There is no question that there will be another influenza pandemic someday. We simply don´t know when it will occur or whether it will be caused by the H5N1 avian influenza virus. But given the number of cases of H5N1 influenza that have occurred in humans to date (251 as of late September 2006) and the rate of death of more than 50%, it would be prudent to develop robust plans for dealing with such a pandemic.

The epicenters of both the Asian influenza pandemic of 1957 and the Hong Kong influenza pandemic of 1968 were in Southeast Asia, and it is in this region that multiple clades of H5N1 influenza virus have already emerged. The Asian H5N1 virus was first detected in Guangdong Province, China, in 1996, when it killed some geese, but it received little attention until it spread through live-poultry markets in Hong Kong to humans in May 1997, killing 6 of 18 infected persons. The culling of all poultry in Hong Kong ended the first wave of H5N1, but the virus continued to circulate among apparently healthy ducks in the coastal provinces of China.

From 1997 to May 2005, H5N1 viruses were largely confined to Southeast Asia, but after they had infected wild birds in Qinghai Lake, China, they rapidly spread westward. The deaths of swans and geese marked H5N1´s spread into Europe, India, and Africa. Infections with highly pathogenic H5N1 viruses were confirmed in poultry in Turkey in mid-October 2005, and the first confirmed human cases in Turkey occurred in early January 2006. Thus, H5N1 influenza viruses continue to emerge from the epicenter.

The H5N1 viruses can be divided into clade 1 and clade 2; the latter can be further subdivided into three subclades. The bad news is that these clades and subclades probably differ sufficiently in their antigenic structure to warrant the preparation of different vaccines. Studies in ferrets suggest that vaccine against one clade will not protect against infection with another clade, though it will protect against influenza-associated death.¹ Thus, the available information supports the notion that a vaccine against H5N1 is worth stockpiling as a "prepandemic" vaccine, since very few persons have been immunologically exposed to H5 antigens and priming with one clade may be beneficial.

Another key question is whether these clades and subclades vary in sensitivity to available anti-influenza drugs. The majority of H5N1 clade 1 viruses (e.g., A/Vietnam/1203/2004) are resistant to the adamantanes (amantadine and rimantadine), but the majority of clade 2 viruses (e.g., A/Indonesia/5/2005) are sensitive. All H5N1 viruses that have been tested are sensitive to the neuraminidase inhibitors; these drugs may be effective when used prophylactically, but the window for effective treatment will probably be limited to 1 to 2 days after initial infection. Kandun et al. make clear in their report in this issue of the Journal (pages 2186–2194) on three clusters of patients with H5N1 infection in Indonesia that the difficulty with the use of a neuraminidase inhibitor (oseltamivir) in those cases was that treatment began 5 to 7 days after initial infection. Such delayed administration of the drug limits its value in decreasing the viral load and might lead to the selection of resistant strains.

The use of rapid diagnostics for H5N1 virus infection can permit specific antiviral treatments to be initiated early. Oner et al. report in this issue of the Journal (pages 2179–2185) that in a human outbreak of H5N1 in Turkey, it was difficult to detect H5N1 virus infection with standard techniques; the authors found that a real-time polymerase-chain-reaction assay performed on nasopharyngeal specimens had the best diagnostic value.

The continuing evolution of H5N1 viruses and the clusters of human infections in Indonesia and Turkey raise important questions. First, can the source of H5N1 be eliminated? And second, is the increasing number of clusters of human infection an indicator of evolution toward consistent human-to-human transmission?

Controlling H5N1 influenza by eradicating it at the source in domestic poultry has worked for some wealthy countries: in 2003, Japan and South Korea eradicated H5N1 through a strategy of quarantine and culling of poultry and implementation of improved biosecurity measures for poultry facilities. In Thailand, however, the same strategy resulted in only a temporary respite; after nearly a year with no H5N1 activity, new cases in humans in July 2006 heralded the resurgence of H5N1 in domestic poultry.

An alternative strategy adopted by China, Indonesia, and Vietnam has been to vaccinate uninfected poultry in conjunction with the quarantine and culling of infected birds. This approach has failed, however, and its critics explain that poultry vaccines are largely of poor quality, do not provide sterilizing immunity, and promote antigenic drift. Yet vaccines against H5N1 influenza virus have been used successfully since 2004 on all poultry sold in Hong Kong, where no H5N1 virus has been isolated from fowl in live-bird markets despite extensive prospective surveillance.

Perhaps the most important experiment in controlling H5N1 is one that is ongoing in Vietnam. Since the country adopted a strategy of vaccinating all poultry with inactivated, oil-emulsion H5N1 vaccine, there have been no additional cases in humans and no reported H5N1 infections in chickens. But in September 2006, H5N1 was reported to have reemerged in ducks and geese in Vietnam. Thus, H5N1 influenza vaccine seems to protect chickens and, indirectly, humans, but probably not waterfowl.

Given that the vaccine predominantly used in Vietnam is prepared in China, where the policy is to vaccinate all poultry, some have questioned why H5N1 is not under control in China. The problem may be the lack of protection in waterfowl. Ducks may be the stealth carriers (the Trojan horses of H5N1 influenza), for wild mallard ducks do not always show signs of disease when infected with any of a range of highly pathogenic H5N1 viruses.² Our knowledge about the efficacy of H5N1 influenza vaccines in domestic waterfowl is limited, and highly pathogenic H5N1 viruses continue to be isolated from waterfowl in the epicenter of the epidemic. If the reservoir of highly pathogenic H5N1 virus is domestic waterfowl, the virus should theoretically be eradicable, but eliminating it would require improved vaccines for waterfowl and draconian prospective surveillance and culling.

Meanwhile, the number of infections in humans continues to increase. By mid-August, 97 humans had been infected in 2006 — the same number as in all of 2005. Perhaps the most surprising thing about highly pathogenic H5N1 is that although more than 230 million domestic birds have died or been killed, only 251 humans have become ill from H5N1 infection, and there has been little or no evidence of subclinical infection in humans. The current H5N1 virus is apparently not well "fitted" to replication in humans, although the genetic makeup of a small proportion of humans supports attachment and replication of the virus, if not its transmission. The specific receptor for the current avian influenza virus (2-3 sialic acid) is found deep in the respiratory tract of humans,³ but it seems likely that only a minority of people have receptors for avian influenza viruses in their upper respiratory tracts. Moreover, receptor specificity is only one of the requirements for human infection; the virus must also find compatible enzyme systems in the infected human cells if the viral polymerase complex is to function. Currently, these conditions are apparently met in only a few persons. But the virus is always changing, and mutations that make it more compatible with human transmission may occur at any time.

The seasonality of H5N1 influenza seems similar to that of human influenza: the virus has apparently been more transmissible among chickens, and consequently to humans, during the cooler months. The cases in humans in Turkey, Iraq, and Egypt occurred during the cooler months and coincided with explosive outbreaks of the disease in wild and domestic poultry. In the tropical areas of Asia, there have been two resurgences of H5N1 during the warmer months of the year — a pattern that resembles that followed by human influenza in the tropics, with its multiple peaks of activity. With winter approaching in the northern hemisphere, H5N1 may spread further. Will it cross from Eurasia to the Americas? Will wild migratory birds carry it from their breeding sites in northern Europe and Siberia to commercial poultry in Europe, Africa, and America? If it is endemic in wild migratory birds that are not rapidly killed by it, then spread to domestic backyard poultry is inevitable. The intermittent spread to humans will continue, and the virus will continue to evolve.

Clearly, we must prepare for the possibility of an influenza pandemic. If H5N1 influenza achieves pandemic status in humans — and we have no way to know whether it will — the results could be catastrophic.