Learning from history in micronutrient research.

George Santayana, the Spanish philosopher, is credited with the often paraphrased quote that " those who cannot remember the past are condemned to repeat it. " In our most cynical moments, these words ring true as a commentary on micro-nutrient supplements for cancer prevention. Four years ago, Tim Byers (1) eloquently summarized in an editorial titled " Anticancer Vitamins du Jour: The ABCED's So Far " that chem-oprevention trials of vitamin supplements have been largely fruitless (pardon the pun). This year, John Potter (2) wrote a review, " The Failure of Cancer Chemoprevention, " which was sharply critical of chemoprevention research. Each of these essays noted that, based on results of trials thus far, a single-agent, supraphysiologic dose of a nutrient does not typically have its intended effect. Missing from the design of these sup-plementation trials is a nuanced understanding of human nutrition. There is reasonable agreement on the intakes of micronutrients that are required to prevent clinical manifestations of deficiency and on intakes high enough to cause acute toxicity; but optimal micronutrient intakes are likely within more narrow ranges than between these two extremes, and studies have demonstrated repeatedly that micronutrient intakes well below those associated with acute toxicity can increase the risks of several cancers. Evidence is now accumulating that selenium is a micronutri-ent that can cause harm at intakes well below the US Tolerable Upper Limit of 400 μg/d. Overall, clinical trials of selenium supplementation have yielded inconsistent results. In the latter half of the 1990s, Clark et al. (3) published results from the Nutritional Prevention of Cancer (NPC) trial, which was a small (n = 980 men and n = 332 women), randomized trial testing whether 200 μg/d (from high-selenium yeast) would reduce the recurrence of nonmelanoma skin cancers. Endpoints considered after the trial was completed include total cancer incidence (excluding nonmelanoma skin cancers) and cancer incidence at several common sites. Although selenium had no effect on squamous or basal cell carcinoma of the skin when the trial had stopped in 1993, it did substantially reduce the risk of prostate cancer. Specifically, there was a 86% (95% confidence interval [CI] = 39% to 97%) reduction in risk among men in the lowest tertile of baseline plasma selenium, a 67% (95% CI = 18% to 87%) reduction among men in the second tertile, and a statistically nonsignificant 14% (95% CI =-49% to 159%) increase among men in …