Correspondence
Aspirin and the Risk of Colorectal Cancer in Women
N Engl J Med 1996; 334:119-122January 11, 1996
- Article
To the Editor:
In the recent report from the Nurses' Health Study about the relation between aspirin use and the risk of colorectal cancer (Sept. 7 issue),1 a key question was whether the investigation inadvertently identified women who were resistant to colorectal cancer. A number of potentially confounding factors, including family history of colorectal cancer, smoking history, alcohol consumption, diet, and level of physical activity, were considered in the data analysis. Unaddressed was the effect of hormone-replacement therapy on the risk of colorectal cancer in the study population. Data from the Nurses' Health Study on hormone-replacement therapy and its relation to colorectal cancer have suggested a protective effect of this therapy against colon cancer.2 Evidence linking hormone-replacement therapy and a reduced risk of colon cancer is also available from other studies.3,4 Could the authors comment on whether women in their study who used aspirin two or more times a week were also more likely to be receiving hormone-replacement therapy?
It was reported previously that the relation between the risk of colorectal cancer and the use of nonsteroidal antiinflammatory drugs (NSAIDs) — almost entirely aspirin — depended more on current use of NSAIDs than on previous use discontinued for at least a year.5 It would be interesting to know more about this aspect of the data from the Nurses' Health Study.
5 ReferencesKaren A. Johnson, M.D., Ph.D., M.P.H.
Sheila A. Prindiville, M.D., M.P.H.
National Cancer Institute, Bethesda, MD 208921
Giovannucci E ,Egan KM ,Hunter DJ , et al. Aspirin and the risk of colorectal cancer in women. N Engl J Med 1995;333:609-614
Full Text | Web of Science | Medline2
Chute CG ,Willett WC ,Colditz GA ,Stampfer MJ ,Rosner B ,Speizer FE . A prospective study of reproductive history and exogenous estrogens on the risk of colorectal cancer in women. Epidemiology 1991;2:201-207
CrossRef | Medline3
Newcomb PA ,Storer BE . Postmenopausal hormone use and risk of large-bowel cancer. J Natl Cancer Inst 1995;87:1067-1071
CrossRef | Web of Science | Medline4
Calle EE ,Miracle-McMahill HL ,Thun MJ ,Heath CW Jr . Estrogen replacement therapy and risk of fatal colon cancer in a prospective cohort of postmenopausal women. J Natl Cancer Inst 1995;87:517-523
CrossRef | Web of Science | Medline5
Rosenberg L ,Palmer JR ,Zauber AG ,Warshauer ME ,Stolley PD ,Shapiro S . A hypothesis: nonsteroidal anti-inflammatory drugs reduce the incidence of large-bowel cancer. J Natl Cancer Inst 1991;83:355-358
CrossRef | Web of Science | Medline
To the Editor:
The paper by Giovannucci et al. provides important data on the optimal frequency and duration of aspirin use required for the chemoprevention of colorectal cancer. I was particularly interested in their graphs showing the relations between relative risk and the frequency and duration of use (Figure 1 and Figure 2, respectively). On first inspection, the graphs are impressive, since reductions in the risk of colorectal cancer related to aspirin dose and duration of use are shown. However, if we consider a dual mechanism for aspirin's protective effect, it can be shown that the graphs have limited meaning.
For purposes of discussion, aspirin protection can be considered as a combination of systemic immunostimulation1 and direct inhibition of colorectal carcinogenesis.2 Figure 1 shows a significant reduction in the risk of colorectal cancer with four or more aspirin tablets per week. Within this level of protection, however, there is a large margin of undefined aspirin use. For example, a dose of eight aspirins could reflect four tablets taken on two days, two tablets taken on four days, or any number of intermediate combinations. These categories need to be more specific, and this bias shows the limitation of Figure 2. Even if the hypothesis of a dual mechanism, mentioned above, is only partially correct, it is likely that the reduction in risk is proportional to the aspirin dose. Consider two patients, both taking eight aspirin tablets per week. Let Patient 1 take two aspirins on alternate days while Patient 2 takes four aspirins on two separate days. It is possible that a 50 percent reduction in cancer risk would be obtained at different times — for example, 5 and 10 years, respectively.
In extension of this principle, I disagree with the editorial in the same issue by Marcus,3 who recommends 325 mg of aspirin on alternate days for patients at risk of colorectal cancer. Currently, there is very little justification for intervention trials, since there is no classification of various risk levels. Indeed, although there is good evidence that aspirin consumption reduces the risk of colorectal cancer, no study has profiled the base-line risk of cancer among aspirin users. It is thus important to develop a carcinogenic grading system, since it is likely that aspirin intervention in low-risk cohorts would be different from that in high-risk cohorts.4
4 ReferencesGareth Morgan, B.Sc.
West Glamorgan Health Authorities, Swansea SA1 1LT, United Kingdom1
Odeh M . Aspirin and risk of fatal colon cancer. Postgrad Med J 1993;69:242-242
CrossRef | Web of Science | Medline2
Morgan GP ,Williams JG . Aspirin and risk of fatal colon cancer. Postgrad Med J 1993;69:893-894
CrossRef | Web of Science | Medline3
Marcus AJ . Aspirin as prophylaxis against colorectal cancer. N Engl J Med 1995;333:656-658
Full Text | Web of Science | Medline4
Morgan G . Nonsteroidal antiinflammatory drugs and colorectal cancer. Cancer 1995;75:2781-2781
CrossRef | Web of Science | Medline
To the Editor:
The article by Giovannucci et al. was a retrospective analysis of an observational study and is naturally endowed with recall bias. It is important that the authors use the word “retrospective” in the abstract and the Methods section, and it is essential that the reviewers demand this of the authors. The authors also defined “nonusers” as those taking fewer than two tablets of aspirin per week. I wonder if this was a preanalysis decision or was based on multiple statistical analysis of the data that allowed them to stumble on this cutoff. Did they control for long-term calcium use by the participants, since calcium may be an effective chemopreventive agent for colorectal cancer?1
The editorial by Dr. Marcus was perplexing. A large part of the editorial dealt with the basic mechanisms of action of aspirin and made no effort to help the readers critique the article. What is more surprising are the recommendations given by Dr. Marcus. Surely he did not base these recommendations on the interesting analysis by Giovannucci et al. First, the only two randomized studies done to date, with all their flaws, have not shown any superiority of aspirin over placebo.2,3 Second, using Table 1 (column 3) of the article by Giovannucci et al., we could predict that 105 cases of colorectal cancer could be prevented after 456,393 person-years of long-term aspirin use. In other words, to prevent 105 cases of colorectal cancer in women, 45,639 women would need to take four to six tablets of aspirin per week for at least 10 years. This also means that, to prevent these 105 cases, at least 94,929,120 tablets of aspirin would be needed. We do not know the effect of aspirin therapy on disease-specific mortality or even its cost effectiveness. Third, various issues such as toxicity3 and problems with surgery and blood donation cannot be forgotten. I do not think there are sufficient data to make such strong recommendations, even for people at high risk.
Finally, regarding the pitch made for such retrospective studies, I can only say that retrospective studies cannot replace well-designed, placebo-controlled prospective studies. This is a fact of science.
3 ReferencesDebasish F. Roychowdhury, M.D.
Veterans Affairs Medical Center, Cincinnati, OH 452201
Scalmati A ,Lipkin M ,Newmark H . Calcium, vitamin D and colon cancer. Clin Appl Nutr 1992;2:67-742
Gann PH ,Manson JE ,Glynn RJ ,Buring JE ,Hennekens CH . Low-dose aspirin and incidence of colorectal tumors in a randomized trial. J Natl Cancer Inst 1993;85:1220-1224
CrossRef | Web of Science | Medline3
Paganini-Hill A ,Chao A ,Ross RK ,Henderson BE . Aspirin use and chronic diseases: a cohort study of the elderly. BMJ 1989;299:1247-1250
CrossRef | Web of Science | Medline
To the Editor:
The study by Giovannucci et al. (and other well-documented clinical and experimental studies) led Marcus in his editorial to recommend regular aspirin use in a rather heterogeneous group of people at risk for colorectal cancer. He suggests that patients with inflammatory bowel disease; breast, ovarian, or endometrial cancer; or a previous adenoma or large-bowel cancer would benefit from 325 mg of aspirin every other day. The study of Giovannucci et al. is indeed supportive of aspirin as prophylaxis against colorectal neoplasms. However, the mechanisms of the demonstrated prophylactic effect of aspirin or other NSAIDs are unclear, and the dose required to obtain this chemopreventive effect is speculative. This makes it doubtful that the presented data should be extrapolated to all the risk groups mentioned by Marcus.
Despite its salutary effects, the use of any dose of aspirin or other NSAID is potentially harmful, especially for the gastrointestinal tract.1 In particular, in patients with inflammatory bowel disease, the use of NSAIDs can lead to flare-ups of colitis, making any long-term beneficial effect of aspirin questionable.1,2
Given these potential risks of NSAIDs, we believe that the recommendation of regular aspirin use in all patient groups at risk for the development of colorectal cancer is premature and should be tempered until further evidence in favor of regular aspirin use is forthcoming.
2 ReferencesA.A. van Bodegraven, M.D.
J. Lourens, M.D., Ph.D.
J.W. Sindram, M.D., Ph.D.
Medical Center Alkmaar, 1815 JD Alkmaar, the Netherlands1
Bjarnason I ,Hayllar J ,MacPherson AJ ,Russell AS . Side effects of nonsteroidal anti-inflammatory drugs on the small and large intestine in humans. Gastroenterology 1993;104:1832-1847
Web of Science | Medline2
Kaufmann HJ ,Taubin HL . Nonsteroidal anti-inflammatory drugs activate quiescent inflammatory bowel disease. Ann Intern Med 1987;107:513-516
Web of Science | Medline
To the Editor:
Dr. Marcus's suggestion that those at high risk for colon cancer take 325 mg of aspirin every second day is problematic for two reasons.
First, we see many people, and not just elderly ones, who have substantial upper gastrointestinal bleeding due to regular small doses of aspirin that they take for its value in prophylaxis against cardiovascular disease. Many such patients have ulcers without Helicobacter pylori gastritis, and these ulcers are almost certainly due to aspirin ingestion. Small doses of aspirin have also been recognized as a source of ulceration of the small intestine beyond the duodenum.1,2
Second, among those at risk for colon cancer are those with inflammatory bowel disease. NSAIDs have been shown to exacerbate inflammatory bowel disease,3 and one must wonder if the risk of recurrent or worsened inflammation from aspirin outweighs the possible benefits of prophylaxis against colon cancer.
I realize one cannot know the answers to the questions that I have raised, but a note of caution should be sounded.
3 ReferencesHerbert J. Kaufmann, M.D.
Katonah Medical Group, Katonah, NY 105361
Allison MC ,Howatson AG ,Torrance CJ ,Lee FD ,Russell RI . Gastrointestinal damage associated with the use of nonsteroidal antiinflammatory drugs. N Engl J Med 1992;327:749-754
Full Text | Web of Science | Medline2
Bjarnason I ,Hayllar J ,MacPherson AJ ,Russell AS . Side effects of nonsteroidal anti-inflammatory drugs on the small and large intestine in humans. Gastroenterology 1993;104:1832-1847
Web of Science | Medline3
Kaufmann HJ ,Taubin HL . Nonsteroidal anti-inflammatory drugs activate quiescent inflammatory bowel disease. Ann Intern Med 1987;107:513-516
Web of Science | Medline
To the Editor:
In his editorial accompanying the paper by Giovannucci and associates, Marcus recommends aspirin therapy for patients at risk for colorectal cancer. Is this conclusion really justified? In past volumes of the Journal, there has been an increase in the publication of epidemiologic surveys from large studies, such as the Nurses' Health Study. Most of these studies were not designed as interventional trials that prospectively compared a certain medication, diet, or behavior in a study group with that in a control group. Therefore, only a coincidence between two variables can be identified as statistically probable or not. A causal relation cannot be deduced, irrespective of the number of variables used in stratifying the sample. Furthermore, many interventional trials — for example, those examining antioxidant vitamins and colorectal adenoma1 — have failed to convert results from epidemiologic surveys into therapeutic measures. Although valuable, epidemiologic studies are the wrong way to define new therapies. At most, they provide clues for the rational design of clinical trials, as does in vitro evidence from the laboratory.
1 ReferencesMartin Schuler, M.D.
Johannes Gutenberg University, D-55101 Mainz, Germany1
Greenberg ER ,Baron JA ,Tosteson TD , et al. A clinical trial of antioxidant vitamins to prevent colorectal adenoma. N Engl J Med 1994;331:141-147
Full Text | Web of Science | Medline
Author/Editor Response
The authors reply:
To the Editor: Dr. Roychowdhury refers to our study as retrospective and that by Paganini-Hill et al. as a randomized trial, but they are, in fact, both prospective cohort studies. Thus, it is difficult to see how this study is “naturally endowed with recall bias,” since aspirin use was reported before the diagnosis of cancer. Regarding the selection of two aspirin tablets per week, we used an identical cutoff in a similar study in men. We chose this same cutoff precisely to avoid multiple statistical analyses. Moreover, we presented the results for all categories of aspirin intake (Figure 2). Calcium was not a confounder, as noted in the paper. In this regard, the Nurses' Health Study is the only study to have assessed calcium intake on the basis of multiple measurements over time, in order to assess long-term use more accurately.1 We agree that a placebo-controlled intervention trial would provide the strongest form of evidence, but our results suggest that such a trial would have to be conducted for 10 to 20 years before an association would become apparent.
Drs. Johnson and Prindiville raise the issue of estrogen-replacement therapy as a potential confounder. The postmenopausal use of estrogens was slightly less frequent among users of aspirin than among nonusers; thus, any protection gained by postmenopausal hormone use could not have accounted for our findings with respect to aspirin. Other evidence supports a benefit from aspirin for men. We believe that any analysis involving the discontinuation of aspirin use within several or more years before the diagnosis of colorectal cancer is extraordinarily prone to bias. For example, among women who are 65 years old, about 1 in 1000 is expected to be given a diagnosis of colorectal cancer in any one year. Assuming that aspirin halves one's risk, we would expect that colorectal cancer would be diagnosed in 1 in 2000 long-term aspirin users in a year. In a group of 2000 long-term aspirin users who had stopped use within the past year, if only 1 of these women had terminated her use of aspirin because of an undiagnosed cancer (or an adenoma that would become malignant), the protection conferred by aspirin would be entirely obscured.
We agree with Mr. Morgan that more data on the dose response would be helpful, and we would add that the apparent benefit with regard to cardiovascular disease also needs to be considered before any recommendations are made.
1 ReferencesEdward Giovannucci, M.D.
Frank E. Speizer, M.D.
Brigham and Women's HospitalKathleen Egan, M.P.H.
Harvard School of Public Health, Boston, MA 021151
Martinez ME ,Giovannucci E ,Colditz G ,Stampfer M ,Willett W ,Speizer F . Relation of calcium, vitamin D, and milk consumption to the risk of colorectal cancer in a prospective study among women. Am J Epidemiol 1995;141:Suppl:S70-S70 abstract.
Web of Science
Author/Editor Response
I thank Drs. Roychowdhury, van Bodegraven et al., Kaufmann, and Schuler for their perceptive and pertinent comments. Prophylaxis is an extremely important consideration in a disease that causes at least 60,000 deaths per year and carries a five-year survival rate of less than 40 percent. Giovannucci et al. identified a cohort of women who took four to six aspirin tablets per week for 10 or more years, with a statistically significant reduction in the occurrence of colorectal cancer. The use of aspirin and other NSAIDs as prophylactic agents against colorectal cancer had been suggested previously by the results of studies in both humans and animals.1
The mechanism of the protective effect of aspirin is unknown. Recent information about the second form of arachidonic acid cyclooxygenase — PGHS-2, the expression of which is up-regulated in colorectal adenomas — indicates its possible involvement in the proliferative process.2,3 Therefore, it follows that the inhibition by aspirin of this enzyme might be clinically useful. To convey this concept, I included a diagram of arachidonic acid metabolism. Lipoxygenase-catalyzed hydroxy fatty acids, which are produced in abundance after the ingestion of aspirin, may also be involved in the protective effect, but they have not been tested in this regard.
A significant reduction in the incidence of colorectal cancer can apparently be achieved with the use of a simple, inexpensive, and relatively safe schedule of aspirin ingestion. Nevertheless, the critically important caveats of the correspondents mentioned above require consideration. People who are at high risk for colorectal cancer by virtue of personal or family history should evaluate, in consultation with their physicians, the risks and benefits of aspirin prophylaxis, as recommended by my editorial. I regard the cost–benefit ratio to be vastly in favor of aspirin treatment when all consequences are considered.
Research is urgently needed in two areas: prospective, placebo-controlled clinical trials, as difficult as they may be, and more studies on the direct or indirect antiproliferative effects of aspirin and other NSAIDs. This is especially appropriate in colorectal cancer. I maintain that, until we have definitive answers, we should act on the implications of the study by Giovannucci et al. and use medically supervised aspirin prophylaxis in people with a high risk of colorectal cancer.
3 ReferencesAaron J. Marcus, M.D.
New York Veterans Affairs Medical Center, New York, NY 100101
Marnett LJ . Aspirin and the potential role of prostaglandins in colon cancer. Cancer Res 1992;52:5575-5589
Web of Science | Medline2
Eberhart CE ,Coffey RJ ,Radhika A ,Giardiello FM ,Ferrenbach S ,Dubois RN . Up-regulation of cyclooxygenase 2 gene expression in human colorectal adenomas and adenocarcinomas. Gastroenterology 1994;107:1183-1188
Web of Science | Medline3
DuBois RN ,Awad J ,Morrow J ,Roberts LJ II ,Bishop PR . Regulation of eicosanoid production and mitogenesis in rat intestinal epithelial cells by transforming growth factor-α and phorbol ester. J Clin Invest 1994;93:493-498
CrossRef | Web of Science | Medline
