Statins and Cancer Risk
Article Outline
Abstract
Objective
Despite numerous randomized clinical trials and observational epidemiologic studies, evidence on the potential effectiveness of statins for prevention of cancer remains controversial. The objective of this study was to investigate the relation between lipophilic statin use and cancer occurrence.
Methods
We conducted a retrospective observational study based on a medical administrative database in the province of Quebec, Canada (1998-2004). Patients aged 45 years or more and discharged from the hospital alive after admission for acute myocardial infarction were included. High- and low-dose statin use were defined as a filled prescription, within 3 days after index discharge, at or above (below) the statin-specific target dose, for any of the 4 lipophilic statin medications: atorvastatin, simvastatin, lovastatin, or fluvastatin. Statin non-use was defined as non-use of any statins while simultaneously using major non-statin cardiac medications. A total of 30,076 patients, including high-dose statin users (n
=6015), low-dose statin users (n=5323), and non-users (n=18,738), were followed for up to 7 years. Multivariable Cox regression analyses were performed to estimate associations between statin dose category and the incidence of admission to hospital with a diagnosis of any type of cancer.
Results
The crude incidence rates of hospital admission with the diagnosis of any type of cancer were 13.9, 17.2, and 26.0 per 1000 person-years in statin high-dose users, low-dose users, and non-users, respectively. The estimated adjusted hazard ratios were 0.75 (95% confidence interval [CI], 0.60-0.95) for statin use at high dose and 0.89 (95% CI, 0.75-1.07) for statin use at low dose. No significant time-dependence of the effect of statins at either dose was detected.
Conclusion
The use of lipophilic statins at sufficiently high dose might be associated with a clinically important reduction in the incidence of cancer.
Keywords: Cohort studies, Drug effects, Hydroxymethylglutaryl-CoA reductase inhibitors, Neoplasms
Statins have been shown to be effective for secondary prevention of coronary heart disease. In addition, several unintended beneficial effects have been proposed,1 including antitumorigenic effects, although original reports had actually suggested the potential opposite, procarcinogenic effects of statins.2 Despite massive amounts of data, the issue remains inconclusive.3, 4, 5, 6, 7 Apart from the fact that the duration of follow-up and the extent of adherence to the regimen across the previous studies have been varied, the reasons for the lack of coherence of the results might be that the effects could depend on the statin type, dose, or cancer at issue. In particular, cerivastatin and atorvastatin have been found to enhance angiogenesis tumor growth and vascularization at high doses, whereas the opposite effects were observed at low doses.8, 9, 10, 11, 12 With respect to statin type, it has been hypothesized that only lipophilic statins can potentially be capable of inhibiting tumor development; in contrast, hydrophilic statins could be expected to actually promote tumor development.13
In this light, recently there have emerged repeated calls for new randomized clinical trials to resolve the controversy.14, 15, 16, 17 Pending the outcome of those trials, we set out to investigate potential antitumorigenic effects of statins in a cohort of patients post-acute myocardial infarction (AMI) in the province of Quebec, Canada.
Materials and Methods
Data Sources
The Quebec hospital discharge summary database, which includes information on all AMI hospitalizations, was linked to provincial physicians and drug claims database. In Quebec, patients aged less than 65 years are typically insured through employment, whereas those who are not, as well as all patients aged 65 years or more, receive prescription coverage at minimal cost through a government program. Available prescription information includes type, dosage, quantity, and days of supply. Vital status information was obtained from the same databases linked to pension, car insurance, and death registry for completeness of the information. The patients’ unique, encrypted health care insurance number was used for the linkage.
Source Population
Patients were included if they had their first recorded AMI-related hospital admission and were discharged between April 1, 1998, and March 31, 2004. All patients had AMI (International Classification of Diseases-9-CM code 410)18 recorded as the most responsible diagnosis (ie, the principal diagnosis contributing to the greatest extent to hospital stay) in the hospital discharge database.
Patients were excluded if they met any of the following criteria: The AMI was coded as an in-hospital complication; the AMI-related hospital admission was a transfer from another hospital; the total length of hospital stay was less than 2 days; the patient was discharged to a long-term care institution or a rehabilitation center or moved out of the province; or the health care number was invalid. More details of the rationale for these criteria can be found elsewhere.19
Study Population and Definition of Study Groups
First, for all instances of statin use, daily dosages for each patient were computed on the basis of the information on dosage category, frequency, and duration for each prescription. Statin high-dose use was defined as a filled prescription, at or within 3 days after hospital discharge, at or above the statin-specific target dose (according to original randomized clinical trials), for any of the 4 statin medications: atorvastatin, simvastatin, lovastatin, or fluvastatin. Statin low-dose use was defined as a filled prescription, at or within 3 days after hospital discharge, below the statin-specific target dose for any of the 4 statins. (The narrow postdischarge time window for statin use/dose definition was chosen in an attempt to achieve a higher degree of comparability among the patients and to minimize the possibility of the “survival bias.”)20 These 4 statin medications were chosen because of their lipophilic properties,21, 22 regarded as essential for potential antitumorigenic effect.23, 24 Thus, subjects whose first statin prescription was for pravastatin or rosuvastatin were excluded because of these statins’ hydrophilic properties and because their use was too infrequent. Further, subjects whose first statin prescription was for cerivastatin were excluded because this statin is no longer used in practice. Next, subjects who were not prescribed a statin at or within 3 days after hospital discharge were identified. Among the statin non-users, those who were not prescribed a major non-statin cardiac medication (ie, a beta-blocker, a calcium-channel blocker, an angiotensin-converting enzyme inhibitor, an angiotensin II receptor blocker, a diuretic, a nitrate, digoxin, acetylsalicylic acid, or clopidogrel) at or within 3 days after hospital discharge were excluded.
Each patient was followed from hospital discharge until the occurrence of the study outcome, end of follow-up (March 31, 2005), loss to follow-up, or discontinuation of the initially adopted statin use/non-use or dose (high or low) regimen, whichever came first. In the latter case, statin high-dose users were considered to have discontinued statin use if they reached at least 60 consecutive days on which they did not fill a prescription for a high-dose statin (as defined above), whereas statin low-dose users were considered to have discontinued statin use if they reached at least 60 consecutive days on which they did not fill a prescription for a low-dose statin (as defined above). Analogously, statin non-users were considered to have discontinued a major non-statin cardiac medication use if they reached at least 60 consecutive days on which they did not fill a prescription for a non-statin cardiac medication (as defined above). The restriction of the category of statin non-use to a subdomain of use of other major cardiac medications and the incorporation of similar criteria of discontinuation of regimens were done in an effort to achieve comparability between the 3 study subpopulations, notably with respect to the distributions of potential confounders not captured by the available covariates, accuracy of outcome ascertainment, and reasons for the initially adopted regimen discontinuation, potentially resulting in informative censoring.
After the study population was thus defined, to ensure that the cohort comprised only cancer-free patients at time zero, we also excluded the patients who, within 1 year before the AMI-related hospitalization, had a diagnosis of any cancer or filled a prescription for an anticancer medication, had a cancer-related visit or hospitalization, or had 2 or more visits to an oncologist or an oncologic radiologist. Finally, patients prescribed a statin within 1 year before the AMI-related hospitalization also were excluded.
Study Outcomes
The primary outcome was the first documented hospitalization with cancer as the admission diagnosis (International Classification of Diseases-9-CM codes 140, 208).18 The secondary outcomes were hospitalizations with specific cancer types: skin, gastrointestinal tract (esophagus, stomach, colon, rectum), pancreas, lung, kidney, bladder, prostate (among men), and breast (among women).
Potential Confounders
Patient demographic characteristics and comorbidities at discharge were determined from the hospital discharge databases. The comorbidities included congestive heart failure, cerebrovascular diseases, chronic obstructive pulmonary disease, chronic kidney and liver conditions, diabetes mellitus, and dementia. Information also was obtained regarding the in-hospital procedure performed, length of hospital stay, calendar year, specialty of the treating physician, teaching status of hospital, rural/urban status of the area of hospital’s location, and annual volume of admissions in the hospital, which was categorized into low-, medium-, and high-volume categories based on tertiles of the overall distribution.
Statistical Analysis
Descriptive statistics were computed to compare baseline characteristics between the study groups. Next, hazard ratios (HRs) associated with statin use/dose were estimated in Cox models. Apart from unadjusted regressions, multivariable Cox models also were fitted, which adjusted for age, sex, marital status, comorbidities, use of non-statin cardiac medications, in-hospital procedure performed, length of hospital stay, calendar year, specialty of the treating physician, hospital’s teaching status, area of location, and annual volume of admissions. The set of covariates adjusted for was decided on a priori and included 2 broad and not mutually inclusive categories of characteristics available in our database: known or suspected risk factors for cancer; and characteristics which are known or suspected of affecting post-AMI prognosis, and which could thus affect decision-making with respect to use/dose of statin therapy. Potential time-dependency of the effect of statin use was explored by adding product-terms between the indicators of statin dose and time elapsed since time zero and by stratifying the analyses by the following segments of follow-up: less than 1 year, 1 to 3 years, and 3 to 8 years. Adjusted survival curves for each study group were constructed on the basis of the method by Ghali et al.25
Next, a series of sensitivity analyses was performed. In particular, we repeated the analyses described above but without censoring observations during follow-up in the case of discontinuation of the initially adopted statin use/dose regimen or, among statin non-users, of use of non-statin cardiac medications; without restricting statin non-users to the domain of use of non-statin cardiac medications; and varying the time interval for the definition of the study groups.
The study was approved by the McGill University Ethics Review Board.
Results
Definition of Study Groups
During the study period, a total of 54,210 patients were discharged from hospital after AMI who were eligible for inclusion in the study. The flow of the patients is shown in Figure 1. After the study group-defining criteria were applied, there were 6015, 5323, and 18,738 patients in the statin high-dose user, low-dose user, and non-user groups, respectively.
Figure 1.
Flow of participants in the study.
Baseline Characteristics and Univariable Analyses
Table 1 presents descriptive statistics on selected characteristics of the patients according to the study group. In general, users of statins at high and low doses had similar characteristics. Statin non-users tended to be slightly older than users. The 3 groups differed in terms comorbidities, the prevalence of which tended to be higher among statin non-users than among users. On the other hand, statin users tended to have undergone invasive cardiac procedures more frequently than non-users. The patterns of statin use/dose are shown in Table 2.
Table 1. Baseline Characteristics of Patients After Acute Myocardial Infarction According to Study Group
| Characteristic | Statin High-dose Users (n=6015) | Statin Low-dose Users (n=5323) | Statin Non-users (n=18,738) |
|---|---|---|---|
| Age, mean (SD), y | 64.2(12.9) | 65.9(12.3) | 71.4(12.7) |
| Men, No. (%) | 4048(67.3) | 3477(65.3) | 10,787(57.6) |
| Marital status: married, No. (%) | 1493(24.8) | 1079(20.3) | 3027 (16.1) |
| Comorbidities, No. (%) | |||
| Congestive heart failure | 798(13.3) | 799(15.0) | 4725(25.2) |
| Cerebrovascular disease | 174(2.9) | 211(4.0) | 1118(6.0) |
| Pulmonary edema | 37(0.6) | 33(0.6) | 289(1.5) |
| Diabetes mellitus | 1155(19.2) | 940(17.7) | 4470(23.9) |
| Acute renal failure | 137(2.3) | 137(2.6) | 965(5.1) |
| Chronic renal failure | 304(5.0) | 283(5.3) | 1981(10.6) |
| Shock | 49(0.8) | 43(0.8) | 182(1.0) |
| Cardiac dysrhythmia | 816(13.6) | 878(16.5) | 4004(21.4) |
| Hypertension | 2205(36.7) | 1831(34.4) | 6602(35.2) |
| COPD | 812(13.5) | 723(13.6) | 3738(19.9) |
| Liver disease | 39(0.6) | 31(0.6) | 165(0.9) |
| Dementia | 49(0.8) | 42(0.8) | 444(247) |
| Admitting physician, No.(%) | |||
| General practitioner | 2230(37.1) | 2297(43.1) | 8768(46.8) |
| Cardiologist | 1393(53.1) | 2424(45.5) | 8043(42.9) |
| Internal medicine | 572(9.5) | 570(10.7) | 1709(9.1) |
| Medication use, No. (%) | |||
| Non-statin lipid-lowering medication | 109(1.8) | 71(1.3) | 1030(5.5) |
| Beta-blockers | 4934(82.0) | 4102(77.1) | 11,706(62.5) |
| ACE inhibitor | 4405(66.6) | 2948(55.4) | 8931(47.7) |
| ARB | 241(4.0) | 159(2.9) | 640(3.4) |
| ASA/clopidogrel | 5555(92.3) | 4516(84.8) | 13,274(70.8) |
| Calcium-channel blocker | 2507(41.7) | 1071(20.1) | 4635(24.7) |
| Nitrates (non-sublingual) | 4627(65.3) | 3630(68.2) | 13,209(70.5) |
| Digoxin | 2160(35.9) | 614(11.5) | 2636(14.1) |
| Diuretics | 1056(17.6) | 952(17.9) | 6050(32.3) |
| Procedures, No. (%) | |||
| PCI | 3035(54.5) | 1935(36.9) | 4110(22.1) |
| CABG | 542(9.7) | 571(10.9) | 1151(6.2) |
| Calendar year period, No. (%) | |||
| 1997-1999 | 147(2.4) | 1110(20.8) | 7411(39.5) |
| 1999-2001 | 504(8.4) | 1480(27.8) | 5558(30.0) |
| 2001-2003 | 1933(32.1) | 1555(29.2) | 3598(19.2) |
| 2003-2005 | 3431(37.0) | 1178(22.1) | 2171(12.6) |
| Length of hospital stay, median (IQR), d | 7(4-11) | 8(5-12) | 9(6-14) |
| Hospital status: teaching, No. (%) | 1297(21.6) | 992(18.6) | 2492(13.3) |
| Hospital location: rural area, No. (%) | 323(5.4) | 300(5.6) | 959(5.1) |
| Hospital volume of admissions, No. (%) | |||
| Low | 124(2.0) | 107(2.0) | 328(1.7) |
| Medium | 365(6.1) | 362(6.8) | 958(5.1) |
| High | 5526(91.9) | 4854(91.2) | 17,452(93.1) |
Table 2. Patterns of Statin Use/Dose
| Statin High-dose Users (n = 6015) | Statin Low-dose Users (n = 5323) | Statin Non-users (n = 18,738) | |
|---|---|---|---|
| Duration of follow-up, median (IQR), y | 0.9(0.4-2.0) | 1.3(0.5-2.8) | 1.0(0.3-2.5) |
| Total population-time of follow-up, y | 8059 | 10,085 | 31,234 |
| Discontinued an initially adopted statin use/non-use regimen | 1240(20.6) | 2405(45.2) | 4197(22.4) |
| Atorvastatin daily dose, median (IQR), mg/d | 20(20-40) | 10(10-10) | 0 |
| Simvastatin daily dose, median (IQR), mg/d | 40(40-40) | 20(10-20) | 0 |
| Lovastatin daily dose, median (IQR), mg/d | 40(30-40) | 20(10-20) | 0 |
| Fluvastatin daily dose, median (IQR), mg/d | 40(40-40) | 20(20-20) | 0 |
A total of 1099 of subjects were admitted to a hospital with cancer as the admission diagnosis during the follow-up, with the overall unadjusted incidence rates being 13.9, 17.2, and 26.0 per 1000 person-years among statin high-dose users, low-dose users, and non-users, respectively. For each of the cancer types examined, the incidence rate was lower among statin users than non-users (Table 3).
Table 3. Number of Cases and Incidence Rates (per 1000 Person-Years) (95% Confidence Interval) for Specific Cancer Types and All Cancers, According to Study Group
| Statin High-dose Users (n = 6015) | Statin Low-dose Users (n = 5323) | Statin Non-users (n = 18,738) | |
|---|---|---|---|
| Skin | 13 | 13 | 49 |
| 1.6(0.7-2.5) | 1.3(0.6-2.0) | 1.6(1.1-2.0) | |
| Gastrointestinal (esophagus, stomach, colon, rectum) | 22 | 39 | 173 |
| 2.7(1.6-3.9) | 3.9(2.6-5.1) | 5.5(4.7-6.4) | |
| Pancreas | 2 | 7 | 29 |
| 0.2(0.0-0.6) | 0.7(0.2-1.2) | 0.9(0.6-1.3) | |
| Lung | 30 | 30 | 179 |
| 3.7(2.4-5.0) | 3.0(1.9-4.0) | 5.7(4.9-6.6) | |
| Kidney | 3 | 9 | 27 |
| 0.4(0.0-0.8) | 0.9(0.3-1.5) | 0.9(0.5-1.2) | |
| Bladder | 9 | 13 | 73 |
| 1.1(0.4-1.8) | 1.3(0.6-2.0) | 2.4(1.8-2.9) | |
| Prostate(men only) | 10 | 15 | 90 |
| 1.8(0.7-3.0) | 2.3(1.1-3.5) | 5.3(4.2-6.4) | |
| Breast (women only) | 5 | 10 | 44 |
| 1.9(0.2-3.5) | 2.8(1.1-4.5) | 3.1(2.2-4.0) | |
| All cancers | 112 | 174 | 813 |
| 13.9(11.3-16.5) | 17.2(14.7-19.8) | 26.0(24.2-27.8) |
Multivariable Analyses
The adjusted survival curves for each study group are shown in Figure 2. The adjusted HR associated with high-dose statin use versus non-use was 0.75 (95% confidence interval [CI], 0.60-0.95), whereas the adjusted HR associated with low-dose statin use versus non-use was 0.89 (0.75-1.07). The addition to the model of the product terms between the indicators of high-dose and low-dose statin use and the duration of follow-up variable suggested that the relative hazards were constant over time (P values for the product-terms=.40 and .59, respectively). The estimates of time segment-specific adjusted HRs (with their corresponding 95% CIs) were mutually compatible for both high-dose use versus non-use (0.77 [0.56-1.07] for<1 year, 0.84 [0.58-1.22] for 1-3 years, and 0.53 [0.25-1.11] for 3-8 years) and low-dose use versus non-use (0.89 [0.67-1.18] for<1 year, 0.91 [0.68-1.22] for 1-3 years, and 0.91 [0.64-1.31] for 3 to 8 years).
Figure 2.
Adjusted survival curves according to study group.
When we compared the effect of high-dose versus low dose, the adjusted HR (95% CI) was estimated to be 0.84 (0.66-1.08). Although the association for this contrast is not statistically significant (P value=.18) and does not, therefore, establish effect heterogeneity with respect to dose, this evidence cannot be treated as supporting the absence of heterogeneity because the lower 95% confidence bound of 0.66 would be consistent with a clinically important difference in the effect between the 2 dose categories. When the 2 dose categories were pooled, the adjusted HR (95% CI) for any use versus non-use was 0.84 (0.72-0.98).
Sensitivity Analyses
When we repeated the analysis without implementing censoring at discontinuation of the initially adopted statin use/dose regimen or, among statin non-users, of use of non-statin cardiac medications, the estimated adjusted HRs associated with high-dose statin use versus non-use and low-dose statin use versus non-use were 0.54 (95% CI, 0.45-0.65) and 0.74 (0.85-0.84), respectively. When we repeated the analyses without restricting statin non-users to the domain of use of non-statin cardiac medications, the results suggested that the restriction had little effect on the estimates of effect of either high dose or low dose. Finally, when we considered alternative time intervals for the definition of the study group, the results obtained with the 7-day and 14-day time intervals were practically identical to those obtained with the 3-day definition.
Discussion
Our findings in this study of 30,076 patients post-AMI suggest that lipophilic statins used at or above the target dose may have clinically important antitumorigenic effects.
The fact that we did not observe a material time-dependence in the effect of statins is in line with some previous studies, which did not find significant differences in the degree of relative rate reduction in cancer incidence across categories of duration of statin use.26, 27, 28 Similarly, a recent study by Setoguchi et al29 suggested modestly protective effects of short-term (<3 years) statin use but modestly harmful effects of longer-term use (≥3 years) with respect to breast cancer incidence. In general, such a pattern of effect could be attributable to potential effects of statins on the rate of progression and growth of latent, subclinical cases of cancer and would thus be consistent with such putative mechanisms of antitumorigenic action of statins as inhibition of tumor angiogenesis and proliferation of cancer cells and impairment of metastatic potential.30, 31, 32 Further, although in another recent study no association was found between statin use and the incidence of prostate cancer overall, the authors did report a strong preventive effect of statins on the incidence of advanced prostate cancer, thereby suggesting that statins could affect cancer progression at advanced stages.33 Still, the brief time lag between the onset of statin therapy and the reduction in the incidence of cancer should be viewed with caution and requires confirmation in future studies.
In one of the most recent and comprehensive published meta-analyses, Dale et al5 pooled results from 26 randomized clinical trials of statins that had cancer-related outcomes documented. The pooled odds ratio point and interval estimates for the outcomes of incidence of cancer and cancer death suggested null associations with statin use. However, a large percentage of the included trials used pravastatin, which is hydrophilic, as the study statin; consequently, the pooling of data for lipophilic statins with those for pravastatin may have obscured the effect.15, 23 Although the authors did perform analyses stratified by statins’ lipophilicity/hydrophilicity status, they classified atorvastatin, for which both point and interval odds ratio estimates were compatible with a moderate cancer death-preventing effect, as a hydrophilic statin even if this statin is generally considered lipophilic.34 The odds ratio point estimates for individual lipophilic statins other than atorvastatin were close to unity, although the corresponding interval estimates were compatible with small to moderate pro- or antitumorigenic effects, depending on the statin. Another issue is that the published reports contained only intention-to-treat analyses, which ignored the incomplete adherence to the randomly assigned regimen and the fact that not all the subjects reached or maintained the target dose. Yet, these issues should be particularly relevant for cancer-related outcomes because sufficiently high dosage is considered important for statins’ putative antitumorigenic effects.11, 13
Several non-experimental studies have been conducted addressing the occurrence of cancer in relation to statins.26, 27, 28, 29, 35, 36, 37, 38, 39, 40 Although the reported findings have been variable, non-experimental studies have tended to suggest stronger antitumorigenic effects of statins than have randomized clinical trials. In particular, both clinically and statistically significant estimates of relative risk reduction associated with statin use versus non-use have been reported for breast cancer,27 non-Hodgkin lymphoma,39 colorectal cancer,38 and prostate cancer.40 Similarly, several studies reported seemingly beneficial effects of statins with respect to all cancers combined.26, 28, 35 One of the most important issues when interpreting the evidence from this source is the likely non-comparability of statin users and non-users with respect to potential confounders. Thus, although some authors have simplistically contrasted users of statins with non-users of statins, others, like us, in an attempt to minimize the possibility of a confounded statin user/non-user contrast, have restricted the statin non-user category to that of users of other medications, such as non-statin cardiac medications,35 non-statin lipid-lowering agents,26, 28 or glaucoma medications.29
Our study had several limitations. First, we relied on prescription data rather than on actual use of statins. However, the misclassification was likely non-differential and could thus be expected to dilute the documented associations. Second, we used hospital admission for cancer as the outcome; consequently, some incident cases of cancer may have been missed. However, it is unlikely that the outcome ascertainment pattern would be materially different between statin users and non-users, as defined for the purposes of our study. Third, we were not able to adequately study the organ-specific effects of statin use on the incidence of cancer. Similarly, we were not able to study the long-term effects of statin use; thus, no evidence was obtained regarding possible “genuine” cancer prevention by potential effects of statins on cancer initiation. Still, today’s reality is that statins are mostly indicated for post–middle-aged individuals, for whom relatively short-term effects on cancer promotion are arguably more relevant than long-term effects on cancer initiation. Finally, a possibility remains that an imbalanced distribution of potential confounders between statin users and non-users could introduce confounding bias in our results despite statistical adjustment for a number of covariates. However, our operationalization of statin non-use as use of major non-statin cardiac medications and application of similar censoring criteria for the 3 study groups should minimize such potential sources of bias under the rubric of confounding as “healthy statin-user effect” and “healthy complier effect.” Moreover, because the unintended putative cancer-preventing effects are at issue, cancer risk would not be expected to play a role in statin-prescribing decision-making on the part of the patient or the physician. Thus, we do not believe confounding by indication presents a serious problem to interpretation of our findings.
Conclusions
This is the first study to suggest a dose-response effect of lipophilic statins on cancer occurrence. Future studies should provide additional evidence allowing the assessment of long-term effects of statins on cancer risk.
References
- . Potential nontraditional applications of statins. Ann Pharmacother. 2003;37:1067–1071
- . Carcinogenicity of lipid-lowering drugs. JAMA. 1996;275:55–60
- . Do statins cause cancer? (A meta-analysis of large randomized clinical trials). Am J Med. 2001;110:716–723
- Statins and cancer risk: a literature-based meta-analysis and meta-regression analysis of 35 randomized controlled trials. J Clin Oncol. 2006;24:4808–4817
- Statins and cancer risk: a meta-analysis. JAMA. 2006;295:74–80
- Statins and cancer prevention. Nat Rev Cancer. 2005;5:930–942
- . Statin use and cancer risk: an epidemiologic review. Cancer Invest. 2006;24:413–424
- Differential sensitivity of various pediatric cancers and squamous cell carcinomas to lovastatin-induced apoptosis: therapeutic implications. Clin Cancer Res. 2001;7:158–167
- Statins differentially regulate vascular endothelial growth factor synthesis in endothelial and vascular smooth muscle cells. Atherosclerosis. 2003;170:229–236
- . Class-specific pro-apoptotic effect of statins on human vascular endothelial cells. Z Kardiol. 2004;93:398–402
- Double-edged role of statins in angiogenesis signaling. Circ Res. 2002;90:737–744
- Statins have biphasic effects on angiogenesis. Circulation. 2002;105:739–745
- . Statins and cancer—beyond the “one drug, one disease” model. N Engl J Med. 2005;352:2238–2239
- . [Letter] Clinical trials are required to prove the chemopreventive worth of statins. Arch Intern Med. 2006;166:1143
- . Mevalonate promotes the growth of tumors derived from human cancer cells in vivo and stimulates proliferation in vitro with enhanced cyclin-dependent kinase-2 activity. J Biol Chem. 2004;279:33079–33084
- . Lipophilic statins merit additional study for breast cancer chemoprevention. J Clin Oncol. 2006;24:2128–2129
- . Breast cancer prevention: time for randomized controlled trials with statins. Arch Intern Med. 2006;166:1143–1144
- International Classification of Diseases, Clinical Modification, 9th Revision (ICD-9-CM). Washington, DC: Public Health Service, US Department of Health and Human Services; 1998;
- . An overview of the methods and data in the CCORT Canadian Cardiovascular Atlas project. Can J Cardiol. 2003;19:655–663
- . Surival bias associated with time-to-treatment association in drug effectiveness evaluation: a comparison of methods. Am J Epidemiol. 2005;162:1016–1023
- . Hydrophilicity/lipophilicity: relevance for the pharmacology and clinical effects of HMG-CoA reductase inhibitors. TiPs. 1998;19:26–37
- . Chemical, pharmacologic and pharmacodynamic properties of statins: an update. Fundam Clin Pharmacol. 2005;19:1897–1898
- . Statins and cancer development. Cancer Epidemiol Biomarkers Prev. 2005;14:1897–1898
- . The role of statins in cancer prevention and treatment. Oncology. 2005;19:739–750
- Comparison of 2 methods for calculating adjusted survival curves from proportional hazards models. JAMA. 2001;286:1494–1497
- . 3-Hydroxy-3-methylglutaryl coenzyme A reductase inhibitors and the risk of cancer: a nested case-control study. Arch Intern Med. 2000;160:2363–2368
- Statin use and breast cancer: prospective results from the Women’s Health Initiative. J Natl Cancer Inst. 2006;98:700–707
- Cancer risk among statin users: a population-based cohort study. Int J Cancer. 2005;643–647
- Statins and the risk of lung, breast, and colorectal cancer in the elderly. Circulation. 2007;115:27–33
- . The statins as anticancer agents. Clin Cancer Res. 2003;9:10–19
- Potential anticancer effects of statins: facts or fiction?. Endothelium. 2003;10:49–58
- The potential of statins as part of anti-cancer treatment. Eur J Cancer. 2005;41:516–522
- Statin drugs and risk of advanced cancer. J Natl Cancer Inst. 2006;98:1819–1825
- . Statins and the risk of cancer. [Letter] JAMA. 2006;21:2720
- The risk of cancer in users of statins. J Clin Oncol. 2004;22:2388–2394
- . Statin use and cancer risk in the General Practice Research Database. Br J Cancer. 2004;90:635–637
- Statin use, hyperlipidaimia, and the risk of breast cancer. Br J Cancer. 2002;86:1436–1439
- Statins and the risk of colorectal cancer. N Engl J Med. 2005;352:2184–2192
- Prior medical conditions and medication use and risk of non-Hodgkin lymphoma in Connecticut United States women. Cancer Causes Control. 2004;15:419–428
- Statins and prostate cancer risk: a case-control study. Am J Epidemiol. 2005;162:318–325
This study was undertaken as part of the Gender and Sex Determinants of Cardiovascular Disease (GENESIS) project, funded by the Canadian Institutes of Health Research Institutes of Gender and Health and Circulatory and Respiratory Health, and Heart and Stroke Foundation of Canada. Dr Pilote is a research scholar of the Canadian Institutes of Health Research and a William Dawson Professor at McGill University. None of the authors have financial interest or conflict with regard to the content discussed in this manuscript. Dr Pilote had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
PII: S0002-9343(08)00074-0
doi:10.1016/j.amjmed.2007.12.011
© 2008 Elsevier Inc. All rights reserved.
