Statins and the Risk of Colorectal Cancer
http://www.100md.com
《新英格兰医药杂志》
To the Editor: Poynter et al. (May 26 issue)1 reported a 47 percent reduction in the risk of colorectal cancer among long-term statin users (i.e., five years or more), as compared with short-term users or nonstatin users. In contrast, a randomized trial2 and meta-analyses3 have found either a slight increase in cancer among statin users or no relation between statins and cancers. Although trials are often limited to comparatively young, low-risk patients and have short follow-up, this difference may result from methodologic limitations in the observational study by Poynter et al.
Long-term statin users tend to be healthier, less physically and cognitively frail, and more adherent to therapy and screening than nonusers.4,5,6 In persons who use statins preventively, it may be more likely that precancerous colorectal polyps will be detected and removed early, further making statins appear to be protective. The current study failed to adjust for these factors, which probably led to residual confounding. Therefore, the lower risk of cancer among long-term statin users may say more about the kind of patients who seek these drugs and comply with statin therapy than about any protective effect. This may be similar to the "cardioprotective" effect once reported for long-term estrogen use.
Soko Setoguchi, M.D., M.P.H.
Jerry Avorn, M.D.
Sebastian Schneeweiss, M.D.
Brigham and Women's Hospital
Boston, MA 02120
ssetoguchi@partners.org
References
Poynter JN, Gruber SB, Higgins PDR, et al. Statins and the risk of colorectal cancer. N Engl J Med 2005;352:2184-2192.
Shepherd J, Blauw GJ, Murphy MB, et al. Pravastatin in elderly individuals at risk of vascular disease (PROSPER): a randomised controlled trial. Lancet 2002;360:1623-1630.
Bjerre LM, LeLorier J. Do statins cause cancer? A meta-analysis of large randomized clinical trials. Am J Med 2001;110:716-723.
Glynn RJ, Knight EL, Levin R, Avorn J. Paradoxical relations of drug treatment with mortality in older persons. Epidemiology 2001;12:682-689.
Benner JS, Glynn RJ, Mogun H, Neumann PJ, Weinstein MC, Avorn J. Long-term persistence in use of statin therapy in elderly patients. JAMA 2002;288:455-461.
Redelmeier DA, Tan SH, Booth GL. The treatment of unrelated disorders in patients with chronic medical diseases. N Engl J Med 1998;338:1516-1520.
To the Editor: Poynter and colleagues conducted a large case–control study to assess the effect of statin use for at least five years on the risk of colorectal cancer. In this study, in contrast to other recent reports,1,2,3 statin use was strongly protective. Although the authors adjusted their results for several known risk factors, there are potentially confounding factors that were not considered. Statins are cholesterol-lowering drugs used for the prevention of heart disease and stroke. For obvious reasons, persons taking a chemotherapeutic, preventive agent over a long period of time would have followed other general health recommendations. For example, it is likely that statin users were advised to consume a healthy diet, to increase physical activity, and to stop smoking (smoking is a recognized risk factor for colorectal cancer that was not considered in the study). Also, screening for colorectal cancer may have been more frequent in the statin group, and this factor alone could be responsible for the low rate of colorectal cancer observed in statin users. As pointed out in the accompanying editorial, only placebo-controlled trials will clarify the statin–cancer question.4
Patrick Maisonneuve, Eng.
European Institute of Oncology
20141 Milan, Italy
patrick.maisonneuve@ieo.it
Albert B. Lowenfels, M.D.
New York Medical College
Valhalla, NY 10595
References
Graaf MR, Beiderbeck AB, Egberts AC, Richel DJ, Guchelaar HJ. The risk of cancer in users of statins. J Clin Oncol 2004;22:2388-2394.
Kaye JA, Jick H. Statin use and cancer risk in the General Practice Research Database. Br J Cancer 2004;90:635-637.
Friis S, Poulsen AH, Johnsen SP, et al. Cancer risk among statin users: a population-based cohort study. Int J Cancer 2005;114:643-647.
Hawk E, Viner JL. Statins and cancer -- beyond the "one drug, one disease" model. N Engl J Med 2005;352:2238-2239.
To the Editor: Our recent journal club applauded the efforts of Poynter and colleagues to confirm the validity of self-reported statin exposure with the use of pharmacy records. However, as shown in Table 1, the data raise more questions than they answer.
Table 1. Self-Reported vs. Documented Statin Use.
Although self-reported statin use for five or more years was nearly twice as likely among controls as among patients with cancer, the documented use of a statin within the previous year was nearly equivalent between the two groups. If the self-reported data are accurate, this raises the question of why there was such a dramatic drop in statin use among the controls (particularly in an era when we all seem to be using these agents more, rather than less, often). In this scenario, it would seem that the controls are highly unusual — and not a valid comparison group. If, on the other hand, the self-reported data are inaccurate, and the pharmacy data more closely reflect the true frequency of exposure, then little association between statin use and colorectal cancer exists.
H. Gilbert Welch, M.D., M.P.H.
Veterans Affairs Medical Center
White River Junction, VT 05009
h.gilbert.welch@dartmouth.edu
The authors reply: One limitation of observational epidemiology is the possibility that differences between patients and controls might lead to inappropriate interpretations if these differences are incompletely accounted for within the design or analysis of the study. Setoguchi et al. suggest that statin users in our study might be healthier and more prevention-oriented than nonusers. Our data do confirm that long-term statin users are more likely to participate in screening than are nonusers: 58 percent of controls who used statins for five or more years reported having had a fecal occult-blood test, sigmoidoscopy, or colonoscopy, as compared with 43 percent of controls who were nonusers. However, this difference had no impact on our results. We analyzed our data again, after adjustment for screening, and found no change in the odds ratio for colorectal cancer (adjusted odds ratio, 0.53; 95 percent confidence interval, 0.40 to 0.70). Similarly, Maisonneuve and Lowenfels suggest that diet, physical activity, smoking, and screening for colorectal cancer were not fully considered in the data presented in our article, even though our analysis was adjusted for diet and physical activity. In our data, smoking status was not associated with the risk of colorectal cancer. We estimated the relative risk associated with statins after simultaneously adjusting for smoking, screening, and vitamin supplementation (as another potential measure of "healthy," prevention-oriented behavior) in addition to the variables we previously described. The adjusted odds ratio for colorectal cancer associated with statin use was 0.54 (95 percent confidence interval, 0.41 to 0.72).
Welch notes that the use of a statin within the previous year, as documented in pharmacy records, was nearly equivalent between patients and controls. Pharmacy records were not sufficiently detailed to provide a lifetime summary of medication use for our participants, since such records were derived from a single health provider over one year. Analyses of any statin use recorded in the pharmacy database between 1998 and 2004 are consistent with an inverse association, with any statin use identified in 17.8 percent of patients and 27.7 percent of controls (odds ratio, 0.59; 95 percent confidence interval, 0.48 to 0.73). However, the limitations of electronic pharmacy records, the ability for persons to switch health plans, and the documented validity of self-reporting led us to conclude that self-reported use of medication was the best measure of exposure in our study.
Despite the consistency of our findings, observational studies are not equivalent to randomized clinical trials, and thus we agree with the conclusion of Maisonneuve and Lowenfels that placebo-controlled trials are required to clarify further the potential relationship between statins and the risk of colorectal cancer.
Jenny N. Poynter, Ph.D.
Stephen B. Gruber, M.D., Ph.D., M.P.H.
University of Michigan
Ann Arbor, MI 48109-0638
sgruber@umich.edu
Long-term statin users tend to be healthier, less physically and cognitively frail, and more adherent to therapy and screening than nonusers.4,5,6 In persons who use statins preventively, it may be more likely that precancerous colorectal polyps will be detected and removed early, further making statins appear to be protective. The current study failed to adjust for these factors, which probably led to residual confounding. Therefore, the lower risk of cancer among long-term statin users may say more about the kind of patients who seek these drugs and comply with statin therapy than about any protective effect. This may be similar to the "cardioprotective" effect once reported for long-term estrogen use.
Soko Setoguchi, M.D., M.P.H.
Jerry Avorn, M.D.
Sebastian Schneeweiss, M.D.
Brigham and Women's Hospital
Boston, MA 02120
ssetoguchi@partners.org
References
Poynter JN, Gruber SB, Higgins PDR, et al. Statins and the risk of colorectal cancer. N Engl J Med 2005;352:2184-2192.
Shepherd J, Blauw GJ, Murphy MB, et al. Pravastatin in elderly individuals at risk of vascular disease (PROSPER): a randomised controlled trial. Lancet 2002;360:1623-1630.
Bjerre LM, LeLorier J. Do statins cause cancer? A meta-analysis of large randomized clinical trials. Am J Med 2001;110:716-723.
Glynn RJ, Knight EL, Levin R, Avorn J. Paradoxical relations of drug treatment with mortality in older persons. Epidemiology 2001;12:682-689.
Benner JS, Glynn RJ, Mogun H, Neumann PJ, Weinstein MC, Avorn J. Long-term persistence in use of statin therapy in elderly patients. JAMA 2002;288:455-461.
Redelmeier DA, Tan SH, Booth GL. The treatment of unrelated disorders in patients with chronic medical diseases. N Engl J Med 1998;338:1516-1520.
To the Editor: Poynter and colleagues conducted a large case–control study to assess the effect of statin use for at least five years on the risk of colorectal cancer. In this study, in contrast to other recent reports,1,2,3 statin use was strongly protective. Although the authors adjusted their results for several known risk factors, there are potentially confounding factors that were not considered. Statins are cholesterol-lowering drugs used for the prevention of heart disease and stroke. For obvious reasons, persons taking a chemotherapeutic, preventive agent over a long period of time would have followed other general health recommendations. For example, it is likely that statin users were advised to consume a healthy diet, to increase physical activity, and to stop smoking (smoking is a recognized risk factor for colorectal cancer that was not considered in the study). Also, screening for colorectal cancer may have been more frequent in the statin group, and this factor alone could be responsible for the low rate of colorectal cancer observed in statin users. As pointed out in the accompanying editorial, only placebo-controlled trials will clarify the statin–cancer question.4
Patrick Maisonneuve, Eng.
European Institute of Oncology
20141 Milan, Italy
patrick.maisonneuve@ieo.it
Albert B. Lowenfels, M.D.
New York Medical College
Valhalla, NY 10595
References
Graaf MR, Beiderbeck AB, Egberts AC, Richel DJ, Guchelaar HJ. The risk of cancer in users of statins. J Clin Oncol 2004;22:2388-2394.
Kaye JA, Jick H. Statin use and cancer risk in the General Practice Research Database. Br J Cancer 2004;90:635-637.
Friis S, Poulsen AH, Johnsen SP, et al. Cancer risk among statin users: a population-based cohort study. Int J Cancer 2005;114:643-647.
Hawk E, Viner JL. Statins and cancer -- beyond the "one drug, one disease" model. N Engl J Med 2005;352:2238-2239.
To the Editor: Our recent journal club applauded the efforts of Poynter and colleagues to confirm the validity of self-reported statin exposure with the use of pharmacy records. However, as shown in Table 1, the data raise more questions than they answer.
Table 1. Self-Reported vs. Documented Statin Use.
Although self-reported statin use for five or more years was nearly twice as likely among controls as among patients with cancer, the documented use of a statin within the previous year was nearly equivalent between the two groups. If the self-reported data are accurate, this raises the question of why there was such a dramatic drop in statin use among the controls (particularly in an era when we all seem to be using these agents more, rather than less, often). In this scenario, it would seem that the controls are highly unusual — and not a valid comparison group. If, on the other hand, the self-reported data are inaccurate, and the pharmacy data more closely reflect the true frequency of exposure, then little association between statin use and colorectal cancer exists.
H. Gilbert Welch, M.D., M.P.H.
Veterans Affairs Medical Center
White River Junction, VT 05009
h.gilbert.welch@dartmouth.edu
The authors reply: One limitation of observational epidemiology is the possibility that differences between patients and controls might lead to inappropriate interpretations if these differences are incompletely accounted for within the design or analysis of the study. Setoguchi et al. suggest that statin users in our study might be healthier and more prevention-oriented than nonusers. Our data do confirm that long-term statin users are more likely to participate in screening than are nonusers: 58 percent of controls who used statins for five or more years reported having had a fecal occult-blood test, sigmoidoscopy, or colonoscopy, as compared with 43 percent of controls who were nonusers. However, this difference had no impact on our results. We analyzed our data again, after adjustment for screening, and found no change in the odds ratio for colorectal cancer (adjusted odds ratio, 0.53; 95 percent confidence interval, 0.40 to 0.70). Similarly, Maisonneuve and Lowenfels suggest that diet, physical activity, smoking, and screening for colorectal cancer were not fully considered in the data presented in our article, even though our analysis was adjusted for diet and physical activity. In our data, smoking status was not associated with the risk of colorectal cancer. We estimated the relative risk associated with statins after simultaneously adjusting for smoking, screening, and vitamin supplementation (as another potential measure of "healthy," prevention-oriented behavior) in addition to the variables we previously described. The adjusted odds ratio for colorectal cancer associated with statin use was 0.54 (95 percent confidence interval, 0.41 to 0.72).
Welch notes that the use of a statin within the previous year, as documented in pharmacy records, was nearly equivalent between patients and controls. Pharmacy records were not sufficiently detailed to provide a lifetime summary of medication use for our participants, since such records were derived from a single health provider over one year. Analyses of any statin use recorded in the pharmacy database between 1998 and 2004 are consistent with an inverse association, with any statin use identified in 17.8 percent of patients and 27.7 percent of controls (odds ratio, 0.59; 95 percent confidence interval, 0.48 to 0.73). However, the limitations of electronic pharmacy records, the ability for persons to switch health plans, and the documented validity of self-reporting led us to conclude that self-reported use of medication was the best measure of exposure in our study.
Despite the consistency of our findings, observational studies are not equivalent to randomized clinical trials, and thus we agree with the conclusion of Maisonneuve and Lowenfels that placebo-controlled trials are required to clarify further the potential relationship between statins and the risk of colorectal cancer.
Jenny N. Poynter, Ph.D.
Stephen B. Gruber, M.D., Ph.D., M.P.H.
University of Michigan
Ann Arbor, MI 48109-0638
sgruber@umich.edu