Abstract
Statins are widely prescribed for cardiovascular disease prevention and also commonly used in patients at high risk for colorectal cancer. We report the results of a planned secondary analysis of the relationship between statin use and colorectal adenoma risk in a large chemoprevention trial. The Adenoma Prevention with Celecoxib (APC) trial randomized 2,035 adenoma patients to receive placebo (679 patients), 200 mg celecoxib twice daily (bid; 685 patients), or 400 mg celecoxib bid (671 patients). The study collected complete medical history and medication use data and performed colonoscopic surveillance to 5 years after study enrollment. The effects of statin use on newly detected adenomas and cardiovascular adverse events were analyzed as time-dependent variables by multivariable Cox regression. Statins were used by 36% (n = 730) of APC trial participants. When adjusted for covariates including cardioprotective aspirin use, age, and sex, participants on the placebo arm who used statins at any time had no benefit over 5 years compared with never users (risk ratio, 1.24; 95% confidence interval, 0.99-1.56; P = 0.065). Statin use for >3 years increased adenoma risk over 5 years (risk ratio, 1.39; 95% confidence interval, 1.04-1.86; P = 0.024). For all comparisons of patients treated with celecoxib, adenoma detection rates for statin users and nonusers were equivalent. Consistent with their use in patients at high risk, cardiovascular serious adverse events were more common among statin users. For patients at high risk of colorectal cancer, statins do not protect against colorectal neoplasms and may even increase the risk of developing colorectal adenomas. Cancer Prev Res; 3(5); 588–96. ©2010 AACR.
Read the Perspective on this article by Baron, p. 573
Introduction
3-Hydroxy-3-methylglutaryl CoA reductase inhibitors, also known as statins, are widely prescribed medications for prevention of hypercholesterolemia and related diseases. Statins show pleotropic effects, with activities both related to and independent of their ability to block cholesterol synthesis. Preclinical studies suggest that statins suppress colorectal cancer (CRC) growth. In model systems, statins inhibit angiogenesis, induce apoptosis in CRC cell lines, and alter cell-cell adhesion (1). Statins also prevent tumor formation in azoxymethane-induced and Apc‐deficient rodent models (2, 3).
Early clinical trials suggested that statin use and/or low cholesterol levels might increase cancer risk (4–7). Cancer as an endpoint was therefore carefully scrutinized in secondary analyses of large statin cardiovascular trials and in a number of case-control and cohort studies (8). The results with respect to CRC incidence are mixed. One population-based case-control study in elderly patients found a 47% reduced CRC risk for those using statins for at least 5 years (9). A second observational study also showed a protective statin effect, superior to that seen with low-dose aspirin (10). A statin-associated reduction in CRC incidence, however, was not confirmed in other prospective cohort or retrospective case-control studies (11–16), nor was it seen by meta-analysis (17).
The first statin, lovastatin, was approved by the Food and Drug Administration in 1987, and statins did not achieve widespread use until about a decade later. Because CRC takes approximately 10 to 20 years to develop, it is possible that lack of consensus on statin use and CRC incidence results from inadequate drug exposure and follow-up intervals. The study with longest follow-up duration was a population-based study from Finland that examined cancer incidence in 472,481 case-control pairs, where cases were statin users with a 3.06-year median duration of medication (16). Over an average follow-up duration of 8.8 years, this study found no difference in incidence of colon or rectal cancers. Adenoma formation is an earlier indicator of CRC risk, but the current literature contains very little statin use data concerning this end point. Two retrospective database studies suggested benefit (18, 19). A secondary analysis of three large colorectal adenoma prevention trials pooled data for 2,915 subjects, of which 8.1% used statins (20). Despite the small numbers, this study could rule out a ≥14% protective effect with 95% confidence and found a null result for an association between statin use and adenoma recurrence.
The Adenoma Prevention with Celecoxib (APC) trial was a randomized placebo-controlled trial of celecoxib for prevention of colorectal adenomas in patients at high CRC risk. This study randomized 2,035 patients to receive placebo (679 patients), 200 mg celecoxib twice daily (685 patients), or 400 mg celecoxib twice daily (671 patients), with follow-up for up to 5 years postrandomization. The primary end points of the APC trial have been reported (21, 22). For patients on placebo, the incidence of newly detected adenomas by year 3 was 60.7%, confirming that the APC trial included a high adenoma risk cohort. New adenoma detection during the 3-year surveillance interval was reduced by 33% for those on 200 mg celecoxib bid and by 45% for those on 400 mg celecoxib bid.
The APC trial captured clinical data relevant to understanding both overall health and adenoma risk. Prospectively gathered data included complete medical history, with a detailed medication use history recording duration of drug use that was updated every 6 months during the active treatment portion of the trial. Because the trial was designed before the cardiovascular toxicity of selective Cox-2 inhibitors and nonsteroidal anti-inflammatory drugs was recognized, patients with a history of cardiovascular disease were eligible. As a result, 75% of participants had at least one cardiovascular risk factor (23) and 36% used statins. The relationship between statin use and adenoma recurrence was a planned secondary objective of the APC trial.
Materials and Methods
Study design
The APC trial tested whether celecoxib reduced the occurrence of endoscopically detected colorectal adenomas, comparing treatment with placebo, celecoxib 200 mg bid, or celecoxib 400 mg bid. Randomization was stratified based on low-dose aspirin use (doses of ≤325 mg qod or 162.5 mg qd) and clinical site. The trial involved 91 sites in the United States, Australia, Canada, and Great Britain. Each site received human subjects committee approval of the study, and all patients provided written consent for participation. During the treatment portion of the trial, an independent data and safety monitoring board reviewed safety data monthly and efficacy data semianually.
Recruitment and randomization
Details of the APC trial have been published (21). Participants ranged from 31 to 88 y of age at enrollment and had a high risk of recurrent colorectal adenomas based on either multiple adenomas or removal of a single adenoma ≥6 mm in diameter. Within 3 mo before enrollment, participants had a colonoscopy with removal of all polyps, one or more of which were histologically confirmed adenomas. Exclusion criteria included a history of familial adenomatous polyposis, hereditary nonpolyposis colon cancer, inflammatory bowel disease, or large bowel resection other than appendectomy.
APC trial participants who completed an initial 3-year study interval were offered the option of continuing on an extension study, allowing them to continue taking study medication in a blinded manner for an additional 2 y. Participants not wishing to remain on study medication were allowed to continue in an additional 2-y surveillance arm. At 5 y after randomization, participants in the extension study underwent a final colonoscopy.
Study treatment
APC trial accrual was completed in March 2002. In late 2004, rofecoxib was withdrawn from the market due to recognition of its association with increased risk of serious cardiovascular adverse events. At this time, APC trial investigators conducted an independent adjudicated review of cardiovascular safety. This analysis showed 2.6- and 3.4-fold increases in selected cardiovascular events compared with placebo in patients receiving celecoxib at doses of 200 and 400 mg bid, respectively (22). In response, study medication use was discontinued on December 17, 2004, approximately 3 mo before the final randomized patients were scheduled to complete the 3-year treatment interval. Patients returned to usual care, which entailed colonoscopic surveillance without chemoprevention, and all participants who wished to do so remained on study for continued collection of safety data and completion of the year 5 colonoscopy.
End-point assessment and follow-up
The APC trial collected detailed medication use data at baseline and during the trial, including the length of statin use before study entry. Subjects on treatment were contacted every 2 mo to update concomitant medication use and report adverse events. Subjects who were no longer using study medication were contacted by telephone every 6 to 12 mo for adverse event reporting. A study investigator performed a complete colonoscopy with visualization of the cecum and endoscopic removal of all polyps at 1, 3, and 5 y after randomization. All polyps removed during these colonoscopies were reviewed by a central study pathologist. Investigator-reported adverse events were classified according to MedRA 8.1 criteria.
Statistical analysis
Associations between baseline characteristics and statin use were assessed using χ2 and ANOVA for categorical and continuous variables, respectively. For these analyses, statin use was subdivided by baseline and postbaseline use according to whether first use of statin medication began by the randomization date. The total duration of statin use was calculated from statin start date until the date of final on-study colonoscopy to estimate the maximum length of use. Specific details of how duration of statin use were quantified are provided in Supplementary Data. The effect of statin use duration on adenoma recurrence was assessed as a time-dependent variable. Analysis of time to adenoma recurrence and cardiovascular adverse events was done using time-dependent Cox models. The significance of parameter estimates from the fitted regression model was assessed by the conventional Wald test. Because adenoma recurrence could only be detected at colonoscopies performed at 1, 3, or 5 y of follow-up, the tied event times were handled by the exact method in PHREG procedure in SAS. All demographic and clinical characteristics associated with statin use were evaluated as covariates. A propensity score for a patient's predicted likelihood of receiving statins conditional on observed factors was also evaluated as a covariate on statin effect. Propensity scores were estimated by a multivariable logistic regression model fitted with statin use–associated baseline factors, including age (≥65 versus <65 y), sex, aspirin use, history of cardiovascular events, hypertension, and diabetes. Because controlling for the propensity score variable yielded the same result on statin effect as directly adjusting for aspirin use, age, and sex, all subsequent analyses on newly detected adenoma were modeled with the three covariates as separate variables (24). Analyses were conducted separately for the placebo group and for the celecoxib groups. All statistical analysis was done using SAS 9.2 (SAS Institute).
Results
Characteristics of the study population
Baseline statin use data were available for 2,028 of 2,035 (99.7%) of participants. Statins were used at baseline by 502 patients (24.7%) and were started during the first 3 years of the trial in 228 patients (11.2%), resulting in a total number of statin users of 730 (36%) as follows: lipophilic statins—atorvastatin n = 406 (55.6%), lovastatin n = 16 (2.2%), and simvastatin n = 214 (29.3%), and hydrophilic statin—pravastatin n = 94 (12.9%). The overall median duration of statin use was 3.6 years (range, 0.01-28.7 years; Table 1). Statin use was more common in participants age ≥65 years at study entry and in men (Table 2). Statin users were more likely to be using cardioprotective aspirin (45.6% of statin users versus 22.7% of statin nonusers; P < 0.001). A baseline history of significant cardiovascular events (e.g., angina, myocardial infarction, congestive heart failure, and cardiovascular or cerebrovascular disease) was present in 14% of participants, and 67.1% of patients with this history used statins. Other subsets with a higher proportion of statin users included those with a history of smoking, hypertension, and diabetes. Statin use was balanced across all three treatment arms and was also balanced for variables associated with adenoma risk, including CRC family history and size and number of adenomas before treatment (Table 2).
Frequency and duration of statin use among APC trial participants
Subjects using statins at baseline, n = 502 (24.7%)* . | ||
---|---|---|
Duration of use before randomization (y)† . | Frequency . | Percent . |
≤1 | 138 | 27.5 |
1-3 | 149 | 29.7 |
3-5 | 102 | 20.3 |
>5 | 102 | 20.3 |
Indeterminate‡ | 11 | 2.2 |
Median (range)§ | 2.4 (0.02-25.7) | |
Subjects beginning statins during APC trial, n = 296 (14.6%) | ||
Duration of use following randomization (y)∥ | Frequency | Percent |
≤1 | 73 | 24.7 |
1-3 | 149 | 50.3 |
3-5 | 6 | 2.0 |
Median (range)¶ | 1.7 (0.01-3.4) | |
All subjects using statins in trial, n = 730 (35.9%) | ||
Total duration of use (y)∥ | ||
≤1 | 85 | 11.6 |
1-3 | 190 | 26.0 |
3-5 | 197 | 27.0 |
>5 | 241 | 33.0 |
Indeterminate | 17 | 2.3 |
Median (range)** | 3.6 (0.01-28.7) |
Subjects using statins at baseline, n = 502 (24.7%)* . | ||
---|---|---|
Duration of use before randomization (y)† . | Frequency . | Percent . |
≤1 | 138 | 27.5 |
1-3 | 149 | 29.7 |
3-5 | 102 | 20.3 |
>5 | 102 | 20.3 |
Indeterminate‡ | 11 | 2.2 |
Median (range)§ | 2.4 (0.02-25.7) | |
Subjects beginning statins during APC trial, n = 296 (14.6%) | ||
Duration of use following randomization (y)∥ | Frequency | Percent |
≤1 | 73 | 24.7 |
1-3 | 149 | 50.3 |
3-5 | 6 | 2.0 |
Median (range)¶ | 1.7 (0.01-3.4) | |
All subjects using statins in trial, n = 730 (35.9%) | ||
Total duration of use (y)∥ | ||
≤1 | 85 | 11.6 |
1-3 | 190 | 26.0 |
3-5 | 197 | 27.0 |
>5 | 241 | 33.0 |
Indeterminate | 17 | 2.3 |
Median (range)** | 3.6 (0.01-28.7) |
*Eight subjects on statins before treatment start date did not have a postrandomization colonoscopy.
†Duration before randomization is calculated from date of first statin use until treatment start date.
‡Eleven statin users were missing a date of first statin use, and for these treatments start date was imputed for statin start date in the analysis.
§n = 491; excludes subjects with indeterminate statin use start date.
∥Duration is calculated from date of first statin use until date of final on-study colonoscopy; indeterminate was due to missing statin start date or having no postrandomization colonoscopy.
¶n = 228; excludes subjects whose statin use began after final on-study colonoscopy.
**n = 713; excludes subjects with indeterminate statin use start date and subjects whose statin use began after final on-study colonoscopy. Eighty-three subjects discontinued statin use during the trial; the median duration when calculated with statin end date is 3.5 y (0.01-28.7 y).
Baseline characteristics and statin use in APC trial participants
Characteristics . | All, n = 2,028 . | Never use, n = 1,298 . | Baseline use,* n = 502 . | Postbaseline use,† n = 228 . | P‡ . |
---|---|---|---|---|---|
Age group (y), n (%) | <0.001 | ||||
<65 | 1404 (69) | 963 (74) | 283 (56) | 158 (69) | |
≥65 | 624 (31) | 335 (26) | 219 (44) | 70 (31) | |
Median age (range) | 59 (31-88) | 57 (31-88) | 63 (37-82) | 60 (39-80) | <0.001 |
Gender, n (%) | <0.001 | ||||
Women | 647 (32) | 459 (35) | 125 (25) | 63 (28) | |
Men | 1381 (68) | 839 (65) | 377 (75) | 165 (72) | |
Smoking history, n (%) | 0.001 | ||||
Current | 335 (17) | 219 (17) | 71 (14) | 45 (20) | |
Former | 907 (45) | 542 (42) | 257 (51) | 108 (47) | |
Never | 785 (39) | 536 (41) | 174 (35) | 75 (33) | |
Body mass index | |||||
Men | 28.7 ± 0.12 | 28.6 ± 0.16 | 29.1 ± 0.22 | 28.9 ± 0.33 | 0.059 |
Women | 28.9 ± 0.25 | 28.8 ± 0.30 | 29.3 ± 0.55 | 29.2 ± 0.74 | 0.410 |
CRC in a parent, n (%) | 427 (21) | 270 (21) | 95 (19) | 62 (27) | 0.708 |
Findings at baseline colonoscopy | |||||
No. of adenomas | 2.1 ± 0.03 | 2.0 ± 0.04 | 2.2 ± 0.07 | 2.0 ± 0.09 | 0.491 |
Adenoma burden, cm§ | 1.5 ± 0.03 | 1.5 ± 0.03 | 1.5 ± 0.06 | 1.5 ± 0.07 | 0.917 |
History of cardiovascular disease,∥ n (%) | 292 (14) | 96 (7) | 167 (33) | 29 (13) | <0.001 |
History of hypertension, n (%) | 833 (41) | 463 (36) | 280 (56) | 90 (39) | <0.001 |
History of diabetes, n (%) | 194 (10) | 87 (7) | 85 (17) | 22 (10) | <0.001 |
Use of low-dose aspirin,¶ n (%) | 627 (31) | 294 (23) | 266 (53) | 67 (29) | <0.001 |
Randomized to placebo, n (%) | 676 (33) | 441 (34) | 167 (33) | 68 (30) | |
Randomized to celecoxib, 200 mg, n (%) | 684 (34) | 434 (33) | 169 (34) | 81 (36) | |
Randomized to celecoxib, 400 mg, n (%) | 668 (33) | 423 (33) | 166 (33) | 79 (35) |
Characteristics . | All, n = 2,028 . | Never use, n = 1,298 . | Baseline use,* n = 502 . | Postbaseline use,† n = 228 . | P‡ . |
---|---|---|---|---|---|
Age group (y), n (%) | <0.001 | ||||
<65 | 1404 (69) | 963 (74) | 283 (56) | 158 (69) | |
≥65 | 624 (31) | 335 (26) | 219 (44) | 70 (31) | |
Median age (range) | 59 (31-88) | 57 (31-88) | 63 (37-82) | 60 (39-80) | <0.001 |
Gender, n (%) | <0.001 | ||||
Women | 647 (32) | 459 (35) | 125 (25) | 63 (28) | |
Men | 1381 (68) | 839 (65) | 377 (75) | 165 (72) | |
Smoking history, n (%) | 0.001 | ||||
Current | 335 (17) | 219 (17) | 71 (14) | 45 (20) | |
Former | 907 (45) | 542 (42) | 257 (51) | 108 (47) | |
Never | 785 (39) | 536 (41) | 174 (35) | 75 (33) | |
Body mass index | |||||
Men | 28.7 ± 0.12 | 28.6 ± 0.16 | 29.1 ± 0.22 | 28.9 ± 0.33 | 0.059 |
Women | 28.9 ± 0.25 | 28.8 ± 0.30 | 29.3 ± 0.55 | 29.2 ± 0.74 | 0.410 |
CRC in a parent, n (%) | 427 (21) | 270 (21) | 95 (19) | 62 (27) | 0.708 |
Findings at baseline colonoscopy | |||||
No. of adenomas | 2.1 ± 0.03 | 2.0 ± 0.04 | 2.2 ± 0.07 | 2.0 ± 0.09 | 0.491 |
Adenoma burden, cm§ | 1.5 ± 0.03 | 1.5 ± 0.03 | 1.5 ± 0.06 | 1.5 ± 0.07 | 0.917 |
History of cardiovascular disease,∥ n (%) | 292 (14) | 96 (7) | 167 (33) | 29 (13) | <0.001 |
History of hypertension, n (%) | 833 (41) | 463 (36) | 280 (56) | 90 (39) | <0.001 |
History of diabetes, n (%) | 194 (10) | 87 (7) | 85 (17) | 22 (10) | <0.001 |
Use of low-dose aspirin,¶ n (%) | 627 (31) | 294 (23) | 266 (53) | 67 (29) | <0.001 |
Randomized to placebo, n (%) | 676 (33) | 441 (34) | 167 (33) | 68 (30) | |
Randomized to celecoxib, 200 mg, n (%) | 684 (34) | 434 (33) | 169 (34) | 81 (36) | |
Randomized to celecoxib, 400 mg, n (%) | 668 (33) | 423 (33) | 166 (33) | 79 (35) |
*Baseline designates statin use on or before randomization (study treatment start) date.
†Postbaseline designates statin use beginning after randomization (study treatment start) date.
‡P value for comparison between statin nonusers and statin users (baseline and postbaseline groups combined).
§Defined as the sum of the diameters of all adenomas present (in centimeters).
∥Defined as a history of angina, myocardial infarction, coronary artery disease, congestive heart failure, or cerebrovascular disease.
¶Low-dose aspirin is defined as doses of ≤100 mg per day or 325 mg every other day.
Relationship between posttreatment adenoma risk and statin use
This planned secondary analysis investigated the association between statin use and adenoma risk during 3- and 5-year surveillance intervals. First, we determined the relationship between on-study adenoma detection and statin use at any time from baseline to study exit (Table 3). For patients receiving placebo, the unadjusted relative risk (RR) of adenoma detection for statin users was 1.27 [95% confidence interval (95% CI), 1.01-1.58; P = 0.038] over 3 years and 1.28 (95% CI, 1.03-1.59; P = 0.024) over 5 years. Of 221 subjects who used statins and received placebo, 149 (67.4%) had one or more adenomas detected over 5 years. Adjustment for significant covariates by different methods, including propensity analysis, produced a 3-year adenoma risk ranging from 1.17 (95% CI, 0.94-1.47) to 1.29 (95% CI, 1.03-1.62) and a 5-year adenoma risk ranging from 1.19 (95% CI, 0.95-1.48) to 1.31 (95% CI, 1.05-1.63; results not shown). Celecoxib treatment seemed to eliminate this risk based on unadjusted analyses for both 200 mg bid (3-year RR, 0.92; 95% CI, 0.70-1.21) and 400 mg bid (3-year RR, 1.13; 95% CI, 0.86-1.50) doses. The conclusions are consistent after adjustment for aspirin, age, and sex for both 200 mg bid (3-year RR, 0.86; 95% CI, 0.65-1.13) and 400 mg bid (3-year RR, 1.06; 95% CI, 0.79-1.41) doses.
Relationship between statin use and adenoma recurrence
Statin use vs no use . | 3-y surveillance interval* . | 5-y surveillance interval† . | ||
---|---|---|---|---|
RR‡ (95% CI) . | P . | RR‡ (95% CI) . | P . | |
Placebo arm | ||||
RR of statin use vs no use | ||||
Unadjusted | 1.27 (1.01-1.58) | 0.038 | 1.28 (1.03-1.59) | 0.024 |
Multivariable-adjusted for aspirin use, age ≥65 y, and sex | 1.22 (0.97-1.55) | 0.094 | 1.24 (0.99-1.56) | 0.065 |
Celecoxib 200 mg bid arm | ||||
RR of statin use vs no use | ||||
Unadjusted | 0.92 (0.7-1.21) | 0.547 | 0.91 (0.7-1.17) | 0.449 |
Multivariable-adjusted for aspirin use, age ≥65 y, and sex | 0.86 (0.65-1.13) | 0.276 | 0.86 (0.66-1.12) | 0.268 |
Celecoxib 400 mg bid arm | ||||
RR of statin use vs no use | ||||
Unadjusted | 1.13 (0.86-1.5) | 0.387 | 1.02 (0.79-1.31) | 0.889 |
Multivariable-adjusted for aspirin use, age ≥65 y, and sex | 1.06 (0.79-1.41) | 0.701 | 0.96 (0.74-1.24) | 0.752 |
Statin use vs no use . | 3-y surveillance interval* . | 5-y surveillance interval† . | ||
---|---|---|---|---|
RR‡ (95% CI) . | P . | RR‡ (95% CI) . | P . | |
Placebo arm | ||||
RR of statin use vs no use | ||||
Unadjusted | 1.27 (1.01-1.58) | 0.038 | 1.28 (1.03-1.59) | 0.024 |
Multivariable-adjusted for aspirin use, age ≥65 y, and sex | 1.22 (0.97-1.55) | 0.094 | 1.24 (0.99-1.56) | 0.065 |
Celecoxib 200 mg bid arm | ||||
RR of statin use vs no use | ||||
Unadjusted | 0.92 (0.7-1.21) | 0.547 | 0.91 (0.7-1.17) | 0.449 |
Multivariable-adjusted for aspirin use, age ≥65 y, and sex | 0.86 (0.65-1.13) | 0.276 | 0.86 (0.66-1.12) | 0.268 |
Celecoxib 400 mg bid arm | ||||
RR of statin use vs no use | ||||
Unadjusted | 1.13 (0.86-1.5) | 0.387 | 1.02 (0.79-1.31) | 0.889 |
Multivariable-adjusted for aspirin use, age ≥65 y, and sex | 1.06 (0.79-1.41) | 0.701 | 0.96 (0.74-1.24) | 0.752 |
*Cumulative risk of adenoma detection over 3 y, based on results of colonoscopies at year 1 and year 3 following randomization.
†Cumulative risk of adenoma detection over 5 y, based on results of colonoscopies at year 1, year 3, and year 5 following randomization.
‡Relative risk of statin use versus no use is unadjusted or multivariable adjusted.
Next, we considered the duration of statin use before colonoscopy, assuming that once statin use started it was not discontinued. This analysis was informed by data showing that only 10% of all statin users discontinued this drug before study exit, with a median duration of statin use among those discontinuing statins of 1.8 years (range, 0-21.4 years). Duration analysis showed that statin use of >3 years was associated with a significantly increased adenoma risk, with rates among placebo users (adjusted for aspirin use, age, and sex) of 1.37 (95% CI, 1.02-1.86; P = 0.038) over 3 years and 1.39 (95% CI, 1.04-1.86; P = 0.024) over 5 years of surveillance (Table 4). As for the prior analysis, statin use was not associated with an increased risk of adenoma detection among patients using celecoxib. A similar analysis was conducted using advanced adenomas as an end point. The power of this analysis was limited, as the advanced adenoma outcome was observed by 5 years for only 21.3% of placebo users overall (23). The RR for advanced adenoma development, adjusted for aspirin use, age, and sex, was 1.31 over 3 years (95% CI, 0.85-2.0; P = 0.217) and 1.37 over 5 years (95% CI, 0.91-2.05; P = 0.130; Supplementary Table A).
Effect of duration of statin use on adenoma recurrence
. | 3-y surveillance interval* . | 5-y surveillance interval† . | ||
---|---|---|---|---|
RR‡ (95% CI) . | P . | RR‡ (95% CI) . | P . | |
Placebo arm | ||||
Unadjusted | ||||
≤1 y | 1.23 (0.77-1.96) | 0.384 | 1.21 (0.76-1.92) | 0.433 |
1-3 y | 1.03 (0.73-1.46) | 0.856 | 1.05 (0.75-1.48) | 0.781 |
>3 y | 1.47 (1.11-1.94) | 0.007 | 1.49 (1.14-1.95) | 0.004 |
Adjusted for aspirin use, age ≥65 y, sex | ||||
≤1 y | 1.26 (0.79-2.02) | 0.327 | 1.24 (0.78-1.98) | 0.363 |
1-3 y | 1.03 (0.72-1.46) | 0.884 | 1.05 (0.74-1.49) | 0.794 |
>3 y | 1.37 (1.02-1.86) | 0.038 | 1.39 (1.04-1.86) | 0.024 |
Celecoxib 200 mg bid arm | ||||
Unadjusted | ||||
≤1 y | 0.94 (0.54-1.66) | 0.835 | 0.95 (0.55-1.64) | 0.860 |
1-3 y | 0.75 (0.48-1.18) | 0.209 | 0.78 (0.51-1.18) | 0.233 |
>3 y | 1.03 (0.74-1.45) | 0.856 | 0.98 (0.71-1.34) | 0.892 |
Adjusted for aspirin use, age ≥65 y, sex | ||||
≤1 y | 0.9 (0.51-1.59) | 0.713 | 0.92 (0.53-1.59) | 0.770 |
1-3 y | 0.7 (0.44-1.1) | 0.122 | 0.75 (0.49-1.14) | 0.170 |
>3 y | 0.96 (0.67-1.36) | 0.796 | 0.92 (0.66-1.28) | 0.629 |
Celecoxib 400 mg bid arm | ||||
Unadjusted | ||||
≤1 y | 1.34 (0.77-2.33) | 0.303 | 1.26 (0.74-2.14) | 0.399 |
1-3 y | 0.98 (0.62-1.56) | 0.935 | 1 (0.67-1.5) | 0.997 |
>3 y | 1.16 (0.82-1.65) | 0.399 | 0.97 (0.71-1.33) | 0.866 |
Adjusted for aspirin use, age ≥65 y, sex | ||||
≤1 y | 1.26 (0.72-2.2) | 0.414 | 1.2 (0.7-2.04) | 0.514 |
1-3 y | 0.93 (0.59-1.48) | 0.764 | 0.95 (0.63-1.44) | 0.817 |
>3 y | 1.07 (0.75-1.54) | 0.708 | 0.91 (0.66-1.25) | 0.541 |
. | 3-y surveillance interval* . | 5-y surveillance interval† . | ||
---|---|---|---|---|
RR‡ (95% CI) . | P . | RR‡ (95% CI) . | P . | |
Placebo arm | ||||
Unadjusted | ||||
≤1 y | 1.23 (0.77-1.96) | 0.384 | 1.21 (0.76-1.92) | 0.433 |
1-3 y | 1.03 (0.73-1.46) | 0.856 | 1.05 (0.75-1.48) | 0.781 |
>3 y | 1.47 (1.11-1.94) | 0.007 | 1.49 (1.14-1.95) | 0.004 |
Adjusted for aspirin use, age ≥65 y, sex | ||||
≤1 y | 1.26 (0.79-2.02) | 0.327 | 1.24 (0.78-1.98) | 0.363 |
1-3 y | 1.03 (0.72-1.46) | 0.884 | 1.05 (0.74-1.49) | 0.794 |
>3 y | 1.37 (1.02-1.86) | 0.038 | 1.39 (1.04-1.86) | 0.024 |
Celecoxib 200 mg bid arm | ||||
Unadjusted | ||||
≤1 y | 0.94 (0.54-1.66) | 0.835 | 0.95 (0.55-1.64) | 0.860 |
1-3 y | 0.75 (0.48-1.18) | 0.209 | 0.78 (0.51-1.18) | 0.233 |
>3 y | 1.03 (0.74-1.45) | 0.856 | 0.98 (0.71-1.34) | 0.892 |
Adjusted for aspirin use, age ≥65 y, sex | ||||
≤1 y | 0.9 (0.51-1.59) | 0.713 | 0.92 (0.53-1.59) | 0.770 |
1-3 y | 0.7 (0.44-1.1) | 0.122 | 0.75 (0.49-1.14) | 0.170 |
>3 y | 0.96 (0.67-1.36) | 0.796 | 0.92 (0.66-1.28) | 0.629 |
Celecoxib 400 mg bid arm | ||||
Unadjusted | ||||
≤1 y | 1.34 (0.77-2.33) | 0.303 | 1.26 (0.74-2.14) | 0.399 |
1-3 y | 0.98 (0.62-1.56) | 0.935 | 1 (0.67-1.5) | 0.997 |
>3 y | 1.16 (0.82-1.65) | 0.399 | 0.97 (0.71-1.33) | 0.866 |
Adjusted for aspirin use, age ≥65 y, sex | ||||
≤1 y | 1.26 (0.72-2.2) | 0.414 | 1.2 (0.7-2.04) | 0.514 |
1-3 y | 0.93 (0.59-1.48) | 0.764 | 0.95 (0.63-1.44) | 0.817 |
>3 y | 1.07 (0.75-1.54) | 0.708 | 0.91 (0.66-1.25) | 0.541 |
*Cumulative risk of adenoma detection over 3 y, based on results of colonoscopies at year 1 and year 3 following randomization.
†Cumulative risk of adenoma detection over 5 y, based on results of colonoscopies at year 1, year 3, and year 5 following randomization.
‡Relative risk of statin use duration versus no use, unadjusted or multivariable adjusted.
Relationship between statin use and cardiovascular events
Previous analyses of the APC trial safety database, using a treatment-emergent approach, showed that 3.8% of placebo users experienced one or more cardiovascular and thrombotic adverse events, compared with 6.0% (RR, 1.6; 95% CI, 1.0-2.5) and 7.5% (RR, 1.9; 95% CI, 1.2-3.1) of those using celecoxib 200 mg bid and celecoxib 400 mg bid, respectively (23). In addition, a baseline history of atherosclerotic heart disease was significantly associated with celecoxib dose and cardiovascular event risk (Pinteraction = 0.004). In this study, we determined the relationship between statin use at baseline and cardiovascular adverse events, using the same definition as the previous analysis (i.e., a combined end point including myocardial infarction, cardiovascular therapeutic procedure, cerebrovascular disease, peripheral vascular disease, peripheral vascular therapeutic procedure, venous thrombosis or thromboembolism, and death or circulatory collapse due to cardiovascular causes; Table 5). Events were included if they occurred during the time from first study medication dose to 30 days after the last study medication dose. Because the majority of patients with year 5 study data had been off study medication for more than a year, we considered this treatment-emergent safety analysis to be the most accurate reflection of drug-associated toxicity. Of 221 patients who used statins and received placebo, 16 (7.2%) experienced cardiovascular adverse events. With this limited number of events, analysis was adjusted for aspirin use and showed that statin use was associated with an increased risk of cardiovascular adverse events (RR, 2.53; 95% CI, 1.11-5.74; P = 0.027). This analysis failed to show an interaction with celecoxib use (P = 0.366).
Relationship between statin use and cardiovascular adverse events
Patients experiencing cardiovascular adverse events* . | Placebo (n = 676) . | Celecoxib 200 mg bid (n = 684) . | Celecoxib 400 mg bid (n = 668) . | ||||
---|---|---|---|---|---|---|---|
Crude n (%) . | Crude n (%) . | Crude n (%) . | |||||
All patients (n = 2,028) | 26/676 (3.9) | 41/684 (6.0) | 50/668 (7.5) | ||||
Statin users† (n = 730) | 16/235 (6.8) | 25/250 (10.0) | 34/245 (13.9) | ||||
Statin nonusers (n = 1,298) | 10/441 (2.3) | 16/434 (3.7) | 16/423 (3.8) | ||||
Statin use vs nonuse | RR (95% CI)‡ | P‡ | RR (95% CI) | P | RR (95% CI) | P | P interaction§ |
Unadjusted | 3.14 (1.44-6.85) | 0.004 | 1.45 (0.77-2.73) | 0.245 | 2.27 (1.30-3.96) | 0.004 | 0.341 |
Adjusted for aspirin use | 2.53 (1.11-5.74) | 0.027 | 1.29 (0.68-2.47) | 0.438 | 2.22 (1.25-3.93) | 0.006 | 0.366 |
Patients experiencing cardiovascular adverse events* . | Placebo (n = 676) . | Celecoxib 200 mg bid (n = 684) . | Celecoxib 400 mg bid (n = 668) . | ||||
---|---|---|---|---|---|---|---|
Crude n (%) . | Crude n (%) . | Crude n (%) . | |||||
All patients (n = 2,028) | 26/676 (3.9) | 41/684 (6.0) | 50/668 (7.5) | ||||
Statin users† (n = 730) | 16/235 (6.8) | 25/250 (10.0) | 34/245 (13.9) | ||||
Statin nonusers (n = 1,298) | 10/441 (2.3) | 16/434 (3.7) | 16/423 (3.8) | ||||
Statin use vs nonuse | RR (95% CI)‡ | P‡ | RR (95% CI) | P | RR (95% CI) | P | P interaction§ |
Unadjusted | 3.14 (1.44-6.85) | 0.004 | 1.45 (0.77-2.73) | 0.245 | 2.27 (1.30-3.96) | 0.004 | 0.341 |
Adjusted for aspirin use | 2.53 (1.11-5.74) | 0.027 | 1.29 (0.68-2.47) | 0.438 | 2.22 (1.25-3.93) | 0.006 | 0.366 |
*These numbers are termed “crude” because the denominators are not known at baseline. The denominators are the final number of statin users or nonusers within each treatment arm at the end of the study.
†Defined as beginning statin use at any time before end of study (baseline + postbaseline use).
‡Risk ratio and P value for comparison between statin users and statin nonusers.
§P value for interaction between statin use and celecoxib treatment.
Discussion
This study represents the largest report of statin use from a prospective randomized trial of patients at high CRC risk. The results show that, at best, statins did not protect against the development of sporadic colorectal adenomas. In fact, even after controlling for multiple covariates, statin users of >3 years on the placebo arm showed a 40% increase in adenoma detection during 5 years of surveillance. Celecoxib use produced significant antitumor effect, which seemed to counteract the tumor-promoting effect of statins. The lack of significant effect of statins against an advanced adenoma end point was likely due to the small number of patients with this outcome. The APC trial showed an increased cardiovascular risk associated with celecoxib use, particularly for the higher dose (22), with a significant association between a baseline history of cardiovascular disease and celecoxib-associated cardiovascular toxicity (23). As expected therefore, because statin use and cardiovascular disease are strongly linked, baseline statin use in participants also emerged as a significant factor in identifying celecoxib-associated cardiovascular risk.
Several factors make these data convincing. First, the APC trial was designed at the outset to assess the effect of concomitant medications, including statins, on study end points. Because of this, detailed medication use histories that included prerandomization treatment duration were obtained from all participants and were updated at 6-month intervals during the 3-year treatment phase of the trial. Second, the APC trial began before the cardiovascular risks of celecoxib were evident. As a result, subjects with significant cardiovascular risk were enrolled on the trial and a substantial proportion of participants used statins, allowing sufficient numbers within the placebo arm to control for important covariates such as age, sex, and cardioprotective aspirin use. Finally, the type of statin used may influence tissue response. Lipophilic statins achieve higher drug levels in nonhepatic tissues and are theoretically more likely to alter the biology of the colorectal mucosa than hydrophilic statins, which are hepatoselective. Of the participants using statins, 87.1% used lipophilic agents; therefore, the results should reflect significant drug effects in the target tissue.
There are some factors limiting the confidence of these results. Although the distribution of treatment assignment was equal for statin users and nonusers and analyses were adjusted for codependent variables, the randomization to treatment for this trial was not stratified for statin use, and it is possible that imbalances affecting the outcome were introduced. In addition, selection for study entry required the presence of a colorectal adenoma at a baseline clearance colonoscopy. As a result, it is possible that patients who entered the study using statins constituted a population preselected to develop adenomas while on statins. We saw a similar effect in the overall APC trial when aspirin use was considered. All of the aspirin users on the APC trial were routinely taking aspirin before study entry and were required to remain on aspirin during the duration of the trial. The study showed that aspirin users had the same overall rate of adenoma recurrence as nonaspirin users on the placebo arm, and we did not see an additive or synergistic effect with celecoxib for the efficacy end points (19). We propose that this occurred because patients on the APC trial who used aspirin were resistant to the beneficial antitumor effects of this nonsteroidal anti-inflammatory drug. An additional minor issue is that statin use was discontinued during the APC trial surveillance period in 10% of users, and the duration analyses that we performed did not correct for this. This assumption, however, should decrease the risk associated with statin use. Finally, these data do not clearly indicate that statin use promotes adenoma development. An increased risk of adenoma development (RR ∼1.27-1.37) does seem to be the case by the overall unadjusted analysis and for both the unadjusted and multivariable analyses in the subset of patients using statins for more than 3 years. However, the significance at both 3- and 5-year end points is lost in the multivariable analysis for all placebo users when adjusted for important covariates of aspirin use, age, and sex, and it is possible that the significance in the duration subset analysis is due to small sample size.
In conclusion, we find no evidence that statin use decreases colorectal adenoma incidence. On the contrary, our data show that long-term statin use may increase the risk of these CRC precursor lesions, although this result is interpreted with caution due to the observational nature of this study.
Appendix 1
The following persons participated in the APC Study: Steering Committee: M.M. Bertagnolli, E.T. Hawk, C.J. Eagle; Statistical Team: A.G. Zauber, K.M. Kim, D. Corle, R. Rosenstein, J. Tang, T. Hess, A. Wilton, M. Hsu; Medical Monitors: W. Anderson, L. Doody; Central Pathology Review: M. Redston, K.R. Geisinger; Project Directors: G.M. Woloj, D. Bagheri, A. Crawford, M. Schietrum, V. Ladouceur; Data and Safety Monitoring Board: S. Rosen (chair), L. Friedman, R. Makuch, R. Phillips, P. Taylor; Principal Investigators, United States: S. Auerbach (California Professional Research, Newport Beach), C.F. Barish (Wake Research Associates, Raleigh, NC), T. Barringer (Carolinas Medical Center, Charlotte, NC), R.W. Bennetts (Northwest Gastroenterology Clinic, Portland, OR), M. Blitstein (Associates in Gastroenterology and Liver Disease, Lake Forest, IL), J. Bruggen (Wake Forest University Baptist Medical Center, Winston Salem, NC), P Carricaburu (Veterans Affairs Hospital, Sheridan, WY), D. Chung (Massachusetts General Hospital, Boston, MA), F. Colizzo (Pentucket Medical Associates, Haverhill, MA), R. Curtis (Newton-Wellesley Hospital, Newton, MA), T. Dewar (Harris Methodist Hospital Fort Worth, Ft. Worth, TX), R. DuBois (Vanderbilt University Medical Center, Nashville, TN), T. Feinstat (Gastroenterology Consultants of Sacramento, Roseville, CA), T.R. Foley (Regional Gastroenterology Associates of Lancaster, Lancaster, PA, D. Gabbaizadeh (Huntington Research Group, Huntington Station, NY), J. Geenen (Wisconsin Center for Advanced Research, Milwaukee, WI), F. Giardiello (Johns Hopkins Hospital, Baltimore, M.D.), A. Goetsch (nTouch Research, Huntsville, AL), M. Goldberg (Regional Gastroenterology Associates of Lancaster, Evanston, IL), J.L. Goldstein (University of Illinois at Chicago, Chicago, IL), W. Harlan, III (Asheville Gastroenterology Associates, Asheville, NC), R. Hogan (Gastrointestinal Associates, Jackson, MS), M. Kamionkowski (Gastroenterology Associates of Cleveland, Mayfield Heights, OH), M. Kelfer (Fallon Clinic, West Boylston, MA), B. Kerzner (Health Trends Research, Baltimore, MD), K. Kim (University of Chicago Medical Center, Chicago, IL), I. Klimberg (Gastroenterology Associates of Ocala, Ocala, FL), G. Koval (West Hills Gastroenterology Associates, Portland, OR), C. Krone (Advanced Clinical Therapeutics, Tucson, AZ), S. Krumholz (Waterside Clinical Research, West Palm Beach, FL), M.W. Layton (South Puget Sound Clinical Research Center, Olympia, WA), C. Lightdale (Columbia-Presbyterian Medical Center, New York, NY), P.J. Limburg (May Clinic, Rochester, MN), C. Lind (Vanderbilt University Medical Center, Nashville, TN), D. Lipkis (Institute for Health Care Assessment, San Diego, CA), M. Lloyd (Idaho Gastroenterology, Meridian, ID), D. Maccini (Spokane Digestive Disease Center, Spokane, WA), F. MacMillan, Sr. (Pentucket Medical Associates, Haverhill, MA), R. Madoff (University of Minnesota, Minneapolis, MN), A. Malik (Advanced Clinical Research, North Providence, RI), A. Markowitz (Memorial Sloane-Kettering Cancer Center, New York, NY), R. Marks (Alabama Digestive Research Center, Alabaster, AL), C.J. McDougall (Manhattan Associates, New York, NY), P. Miner (Oklahoma Foundation for Digestive Research, Oklahoma City, OK), M. Murphy (Southern Digestive and Liver Disease Institute, Savannah, GA), A. Namais (Gastrointestinal Physicians, Salem, MA), N. Nickl (University of Kentucky Medical Center, Lexington, KY), M. Pochapin (Jay Monahan Center for Gastrointestinal Health, New York, NY), R.E. Pruitt (Nashville Medical Research Institute, Nashville, TN), J Puolos (Cumberland Research Associates, Fayetteville, NC), D.S. Riff (AGMG Clinical Research, Anaheim, CA), R. Roman (South Denver Gastroenterology, Englewood, CO), L. Rubin (New Jersey Physicians, Passaic, NJ), D. Ruff (Healthcare Discoveries, San Antonio, TX), M. Safdi (Consultants for Clinical Research, Cincinnati, OH), J. Saltzman (Brigham and Women's Hospital, Boston, MA), B. Salzberg (Atlanta Gastroenterology Associates, Atlanta, GA), J.A. Sattler (Western Clinical Research, Torrence, CA), P. Schleinitz (Americas Doctors Research, Medford, OR), J. Schwartz (Northwest Gastroenterologists, Arlington Heights, IL), M. Schwartz (Jupiter Research Association, Jupiter, FL), M. Silpa (Gastroenterology Associates of the East Bay Medical Group, Berkeley, CA), D. Silvers (Drug Research Services, Metairie, LA), D. Smoot (Howard University Cancer Center, Washington, DC), S. Sontag (Veterans Affairs Medical Center, Hines, IL), R.J. Sorrell (Gastroenterology Specialties, Lincoln, NE), D. Stanton (Community Clinical Trials, Orange, CA), J. Sturgeon (Americas Doctors Research, Shawnee Mission, KS), J.P. Tracey (Hawthorne Medical Associates, North Dartmouth, MA), T. Werth (Charlotte Gastroenterology and Hepatology, Charlotte, NC), C.M. Wilcox (University of Alabama at Birmingham, Birmingham, AL), R. Wohlman (Northwest Gastroenterology Associates, Bellevue, WA), S. Woods (Gastroenterology Associates of Fairfield County, Bridgeport, CT); United Kingdom: J. Burn (South Cleveland Hospital, Middlesbrough); Australia: H. Ee (Sir Charles Gairdner Hospital, Nedlands, W.A.), M. Korman (Monash Medical Centre, Clayton, Victoria), A. Lee (Concord Repatriation and General Hospital, Concord, NSW), B. Leggett (Royal Brisbane Hospital, Herston, Queensland), F. Macrae (Royal Melbourne Hospital, Melbourne, Victoria), L. Mollison (Freemantle Hospital, Freemantle, WA), N. Yeomans (Western Hospital, Footscray, Victoria), G. Young (Flinders Medical Center, Bedford, SA); Canada: G. Aumais (Hospital Maisonneuve-Rosemont, Montreal), R. Bailey (Hys Medical Center, Edmonton, Alberta), C. Bernstein (Winnipeg Health Sciences Centre, Winnipeg, Manitoba), L. Cohen (Sunnybrook and Women's Hospital, Toronto), C. Dallaire, R. Dube (Centre Hospitalier Universitaire de Quebec, Quebec), D. Morgan (McMaster University, Hamilton, Ontario), T. Sylwestrowicz (St. Paul's Hospital, Saskatoon, Saskatoon), G. Van Rosendaal (University of Calgary, Alberta), S.J. Van Zantan (Queen Elizabeth II Health Sciences Centre, Halifax, NS).
Disclosure of Potential Conflicts of Interest
M.M. Bertagnolli, A.G. Zauber, and M. Hsu received research funding from Pfizer, Inc., as well as the National Cancer Institute (CN-95015) for this study. C.J. Eagle is an employee of Pfizer, Inc. E.T. Hawk disclosed no potential conflicts of interest.
Acknowledgments
Grant Support: National Cancer Institute grant CA-N01-95015 (Monica Bertagnolli, Principal Investigator) and Pfizer, Inc.