|
|
Background Regular use of aspirin reduces the risk of a colorectal neoplasm, but the mechanism by which aspirin affects carcinogenesis in the colon is not well understood.
Methods We estimated cyclooxygenase-2 (COX-2) expression by immunohistochemical assay of sections from paraffin-embedded colorectal-cancer specimens from two large cohorts of participants who provided data on aspirin use from a questionnaire every 2 years. We applied Cox regression to a competing-risks analysis to compare the effects of aspirin use on the relative risk of colorectal cancer in relation to the expression of COX-2 in the tumor.
Results During 2,446,431 person-years of follow-up of 82,911 women and 47,363 men, we found 636 incident colorectal cancers that were accessible for determination of COX-2 expression. Of the tumors, 423 (67%) had moderate or strong COX-2 expression. The effect of aspirin use differed significantly in relation to COX-2 expression (P for heterogeneity=0.02). Regular aspirin use conferred a significant reduction in the risk of colorectal cancers that overexpressed COX-2 (multivariate relative risk, 0.64; 95% confidence interval [CI], 0.52 to 0.78), whereas regular aspirin use had no influence on tumors with weak or absent expression of COX-2 (multivariate relative risk, 0.96; 95% CI, 0.73 to 1.26). The age-standardized incidence rate for cancers that overexpressed COX-2 was 37 per 100,000 person-years among regular aspirin users, as compared with 56 per 100,000 person-years among those who did not use aspirin regularly; in contrast, the rate for cancers with weak or absent COX-2 expression was 27 per 100,000 person-years among regular aspirin users, as compared with 28 per 100,000 person-years among nonregular aspirin users.
Conclusions Regular use of aspirin appears to reduce the risk of colorectal cancers that overexpress COX-2 but not the risk of colorectal cancers with weak or absent expression of COX-2.
B,15 induction of apoptosis by activation of p38 kinase,16 and catabolism of polyamines.17 If aspirin exerts its effect on the formation of adenomas and cancers by inhibiting COX-2 or its downstream effectors, then the use of aspirin should preferentially reduce the risk of tumors for which growth depends on COX-2 function. We investigated whether the influence of aspirin on the risk of colorectal cancer varied with the expression of COX-2 in the tumor. For this purpose, we used tumor specimens from two large prospective studies in which an association was found between the regular use of aspirin and a reduced risk of colorectal cancer.18,19,20
Methods
Study Population
The Nurses' Health Study (NHS) was initiated in 1976 when 121,701 U.S. female registered nurses, 30 through 55 years of age, completed a mailed questionnaire.19,20 The Health Professionals Follow-up Study (HPFS) was established in 1986 with a parallel cohort of 51,529 U.S. male dentists, optometrists, osteopaths, podiatrists, pharmacists, and veterinarians who were 40 through 75 years of age at entry.18 The cohorts had a follow-up rate of 92%. We mailed questionnaires every 2 years to obtain updated information and identify new cases of cancer. In 1980, the NHS questionnaire was expanded to include a validated assessment of diet and aspirin use21; a similar instrument had been used in the 1986 HPFS questionnaire.22 The overall response rate for these questionnaires was 90%. The institutional review boards at Brigham and Women's Hospital and the Harvard School of Public Health approved this study; completion of the questionnaire was considered to imply informed consent.
Assessment of Medication Use
Assessment of aspirin use in both the NHS and the HPFS has been described in detail previously.18,19,20,23 Briefly, in 1980 we asked NHS participants whether they used aspirin, the number of pills taken each week, and the number of years of use. We updated this information every 2 years except for 1986 with specific questions on the number of aspirin tablets taken per week (in categories of number taken). In the 1986 HPFS questionnaire and in questionnaires every 2 years thereafter, we inquired whether the men in the study used aspirin two or more times per week. Beginning in 1992, we also asked these men the average number of tablets taken per week (in categories of number taken). In both cohorts, we specifically inquired about standard-dose (325-mg) aspirin tablets. However, to reflect secular trends in consumption of low-dose (baby) aspirin, the questionnaires after 1992 asked participants to convert four baby aspirin tablets to one adult standard-dose tablet when responding. We did not collect consistent data on use of nonaspirin NSAIDs in either cohort. We did not evaluate the use of selective inhibitors of COX-2 that were introduced in the United States in 1999.
Ascertainment of Cases
We requested permission to acquire medical records and pathology reports from participants who reported colorectal cancer on our biennial questionnaire. We identified deaths from the National Death Index and from next of kin.24 A study physician who was unaware of information on the participants' intake of aspirin reviewed medical and pathological records to extract information on histologic types and anatomic locations of the cancers. A single study pathologist reviewed all of the cases that were retrieved for determination of COX-2 expression.
Assessment of COX-2 Expression
Beginning in 1997 in the HPFS and 2001 in the NHS, we began retrieving, from the pathology departments of treating hospitals, available pathological specimens from participants whom we confirmed had received a diagnosis of colorectal cancer. We obtained specimens from 648 cases (76%) over 16 years of follow-up in the HPFS and 662 cases (58%) over 22 years of follow-up in the NHS. We limited our analysis of the expression of COX-2 to unstained paraffin blocks with amounts of tumor tissue and adjacent mucosa that were sufficient for immunohistochemical analysis (636 cases: 368 from the NHS and 268 from the HPFS). The baseline characteristics of participants with colorectal cancer whose tumors we analyzed were similar to those of participants whose tumors we did not analyze (mean age, 54.4 vs. 54.9 years; nonwhite, 3% vs. 3%; former or current smoker, 60% vs. 58%; mean body-mass index [the weight in kilograms divided by the square of the height in meters], 25.5 vs. 25.3; mean metabolic equivalent [MET; i.e., exercise intensity] score per week, 16.6 vs. 15.3; current multivitamin use, 36% vs. 33%; previous lower gastrointestinal endoscopy, 13% vs. 12%; folate intake, 401 vs. 408 µg per day; calcium intake, 798 vs. 786 mg per day; alcohol intake, 9.9 vs. 10.2 g per day; P>0.25 for all comparisons).
We used a previously described method for COX-2 immunostaining on whole tissue sections or on microassays constituted from specimens.25,26 We incubated deparaffinized tissue sections with citrate buffer (BioGenex) in a microwave oven for 15 minutes, then cooled them for 40 minutes. Tissue sections were then incubated with 3% hydrogen peroxide for 20 minutes, avidin block for 15 minutes, and biotin block for 15 minutes. A mouse anti–COX-2 antibody (Cayman Chemical) diluted 1:300 in phosphate-buffered saline was applied, and the slides were maintained overnight at 4°C. Next, we applied an antimouse horse IgG antibody (Vector Laboratories) for 20 minutes, followed by an avidin–biotin complex conjugate (Vector Laboratories). The immunochemical reaction was revealed by diaminobenzidine and methyl-green counterstain. For each assay run, we included a cancer tissue with COX-2 overexpression and a negative control (normal colonic tissue). We also examined a specimen with known COX-2 overexpression treated with phosphate-buffered saline, but not the anti-COX-2 antibody, as a control for nonspecific binding of the immunohistochemical marker.
A pathologist who was unaware of any data concerning the participants interpreted COX-2 expression, using a standardized grading system (absent, weak, moderate, or strong). The pathologist classified staining of tumor cells as "absent" if COX-2 expression was at the same level of intensity as that in adjacent normal colonic epithelium; weak, moderate, or strong staining indicated progressively increasing degrees of overexpression. A random sample of 108 cancers was reread by a second pathologist unaware of data on the participants; the concordance between the two pathologists was 0.92 (=0.62, P<0.001).26 If the intensity of immunostaining was moderate or strong, tumors were classified as cancers with COX-2 overexpression (COX-2–positive). If immunostaining intensity was weak or absent, tumors were classified as negative for COX-2 overexpression (COX-2–negative) (Figure 1).
|
At baseline, we excluded men and women who did not complete the baseline dietary questionnaire or medication questions, who recorded implausible dietary or aspirin data (e.g., responded "yes" to use but then recorded use of zero aspirin tablets per week), or reported a baseline history of cancer (except nonmelanoma skin cancer). We also excluded participants with inflammatory bowel disease, familial polyposis, or hereditary nonpolyposis colon cancer. After these exclusions, 82,911 women and 47,363 men were eligible for analysis. They accrued follow-up time beginning on the month of return of their baseline questionnaire and ending on the month of diagnosis of colorectal cancer, death from other causes, or June 2002, whichever came first. Participants in whom colorectal cancer was adjudicated as the cause of death by study physicians after review of medical records were included as cases. Data from cases of colorectal cancer for which we were unable to assay COX-2 expression in the tumor were censored from the analysis at the date of diagnosis.
To reduce within-person variation and to better estimate long-term intake, we used the cumulative average intake of aspirin as reported on all available questionnaires up to the start of each 2-year follow-up interval.27 In the NHS, women who reported taking two or more standard (325-mg) aspirin tablets per week were classified as regular users, whereas those who reported less aspirin use were classified as nonusers. We also estimated the duration of the regular use of aspirin by NHS participants from the number of years of regular use reported in 1980, with updating of this variable every 2 years.19,20,23 In the HPFS, men who reported using aspirin at least two times per week were classified as regular users, whereas those who did not report aspirin use at least two times per week were classified as nonusers.18 For our main analyses, we pooled data from both cohorts. We conducted tests of heterogeneity using the Q statistic before pooling.28 In this analysis, we observed no heterogeneity between the cohorts regarding the association of regular aspirin use and risk of colorectal cancer (P=0.33 for Cochran's Q test).
We calculated the incidence rates of colorectal cancer for participants in a specific category of aspirin use by dividing the number of incident cases by the number of person-years of follow-up. We calculated relative risks by dividing the incidence rate in one category by the incidence rate in the reference category; we used Cox proportional-hazards modeling to control for multiple variables simultaneously and to calculate 95% confidence intervals. Multivariate relative risks are adjusted for age, sex, smoking before the age of 30 years (0, 1 to 4, 5 to 10, 11 to 15, or more than 15 pack-years); body-mass index (in quintiles); regular vigorous exercise (in quintiles of MET task score per week); colorectal cancer in a parent or sibling (yes or no); previous lower gastrointestinal endoscopy (yes or no); history of polyps (yes or no); current multivitamin use (yes or no); consumption of beef, pork, or lamb as a main dish (0 to 3 times per month, 1 time per week, 2 to 4 times per week, or 5 or more times per week); alcohol consumption (0, 0.1 to 4.9, 5.0 to 14.9, or 15 g or more per day); and energy-adjusted folate and calcium intake (in quintiles). For the analyses restricted to women, the multivariate models additionally adjusted for postmenopausal hormone-replacement therapy (premenopausal women and postmenopausal women who never used, formerly used, or were currently using hormone-replacement therapy) and did not include sex. For the analyses restricted to men, the multivariate models also did not include sex. We used the most recent information for all covariates.
To compare the association between aspirin use and the risk of colorectal cancer in relation to COX-2 expression in the cancer, we used a competing-risk analysis that uses a duplication-method Cox regression.29,30 This method permits estimation of separate regression coefficients for aspirin use stratified according to the type of outcome (e.g., cancer with COX-2 overexpression vs. cancer with weak or absent COX-2 expression). We assessed the statistical significance of the difference between the risk estimates according to tumor type by a likelihood-ratio test that compared the model that allowed for separate associations of aspirin according to COX-2 expression with a model that assumed a common association. We used SAS software (version 9.1.3) for all analyses. All P values are two-sided.
Results
Among the 82,911 eligible women and 47,363 eligible men at the time of the baseline questionnaire, participants reporting the regular use of aspirin (taking two or more standard aspirin tablets per week or using aspirin at least two times per week) were older, were slightly less physically active, were more likely to be current or former smokers, were more likely to use multivitamins regularly, were more likely to use postmenopausal hormones (among women), consumed more alcohol, and consumed more folate (P<0.001 for all unadjusted comparisons) (Table 1). On the basis of the updated data on aspirin use we obtained on biennial questionnaires, we observed that patterns of aspirin use changed over time. Averaged over the total number of person-years of follow-up (2,446,431), the difference in mean age between regular aspirin users and nonusers gradually increased (59.0 years for regular aspirin users and 56.5 years for nonusers, P<0.001). For this analysis, we identified 636 incident cases of colorectal cancer among users and nonusers of aspirin that were available for analysis of COX-2 expression (Table 1). Among these 636 tumors, 423 (67%) had moderate or strong COX-2 expression (i.e., were COX-2–positive), whereas 213 (33%) had weak or absent COX-2 expression (i.e., were COX-2–negative).
|
|
Table 3 shows the association between the duration of regular use of aspirin and the risk of a COX-2–positive or COX-2–negative colorectal cancer. A statistically significant reduction in the number of cases of COX-2–positive cancer was not evident until aspirin had been used regularly for more than 10 years (multivariate relative risk, 0.59; 95%, CI, 0.42 to 0.82; P for trend, <0.001); there was no statistically significant reduction in the number of COX-2–negative cancers with increasing duration of aspirin use (P for trend=0.25).
|
|
Discussion
We found that participants in two large prospective cohorts who reported long-term, regular aspirin use (at least two standard tablets per week or use of aspirin at least two times per week) had a significant reduction in the relative risk of colorectal cancer, as compared with participants who reported less use of aspirin. This reduction in risk was due almost entirely to a reduction in the number of COX-2–positive colorectal cancers among regular aspirin users. In contrast, the difference in the incidence of COX-2–negative colorectal cancers between regular users and nonusers of aspirin was not statistically significant. A reduction of the risk of COX-2–positive tumors was also found with increasing aspirin dose and increasing duration of use. No significant dose or duration effect was observed for COX-2–negative tumors.
These results are consistent with a cross-sectional case–control study in which the association of NSAIDs with a reduced risk of adenomatous polyps was strongest in cases with high expression of COX-2 messenger RNA.31 Our finding that the association with a reduced risk of colorectal cancer is most apparent at intakes of more than five aspirin tablets per day is consistent with the results of studies in which higher doses of aspirin were required to inhibit COX-2 than to inhibit COX-1.1,5 Moreover, the association with a reduction in cancer risk was found only after several years of aspirin use, a finding that suggests an effect of aspirin on the early stages of colorectal adenoma or cancer.1,2,3,4
COX-2 is progressively overexpressed during the stepwise sequence from adenoma to carcinoma,7 and randomized, placebo-controlled trials have shown that selective COX-2 inhibitors prevent recurrence of adenoma among patients with a history of adenoma or familial polyposis.10,11,12,13 Our data suggest that the anticancer benefit of aspirin is mediated, at least in part, by inhibition of COX-2. Experimental studies have shown that aspirin and NSAIDs, especially at high doses, have a range of other potentially antineoplastic actions15,16,17,32,33,34,35,36,37; these results indicate that further work is needed to elucidate the effects of these agents and COX-2 (or its downstream effectors) on the development of colorectal cancer.38,39 COX-2, possibly through production of inflammatory prostaglandins or eicosanoids, may regulate angiogenesis, apoptosis, or tumor-cell invasiveness.6
Our study has several strengths. First, because we collected detailed, updated information on aspirin use during 22 years of follow-up, we were able to evaluate the effects of long-term use across a broad range of intakes. Second, we obtained aspirin data prospectively, before the diagnosis of colorectal cancer. Any errors in recall would have tended to attenuate rather than exaggerate true associations. Third, because the participants were all health professionals, the accuracy of self-reported aspirin use is likely to be high. Fourth, we collected detailed data on potential confounders and had a high follow-up rate (92% of participants returned questionnaires or were identified as deceased). Finally, our results were remarkably consistent between the two independent cohorts.
There are several limitations to our study. The study was observational, and aspirin use was self-selected; thus, among aspirin users and nonusers, there were small, albeit statistically significant, differences in risk factors for colorectal cancer, including smoking history, physical activity, and use of multivitamins. However, adjustment for these characteristics, as well as for a wide range of other potential risk factors, had minimal influence on our findings, suggesting little potential for residual or uncontrolled confounding. We did not have sufficient data on use of nonaspirin NSAIDs in both cohorts to reliably examine their use in relation to COX-2–expressing tumors. However, it is improbable that use of nonaspirin NSAIDs influenced our findings with aspirin; our previous study did not demonstrate any modification by NSAID use of the reduced risk of colorectal cancer observed with aspirin.20 We were unable to obtain tumor tissue from all cases of confirmed colorectal cancer detected in the two cohorts, but the risk factors in these cases did not appreciably differ from those in cases with tumor tissue available.
Another limitation of our study is the lack of a widely accepted, standardized classification scheme for COX-2 expression in colorectal cancer. Nevertheless, previous studies have demonstrated that the results of Western and Northern blot analyses correlate well with immunohistochemical grading of COX-2.40 With our classification of COX-2 expression, we found, in our overall population, a proportion of tumors that overexpressed COX-2 similar to those found by other investigators.8,9,41 In validation studies of the central, blinded review of tumor specimens, we observed substantial interobserver agreement (92%). Any significant misclassification of COX-2 overexpression would be expected to bias our results toward finding no significant differences in the effect of aspirin on tumors in relation to COX-2 expression.
Our results support the importance of continued investigation into COX-2 and related pathways for the development of new treatments and the potential use of COX-2 as a molecular marker for tailoring chemoprevention in participants with a history of colorectal cancer.
Supported by grants from the National Cancer Institute (CA87969, CA55075), the National Institutes of Health, the Marshall S. Kates Memorial Fund, the Bennett Family Fund for Targeted Therapies Research, and the Entertainment Industry Foundation National Colorectal Cancer Research Alliance. Dr. Chan is a recipient of the American Gastroenterological Association/Foundation for Digestive Health and Nutrition Research Scholar Award and a career development award from the National Cancer Institute (CA10741). The National Cancer Institute, the National Institutes of Health, the Marshall S. Kates Memorial Fund, the Bennett Family Fund for Targeted Therapies Research, the Entertainment Industry Foundation National Colorectal Cancer Research Alliance, the American Gastroenterological Association, and the Foundation for Digestive Health and Nutrition had no role in the collection, management, analysis, or interpretation of the data and had no role in the preparation, review, or approval of the manuscript.
Dr. Chan reports receiving a career development award from the Glaxo Institute for Digestive Health for an unrelated study. No other potential conflict of interest relevant to this article was reported.
We thank Mr. Gregory Kirkner for assistance in accessing tumor specimens, Ms. Mami Cantor for assistance in immunohistochemical assays, Dr. Reiko Dehari for interobserver variability studies, Dr. Eva Schernhammer for statistical expertise, and Dr. Edward Giovannucci and Dr. Walter Willett for scientific insights.
Source Information
From the Gastrointestinal Unit, Massachusetts General Hospital and Harvard Medical School (A.T.C.); the Department of Pathology, Brigham and Women's Hospital and Harvard Medical School (S.O.); the Department of Medical Oncology, Dana–Farber Cancer Institute (S.O., C.S.F.); and the Channing Laboratory, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School (A.T.C., C.S.F.) — all in Boston.
Address reprint requests to Dr. Chan at the Gastrointestinal Unit, Massachusetts General Hospital, 55 Fruit St., GRJ 728A, Boston, MA 02114, or at achan{at}partners.org.
References
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Related Letters:
Aspirin, COX-2, and the Risk of Colorectal Cancer
Benelli R., Murashige N., Kami M., Ikeda M., Chan A. T., Ogino S., Fuchs C. S.
Extract |
Full Text |
PDF
N Engl J Med 2007;
357:824-825, Aug 23, 2007.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
