A low-fat diet can decrease the risk of breast cancer recurrence by more than 40% in patients with a form of the cancer that is not sensitive to levels of the hormone estrogen, researchers said Monday.

Those patients account for a third of all breast cancer cases. The other two-thirds — those whose tumor growth is stimulated by estrogen — showed little benefit from a reduction in fat intake, Dr. Rowan T. Chlebowski of Harbor-UCLA Medical Center told a meeting of the American Society of Clinical Oncology in Orlando, Fla.

Some researchers hailed the findings as the first evidence that lifestyle interventions could reduce the risk of developing cancer.

"Doctors have been nihilistic about trying to do lifestyle changes because they are notoriously difficult to effect and to measure," said Dr. Robert J. Morgan Jr. of City of Hope Medical Center in Duarte.

Although the findings need to be confirmed, he said, the study shows that "doing interventions on lifestyle changes is effective…. The concept has been demonstrated."

Other researchers cautioned that the results were not definitive, that the beneficial effects of the diet might have resulted from an increased consumption of fruits and vegetables or from weight loss associated with changed eating habits. On average, the women on the low-fat diets lost about 4 pounds.

Other studies have shown that losing weight can reduce the risk of recurrence, said Dr. Gabriel N. Hortobagyi of the M.D. Anderson Cancer Center in Houston, so it is important to determine which factor was most important. "After all, it's easier to lose weight than to reduce the amount of fat in your diet," he said.

Chlebowski acknowledged that "we can't separate those components out," but he argued that the change in diet was beneficial regardless.

Dr. Len Lichtenfeld of the American Cancer Society, responding to critics' skepticism, said that "if this kind of reduction were found in a preventive chemotherapy study, it would be the cause of much excitement."

For now, he said, women with breast tumors that are not sensitive to estrogen "should consider adopting a low-fat diet."

About 211,240 women in the United States will be diagnosed with invasive breast cancer this year, with about 40,410 of them dying from the disease, according to the American Cancer Society. There are more than 2 million women in the U.S. who have been treated for breast cancer.

The putative link between dietary fat and breast cancer has a long and checkered history. Evidence of the link has come from animal studies, analyses of breast cancer rates in different nations, and U.S. comparisons of women who did and did not develop breast cancer.

But a major epidemiological study from Brigham and Women's Hospital in Boston published in 1999 found no firm link, even hinting that a low-fat diet might increase the risk of breast cancer.

Studies on both sides of the issue, however, have generally involved questionnaires in which women described their diets after the fact.

The new study, sponsored by the National Cancer Institute, is the first prospective, randomized, controlled study — generally considered the gold standard for teasing out facts.

Chlebowski and his colleagues at 37 hospitals enrolled 2,437 women, ages 48 to 79, who had been successfully treated for early-stage breast cancer. Two-thirds were told to follow their normal diet, and the rest were given intensive nutritional counseling. The diet of the women in the control group averaged about 51.3 grams of fat per day, whereas the counseled group averaged 33.3 grams per day.

After an average of five years, 9.8% of the women with the low-fat diet had a recurrence of their cancer, compared with 12.4% of those on the standard diet.

Closer analysis showed that virtually all of the improvement came in the women whose tumors were not sensitive to estrogen. They had a 42% reduction in recurrence, whereas those whose tumors were hormone sensitive did not have a statistically significant reduction.

That finding is counterintuitive, Hortobagyi said. "If you had asked me before the results were presented, I would have predicted that the only beneficial effect would have been in" estrogen-sensitive tumors, for which a clear link to obesity has been shown.

"That makes them very interesting, but not terribly reliable until we see confirmation," he said.