The selenium surprise.

"It's the most exciting finding we've ever had in nutrition and cancer," says Tim Byers, professor of preventive medicine at the University of Colorado in Denver.

That's quite a lot to say about a study that "failed."

More than a decade ago, researcher Larry Clark -- then at Cornell University -- launched a study with colleagues Gerald Combs and Bruce Turnbull. Their goal: to see if supplements of the trace element selenium could keep skin cancers from recurring in people in the Southeastern U.S.(1)

It didn't.

"Selenium supplementation had no effect on the recurrence of skin cancer," said Clark, now with the Arizona Cancer Center at the University of Arizona in Tucson. "But to incidence, total cancer mortality, and the three leading sites of cancer -- lung, prostate, and colon -- all decreased."

By roughly half.

"It's been a surprise to everyone that the effect was so big," says Byers. He served on the safety committee that stopped Clark's study in early 1996, almost three years before it was due to end, because the selenium group fared so much better than the group taking a (lookalike but inactive) placebo.

But Byers and Clark are cautious.

"Selenium isn't a miracle drug," says Clark. "It's not going to help everybody, and it's not going to cure all cancer."

Among the limitations: The study was done in the Southeast, where many people don't get much selenium from their food, because it's grown in selenium-poor soil.

"In parts of the country where people get more selenium from their food, the effect may be less dramatic," says Clark.

THE GOLD STANDARD

It's not just the drastic drop in cancer that sets Clark's results apart. It's the type of study that found the drop. Most studies "observe" what people eat or the supplements they take and then search for links with different diseases.

"The disadvantage of observational studies is that there may be differences in the lifestyles, diets, or medical histories of people who choose to eat a diet that's high or low in a nutrient," says Julie Buring, an epidemiologist at Harvard Medical School. And those differences -- not the nutrient -- could explain their different risks of disease.

Clark's study was a randomized clinical intervention trial. That means he randomly assigned people to take either selenium or a placebo, and then wait@ ed to see how many people in each group got cancer.

"The strength of a clinical trial is that the only difference between the groups is that the people in one get selenium and the people in the other don't," says Buring. So it's unlikely that anything other than selenium explains the differences in their risk.

That's clinical trials are considered the "gold standard" of studies.

TOO GOOD TO BE TRUE?

Nevertheless, some researchers are skeptical of Clark's results. "From what we could see, there wasn't anything wrong with the study," says Walter Willett, who chairs the nutrition department at the Harvard School of Public Health.

"But it's almost too good to be true," he adds. "If the effect is this large, it would be more important than anything else we know about in cancer prevention."

Willett is also troubled by how quickly cancer rates dropped in the selenium-takers. "It's not impossible, but it's pretty unusual to see cancer rates go down right away."

While selenium's ability to prevent tumors in animals is impressive, studies on people -- at least the few that have been done -- have yielded mixed results.(2-5)

"The observational studies on breast cancer very consistently show no benefit of selenium," says Willett. In fact, Clark didn't see any drop in breast cancer risk in his selenium-takers, though there were too few cases to be certain.

"The data from observational studies are somewhat supportive for selenium's protecting against lung cancer, so that wouldn't be too surprising," says Willett. "For prostate and colon cancer, there's little evidence one way or the other."

Selenium may not have appeared to reduce the risk of cancer in some earlier studies because there wasn't enough difference between high-selenium and low-selenium intakes. "It could be that you only see the effect of selenium when there's a starker contrast," says Byers. "This trial doubled the participants, intakes."

How might selenium work? Researchers used to think that its antioxidant properties could prevent cancer. But there's a newer theory.

"Cancer cells are more sensitive to selenium-induced programmed cell death," explains Clement Ip, a selenium expert at Roswell Park Memorial Cancer Center in Buffalo, New York.

"So it's possible that selenium eliminates early precancerous lesions." But, he adds, "we need more evidence from intact organisms, not just cell cultures."

ONCE ISN'T ENOUGH

Scientists may disagree about the odds that selenium prevents cancer, but they all agree on one thing: "This trial demands follow-up," says Clark.

Among other things, Clark's study can't provide a definitive answer because it was done on one specific population-patients with non-melanoma skin cancer in the Southeast who don't get much selenium from their food.

"If the study had been done on a more-affluent, health-conscious group like physicians, it might have found nothing because they may get plenty of selenium," says Clark.

But since the risk of cancer dropped quickly in the selenium-takers, a second trial wouldn't take too long. "We can think about doing trials that would last four years or so," says Byers.

What to do until then?

"We don't have enough evidence to make public health recommendations," says Clark. "But if informed people wanted to supplement themselves as a precaution, I wouldn't discourage them."

Clark and others add two caveats: "People have to know that selenium's benefits haven't been confirmed and it may end up that the effects aren't there," says Byers.

What's more, says Ip, "we have to caution people not to indiscriminately take selenium, because there's a danger of overdosing."

TOO MUCH IS TOXIC

In Clark's study, people were given 200 micrograms (mcg) a day of selenium-that's the upper end of the "Estimated Safe and Adequate Daily Intake" set by the National Academy of Sciences for everyone.

"At levels of about 1,000 mcg -- or 1 milligram -- day people start to lose their hair, they lose their fingernails, and they get gastrointestinal problems," says Ip.

People who ignore those signs can get a lethal overdose. Everybody's scared about the results of this study because you could kill yourself with selenium," says Byers.

Trying to get more selenium from foods is a bit tricky because levels vary with the selenium content of the soil in which the foods were grown. "Most grains are a good source," Byers adds, "because most of our grains are grown in the upper Great Plains, which have a fairly selenium-rich soil."

A selenium or multivitamin supplement with no more than 200 mcg of selenium doesn't worry most experts. Our study shows that this dose is pretty safe," says Clark.

But, he adds, the source of the selenium used in his study was yeast raised on a high-selenium diet (see illustration). In contrast, "most of the selenium on the market is inorganic sodium selenite. And sometimes they throw sodium selenite into yeast and bottle it as high-selenium yeast."

Clark used high-selenium yeast because, just like the selenium in food, it is mostly incorporated into an amino acid called methionine.

And in earlier studies of a malnourished population in China, a supplement of high-selenium yeast (along with vitamin E and beta-carotene) seemed to lower the risk of stomach cancer, while a multivitamin with sodium selenite did not.(6)

A WORD OF CAUTION

How does it feel to maybe -- just maybe -- stumbled upon something do could slash cancer rates for millions of people? Clark calls himself "excited and cautiously optimistic."

His caution is well-founded. It doesn't take much imagination to envision headlines like "Another-Flip-Flop" if a future study disagrees with Clark's.

"I don't think anyone should get carried away," says Harvard's Walter Willett. "These results are interesting, and it's possible that they're right but we need to see if they're reproducible."

1. J. Amer. Med. Assoc. 276: 1957, 1996.

2. European J. Cancer Prevention 4:91, 1995.

3. Cancer Research 53:4860, 1993.

4. J. National Cancer Inst. 85:224, 1993.

5. J. Amer. Med. Assoc. 264: 1128, 1990.

6. J. National Cancer lnst. 85: 1485, 1492, 1993.