Evidence Still Weak For Cancer Prevention By Nutrition

BETHESDA, Md., July 19 -- Dietary supplements such as antioxidants, vitamins, and garlic do little if anything to prevent cancer, but antibiotics aimed at cancer-causing bacteria appear to prevent some tumors.

Those are the conclusions of two studies reported in the July 19 issue of the Journal of National Cancer Institute - one a large, long-term randomized trial in pre-cancerous gastric lesions and the other a meta-analysis of studies in a range of cancer types and pre-cancerous lesions.

Researchers here and in China conducted a randomized, placebo-controlled, factorial study, comparing three possible chemopreventive treatments for gastric lesions in a region known for a high incidence of gastric cancer, according to Mitchell Gail, M.D., Ph.D., of the National Cancer Institute here.

The study, carried out in the Chinese province of Shandong, began in 1994, when the researchers enrolled 3,365 participants, ages 35 to 64, and randomly assigned them, in a factorial fashion to three interventions or placebos:

• Amoxicillin and omeprazole for two weeks in 1995, to treat Helicobacter pylori infection
• A vitamin supplement consisting of vitamin C, vitamin E, and selenium, for 7.3 years
• Or a garlic supplement, consisting of an aged garlic extract and steam-distilled garlic oil, for 7.3 years.

The participants underwent a baseline endoscopy, with follow-up procedures in 1999 and 2003. At baseline, they included all histologic categories except gastric cancer, although only 0.18% had normal gastric mucosa.

The study found that H. pylori treatment resulted in statistically significant decreases in the combined prevalence of severe chronic atrophic gastritis, intestinal metaplasia, dysplasia, or gastric cancer in 1999 and again in 2003. The odds ratios were 0.77 in 1999, with a 95% confidence interval from 0.62 to 0.95, and 0.60 in 2004, with a 95% confidence interval from 0.47 to 0.75.

The treatment didn't reduce the combined prevalence of dysplasia or gastric cancer, although 19 of the 1,130 participants getting antibiotics (1.7%) developed gastric cancer compared to 27 of 1,128 getting placebo (or 2.4%). The trend was not significant, however.

On the other hand, no significant favorable effects were seen for garlic or vitamin supplements, the researchers reported.

The randomized trial had outcomes similar to that seen in a meta-analysis of 59 studies, in that the vitamin supplements and antioxidants did not appear to have a consistent beneficial effect.

Researchers led by Steven Thomas, M.D., Ph.D., of the University of Bristol, in England, identified 25 studies in patients with cancer and 34 in patients with pre-invasive lesions, but overall there was no association between either the antioxidants or vitamin A supplements and all-cause mortality.

On the other hand, Dr. Thomas and colleagues found a trend toward benefit for a "healthy diet." The combined odds ratio for such a diet -- alone or with dietary supplements, weight loss, or exercise - on all-cause mortality was 0.90. However, the 95% confidence interval crossed unity, so the association was not statistically significant.

The quality of the studies analyzed was low, the researchers said, which combined with the limited number of such studies, makes it hard to estimate the effects of nutritional interventions. Nevertheless, "there is no evidence that dietary modification by cancer patients improves survival and benefits disease prognosis," the researchers concluded.

The two studies "well illustrate the contemporary status of chemoprevention," said John Baron, M.D., of Dartmouth Medical School, in Lebanon, N.H.., in an accompanying editorial. Chemoprevention is "hard to summarize, (with) many negative findings, but some hopeful nuggets of progress," Dr. Baron summed up.

He noted that Dr. Thomas and colleagues used a relatively narrow definition of quality, focusing on randomization and blinding, but "most of the negative assessments actually reflected a lack of reporting, so arguably a better characterization of the trial quality might be 'unknown.'"