Can selenium prevent colorectal cancer? A signpost from epidemiology.

The limited epidemiology published heretofore on associations between selenium and colorectal adenomas or cancer has been mixed, inconclusive, and based predominantly on small studies. The epidemiology studies have been about evenly split between those showing a statistically significant (1–5) or suggestive (6,7) protective association and those showing a null (8–11) or suggestive harmful (12,13) association between selenium and colorectal adenomas or cancer. The strongest (and only clinical) previous support for a possible protective effect of selenium intake came from a large, randomized controlled trial of selenium in preventing nonmelanoma skin cancer (14). This trial showed a statistically significant reduction of colorectal cancer as a secondary endpoint (relative risk [RR] 0.58, 95% confidence interval [CI] 0.18 to 0.95) that attenuated after longer follow-up (RR 0.45, 95% CI 0.19 to 1.08) (15). A marginally statistically significant overall reduction in colorectal adenomas, which became stronger in association with the lowest tertile of baseline selenium, was also observed again, as a secondary endpoint in the trial (16). In this issue of the Journal, Jacobs et al. report their pooled analysis of three randomized clinical trials of dietary or nutrient interventions in preventing colorectal adenomas (17). Individuallevel data from each of the Wheat Bran Fiber (WBF) Trial (18), the Polyp Prevention Trial (PPT) (19), and the Polyp Prevention Study (PPS) (20) were statistically reanalyzed to provide a more precise estimate of the association between plasma selenium concentration and adenoma risk than would be possible from any of the three studies alone. None of the trials tested selenium as an intervention. This pooled analysis, the largest epidemiologic investigation of the role of selenium in preventing colorectal neoplasia, indicates that individuals in the highest quartile of plasma selenium concentration (median 150 ng/mL) (plasma collected at enrollment for PPT and PPS and 1 year after randomization for WBF) had 34% lower odds (OR 0.66, 95% CI 0.50 to 0.87, Ptrend .006) of developing a new adenoma throughout the follow-up, compared with those in the lowest quartile of plasma selenium (median 113 ng/mL). Plasma selenium concentrations were not statistically significantly associated with adenoma size, location, or histology in the pooled analysis. There was a trend, however, of fewer advanced adenomas ( 1 cm in diameter and/or villous histology) in persons within the highest quartile of plasma selenium. This suggestion of activity in higher–cancer risk, advanced adenomas supports the promise of selenium for reducing colorectal cancer risk. The association of selenium with colorectal cancer development (and mortality) will be assessed as a prespecified secondary endpoint of the Selenium and Vitamin E [prostate] Cancer Prevention Trial (SELECT) currently under way in more than 35 000 men (21). Although the three source studies were mutually independent and executed prior to the pooling project, they had similar designs and outcomes that strengthened the pooled analysis. All participants were at high adenoma risk after recent colonoscopic adenoma resection; each of the three trials assessed colorectal adenoma endpoints in over 85% of its patients, at year 1 and either year 3 (WBF) or year 4 (PPT and PPS) after randomization; and each intervention did not statistically significantly alter the adenoma risk. The pooled analysis also had limitations, however, which were adequately discussed by the authors and included the possibility that some events (new adenomas) detected at year 1 were not actual events but preexisting adenomas missed at the index colonoscopy, varied blood collection times, and different methodologies used to determine plasma selenium concentrations. The biologic plausibility of the pooled analysis and earlier secondary clinical data is provided by consistent animal studies showing the activity of selenium in colorectal neoplasia (22–28) and by other data that address relevant selenium mechanisms. Selenium has effects on key cellular events of tumorigenesis, such as cell proliferation and apoptosis, and the possible complex mechanisms underlying these effects are emerging in increasing numbers (28–30). Important relevant selenium mechanisms may involve gene promoter methylation and polyunsaturated fatty acid metabolism. Cancer cells commonly have aberrant methylation patterns characterized by global hypomethylation coupled with hypermethylation of CpG islands that appear to contribute importantly to tumorigenesis (31). The enzyme DNA cytosine methyltransferase 1 (DNMT1) is increased in tumor progression in association with regional hypermethylation (31). Preclinical studies suggest that selenium inhibits DNMT1 in various cell lines, including the HCT116 and HT-29 human colonic carcinoma cell lines (22,32–34). Oxidative metabolism of the n-6 polyunsaturated fatty acids arachidonic and linoleic acids contributes importantly to colonic tumorigenesis (35), and current data suggest that selenium and arachidonic and linoleic acid metabolism are linked in colonic tumorigenesis. Selenomethionine inhibited growth in four hu-