Cancer Causes & Control

, Volume 24, Issue 2, pp 365–371

Use of antihypertensive medications and breast cancer risk

Authors

    • Division of Public Health SciencesFred Hutchinson Cancer Research Center
    • Department of EpidemiologyUniversity of Washington
    • Seattle Children’s Hospital
  • Noel S. Weiss
    • Division of Public Health SciencesFred Hutchinson Cancer Research Center
    • Department of EpidemiologyUniversity of Washington
  • Weiva Sieh
    • Department of Health Research and PolicyStanford University School of Medicine
  • Annette L. Fitzpatrick
    • Department of EpidemiologyUniversity of Washington
  • Anne McTiernan
    • Division of Public Health SciencesFred Hutchinson Cancer Research Center
    • Department of EpidemiologyUniversity of Washington
  • Janet R. Daling
    • Division of Public Health SciencesFred Hutchinson Cancer Research Center
    • Department of EpidemiologyUniversity of Washington
  • Christopher I. Li
    • Division of Public Health SciencesFred Hutchinson Cancer Research Center
    • Department of EpidemiologyUniversity of Washington
Original paper

DOI: 10.1007/s10552-012-0122-8

Cite this article as:
Saltzman, B.S., Weiss, N.S., Sieh, W. et al. Cancer Causes Control (2013) 24: 365. doi:10.1007/s10552-012-0122-8

Abstract

Purpose

Use of specific antihypertensive medications (AHTs) has been hypothesized to increase breast cancer risk, but results across published studies are inconsistent.

Methods

We re-evaluated the relationship between AHT use and breast cancer risk in a prospective cohort of 3,201 women ≥65 years of age at recruitment now with more than double the length of follow-up (12 vs. 5 years) and substantially more breast cancer diagnoses (188 compared with 75 cases). We estimated the association between AHT use overall as well as use of specific formulations (based on data collected annually) and breast cancer risk using multivariate-adjusted Cox regression.

Results

Compared with women who reported no use of AHTs, women who had used calcium channel blockers (CCB) within the past two years had a 1.6-fold increased risk of breast cancer (95 % confidence interval (CI): 1.0–2.5), and in particular, recent users of immediate-release CCBs had a 2.4-fold increased risk (95 % CI: 1.3–4.5). Neither ever nor recent use of any other type of AHT was associated with breast cancer risk.

Conclusions

While the observed association between immediate-release CCBs and breast cancer risk is based on a small sample size and needs to be interpreted cautiously, this result is consistent with others in the literature. However, given declines in use of these preparations in favor of sustained-release CCBs, which was not related to risk, the potential clinical and public health impact of this association is limited. This study also adds to the evidence that other commonly used AHTs are not strongly related to breast cancer risk.

Keywords

Breast cancerEtiologyEpidemiologyRisk factorsAntihypertensive medications

Background

Studies evaluating the relationship between various antihypertensive therapies and breast cancer risk have yielded inconsistent results. The two classes of antihypertensive medications of greatest interest with respect to breast cancer have been calcium channel blockers and diuretics, given the former’s potential to inhibit apoptosis and the latter’s influence on insulin synthesis and metabolism. While some studies observed that use of calcium channel blockers [1, 2] or diuretics [35] were positively related to breast cancer risk, most did not [613]. The inconsistent findings of calcium channel blocker’s influence on breast cancer risk may reflect the changing prevalence of immediate and sustained-release formulations; only two of the previously published reports differentiated the two [5, 8]. Herein, we present a re-analysis of one of the studies that found that calcium channel blocker use was associated with a 2.57-fold (95 % CI 1.17–4.49) increased risk of breast cancer [1]. In this re-analysis of this prospective cohort of individuals, 65 years of age and older at enrollment, there is now more than twice the length of follow-up time, more than three times the number of breast cancer cases, and cancer outcomes data are now based on linkages with cancer registries rather than self-report.

Methods

We utilized data from the cardiovascular health study (CHS), a prospective cohort of adults 65 years of age and older that has been described in detail elsewhere [14, 15]. Briefly, CHS recruited subjects living in four communities: Forsyth County, North Carolina; Sacramento County, California; Washington County, Maryland; and Pittsburgh, Pennsylvania. CHS initially recruited 5,201 men and women into the study between the years of 1989 and 1990. An additional 687 African Americans were added to the cohort from 1992 to 1993. CHS excluded potential participants who were institutionalized, dependent on a wheelchair, in hospice care, or were receiving radiation or chemotherapy at the time of study enrollment. We also excluded 45 cohort members with a history of breast cancer at study entry, and 142 who were diagnosed with congestive heart failure (CHF) at study begin, since CHF is a contraindication for use of calcium channel blockers. Analyses were performed on 3,201 remaining women. All participants provided written informed consent.

Data collection was conducted annually and involved several components. The baseline medical exam included measurement of resting blood pressure. Information about medication use was abstracted directly from the prescription labels on subjects’ medication bottles at baseline and annual follow-up visits. Self-administered questionnaires inquiring about demographic characteristics, co-morbidities, previous cancer history, and tobacco and alcohol use were collected at baseline and annually. Participants were considered hypertensive if their baseline or follow-up measured systolic pressure was ≥140 or if their diastolic pressure was ≥90. An individual was also classified as hypertensive if she took antihypertensive medications and reported a history of hypertension through the self-administered study questionnaires.

Information regarding the diagnosis of breast cancer was obtained by linking data from CHS to cancer registries serving the regions where participants resided. These registries included the California Cancer Registry; Pennsylvania Cancer Registry; the Central Cancer Registry in North Carolina; and the Maryland Cancer Registry. CHS participants were also linked with the Johns Hopkins Training Center cancer registry to obtain cancer diagnoses in Maryland prior to 1992, at which time the Maryland registry began collecting complete data. We obtained data on cancer diagnoses through December 2001.

Mortality and vital status of CHS participants were ascertained at annual examinations and 6-month telephone interviews. If the participant was unavailable, a proxy was interviewed. Deaths were also identified from obituaries. Follow-up for vital status was 100 % complete. To correspond with surveillance of cancer diagnoses available through our cancer registry linkages women who were breast cancer free and still alive were censored at the end of 2001.

We assessed the risk of breast cancer associated with antihypertensive medication use through Cox proportional hazard models using STATA (version 9.0 for windows; Stata Corp, College Station, Texas). In all analyses, antihypertensive medication use was treated as a time-varying exposure. Women contributed person time until they were either diagnosed with breast cancer or were censored (at death or end of 2001); “unexposed” time was contributed before initiation of any antihypertensive medication, and “exposed” person time began when a woman reported use of antihypertensive medication and continued irrespective of antihypertensive medication discontinuation. To assess the association with recent use, analyses were carried out in which person time and events were censored 2 years after the follow-up visit where data about antihypertensive medication use or cessation of use were recorded, and women were considered “exposed” for these analyses only if they reported use within the previous 2 years and unexposed if they reported no use within the previous 2 years and had no prior reported use. All analyses were repeated among hypertensive women who reported use of any antihypertensive medication in order to assess potential confounding by indication or detection bias; women reporting use of specific types of antihypertensive medications were compared with users of other antihypertensive medications. A woman may have been included in more than one exposure group if she reported use of more than one type of antihypertensive medication.

We assessed confounding by testing whether inclusion of potential confounders in the regression model changed the risk estimate by more than 10 %. Age, average weekly alcohol consumption, age at menopause, and waist–hip ratio confounded the relationship between antihypertensive medication use and breast cancer risk, and so we retained these terms in all models as categorical variables as defined in Table 1; race, smoking history, education and income, body mass index, use of estrogen replacement therapy, and diabetes were not found to have an appreciable effect on the association between AHT use and breast cancer risk and were therefore not retained in the final analytic models. We compared the risks of any breast cancer associated with AHT use, as well as only invasive breast cancers; since the pattern of results was similar, we chose to retain in situ cancers in our analyses in order to maximize the precision of our risk estimates. We performed likelihood-ratio tests to evaluate the presence of an interaction on a multiplicative scale between antihypertensive medication use and the breast cancer risk factors listed in Table 1. We did not find any statistically significant effect modifiers (all p values for interaction were >0.05).
Table 1

Distribution of demographic and anthropometric characteristics at baseline among women according to their use of antihypertensive medication (AHT) at baseline: Cardiovascular Health Study

Variable

No use of AHT medication

Use of any AHT medication

Use of calcium channel blocker

Use of diuretic

 

[n = 1,675]

n (%)

[n = 1,526]

n (%)

[n = 392]

n (%)

[n = 993]

n (%)

Age at baseline (years)a

 65–74.9

1220 (72.8)

976 (64.0)

253 (64.5)

617 (62.1)

 75–84.9

411 (24.5)

491 (32.2)

124 (31.6)

338 (34.0)

 85–94.9

42 (2.5)

58 (3.8)

15 (3.8)

37 (3.7)

 95+ 

2 (0.1)

1 (0.1)

0 (0.0)

1 (0.1)

Raceb

    

 White

1488 (88.8)

1160 (76.0)

257 (65.6)

759 (76.4)

 Black

177 (10.6)

357 (23.4)

133 (33.9)

226 (22.8)

 Other

10 (0.6)

9 (0.6)

2 (0.5)

8 (0.8)

Education attainedb

 Less than high school

414 (24.8)

502 (33.0)

140 (36.0)

335 (33.9)

 High school or higher

1258 (75.2)

1018 (67.0)

249 (64.0)

654 (66.1)

 Missing

3

6

3

4

Earns over $25 k/yearb

 No

955 (61.9)

1004 (71.4)

260 (72.0)

677 (74.2)

 Yes

587 (38.1)

403 (28.6)

101 (28.0)

236 (25.8)

 Missing

133

119

31

80

Cigarette useb

 Never smoked

943 (56.3)

877 (57.5)

217 (55.5)

581 (58.6)

 Former smoker

489 (29.2)

488 (32.0)

133 (34.0)

309 (31.1)

 Current smoker

242 (14.5)

159 (10.4)

41 (10.5)

102 (10.3)

 Missing

1

2

1

1

Alcohol use (average number of drinks per week)b

 None

847 (50.9)

931 (61.1)

247 (63.0)

631 (63.7)

 Fewer than 2

529 (31.8)

390 (25.6)

95 (24.2)

232 (23.4)

 Two or more

289 (17.4)

202 (13.3)

50 (12.8)

128 (12.9)

 Missing

10

3

0

2

Body mass index (kg/meter2)b

 Normal weight

792 (47.5)

505 (33.2)

139 (35.7)

293 (29.6)

 Overweight

601 (36.0)

571 (37.5)

141 (36.2)

365 (36.9)

 Obese

276 (16.5)

445 (29.3)

109 (28.0)

332 (33.5)

 Missing

6

5

3

3

Waist–hip ratiob

 0.61–82

476 (28.6)

328 (21.6)

78 (20.1)

215 (21.8)

 0.83–89

412 (24.8)

375 (24.7)

87 (22.4)

258 (26.2)

 0.90–95

389 (23.4)

400 (26.4)

107 (27.6)

242 (24.5)

 0.96+

385 (23.2)

413 (27.2)

116 (29.9)

271 (27.5)

 Missing

13

10

4

7

Age at menopauseb

 <49

824 (52.6)

780 (56.1)

215 (60.1)

488 (54.0)

 >50

743 (47.4)

610 (43.9)

143 (39.9)

416 (46.0)

 Missing

108

136

34

89

Parityb

 None

273 (16.6)

273 (18.5)

66 (17.5)

182 (18.9)

 One

280 (17.0)

267 (18.1)

66 (17.5)

185 (19.2)

 Two or more

1095 (66.4)

938 (63.5)

246 (65.1)

595 (61.9)

 Missing

27

48

14

31

History of hysterectomyb

 No

1055 (63.9)

847 (57.6)

201 (53.5)

559 (58.3)

 Yes

595 (36.1)

624 (42.4)

175 (46.5)

400 (41.7)

 Missing

25

55

16

34

One or both ovaries removedb

 No

1184 (75.3)

976 (70.8)

239 (68.1)

642 (71.9)

 Yes

388 (24.7)

402 (29.2)

112 (31.9)

251 (28.1)

 Missing

103

148

41

100

Use of estrogena

 No

1449 (86.5)

1356 (88.9)

352 (89.8)

886 (89.2)

 Yes

226 (13.5)

170 (11.1)

40 (10.2)

107 (10.8)

Use of progestina

 No

1627 (97.1)

1497 (98.1)

388 (99.0)

981 (98.8)

 Yes

48 (2.9)

29 (1.9)

4 (1.0)

12 (1.2)

Hypertensiona

 Normotensive

1036 (66.7)

159 (12.8)

57 (20.7)

79 (9.6)

 Borderline hypertensivec

353 (22.7)

46 (3.7)

14 (5.1)

27 (3.3)

 Hypertensived

165 (10.6)

1038 (83.5)

204 (74.2)

714 (87.1)

 Missing

121

283

117

173

Diabetesa

 No

1513 (91.2)

1218 (81.1)

306 (78.9)

790 (80.7)

 Yes

146 (8.8)

284 (18.9)

82 (21.1)

189 (19.3)

 Missing

16

24

4

14

aTime-varying characteristic: these characteristics assessed at each follow-up visit and are time varying throughout the analysis; age assessed at baseline, included in analysis as time varying

bThese characteristics assessed at study entry, values fixed throughout the analysis

cSeated blood pressure systolic = 140–159 mmHG, OR Seated blood pressure diastolic = 90–94 mmHG

dSeated blood pressure average systolic > = 160 mmHG, OR seated blood pressure diastolic > = 95 mmHG, OR self-reported history of hypertension AND participant reported use of antihypertensive medication

Results

Table 1 shows the distribution of baseline demographic and anthropometric characteristics stratified by baseline use of antihypertensive medications. Higher proportions of women who used antihypertensive medications were African American, had less than a high school education, and were less likely to earn more than $25,000 annually. Slightly higher proportions of non-users were smokers and reported consumption of two alcoholic drinks per week than AHT-users, but antihypertensive medication users had higher waist–hip ratios, and were much more likely to be hypertensive, and to have diabetes. Compared with calcium channel blocker users, diuretic users were somewhat more likely to be obese and less likely to have diabetes.

Ever use of antihypertensive medications (regardless of timing) was not related to breast cancer risk, nor was use of individual types of antihypertensive medications (Table 2). Additionally, no associations between antihypertensive medication use and risk of any particular breast cancer subtype (defined by estrogen receptor status, histology, or stage) were observed (data not shown). While recent antihypertensive medication use (use within the past 2 years) was also not related to breast cancer risk overall, recent calcium channel blocker use was associated with a 60 % increased risk of breast cancer (95 % CI 1.0–2.5), and in particular, recent use of immediate-release calcium channel blockers was associated with a 2.4-fold increase in risk (95 % CI 1.3–4.5). To account for potential confounding by indication or detection bias, we restricted the analysis to those women who had been diagnosed with hypertension and had ever used any class of antihypertensive medication. While ever use of any particular AHT subclass was not related to breast cancer risk in this sub-analysis (data not shown), there was a suggestion that recent use of immediate-release calcium channel blockers was associated with an 80 % increase in risk (95 % CI 0.9–3.5).
Table 2

Breast cancer risk in users of particular types of antihypertensive medication (AHT)

 

Ever usea

Use in past 2 yearsb

Cases

Person

Years

Incidence

Ratec

HR (95 % CI)d

Cases

Person

Years

Incidence

Ratec

 

HR (95 % CI)d

No AHT usee

69

14,081

4.9

1.00 (ref)

53

11,404

4.6

1.00 (ref)

Any AHT use

119

22,584

5.3

1.1 (0.8–1.5)

90

16,057

5.6

1.2 (0.9–1.8)

CCBs

 All types

55

9,916

5.5

1.1 (0.7–1.6)

40

5,740

7.0

1.6 (1.0–2.5)*

 Immediate release

24

3,995

6.0

1.2 (0.7–2.0)

16

1,453

11.0

2.4 (1.3–4.5)

 Sustained release

43

7,987

5.4

1.1 (0.7–1.6)

28

4,506

6.2

1.4 (0.8–2.3)

Diuretics

 Diuretics- all types

83

15,273

5.4

1.2 (0.8–1.6)

48

9,276

5.2

1.2 (0.8–1.9)

 HCTZ

58

11,529

5.0

1.0 (0.7–1.5)

33

6,103

5.4

1.2 (0.7–2.0)

 KSPR

36

6,031

6.0

1.2 (0.7–1.8)

12

2,977

7.1

1.2 (0.6–2.4)

 LOOP

22

4,476

4.9

1.1 (0.7–1.9)

21

2,602

4.6

1.5 (0.8–2.7)

Beta-blockers

47

7,601

6.2

1.1 (0.7–1.7)

31

4,193

7.4

1.5 (0.9–2.4)

ACE inhibitors

31

6,415

4.8

1.0 (0.7–1.6)

19

3,411

5.6

1.4 (0.8–2.4)

HR hazard ratio, CCB calcium channel blocker, HCTZ hydrochlorothiazide, KSPR potassium sparing diuretics, LOOP loop diuretics

p < 0.05

aDuring the course of the CHS study

bIncludes only person time and events following informative follow-up visit resulting in a classification of either exposed or unexposed to any and particular AHT types

cPer 1,000 person years

dEstimates adjusted for age, alcohol use, income, age at menopause, waist–hip ratio; all compared against common reference group of never AHT users

ePerson years and events accrued during course of study prior to starting any AHT-common ref group for all exposure categories

Discussion

Before interpreting the results of this study, it is important to acknowledge its limitations. We had data on antihypertensive medication use only during, but not before, cohort follow-up. This prohibited us from assessing the potential influence of duration of use on breast cancer risk. Though the frequency of antihypertensive medication use was high in this older population, due to our study’s relatively small size we had limited statistical power to identify more modest elevations in risk in the range of 20–50 % and to conduct sub-analyses. We were unable to evaluate the potential effect of AHT dose on breast cancer risk. Finally, participants who moved to a different state during the study period would not have been captured by the cancer registries serving the four CHS communities, and as a result some cancer diagnoses may have been missed. However, considering the age of this elderly cohort, the number of missed cancers because of migration is likely to have been small and non-differential with respect to medication use, and therefore would be expected to have a negligible effect on the results.

In 1996, a prospective cohort study reported that baseline use of calcium channel blockers was associated with a 65 % increase in breast cancer risk (95 % CI 0.49–5.55) [2]. In an early report utilizing CHS data (the same cohort evaluated in the current analysis), researchers observed that ever use of calcium channel blockers during cohort follow-up was associated with a 2.57-fold (95 % CI 1.47–4.49) increased risk of breast cancer with some suggestion that this association was stronger for use of immediate-release calcium channel blockers [1]. However, subsequent cohort and case–control studies have not duplicated these early strong associations, with risk ratios ranging from 0.8 to 1.45 [3, 513]. Here, we observed that while there was some suggestion that calcium channel blockers increase breast cancer risk, this association appears to be confined to recent use of immediate-release calcium channel blocker formulations. The fact that studies published subsequent to the initial CHS report did not replicate earlier positive findings may have been due to the result of shifting patterns of CCB use, with reduced use of immediate-release calcium channel blockers as a result of the availability of sustained-release formulations and concerns about the safety of immediate-release formulations [16, 17]. The CHS data illustrate this point: during years 1–3 of CHS follow-up, 44 % of reported calcium channel blocker use was with an immediate-release formulation; during years 6–10 of follow-up, this proportion was only 10 %. Only one prior study [5] has specifically evaluated the relationship between immediate-release calcium channel blocker use and breast cancer risk, and it too observed a positive association (odds ratio: 1.4, 95 % 1.0–2.1). While our results should be interpreted with caution since our risk estimate was based on only 16 cases who were recent immediate-release calcium channel blocker users, it is noteworthy that an elevation in risk persisted when we restricted the analysis to those women who had ever used antihypertensive medication, suggesting that the finding is not the result of confounding by indication or detection bias resulting from increased surveillance among women with treated hypertension. If immediate-release calcium channel blocker use is indeed related to breast cancer risk, this may explain in part the inconsistencies in the literature related to calcium channel blocker use given the higher proportion of immediate-release use among earlier studies and the likely smaller proportions of immediate-release users in the more recent studies. These findings differentiating the risk between users of immediate- and sustained-release formulations of calcium channel blockers may help hypertensive women and their providers who are concerned about breast cancer risk feel more confident about sustained-release CCBs as one of their AHT options.

The biological mechanism underlying a possible association between calcium channel blocker use, and immediate-release calcium channel blocker use in particular, and breast cancer risk is not understood. It has been hypothesized that CCBs may inhibit apoptosis through a rise in intracellular calcium levels [1, 18, 19]. However, subsequent evaluations of this hypothesis have concluded that this is not supported by the scientific literature [20]. In fact, the antineoplastic effects of the calcium channel blockers diltiazem, amlodipine, and verapamil are being evaluated for use in conjunction with traditional chemotherapeutic agents in the treatment of drug-resistant breast cancers [2123]. However, calcium channel blockers have a broad spectrum of physiologic effects, many of which are still unidentified, that could ultimately affect breast cancer risk. Compared with sustained-release formulations, immediate-release formulations result in greater fluctuations of circulating CCBs. Consequently, circulating levels of CCBs in users of immediate-release formulations may surpass some threshold required for tumor promotion and result in an increased risk of breast cancer associated with their use.

In the current analysis, we found no clear evidence to support the hypothesis that use of either diuretics overall or any subclass of diuretics was related to breast cancer risk. Findings from the five previous studies evaluating the relationship between diuretic use and breast cancer risk have been mixed, although four observed that diuretic use was associated with breast cancer risk to at least some degree (odds ratios ranging from 1.2 to 1.79) [1, 35, 24]. While our the findings we report here regarding the potential association between diuretic use and breast cancer risk are not inconsistent with a risk of this magnitude, they may also be due to chance. The fifth study evaluating the association, using prospectively collected data from a cohort of women in Denmark, observed no association between diuretic use and breast cancer risk (odds ratio: 0.95, 95 % CI 0.8–1.12) [13].

In conclusion, findings from the current study do not provide evidence that there is an association between diuretic, ACE-inhibitor, or beta-blocker use and breast cancer risk among post-menopausal women. While there may be an association between use of immediate-release calcium channel blockers and breast cancer risk, the public health impact of this association is likely to be small given the decreasing prevalence of immediate-release calcium channel blocker use.

Acknowledgments

The Cardiovascular Health Study (http://www.chs-nhlbi.org) provided all of the subjects and specimens for this project, and Dr. Mary Lou Biggs was particularly supportive throughout. Cancer incidence data were supplied by the California Cancer Registry, North Carolina Central Cancer Registry, Pennsylvania Bureau of Health Statistics, Maryland Cancer Registry, and Johns Hopkins Training Center for Public Health Research; these institutions specifically disclaim responsibility for any analyses, interpretations, or conclusions expressed herein. This research was supported in part by a pilot grant from the Avon Foundation for women.

Conflict of interest

None to disclose.

Copyright information

© Springer Science+Business Media Dordrecht 2012