All products featured on WIRED are independently selected by our editors. However, we may receive compensation from retailers and/or from purchases of products through these links.
In 1962, the Nobel prizewinner McFarlane Burnet opened the latest edition of his book Natural History of Infection Disease with the kind of grandiose statement that, sooner or later, is bound to make even the most famed immunologist look a little naive.
“At times one feels that to write about infectious disease is almost to write of something that has passed into history,” he wrote. At the time, Burnet had a point. Thanks to widespread vaccination regimes, smallpox – which by some estimates killed 300 million in the 20th century alone – was just 17 years away from total eradication. In the preceding 20 years vaccines for diphtheria, tetanus, polio and whooping cough had become routine in many countries, dramatically reducing the prevalence of deadly diseases that had once been widespread.
But for all of Burnet’s bravado, one infection has remained stubbornly difficult to get under control through vaccination. Flu. In the United States, last winter’s flu was particularly severe, leading to 80,000 deaths, the deadliest season since 1976, according to a preliminary figure from the Centers for Disease Control and Prevention.
When it comes to flu, vaccination is no guarantee of protection. Last winter in the UK, only 15 per cent of people who had the vaccine were protected against infection. In good years – such as the 2015/16 season – the percentage of people protected can be as high as 52 per cent, but often the figure hovers around 40 per cent or lower. But now scientists are racing to re-engineer the flu vaccine so it protects everybody, all the time.
At the moment, devising flu vaccines involves a surprising amount of luck. Twice a year the World Health Organisation meets to decide which flu strains will be protected against by that year’s vaccine. The resulting mixture – which usually protects against four strains – tends to read like a particularly exotic and unappetising cocktail.
This year’s flu vaccine, which was finalised way back in February to give pharmaceutical firms enough time to manufacturer and distribute it, is a good example. Each flu strain is named after the place it was first identified, and this year’s shot defends against strains from Michigan, Singapore, Colorado and Phuket, because these were identified as some of the most commonly circulating strains in the previous winter.
But a lot can happen between February and October – when the flu season in the northern hemisphere starts ramping up. Older flu strains that have been left out of this year’s vaccine might return, or as-yet-unidentified strains may rear their heads. Since each vaccine within the flu shot cocktail only protects against one flu strain, when either of these things happen the number of people protected by a vaccine starts plummeting.
Enter the universal flu vaccine. The hope is to create a vaccine that in one – or at a push, two – jabs that would give people immunity against all strains of flu. “The world needs a universal flu vaccine, because it just causes so many problems, so many deaths, especially in developing countries,” says Craig Thompson, who studies the evolution of infectious diseases at the University of Oxford.
Our current flu vaccines work by getting our immune systems to respond to a very specific protein embedded in the surface of each flu virus cell. Our immune system recognises these proteins – called epitopes - and produces antibodies to clear up the infection. Vaccines contain deactivated versions of the virus so your immune system memorises what those epitopes look like, and next time it encounters the virus for real, it can quickly produce those antibodies before you get ill.
The problem is that just one of these surface proteins comes in 18 different varieties – every one provoking a slightly different type of immune response. That’s why each vaccine has to be tailored to the particular flu varieties that the WHO hopes will be circulating that season. It’s just not possible to pack in resistance to every single possible strain in a single flu shot using our current way of creating vaccines.
Read more: The startup fighting cancer with AI
Thompson thinks there might be a different way to solve this problem. Like any species, flu strains evolve over time, and because humans are relatively good at acquiring immunity to a particular flu strain, these viruses evolve fairly rapidly in order to remain infectious. “It's clever in a sense that it can evade the immunity in a population,” says Thompson. “But people haven't been able to predict how it evolves, or capture that variability in the vaccine before.”
By running mathematical models examining the way that historic flu strains evolve over time, Thompson and his colleagues have worked out that the flu changes in a predictable way, with every flu virus – and the proteins within them – going through the same four stages of evolution every ten years or so. This brings the number of variations that the vaccines can be targeted towards down from 18 to four.
And four strains is a small enough number to squeeze into one or two flu shots, Thompson says. “Then you'd have lifetime to twenty years immunity, which is currently the way we're going forward with it,” he says. Although Thompson and his team have already completed initial trials in mice, they are getting started on the process of developing this new technique into a full flu vaccine to be tested in humans.
Another Oxford company called Vaccitech is hoping to build a universal vaccine by targeting the proteins within the flu virus. Off the back of a £20 million funding round led by Google ventures, the firm is nearing the end of a two-year trial involving 2,000 participants.
But other researchers think that rather than starting by looking at vaccines, we should be thinking about our biological makeup to work out why vaccines protect some people and not others. “How do we use technology to figure out how we are different as individuals, and then use that knowledge of that difference to make these vaccines more specific and effective?” says Niven Narain, the co-founder and CEO of Berg, a Boston-based pharmaceutical startup that uses AI to develop new drugs.
“You have to fundamentally understand the biology of a population before you can embark on these types of projects,” he says. Along with the French drug firm Sanofi, Berg is using AI to to compare people who were protected against flu by vaccines with those who weren’t. Narain thinks that by studying particular biomarkers in individuals, we might be able to tailor vaccine to particular populations.
Narain likens immunity profiles to blood types. “Basically we would all have a certain immuntype, and based on your immunitype you might get vaccine a, versus vaccine b or c,” he says. He’s already trying to work out whether the presence of particular proteins in the bloodstream is correlated with vaccine success, and whether that might be enough to influence what vaccine someone could get.
“AI's not going to come in and solve the world's problems. It's going to help us get better, it's going to help us be more specific and precise, it's going to guide us to areas of biology that we haven't learned about before,” Narain says. He’s hoping that by looking at human biology as well as the flu Virus, we’ll be able to come up with new ways of tweaking vaccines to make them more effective.
Although their methodologies differ sharply, Thompson and Narain are agreed on one thing: something needs to change with our current flu vaccines. With vaccine coverage so variable, the spectre of a devastating flu epidemic – like the Spanish flu that killed as many as 100 million people between 1918 and 1920 – looms large. “That is the biggest risk to the human population right now,” says Narain.
This article was originally published by WIRED UK
