The below was originally written for The Rules of Contagion, but it was cut from the final draft in late 2019 – the book was already too long and it was a stand-alone section that didn’t really link together with the topics around it.
Six years before SARS appeared in Hong Kong, a three-year-old boy with a fever arrived at Queen Elizabeth Hospital in Kowloon1. It was May 1997, a few weeks before the handover of Hong Kong from Britain to China. It would turn out to be a new biological era as well as a political one. The boy was infected with H5N1, a strain of flu that had until that point been a bird virus.
The realisation that H5N1 could infect humans made health agencies wonder whether it might one day cause a pandemic. Since that first case in 1997, there have been over eight hundred human H5N1 infections reported globally2. However, for a long time it wasn’t clear whether these bird viruses would ever be able to spread between humans, or even mammals. That changed in 2011, when teams at Erasmus Medical Center and the University of Wisconsin–Madison announced that they had each created new strains of H5N13. Unlike circulating viruses – which had only spread between people who’ve come into close contact with each other – these mutant versions were airborne, successfully spreading between ferrets in a lab.
Why could these new bird flu viruses spread when others had failed? Both teams found that three crucial changes made transmission possible4. First, the new viruses could infect the easy-to-access upper portion of the human airway. Second, they could survive at lower temperatures (in birds, flu infects the warm inner gut rather than the cooler airway). And third, the viruses had picked up a mutation that made them more stable, helping them make the trip from one host into another intact. Scientists had already suspected the first two changes were important for transmission5, but it was the addition of the stability mutation that apparently allowed the virus to spread successfully between ferrets6.
The experiments had undoubtedly improved our knowledge of bird flu viruses. However, the research soon triggered an intense debate about whether the scientific insights from such studies were worth the potential public health risks. What would happen if an H5N1 virus like this escaped from a lab? Although scientists regularly study dangerous viruses found in nature, here was a virus that – as far as we know – did not yet exist in the natural world.
The idea of a non-natural flu outbreak wasn’t just a hypothetical concern. When a seemingly new flu virus emerged in 1977, researchers noticed some unusual features: the disease was surprisingly mild, and most of the cases were young. Why weren’t older age groups getting ill? It turned out that they’d already been exposed to the virus. The ‘new’ 1977 strain was in fact an old one: it was the same as a virus that had been spreading in the 1950s. We still don’t know the exact reason it reappeared. It may have been a lab accident, or a careless human infection study. But we do know what happened afterwards. Variants of the virus would continue to cause outbreaks for the next thirty years.
Could the same thing happen again with an H5N1 virus? In 2014, epidemiologist Marc Lipsitch and biosecurity specialist Thomas Inglesby used a simple model to estimate the risk the H5N1 experiments posed. For a lab virus to harm the wider population, a scientist would first need to get infected, then that this infection would need to spread. Based on historical data, Lipsitch and Inglesby reckoned that if a single lab studied high-risk viruses for one year, there was between a 1 in 1,000 and 1 in 10,000 chance the experiments could lead to an accidental pandemic7.
Ron Fouchier, who’d led the H5N1 project at Erasmus, disagreed with their calculation. He argued that the Erasmus lab had many additional safeguards in place, so the annual chance of a lab-induced pandemic would be less than one in 33 billion8. Lipsitch and Inglesby disputed this value, pointing out that some safeguards had effectively been double-counted in the calculation9. The debate shows how difficult it can be to agree on the chance of unlikely events: subtle tweaks to assumptions can have a big effect on the overall probability10.
These disagreements might seem irrelevant, given that Lipsitch and Inglesby’s most pessimistic estimate works out at one accidental pandemic every 1,000 years. However, such probabilities need to be weighed against the massive amount of global damage a flu pandemic could cause11. From a risk assessment point of view, unlikely events that result in a large amount of harm are more concerning than common events with limited consequences.
If these are the risks of making H5N1 viruses more transmissible, what are the benefits? The most obvious one is a proof of concept: we now know that H5N1 can adapt to spread between mammals like ferrets. Unless the virus had evolved in nature, this is not something that could have been discovered in another way. It’s also been suggested that knowledge of specific risky H5N1 mutations could help guide disease surveillance12. However, the potential benefits here are less clear. Health agencies were concerned about H5N1 before they knew it could transmit between ferrets, and there are many other ways flu could adapt to spread among humans; while attention was on bird flu in Asia in the early 21st Century, the 2009 flu pandemic emerged in pigs in Mexico. A more tangible benefit of the H5N1 experiments has been an improved understanding of fundamental virus biology, like the importance of that stability mutation. Even so, discovering that mutation didn’t necessarily require high-risk experiments on H5N1. Researchers have since found the same thing by studying the 2009 pandemic virus13.
As a result of the debate around the H5N1 studies, the US National Institutes of Health (NIH) would eventually suspend funding for experiments that aimed to make risky viruses more transmissible14. It didn’t help that in 2014, there had been several high profile reports of dangerous viruses being mislabelled at supposedly secure US government labs, potentially putting staff at risk15. The funding suspension would remain in place until late 2017, when NIH announced they would consider studies involving ‘potential pandemic pathogens’ if they met new risk-reduction criteria. As well as being assessed on their scientific merit, future proposals would receive extra scrutiny from specialists in biosecurity and ethics. ‘The life sciences have reached a crossroads,’ noted the NIH advisory board who oversaw the publication of the two H5N1 transmission studies16. ‘The direction we choose and the process by which we arrive at this decision must be undertaken as a community and not relegated to small segments of government, the scientific community, or society.’
Parkin S. Fowl Plague: Episode 1, Inception: The Avian Flu Outbreak in Hong Kong, 1997. How We Get To Next, 2017
World Health Organization. Cumulative number of confirmed human cases for avian influenza A(H5N1) reported to WHO, 2003-2018. WHO/GIP, 2018
Enserink M. Public at Last, H5N1 Study Offers Insight Into Virus’s Possible Path to Pandemic. Science, 2012
Imai M et al. Experimental adaptation of an influenza H5 HA confers respiratory droplet transmission to a reassortant H5 HA/H1N1 virus in ferrets. Nature, 2012; Herfst S et al. Airborne Transmission of Influenza A/H5N1 Virus Between Ferrets. Science, 2012
Auewarakul P et al. An Avian Influenza H5N1 Virus That Binds to a Human-Type Receptor. J Virol, 2007; Le QM et al. Selection of H5N1 Influenza Virus PB2 during Replication in Humans. J Virol, 2009
Russell CJ. Influenza Hemagglutinin Protein Stability, Activation, and Pandemic Risk. Trends in Microbiology, 2018
Lipsitch M et al. Moratorium on Research Intended To Create Novel Potential Pandemic Pathogens. mBio, 2014
Fouchier RAM. Studies on Influenza Virus Transmission between Ferrets: the Public Health Risks Revisited. mBio, 2015
Lipstick M et al. Reply to “Studies on Influenza Virus Transmission between Ferrets: the Public Health Risks Revisited”. mBio, 2015
Lipsitch M et al. Underprotection of Unpredictable Statistical Lives Compared to Predictable Ones. Risk Analysis, 2016
UK Cabinet Office. National Risk Register, 2017
Davis CT et al. Use of Highly Pathogenic Avian Influenza A(H5N1) Gain-Of-Function Studies for Molecular-Based Surveillance and Pandemic Preparedness. mBio, 2014
Russell CJ. Influenza Hemagglutinin Protein Stability, Activation, and Pandemic Risk. Trends in Microbiology, 2018
Kaiser J. NIH lifts 3-year ban on funding risky virus studies. Science, 19 Dec 2017
Kaiser J. Lab incidents lead to safety crackdown at CDC. Science, 11 July 2014
Berns KI et al. Adaptations of Avian Flu Virus Are a Cause for Concern. Science, 2012