The severe disease that H5N1 avian flu provokes in people appears to be caused by the virus's ability to replicate at unusually high levels for a prolonged period— an overwhelming assault that triggers a massive and devastating immune system response, a new scientific paper suggests.

Interrupting that process before it reaches the tipping point is critical, say the authors of the article, who based their observations on the detailed study of 18 H5N1 patients in Vietnam.

But experts unrelated to the research wonder whether the current anti-flu arsenal — mainly neuraminidase inhibitors such as Tamiflu— is equipped to do that job.

"The question I worry about is whether treating patients now with neuraminidase inhibitors may be very much like shutting the barn door after the horse is already out," said Dr. Michael Osterholm, director of the Center for Infectious Diseases Research and Policy at the University of Minnesota.

"I don't know that. But we've got to find that out."

The World Health Organization is in the process of setting up a research network across countries afflicted with H5N1, in the hope that joint studies will answer questions like the one Osterholm poses.

Dr. Frederick Hayden, an antiviral expert who heads the effort, said the first task will be to investigate whether different dosing regimes for Tamiflu might achieve better results. They alsohope to look at whether injectable forms of drugs are more effective at combating H5N1 infection.

The paper,published Monday in the journal Nature Medicine, is the first detailed record of the disease process or pathogenesis of H5N1 in humans. The observations are based on 18 H5N1 patients who were treated in Vietnam during 2004 and 2005. Thirteen of them died.

The researchers charted the way H5N1 replicated in those cases and how their immune systems responded. The massive amounts of virus generated triggered production of excessive levels of some chemokines and cytokines, chemicals used to attract white blood cells— the body's cleanup crew— to the site of infection.

This hyperactive immune response— called a cytokine storm —actually does more damage than it sets out to fix. And scientists aren't clear on how to suppress the damaging parts of that process without disabling the immune system's ability to fight the virus.

Lead author Dr. Menno de Jong suggested the key is early treatment with antivirals, perhaps combined with drugs that modify the immune response.

"The paradigm 'hit hard and hit early,' which has been used for treatment of HIV-AIDS in the past, may well be very true for the treatment of avian flu," said de Jong, a physician and virologist with Oxford University's Clinical Research Unit at the Hospital for Tropical Diseases in Ho Chi Minh City.

"Considering the likely role of the immune response in causing damage during the later stages of infection, it obviously makes sense to think about the use of anti-inflammatory or immune-modulatory treatment," he said in an e-mail interview.

Both de Jong and Dr. Anthony Fauci, director of the U.S. National Institutes of Allergy and Infectious Diseases, said more needs to be known about how the virus interacts with the immune system of a human host totime the use of drugs.

"To me that's almost a catch-22," Fauci said in an interview. "Because if you know the person is infected and has virus replication only when they become symptomatic, then depending on how long that peak lasts, will you or will you not be able to have a substantial impact on it by [administering] antivirals?"

Finding that optimal point may not be easy, he suggested, because influenza replication typically peaks as people are becoming symptomatic and the immune response is already kicking into gear.

Osterholmsuggestedneuraminidase inhibitors like Tamiflu may be better suited to preventing H5N1 infection rather than treating the ensuing disease once it's apparent someone is infected. If so, he advocates stockpiling drugs to protect health-care workers during a pandemic rather than for treatment purposes,as stressed by the pandemic plans of many governments, including Canada's.

De Jong's paper also clearly underscores the differences between the disease provoked by H5N1 and that caused by human flu strains.

It notes, for instance, that swabbing the pharynx to look for virus is a more effective way to diagnose H5N1 than taking a nasal swab. With human flu strains, the reverse is true.

The researchers also found traces of the virus in the blood and rectum of critically ill patients. While that could be a hint the virus replicates in organs outside the respiratory tract and the lungs, it isn't proof positive, de Jong said. Autopsies of H5N1 patients would be needed to determine that, butfew have been performed because of cultural objections.