|
Correction
to
Greenberg et al.,
|
| |||||||||||||||||||||||||||||||||
In order to determine before the study that the appropriate surrogate outcome has been selected, the connection between that outcome and the outcome of ultimate interest must be determined -- that is, the accuracy with which the three-year occurrence of adenomas predicts colorectal cancer must be assessed. If this predictive accuracy is low, it is unlikely that an intervention whose benefit is indexed by this surrogate outcome will produce a detectable benefit.
Harry B. Burke, M.D., Ph.D.
University of Nevada School of Medicine
Reno, NV 89520
References
Carotenoids are mostly metabolized in the wall of the small intestine to retinaldehyde and are then converted to retinol1; the rate and extent of these metabolic steps, however, are not clearly defined. Moreover, retinoids, including beta carotene, can show some but not necessarily all of the biologic properties of retinol (vitamin A). It is generally assumed that 1 µg of retinol (approximately 3 units) is biologically equivalent to 6 µg of beta carotene.2 It follows that the amount of "biologically active" vitamin A given by Greenberg et al. to their patients was considerably lower than that administered in other studies in which the treatment (10 mg of retinol) was associated with a reduced occurrence of new colorectal adenomas3 or with a marked inhibition of cell proliferation in the upper portions of colorectal crypts.4
In our opinion, therefore, the negative results obtained by Greenberg and collaborators should not stop the ongoing investigations into the possible role of carotenoids (and other antioxidant compounds) in cancer chemoprevention but, rather, should prompt a more systematic approach, based on a careful evaluation of the effect of different doses of retinol or carotenoids on the incidence of colorectal adenomas or other intermediate end points.
Maurizio Ponz de Leon, M.D.
Luca Roncucci, M.D.
University of Modena
41100 Modena, Italy
References
If this trial had been large enough for a smaller effect of antioxidant vitamins to be detectable, the negative results would have been interpretable. One could then have said that it is unlikely that the negative result occurred by chance and that antioxidant vitamins probably do not affect the progression of adenomas over a four-year period. As this study is written, with no mention of power, the negative results can be incorrectly interpreted to mean that antioxidant vitamins are not efficacious, which may or may not be true.
Without mention of the power of a study or calculations of sample size, it is difficult for the reader to make an informed judgment about the clinical relevance of negative results in randomized, controlled trials.1
Thomas H. Wroth
Columbia University
New York, NY 10032
References
To the Editor: Mr. Wroth believes our study lacked power to find a clinically important reduction in the risk of adenoma. We designed the trial to have 80 percent power to detect a 35 percent reduction from an estimated three-year risk of adenoma of 0.27 in the control group, assuming there was no interaction between the two treatments. The observed risk in the control group was higher, there were fewer losses to follow-up than expected, so the power of the study was actually higher. However, power is not especially informative once a study is completed, its results (with confidence intervals) are known. Our main results excluded a reduction in the relative risk of more than 15 percent for beta carotene and 10 percent for vitamins C and E.
We agree with Dr. Burke that the interpretation of studies of adenoma prevention has limitations. We addressed this issue in our report, and a more complete discussion has been published elsewhere.1 However, it is not feasible to determine whether the recurrence of adenomas predicts colorectal cancer, since following patients and removing their adenomas probably prevents subsequent cancer.2 On balance, we still believe that adenomas are the most practical end points for an intervention study.
In reply to Drs. Ponz de Leon and Roncucci: the 25 mg of beta carotene in our daily supplements greatly exceeded the level of intake associated with a reduced risk of cancer in epidemiologic studies. The rationale for using beta carotene is that it is nontoxic and has antioxidant properties that are not manifested through a conversion to retinol. In people who are not deficient in vitamin A, relatively little of an oral dose of beta carotene is metabolized to retinol in the small intestine, and retinol levels do not increase in blood.3 Indeed, if the effects of beta carotene were mediated only through retinol, it would be simpler and less expensive to give retinol itself. However, long-term daily doses of 10 mg of retinol approach the levels at which toxicity may occur.3
Although we provided evidence that the vitamin supplements had no effect, it is certainly possible that with higher doses or longer use a benefit would be seen. Indeed, no one study can prove that a given intervention does not prevent cancer. Instead, it is up to the proponents of vitamin supplements to prove that the supplements are effective. Until this is done, vitamins should not be promoted for cancer prevention. However, their continued study in clinical trials appears warranted.
Finally, there was a typographic error in our manuscript. Our supplement contained dl-alpha-tocopherol. All agents were provided at no cost by BASF of Wyandotte, Michigan.
E. Robert Greenberg, M.D.
John A. Baron, M.D.
Tor D. Tosteson, Sc.D.
Dartmouth-Hitchcock Medical Center
Lebanon, NH 03756
References
| |||||||||||||||||||||||||||||||||
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
