The Future Evolution of Bird Flu

It’s been a rough flu season this winter in the United States and Europe, but it could be worse. A lot worse. The flu viruses that are making us sick go by names like H1N1 and H3N2, referring to the kinds of proteins that stud their surface. There’s another sort of flu lurking in other parts of the world, like Egypt, India, and Cambodia, known as H5N1. Since 2003, 615 people have come down with H5N1, and, as of Feburary 1, 364 of them had died. In January alone, 5 people in Cambodia were diagnosed with H5N1. Four of them died.

There’s a lot of debate about precisely how bad H5N1 is. It’s possible that a lot of people are getting sick with H5N1 without making it onto the official records. They’re crawling off to bed for a week, recuperating, and then getting on with their lives. So the 59 percent death rate you get from the official numbers (what’s known as the case-fatality rate) may be a serious overestimate. Still, even if the true rate was only half as high, H5N1 would not be a virus you’d want to cross paths with. The most famous flu outbreak of all, the so-called Spanish flu of 1918, is estimated to have killed 50 to 100 million people worldwide. But it infected billions, with a death rate of roughly 2 percent. If H5N1 could somehow take off and become a global pandemic, it would become an unparalleled catastrophe even if its official 59 percent case-fatality rate was chopped down by a factor of ten.

Right now, H5N1 does not have what it takes to race around the world. It might someday, although nobody can say for sure what the odds are. And, as a team of scientists now report in the Philosophical Transactions of the Royal Society, H5N1 would continue to evolve as it spread. Its ability to spread would evolve, along with its ability to kill. The scientists can’t look into their crystal ball and say for sure how many people would die. But they can say this: what we do when and if we face an H5N1 pandemic could alter the evolution of the virus itself. And thousands of lives could be saved or lost as a result.

All of the flu strains that make us humans sick get their start in birds. Our feathered friends harbor a huge diversity of flu viruses in their guts, which they shed in their droppings. Most bird flu strains are fairly harmless to the birds themselves–they don’t overhwhelm their immune system, in other words. Most of them can’t infect animals other than birds. But on rare occasion, a strain of bird flu evolves that can also latch onto cells in another species, such as seals or humans.

That’s what H5N1 has done. In at least some of the people who are exposed to it–from their chickens or from wild waterfowl–H5N1 can latch onto cells in the airway and invade. What H5N1 can’t do is spread from one person to another. There may be a number of reasons for this. A successful human flu virus needs to find a way to enter human cells quickly, it needs to hijack human biology to make new viruses, it needs to escape human immune attacks, and it needs to be able to spread efficiently from one person to the next.

Exactly what it would take for H5N1 to become a true human flu got two teams of scientists in hot water for back in 2011. They introduced mutations into H5N1 and then infected ferrets with it. (Ferrets get sick from the flu a lot like we do, so they’re pretty good models to study.) They then swabbed viruses from the noses of the ferrets and then infected healthy ferrets. Within several rounds of this transfer, the viruses evolved the ability to spread on their own between ferrets through the air. Intriguingly, the H5N1 viruses became much less dangerous along the way. In its mammal-adapted form, H5N1 didn’t kill any of the ferrets.

The prospect of a mammal-adapted H5N1 was so alarming that the researchers agreed to a moratorium while the scientific community debated the risks and benefits of the research. At the end of January, the moratorium came to an end, and research will continue in countries that have established clear guidelines for studying the virus.

Their experiments revealed a fairly short list of mutations that were sufficient to make H5N1 a mammal flu. That was a sobering discovery, because H5N1 doesn’t depend on scientists to acquire new mutations. It’s mutating all the time in the wild.

In the February issue of EMBO Reports, three H5N1 experts, Yohei Watanabe, Madiha S. Ibrahim, and Kazuyoshi Ikuta, describe how H5N1 has been evolving as it has spread to new countries. In Egypt, for example, the virus first emerged in 2006, and doctors there have documented 169 infections of humans so far. Since then, Egyptian viruses have evolved that are better adapted to infecting humans. Their new adaptations include proteins that let them latch more tightly to human cells than their ancestor could. Watanabe and his colleagues suggest that this evolution may explain why Egypt has had 50 percent of all the human cases of H5N1 since 2009. They warn that Egypt may now have the highest potential for an H5N1 pandemic.

Let us, then, imagine the worst. A chicken pecking in the dirt on the outskirts of Alexandria picks up the flu. Inside its gut, the viruses replicate, picking up some crucial mutations for infecting humans, and are then shed in the chicken’s droppings and blown around in the dust.

An unlucky farmer breathes in the virus, which infects his airway and leaves him deathly ill. Inside his body, the H5N1 further mutates, and one of viruses now has the wherewithal to fly out of his body and infect his wife as she changes his bedding. It starts spreading from village to village, from Egypt to other countries, until we are in the grip of a global H5N1 epidemic.

A group of flu experts led by Maciej Boni of the Oxford University Clinical Research Unit in Vietnam have now tried to make some rough predictions about that scenario. They didn’t simulate viruses infecting millions of people. Instead, they created mathematical models based on the observations doctors have made on H5N1 infections. Unlike most epidemiological models, theirs takes into account the fact that viruses never stop evolving.

As with other pathogens, natural selection sculpts the genes of flu viruses in many different ways. On the one hand, natural selection should favor mutations that speed up the replication of viruses in their host. But over the long term, flu viruses also have to get from one host to another. Those two demands can sometimes come into conflict–if faster replication leads to less spreading from host to host, for example. This tradeoff can arise if, for example, fast-replicating viruses also tend to be deadlier. Kill of your hosts too soon, and you ultimately kill off yourself.

A flu virus like H5N1 experiences a special form of this tradeoff. In their current form, H5N1 grabs more tightly to receptors in the lower respiratory tract than the upper respiratory tract. A number of flu experts suspect that this preference is what makes the flu so deadly, because it’s so far down the airway that it can wreak more havoc with the lungs. Because the virus is so deadly, it has less time to spread to a new host. And being nestled so deep in the respiratory tract also means that relatively few of the viruses can leap up the airway and escape their host.

Human flu viruses have this tradeoff flipped. They tend to latch onto cells in the upper respiratory tract. They’re less deadly than H5N1, but they also have an easier time getting sneezed or coughed into the air. Instead of dying in a few days, people with human flu are far more likely to remain alive and infective. They can even walk around, transporting their viruses to stores, schools, and subways.

Boni and his colleagues created a model of an H5N1 outbreak spreading through a population of 10 million people. In their model, the virus could evolve a stronger or weaker preference for the lower or upper airway. That preference would influence how deadly it was and how easily it could spread.

They also included the human factor. When flu outbreaks occur, one of the most effective strategies is to keep people away from each other. No school, no movies, no parades. Boni and his colleagues analyzed the course of the outbreak if this isolation did or didn’t happen. They could examine the difference between a brief “flu weekend” or a long shut-down.

First the scientists looked at what it would take for a human-adapted H5N1 virus to trigger a full-blown pandemic. If it had the 60 percent case-fatality rate that doctors have seen in bird flu, it would fail. People would die so fast that the virus wouldn’t have enough time to replicate and spread to other people.

Sounds great, right? Not so fast. If the mortality rate dropped to roughly 15 percent, the scientists found, the flu would take off. Their model suggests we might face a fast-spreading virus that was far deadlier than the 1918 pandemic. Boni and his colleagues point out that in Egypt, the fatality rate of H5N1 is dropping to about 36 percent. They call the drop “worrying.” Normally, you’d think a milder pathogen is a better one. But thanks to the quirks of the flu, it might actually be entering its pandemic sweet spot.

In the early part of an epidemic, natural selection will favor viruses that can spread quickly. They can do so if they evolve the ability to colonize the upper airway and get coughed onto new hosts. But this preference is not an all-or-nothing choice. As the flu shifts to the upper airway, it could still continue to infect the lower airway, albeit at a lower rate. So the virus could continue to cause a lot of deaths even as it began spreading faster through the human population.

As the outbreak progressed, evolution would shift in a new direction. What started out as one huge population of vulnerable people, would become more complex: a population in which many people had already been exposed and fought the virus off, some people had died, and some were in hospitals. Now it would take longer for the virus to successfully jump from host to host. Instead of transmitting quickly, natural selection would favor viruses that were less deadly, but also able to cause longer-lasting infections.

Public health officials like to isolate people from each other in order to drive a virus to extinction. If it gets too hard for viruses to find new hosts, they eventually die out like a sputtering fire. But they usually don’t consider the fact that this isolation can also prompt the virus to evolve in important ways.

If people are suddenly isolated from each other, many of the viruses will die out. Selection will become intense for viruses that can survive longer. Instead of killing their hosts, they will make them ill for many days. The viruses may be able to survive the isolation; once life returns to normal, the flu reemerges and triggers a second epidemic–albeit a milder one.

It’s not easy to get people to stay home for very long, so public health workers may try to establisher an isolation plan that’s less strict but lasts longer. Under these conditions, the virus may not die out quickly, but selection may produce tamer forms. As a result, most people who do get sick won’t face as grave a risk as before.

This study is too preliminary to serve as a guide for dealing with an outbreak. But it does help to sharpen our senses to the potential that H5N1 has to evolve into something new. Whether the surprise is pleasant or horrifying may be, in part, up to us.

(For more information on viruses, see my book, A Planet of Viruses.)

[Update 2/8: A few tweaks made at the suggestion of Boni.]