WASHINGTON - Blood donated by four survivors of bird flu seems to harbor a potent protection against the deadly virus.

Scientists have long suspected that culling immune-system molecules from survivors could provide a new therapy for the hard-to-treat H5N1 flu strain. Monday, an international team of researchers reported the first evidence, albeit from tests in mice, that it really may work.

If the research pans out, it could be possible to stockpile these antibodies, the immune system's search-and-destroy force, as an additional way to treat or even prevent H5N1 in case the worrisome flu strain ever mutates to spark a worldwide epidemic.

“Obviously we're interested and excited about this potential,” said Dr. Anthony Fauci, infectious disease chief at the National Institutes of Health.

The research started when four Vietnamese adults who survived bouts of H5N1 in 2004 agreed to donate blood to the Hospital for Tropical Diseases in Ho Chi Minh City.

At Switzerland's Institute for Research in Biomedicine, Dr. Antonio Lanzavecchia created a way to cull antibody-producing cells from the blood and keep them churning out the molecules in laboratory dishes.

In the U.S., the NIH's Dr. Kanta Subbarao tested thousands of those antibodies to tease out the handful able to kill H5N1. They were purified to better target the virus.

Then came the real tests: Subbarao's lab infected mice with H5N1. Some were given the antibodies before they were exposed, others after they already were infected; still others were given antibodies that target different diseases, not influenza.

Mice given the non-H5N1 antibodies died. The H5N1-targeting antibodies protected mice, both when they were administered as a vaccine-like preventive or after infection. Importantly, they worked against both the same 2004 strain that the people had survived and against a different H5N1 strain that circulated in 2005.

The work is reported Monday in the online journal PLoS Medicine.

This approach is called “passive immunotherapy,” and more crude forms of the approach have long been used to protect against certain viruses. Before hepatitis A vaccines, for example, antibody-containing shots were common for tourists heading to developing countries.

And during the 1918 flu pandemic, the worst in history, doctors sometimes transfused blood directly from survivors to the newly sick, sometimes with good results.

The mouse study is “a very lovely, elegant proof of principle,” said Dr. William Schaffner, a flu expert at Vanderbilt University.

More work is needed before trying these purified antibodies in people. It's standard to test flu vaccines and treatments in ferrets, who respond to influenza more like people do. Then the antibodies would need testing in healthy people, to see if they're safe.

If so, they might be tried as a treatment for people still falling ill with H5N1 in parts of Asia. The only treatment now is the drug Tamiflu, which doesn't always save them.

But Schaffner points to a more immediate use: If antibodies can help the notorious bird flu, why not cull some specific to the regular, but still too often deadly, influenza that spreads every winter?

“This has the dual potential of being useful potentially in a pandemic, but perhaps more so on an annual basis,” Schaffner said. “That's where I think the real excitement is.”