The Evolution of Bird Flu, and the Race to Keep Up

On May 20, a 10-year-old girl in rural Cambodia got a fever. Five days later, she was admitted to a hospital, and after two days of intensive care she was dead.

The girl was the most recent documented victim of the influenza virus H5N1, a strain that has caused 606 known human cases and 357 deaths since it re-emerged in 2003 after a six-year absence.

H5N1 can race through bird populations, and the World Health Organization suspects the girl was infected while preparing chicken for a meal.

While humans are not ideal hosts for H5N1, bird flu viruses do sometimes manage to adapt for easy transmission from human to human, and the results can be devastating. In 1918, one such transformation led to the Spanish flu pandemic, a global outbreak that claimed an estimated 50 million people.

To better understand the possibility of H5N1 making a similar transformation, two teams of scientists recently manipulated the virus until it could spread through the air from one ferret to another. If a flu virus can infect a ferret, then it could theoretically infect other mammals, including humans.

Last fall, the scientific community became embroiled in a debate about whether the details of this research should be published; security experts, among others, feared that the information could be used to develop a biological weapon.

After months of arguments, a federal advisory board recommended in March that the results be published in full. In May, Nature published the first of the two controversial papers; now the second team has published the results of their experiments in the journal Science.

The experiments offer two examples of how H5N1 could theoretically become a human flu.

“It’s important work,” said Marc Lipsitch, a professor of epidemiology at Harvard. “The most important thing it did was put to rest the claims that H5N1 just couldn’t be transmitted between mammals. We know this is a serious possibility.”

But just because such a virus can exist doesn’t necessarily mean that it will evolve in the wild. How flu viruses cross the species barrier remains deeply mysterious. “We know that it happens, but we also know it must not be easy,” said Dr. Jeffery K. Taubenberger, a virologist at the National Institutes of Health who has studied the 1918 flu.

Dr. Taubenberger and other researchers warn that we don’t know enough about the flu to say how likely it is that H5N1 will take the two paths presented in the new papers — or a different path. “We are still woefully ignorant about how all of this works,” he said.

The new paper describes the more straightforward of the two experiments. Ron Fouchier of the Erasmus Medical Center in the Netherlands and colleagues started off with H5N1 isolated from an Indonesian patient and introduced mutations to the virus’s genes. They then transferred the virus from one ferret to another, allowing more mutations to accumulate. After five mutations, viral strains emerged that were able to spread from one ferret to the next through the air.

Acquiring a series of mutations is also probably how a bird flu gave rise to the 1918 pandemic. In another paper in Science, Derek Smith, an evolutionary biologist at the University of Cambridge, and his colleagues investigated how easy it would be for natural H5N1 to acquire the five mutations identified by Dr. Fouchier.

They found that two of the mutations have arisen numerous times in the wild. In one group of closely related H5N1 viruses, both mutations are present in their genes, so they would need to acquire only three more mutations to spread between mammals.

To judge the likelihood of that, they developed a mathematical model based on what is known about flu viruses.

A flu infection can produce 100 trillion new viruses in a matter of days. And because these viruses have a high mutation rate, just about every new virus will be slightly different from its ancestor. Dr. Smith and his colleagues found that under plausible conditions, there was a small chance that the flu viruses could gain the final three mutations in a single person during a single infection.

But figuring the exact odds is tricky.

“It’s possible that the chances are one in a thousand, and we’ve just gotten lucky,” Dr. Smith said. “Or it’s possible that the chance is one in a million and it might not happen for a long time.”

The Nature paper, by Yoshihiro Kawaoka of the University of Wisconsin and colleagues, offers a different path to a human form of H5N1 — a path based on a kind of viral sex.

Sometimes two viruses simultaneously infect a host cell. The cell makes new copies of both viruses’ genes, and they are mixed together as they are packaged into protein shells.

This process, called reassortment, occurs every year when different human flu strains infect the same person at once. It also happens — more rarely — between viruses that normally infect different species. Over the past century, the 1918 strain has periodically mixed with bird flu viruses, leading to new pandemics.

Dr. Kawaoka and his colleagues combined human flu genes with an H5N1 gene that directs production of a protein the virus uses to enter host cells. They found that after this reassortment, it took only four mutations for the virus to spread between ferrets.

No one knows the probability that H5N1 would reassort with a human flu. All researchers know is that a hybrid between H5N1 and a human flu has not been discovered — yet.

“We really don’t talk about it, because so little is known about the probability of it,” said Dr. Smith.

This ignorance is nothing new when it comes to flu. In a commentary in Science, Dr. Lipsitch and his colleagues point out just how often flu evolution delivers us unpleasant surprises. Resistance to antiviral drugs like Tamiflu, for example, arose without warning.

Dr. Lipsitch doesn’t take much comfort from the fact that the engineered viruses are relatively mild in ferrets. The scientists can’t explain why their viruses are mild, and if they escape they might evolve further, becoming both easily spread and quite deadly.

While scientists have offered two possible ways in which H5N1 might become a human flu, they’re almost certainly not the only two. There is no checklist of mutations that any bird flu must acquire to start infecting humans.

Dr. Anthony S. Fauci, director of the National Institute of Allergy and Infectious Diseases, hopes scientists will be able to amass a longer list of potential mutations, and even find a common denominator in how they alter H5N1. It might then be possible to monitor emerging strains for signs that they are about to cross over into humans.

“That may be the real home run for surveillance,” he said. “I don’t know that, but the only way you find out is by accumulating knowledge and making it available to the scientific community.”