regulation of tumor angiogenesis [21]. In recent decades, the use of RAAS inhibitors such as ACEIs and ARBs has become increasingly popular [22–25]. Through the continuous process of medical innovation, the use of different classes of antihypertensive drugs has changed significantly over time. Currently, hypertensive patients typically require two or more concomitant medications to achieve an optimal blood pressure level [22,23,26]. This is why we justify our combination of drug classes into one exposure, despite their different mechanisms related to cancer development, which resembles more to a naturalistic clinical setting. Furthermore, both prevalence and incidence of hypertension are steadily escalating with the aging population in Taiwan, making prescription of antihypertensive drugs increasingly common [22,23,25,27]. Our findings show that patients with hypertension are at higher risk of developing PC as compared to the general population. The main mechanism of hypertension in association with cancer lies in the cellular level, particularly at the interplay between cell proliferation and apoptosis [28]. In studies with hypertensive animal models, a defective growth stimulatory–inhibitory control can result due to abnormal proliferation induced by hypertension [29]. For PC, components of metabolic syndrome have been suggested to be the risk factors for its development [30]. Hyperinsulinemia has been specifically mentioned to be an effective predictor of PC development, and insulin level can be used as an indicator for its prognosis and tumor aggressiveness [31]. Moreover, plasma insulin levels can be modified through dietary and pharmaceutical behaviors [32]. We also find that among hypertensive patients, those with long term antihypertensive drug use are not at heightened risk of developing PC than non-users of these drugs. The results were robustly tested for different age groups. Therefore, our results suggest that hypertensive patients, regardless of age or comorbidity, should be prescribed antihypertensive drugs on a continuous basis, without the fear of increased PC risk, to avoid possible complications associated with hypertension. The lack of significant impact on the development of PC from using antihypertensive drugs (vs. non-users) is consistent with findings of a previous Finnish study which demonstrated only a slightly elevated risk for any antihypertensive drug users [6]. This also resonates with results found in the 15 clinical trials conducted by the ARB Trialists Collaboration which revealed no significant harm with long term ARB use [12]. Contrary to what we find, there are also existing studies that indicate an increased PC risk with the use of antihypertensive drugs [8,10,11, 33]. For example, Bhaskaran et al. [33] found a significant difference in the risk of developing PC among users of ARBs (HR = 1.10, 95% CI = 1.00–1.20). In addition, Sipahi et al. [11] conducted a meta-analysis and found a relative risk of 1.15 (95% CI = 0.99–1.34) for PC compared with the controls. To the best of our knowledge, most past research, including those mentioned above, has been limited by short-term follow-up periods, non-pragmatic settings (i.e., clinical trials), smaller sample sizes, and lack of clearly calculated cumulative drug doses, and most importantly, they have not estimated the risk of PC occurrence in hypertensive patients without use of any antihypertensive drug. Our study used a nationwide database and followed our subjects up to a period of nine years, using at least one year's worth of cumulative dosage within the follow-up period for the precise definition of hypertension patients undergoing antihypertensive drug therapy. Drug use was then tracked carefully on a daily basis. A notable strength of our study is that we included daily concentrations of antihypertensive drugs as time-dependent variables in the statistical models, allowing us to adjust for possible temporal and accumulative effects of antihypertensive drug use on the risk of PC. Finally, in order to allow for a more accurate estimation of the risk of PC development, we also considered death as a competing variable and allotted a time window for potential immortal time bias which is appropriate for our follow-up study of up to nine years.

6

P.-Y. Pai et al. / International Journal of Cardiology 193 (2015) 1–7

Table 5 Sensitivity analyses with different age groups using the Fine and Gray survival model. 50–59 years

Drug use Non-use 1–5 years N5 years Comorbidities Congestive heart failure Diabetes Hyperlipidemia CVD

60–69 years

sHR

p-Value

1.00 (reference) – 0.22 (0.02–3.30)

– 0.275

– 1.79 (0.06–50.10) 1.62 (0.08–34.51) –

– 0.732 0.756 –

sHR

≥80 years

70–79 years p-Value

sHR

p-Value

sHR

p-Value

1.00 (reference) 1.17 (0.11–12.40) 0.39 (0.05–3.45)

0.895 0.400

1.00 (reference) 0.80 (0.11–5.67) 1.24 (0.22–6.84)

0.827 0.806

1.00 (reference) 2.70 (0.16–44.68) 1.58 (0.10–24.79)

0.489 0.744

0.39 (0.04–3.54) 2.33 (0.32–16.75) 13.12 (0.62–276.10) 0.97 (0.08–12.25)

0.404 0.402 0.098 0.982

0.73 (0.18–2.86) 0.09 (0.01–0.83) 2.48 (0.53–11.66) 1.27 (0.29–5.60)

0.646 0.034 0.249 0.754

0.92 (0.09–9.51) 0.71 (0.04–14.39) – –

0.943 0.824 – –

sHR: subdistribution hazard ratio; CVD: cardiovascular disease.

4.1. Study limitations There are several limitations to our study. First, the severity of hypertension cannot be stratified using this claims-based data. In our hypertensive drug users (n = 79,615), the average number of antihypertensive drug classes jointly used is around 1.8, which is similar to previous reports. Prescription behavior of physicians, which is being monitored and kept under strict NHI regulations, has also been reported to be analogous with the western world guidelines [23,26]. Second, PC is a slowly-progressing cancer with a relatively long latent period. Many environmental and genetic factors can attribute to its development during this timeframe, which can also be used to explain for the racial and ethnic differences across countries and populations [34,35]. This can potentially lead to differences in outcomes observed. Third, all hypertension subjects included in the study were aged 50 and above, and were followed for up to nine years. It is important to note that unknown confounding factors may have existed during the follow-up period. Also, we did not have access to, and therefore did not consider personal information such as smoking, alcohol drinking, family history, body mass index, etc. Nevertheless, we included comorbidities (hyperlipidemia, congestive heart failure, diabetes, CVD) in the estimation model in an effort to make up for the missing information. Fourth, long term use of hypertensive agents can link to depression that exacerbates the general well-being of comorbid individuals (loss of appetite, quality sleep, etc.) and possibly contribute to the progression of PC [35]. The cut-off points for years of cumulative dose were arbitrary. From our results, however, cumulative intake of antihypertensive drugs did not increase one's risk of getting PC. Finally, we did not have clinical lab data to confirm our findings and to understand the physiological interactions of the drugs and their possible impacts on the risk of developing PC. Experimental data suggests the role of angiotensin II receptors in the development of cancer, but the direction of association remains unclear — both promotion and inhibition of angiogenesis have been detected for losartan in animal experiments [33,36,37]. Further studies are required to provide additional support for our findings.

5. Conclusions We conducted a nationwide cohort study with up to a nine-year follow up period. Our findings suggest that patients with hypertension are at significantly increased risk for the development of prostate cancer when compared to a matching general population. Among hypertensive patients, those with long term antihypertensive drug use are not at elevated risk of developing prostate cancer when compared to the non-users. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ijcard.2015.05.042.

Conflict of interest The authors have declared that no competing interests exist. Acknowledgements This work was made possible with the support of a research grant from the China Medical University, Taiwan (No. 10246T58). We are grateful for the data provided by the Collaboration Center of Health Information Application (CCHIA), Ministry of Health and Welfare, Taiwan. They had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. References [1] P.S. Vokonas, W.B. Kannel, L.A. Cupples, Epidemiology and risk of hypertension in the elderly — the Framingham-Study, J. Hypertens. 6 (1988) S3–S9. [2] J. Marin, Age-related-changes in vascular-responses — a review, Mech. Ageing Dev. 79 (1995) 71–114. [3] S. Lewington, R. Clarke, N. Qizilbash, R. Peto, R. Collins, et al., Age-specific relevance of usual blood pressure to vascular mortality: a meta-analysis of individual data for one million adults in 61 prospective studies, Lancet 360 (2002) 1903–1913. [4] 2013 ESH/ESC guidelines for the management of arterial hypertension, J. Hypertens. (2013) 1281–1357. [5] A.V. Chobanian, G.L. Bakris, H.R. Black, W.C. Cushman, L.A. Green, et al., The seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure: the JNC 7 report, JAMA 289 (2003) 2560–2572. [6] K.J. Kemppainen, T.L. Tammela, A. Auvinen, T.J. Murtola, The association between antihypertensive drug use and incidence of prostate cancer in Finland: a population-based case–control study, Cancer Causes Control 22 (2011) 1445–1452. [7] M.A. Poch, D. Mehedint, D.J. Green, R. Payne-Ondracek, E.T.H. Fontham, et al., The association between calcium channel blocker use and prostate cancer outcome, Prostate 73 (2013) 865–872. [8] M.A. Pfeffer, K. Swedberg, C.B. Granger, P. Held, J.J.V. McMurray, et al., Effects of candesartan on mortality and morbidity in patients with chronic heart failure: the CHARM-Overall programme, Lancet 362 (2003) 759–766. [9] F. Deshayes, C. Nahmias, Angiotensin receptors: a new role in cancer? Trends Endocrinol. Metab. 16 (2005) 293–299. [10] Cardiovascular and Renal Drugs Advisory Committee Briefing DocumentJuly 29, 2009. Available at: http://www.fda.gov/downloads/AdvisoryCommittees/ CommitteesMeetingMaterials/Drugs/CardiovascularandRenalDrugsAdvisory Committee/UCM173529.pdf (Accessed Aug 27 2013). [11] I. Sipahi, S.M. Debanne, D.Y. Rowland, D.I. Simon, J.C. Fang, Angiotensin-receptor blockade and risk of cancer: meta-analysis of randomised controlled trials, Lancet Oncol. 11 (2010) 627–636. [12] ARB Trialists Collaboration, Effects of telmisartan, irbesartan, valsartan, candesartan, and losartan on cancers in 15 trials enrolling 138,769 individuals, J. Hypertens. 29 (2011) 623–635. [13] Y.L. Chen, K.C. Cheng, S.W. Lai, I.J. Tsai, C.C. Lin, et al., Diabetes and risk of subsequent gastric cancer: a population-based cohort study in Taiwan, Gastric Cancer 16 (2013) 389–396. [14] Y.N. Lin, Y.J. Chang, Y.H. Chen, P.Y. Ko, C.L. Lin, et al., Nationwide population-based cohort study on the association of acute coronary syndrome in patients with malignancies, Support Care Cancer 22 (10) (2014) 2707–2713. [15] H.H. Grytli, M.W. Fagerland, S.D. Fosså, K.A. Taskén, L.L. Håheim, Use of β-blockers is associated with prostate cancer-specific survival in prostate cancer patients on androgen deprivation therapy, Prostate 73 (2013) 250–260. [16] C.R. Cardwell, Ú.C. Mc Menamin, B.M. Hicks, C. Hughes, M.M. Cantwell, L.J. Murray, Drugs affecting the renin–angiotensin system and survival from cancer: a population based study of breast, colorectal and prostate cancer patient cohorts, BMC Med. 12 (2014) 28.

P.-Y. Pai et al. / International Journal of Cardiology 193 (2015) 1–7 [17] S. Yusuf, Effect of enalapril on survival in patients with reduced left-ventricular ejection fractions and congestive-heart-failure, N. Engl. J. Med. 325 (1991) 293–302. [18] M.D. Flather, S. Yusuf, L. Kober, M. Pfeffer, A. Hall, et al., Long-term ACE-inhibitor therapy in patients with heart failure or left-ventricular dysfunction: a systematic overview of data from individual patients, Lancet 355 (2000) 1575–1581. [19] S. Yusuf, P. Sleight, J. Pogue, J. Bosch, R. Davies, et al., Effects of an angiotensinconverting-enzyme inhibitor, ramipril, on cardiovascular events in high-risk patients, N. Engl. J. Med. 342 (2000) 145–153. [20] H.C. Gerstein, S. Yusuf, J.F.E. Mann, B. Hoogwerf, B. Zinman, et al., Effects of ramipril on cardiovascular and microvascular outcomes in people with diabetes mellitus: results of the HOPE study and MICRO-HOPE substudy, Lancet 355 (2000) 253–259. [21] T. Kosaka, A. Miyajima, E. Kikuchi, S. Mikami, S. Shirotake, et al., Ets-1 and hypoxia inducible factor-1 a induced by angiotensin II: novel new molecular targets of angiotensin II type-1 receptor blocker (ARB) as an angiogenic inhibitor, J. Urol. 181 (2009) 261–262. [22] B. Dahlof, P.S. Sever, N.R. Poulter, H. Wedel, D.G. Beevers, et al., Prevention of cardiovascular events with an antihypertensive regimen of amlodipine adding perindopril as required versus atenolol adding bendroflumethiazide as required, in the Anglo-Scandinavian Cardiac Outcomes Trial-Blood Pressure Lowering Arm (ASCOTBPLA): a multicentre randomised controlled trial, Lancet 366 (2005) 895–906. [23] T.C. Su, C.H. Bai, H.Y. Chang, S.L. You, K.L. Chien, et al., Evidence for improved control of hypertension in Taiwan: 1993–2002, J. Hypertens. 26 (2008) 600–606. [24] C.C. Huang, W.L. Chan, Y.C. Chen, T.J. Chen, S.J. Lin, et al., Angiotensin II receptor blockers and risk of cancer in patients with systemic hypertension, Am. J. Cardiol. 107 (2011) 1028–1033. [25] W.H. Barker, J.P. Mullooly, K.L.P. Linton, Trends in hypertension prevalence, treatment, and control — in a well-defined older population, Hypertension 31 (1998) 552–559. [26] C.E. Chiang, T.D. Wang, Y.H. Li, T.H. Lin, K.L. Chien, et al., 2010 guidelines of the Taiwan Society of Cardiology for the Management of Hypertension, J. Formos. Med. Assoc. 109 (2010) 740–773.

7

[27] S.T. Chiou, T.L. Chen, H.Y. Wei, Epidemiology of hypertension in Taiwan: the results from a nationwide survey [abstract], J. Clin. Hypertens. (2013). [28] P. Hamet, Cancer and hypertension. An unresolved issue, Hypertension 28 (1996) 321–324. [29] V. Hadrava, J. Tremblay, P. Hamet, Abnormalities in growth characteristics of aortic smooth muscle cells in spontaneously hypertensive rats, Hypertension 13 (1989) 589–597. [30] J. Hammarsten, B. Högstedt, Clinical, haemodynamic, anthropometric, metabolic and insulin profile of men with high-stage and high-grade clinical prostate cancer, Blood Press. 13 (2004) 47–55. [31] J. Hammarsten, B. Högstedt, Hyperinsulinaemia: a prospective risk factor for lethal clinical prostate cancer, Eur. J. Cancer 41 (2005) 2887–2895. [32] E. Giovannucci, Nutrition, insulin, insulin-like growth factors and cancer, Horm. Metab. Res. 35 (2003) 694–704. [33] K. Bhaskaran, I. Douglas, S. Evans, T. van Staa, L. Smeeth, Angiotensin receptor blockers and risk of cancer: cohort study among people receiving antihypertensive drugs in UK General Practice Research Database, BMJ 344 (2012) e2697. [34] N. Breslow, C.W. Chan, G. Dhom, R.A.B. Drury, L.M. Franks, et al., Latent carcinoma of prostate at autopsy in 7 areas — collaborative study organized by InternationalAgency-for-Research-on-Cancer, Lyons, France, Int. J. Cancer 20 (1977) 680–688. [35] A.W. Hsing, L. Tsao, S.S. Devesa, International trends and patterns of prostate cancer incidence and mortality, Int. J. Cancer 85 (2000) 60–67. [36] D.H. Munzenmaier, A.S. Greene, Opposing actions of angiotensin II on microvascular growth and arterial blood pressure, Hypertension 27 (1996) 760–765. [37] T. Walther, A. Menrad, H.D. Orzechowski, G. Siemeister, M. Paul, et al., Differential regulation of in vivo angiogenesis by angiotensin II receptors, FASEB J. 17 (2003) 2061–2067.