Outsmarting Influenza's Rapid Evolution

Twice yearly, World Health Organization health officials meet to strategize against influenza, a malady that kills at least 250,000 people each year. In a chess match of sorts, they work to predict their opponent's next move, in this case by modifying vaccines to compensate for changes in the critical viral antigen hemagglutinin, which triggers the host's long-term immune memory. It takes several months to manufacture and distribute flu vaccines in sufficient quantities to inoculate vulnerable individuals before an epidemic occurs, making it necessary to anticipate critical antigenic changes well in advance. Until now, this prediction has been a qualitative rather than a quantitative process, performed by WHO's influ-enza committee in February and September. These predictions have become much more accurate since molecular data for amino acid sequences of the influenza A and B viruses became available in the mid-1980s, but the vaccines still fail to confer adequate protection in a significant number of cases. Recently developed predictive mathematical models, however, foretell a further leap forward in flu vaccine effectiveness, say researchers and senior WHO officials. It is possible, they say, to isolate key parameters of their mutations and interactions with the environment, and to predict some of the antigenic changes within six months to a year. But few predictions are perfect. Others say virus-host interactions are far too complex to simulate in a computer model. Uncertainties about the nature of the immune response and the hemagglutinin mutation rate gnaw at the parameters of the model. Unlike in chess, computers fall far short of their biological opponents. COMPUTING A DISEASE Computational models can do two things, according to Neil Ferguson, Department of Infectious Disease Epidemiology, Imperial College London. First, models that couple flu virus evolution to the host immune response could yield more accurate predictions of which viral strains will cause the world's next epidemic. The second, even more exciting possibility: Models may lead to a vaccine that provides generic immunity against all influenza strains, by triggering the host's temporary immune response to groups of related microbes. The exact biological mechanism of temporary immunity is still unclear. In the case of influenza, it protects against reinfection from related flu bugs while the patient's overall immunity against pathogens may be weakened. The possibility of exploiting temporary immunity has emerged from the work of Ferguson and others,1 which centers on why antigenic drift in hemaggluttinin-which some say mutates at an elevated rate-has not led to greater diversity in the flu virus than that observed consistently over the 25 years or more since data became available. "There is a natural tendency for viruses to get more and more diverse, to get as far apart genetically as they can," says Ferguson. Thus, they would avoid easy detection by the host immune system. "But that clearly has not happened in the case of the influenza virus." Ferguson and colleagues offer a possible explanation: The temporary immune response, by giving hosts short-term protection, inhibits diversity by reducing the number of strains that can be in active circulation at a given time. Once an epidemic starts, the strain that caused it spreads fairly rapidly within a population, reducing the number of potential hosts for other strains during the temporary immunity's time window. Similarly, if there were a subsidiary epidemic running concurrently, just one or two strains would saturate the available population, so that other strains die out for want of potential hosts. Consequently, argues Ferguson, other strains lose their foothold in a population and become extinct. Any good strategy requires patience. The model created by Neil Ferguson and colleagues at the Department of Infections Disease Epidemiology, Imperial College London,1 required 8 GB RAM, about 20 times that found in a modern PC. Running it on a 2GHz processor would have taken in effect 10,000 hours, so the task was split between 10 dual processor PCs, reducing the time required to 500 hours, or roughly three weeks. Ferguson's conclusions are based on one of the most sophisticated mathematical models yet devised for studying the transmission of influenza at a global level. The model that he and his colleagues devised simulates transmission within a constant population of 12 million hosts, divided into multiple zones. Each individual's immune history and age are taken into account, and the possibility of transmission between individuals is expressed as a combination of whether they are in the same zone and their proximity within that zone. Also an allowance is given to the "air travel effect," admitting a smaller chance of transmission between individuals in different zones. According to Ferguson, the model consumes huge computing resources, but a simpler version with fewer variables may actually yield deeper insights into the precise relative importance of the different factors. THE SIXTH SENSE Identifying whether and how short-term immunity could be exploited to develop a generic vaccine will depend on the nature of short-term immunity.2-5 If it is a form of innate immunity, it might not be a suitable candidate, because a vaccine could elicit an inflammatory response. But if short-term immunity is an adaptive response to specific proteins, the potential exists to harness it and produce a vaccine that would itself yield at least temporary immunity, says Ferguson. Most members within the influenza community consider it too early to talk about radically new vaccines, but the work has been welcomed. "In the long run this approach will be very useful, because we'll understand in much greater detail what is actually happening, how new variants are selected, and what the dynamics of their evolution are," says Nancy Cox, chief, influenza branch at the Centers for Disease Control and Prevention. Cox says she believes that inherent unpredictability of the flu virus will prevent models from ever replacing the gut instinct of human experts entirely. "But it may be that, although the prediction method won't be 100% successful, combining it with the experience of people in the field will work well," says Cox. Others are more skeptical, such as Alan Hay, director of WHO's Center for Reference and Research on Influenza in London. Hay says that even if models can predict amino acid changes in the hemagglutinin antigen, this is not sufficient to determine the precise vaccine formulation. The immune response is a function of time as well as antigenic spatial configuration, says Hay. "At a different point in time, a given change [in hemagglutinin amino acid sequence] may have a different effect on antigenic characteristics." The model might be improved by coupling host immune-system changes to viral antigenic drift. Another caveat: The models fail to account for genetic recombination, a significant factor in viral evolution during a host cell's coinfection by multiple strains, says Paul Sniegowski, assistant professor, Department of Biology, University of Pennsylvania. Sniegowski refers to earlier research on the evolution of high hemagglutinin mutation rates by the flu virus to elude the immune system, notably seminal work by Walter Fitch, Nancy Cox, and others.6 "It's difficult to see how you can evolve an elevated [mutation] rate for a system that has a lot of genetic recombination," says Sniegowski. Viral genomes tend to recombine with other strains, while coresident within the host cell. But, an elevated mutation rate at a given locus would evolve through a change elsewhere on the genome. Selective advantage for such a mutation would come only through association with an increased number of beneficial mutations on the target loci. Recombination tends to sever such linkages, limiting selection for a gene that confers a benefit elsewhere on the genome. Sniegowski agrees that this is not a fatal flaw in the model. There is still potential with recombination, as occurs in sexually reproducing higher organisms, for selection of linked genes. "But we raised it as a caveat for further thinking," says Sniegowski. The first strong empirical evidence that rapid mutation of hemagglutinin antigen made the influenza virus more likely to be successful came in 1997.6 An 11-year study of influenza A evolution gave strong statistical evidence that strains with high hemagglutinin mutation rates were more likely to be the progenitors of future successful strains. Fitch and colleagues have since repeated the research with the less virulent influenza B virus, and have come to the same conclusion, in a paper soon to be published. "We have found that what is true for A is true for B," Fitch confirms. This further strengthens the hypothesis that the host immune response drives evolution of fast hemagglutinin mutation in the influenza vaccine. But, unknowns in the complexity of the host-virus relationship remain, says Ferguson. In other words, checkmate lies far in the future. Philip Hunter (phunter@philiphunter.com) is a freelance writer in London. References 1. N.M. Ferguson et al., "Ecological and immunological determinants of influenza evolution," Nature, 422: 428-33, March 27, 2003. 2. T. Sonoguchi et al., "Cross-subtype protection in humans during sequential, overlapping, and/or concurrent epidemics caused by H3N2 and H1N1 influenza viruses," J Infect Dis, 151:81-8, 1985. 3. D.P. Skoner et al., "Effect of influenza A virus infection on natural and adaptive cellular immunity," Clinical Immunol Immunopathol, 79:294-302, 1996. 4. S.H. Seo et al., "Protective cross-reactive cellular immunity to lethal A/Goose/Guandong/1/96-like H5N1 influenza virus is correlated with the proportion of pulmonary CD8(+) T cells expressing gamma interferon," J Virol, 76:4884-90, 2002. 5. S. Sambhara et al., "Heterosubtypic immunity against human influenza A viruses, including recently emerged avaian H5 and H9, induced by FLU-ISCOM vaccine in mice requires both cytoxic T-lymphocyte and macrophage function," Cell Immunol, 211:143-53, 2001. 6. R.M. Bush et al., "Positive selection on the H3 hemagglutinin gene of human influenza virus A," Mol Biol Evol, 16:1457-65, 1999. function sendData() { document.frm.pathName.value = location.pathname; result = false if (document.frm.score[0].checked) result = true; if (document.frm.score[1].checked) result = true; if (document.frm.score[2].checked) result = true; if (document.frm.score[3].checked) result = true; if (document.frm.score[4].checked) result = true; if (!result) alert("Please select") return result; } .myradio{background-color : #94bee5} Please indicate on a 1 - 5 scale how strongly you would recommend this article to your colleagues? Not recommended 1 2 3 4 5 Highly recommended Please register your vote

Outsmarting Influenza's Rapid Evolution

Interested in reading more?

Become a Member of