Amanda N. Koch
The James Martin Center for Nonproliferation Studies
Is the Avian Influenza Virus a Suitable Agent for a Biological Weapon?
Increased public awareness of a possible influenza pandemic has raised many questions among the general public regarding the virus, its transmissibility, and public health preparedness and response. The inadequate international efforts to respond to the SARS epidemic in 2003, Indian Ocean tsunami of 2004, and hurricanes of 2005, have prompted many nations and international bodies to ask whether they are prepared for a natural outbreak of the flu, or other disease-related incidents. Many states have formulated preparedness plans, which are at various stages of implementation. However, other questions have arisen over whether this virus could be used by terrorists or other entities in a biological weapon. In this briefing, we will discuss the different strains of the influenza virus, their diverse means of transmission and genetic variability, and conclude by analyzing the potential use of the virus in a biological weapons system.
The influenza virus has its natural reservoir in wild birds, especially ducks, and pigs. There are three types of the virus: A, B, and C. Most remain in their natural host populations, but some are transmissible to and among humans and other animals. The influenza virus mutates quickly, as do most RNA viruses, and has varying levels of virulence according to its current genetic makeup.
Each influenza virus type has two distinct glycoproteins located on its surface; hemagglutinin (H or HA) and neuraminidase (N). There are 15 known hemagglutinin (H) serotypes and nine known neuraminidase (N) serotypes. Depending on which combination of glycoproteins is found on the virus, as well as other genetic differences, the influenza virus will have greater or lesser infectivity and virulence, and affect different animal species.
A highly virulent subtype of the influenza virus type A, H5N1, which was first identified in China in 1996, became a cause of concern when it began spreading to humans in the live-poultry markets of Hong Kong in May 1997. Six of 18 infected individuals died, indicating how potentially devastating the disease caused by H5N1 could become if it were to spread through human populations. The disease, now termed avian influenza, was apparently brought under control by the aggressive culling of poultry in Hong Kong, but it soon became clear that the H5N1 virus continued to spread among the wild duck populations of the Chinese coastal areas. These birds can carry and spread the virus but do not themselves become ill.
Thus far, four genetically and antigenically distinct forms of the H5N1 virus have been identified, referred to as clades and subclades. Clade 1 spread to South Korea and Japan in 2003 and 2004. These countries responded by imposing quarantine, culling domestic poultry, and imposing stringent microbial security at poultry facilities. No cases of human infection were identified during this outbreak. However, beginning in May 2005, apparently because of the spread of the virus to wild fowl in Qinghai Lake in China, clade 2 began to spread rapidly throughout Southeast Asia and Indonesia, and westward to India, Russia, Turkey, and into Africa. Clusters of human infection have been reported in several places affected by clade 2, with the number of known cases exceeding 250 to date. The mortality rate is around 50% in hospitalized individuals who exhibit symptoms and/or who have been positively diagnosed to be infected with the H5N1 virus. However, the overall mortality rate is probably much lower if one takes into account individuals with either no symptoms or mild symptoms who have not sought medical help.
Because the virus can apparently infect several types of wild migratory birds, often without symptoms, thus providing a rapidly dispersible viral reservoir, there is great concern that the virus could continue to spread to new areas of the world, with an increasing chance of infecting humans, possibly giving rise to a very serious global influenza pandemic.
Avian influenza in humans is characterized by an incubation period of four to seven days following exposure to infected birds, with the symptoms including fever and lower respiratory tract symptoms (eg. phlegm, coughing, discomfort) in nearly all patients. Several of the following symptoms are also usually present: headache, muscle ache, shortness of breath, cough, sore throat, bleeding gums, irritated eyes, runny nose, and diarrhea. Of these, the upper respiratory tract symptoms are often absent. Initial exposure is believed in most cases to be through the inhalation of droplets containing the virus, during which the virus comes in contact with the respiratory mucosa. Pneumonia is common, consistent with the replication of the virus beginning in the lower respiratory tract, becoming severe in about half of the patients, in which case, death is likely. The failure of other organs can also occur.
Most patients studied were hospitalized five to seven days from the onset of symptoms. In moribund patients, death ensues 10-15 days from the onset of symptoms.
The symptoms observed during infection by the H5N1 virus are highly variable: diagnostic surveillance of people in a few areas where human cases were reported reveal that some individuals become infected, but remain asymptomatic or with only mild symptoms. There also seem to be differences in the prevalence of particular symptoms exhibited by patients infected with different clades of the virus.
The H5N1 virus is difficult to diagnose. Immunological methods, such as rapid influenza antigen testing and Enzyme-linked immunosorbent assay (ELISA), usually fail to detect the presence of the virus and even reverse transcriptase polymerase chain reaction assays (RT-PCR) can give false negatives. Therefore repeated testing of samples from different sites (e.g., serum, nasal and throat or pharyngeal swabs) is often necessary to make a definitive diagnosis of the presence of the H5N1 virus. Direct viral isolation and sequencing, viral culture and western blot analyses have been used to confirm the presence of the virus in serum and respiratory samples, but these are not useful methods for making a rapid diagnosis of a sick person suspected of having the disease. Because of the speed and reliability of RT-PCR testing, this is probably the method of choice for the clinical confirmation of infection by H5N1, using samples from deep tracheal aspiration.
Because of a relatively long incubation period before symptoms appear, treatment with antiviral agents, such as the neuraminidase inhibitor oseltamivir after symptoms appear is generally not very effective in reducing viral load. Moreover, such delayed treatment increases the probability of selecting for resistant strains, since only one mutational event is needed to confer resistance to oseltamivir. A significant fraction of the H5N1 virus isolates from severely infected patients are resistant to oseltamivir. Neuraminidase inhibitors given immediately after exposure can be quite effective, so it is logical to administer the drug prophylactically in areas where exposure to the virus is suspected. A variety of other drugs, including antibiotics and corticosteroids (albuterol, fluticasone, ceftriaxone, meropenem, ciprofloxacin, vancomycin, gentamycin, amikacin, linezolid, budesonide, aminophylline, dexamethasone) have been tried unsuccessfully, apparently in desperation, to save severely ill patients. Although antibiotics have no affect on viruses and should be used solely to treat bacterial infections, some doctors may have administered these in an attempt to treat suspected secondary bacterial infections that could be associated with influenza. Mechanical ventilation has been used to treat patients with respiratory failure, but when the illness has progressed this far, death has always followed.
Some clade 2 H5N1 viruses appear to be sensitive to adamantanes (amantadine and rimantadine), which interfere with a key step in the process of infection, whereas most of the clade 1 viral isolates are resistant. Because adamantane resistance is conferred by a single mutational event, and resistant strains are becoming more common, WHO recommends against using this class of agents to treat avian influenza in humans.
While antiviral agents are of limited value generally, in order to be effective they must be administered as soon as possible following exposure and infection. Realistically, this will nearly always be well before the appearance of symptoms. Short of coercive prophylactic measures by public health authorities in regions where there are risks of outbreaks of avian influenza, it is unlikely that those who are most at risk, i.e., people who live in close contact with domestic poultry, will receive such treatment until they become sick, in which case it is probably too late to do much good.
In principle, the H5N1 virus can infect humans through three possible routes: most commonly from infected birds (or other animals), less commonly from an infected person, or, rarely, from the environment. Thus far, it appears that the vast majority of human infections described to date have come from direct contact with infected birds, often under conditions where people and poultry share the same space, including the same living quarters. There have also been several instances of clusters of human infections (usually in the same family) where there is compelling circumstantial evidence that human to human transmission has occurred. While the virus is quite stable in the environment and exposure and infection is theoretically possible, for example from contaminated clothing, towels, or doorknobs, it is unlikely that casual environmental exposure could create a high enough viral titer coming in contact with the mucosal surfaces deep in the bronchi and alveoli where the concentration of avian influenza virus receptors are highest.
There is considerable information available on the structure, genetics, immunogenicity, and mechanisms of infections of the type A influenza viruses. Human and avian influenza viruses infect the human body by binding to hemagluttinin (HA) receptors located in the respiratory tract, then replicating within invaded cells, and producing an infection. These viruses both are type A influenza viruses, but bind to different receptors, which are found in different locations in the host. The human influenza virus finds its receptor in the upper respiratory tract of humans, in the nose and throat, whereas the avian influenza virus must travel further down the respiratory tract to the receptors in the lower bronchi and alveoli in the lungs. Therefore, being infected with the avian influenza virus through the inhalation of droplets containing the virus through casual exposure to infected poultry is unlikely, since most contact would occur at the nasal mucosal surfaces, where there are few avian flu receptors. But in situations where there are many infected birds sharing a small, closed space with people, the number and concentration of airborne viruses is much higher, so the probability of inhaling the virus particles deeper into the lungs, where they can bind efficiently to receptors on the cells of the mucosal surface, is much higher. A large fraction of the human cases reported so far fit this picture.
It should be noted that because diarrhea is common among patients infected with H5N1 viruses, there are probably receptors for the avian influenza virus in the gut, allowing it to replicate there as well.
There are several features of the H5N1 virus that affect virulence, including the binding specificity of the HA and its activation by cellular proteases, the rate of replication of the virus, and the structure and properties of the surface/capsid proteins that influence the host immune responses, both cellular and subcellular. A number of mutations have been mapped and characterized that affect all of the above properties of the H5N1 virus., 
One obvious possible corollary of the primary site of initial infection being deeper in the lungs is that pneumonia would be expected to develop early and be quite severe, whereas, if the nasal mucosa were the initial site, there is the possibility that some immunity could be induced before there was significant invasion of the lungs. This could explain the 50% mortality rate among the reported symptomatic human cases, where death is nearly always the result of pneumonia and respiratory failure. This very high mortality rate must be understood in the context that many individuals may contract influenza, but with light to moderate symptoms that they treat at home without a doctor. Thus the number of reported cases, and their high mortality rate only include the most severe instances, where medical assistance was sought.
The subtypes of influenza A viruses responsible for the 1918 global pandemic, as well as those in 1957 and 1968, derived from avian influenza viruses, in which the HA proteins typically bind to the receptors located in the lower respiratory tract. However, HA from viral isolates from infected patients from these pandemics was found to bind to receptors in the upper respiratory tract preferentially in all cases. Thus the historic pandemic influenza viral subtypes must have undergone one or more mutations (often called genetic shift), resulting in strains that bind to a different set of receptors. This is an important adaptive step in the formation of a highly virulent and potentially contagious strain from one that infects mainly birds.
Genetic mapping of H5N1 viruses from infected humans has revealed a great deal of variation. Of particular relevance is the appearance of mutations affecting the HA protein, changing its preference for which cells to bind to in the host. The HA from several human isolates of the H5N1 virus can bind to receptors found both in the upper and lower respiratory tracts of humans. However, viral isolates from infected birds bind only to the receptor found in the lower respiratory tract of humans. Studies have shown that this change in binding preference can occur with only one minor mutation in the flu virus' RNA. In the infection process, the flu virus may replicate several times, increasing the probability that it will undergo a mutation. This could account for some of the demonstrated cases of human-to-human spread, although in at least one instance, the viral isolates from those infected through human contact showed no adaptive change in their specificity for binding sites.
It is also noteworthy that recent H5N1 viral isolates differ considerably in both the HA and N genes from those obtained in the 1997. Moreover, the differentiation into two distinct clades (one with 3 subclades), is testimony to the instability of this influenza A viral genome. There is sufficient antigenic variation among the different clades and subclades to suggest that a vaccine against one type is likely not to be nearly as effective against the others.
Essentially any factor or circumstance that affects the survival or the propagation of the H5N1 virus creates a selective pressure with inevitable genetic drift or adaptation to the new circumstances. For example, treating patients with antiviral drugs, such as the neuraminidase inhibitor oseltamivir only after the appearance of symptoms, while it may be of some value in treating the disease, also creates a selective pressure toward the creation of a resistant strain.
Another phenomenon that certainly must be considered is the possibility of genetic mixing between the H5N1 viral genome and other, less virulent strains that are already adapted to the human host. While the probability is small that a single individual would become co-infected with both a human influenza A virus and the H5N1 virus simultaneously, the probability of this rare event increases as the virus spreads around the world, as it now seems to be doing. The outcome to be feared is of course that a viral strain with the virulence and mortality rate of the H5N1 virus and ease of infecting humans contagiously would emerge. This scenario is believed to be the critical event in the evolution of the 1918, 1957, and 1968 pandemic influenza A strains.
Curiously, the strain of H5N1 virus, clade 2, subclade 2, that has spread westward from China to Russia, Turkey, and Africa has shown almost no genetic variation, in sharp contrast to the strains that have spread to Indonesia and throughout Southeast Asia. While it was undoubtedly spread through wild bird species that did not become sick, it appears to infect relatively few species, which could account for the lack of selective or adaptive pressure present in some of the other strains.
In any case, the viral isolates obtained during the past year indicate that rapid adaptive evolution is taking place. They are significantly different in many genes and apparently more virulent than those obtained in the first outbreak in 1997. The host range in birds has expanded, the virus has adapted to be able to infect members of the cat family, pathogenicity in experimental animals is substantially greater than in the earlier strains, leading to systemic infections, and the viruses themselves have evolved to become more stable in the environment.
There is currently great concern that the H5N1 virus is undergoing rapid evolutionary changes that may (1) increase the capacity of the virus to infect humans and (2) increase its virulence. This constitutes a recipe for a potential global pandemic. If the mortality rate of 50% were to prevail in a strain that was readily contagious from person to person, the outcome could be catastrophic. Public health authorities around the world are now engaged in measures designed to prevent this from happening, including responding quickly to any outbreaks among domestic or wild fowl, and working to develop and produce vaccines that will be effective against all of the known antigenic variants (clearly, a moving target). At present, the virus has not become highly contagious among humans, but given the instability of the viral genome, it is probable that, given enough time, this will happen.
How, then, could the H5N1 virus be considered a serious candidate for a BW agent? The question of whether flu viruses generally may be useful for biological weapons purposes has been considered previously by Pouliot. This discussion deals specifically with the H5N1 subtype of the influenza virus. To be useful for BW purposes, the H5N1 virus would need to undergo significant modifications. The virus would need to be engineered to affect the binding preference of the HA protein, viral transmissibility from human to human would need to be considered, formulation to maintain viability and virulence would need to be undertaken, and the resulting product would need to be prepared in a way to ensure effective dissemination to the target population. The latter would most probably be in the form of an aerosol, but since there seem to be HA receptors in the gut, an oral route of dissemination might also be considered. Moreover, genetic stability is essential in order to guarantee predictable results in the use of the virus as a BW agent.
For use as an aerosolized BW agent, the H5N1 virus would have to be modified so that its HA would bind preferentially to the receptors in the upper respiratory tract of the target population. This would guarantee a higher infection and incapacitation rate, since fewer viruses would need to be dispersed, and the target population would not need to inhale deeply in order to become infected. In terms of transmissibility, historically there has been no consensus on whether a BW agent should be contagious or not. The USSR, for example, weaponized and researched for BW use many highly contagious agents, such as the smallpox and hemorrhagic fever viruses, whereas the United States made a conscious decision to not use contagious agents in its BW program. In the case of the H5N1 virus as a BW agent, the characteristic of transmissibility would be desirable only under certain circumstances, depending on the intended application of the agent and desired consequences. It could be engineered to be more contagious from person to person, but only if it also gained a highly stable genetic composition and if its developer possessed a functional vaccine. In this case the virus could be released and spread throughout all unisolated and unvaccinated target populations, with no risk of unwanted mutation and unintended infection of vaccinated populations. In the case that genetic stability and an effective vaccine were not achieved, it would be better to engineer the virus to not be transmissible from human to human, thus limiting infection to only the initially targeted population and controlling the spread of the disease. At the same time, it would be desirable for the modified virus to be incapable of infecting birds or other animals, thus eliminating the risk of inadvertent spread of the virus in, e.g., asymptomatic migratory birds.
Because effective vaccines against a genetically stable H5N1 virus and its derivatives could be made readily, the safety of workers could be controlled through vaccination. But this fact may also render the virus ineffective as a BW agent, since there already would be measures in place to vaccinate populations against any incipient H5N1 epidemic. Even though there is great antigenic variation, there is usually some cross reactivity, which, while insufficient to prevent infection, may nevertheless greatly reduce the mortality rate. To engineer a strain of the virus that was resistant to the common vaccinations and antigenically different from all the wild strains but still retained high virulence might be difficult or impossible to do.
The real challenge in trying to adapt the H5N1 virus to satisfy the criteria of a suitable BW agent would be to stabilize the genome. The influenza A viruses historically are maddeningly variable, which is why they pose a continuing challenge to public health organizations, and why the flu shot developed for use this year may be quite useless against next year's flu strain. From the virus' point of view, this property is an important evolutionary adaptation. To our knowledge, nobody knows how to change the innate frequency of mutations in this virus. It could, for example, lie in the polymerase(s) that are responsible for replicating the viral genome, where on thermodynamic grounds, there is a built-in error frequency that allows evolutionary adaptation to proceed rapidly, and even improves the probability of purely random, non-adaptive, events. It would take a great deal of effort to determine the source or sources of the genetic instability, and a great deal more to engineer changes to prevent it, raising the question of whether or not the effort is even worth it, given the abundance of other natural infectious agents that would be more suitable as agents for weaponry.
In spite of its appeal as a highly virulent and relatively stable infectious agent, there are several important reasons why the avian influenza A H5N1 virus would be a poor choice as BW agent.
The authors consider it highly improbable that any party will choose to pursue the development of BW using this virus. However, like the arsonist with a box of matches, an irrational individual with access to a highly virulent strain of the virus could still cause a great deal of harm.
 WHO Writing Committee, "Avian Influenza A (H5N1) Infection in Humans," New England Journal of Medicine 353 (2005), pp. 1374-1385; Tran Tien Hien, Menno de Jong, and Jeremy Farrar, "Avian Influenza – A Challenge to Global Health Care Structures," New England Journal of Medicine 351 (2005), pp. 2363-2365.
 Robert G. Webster and Elena A. Govorkova, "H5N1 Influenza – Continuing Evolution and Spread," New England Journal of Medicine 355 (2006), pp. 2174-2177.
 D.J. Hulse-Post, K.M. Sturm-Ramierez, and J. Humberd, et al. "Role of Domestic Ducks in the Propagation and Biological Evolution of Highly Pathogenic Influenza Viruses in Asia," Proceedings of the National Academy of Sciences of the United States 102 (2005), pp.10682-7.
 A.F. Oner, et al. "Avian Influenza A (H5N1) Infection in Eastern Turkey in 2006," New England Journal of Medicine 355 (2006), pp. 2179-85; Nyoman Kandun, et al. "Three Indonesian Clusters of H5N1 Virus Infection in 2005," New England Journal of Medicine 355 (2006), pp. 2186-94.
 T. Chotpitayasunondh, K. Ungchusak, W. Hanshaoworakul, et al., "Human Disease from Influenza A (H5N1), Thailand, 2004," Emerging Infectious Diseases 11 (2005), pp. 201-209.
 P. Ward, I. Small , J. Smith, P. Suter, and R. Dutowski, "Oseltamivir (Tamiflu) and its Potential Use in the Event of and Influenza Pandemic," Journal of Antimicrobial Chemotherapy 55, (2005), pp. i5-i21.
 Oner, et al, "H5N1 in Eastern Turkey"; Kandun, et al., "Indonesian Clusters of H5N1."
 Kandun, et al., "Indonesian Clusters of H5N1."
 Anne Moscona, "Neuraminidase inhibitors for influenza," New England Journal of Medicine 353 (2005), pp.1363-1373.
 Tran Tinh Hien, Thanh Liem Nguyen, Thi Dung Nguyen, et al., "Avian Influenza A (H5N1) in 10 Patients in Vietnam," New England Journal of Medicine 350 (2004), pp. 1179-1188
 World Health Organization, "WHO rapid advice guidelines on pharmacological management of humans infected with avian influenza A (H5N1) virus," (Geneva: WHO Press, 2006), www.who.int.
 K. Ungchusak, et al. Probable Person-to-Person Transmission of Avian Influenza A (H5N1), New England Journal of Medicine 352 (2005), pp. 333-340.
 K. Shinya, M. Ebina, S. Yamada, M. Ono, N. Kasai, and Y. Kawaoka, "Avian Flu: Influenza Virus Receptors in the Human Airway," Nature 440 (2006), pp. 435-436.
 M. Hatta, P. Gao, P. Halfmann, and Y. Kawaoka, "Molecular Basis for High Virulence of Hong Kong H5N1 Influenza A Viruses." Science 293 (2001), pp. 1840-1842.
 K. Shinya, S. Hamm, M. Hatta, H. Ito, T. Ito, and Y. Kawaoka, "PB2 Amino Acid at Position 627 Affects Replicative Efficiency, but not Cell Tropism, of Hong Kong H5N1 Influenza A Viruses in Mice," Virology 320 (2004), pp. 258-266.
 S. Yamada, et al., "Haemagglutinin Mutations Responsible for the Binding of H5N1 Influenza A Viruses to Human-type Receptors," Nature 444, (2006), pp. 378-382.
 R. Harvey, A.C. Martin, M. Zambon, and W.S. Barclay, "Restrictions to the Adaptation of Influenza A virus H5 Hemagglutinin to the Human Host," Journal of Virology 78 (2004), pp. 502-7.
 K. Ungchusak, et al. "Probable Person-to-Person Transmission of Avian Influenza A (H5N1)," New England Journal of Medicine 352 (2005), pp. 333-340.
 World Health Organization, "Influenza research at the human and animal interface: Report of a WHO Working Group," (Geneva: WHO Press, 2006), www.who.int.
 Martin Enserinck, "H5N1 Moves Into Africa, European Union, Deepening Global Crisis," Science 311 (2006), p. 921.
 Yannick Pouliot, "Influenza (Flu)" in Richard D. Pilch and Raymond A. Zilinskas, eds., Encyclopedia of Bioterrorism Defense (Hoboken, NJ: Wiley-Liss, 2005), pp. 259-262.
Sign up for our newsletter to get the latest on nuclear and biological threats.
Practical solutions for organizations involved in life science research to manage risks and prevent accidents, misuse, and other adverse outcomes stemming from their work.
In a new opinion piece for The Economist, Jaime Yassif makes the case for urgent action to create stronger guardrails for bioscience and biotechnology.
NTI | bio convened senior leaders from around the world for a scenario-based tabletop exercise designed to identify gaps in global capabilities to prevent and respond to high-consequence biological events.