Quantifying pandemic risk caused by within-host evolution of influenza A/H5N1 viruses

Throughout history, influenza pandemics have caused widespread illness and death. The ability of influenza to transmit between humans through the air allows it to spread rapidly, and makes it difficult to design efficient countermeasures. Understandably, the possibility of a next influenza pandemic is therefore a constant cause of concern. Influenza viruses found in birds are particularly suspect, as they have been the source of previous pandemic viruses.

Since 1997, the primary influenza strain of pandemic concern has been A/H5N1, often referred to as "bird flu". The H5N1 virus is common in countries such as China, Indonesia, Egypt and Thailand, and kills all birds on a poultry farm within 48 hours after it arrived. There have been over 600 instances where a human was infected with an H5N1 virus, often as a result of very close contact with sick animals. In over half of the cases of human infection with H5N1, the patient dies. However, to date, there is no hard evidence of an H5N1 infected human infecting another human. If the virus could spread easily from human to human and travel over the world, it could be a disaster. Therefore, it is important to understand why we do not see transmission between humans, even though human infections with H5N1 are possible.

Until recently, one explanation for the lack of human-to-human transmission was that maybe it just was not possible for the H5N1 virus to spread through air. However, earlier this year, two research groups showed that with a specific set of mutations in the virus genome, the virus can transmit between mammals. This puts us ahead of the game for pandemic preparedness, and we could do everything in our power to stop an H5N1 pandemic from ever starting. But an important question remains: will a virus with the key mutations ever evolve in nature? The elimination of the virus would be time-consuming, expensive and difficult. If it is too difficult for the virus to actually get this set of mutations, a pandemic may never occur. And then we should have spent our time and research and healthcare money differently.

I addressed this question of what is the risk of a transmissible H5N1 virus evolving in nature, by building a mathematical model that the described the evolution of the virus during infection of a human. The results from this model were published in the highly-respected scientific journal Science, and also used to inform advisory boards and governments in the USA and the Netherlands on the risk of transmissible H5N1 viruses evolving in nature. This is exactly what I want to do: researching fundamental principles, whilst having a direct translational impact.

For this proposal, I will refine this risk assessment of an H5N1 pandemic. I will work closely with the world's leading research groups on influenza virus transmission to assess if which mutations appear during human H5N1 infections, and how often. I will refine our estimates and understanding of key parameters and processes. While doing this, I will gain an understanding of fundamental processes of evolution that occur within an infected host. For these studies, I will use advanced genetic sequencing technology to study the evolution of H5N1 viruses within the human host. I will develop new quantitative methods to analyse the data during a training period at the UK's prominent institute for genetic data analysis: the Sanger Institute.

I have a wide range of collaborators to obtain the information I need, will be working in a host group specialised in evolution and influenza, and have designed a training programme to improve my skills. With this construction, I will be able to combine the outcomes of these studies and to provide input to basic science and knowledge on evolution of viruses during infection. By refining the model on evolution of H5N1 viruses during infection, I will be able to provide the much-needed estimate for the risk of H5N1, and contribute to improvements in global health.

I want to assess the pandemic risk posed by the evolution of A/H5N1 influenza viruses within an infected host, through acquiring respiratory droplet transmissibility between humans. The description of the probability of a transmissible H5N1 virus emerging in nature is critical to estimate and mitigate the risk for H5N1 viruses to create a pandemic, and is important for international policy decisions on the strategies for surveillance, intervention and global eradication of the virus.

I have built a mathematical model describing the within-host evolution of pathogens, which was published in Science earlier this year. I propose to assess the risk posed by H5N1 viruses by refining this model and through working closely with a wide network of expert collaborators, and designing experiments with them. I will design a framework for the analyses of the deep sequencing data, and use this to study infections with fully adapted viruses to increase our understanding of viral dynamics and within-host evolutionary processes. Through analysis of deep sequencing data from experimental samples and precious human H5N1 infection data, the evolutionary mechanisms will be complemented with information on the viral diversity at the start and during infection, fitness gains and losses for different mutants, the identification of functionally equivalent mutations, knowledge on the effects of glycan heterogeneity in the respiratory tract and studies on the mechanisms of respiratory droplet transmission.

I will synthesise the gained fundamental insights into within-host pathogen evolution and the contributions to fundamental scientific knowledge arising from refined estimates for critical parameters, to design further experiments and, most importantly, to integrate this information and use it to continuously refine the risk assessment for a human pandemic through evolution of currently circulating H5N1 viruses.

The work I led on the within-host evolution of H5N1 influenza viruses earlier this year has had a wide impact, with a scientific publication in Science and presentation of the work at international policy meetings on the publication of the H5N1 transmission experiments. The publication highlighted areas of investigation that are critical to assess, monitor and mitigate the pandemic risk posed by H5N1 viruses. I will add to these implementable actions with the proposed research, which will directly enable a refined assessment of this risk. The risk assessment is critical for health officials and policy makers, as it enables us to mitigate pandemic risks, and thus avoid extensive morbidity and mortality.

Influenza H5N1 viruses currently affect a large part of the world, in particular countries such as Egypt, Indonesia, China and Thailand. Most strategies to address H5N1 infections in poultry to prevent further spread of the virus, and thus prevent infection of humans, include vaccination of livestock and culling. Blanket solutions to mitigate pandemic risk include human vaccination programs, stockpiling of antivirals and the eradication of H5N1 in the world through the systematic culling of poultry. All of these measures are extraordinarily expensive, logistically challenging and may not always be effective (e.g. if the pandemic strain is not responsive to antivirals, or not covered by the immune response elicited by the vaccine). Moreover, the culling of poultry has a large impact on human wellbeing in affected countries, as chicken is a main source of protein in their nutrition. If current surveillance practices are sufficient to mitigate pandemic risk, then there will be the potential for huge economic and environmental savings. The proposed research additionally enables us to target surveillance efforts more specifically, by providing a quantitative basis to inform decisions on which species to monitor, when to sample and how deep to sequence.

There are actuarial calculations that will have to be made, by governments as well as insurance companies, to weigh the costs and benefits of these approaches. More accurate risk assessments will feed into these calculations and can aid the evaluation of the need for large-scale efforts and the design of more targeted prevention and mitigation strategies. Knowing the risks posed by H5N1 will additionally allow an evaluation on whether the benefits of transmission research outweigh the risk, and whether it is the best investment of infectious disease research money; it might provide guidelines for governments and funding agencies for research and stimulus prioritisation.

The broader implications of this work are relevant to H5N1 and for seasonal influenza, as well as for other infectious diseases. Mechanisms for within-host evolution are relevant for the viral escape of antivirals or antibodies. Therefore, this research is also useful for pharmaceutical companies developing drugs or vaccines and regulators of such drugs, not just for H5N1 and other influenza strains, but also to combat other pathogens for which within-host evolution complicates effective treatment. Knowing how many mutations the virus can gain, and how many would be too much, over a specified amount of time, would be of great value. This information can, for example, provide input on the use of combination therapy and suggest further improvements to vaccine formulation. Improved antiviral drug therapy regiments, anticipation of antibody escape in the context of vaccination, and novel mechanisms to counter the spread of H5N1 and other pathogens would additionally have a financial impact due to the more effective use of money. These highly important developments in infectious disease control and prevention strategies have great potential to influence regulators, day-to-day medical professionals and the NHS, by providing targeted strategies to enhance national and global health.