Selenium and Cancer Prevention

Promising

Results Indicate Further Trials

randomized
women

Required
(approximately 75% of the participants) and
women

Despite advances in diagnosis and treatment, morbidity and mortality from cancer continue to represent a major health burden. Given the potentially devastating consequences and the fear associated with a diagnosis of cancer, it is not surprising that physicians, patients, and the public may hold substantial optimism for and have considerable interest in seemingly simple and relatively easily implemented measures, such as dietary modification or use of vitamin and trace element supplementation, that appear rapidly effective for cancer prevention. Previous studies examining the potential effects of dietary supplements, such as antioxidant vitamins1,2 for prevention of cancer, have produced mixed results. In this
issue of THE JOURNAL, Clark and colleagues3 present promising but preliminary findings from a randomized trial evaluating the possible effectiveness of selenium supplementation for cancer prevention.

men

See also p 1957.

The Nutritional Prevention of Cancer Study by Clark et al3 randomized 1312 patients to receive placebo or 200 \g=m\gof selenium per day and was designed to evaluate the effect of this supplement on risk of developing new basal cell and squamous cell skin cancers. In analysis of the primary out comes, selenium supplementation had no effect on reducing the incidence of these skin cancers. However, after prelimi nary analyses showed a reduction in total carcinoma, the protocol was modified in 1990 to add total and cancer mor tality, as well as the incidence of lung, colon, rectum, and prostate cancers as secondary end points. Analysis of these end points indicates apparent rapid and large preventive effects for cancers of the lung, prostate, colon, and rectum, but no reduction in risk of breast or bladder cancer. The overall effect, however, when all carcinomas are combined into a secondary end point, is a statistically significant 39% reduction in incidence and 48% reduction in mortality. Interpretation of secondary end points, like the interpreta tion of subgroups in randomized trials, requires caution.4 Clark et al specified a priori the secondary end points for follow-up from 1991 to 1993. In this period, the results remained consis tent with those observed during the first 6 years; incidence and mortality were both significantly reduced. Additional caution in interpretation is needed given that pathology reports were available for only 76% of cancers. And, importantly, this trial
From the Department of Medicine, Brigham and Women's Hospital and Harvard Center for Cancer Prevention, Harvard Medical School, Boston, Mass. Corresponding author: Graham A. Colditz, MBBS, DrPH, Channing Laboratory, 181 Longwood Ave, Boston, MA 02115.

included and the existing epidemiologie data that show no con sistent reduction in risk of breast and other cancers specific to women with higher selenium intake,5 the applicability of the results to women remains uncertain. How should the results of prevention studies be inter preted? Preventive inference should not be based solely on randomized trials. Randomization is helpful but not neces sary,6 and consistency with biologic knowledge of the agent and of the time course of disease also should be considered. Importantly, the time course of prevention should be bio logically plausible. That is, the time from initiating a pre ventive measure should fit with the known biology of specific tumors and with the potential mechanisms of action for the preventive agent, when these have been described. The study by Clark et al3 reports that selenium supplementation was associated with a reduction in risk of lung cancer within 5 years of randomization that is substantially greater than the effect of smoking cessation.7 As justifications for belief in its results, the authors of this randomized trial focus not only on its study design, but also on the potential biologic mechanism whereby selenium acts through molecular mechanisms that regulate apoptosis (ie, programmed cell death). Reliance on study design alone as justification for change in prevention recommendations is misguided. Results from randomized trials must be considered in the context of all the available evidence. Other randomized prevention trials using lower doses of selenium have not shown such substantial protection against total cancer.8 Whether this inconsistency is the result of the dose of selenium, the form of selenium, or the specific cancers studied (cancers of the esophagus and stomach in China vs cancers of the lung, colon, rectum, and prostate in the United States) is unclear. As in clinical treat ment trials, decisions are best made when a number of studies show consistent findings. Further, in interpreting the results, the authors emphasize the effectiveness of selenium in animal tumor model studies, perhaps because of the large effects observed. These studies suggest a specificity of the molecular mechanisms that is per plexing. Laboratory studies cited to justify these mechanisms include breast cancer cell lines,9 yet this trial and numerous epidemiologie studies found no relation between selenium and risk of breast cancer.1011 If this possible magic bullet acts through mechanisms suggested by the authors, it is not clear why it should be specific to only some cancers. Moreover, human populations behave differently from animals, and the laboratory findings are not consistent with all the human data. Thus, further work is required to identify markers for the molecular changes of apoptosis and cell cycle regulation within

(25%). But, because of the relatively fewer

humans that may better reflect the mechanism of action for selenium, rather than reliance on animal models or cell lines. The specificity of the relation for only some among many cancers also raises questions of interpretation of secondary end points in trials. When multiple subgroups, here specific cancer sites, are examined, the probability of finding chance differences increases.12 Importantly, even though the results for lung and colorectal cancers agree with previous epide miologie studies, little evidence exists to support a relation with prostate cancer. In addition, a better understanding of the relation between selenium supplementation at this level and risk of cancers specific to women is required to rule out adverse effects and safely generalize the results to women. The dose of selenium used in this trial is not substantially greater than the comparisons that can be achieved in analytic epidemiologie studies among free-living individuals in the United States. Individuals in high selenium areas (ie, those with high selenium soil content, such as North and South Dakota) will have higher blood levels of selenium than those achieved in this trial which selected a low-intake area for recruitment of study participants. When we have heteroge neity among the results from studies, the variation in dose is clearly one plausible explanation.13 If the dose is central to the magnitude of benefit observed by Clark et al, then additional studies will be required to replicate this finding. Simulta neously, a systematic reexamination ofthe dose relation within the existing epidemiologie data would also be informative. While it is clear that additional randomized trials are nec essary to confirm or refute the findings of Clark et al,3 pos sible strategies to implement prevention with selenium for the broader US population also should be explored. Preven tion goals that focus on risk reduction across the entire com munity, rather than merely cancer-specific approaches, have the greatest likely benefit.14 Population-wide approaches that include small changes in risk for individuals result in major changes in the population burden of disease.15 How to achieve such population-wide prevention should selenium prove to be such a powerful preventive agent requires focused study. A preventive strategy may be implemented through health care professionals (who can administer diagnostic tests, pre scribe drugs, or counsel patients about dietary change), through government regulations that reduce the risks of harm to the population, and through local activities that promote a healthier lifestyle and environment.16 The social strategy developed to increase folate intake offers one useful example. In March 1996, the Food and Drug Administration changed the recommended daily allowance of this nutrient back to 400 g per day. To ensure adequate intake, US food manufacturers will add the nutrient folie acid to most enriched breads, flours, corn meals, pastas, rice, and other grain products. This simple regulation aims to raise intake for the majority of the US population. While it is premature to market selenium based on the results re ported by Clark et al,3 a similar population-wide approach should likely be considered for selenium if the results of the current trial are substantiated and benefits are also observed among women. However, selenium has known dose-dependent ad verse effects, which also must be considered. Supplementation through bread flour already has been shown to increase levels

of selenium.17 Such population-wide approaches are most ef fective if we are to maximize the benefits of prevention strat egies and achieve our national goals of reducing the burden of cancer. Meanwhile, as we await the results of further preven tion research, known lifestyle changes that can reduce cancer risk (such as smoking cessation, consuming adequate amounts of fruits and vegetables each day, reducing intake of animal fat, and increasing physical activity)18 should be implemented. In conclusion, this promising set of results based on sec ondary end points from a skin cancer prevention trial by Clark et al3 require confirmation in further randomized trials designed to test the effect of selenium supplementation on cancer incidence and mortality. Given the results of this study, the expected effect will be large and the time frame will be short. Therefore, such trials should be feasible, and results should be available in the near future. These trials must evaluate the influence of dose and must include women to determine whether these benefits apply to the common car cinomas in women, including breast and ovarian cancers, and so that explicit comparisons of the efficacy of selenium in women and men can be made. For now, it is premature to change individual behavior, to market specific selenium supple ments, or to modify public health recommendations based on the results of this one randomized trial.
Graham A. Colditz, MBBS, DrPH
1. Hennekens CH, Buring JE, Manson JE, et al. Lack of effect of long-term supplementation with beta carotene on the incidence of malignant neoplasms and cardiovascular disease. N Engl J Med. 1996;334:1145-1149. 2. Omenn GS, Goodman GE, Thornquist MD, et al. Effects of a combination of beta carotene and vitamin A on lung cancer and cardiovascular disease. N Engl J Med.

LC, Combs GF, Turnbull BW, et al, for the the Nutritional Prevention of Cancer Study Group. Effect of selenium supplementation for cancer prevention with carcinoma of the skin: a randomized controlled trial. JAMA. 1996;276:1957-1963. 4. Pocock S, Hughes M. Estimation issues in clinical trials and overviews. Stat Med.
5. Garland M, Morris J, Stampfer M, et al. Prospective study of toenail selenium levels and cancer among women. J Natl Cancer Inst. 1995;87:497-505. 6. Hill AB. The environment and disease: association or causation? Proc R Soc Med.

3. Clark

1996;334:1150-1155.

1990;9:657-671.

1965;58:295-300. Dept of Health and Human Services. The Health Benefits ofSmoking CessaWashington, DC: US Dept of Health and Human Services, Public Health Service, Centers for Disease Control, Center for Chronic Disease Prevention and Health Promotion, Office of Smoking and Health; 1990. Dept of Health and Human Services publication CDC 90-8416. 8. Blot WJ, Li JY, Taylor PR, et al. Nutrition intervention trials in Linxian, China: supplementation with specific vitamin/mineral combinations, cancer incidence, and disease-specific mortality in the general population. J Natl Cancer Inst. 1993;85:
7. US

tion.

H, Wilson A, Lu J, et al. Comparison of the effects of an organic and inorganic form of selenium on a mammary carcinoma cell line. Carcinogenesis. 1994;
10. Hunter DJ, Morris JS, Stampfer MJ, Colditz GA, Speizer FE, Willett WC. A prospective study of selenium status and breast cancer risk. JAMA. 1990;264:1128\x=req-\

1483-1492. 9. Thompson

15:183-186.

PA, Collette HJ, Maas MJ, de Waard F. Selenium levels in nails of premenopausal breast cancer patients assessed prediagnostically in a cohort-nested case-referent study among women screened in the DOM project. Int J Epidemiol.
12. Ingelfinger J, Mosteller F, Thibodeau L, Ware J. Biostatistics in Clinical Medicine 3rd ed. New York, NY: McGraw-Hill Inc; 1994. 13. Hertz-Picciotto I, Neutra R. Resolving discrepancies among studies: the influence of dose in effect size. Epidemiology. 1994;5:156-163. 14. Colditz G, Gortmaker S. Cancer prevention strategies for the future: risk identification and prevention intervention. Milbank Q. 1995;73:621-651. 15. Rose G. Strategy of prevention: lessons from cardiovascular disease. Br Med J

1131. 11. van Noord

1987;16:318-322.

Clin Res. 1981;282:1847-1851. 16. Richmond J, Kotelchuck M. Co-ordination and development of strategies and policy: the United States example In: Holland W, ed. Textbook of Public Health. Oxford, England: Oxford University Press; 1984. 17. Longnecker M, Stampfer M, Morris J, et al. A 1-year trial of the effect of highselenium bread on selenium concentrations in blood and toenails. Am J Clin Nutr.

1993;57:408-413. 18. Willett W, Coloditz G, Mueller N. Strategies for minimizing cancer risk. Sci Am. 1996;275:88-95.