(The Associated Press) -- Measuring levels of a certain protein in a tumor could give doctors an astonishingly accurate way of predicting whether early breast cancer is likely to spread to the rest of a woman's body, a study suggests.

If the preliminary findings hold up, doctors could someday use a growth protein called cyclin E to tell which women need surgery plus chemotherapy and which ones just need the tumor cut out.

Cyclin E appeared six times more powerful a predictor than the current methods - measuring tumor size and how far cancer cells have spread, said biochemist Khandan Keyomarsi of the M.D. Anderson Cancer Center in Houston.

In women with early-stage cancer that showed no sign of spreading, cyclin E was 100 percent accurate in telling who would be alive six years later and who would die within that time. It was more than 90 percent accurate in those whose cancer had spread to the lymph nodes.

However, Keomarsi cautioned that the results might not be as impressive in a bigger study.

Cyclin E "could potentially be very useful in helping to identify those patients who may not need the grueling nature of chemotherapy, as well as those who should be treated more aggressively," said Keyomarsi, whose findings were reported in Thursday's New England Journal of Medicine.

Cyclin E is seen as a good marker, or predictor, because it plays a role in cell growth, and cancer is essentially cell growth gone wild. Keyomarsi said the protein could be the first reliable biological marker that can distinguish between aggressive and non-aggressive breast cancer before it spreads.

Other researchers described the research as intriguing, but they, too, cautioned that it must be confirmed through additional studies and longer follow-up of patients.

"It's very provocative, and I look forward to evolution of this particular marker," said Dr. Larry Norton, head of solid tumor oncology and director of the Breast Center at Memorial Sloan Kettering Cancer Center in New York.

Keyomarsi said earlier, conflicting studies of cyclin E used antibodies to find the protein. She instead employed a test that uses electric current to separate proteins, and said it did better in head-to-head comparisons against the antibody technique.

However, clinical laboratories generally do not have equipment for that test. Moreover, breast tumor tissue is usually embedded in wax for diagnosis, and the test that Khandan found to be superior needs fresh or fresh-frozen tissue.

That could complicate efforts to develop a test for cycylin E, since fresh tissue decays rapidly. Also, because mammograms are spotting cancer earlier, many tumors are so small there is nothing left over for analysis after the diagnosis, Norton said.

"If it turns out to be this useful as a prognostic factor, it would be worth it," Norton said. "We'll have to wrestle with ways to accomplish this."

James Roberts, a scientist at the Fred Hutchinson Cancer Research Center in Seattle, said researchers there did some early studies that suggested cyclin E is a good marker for aggressive breast cancer. But he said more standard markers of excessive cell growth may be just as good.

Keyomarsi's study looked at cyclin E levels in tissue from 395 women diagnosed between 1990 and 1995.

A total of 114 of the women were diagnosed with early cancer that had not yet spread. Twelve of them died, and all 12 had high levels of the protein. The other women had low levels of the protein, and all of them survived.

The difference was not quite as clear-cut for cancer in its later stages, and there was no difference at all for cancer that had already spread past the lymph nodes.

Generally, if cancer has not spread and the tumor is less than a centimeter - smaller than a fingernail - doctors just remove it without radiation or chemotherapy.

Copyright 2002 Associated Press. All rights reserved. This material may not be published, broadcast, rewritten, or redistributed.