SAN ANTONIO — U.S. breast cancer rates plunged an unprecedented 7 percent in 2003, the year after millions of women stopped taking menopause hormones when a study showed the pills raise the risk of tumors.

The startling new analysis, reported yesterday at the San Antonio Breast Cancer Symposium, does not prove a link between hormone therapy and breast cancer, but strongly suggests it, many experts said.

"When I saw it, I couldn't believe it," statistician Donald Berry said of the drop. Berry is with the University of Texas M.D. Anderson Cancer Center, whose researchers presented the study.

Cancers take years to form, so going off hormones would not instantly prevent new tumors. But tumors that had been developing might stop growing, shrink or disappear so they were no longer detected by mammograms, doctors theorized.

Cases dropped most among women 50 and older — the age group taking hormones. The decline was biggest for tumors whose growth is fueled by estrogen — the type most affected by hormone use.

The drop was seen in every single cancer registry that reports information to the federal government, and no big change occurred with any other major type of cancer. These are strong signs that the breast cancer decline is no statistical fluke or error.

"It's a big deal ... amazing, really," said another of the researchers, Dr. Rowan Chlebowski of Harbor-UCLA Medical Center in Los Angeles. "It's better than a cure" because these are cases that never occurred, he said.

About 200,000 cases of breast cancer had been expected that year; the drop means that about 14,000 fewer women actually were diagnosed with the disease.

A separate study by the American Cancer Society, currently in press with a medical journal, also documents the drop. Lead author Ahmedin Jemal attributes two-thirds of it to a decline in hormone use and the rest to mammography use leveling off, resulting in fewer tumors being detected.

"We are really trying to look at the big picture," he said. "You cannot rule out the effect of screening."

COMMON CANCER

Breast cancer is the most common major cancer in American women and the second leading cause of cancer deaths in women. About 213,000 new cases are expected to occur in the United States this year and more than 1 million worldwide.

Incidence in the United States rose almost 2 percent per year from 1990 to 1998, then began to slightly decrease, said Dr. Peter Ravdin, the M.D. Anderson doctor who led the new analysis.

In July 2002, the federal Women's Health Initiative study was stopped after more breast cancers and heart problems occurred among women taking estrogen-progestin pills.

That led to new warning labels on the drugs and doctor groups urging women to use the lowest dose for the shortest time possible for hot flashes and other menopause symptoms.

Within a year, about half of women who had been taking hormones stopped. Prescriptions had been steady at around 22 million each quarter, but plummeted to 12.7 million in the last quarter of 2003, according to IMS Health, which tracks drug sales.

Breast cancer rates declined, too. In 2002, there were roughly 134 cases per 100,000 women — a 2.5 percent drop from about 137 the previous year. In 2003, there were only 124 cases per 100,000 women — about a 7 percent drop over 2002. That is the most significant decline in the breast cancer rate since records have been kept beginning in the 1970s.

(Rates are used rather than raw numbers because they give a better picture of the situation as the population shifts and ages.)

Researchers saw an even stronger trend when they looked month-to-month. Cases dropped 6 percent in the first half of 2003 and 9 percent in the second half.

Estrogen-sensitive tumors declined twice as much as tumors that are not fueled by estrogen. The decline in incidence among women ages 50 to 69 was three times that of other age groups.

The numbers come from the National Cancer Institute's surveillance database, which uses cancer registries around the country to project national incidence and death rates.

CAUTIOUS RESPONSE

When the 2003 numbers were first released a few months ago, they were grouped with 2001 and 2002 and portrayed as a leveling off of breast cancer after decades of steady rise. The big single-year drop was not pointed out.

"You don't want to overinterpret one point" without knowing whether it is a trend, said Kathy Cronin, a National Cancer Institute statistician who worked on the new analysis.

"The major health organizations have been cautious because of not wanting to call attention to something of this much interest to everyone prematurely," said Dr. Michael Thun of the cancer society.

Ravdin disagreed.

"It doesn't have to be a trend to be real," he said. "Such a rapid effect is most consistent with the idea that cancers that were already there ... were actually being stopped in their growth to the point where they would not be detected."

Federal statistics for 2004 are expected in April. Information from one large registry, California's, published recently in the Journal of Clinical Oncology, hints that the trend is continuing.