The experimental drug, Tykerb, worked so well that an international study of it was stopped early, in March, and all participants were offered the drug.

In the study, women who received Tykerb plus the chemotherapy drug Xeloda had no growth of their tumors for an average of 8 1/2 months. That compares to 4 1/2 months for those given only Xeloda, said Dr. Charles Geyer Jr. of Allegheny General Hospital in Pittsburgh.

He led the study and reported results Saturday at a meeting in Atlanta of the American Society of Clinical Oncology.

Tykerb's manufacturer, British-based GlaxoSmithKline PLC, paid for the study and said it would expand global access to the drug under compassionate use provisions. The company plans to seek approval to sell Tykerb in the United States and elsewhere later this year.

"This is huge," said Dr. Roy Herbst, a cancer specialist at the University of Texas' M.D. Anderson Cancer Center in Houston, who had no role in the study but has consulted for Glaxo in the past.

"The next step will be to use it in patients instead of Herceptin up front," to see whether it is more effective, he said.

Herceptin and Tykerb are members of a new generation of cancer medicines that more precisely target tumors without killing lots of healthy cells. Herceptin has been an important option for many women with advanced breast cancer, but eventually it stops working and women succumb to the disease.

Tykerb works in a similar yet completely novel way. Like Herceptin, it targets a protein called HER-2/neu, which is made in abnormally large quantities in roughly one-fourth of all breast cancers.

Herceptin blocks the protein on the cell's surface; Tykerb does it inside the cell, and blocks a second abnormal protein, too.

The benefits seemed to come without serious side effects - at least in this study of 321 women, Geyer said. Diarrhea, mostly mild, and rash were more common in women taking Tykerb.

No patients developed heart failure, but four of the 160 on the drug combination had a modest decrease in pumping power of the main chamber of the heart - side effects that also have been seen with Herceptin.

Tykerb has one big advantage over Herceptin - it's a pill instead of an intravenous drug, which should make it cheaper and easier to use, doctors said.

But Dr. Pamela Klein, a vice president at Genentech, Herceptin's maker, said Tykerb's real value may be not necessarily as a competitor. She said the drug may be even more effective in combination with Herceptin, to attack the abnormal protein from inside and outside a cancer cell at the same time. Studies are being planned to test this and other possibilities.

"Both of them together may be better than either of them alone," said Dr. Julie Gralow, specialist in breast cancer at the University of Washington and Fred Hutchinson Cancer Center in Seattle.

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.

Between 10 and 20 percent of breast cancers are advanced or have already started to spread at the time they are diagnosed. Average survival with this type of cancer is about two years.

Tykerb produced mixed results when tested in 416 patients with advanced kidney cancer. It made no difference in survival or disease progression for the group as a whole, but a subset of patients with high levels of an abnormal protein the drug targets had more time before their tumors started to grow, said study leader Dr. Alain Ravaud of University Hospital of Bordeaux, France.

Doctors at the conference also said that Sutent, a Pfizer drug recently approved for treating certain stomach tumors and advanced kidney cancer, showed promise in a small experiment involving 63 patients with lung cancer.

Tumors shrank in six patients and stabilized in another 26 roughly two months after treatment with the drug, said Dr. Mark Socinski of the University of North Carolina at Chapel Hill. However, three patients died of bleeding problems, at least one of which was thought to be related to treatment rather than the disease.