Cardiovascular Medication Use and Risk for Colorectal Cancer

Abstract

Objective

To evaluate the association between lipid-lowering agents, antihypertensive medications, and colorectal cancer risk. We hypothesized a reduction in colorectal cancer risk with 3-hydroxy-3-methylglutaryl coA reductase inhibitors (statins) and angiotensin-converting enzyme inhibitors.

Methods

We conducted a case-control study at Group Health Cooperative, an integrated delivery system in Washington State. Incident colorectal cancer cases diagnosed between January 1, 2000, and December 31, 2003, were identified from the western Washington Surveillance, Epidemiology, and End Results cancer registry. Controls were matched by age, sex, and duration of enrollment. Data on medication use and potential confounders were obtained from health plan records. We estimated odds ratios and 95% confidence intervals (95% CI) using multivariate conditional logistic regression.

Results

Risk for colorectal cancer was not associated with use of statins (odds ratio, 1.02; 95% CI, 0.65–1.59), other lipid-lowering agents (odds ratio, 1.31; 95% CI, 0.70–2.47), angiotensin-converting enzyme inhibitors (odds ratio, 0.98; 95% CI, 0.67–1.43), calcium channel blockers (odds ratio, 1.06; 95% CI, 0.72–1.55), or diuretics (odds ratio, 1.00; 95% CI, 0.70–1.44). Risk did not differ by duration of medication use, including long-term use.

Conclusions

Risk for colorectal cancer was not reduced by use of statins or angiotensin-converting enzyme inhibitors. Other lipid-lowering and antihypertensive medications were also not associated with colorectal cancer risk.

Introduction

Colorectal cancer is the fourth most common cancer in the United States overall and the third leading cause of cancer mortality in men and women (1). Despite increased understanding of colorectal cancer risk factors and the availability of effective screening tests, a considerable increase in cases is expected in the aging populations of the world (2). Acceptance of screening (that is, endoscopic and fecal-based tests) is less than optimal, and (3, 4) as a result, there is interest in prevention, including chemoprevention (5, 6). Studies evaluating medication use and colorectal cancer risk have mainly focused on the chemopreventive effects of aspirin and other nonsteroidal anti-inflammatory drugs (NSAID; ref. 7) and hormone replacement therapy (8, 9). Other therapeutic drug classes including angiotensin-converting enzyme inhibitors (10–12) and 3-hydroxy-3-methylglutaryl coA reductase inhibitors (statins; refs. 13, 14) may also have chemopreventive effects, but few studies have evaluated the association of angiotensin-converting enzyme inhibitors and the risk for colorectal cancer (15–18), and the results of studies on statins and colorectal cancer risk are inconsistent (19–31).

We conducted a case-control study on the use of angiotensin-converting enzyme inhibitors and statins in relation to risk for colorectal cancer. We also evaluated other lipid-lowering and antihypertensive medications in relation to colorectal cancer risk to evaluate whether any observed associations with statins and angiotensin-converting enzyme inhibitors is confounded by indication for the medication (that is, hypertension and high cholesterol) and because limited evidence suggests a possible increase in the risk for various cancers with the use of calcium channel blockers (32–38) and diuretics (34, 39).

Materials and Methods

Study Setting and Population

Cases were eligible for this study if they were diagnosed with colorectal cancer between January 1, 2000, and December 31, 2003, and enrolled in Group Health Cooperative (Group Health), a nonprofit integrated delivery system that provides comprehensive health care to ~550,000 individuals throughout western Washington State. Cancer cases were identified from the western Washington Surveillance, Epidemiology, and End Results cancer registry (40). We restricted analyses to patients 40 years of age and older at diagnosis.

Controls were randomly sampled from the Group Health enrollment file and were matched 1:1 to cases on age (month, year), gender, and duration of Group Health enrollment. Controls were assigned a reference date (month, year) to correspond to the case diagnosis date. Subjects (cases or controls) were ineligible for study if they were enrolled in Group Health for <2 years or had a previous diagnosis of colorectal cancer. The institutional review board of Group Health approved the study.

Data Collection

Medical records, including electronic pharmacy records, were abstracted for medication use and potential covariates, including height, weight, race, use of NSAIDs including aspirin and cyclooxygenase 2 inhibitors, diagnoses of diabetes and hyperlipidemia, and smoking status. Since 1976, the Group Health pharmacy database has included a record for all medications dispensed to Group Health enrollees. Each record includes a patient identifier, drug name, strength, date dispensed, quantity dispensed, instructions for use, and form. The pharmacy data of Group Health are considered a complete source of medication use, and it is estimated that Group Health enrollees obtain 97% of their prescription medications at Group Health pharmacies (41, 42). Data collection on exposures and covariates was restricted to the 10 years before the year preceding diagnosis (cases) or reference date (controls).

Medication Use

Use of the medication classes of interest—statins, other lipid-lowering agents, angiotensin-converting enzyme inhibitors, calcium channel blockers, and diuretics—was defined as two or more prescriptions for any drugs belonging to the medication class of interest during a 6-month period. Other lipid-lowering agents included bile acid sequestrants, fibrates, and niacin. Among users, we calculated the cumulative duration of use for each drug class. For each dispensing, we estimated the date when the pills should have run out (runout date) based on quantity dispensed and instructions for use. A new runout date was set with each successive dispensing. Successive dispensing with a <60-day lag period between the runout date of one dispensing and fill date of the subsequent dispensing were considered continuous use. Periods of continuous use were summed for total duration of use.

Statistical Analyses

Analyses were done using Stata 9.0 (Stata Corporation). Using conditional logistic regression, we examined the association between the therapeutic medication classes of interest and colorectal cancer. Adjusted odds ratio estimates and respective 95% confidence intervals (95% CI) were computed for statins, other lipid-lowering agents, calcium channel blockers, diuretics, and angiotensin-converting enzyme inhibitors. Multivariate models included the following set of potential confounders that were identified a priori: age (continuous), body mass index (continuous), diabetes (yes or no), smoking status (ever or never), use of hormone replacement therapy among women (ever or never), and use of aspirin or other NSAIDs (yes or no). Risk for colorectal cancer was estimated in relation to ever use of medication and duration (categorized as <2 and 2+ years). The duration of use analysis was restricted to subjects enrolled in Group Health for 2+ years. In addition, we examined the data separately for cancers of the colon and rectum in exploratory analyses.

Results

A total of 665 patients who were diagnosed with colorectal cancer and met inclusion criteria during the study period were matched with 665 controls. Selected characteristics of cases and controls are given in Table 1. Study participants were 70 years of age on average and enrolled in Group Health for ~19 years. Of the 665 colorectal cancer cases, 76% were colon cancer, 23% were rectal cancer, and 1% were coded as both; 3% were in situ; 28%, stage I; 30%, stage II; 24%, stage III; 13%, stage IV; and 2%, unknown stage. Colorectal cancer cases were more likely to have had a previous cancer diagnosis, a diagnosis of diabetes, and a higher body mass index than controls. Most subjects were Caucasian, reflecting the Puget Sound geographic area that Group Health serves.

Table 1.

Demographic characteristics of the study population by colorectal cancer status

Characteristics at index date	Cases	Controls
	n = 665	n = 665
Females	342 (51.4)	342 (51.4)
Mean age, y (SD)	69.9 (12.3)	70.0 (12.3)
Mean enrollment, y (SD)	18.59 (11.45)	18.87 (11.44)
Race
White	594 (89.5%)	508 (90.4%)
African American	29 (4.4%)	23 (4.1%)
Asian or Pacific Islander	41 (6.2%)	31 (5.5%)
Other or unknown	1	103
Mean BMI in kg/m2 (SD)	28.5 (7.0)	27.4 (5.9)
Missing	192	182
Diabetes*	161 (24.2%)	118 (17.7%)
Hypercholesterolemia†	281 (42.3%)	291 (43.8%)
Previous cancer diagnosis	167 (25.1%)	121 (18.2%)
Ever smoker	387 (58.2%)	327 (49.2%)
Ever use of NSAID or aspirin	402 (60.5%)	426 (64.1%)
Ever use of hormone	130 (38.0)	169 (49.4)
therapy among women
Menopausal status‡
Pre	5 (1.5%)	6 (1.8%)
Peri or post	337 (98.5%)	336 (98.3%)
Ever use of statins	99 (14.9)	97 (14.6)
Ever use of other	41 (6.2)	44 (6.6)
lipid-lowering drugs
Ever use of	190 (28.6)	174 (26.2)
angiotensin-converting
enzyme inhibitors
Ever use of calcium channel	138 (20.8)	126 (19.0)
blockers
Ever use of diuretics	208 (31.3)	195 (29.3)

Abbreviation: BMI, body mass index.

^*Diabetes was defined as one or more of the following: 2+ dispensings for a medication used to treat diabetes, fasting glucose of >125 mg/dL confirmed by a second out-of-range test within 1 y, random glucose of >200 mg/dL confirmed by a second test within 1 y, or hospital discharge of diabetes or 2 outpatient diagnosis of diabetes (International Classification of Diseases 9 = 250, 250.0, 250.1, 250.2, 250.3, 250.4, 250.5, 250.6, 250.7, 250.8, 250.9).

^†Hypercholesterolemia was defined as low density lipoprotein of ≥160 mg/dL or total cholesterol of ≥240 mg/dL.

^‡Peri- or postmenopausal status was defined as women with hormone replacement therapy (HT) use, bilateral oophorectomy, natural menopause, or who were ≥55 y old.

Statins

Approximately 14% of cases and controls used statins during the 10 years before index date and use was not associated with risk for colorectal cancer (Table 2). Risk seemed to increase with duration of use; however, this finding was not statistically significant and seemed to be limited to rectal cancer.

Table 2.

Association between cardiovascular medication use and colorectal cancer risk

Medication use	Cases (n)	Controls (n)	Colorectal	Colon*	Rectal†
			Adjusted odds ratio (95% CI)†
Statins
Non-user	297	303	1.0 (ref)	1.0 (ref)	1.0 (ref)
Ever user	60	54	1.02 (0.65–1.59)	0.91 (0.55–1.50)	1.47 (0.50–4.29)
Duration among enrolled 2+ y‡
Non-users	293	300	1.0 (ref)	1.0 (ref)	1.0 (ref)
<2 y	20	22	0.80 (0.40–1.59)	0.78 (0.37–1.65)	1.04 (0.36–2.98)
2+ y	40	31	1.22 (0.70–2.12)	1.06 (0.57–1.98)	1.69 (0.77–3.70)
Other lipid-lowering drugs
Non-user	329	334	1.0 (ref)	1.0 (ref)	1.0 (ref)
Ever user	28	32	1.31 (0.70–2.47)	1.43 (0.73–2.81)	Cannot estimate
Duration among enrolled 2+ y‡
Non-users	325	330	1.0 (ref)	1.0 (ref)	1.0 (ref)
<2 y	17	16	1.05 (0.49–2.26)	1.21 (0.54–2.68)	Cannot estimate
2+ y	11	7	1.96 (0.72–5.35)	2.00 (0.67–5.98)
Angiotensin-converting enzyme inhibitors
Non-user	244	255	1.0 (ref)	1.0 (ref)	1.0 (ref)
Ever user	113	102	0.98 (0.67–1.43)	1.00 (0.67–1.53)	0.73 (0.29–1.79)
Duration among enrolled 2+ y‡
Non-users	240	252	1.0 (ref)	1.0 (ref)	1.0 (ref)
<2 y	53	45	1.04 (0.63–1.71)	1.05 (0.59–1.86)	0.83 (0.31–2.26)
2+ y	60	56	0.96 (0.61–1.53)	1.02 (0.62–1.68)	0.54 (0.15–1.97)
Calcium channel blockers
Non-user	270	279	1.0 (ref)	1.0 (ref)	1.0 (ref)
Ever user	87	78	1.06 (0.72–1.55)	1.08 (0.71–1.66)	0.63 (0.22–1.76)
Duration among enrolled 2+ y‡
Non-users	266	275	1.0 (ref)	1.0 (ref)	1.0 (ref)
<2 y	33	26	1.23 (0.68–2.23)	1.06 (0.55–2.02)	2.42 (0.44–13.5)
2+ y	54	52	0.98 (0.63–1.52)	1.09 (0.67–1.77)	0.24 (0.05–1.06)
Diuretics
Non-user	229	233	1.0 (ref)	1.0 (ref)	1.0 (ref)
Ever user	118	124	1.00 (0.70–1.44)	1.00 (0.68–1.48)	1.06 (0.42–2.70)
Duration among enrolled 2+ y‡
Non-users	225	229	1.0 (ref)	1.0 (ref)	1.0 (ref)
<2 y	55	47	1.13 (0.71–1.80)	1.08 (0.65–1.81)	1.39 (0.44–4.33)
2+ y	73	77	0.91 (0.60–1.40)	0.95 (0.60–1.50)	0.77 (0.23–2.60)

NOTE: Odds ratios were adjusted for age, body mass index, diabetes, smoking status, hormone therapy use, and NSAID use.

Abbreviation: ref, reference.

^*570 observations in ever or never analyses and 562 observations in duration of use analyses.

^†150 observations in ever or never analyses and 150 observations in duration of use analyses.

^‡Duration of use analyses were restricted to subjects enrolled in Group Health for 2+ y.

Other Lipid-Lowering Agent

Long-term users of other lipid-lowering agents seemed to be at excess risk for colorectal cancer; however, use was low overall in this population and these findings were consistent with chance.

Angiotensin-Converting Enzyme Inhibitors

The odds ratio from multivariate models suggested a 27% decreased risk for rectal cancer with ever use of angiotensin-converting enzyme inhibitors compared with non-users, but the wide confidence limits were consistent with chance (no reduction deducible). We found no association in the analyses of colorectal and colon cancer (Table 2). Risk for colorectal cancer did not vary with increasing duration of angiotensin-converting enzyme inhibitor use, but a nonstatistically significant trend in lower rectal cancer risk was observed by duration of angiotensin-converting enzyme inhibitor use.

Calcium Channel Blockers

No association between use of calcium channel blockers and risk for colorectal cancer was observed, but a point estimate of odds ratio of 0.63 (95% CI, 0.22–1.76) was suggestive of a reduced risk for rectal cancer among ever users of calcium channel blockers compared with non-users (Table 2). However, the wide confidence limits of rectal cancer risk indicate that the risk might be consistent with chance (no reduction deducible), and we did not observe a trend in colon, rectal, or the combined endpoint of colorectal cancer risk by duration of calcium channel blocker use.

Diuretics

We found no difference in colorectal cancer risk by ever use of diuretics (Table 2).

Discussion

The results of this population-based case-control study do not indicate a reduced risk for colorectal cancer risk among users of either statins or angiotensin-converting enzyme inhibitors. We also did not find an association between other lipid-lowering drugs, calcium channel blockers, or diuretics and colorectal cancer risk. Strengths of our study include extensive and unbiased ascertainment of medication use, a population with stable enrollment over time who receive almost all of their care within an integrated group practice, and reliable data on colorectal cancer incidence. Although subjects were from a single healthcare system in western Washington State and may not be representative of other populations, use of a well-defined population where controls represent the base that produced the cases enhances the internal validity of the study.

Although laboratory studies suggest that statins inhibit cancer cell growth and lead to apoptotic cell death through inhibition of the mevalonate pathway, (13) our study and numerous other studies in humans, (25–30) including a meta-analysis of >1.5 million subjects do not support a reduced risk for colorectal cancer risk with statin use (14). However, 3 observational studies support a 35% to 47% reduced risk for colorectal cancer risk with statin use, (23, 24, 43) and meta-analyses of case-control studies show a consistent and modest reduction in colorectal cancer risk with statin use (19, 20). At least one study observed a stronger reduction in risk for rectal cancers compared with tumors that arise in the colon, (31) but the finding is not consistent and other reports, including ours, show no association (30). The National Cancer Institute recently sponsored a phase II trial of statin use and colorectal cancer risk among patients with an increased risk for colorectal cancer (44).

It is also important to note that any observed reductions in risk associated with statin use may be at least partially attributable to healthy behaviors, including diet, exercise, and colorectal cancer screening among statin users (45). Although not supported by our data, higher screening rates among statin users than non-users would result in a greater opportunity to identify premalignant lesions and bias results toward a protective association between statins and colorectal cancer. A healthy user bias may partly explain any observed associations between antihypertensive medication use and cancer risk for similar reasons. Our study did not collect or adjust for diet, exercise, socioeconomic status, or colorectal screening history, which are potential confounders of the association between all the medication classes of interest and colorectal cancer risk.

A few studies examined other lipid-lowering agents in relation to colorectal cancer risk, (24, 46, 47) and none found evidence that these drugs alter risk. Nevertheless, these studies were limited by small sample size and, similar to our results, the low prevalence of the use of lipid-lowering medications overall.

Colorectal cancer cells express angiotensin II receptors, (12) and previous animal studies suggest that angiotensin-converting enzyme inhibitors may reduce cancer risk by inhibiting tumor growth and angiogenesis (10). However, our finding of no association between angiotensin-converting enzyme inhibitor use and colorectal cancer risk is in agreement with the results of several other observational studies (15–17). One previous cohort study by Lever and colleagues (18) reported a reduction in the incidence of overall cancer, lung cancer, and breast cancer among angiotensin-converting enzyme I users. The investigators reported a 65% reduction in colorectal cancer among angiotensin-converting enzyme inhibitor users, although this finding did not achieve statistical significance.

Whereas some studies report calcium channel blockers are associated with an increased overall risk for cancer (32, 33) and breast cancer, (34, 35) our study and others find no association between the use of calcium channel blockers and risk for colorectal cancer (32, 48) calcium channel blockers are hypothesized to increase cancer risk by inhibiting apoptosis (36–38). However, there are other reports that calcium channel blockers actually promote apoptosis (49, 50) and inhibit cell growth in colon and breast carcinoma cells (51). We were unable to look at individual classes of calcium channel blockers because of small numbers, but studies suggest that immediate-release nondihydropyridine calcium channel blockers are associated with an increased risk for breast cancer (34, 35) and verapamil is associated with an increased risk for overall cancer (32). The predominant agents used at Group Health during the study period were immediate-release nondihydropyridine calcium channel blockers (diltiazem, verapamil) and two dihydropyridine calcium channel blockers (sustained-release felodipine and immediate-release nifedipine; ref. 52).

We are aware of only one other report on diuretic use and colorectal cancer risk. Tenenbaum et al. (39) found an increased risk for colorectal cancer with diuretic use compared with non-users, but this effect was limited to non-users of aspirin (hazard ratio, 2.0; 95% CI, 1.2–3.2). We did not find an association between diuretics and colorectal cancer, and we did not have sufficient sample size to evaluate aspirin or other NSAID use as an effect modifier.

There are several limitations in our study. First, our final sample size may be inadequate to detect small to moderate associations between the medications of interest and colorectal cancer risk, especially for exploratory subgroup analyses. The relatively low prevalence of long-term medication use, especially statin use, was also a limitation. Second, it is possible that exposure to medications was misclassified as patients who fill prescriptions but do not subsequently take the medication may be misclassified as users. However, misclassification of medication use is relatively unlikely because previous Group Health studies report that enrollees obtain 97% of their medications at Group Health pharmacies, (41, 42) and we required ≥2 dispensings over a 6-month period to be considered a user of the medication classes of interest.

In conclusion, our study indicates that statins and angiotensin-converting enzyme inhibitors are safe in relation to colorectal cancer risk, but any chemopreventive effects remain to be established. Reassuringly, other lipid-lowering and antihypertensive medication classes studied also seem to have no measurable impact on colorectal cancer risk.