Sections

Services

Information

UK United Kingdom

Scientists create accurate predictor of the next flu virus

Influenza viruses rapidly evolve in shape, making it hard to develop protective vaccines against them. Despite a great deal of study, scientists have been at a loss to forecast their evolution in any detail…

That crystal ball we ordered? New flu model helps predict future strains. April Mo, CC BY-SA

Influenza viruses rapidly evolve in shape, making it hard to develop protective vaccines against them. Despite a great deal of study, scientists have been at a loss to forecast their evolution in any detail for decades. Now, thanks to improvements in our ability to study viruses and a new mathematical model, anticipating influenza’s next move appears possible.

Making the jump

Influenza presents two kinds of evolutionary challenges. One is its ability to jump from animals to humans and spark global pandemics, for example swine flu which killed thousands from Mexico to China in 2009.

But a pandemic does not always follow from these jumps. For example there have been reports of humans being infected with the H5N1 bird flu, without it spreading en masse. Researchers are still trying to understand the reason why some influenza viruses are unable to spread, while others have taken off.

Year to year

One of the successful jumps was made in 1968 by a subtype called H3N2. H3N2 evolves quickly enough that its entire population is replaced every few years. It exemplifies the second type of evolutionary challenge: predicting, from year to year, which of the many circulating strains will take over.

Because influenza is a major cause of death as a result of pneumonia, understanding this evolution is more than an academic exercise. Predicting which strain will be widespread nearly a year into the future is the central challenge of the World Health Organisation’s vaccine strain selection committee. It meets twice a year to review the evidence on circulating viruses and pick the likeliest candidates for the next flu seasons.

The problem is that it is difficult to build a compelling case against any one strain. Influenza’s rapid evolution in humans is partly driven by its fight to survive against our immune systems. Viruses with mutations that allow them to escape antibodies tend to spread faster, leaving a trail of immunity that helps drive their weaker ancestors to extinction.

Slow and costly

But measuring how well different strains have escaped immunity has been a slow and costly science. Traditionally, ferrets are experimentally infected with common strains to see how well they develop antibodies that react with the candidate vaccine strains. These are then compared with measures from human samples. Often, several strains show some ability to escape immunity.

For more than a decade, different groups have attempted to find genetic shortcuts to predict the winners in advance. Influenza researchers have long known that mutations in certain parts of the virus are more likely to lead to immune escape than others. But in the past, every time a rule was derived, the virus seemed to break it.

Over time, however, this pattern gradually enforced a recurring theme in evolution: the impact of a mutation depends heavily on the genetic background in which it occurs. For fast-evolving viruses such as influenza, the combinatorial possibilities of mutations and backgrounds has made prediction seem like a daunting task.

A new model

But a recent study published in Nature shows that, in the case of H3N2, we perhaps can predict its evolution after all. The study authors, Marta Łuksza and Michael Lässig, showed that the future success of related H3N2 strains, known as clades, could be predicted by a relatively simple model.

The model considers only three types of information when assessing a clade’s future: mutations in sites that bind antibodies (generally thought to be beneficial), mutations in sites not binding antibodies (generally thought to be harmful), and the recent frequencies of the clade and competing clades. Most powerfully, the authors showed the model can be used to predict strain frequencies on a time scale useful for creating vaccines. This could greatly increase the effectiveness of flu jabs in protecting against the virus.

In addition to its ability to select strains, the model also reveals important information about the way influenza evolves. The immunity acquired by host populations from infections with different strains shapes the virus' evolution. This suggests that widespread vaccination could shape the evolution of influenza.

The study also contributes evidence to the idea that strains emerging from Asia tend to be inordinately successful. Why this is, and whether the trend will continue, are unanswered questions. Finally, the model reveals that as it evolves, influenza follows a narrow path between beneficial mutations to escape immunity and harmful ones that affect its functional stability.

How do we know the virus won’t break the rules of this model too? In a way, we don’t, but the authors took steps to show that their model balanced the trade off between complexity and predictive power. Their ability to find this sweet spot comes entirely from the large and growing number of publicly available influenza sequences. It gives hope that other evolutionary challenge, such as predicting whether H5N1 or H7N9 might make the jump, could be solvable too.