Abstract
Inflammation, as measured by the circulating inflammatory marker high-sensitivity C-reactive protein (hsCRP), has been associated with cardiovascular disease. However, data about CRP and risk of colorectal cancer have been conflicting. The Adenoma Prevention with Celecoxib (APC) trial showed that the anti-inflammatory drug celecoxib prevents recurrence of colorectal adenoma but increases risk of cardiovascular events. We examined whether serum hsCRP modified these results. We measured hsCRP from serum specimens provided at study entry by patients enrolled in the APC trial. Patients were stratified according to use of low-dose aspirin, randomized to receive 3 years of treatment with placebo, 200-mg-bid celecoxib, or 400-mg-bid celecoxib, and underwent follow-up colonoscopies at years 1 and 3. Among 1,680 patients, the estimated 3-year cumulative incidence of adenoma was 42% for patients with hsCRP <1 mg/L, compared with 43% [relative risk (RR) = 1.02; 95% CI = 0.85–1.22] for hsCRP 1–3 mg/L, and 41% (RR = 1.10; 95% CI = 0.90–1.34) for hsCRP >3 mg/L. The effect of celecoxib on adenoma recurrence did not vary among patients with high (>3 mg/L) compared with low (≤3 mg/L) hsCRP. However, among patients with high hsCRP, the RR of cardiovascular events compared with placebo was 2.27 (95% CI = 0.72–7.14) for those randomized to celecoxib 200-mg-bid and 3.28 (95% CI = 1.09–9.91) for 400-mg-bid. In contrast, among patients with low hsCRP, the corresponding RRs were 0.99 (95% CI = 0.53–1.83) and 1.11 (95% CI = 0.61–2.02). hsCRP may predict risk of celecoxib-associated cardiovascular toxicity but not adenoma recurrence or celecoxib treatment efficacy. Patients with low hsCRP may be a subgroup with a favorable risk-benefit profile for celecoxib chemoprevention. Cancer Prev Res; 4(8); 1172–80. ©2011 AACR.
See perspective on p. 1145
Introduction
Aspirin and selective COX-2 inhibitors, such as celecoxib, reduce risk of colorectal adenoma and cancer (1–5). This effect may be mediated through abrogation of inflammation (6–8). High-sensitivity C-reactive protein (hsCRP) is a circulating inflammatory biomarker of chronic conditions including cardiovascular disease (9, 10). However, prospective studies relating hsCRP to risk of colorectal cancer and adenoma have been equivocal (11–29).
In the Adenoma Prevention with Celecoxib (APC) trial, patients who had recently undergone colonoscopic removal of an adenoma were randomly assigned to receive placebo, 200 mg twice daily (200-mg-bid) of celecoxib, or 400 mg twice daily (400-mg-bid) of celecoxib and underwent follow-up colonoscopies at 1 and 3 years. The relative risk (RR) of the detection of 1 or more new adenomas by year 3 compared with placebo was 0.67 (95% CI = 0.59–0.77) for those receiving 200-mg-bid celecoxib and 0.55 (95% CI = 0.48–0.64) for those receiving 400-mg-bid celecoxib (4). Unfortunately, in a separate, adjudicated safety analysis, the APC trial also revealed unexpected dose-related cardiovascular toxicity (30). Because hsCRP may be related to both risk of neoplasia and cardiovascular events and celecoxib has been shown to reduce hsCRP levels (31, 32), we examined baseline hsCRP in relation to (i) risk of recurrent adenoma, (ii) celecoxib-related chemopreventive efficacy, and (iii) celecoxib-related cardiovascular toxicity.
Methods
Study population
Jointly sponsored by the National Cancer Institute and Pfizer, Inc., the APC trial was a randomized, placebo-controlled trial which enrolled patients within 6 months of colonoscopic removal of multiple adenomas or a single adenoma larger than 5 mm in diameter (ClinicalTrials.gov NCT00005094; ref. 4). Beginning in November 1999, 2,457 potential participants at 91 clinical sites were entered into a 30-day placebo run-in period during which they were required to have at least 80% adherence to medication use. After the run-in period, 2,035 patients were subsequently randomly assigned to placebo, 200-mg-bid of celecoxib, or 400-mg-bid of celecoxib. Randomization was stratified on the basis of the use or nonuse of low-dose aspirin (≤325 mg every other day or ≤162.5 mg every day) and clinical site. For the duration of the study, patients were required to abstain from long-term use of non–steroidal anti-inflammatory drugs. Patients were excluded if they had a history of familial adenomatous polyposis, hereditary nonpolyposis colon cancer, inflammatory bowel disease, or large-bowel resection other than appendectomy. Other exclusion criteria included a history of a renal or hepatic disorder, a clinically significant bleeding disorder, or treatment for a gastrointestinal ulcer before study entry. Study drug treatment was initially planned for 3 years for all participants. However, at the recommendation of the APC trial Data Safety Monitoring Board (DSMB), treatment was terminated prematurely on December 17, 2004, based on the results of an unscheduled safety analysis conducted by an independent cardiovascular safety committee. At that time 1,762 patients (86.6%) had completed 3 years of treatment and 273 patients had 1 to 3 months of treatment remaining. In addition, 639 patients had begun participation in the extension study in which study medication was continued in a blinded manner for an additional 2 years. The median duration of treatment exposure in the extension study was 3.5 years (33). All patients provided written informed consent and the human subjects committee at each site approved the study. This analysis was approved by the Human Subjects Committee of Partners HealthCare.
hsCRP measurements
At baseline randomization, 1,707 participants provided a serum specimen at baseline randomization which was subsequently stored at −70°C. Personnel blinded to quality control and outcome data assayed for hsCRP using a high-sensitivity, latex-enhanced immunoturbidimetric assay (intra-assay coefficient of variation = 2.9%; Denka Seiken). Among the 1,707 specimens, 27 could not be measured for technical reasons (Fig. 1). Thus, this analysis included 1,680 participants, which were similar according to baseline characteristics to those for whom we did not measure hsCRP (data not shown).
Flow of patients through the study. Patients who violated study entry criteria were those for whom the presence or absence of adenoma on the baseline colonoscopy could not be confirmed. Adherence to the use of study medication was calculated as duration of use in days, divided by 1,095.
Outcome ascertainment
A study investigator conducted follow-up colonoscopies with endoscopic removal of polyps at 1 and 3 years after randomization. A central study pathologist examined, in a blinded fashion, all polyps removed during these colonoscopies. Adverse events were reported by investigators and classified according to criteria from the Medical Dictionary for Regulatory Activities (MedDRA), version 8.1 (4, 33).
Statistical analysis
As in prior studies (21), we examined hsCRP levels according to cutoff points proposed in clinical guidelines (<1, 1–3, and >3 mg/L; ref. 34). Consistent with the intent-to-treat principle, we used the detection of an adenoma during a postrandomization colonoscopy, regardless of whether the patient adhered to the treatment regimen, as the primary endpoint. We estimated the cumulative incidence of adenoma at 3 years within different subgroups using Kaplan–Meier method. The effect of hsCRP levels on having a recurrent adenoma at a postrandomization colonoscopy was estimated by RR derived from Cox proportional hazards modeling with ties handled by the exact method, stratified by age (≥65 vs. <65), sex, time, baseline use of low-dose aspirin (≤325 mg/every other day or ≤162.5 mg/d), postmenopausal hormones, and duration of statin use (≤3 vs. >3 y) as a time-dependent variable (35). We used Cox proportional hazards models, adjusted for the same variables as the efficacy analyses, to estimate the RR of an investigator-reported adverse event after the first dose and up to 30 days after the last dose of study medication including events among patients who continued study medication in the 24-month extension study (33, 35). We used the SAS version 9.1 (SAS Institute) for all analyses. All significance tests were 2 sided at a 5% level of significance.
Results
Among the 1,680 participants, the median age was 59 [range = 31–88] years, 92% were white, and 68% were men, and the median hsCRP level was 1.6 mg/L (range = 0.1–2.72 mg/L). Patients with elevated CRP levels (>3 mg/L) more frequently smoked, had a higher body mass index, had a prior history of cardiovascular events, hypertension, and diabetes mellitus, and used low-dose aspirin and statin drugs. Consistent with the known effect of postmenopausal hormones on CRP levels (36), the prevalence of postmenopausal hormone use was higher among women with elevated CRP. The number of adenomas (P = 0.24) or adenoma burden (sum of diameter of all adenomas; P = 0.61) at the baseline qualifying examination did not vary according to hsCRP (Table 1).
Baseline characteristics of the patients according to serum hsCRP levelsa
Baseline hsCRP was not statistically significantly associated with risk of recurrent colorectal adenoma. The RR of adenoma through year 3 associated with a 1 SD increase in log hsCRP was 0.96 (95% CI = 0.89–1.03). The estimated 3-year cumulative incidence of adenoma was 42% for patients with hsCRP <1 mg/L, compared with 43% (RR = 1.02; 95% CI = 0.85–1.22) for hsCRP 1–3 mg/L, and 41% (RR = 1.1; 95% CI = 0.90–1.34) for hsCRP >3 mg/L. These results did not vary according to strata defined by celecoxib assignment, aspirin use, sex, use of postmenopausal hormones, body mass index, or use of statin drugs (Table 2). When we analyzed data according to quartile cutoff points of the distribution of hsCRP, our results were essentially unchanged (data not shown). We also examined hsCRP in relation to risk of advanced adenoma. The estimated 3-year cumulative incidence of advanced adenoma was 8.2% for patients with hsCRP <1 mg/L, compared with 7.5% (RR = 0.90; 95% CI = 0.59–1.38) for hsCRP 1–3 mg/L, and 5.6% (RR = 0.72; 95% CI = 0.44–1.17) for hsCRP >3 mg/L.
Risk of adenoma according to serum hsCRP levels
The overall reduction in cumulative incidence of adenoma associated with celecoxib did not appear to vary among patients with high (>3 mg/L) compared with low (≤3 mg/L) hsCRP (Table 3). However, among patients with high hsCRP, the RR of cardiovascular events compared with placebo was 2.27 (95% CI = 0.72–7.14) for those randomized to celecoxib 200-mg-bid and 3.28 (95% CI = 1.09–9.91) for 400-mg-bid. In contrast, among patients with low hsCRP, the corresponding RRs were 0.99 (95% CI = 0.53–1.83) and 1.11 (95% CI = 0.61–2.02; Table 4; Fig. 2). A formal test of interaction between high hsCRP and 400-mg-bid celecoxib approached statistical significance (P = 0.11). We considered the possibility that concurrent aspirin use may modify the interaction between high hsCRP levels and celecoxib-associated cardiovascular events. Among patients with hsCRP >3 mg/L and were not taking low-dose aspirin, the RR for a cardiovascular event compared to those randomized to placebo was 2.74 (95% CI = 0.54–13.70) for those randomized to celecoxib 200-mg-bid and 4.98 (95% CI = 1.10–22.59) for those randomized to celecoxib 400-mg-bid. In contrast, among patients with hsCRP >3 mg/L and were taking low-dose aspirin, the RR for a cardiovascular event compared with those randomized to placebo was 1.72 (95% CI = 0.33–8.96) for those randomized to celecoxib 200-mg-bid, and 1.65 (95% CI = 0.30–9.0) for those randomized to celecoxib 400-mg-bid. Although these results do suggest a possible attenuation of cardiovascular risk associated with low-dose aspirin, a formal test of interaction between celecoxib treatment and use of low-dose aspirin was not statistically significant (P = 0.57). Moreover, because there were limited numbers of events within each aspirin strata (19 events in the group not taking aspirin and 12 events in the group taking low-dose aspirin), these results should be considered exploratory. Finally, among all patients, there did not appear to be an overall dose-related increase in either renal and hypertensive disorders or gastrointestinal ulceration and hemorrhage, consistent with the overall results of the trial (4).
Relative risk of cardiovascular events according to baseline hsCRP. Error bars correspond to 95% CIs.
Risk of adenoma according to celecoxib treatment, stratified by hsCRP levela
Risk of adverse events according to celecoxib treatment, stratified by hsCRP levela
Discussion
The divergent results of previous studies relating hsCRP to colorectal adenoma or cancer may be related to the timing at which hsCRP was measured (11–29). Specifically, positive studies may reflect hsCRP elevations associated with occult disease including adenoma (11, 12, 14, 22). In support of this explanation, cross-sectional data measuring hsCRP at the time of colonoscopy have shown a modest association with prevalent adenoma (24, 25), whereas a prospective study of incident adenoma and a cross-sectional study measuring hsCRP several years before diagnosis of adenoma showed inverse or no association (26, 28). In our study, because we examined hsCRP levels among individuals within 6 months of a clearing colonoscopy and endpoints were ascertained at protocol-defined surveillance colonoscopies, our results more closely reflect the association between hsCRP and incident neoplasia. Our findings are largely consistent with those of the similarly designed Aspirin Polyp Prevention Study (PPS; ref. 37). However, the results from both the APC and Aspirin PPS trials do not exclude a potential association between hsCRP and the initial development of adenoma rather than recurrence. A limitation of our study is that we only had a single baseline measure of hsCRP and we could not correlate change in hsCRP with celecoxib treatment. However, celecoxib treatment reduced adenoma recurrence irrespective of baseline levels of hsCRP. Taken together with data from the Aspirin PPS which did measure hsCRP levels at baseline and at year 3, our findings support the conclusion that hsCRP does not mediate the chemopreventive effect of aspirin or celecoxib (37).
Among individuals with high hsCRP levels (>3 mg/L), celecoxib treatment was associated with a 3-fold higher risk of an adverse cardiovascular event; in contrast, among those with low hsCRP (≤3 mg/L), celecoxib treatment was not associated with higher risk. Although a formal test for interaction was not significant, this is likely due to the limited number of adverse cardiovascular events. These findings are consistent with emerging data that individuals can be stratified for celecoxib cardiovascular toxicity according to baseline cardiovascular risk. A prior pooled analysis of 6 placebo-controlled trials of celecoxib observed that patients with high baseline cardiovascular risk scores had the greatest risk of celecoxib-related adverse events (38). Similarly, a 5-year safety analysis of an extension of the APC trial showed a significant interaction between a baseline history of atherosclerotic heart disease and risk of celecoxib-associated cardiovascular events (33). Further studies are needed to determine whether baseline hsCRP alone can be used to identify patients for whom long-term use of celecoxib is relatively safe.
In the APC trial, patients were randomized to either 200-mg-bid or 400-mg-bid doses of celecoxib. Thus, it is unclear if baseline hsCRP may also predict risk of cardiovascular events among patients who take celecoxib once daily. In a parallel randomized, placebo-controlled trial, treatment with 400 mg of daily celecoxib was associated with an overall nonsignificant increase in risk of cardiovascular events (RR = 1.30; 95% CI = 0.65–2.62; ref. 5). Thus, it is possible that use of a once daily dosing of celecoxib among patients with low hsCRP may be a particularly effective strategy to minimize cardiovascular risk.
Finally, in our analysis, we did observe a nonsignificant increase in risk of renal and hypertensive events in the subgroup of patients with low hsCRP randomized to 200-mg-bid. In contrast, there was no association among those randomized to 400-mg-bid. On the basis of this lack of a dose response, it is unlikely that there is a unique biological interaction between celecoxib and low baseline inflammatory state and renal or hypertensive events. Nonetheless, larger studies with a greater number of such endpoints are needed to make definitive conclusions.
In this large, randomized, placebo-controlled trial, baseline hsCRP was not associated with overall risk of adenoma recurrence or celecoxib chemopreventive benefit after 3 years of treatment. However, individuals with high hsCRP appeared to have the greatest risk of celecoxib-related cardiovascular toxicity. Further studies are needed to determine the role of hsCRP in relation to other markers of cardiovascular risk to evaluate the risk-benefit profile of celecoxib treatment for a range of clinical indications.
Disclosure of Potential Conflicts of Interest
A.T. Chan is a Damon Runyon Cancer Research Foundation Clinical Investigator and has served as a consultant to Bayer HealthCare. M.M. Bertagnolli is the recipient of research funding from Pfizer Inc. E.T. Hawk has served as a consultant for Pozen Pharmaceutical Development Company. P.M. Ridker is listed as a coinventor on patents held by Brigham and Women's Hospital that relate to the use of inflammatory biomarkers in cardiovascular disease and diabetes that have been licensed to Siemens and AstraZeneca. C.S. Sima has no conflicts of interest. The statistical analysis of the entire data sets pertaining to efficacy and safety has been independently confirmed by A.G. Zauber, who is not employed by any corporate entity. The corresponding author had full access to all of the data and takes full responsibility for the veracity of the data and analysis.
Grant Support
This work was supported by the National Cancer Institute at the NIH (grant number R01 CA137178 to A.T. Chan and N01 CN95015 to M.M. Bertagnolli).
Acknowledgments
The authors thank Dr. Nader Rifai and Mr. Gary Bradwin for technical assistance with the CRP assays.
Footnotes
Note: Portions of this data were previously presented in abstract form at Digestive Disease Week in Chicago, IL, on May 30–June 4, 2009.
- Received December 30, 2010.
- Revision received February 18, 2011.
- Accepted March 29, 2011.
- ©2011 American Association for Cancer Research.
References
Volume 4, Issue 8
