pharmacoepidemiology and drug safety 2004; 13: 581–585Published online 15 December 2003 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/pds.910

ORIGINAL REPORT

Breast cancer incidence and use of antihypertensivemedication in women{

Antonio Gonzalez-Perez1*, Gunnar Ronquist2 and Luis Alberto Garcıa Rodrıguez1

1Centro Espanol de Investigacion Farmacoepidemiologica (CEIFE), Madrid, Spain2Department of Medical Sciences, Clinical Chemistry, University Hospital, Uppsala, Sweden

SUMMARY

Purpose Breast cancer is the most common cancer in women worldwide. We attempted to investigate the associationbetween the risk of breast cancer and use of captopril and other antihypertensive medication.Methods We performed a cohort study with a nested case-control analysis using the General Practitioner Research Data-base (GPRD) from the UK. We obtained adjusted estimates by fitting logistic regression models.Results The incidence rate of breast cancer in our cohort of women aged 30–79 years was 156 per 100 000 person-years.Overall, incidence of breast cancer among users of antihypertensive drugs was no different from the one among non-users(odds ratio (OR): 1.0; 95% confidence interval (CI): 0.9–1.1). Captopril was not associated with a reduced risk of breastcancer (OR: 0.9; 95%CI: 0.6–1.3).Conclusions We did not find any clear association between antihypertensive drugs and risk of breast cancer. Similarly,captopril was not associated with a reduced risk of breast cancer risk. Copyright # 2003 John Wiley & Sons, Ltd.

key words— captopril; antihypertensive drugs; breast cancer

INTRODUCTION

Breast cancer is by far the most common cancer inwomen world-wide, accounting for almost 20% ofall malignancies. The age standardised incidenceincreased by 40% from 1979 to 1992 in Englandand Wales.1 There is a growing interest in breast can-cer prevention and several strategies have been evalu-ated with some promising results.2

There has been some debate on the associationbetween hypertension, antihypertensive drugs andcancer. Some reports have associated angiotensin-I-converting enzyme (ACE) inhibitors with a reducedrisk of cancer as compared to other antihypertensivedrugs.3,4 ACE is responsible for the formation of thevasoconstrictor angiotensin II. This peptide stimulatesneovascularization,5 which is one of the requirements

for tumour growth.6 Also, angiotensin II can act as agrowth factor with stimulation of cell replication intissue culture7 and is associated with an increase in theexpression of genes that control cell growth.8 Captoprilis an ACE inhibitor that distinguishes itself from allother ACE inhibitors in the way it binds to the centralzinc atom of ACE. In a previous study, we found thatuse of captopril, but not other antihypertensive drugs,could be associated with a reduction on the occurrenceof prostate cancer.9

We performed a cohort study with nested case-control analysis to assess the association between therisk of breast cancer and use of captopril. We alsoexamined the association between the occurrence ofbreast cancer and use of other ACE inhibitors, calciumchannel antagonists and other antihypertensive drugs.

METHODS

We used data from the General Practice ResearchDatabase (GPRD). The GPRD contains computerisedinformation entered by general practitioners in the

Received 30 July 2003Revised 6 October 2003

Copyright # 2003 John Wiley & Sons, Ltd. Accepted 13 October 2003

* Correspondence to: A. Gonzalez-Perez, Centro Espanol deInvestigacion Farmacoepidemiologica, c/Almirante 28, 28 28004-Madrid, Spain. E-mail: agonzalez@ceife.es{No conflict of interest was declared.

UK.10 Data on over two million patients are systema-tically recorded and sent anonymously to the Medi-cines and Healthcare Products Regulatory Agency(MHRA) that collects and organises this informationin order to be used for research projects. The compu-terised information includes demographics, detailsfrom general practitioner’s visits, diagnoses from spe-cialist’s referrals and hospital admissions, results oflaboratory tests and a free text section. Prescriptionsissued by the general practitioner are directly gener-ated from the computer. Several studies with theGPRD have documented the validity and complete-ness of this database.11

Study population

We identified all females 30–79 years old betweenJanuary 1995 and December 2001. Women becamemembers of the study population on the first day ofthe study period when they met the criteria of at least1 year enrolment with the general practitioner and1 year since the first computerised prescription. Thatdate was their start date. Study members with a codefor cancer before start date were excluded. We alsoexcluded women 70 years and more at start datewho had a follow-up greater than 1 year and no datarecorded during their total follow-up time. Our finalstudy cohort comprised 734 899 women.

Follow-up

All study members were followed from start date untilthe earliest occurrence of one of the following end-points: recorded diagnosis of breast cancer, any cancerother than breast, age of 80 years, death or end ofstudy period (December 2001).

Case ascertainment and validation

We identified 4005 patients with a code of breast can-cer and manually reviewed their computerised patientprofiles. Information included demographic data andall clinical information with no personal identifiers.We excluded 297 women: the main reasons were thediagnosis was not confirmed (60%) and prevalentcases (31%). Based on the review of the computerisedinformation, 3708 patients were considered incidentcases of breast cancer. A previous study validating alarge number of cancer cases documented a high relia-bility of cancer diagnoses in the GPRD.12 Anotherstudy using the same database validated a small ran-dom sample of breast cancer cases and the diagnosiswas confirmed in all cases.13

Cohort and nested case-control analysis

Incidence rates of breast cancer were calculated invarious age groups. Then, a nested case-control analy-sis was performed to assess risk factors of breast can-cer. All cases of breast cancer (n¼ 3708) identified inthe study cohort were used in the nested case-controlanalysis and we considered their date of initial diagno-sis as index date. A date during the study period wasgenerated at random for every member of the studycohort. If the random date of a study member wasincluded in her eligible person-time, we used her ran-dom date as the index date and marked that woman asan eligible control. This selection mechanism allowsthat the likelihood of being selected as a control isproportional to the person-time at risk. The sameexclusion criteria were applied to controls as to cases.Twenty thousand controls were frequency-matched byage (interval of 1 year) and calendar year from the listof all eligible controls.

Estimates of odds ratio (OR), assumed to be validestimates of the rate ratio, and 95% confidence interval(CI) associated with use of ACE-inhibitors, calciumantagonists and other antihypertensive drugs comparedto non-users were computed using unconditionallogistic regression. We collected information on priorhistory of hypertension, diabetes, ischaemic heartdisease and cerebrovascular disease. We also ascer-tained patients with a previous breast lump and/orbreast biopsy recorded at least more than 1 year beforethe index date. Other risk factors like alcohol intake,body mass index (BMI) and hormone replacementtherapy (HRT) were also ascertained. All estimates ofOR were adjusted for age, calendar year, hypertension,BMI, alcohol intake, smoking status, HRT use andprior breast lump and/or breast biopsy.

Exposure definition

We studied the different groups of antihypertensivedrugs: diuretics, beta-adrenoceptor blocking drugs(beta-blockers), ACE inhibitors, calcium channelblockers (CCB) and alpha-adrenoceptor blockingdrugs (alpha-blockers). We defined three time win-dows of exposure for each class of drugs and the mostwidely prescribed individual antihypertensive drugssuch as captopril: current use, past use and no use.Current use was defined as use that lasted until theindex date or ended in the year prior to the index datebased on the supply of drug therapy as prescribed bythe general practitioner. Past use was use that endedmore than 1 year before the index date. Finally, thetime window of no use was defined as no use of each

582 a. gonzalez-perez ET AL.

Copyright # 2003 John Wiley & Sons, Ltd. Pharmacoepidemiology and Drug Safety, 2004; 13: 581–585

respective drug group at any time before the indexdate. Current users were further subdivided into cate-gories by treatment duration. An additional analysisusing 1-year time lag (advancing the index date by1 year in cases and controls) was performed.

RESULTS

The overall incidence rate of breast cancer in ourstudy population was 156 per 100 000 person-yearsamong women aged 30–79 years old. The incidenceincreased steeply until the age of 60 years and tendedto plateau thereafter (Figure 1).

Overall, 29% of cases and 28% of controls received aprescription for an antihypertensive drug within 1 yearof index date. Current users of antihypertensive drugspresented a OR of 1.0 (95%CI: 0.9–1.1) compared tonon-users. Hypertension was not associated withan increased risk of breast cancer (OR: 1.0; 95%CI:0.9–1.1).

Table 1 shows the results obtained for the differentclasses of antihypertensive drugs. None of them wereassociated with a significant reduction in breast cancerrisk. We did not find any clear trend towards increasedor decreased risk when we looked at treatmentduration. As a class, use of alpha-blockers with aduration greater than 1 year presented the smallestestimate of risk (OR: 0.6; 95%CI: 0.3–1.1). In Table 2,we report the estimates for the most widely prescribedindividual antihypertensive drugs. Use of captopril was

not associated with a reduced risk of breast cancer (OR:0.9; 95%CI: 0.6–1.3). Long-term use of captopril wasassociated with an OR of 0.8 (95%CI: 0.5–1.3). Westudied the effect of dose among long-term users.Those taking up to 50 mg per day presented an OR of0.7 (95%CI: 0.4–1.4) whereas those taking a dailydose higher than 50 mg had an OR of 0.9 (95%CI: 0.4–1.9). All other individual antihypertensive drugs hadestimates compatible with either a marginallyincreased or reduced risk of breast cancer.

When exposure to drugs occurring 1 year before theindex date was excluded (lag time analysis), the resultswere virtually the same (data not shown). Also, whenwe analysed the effect of antihypertensive drugsseparately by sex and among patients with and withouthypertension, the results were virtually identical (datanot shown).

DISCUSSION

In our study, we did not observe any clear associationbetween hypertension or the use of antihypertensivedrugs and the risk of breast cancer. In particular, cap-topril did not seem to be associated with a reducedbreast cancer risk, although a slight risk reductionwith long-term use can not be completely excluded.These results are in agreement with a previous studyusing the same database.13

It could be argued that early symptoms of breastcancer could have influenced in some way the pattern

Figure 1. Breast cancer incidence rate by age group

breast cancer and antihypertensive drugs 583

Copyright # 2003 John Wiley & Sons, Ltd. Pharmacoepidemiology and Drug Safety, 2004; 13: 581–585

of drug use in those patients (protopathic bias). Yet, it issomewhat difficult to predict the direction of this bias ifpresent. We explored the potential for this bias byperforming a subanalysis in which drug exposureoccurring 1 year before the index date was removed inboth cases and controls, and this analysis yieldedidentical results.

We found that incidence of breast cancer was 156 per100 000 person-years in our population of 30–79 year-old females. This estimate is well in line with otherreports from the UK. The incidence of breast cancer inEngland in 1998 based on cancer registries data was131 per 100 000 person-years.14

We recorded information regarding the most rele-vant risk factors for breast cancer including: age,alcohol use, smoking, BMI, HRT use and previousbreast abnormalities. However, there are some poten-tial confounders that we were not able to ascertainthrough our database including: age at menarche,

parity, family history, age at first child or germ linemutation. However, it seems highly unlikely that thislimitation of our study would have changed signifi-cantly the null results. Indeed, one would expect thedistribution of these potential confounders to be ratherevenly distributed between users of antihypertensiveand non-users.

In the early seventies, three studies published jointlysuggested that Rauwolfia alkaloids (used for treatinghypertension) were associated with an increased risk ofbreast cancer.15–17 Further studies failed to replicatethis finding that generated much controversy. The useof selected hospital controls in the first studies(excluding those subjects with cardiovascular relatedconditions as controls) was proposed as the main factorunderlying the spurious results.18 The design in thepresent study effectively avoids this so-called exclu-sion bias by applying identical eligibility criteria toboth cases and controls.

Table 1. Risk of breast cancer associated with use of antihypertensive drugsa

Cases n¼ 3708 (%) Controls n¼ 20 000 (%) Odds ratio (OR)b 95%CI

Diuretics (duration)No use 2699 (72.8) 14 797 (73.8) 1Current (<1 year) 206 (5.6) 925 (4.6) 1.2 1.0–1.4Current (1–3 years) 152 (4.1) 754 (3.8) 1.1 0.9–1.4Current (>3 years) 331 (8.9) 1698 (8.5) 1.1 0.9–1.2Past use 320 (6.6) 1826 (9.0) 1.0 0.8–1.1

Beta-blockers (duration)No use 2949 (79.5) 15 952 (79.8) 1Current (<1 year) 129 (3.5) 582 (2.9) 1.1 0.9–1.4Current (1–3 years) 75 (2.0) 422 (2.1) 1.0 0.8–1.3Current (>3 years) 231 (6.2) 1176 (5.9) 1.1 0.9–1.2Past use 324 (8.7) 1868 (9.3) 1.0 0.8–1.1

ACE inhibitors (duration)No use 3445 (92.9) 18 478 (92.4) 1Current (<1 year) 65 (1.8) 307 (1.5) 1.0 0.8–1.4Current (1–3 years) 60 (1.6) 364 (1.8) 0.8 0.6–1.1Current (>3 years) 82 (2.2) 474 (2.4) 0.9 0.7–1.2Past use 56 (1.5) 377 (1.9) 0.8 0.6–1.1

Calcium channel blockers (CCB) (duration)No use 3333 (89.9) 17 844 (89.2) 1Current (<1 year) 63 (1.7) 377 (1.9) 0.8 0.6–1.1Current (1–3 years) 66 (1.8) 419 (2.1) 0.8 0.6–1.1Current (>3 years) 128 (3.5) 703 (3.5) 1.0 0.8–1.2Past use 118 (3.2) 657 (3.3) 0.9 0.8–1.2

Alpha-blockers (duration)No use 3668 (98.9) 19 754 (98.8) 1Current (<1 year) 15 (0.4) 60 (0.3) 1.5 0.8–2.8Current (1–3 years) 6 (0.2) 54 (0.3) 0.6 0.3–1.5Current (>3 years) 6 (0.2) 56 (0.3) 0.5 0.2–1.3Past use 13 (0.4) 76 (0.4) 0.8 0.4–1.5

aEstimates are adjusted for age, calendar year, BMI, alcohol intake, smoking status, HRT use, prior breast lump and/or biopsy, hypertension

and all the variables in the table using logistic regression.bCurrent use was the one that lasted to index date or ended in the year prior to the index date. Current users were subdivided into categories by

treatment duration. Past use was the one that ended more than 1 year before the index date.

584 a. gonzalez-perez ET AL.

Copyright # 2003 John Wiley & Sons, Ltd. Pharmacoepidemiology and Drug Safety, 2004; 13: 581–585

In summary, we did not find any association betweenantihypertensive drugs and risk of breast cancer. In thesame line, captopril was not associated with a reducedrisk of breast cancer risk, although a slight riskreduction with long-term use cannot be safelyexcluded.

ACKNOWLEDGEMENTS

We thank Saga Johansson and Mari-Ann Wallanderfor their helpful comments on the manuscript. We

thank the staff at GPRD and the participating generalpractitioners for their collaboration. We also thank theBoston Collaborative Drug Surveillance Program(BCDSP) for providing access to the database.

REFERENCES

1. Quinn M, Allen E. Changes in incidence of and mortality frombreast cancer in England and Wales since introduction ofscreening: United Kingdom association of cancer registries.Br Med J 1995; 311(7017): 1391–1395.

2. Chlebowski RT. Reducing the risk of breast cancer. N Engl JMed 2000; 343(3): 191–198.

3. Pahor M, Guralnik JM, Salive ME, Corti M-C, Carbonin P,Havlik RJ. Do calcium channel blockers increase the risk ofcancer? Am J Hypertens 1996; 9: 695–699.

4. Jick H, Jick S, Derby LE, Vasilakis C, Meyers MW, Meier CR.Calcium-channel blockers and risk of cancer. Lancet 1997;349: 525–528.

5. Fernandes LA, Twickler J, Mead A. Neovascularizationproduced by angiotensin II. J Lab Clin Med 1985; 105:141–145.

6. Folkman J, Watson K, Ingber D, Hanahan D. Induction ofangiogenesis during the transition from hyperplasia to neopla-sia. Nature 1989; 339: 58–61.

7. Daemen MJAP, Lombardi DM, Bosman FT, Schwartz SM.Angiotensin II induces smooth muscle cell proliferation inthe normal and injured rat arterial wall. Circ Res 1991; 68:450–456.

8. Sadoshima J-I, Izumo S. Signal transduction pathways ofangiotensin II-induced c-fosgene expression in cardiac myo-cytes in vitro: roles of phospholipid-derived second messen-gers. Circ Res 1993; 73: 424–438.

9. Ronquist G, Garcıa Rodrıguez LA, Ruigomez A, et al. Asso-ciation between captopril, other antihypertensive drugs andrisk of prostate cancer. The Prostate (in press).

10. Garcıa Rodrıguez LA, Perez Gutthann S. Use of the U.K. gen-eral practice research database for pharmacoepidemiology. BrJ Clin Pharmacol 1998; 45: 419–426.

11. Jick H, Jick SS, Derby LE. Validation of information recordedon general practitioner based computerised data resource inthe United Kingdom. Br Med J 1991; 302: 766–768.

12. Jick H, Jick S, Derby LE, Vasilakis C, Myers MW, Meier CR.Calcium channel blockers and risk of cancer. Lancet 1997;349: 525–528.

13. Meier CR, Derby LE, Jick SS, Jick H. Angiotensin-convertingenzyme inhibitors, calcium channel blockers, and breast can-cer. Arch Int Med 2000; 160: 349–353.

14. Office for National Statistics. Cancer Statistics—Registrationsof Cancer Diagnosed in 1998, England. Series MB1 no.29. The Stationery Office: London, 2002.

15. Boston Collaborative Drug Surveillance Program. Reserpineand breast cancer. Lancet 1974; 2: 669–671.

16. Armstrong B, Stevens N, Doll R. Retrospective study of theassociation between use of rauwolfia derivatives and breastcancer in English women. Lancet 1974; 2: 672–675.

17. Heinonen OP. Shapiro S. Tuominen L. Turunen MI. Reserpineuse in relation to breast cancer. Lancet 1974; 2: 675–677.

18. Horwitz RI, Feinstein AR. Exclusion bias and the false rela-tionship of reserpine and breast cancer. Arch Intern Med1985; 145: 1873–1875.

KEY POINTS

� Incidence of breast cancer in our population of30–79 year-old females was 156 per 100 000person-years.

� Use of captopril and other antihypertensive drugswas not associated with breast cancer incidence.

Table 2. Risk of breast cancer associated with current use ofindividual antihypertensive drugsa

Cases(n¼ 3708)

Controls(n¼ 20 000)

ORb 95%CI

CaptoprilCurrent shorta 13 62 1.1 0.6–2.0Current long 20 131 0.8 0.5–1.3

EnalaprilCurrent short 36 185 0.9 0.6–1.4Current long 35 250 0.7 0.5–1.1

LisinoprilCurrent short 30 198 0.8 0.5–1.2Current long 27 149 1.1 0.7–1.6

AtenololCurrent short 174 824 1.0 0.8–1.3Current long 261 1356 1.1 0.9–1.3

PropanololCurrent short 72 322 1.1 0.8–1.6Current long 40 191 1.0 0.7–1.5

NifedipineCurrent short 36 216 0.9 0.6–1.3Current long 87 459 1.1 0.8–1.4

AmlodipineCurrent short 42 286 0.7 0.5–1.0Current long 28 160 1.0 0.7–1.5

DiltiazemShort duration 20 137 0.8 0.5–1.3Long duration 19 134 0.7 0.4–1.2

aCurrent short-term users were current users whose duration of treat-

ment was shorter than 2 years and current long-term users were cur-

rent users treated for a period longer than 2 years.bEstimates are adjusted for age, calendar year, BMI, alcohol intake,

smoking status, HRT use, prior breast lump and/or biopsy, hyperten-

sion and all the variables in the table using logistic regression.

breast cancer and antihypertensive drugs 585

Copyright # 2003 John Wiley & Sons, Ltd. Pharmacoepidemiology and Drug Safety, 2004; 13: 581–585