|Home||Books Journals Flying Publisher Guides|
PDF, 225 pages, 2.8 MB
Publish this book under your own name
(Green links: Free full-text articles)
Highly pathogenic avian influenza, or, as
it was termed originally,
Highly pathogenic avian influenza, or, as it was termed originally,'fowl plague', was initially recognised as an infectious disease of birds in chickens in Italy, 1878 (Perroncito 1878). Due to a former hot spot in the Italian upper Po valley it was also referred to as 'Lombardian disease'. Although Centanni and Savonuzzi, in 1901, identified a filtrable agent responsible for causing the disease, it was not before 1955 that Schäfer characterised these agents as influenza A viruses (Schäfer 1955). In the natural reservoir hosts of avian influenza viruses, wild water birds, the infection generally runs an entirely asymptomatic course as influenza A virus biotypes of low pathogenicity co-exist in almost perfect balance with these hosts (Webster 1992, Alexander 2000).
When low pathogenic avian influenza virus (LPAIV) strains are transmitted from avian reservoir hosts to highly susceptible poultry species such as chickens and turkeys (i.e., a transspecies transmission step!), only mild symptoms are induced in general. However, in cases where the poultry species supports several cycles of infection, these strains may undergo a series of mutation events resulting in adaptation to their new hosts. Influenza A viruses of the subtypes H5 and H7 not only run through a host adaptation phase but may have the capability to saltatorily switch by insertional mutations into a highly pathogenic form (highly pathogenic avian influenza viruses, HPAIV) inducing overwhelming systemic and rapidly fatal disease. Such HPAI viruses may arise unpredictably de novo in poultry infected with LPAI progenitors of H5 and H7 subtypes.
HPAI in poultry is characterised by a sudden onset, severe illness of a short duration, and a mortality approaching virtually 100 % in vulnerable species. Due to excessive economical losses to the poultry industry, HPAI receives immense attention in the veterinary world and is globally treated as a disease immediately notifiable on suspicion to the authorities. Because of their potential to give rise to HPAIV, LPAI caused by subtypes H5 and H7 is also considered notifiable (OIE 2005). Before 1997, HPAI was fortunately a rare disease, with only 24 recorded primary outbreaks globally since the 1950s (Table 1).
Recently, however, avian influenza acquired world-wide attention when a highly pathogenic strain of the subtype H5N1, which probably arose before 1997 in Southern China, gained enzootic status in poultry throughout South East Asia and unexpectedly 'traversed interclass barriers' (Perkins and Swayne 2003) when transmitted from birds to mammals (cats, swine, humans). Although not an entirely unprecedented event (Koopmans 2004, Hayden and Croisier 2005), the substantial number of documented cases in humans, associated with severe disease and several fatalities raised serious concerns about a pandemic potential of the H5N1 strain (Klempner and Shapiro 2004; Webster 2006). There are several further lines of evidence - which will be discussed below - suggesting that the H5N1 virus has acquired increased pathogenic potency for several mammal species. Justifiably, this has caused world-wide public concern (Kaye and Pringle 2005).
1Modified from Capua and Mutinelli, 2001
* Outbreaks with significant spread to numerous farms, resulting in great economic losses. Most other outbreaks were associated with only restricted or no spread from the index farms.
Influenza viruses are spherically or longitudinally shaped enveloped particles with an up to eight-fold segmented, single-stranded RNA genome of negative polarity. Influenza viruses hold generic status in the Orthomyxoviridae family and are classiﬁed into types A, B or C based on antigenic differences of their nucleo- and matrix proteins. Avian influenza viruses (AIV) belong to type A. Excellent reviews on the structure and replication strategy of influenza viruses have been published recently (e.g. Sidoronko and Reichl 2005).
The main antigenic determinants of influenza A and B viruses are the haemagglutinin (H or HA) and the neuraminidase (N or NA) transmembrane glycoproteins, capable of eliciting subtype-speciﬁc and immune responses which are fully protective within, but only partially protective across, different subtypes. On the basis of the antigenicity of these glycoproteins, influenza A viruses currently cluster into sixteen H (H1 - H16) and nine N (N1 - N9) subtypes. These clusters are substantiated when phylogenetically analysing the nucleotide and deduced amino acid sequences of the HA and NA genes, respectively (Fouchier 2005).
The conventional nomenclature for influenza virus isolates requires connotation of the influenza virus type, the host species (omitted in the case of human origin), the geographical site, serial number, and year of isolation. For influenza virus type A, the haemagglutinin and neuraminidase subtypes are added in brackets. One of the parental avian strains of the current outbreaks of H5N1 of Asian lineage was isolated from a goose in the Chinese province, Guangdong: accordingly, it is designated A/goose/Guangdong/1/96 (H5N1) (Xu 1999) while the isolate originating from the ﬁrst-documented human case of Asian lineage H5N1 infection from Hong Kong (Claas 1998) is referred to as A/HK/156/97 (H5N1).
The haemagglutinin, a glycosylated and acylated protein consisting of 562 - 566 amino acids, is incorporated in the viral envelope. The globular head of its membrane-distal, knob-like external domain is associated with binding to cellular receptors composed of oligosaccharides which terminally carry derivates of neuraminic acid (Watowich 1994). The exodomain of the second transmembrane glycoprotein, the neuraminidase (NA), exerts sialolytic enzymatic activity and liberates virus progeny captured at the surface of infected cells during egress. This function prevents viral aggregation during egress, and possibly also facilitates the drifting of the virus through the mucus layers of the targeted epithelial tissues leading to viral attachment (Matrosovich 2004a). This renders the neuraminidase an interesting target of antiviral agents (Garman and Laver 2004). Mutually attuned and co-ordinated actions of the antagonistic glycoprotein species HA and NA of a viral strain are pivotal for effective attachment and release processes of the virions (Wagner 2002).
Attachment to cell surface proteins of inﬂuenza A virions is achieved through mature trimerised viral HA glycoproteins. Attachment is stratiﬁed by recognition of distinct terminal sialic acid species (N-acetyl- or N-glycolylneuraminic acid), the type of glycosidic linkage to penultimate galactose (α2-3 or α2-6) and the composition of further inner fragments of sialyloligosaccharides present at the cell surface (Herrler 1995, Gambaryan 2005). A variety of different sialyloligosaccharides are expressed with restriction to tissue and species origin in the different hosts of influenza viruses. Adaptation in both the viral HA and the NA glycoprotein to the speciﬁc receptor type(s) of a certain host species is a prerequisite for efficient replication (Ito 1999, Banks 2001, Matrosovich 1999+2001, Suzuki 2000, Gambaryan 2004). This implies a re-shaping of the receptor binding units of the HA protein following interspecies transmission (Gambaryan 2006). A mechanistic overview of the diverse receptor types is given in ﬁgure 1. Avian influenza viruses generally show the highest affinities for α2-3 linked sialic acid as this is the dominating receptor type in epithelial tissues of endodermic origin (gut, lung) in those birds that are targeted by these viruses (Gambaryan 2005a, Kim 2005). Human-adapted influenza viruses, in contrast, primarily access 2-6 linked residues which predominate on non-ciliated epithelial cells of the human airway. These receptor predilections define part of a species barrier preventing hassle-free transmission of avian viruses to humans (Suzuki 2000, Suzuki 2005). Yet recently, it has been shown that there is a population of ciliated epithelial cells in the human trachea which also carry avian receptor-like glycoconjugates at lower densities (Matrosovitch 2004b), and also chicken cells carry human-type sialyl receptors at low concentrations (Kim 2005). This might explain why humans are not entirely refractory towards infection with certain avian strains (Beare and Webster 1991). In pigs, and also in quails, both
receptor types are present at higher densities which renders these species putative mixing vessels for avian and human strains (Kida 1994, Ito 1998, Scholtissek 1998, Peiris 2001, Perez 2003, Wan and Perez 2005).
Figure 1. Overview of receptor predilections of influenza A viruses (based on data by Gambaryan 2005)
Once successfully attached to a suitable receptor, the virion is internalised into an endosomal compartment by clathrin-dependent and -independent mechanisms (Rust 2004). The virus escapes degradation in this compartment by fusing viral and endolysomal membranes: mediated by proton transport through the viral matrix-2 (M2) tunnel protein at pH values in the endosome of around 5.0, a cascade of steric rearrangements in the matrix-1 (M1) proteins andthe homotrimeric HA glycoprotein complex commence. As a result, a highly lipophilic, fusogenic domain of each HA monomere is exposed which inserts itself into the endolysosomal membrane, thereby initiating fusion of viral and lysosomal membranes (Haque 2005, Wagner 2005). In turn, the eight viral genomic RNA segments, enclosed in a protective layer of nucleocapsid (N) proteins (ribonucleoprotein complex, RNP) are released into the cytoplasm. Here they are transported to the nucleus for transcription of viral mRNAs and replication of genomic RNA in a complex process which is delicately regulated by viral and cellular factors (Whittaker 1996). The RNA-dependent RNA polymerase (RdRp) is formed by a complex of the viral PB1, PB2 and PA proteins, and requires encapsidated RNA (RNPs) for this task. Upon translation of viral proteins and assembly of nucleocapsids harbouring replicated genomic RNA, progeny virions bud from the cellular membrane into which the viral glycoproteins have previously been inserted. Arrangements between helical nucleocapsids and viral envelope proteins are mediated by the viral matrix-1 (M1) protein which forms a shell-like structure just beneath the viral envelope. Viral reproduction in fully permissive cells is a fast (less than ten hours) and efﬁcient process, provided an 'optimal' gene constellation is present (Rott 1979, Neumann 2004).
Due to the error-prone activity of the viral RdRp, a high mutation rate of≥ 5 x 10-5 nucleotide changes per nucleotide and replication cycle, thus approaching almost one nucleotide exchange per genome per replication, is observed among the influenza viruses (Drake 1993). In case selective pressures (such as neutralising antibodies, suboptimal receptor binding or chemical antivirals) are acting during viral replication on a host or population scale, mutants with corresponding selective advantages (e.g. escape from neutralisation, reshaped receptor-binding units) may be singled out and become the dominant variant within the viral quasispecies in that host or population. If antigenic determinants of the membrane glycoproteins HA and NA are affected by mechanisms driven by immunity, such a (gradual) process is referred to as antigenic drift (Fergusson 2003).
Antigenic shift, in contrast, denotes a sudden and profound change in antigenic determinants, i.e. a switch of H and/or N subtypes, within a single replication cycle. This occurs in a cell which is simultaneously infected by two or more influenza A viruses of different subtypes. Since the distribution of replicated viral genomic segments into budding virus progeny occurs independently from the subtype origin of each segment, replication-competent progeny carrying genetic information of different parental viruses (so-called reassortants) may spring up (Webster and Hulse 2004,WHO 2005). While the pandemic human influenza viruses of 1957 (H2N2) and 1968 (H3N2) clearly arose through reassortment between human and avian viruses, the influenza virus causing the 'Spanish flu' in 1918 appears to be entirely derived from an avian source (Belshe 2005).
Wild aquatic birds, notably members of the orders Anseriformes (ducks and geese) and Charadriiformes (gulls and shorebirds), are carriers of the full variety of influenza virus A subtypes, and thus, most probably constitute the natural reservoir of all influenza A viruses (Webster 1992, Fouchier 2003, Krauss 2004,Widjaja 2004). While all bird species are thought to be susceptible, some domestic poultry species - chickens, turkey, guinea fowl, quail and pheasants - are known to be especially vulnerable to the sequelae of infection.
Avian influenza A viruses generally do not cause disease in their natural hosts. Instead, the viruses remain in an evolutionary stasis, as molecularly signalled by low N/S (non-synonymous vs. synonymous) mutation ratios indicating purifying evolution (Gorman 1992, Taubenberger 2005). Host and virus seem to exist in a state of a meticulously balanced mutual tolerance, clinically demonstrated by absence of disease and efﬁcient viral replication. Large quantities of virus of up to 108.7 x 50% egg-infective dose (EID50) per gram faeces can be excreted (Webster 1978). When transmitted to highly vulnerable poultry species, usually mild, if any, symptoms ensue. Viruses of this phenotype are referred to as low pathogenic (LPAIV) and, in general, only cause a slight and transient decline in egg production in layers or some reduction in weight gain in fattening poultry (Capua and Mutinelli 2001). However, strains of the subtypes H5 and H7 carry the potential to mutate to a highly pathogenic form after transmission and adaptation to the new poultry hosts. Nascency of highly pathogenic forms of H5 and H7 or of other subtypes has never been observed in wild birds (Webster 1998). Therefore, one may even come to look at the highly pathogenic forms as something artiﬁcial, made possible only as a result of man-made interference with a naturally balanced system.
Once HPAIV phenotpyes have arisen in domestic poultry, they can be transmitted horizontally from poultry back into the wild bird population. The vulnerability of wild birds towards HPAIV-induced disease appears to vary grossly according to species, age and viral strain. Until the emergence of the Asian lineage H5N1 HPAI viruses, spill-overs of HPAIV into the wild bird population occurred sporadically and were locally restricted (with the single exception of a die-off among terns in South Africa in 1961 [Becker 1966]), so that wild birds had not been assigned an epidemiologically important function in the spread of HPAIV (Swayne and Suarez 2000). This might have changed fundamentally since early 2005, when a large outbreak of the Asian lineage H5N1-related HPAI was observed among thousands of wild aquatic birds in a nature reservation at Lake Qinghai in the North West of China (Chen 2005, Liu 2005). As a result of this, further spread of this virus towards Europe during 2005 may have been founded (OIE 2005). The details and consequences of this process are described below.
Figure 2. Scheme of avian influenza pathogenesis and epidemiology
LPAIV - low pathogenic avian influenza virus; HPAIV - highly pathogenic avian influenza virus; HA - haemagglutinin protein; dotted lines with arrows represent species barriers
Pathogenesis of HPAI
Pathogenicity as a general viral property in influenza A viruses is a polygenic trait and depends largely on an'optimal' gene constellation affecting host and tissue tropism, replication efﬁcacy and immune evasion mechanisms, amongst others. In addition, host- and species-speciﬁc factors contribute to the outcome of infection, which, after interspecies transmission, is therefore unpredictable a priori. The highly pathogenic form of avian influenza has been caused to date by influenza A viruses of the H5 and H7 subtypes exclusively. However, only a few representatives of the H5 and H7 subtypes in fact display a highly pathogenic biotype (Swayne and Suarez 2000). Usually, H5 and H7 viruses are stably maintained in their natural hosts in a low pathogenic form. From this reservoir, the viruses can be introduced by various pathways (see below) into poultry ﬂocks. Following a variable and indecisive period of circulation (and, presumably, adaptation) in susceptible poultry populations, these viruses can saltatorily mutate into the highly pathogenic form (Rohm 1995).
Nucleotide sequencing studies have shown that most HPAIVs share a common feature in their HA genes which can serve, in poultry, as a virulence marker (Webster 1992, Senne 1996, Perdue 1997, Steinhauer 1999, Perdue and Suarez 2000):
In order to gain infectivity, influenza A virions must incorporate HA proteins which have been endoproteolytically processed from a HA0 precursor to a disulphide-linked HA1,2 dimer (Chen 1998). The newly created N-terminus of the HA2 subunit harbours a fusogenic peptide, composed of a highly lipophilic domain (Skehel 2001). This domain is vitally required during the fusion process of viral and lysosomal membranes because it initiates the penetration process of viral genomic segments into the host cell cytoplasm. The cleavage site of the HA of low pathogenic viruses is composed of two basic amino acids at positions -1/-4 (H5) and -1/-3 (H7) (Wood 1993). These sites are accessible to tissue-speciﬁc trypsin-like proteases which are preferentially expressed at the surface of respiratory and gastrointestinal epithelia. Therefore, efﬁcient replication of LPAIVs is believed to be largely confined to these sites, at least in their natural hosts. In contrast, the cleavage site of HPAI viruses generally contains additional basic amino acids (arginine and/or lysine) which renders it processible for subtilysin-like endoproteases speciﬁc for a minimal consensus sequence of -R-X-K/R-R- (Horimoto 1994, Rott 1995). Proteases of this type (e.g. furin, proprotein-convertases) are active in virtually every tissue throughout the body. Therefore, viruses carrying these mutations have an advantage for replicating unrestrictedly in a systemic manner. This process has been documented in the field on several occasions. In Italy, for example, an LPAI H7N1 virus circulated for several months in the turkey and chicken population before, in December 1999, an HPAI H7N1 virus, distinguishable from its precursor only by its polybasic cleavage site, sprang up and caused devastating disease (Capua 2000).
It has been hypothesised that the HA gene of the H5 and H7 subtypes harbour distinct secondary RNA structures which favour insertional mutations (codon duplications) by a re-copying mechanism of the viral polymerase unit at a purine-rich sequence stretch encoding the endoproteolytic cleavage site of these HA proteins (Garcia 1996, Perdue 1997). This, and probably other mechanisms too, such as nucleotide substitutions or intersegmental recombination (Suarez 2004, Pasick 2005), may lead to the incorporation of additional basic amino acid residues. The latter has been experimentally proven by the generation of HPAIV from LPAIV precursors following repeated passaging in vitro and in vivo by site-directed mutagenesis (Li 1990, Walker and Kawaoka 1993, Horimoto and Kawaoka 1995, Ito 2001). Conversely, removal by reverse genetics of the polybasic cleavage site attenuates the HPAI phenotype (Tian 2005).
There are, however, viral strains in which the nucleotide sequence encoding the HA cleavage site and the pheno-/pathotype did not match in the predicted way: a Chilean H7N3 HPAIV which arose by intersegmental recombination displayed basic amino acid residues only at positions -1, -4 and -6 (Suarez 2004). Comparable examples exist for the H5 lineage (Kawaoka 1984). On the other hand, an H5N2 isolate from Texas was shown to harbour the HPAIV cleavage site consensus sequence, yet was clinically classiﬁed as LPAI (Lee 2005). These data re-emphasise the polygenic and intricate nature of influenza virus pathogenicity.
Fortunately, nascency of HPAI phenotypes in the field appears to be a rare event. During the last ﬁfty years, only 24 primary HPAI outbreaks caused by HPAIV, which likely arose de novo in this way in the field, have been reported world-wide (Table 1).
In addition, HPAIV have been shown to be able to infect mammals, and humans in particular. This has especially been observed for the Asian lineage H5N1 (WHO 2006). Host-dependent pathogenicity of HPAIV H5N1 for mammals has been studied in several model species: mice (Lu 1999, Li 2005a), ferrets (Zitzow 2002, Govorkova 2005), cynomolgous monkeys (Rimmelzwaan 2001) and pigs (Choi 2005). The outcome of infection was dependent on the viral strain and species of host. Ferrets appeared to mirror pathogenicity in humans better than mice (Maines 2005).
A number of genetic markers believed to be involved in pathogenicity have been located in different segments of the Z genotype of H5N1 (Table 2). Among these, mechanisms of interference with ﬁrst-line defence mechanisms of the host, such as the interferon system, through the NS-1 gene product have received marked interest. Experimentally, it has been demonstrated using reverse genetics, that NS-1 proteins of some H5N1 strains carrying glutamic acid at position 92 are capable of circumventing the antiviral effects of interferon and tumour necrosis factor-alpha, eventually leading to enhanced replication in, and reduced clearance from, the infected host (Seo 2002+2004). In addition, immune-mediated damage resulting from NS-1-mediated disruption of cytokine networks may account for parts of the lung lesions (Cheung 2002,Lipatov 2005). However, none of the mutations (Table 2) on its own represents a true prerequisite for pathogenicity in mammals (Lipatov 2003). Therefore, optimal gene constellations, to a large extent, appear to drive pathotype speciﬁcities in a host-dependent manner in mammals (Lipatov 2004).
Following an incubation period of usually a few days (but rarely up to 21 days), depending upon the characteristics of the isolate, the dose of inoculum, the species, and age of the bird, the clinical presentation of avian influenza in birds is variable and symptoms are fairly unspeciﬁc (Elbers 2005). Therefore, a diagnosis solely based on the clinical presentation is impossible.
The symptoms following infection with low pathogenic AIV may be as discrete as rufﬂed feathers, transient reductions in egg production or weight loss combined with a slight respiratory disease (Capua and Mutinelli 2001). Some LP strains such as certain Asian H9N2 lineages, adapted to efﬁcient replication in poultry, may cause more prominent signs and also significant mortality (Bano 2003, Li 2005).
In its highly pathogenic form, the illness in chickens and turkeys is characterised by a sudden onset of severe symptoms and a mortality that can approach 100 % within 48 hours (Swayne and Suarez 2000). Spread within an affected ﬂock depends on the form of rearing: in herds which are litter-reared and where direct contact and mixing of animals is possible, spread of the infection is faster than in caged holdings but would still require several days for complete contagion (Capua 2000). Often, only a section of a stable is affected. Many birds die without premonitory signs so that sometimes poisoning is suspected in the beginning (Nakatami 2005). It is worth noting, that a particular HPAI virus isolate may provoke severe disease in one avian species but not in another: in live poultry markets in Hong Kong prior to a complete depopulation in 1997, 20 % of the chickens but only 2.5 % of ducks and geese harboured H5N1 HPAIV while all other galliforme, passerine and psittacine species tested virus-negative and only the chickens actually showed clinical disease (Shortridge 1998).
In industrialised poultry holdings, a sharp rise followed by a progressive decline in water and food consumption can signal the presence of a systemic disease in a ﬂock. In laying ﬂocks, a cessation of egg production is apparent. Individual birds affected by HPAI often reveal little more than severe apathy and immobility (Kwon 2005). Oedema, visible at feather-free parts of the head, cyanosis of comb, wattles and legs, greenish diarrhoea and laboured breathing may be inconsistently present. In layers, soft-shelled eggs are seen initially, but any laying activities cease rapidly with progression of the disease (Elbers 2005). Nervous symptoms including tremor, unusual postures (torticollis), and problems with co-ordination (ataxia) dominate the picture in less vulnerable species such as ducks, geese, and ratites (Kwon 2005). During an outbreak of HPAI in Saxonia, Germany, in 1979, geese compulsively swimming in narrow circles on a pond were among the ﬁrst conspicuous signs leading to a preliminary suspicion of HPAI.
The clinical presentation of avian influenza infection in humans is discussed in detail in the chapter entitled'Clinical Presentation of Human Influenza'.
Lesions vary with the viral strain and the species and age of the host. In general, only turkeys and chickens reveal any gross and microscopic alterations especially with strains adapted to these hosts (Capua and Mutinelli 2001). In turkeys, sinusitis, tracheitis and airsacculitis have been detected, although secondary bacterial infections may have contributed as well. Pancreatitis in turkeys has been described. In chickens, mild involvement of the respiratory tract is most commonly seen. In addition, lesions concentrate on the reproductive organs of layers (ovaries, oviduct, yolk peritonitis).
Gross pathological and histopathological alterations of HPAI reveal similar dependencies to those listed for the clinical presentation. Four classes of pathological alterations have been tentatively postulated (Perkins and Swayne 2003):
(i) Peracute (death within 24-36 hours post infection, mainly seen in some galliforme species) and acute forms of disease reveal no characteristic gross pathological alterations: a discrete hydropericardium, mild intestinal congestion and occasionally petechial bleedings of the mesenterical and pericardial serosa have been inconsistently described (Mutinelli 2003a, Jones and Swayne 2004). Chickens infected with the Asian lineage H5N1 sometimes reveal haemorrhagic patches and significant amounts of mucus in the trachea (Elbers 2004). Serous exudates in body cavities and pulmonary oedema may be seen as well. Pinpoint bleedings in the mucosa of the proventriculus, which were often described in text books in the past, have only exceptionally been encountered in poultry infected with the Asian lineage H5N1 (Elbers 2004). Various histological lesions together with the viral antigen can be detected throughout different organs (Mo 1997). The virus is ﬁrst seen in endothelial cells. Later on virus-infected cells are detected in the myocardium, adrenal glands and pancreas. Neurons as well as the glial cells of the brain also become infected. Pathogenetically, a course similar to other endotheliotropic viruses may be assumed, where endothelial and leukocyte activation leads to a systemic and uncoordinated cytokine release predisposing to cardiopulmonary or multi-organ failure (Feldmann 2000, Klenk 2005).
(ii) In animals which show a protracted onset of symptoms and a prolonged course of disease, neurological symptoms and, histologically, non-suppurative brain lesions predominate the picture (Perkins and Swayne 2002a, Kwon 2005). However, virus can also be isolated from other organs. This course has been described in geese, ducks, emus and other species experimentally infected with an Asian lineage HPAI H5N1 strain. In laying birds, inﬂammation of the ovaries and oviducts, and, after follicle rupture, so-called yolk peritonitis, can be seen.
(iii) In ducks, gulls and house sparrows, only restricted viral replication was found. These birds showed mild interstitial pneumonia, airsacculitis and occasionally lymphocytic and histiocytic myocarditis (Perkins and Swayne 2002a, 2003).
(iv) In the experiments described by Perkins and Swayne (2003), pigeons and starlings proved to be resistant against H5N1 infection. However, Werner et al. (to be published) were able to induce protracted neurological disease, due to non-suppurative encephalitis (Klopﬂeisch 2006), in 5/16 pigeons using a recent Indonesian HPAI H5N1 isolate.
The following diseases must be considered in the differential diagnosis of HPAI because of their ability to cause a sudden onset of disease accompanied by high mortality or haemostasis in wattles and combs:
Less severe forms of HPAI can be clinically even more confusing. Rapid laboratory diagnostic aid, therefore, is pivotal to all further measures (Elbers 2005).
Collection of Specimens
Specimens should be collected from several fresh carcasses and from diseased birds of a ﬂock. Ideally, adequate sampling is statistically backed up and diagnosis is made on a ﬂock basis. When sampling birds suspected of HPAI, safety standards must be observed to avoid exposure of the sample collectors to potentially zooanthroponotic HPAIV (Bridges 2002). Guidelines have been proposed by the CDC (CDC 2005).
For virological assays, swabs obtained from the cloaca and the oropharynx generally allow for a sound laboratory investigation. The material collected on the swabs should be mixed into 2-3 ml aliquots of a sterile isotonic transport medium containing antibiotic supplements and a protein source (e.g. 0.5 % [w/v] bovine serum albumin, up to ten percent of bovine serum or a brain-heart infusion).
At autopsy, carried out under safe conditions and avoiding spread of disease (see above), unpreserved specimens of brain, trachea/lung, spleen and intestinal contents are collected for isolation of the virus.
For serological purposes, native blood samples are taken. The number of samples collected should sufﬁce detection with a 95 % conﬁdence interval for a parameter with a prevalence of 30 %.
Transport of Specimens
Swabs, tissues and blood should be transported chilled but not be allowed to freeze. If delays of greater than 48 hours are expected in transit, these specimens should be frozen and transported on dry ice. In all cases, transport safety regulations (e.g. IATA rules) should be punctiliously observed to avoid spread of the disease and accidental exposure of personnel during transport. It is highly advisable to contact the assigned diagnostic laboratory before sending the samples and, ideally, even before collecting them.
Direct Detection of AIV Infections
Basically, there are two (parallel) lines of diagnostic measures that attempt to (i) isolate and subtype the virus by classical methods (seeOIE Manual 2005) and (ii) molecularly detect and characterise the viral genome.
(i) Conventionally, AI virus is isolated by inoculation of swab ﬂuids or tissue homogenates into 9- to 11-day-old embryonated chicken eggs, usually by the chorioallantoic sac route (Woolcock 2001). Depending on the pathotype, the embryos may or may not die within a ﬁve-day observation period and usually there are no characteristic lesions to be seen in either the embryo or the allantois membrane (Mutinelli 2003b). Eggs inoculated with HPAIV-containing material usually die within 48 hours. The presence of a haemagglutinating agent can be detected in harvested allantoic fluid. Haemagglutination (HA) is an insensitive technique requiring at least 106.0 particles per ml. If only a low virus concentration is present in the inoculum, up to two further passages in embryonated eggs may be neccessary for some LPAIV strains, in order to produce enough virus to be detected by HA. In the case of HPAIV, a second passage using diluted inoculum may be advantageous for the optimal production of haemagglutinating .
Haemagglutinating isolates are antigenically characterised by haemagglutination inhibition (HI) tests using (mono-) speciﬁc antisera against the 16 H subtypes and, for control, against the different types of avian paramyxoviruses which also display haemagglutinating activities. The NA subtype can be subsequently determined by neuraminidase inhibition assays, again requiring subtype-speciﬁc sera (Aymard 2003). In case isolates of the H5 or H7 lineages are encountered, their intravenous pathogenicity index (IVPI) needs to be determined to distinguish between LP and HP biotypes (Allan 1977). This is achieved by iv inoculation of ten 6-week old chickens with the egg-grown virus isolate (0.1 ml of a 1 in 10 dilution of allantoic fluid containing a HA titre greater than 1 in 16). The chickens are observed over a period of ten days for clinical symptoms. Results are integrated into an index which indicates a HPAI virus when values greater than 1.2 are obtained. Alternatively, a HPAI isolate is encountered when at least seven out of ten (75 %) inoculated chickens die within the observation period.
The described classical procedures can lead to a diagnosis of HPAI within five days but may demand more than a fortnight to rule out the presence of AIV. In addition, high quality diagnostic tools (SPF eggs, H- and N-subtype speciﬁc antisera) and skilled personnel are a prerequisite. Currently, there are no cell culture applications for the isolation of AIV that can achieve the sensitivity of embryonated hen eggs (Seo 2001).
(ii) A more rapid approach, especially when exclusion of infection is demanded, employs molecular techniques, which should also follow a cascade style: the presence of influenza A speciﬁc RNA is detected through the reverse transcription-polymerase chain reaction (RT-PCR) which targets fragments of the M gene, the most highly conserved genome segment of influenza viruses (Fouchier 2000, Spackman 2002), or the nucleocapsid gene (Dybkaer 2004). When a positive result is obtained, RT-PCRs amplifying fragments of the haemagglutinin gene of subtypes H5 and H7 are run to detect the presence of notiﬁable AIVs (Dybkaer 2004, Spackman 2002). When positive again, a molecular diagnosis of the pathotype (LP versus HP) is feasible after sequencing a fragment of the HA gene spanning the endoproteolytic cleavage site. Isolates presenting with multiple basic amino acids are classified as HPAI. PCRs and other DNA tion techniques are being designed for the detection of Asian lineage H5N1 strains (Collins 2002, Payungporn 2004,Ng 2005). Non-H5/H7 subtypes can be identified by a canonical RT-PCR and subsequent sequence analysis of the HA-2 subunit (Phipps 2004). There are also speciﬁc primers for each NA subtype. A full characterisation might be achievable within three days, especially when real time PCR techniques are used (Perdue 2003, Lee and Suarez 2004). However, DNA chips are in development which should further streamline the typing of AI viruses (Li 2001, Kessler 2005). An exclusion diagnosis is possible within a single working day.
The disadvantages of molecular diagnostics are the price one has to pay for purchasing equipment and consumables, although, if available, many samples can be analysed by less personnel in grossly shorter times in comparison to virus isolation in eggs. However, it should not be kept secret that each PCR or hybridisation reaction, in contrast to virus isolation in eggs, harbours an intrinsic uncertainty related to the presence of speciﬁc mutations in a given isolate at the binding sites of primers and/or probes which might render the assay false negative.
Thus, a combination of molecular (e.g. for screening purposes) and classical methods (e.g. for final characterisation of isolates and confirmation of diagnosis of an index case) may help to counterbalance the disadvantages of the two principles.
Rapid assays have been designed for the detection of viral antigen in tissue impression smears and cryostat sections by use of immunofluorescence, or by antigen-capture enzyme-linked immunosorbent assay (ELISA) and dip-stick lateral flow systems in swab fluids. So far, these techniques have been less sensitive than either virus isolation or PCR, and therefore might be difficult to approve for a legally binding diagnosis, especially of an index case (Davison 1998, Selleck 2003, Cattoli 2004). The use of pen side tests in the veterinary field is still in its infancy and needs further development.
Indirect Detection of AIV Infections
Serology on a herd basis may be useful for screening purposes (Beck 2003). For the detection of AIV-speciﬁc antibodies in serum samples from birds, or in egg yolk in the case of laying ﬂocks, the haemagglutination inhibition (HI) assay using reference subtype antigens still represents the gold standard. Group-speciﬁc antibodies (influenza virus type A) against the nucleocapsid protein can also be detected by agar gel immunoprecipitation and by enzyme-linked immunosorbent assays (ELISA) (Meulemans 1987, Snyder 1985, Jin 2004). Competitive ELISA formats allow the examination of sera of all bird species, independent from the availability of species-speciﬁc conjugates (Shafer 1998, Zhou 1998). An ELISA format for the detection of H7-speciﬁc antibodies has been reported (Sala 2003), but there is no such assay presently available for the detection of H5-speciﬁc antibodies in avian sera.
Subtype-speciﬁc antibody kinetics depend on the viral strain characteristics and, primarily, on the host species. In gallinaceous birds, AIV-speciﬁc antibodies reliably become detectable during the second week following exposure; antibodies in egg yolk are detectable after a delay of a few days (Beck 2003). The production and detection of antibodies in Anatidae species are much more variable (Suarez and Shultz-Cherry 2000).
Transmission between Birds
Avian influenza viruses of low pathogenicity circle genetically stable in wild water fowl (Webster 1992). The infection cycle among birds depends on faecal-oral transmission chains. Apart from being directly transmitted from host to host, indirect spread via virus-contaminated water and fomites is an important route in contrast to influenza virus infections in mammals (humans, swine, and horses) where transmission by aerosols prevails. In birds, peak excretion titres of up to 108.7 x 50 % egg-infective dose (EID50) per gram faeces have been measured (Webster 1978). Average titres will be grossly lower. Avian influenza viruses reveal an astonishing capability to retain infectivity in the environment and particularly in surface water in spite of their seemingly delicate morphology (Stallknecht 1990a+b, Lu 2003). Virus suspensions in water have been shown to retain infectivity for more than 100 days at 17°C. Below -50°C the virus can be stored indefinitely. Data provided by Ito et al. (1995) and Okazaki et al. (2000) provided evidence that in the palearctic regions, avian influenza viruses are preserved in frozen lake water during the winter in the absence of their migrating natural hosts. Upon return for breeding purposes during the subsequent season, returning birds or their (susceptible) offspring are re-infected with viruses released by chance from melting environmental water. Along these lines, it has been hypothesised that influenza viruses can be preserved in environmental ice for prolonged time periods (Smith 2004), and that ancient viruses and genotypes might be recycled from this reservoir (Rogers 2004).
The introduction of H5 or H7 subtypes of LPAI viruses to susceptible poultry ﬂocks is the basis of a chain of infection events which may lead to the de novo development of highly pathogenic biotypes. The risk that infection will be transmitted from wild birds to domestic poultry is greatest where domestic birds roam freely, share a water supply with wild birds, or use a water or food supply that might become contaminated by droppings from infected wild bird carriers (Capua 2003, Henzler 2003). Birds are infected by direct contact with virus-excreting animals and their excretions or through contact with (abiotic) vectors which are contaminated with virus-containing material. Once introduced into domestic ﬂocks, LPAIV may or may not depend on a phase of adaptation to poultry species before they are excreted in amounts large enough to ensure sustained horizontal transmission within and between ﬂocks. HPAIV, once it has arisen from an LPAIV infected ﬂock, spreads by similar means. So-called 'wet' markets, where live birds are sold under crowded conditions, are multiplicators of spread (Shortridge 1998, Bulaga 2003).
Biosecurity measures, aiming at the isolation of large poultry holdings, effectively prevent transmission from farm to farm by mechanical means, such as by contaminated equipment, vehicles, feed, cages, or clothing - especially shoes. An analysis of the Italian HPAI epizootic in 1999/2000 revealed the following risks for transmission: movements of infected ﬂocks (1.0 %), mediated contacts during transport of poultry to slaughter houses (8.5 %), neighbourhood within a one kilometre radius around infected premises (26.2 %), lorries used for transport of feed, bedding or carcasses (21.3 %), other indirect contacts through exchange of farm staff, working machines, etc. (9.4 %) (Marangon and Capua 2005). There were no hints at aerogenic spread obtained during the Italian epizootic. However, during outbreaks in the Netherlands (2003) and Canada (2004), airborne spread has been considered (Landman and Schrier 2004,Lees 2004). The role of live vectors such as rodents or flies, which may act as 'mechanical vectors' and are not themselves infected, is largely indetermined but certainly does not constitute a major factor.
Until the emergence of the Asian lineage H5N1 HPAIV, a re-introduction of HPAIV from poultry into the wild bird population had not played any significant role. In April 2005, however, Asian lineage H5N1-associated disease surfaced at Lake Qinghai in North Western China affecting thousands of bar-headed geese and other migratory species of ducks, cormorants and gulls (Chen 2005, Liu 2005). Therefore, transmission of Asian lineage H5N1 viruses by wild birds must be taken into account in future preventive concepts (discussed below).
Since late 2003, some H5N1 viruses have been encountered in Asia which were highly pathogenic for chickens but not for ducks (Sturm-Ramirez 2005). Experimental infections using these isolates revealed a heterogeneous mixture with respect to genetic analysis and plaque formation capacities in cell culture (Hulse Post 2005). Ducks that survived infection with these isolates were shown to shed a virus population on day 17 that had lost its pathogenic potential for ducks. When clinical signs are used to screen for the presence of HPAIV H5N1 in the field, ducks may become the 'Trojan horse' of this virus (Webster 2006).
Transmission to Humans
Transmission of avian influenza viruses to humans, leading to the development of clinically overt disease is a rare event (Table 3). Given the potential exposure of millions of people to HPAIV H5N1 in South East Asia, the actual number of documented human cases, although steadily growing over the past years, must still be considered as being comparatively low (http://www.who.int/csr/disease/avian_influenza/country/en).
The first association of the Asian lineage HPAIV H5N1 with respiratory illness in human beings was observed in Hong Kong in 1997, when six out of 18 H5N1 infected human cases died. These cases were epidemiologically linked to an outbreak of highly pathogenic H5N1 in live-bird markets (Yuen 1998, Claas 1998,Katz 1999). The risk of direct transmission of the H5N1 virus from birds to humans seems to be greatest in persons who have close contact with live infected poultry, or surfaces and objects heavily contaminated with their droppings. Exposure risk is considered substantial during slaughter, defeathering, butchering and preparation of poultry for cooking (http://www.who.int/csr/don/2005_08_18/en/). The Asian lineage HPAI H5N1 virus can be found in all tissues - including the meat - throughout the bird's carcass. In several such instances, it was reported that the person who slaughtered or prepared a sick bird for consumption developed fatal illness, while family members who participated in the meal did not (http://www.who.int/csr/don/2005_10_13/en/index.html).
* Source: Avian influenza - assessing the pandemic threat. WHO, http://www.who.int/csr/disease/influenza/WHO_CDS_2005_29/en/, accessed 06 January 2006.
** Highly pathogenic for poultry
A H9N2 strain caused mild, influenza-like symptoms in two children in Hong Kong SAR in 1999, and in one child in mid-December 2003 (Saito 2001, Butt 2005). The H9N2 strain circulating in poultry at these times provoked significant symptoms and lethality rates in highly vulnerable species such as turkeys and chickens.
To date, there is no evidence that properly cooked poultry meat or poultry products are a source of human infection by the Asian lineage H5N1. As a general rule, the WHO recommends that meat be thoroughly cooked, so that all parts of the meat reach an internal temperature of 70°C. At this temperature, influenza viruses are inactivated, thus rendering safe any raw poultry meat contaminated with the H5N1 virus (WHO 2005).
Transmission to other Mammals
Avian influenza viruses have been transmitted to different mammal species on several occasions. Here, following cycles of replication and adaptation, new epidemic lineages can be founded. Pigs, in particular, have been frequently involved in such 'interclass transversions'. In European pig populations, avian-like H1N1 viruses are highly prevalent (Heinen 2002) and an H1N2 virus, a human-avian reassortant virus, ﬁrst isolated in the U.K. in 1992, is constantly gaining ground (Brown 1998). In the U.S., a triple reassortant (H3N2) between the classical H1N1, the human H3N2 and avian subtypes is circulating (Olsen 2002). Other subtypes of presumably avian origin (e.g. H1N7, H4N6) have been found mainly anecdotally in swine (Brown 1997, Karasin 2000). A H9N2 virus of avian provenance is moderately prevalent in swine populations in the East of China (Xu 2004). In addition to swine, marine mammals and horses have been shown to acquire influenza A viruses from avian sources (Guo 1992, Ito 1999).
Natural infection with H5N1 was described in tigers and other large cats in a zoo in Thailand after the animals were fed with virus-positive chicken carcasses (Keawcharoen 2004, Quirk 2004, Amosin 2005). Severe disease accompanied by high mortality ensued. Also, cat-to-cat transmission has apparently occurred in the same zoo (Thanawongnuwech 2005). This was the ﬁrst report of influenza virus infections in Felidae. Household European short hair cats can experimentally be infected with the H5N1 virus (Kuiken 2004).
In 2004, 3,000 serum samples obtained from free roaming pigs in Vietnam were tested serologically for evidence of exposure to the H5N1 influenza virus (Choi 2005). Virus neutralisation assay and Western blot analysis confirmed that only 0.25 % of the samples were seropositive. In experimental infections, it was shown that pigs can be infected with H5N1 viruses isolated in Asia in 2004 from human and avian sources. A mild cough and elevated body temperature were the only symptoms observed for four days post infection. Virus could be isolated from tissues of the upper respiratory tract for at least 6 days. Peak viral titres from nasal swabs were found on day 2 post infection, but none of the experimentally infected animals transmitted the infection to contact pigs. The highly lethal H5N1 viruses circulating in Asia seem to be capable of naturally infecting pigs. However, the incidence of such infections has been apparently low. None of the avian and human H5N1 viruses tested were readily transmitted between pigs under experimental conditions (Choi 2005). Based on these observations, pigs probably do not currently play an important role in the epidemiology of the Asian lineage H5N1.
An outbreak of the highly pathogenic H7N7 avian influenza in poultry, in the Netherlands, Belgium and Germany in Spring 2003, caused infection and mild illness, predominantly conjunctivitis, in 89 poultry workers exposed to infected animals and carcasses (Koopmans 2004). The infection of one veterinarian caused an acute respiratory distress syndrome and took a fatal course (Fouchier 2004). In addition, during the Dutch outbreak, H7N7 infection was virologically and serologically confirmed in several household contacts, four of which showed conjunctivitis (Du Ry van Beest Holle 2005). Evidence for (asymptomatic) natural infection with LPAIV strains of H9, H7 and H5 subtypes in humans has also been reported on other occasions in Italy and Japan (Zhou 1996, Puzelli 2005, Promed 20060110.0090).
In an anecdotal report (Promed Mail 20050826), a fatal infection due to H5N1 influenza in three rare civet cats born in captivity at a national park in Vietnam was mentioned. The source of the infection remained obscure. Another 20 civets of the same species, housed in adjacent cages, did not become sick.
Avian influenza viruses have never been detected in rats, rabbits and various other mammals present at live bird markets in Hong Kong where 20 % of the chickens were found positive for the Asian lineage H5N1 (Shortridge 1998).
Up to the end of 2003, HPAI was considered a rare disease in poultry. Since 1959, only 24 primary outbreaks had been reported world-wide (Table 1). The majority occurred in Europe and the Americas. Most outbreaks were geographically limited, with only five resulting in significant spread to numerous farms, and only one which spread internationally. None of the outbreaks had ever approached the size of the Asian outbreaks of H5N1 in 2004 (WHO 2004/03/02). To date, all outbreaks of the highly pathogenic form have been caused by influenza A viruses of the subtypes H5 and H7.
In the past outbreaks, illegal trade or movements of infected live birds or their unprocessed products, and unintended mechanical passing-on of virus through human movements (travellers, refugees, etc.) have been the main factors in the spread of HPAIV.
A new dimension of HPAI outbreaks became evident late in 2003. From mid-December 2003 through to early February 2004, outbreaks in poultry caused by the Asian lineage HPAI H5N1 virus were reported in the Republic of Korea, Vietnam, Japan, Thailand, Cambodia, Lao People's Democratic Republic, Indonesia, and China. The simultaneous occurrence in several countries of large epidemics of highly pathogenic H5N1 influenza in domestic poultry is unprecedented. All efforts aimed at the containment of the disease have failed so far. Despite the culling and the pre-emptive destruction of some 150 million birds, H5N1 is now considered endemic in many parts of Indonesia and Vietnam and in some parts of Cambodia, China, Thailand, and possibly also the Lao People's Democratic Republic.
The original virus, encountered for the first time in 1997, was of a reassortant parentage, including at least a H5N1 virus from domestic geese (A/goose/Guangdong/1/96, donating the HA) and a H6N1 virus, probably from teals (A/teal/Hong Kong/W312/97, donating the NA and the segments for the internal proteins), which underwent many more cycles of reassortation with other unknown avian influenza viruses (Xu 1999,Hoffmann 2000, Guan 2002b). Several different genotypes of the H5N1 lineage have been described (Cauthen 2000, Guan 2002a+2003). The so-called genotype 'Z' has dominated the outbreaks since December 2003 (Li 2004).
In April 2005, yet another level of the epizootic was reached, when, for the ﬁrst time, the H5N1 strain obtained access to wild bird populations on a larger scale (Chen 2005, Liu 2005). At Lake Qinghai in North Western China several thousands of bar-headed geese, a migratory species, succumbed to the infection. Several species of gulls as well as cormorants were affected as well at this location. When, in the summer and early autumn of 2005, H5N1 outbreaks were reported for the ﬁrst time from geographically adjacent Mongolia, Kazakhstan, and Southern Siberia, migratory birds were suspected of spreading the virus. Further outbreaks along and between overlapping migratory flyways from inner Asia towards the Middle East and Africa hit Turkey, Romania, Croatia, and the Crimean peninsula in late 2005. In all instances (except those in Mongolia and Croatia) both poultry and wild aquatic birds were found to be affected. Often the index cases in poultry appeared to be in close proximity to lakes and marshes inhabited by wild aquatic birds. While this seems to suggest a direct hint towards migratory aquatic birds spreading the virus, it should be clearly noted that Asian lineage HPAI H5N1 virus has so far only been detected in moribund or dead wild aquatic birds. The true status of H5N1 in the populations of wild water birds and their role in the spread of the infection remains enigmatic. Presently, it can only be speculated as to whether wild aquatic birds can carry the virus over long distances during the incubation period, or whether some species indeed remain mobile despite an H5N1 infection.
Meanwhile, however, studies in China have revealed the presence of more new genotypes of the Asian lineage H5N1 virus in tree sparrows (Kou 2005). Neither the sparrows from which the viruses were isolated, nor the ducks that were experimentally infected with these viruses, showed any symptoms. However, upon transmission to chickens, full-blown HPAI was provoked. Since different sparrows of the same ﬂock carried several distinguishable genotypes, which likely arose by reassortment with different AI viruses of unknown provenance, it was suspected that H5N1-like viruses had already been transmitted to these birds some time (months?) ago. These data mark another step of aggravation: sparrows, because of their living habits, are ideal mediators between wild birds and domestic poultry and may shuttle HPAI viruses between these populations. Locally restricted infection with HP H5N1 in individual (diseased or dead) sparrows has also been reported from Thailand and Kong Kong. Endemicity of HPAIV in passerine birds such as sparrows, starlings or swallows which live in close connection to human settlements would not only impose a huge pressure on local poultry industries but also increase the exposure risks for humans (Nestorowicz 1987).
Up until the 30th December 2005, 142 H5N1 cases in humans had been reported. The human epidemic is currently limited to Cambodia, Indonesia, Thailand, and the epicentre Vietnam (65.5 % of all cases). 72 (50.7 %)persons have died.
For more detailed information, see the chapter entitled "Epidemiology".
Outbreaks of highly pathogenic avianinfluenza can be catastrophic for single farmers and for the poultry industry of an affected region as a whole (see Table 1). Economical losses are usually only partly due to direct deaths of poultry from HPAI infection. Measures put up to prevent further spread of the disease levy a heavy toll. Nutritional consequences can be equally devastating in developing countries where poultry is an important source of animal protein. Once outbreaks have become widespread, control is difficult to achieve and may take several years (WHO 2004/01/22).
Control Measures against HPAI
Due to its potentially devastating economic impact, HPAI is subject world-wide to vigilant supervision and strict legislation (Pearson 2003,OIE Terrestrial Animal Health Code 2005). Measures to be taken against HPAI depend on the epidemiological situation of the region affected. In the European Union (EU) where HPAIV is not endemic, prophylactic vaccination against avian influenza is generally forbidden. Thus, outbreaks of HPAI in poultry are expected to be conspicuous due to the clinically devastating course of the disease. Consequently, when facing such an outbreak, aggressive control measures, e.g. stamping out affected and contact holdings, are put in place, aiming at the immediate eradication of HPAI viruses and containing the outbreak at the index holding.
For these purposes, control and surveillance zones are erected around the index case with diameters varying from nation to nation (3 and 10 kilometres, respectively, in the EU). The quarantining of infected and contact farms, rapid culling of all infected or exposed birds, and proper disposal of carcasses, are standard control measures to prevent lateral spread to other farms (OIE - Terrestrial Animal Health Code). It is pivotal that movements of live poultry and also, possibly, poultry products, both within and between countries, are restricted during outbreaks.
In addition, control of H5 and H7 subtypes of LPAI in poultry, by testing and culling of acutely infected holdings, may be advisable in non-endemic areas in order to reduce the risk of a de novo development of HPAIV from such holdings.
Speciﬁc problems of this eradication concept may arise in areas (i) with a high density of poultry populations (Marangon 2004,Stegemann 2004, Mannelli 2005) and (ii) where small backyard holdings of free roaming poultry prevail (Witt and Malone 2005). Due to the close proximity of poultry holdings and intertwining structures of the industry, spread of the disease is faster than the eradication measures. Therefore, during the Italian outbreak of 1999/2000 not only infected or contact holdings were destroyed, but also ﬂocks with a risk of infection within a radius of one kilometre from the infected farm were pre-emptively killed. Nevertheless, eradication required four months and demanded the death of 13 millions birds (Capua 2003). The creation of buffer zones of one to several kilometres around infected farms completely devoid of any poultry was also behind the successful eradication of HPAIV in the Netherlands in 2003 and in Canada in 2004. So, not only the disease itself, but also the pre-emptive culling of animals led to losses of 30 and 19 million birds, respectively. In 1997, the Hong Kong authorities culled the entire poultry population within three days (on the 29th, 30th, and 31st December; 1.5 million birds). The application of such measures, aimed at the immediate eradication of HPAIV at the cost of culling also non-infected animals, may be feasible on commercial farms and in urban settings. However, this will afflict the poultry industry significantly and also prompts ethical concern from the public against the culling of millions of healthy and uninfected animals in the buffer zones.
Such measures are most difficult to implement in rural areas with traditional forms of poultry holdings where chickens and ducks roam freely and mingle with wild birds or share water sources with them. Moreover, domestic ducks attract wild ducks and provide a significant link in the chain of transmission between wild birds and domestic ﬂocks (WHO 2005). These circumstances may provide the grounds for HPAI viruses to gain an endemic status.
Endemicity of HPAI in a certain region imposes a constant pressure on poultry holdings. As the above mentioned restrictions can not be upheld over prolonged periods without vital damage to a country's poultry industry or, in the developing world, leading to a serious shortage of protein supply for the population, other measures must be considered.
Vaccination has been widely used in these circumstances and may also be a supplementary tool in the eradication process of outbreaks in non-endemic areas.
Vaccination in the veterinary world pursues four goals: (i) protection from clinical disease, (ii)protection from infection with virulent virus, (iii) protection from virus excretion, and (iv) serological differentiation of infected from vaccinated animals (so-called DIVA principle).
In the field of influenza vaccination, neither commercially available nor experimentally tested vaccines have been shown so far to fulfil all of these requirements (Lee and Suarez 2005). The ﬁrst aim, which is the protection from clinical disease induced by HPAIV, is achieved by most vaccines. The risk of infection of vaccinees with, and excretion of, virulent field virus is usually reduced but not fully prevented. This may cause a significant epidemiological problem in endemic areas where exhaustive vaccination is carried out: vaccinated birds which appear healthy may well be infected and excrete the field virus 'under cover' of the vaccine. The effectiveness of reduction of virus excretion is important for the main goal of control measures, that is, the eradication of virulent field virus. The effectiveness can be quantified by the replication factor r0. Assuming a vaccinated and infected ﬂock passes on the infection on average to less than one other ﬂock (r0 < 1), the virulent virus is, on mathematical grounds, prone to be extinguished (van der Goot 2005). When dealing with vaccination against the potentially zoonotic H5N1 virus, reduction of virus excretion also reduces the risks of transmission to humans, since a significant dose of virus seems to be required to penetrate the species barrier between birds and humans. Last but not least, a DIVA technique allows the tracing of field virus infections by serological means in vaccinated birds too.
For practical use several requirements must be observed (Lee and Suarez 2005):
A bunch of different vaccine concepts has been developed. Most are still based on inactivated, adjuvanted whole virus vaccines which need to be applied by needle and syringe to each animal separately.
Inactivated homologous vaccines, based on the actual HPAI strain, induce proper protection but do not allow a distinction of vaccinees and infected birds serologically. Since the vaccine is made from the current HPAI virus, there is an inherent delay before such vaccines can be used in the field.
Inactivated heterologous vaccines, in contrast, can be used as marker vaccines when the vaccine virus expresses the same HA- but a different NA-subtype compared to the field virus (e.g. H5N9 vaccine vs, H5N2 HPAI). By detection of NA subtype-speciﬁc antibodies, vaccinees and infected birds can be distinguished (Cattoli 2003). However, these methods can be laborious and may lack sensitivity. Nevertheless, such vaccines can be kept in vaccine banks comprising several H5- and H7-subtypes with discordant NA subtypes. Reverse genetics will greatly aid in producing vaccines both for veterinary and medical use with the desired HxNy combinations in a favourable genetic background (Liu 2003, Neumann 2003, Subbarao 2003, Lee 2004, Chen 2005, Stech 2005). Currently, inactivated heterologous vaccines are in field use in the H5N1 hot spots of South East Asia as well as in Mexico, Pakistan and Northern Italy (e.g. Garcia 1998, Swayne 2001). As an alternative DIVA system for use with inactivated vaccines, the detection of NS-1 speciﬁc antibodies has been proposed (Tumpey 2005). These antibodies are generated at high titres by naturally infected birds, but at considerably lower titres when inactivated vaccines are used.
Recombinant live vector-engineered vaccines express a H5 or H7 HA gene in the backbone of viruses or bacteria capable of infecting poultry species (e.g. fowl pox virus [Beard 1991, Swayne 1997+2000c], laryngotracheitis virus [Lueschow 2001, Veits 2003] or Newcastle Disease virus [Swayne 2003] among others). Being live vaccines, mass application via water or sprays is often feasible. While allowing for a clear-cut DIVA distinction, a pre-existing immunity towards the vector virus, however, will grossly interfere with vaccination success. Some field experience with fowl pox recombinants has been collected in Mexico and the U.S.
Finally, successful use of recombinantly expressed HA proteins and of DNA vaccination using HA-expressing plasmids has been experimentally proven (Crawford 1999,Kodihalli 1997).
Vaccination is now planned to be used on a nation wide scale in several countries in South East Asia (Normile 2005).
Three conditions need to be met for a new pandemic to start:
This shows that a threat of a new human influenza pandemic is not uniquely linked to the emergence of HPAI H5N1. So far, H5N1 only meets two of these conditions: it is, for the vast majority of the human population, a new subtype and it has infected and caused severe illness and high lethality in more than 140 humans to date. There is no immunity against a H5N1-like virus in the vast majority of the human population. A new pandemic would be at the brink should the Asian lineage H5N1 acquire properties, by stepwise adaptation or by reassortment with an already human-adapted virus, for an efﬁcient and sustained human-to-human transmission (Guan 2004). In vitro, it has been shown that two simultaneous amino acid exchanges in the receptor binding site of the HA protein of the Asian lineage HPAIV H5N1 (Q226L and G228S) optimises binding to human receptors of the 2-6 type like that of other human adapted influenza A viruses (Harvey 2004). Gambaryan et al. (2006) have already identified two human isolates originating from a father and his son infected with H5N1 in Hong Kong in 2003, which, in contrast to all other H5N1 isolates from humans and birds, showed a higher affinity for 2-6 receptors due to a unique S227N mutation at the HA1 receptor binding site.
This instance might be just around the corner or might already have occurred while reading this article - no one knows or can foretell. The chances for such an event to occur are directly correlated to the amount of virus circulating in poultry and, thus, the exposure risks of humans. Therefore, fighting H5N1 at its source would also reduce pandemic risks posed by this virus. Heretically, it has been proposed in one of the internet mail- and discussion-forums that the investment of only ten percent of the money that is scheduled to be spent for the development of H5-speciﬁc human vaccines in the eradication of H5N1 in poultry would have a greater effect than human vaccination in the protection of the human population from a H5N1 epidemic.
Since its ﬁrst isolation in humans in 1997, H5N1 has failed to perform this last step towards pandemicity in human hosts. Recent studies, however, suggest that over the years, the virulence of H5N1 for mammals has increased and the host range has expanded:
However, it should not be overlooked that while staring at the H5N1 situation in Asia, other influenza viruses with possibly even greater pandemic potential may emerge or may already have emerged in the meantime. For example, strains of the H9N2 subtype which was not found in Asia prior to the 1980s have not only become widespread in Asian poultry populations, but also have crossed efficiently into pig populations in South Eastern and Eastern China (Shortridge 1992,Peiris 2001, Xu 2004). The receptor of these viruses revealed specificities similar to human-adapted viruses (Li 2005b, Matrosovich 2001). These H9 viruses have a broad host range, are genetically diverse and can directly infect man. The H9N2 strain, which was responsible for these human infections in Hong Kong, even revealed a genotype akin to that of the H5N1 viruses of 1997 (Lin 2000).
The importance of highly pathogenic avian influenza (AI) as a devastating disease of poultry has markedly increased during the last decade. The introduction of AI viruses of the subtypes H5 and H7 of low pathogenicity (LP) from a reservoir in wild water birds has been at the base of this process. It remains to be elucidated whether and, if so, why, the prevalence of LP H5 and H7 in their reservoirs has also been changing. With regard to the endemic status of the Asian lineage HPAI H5N1 in domestic poultry populations in South East Asia, causing frequent spill-overs into populations of migratory birds, a paradigm shift in the epidemiology of HPAI towards endemicity in migratory wild bird populations seems to be imminent. This would have grave consequences for the poultry industry on a transcontinental scale. Exposure risks for humans are directly linked to the increased presence of potentially zooanthroponotic viruses in domestic poultry.
With respect to the avian and veterinary side of the story, many questions still remain unanswered:
In particular, the ﬁrst question is of overwhelming importance - not only for the veterinary world. Endemicity of the Asian lineage HPAIV H5N1 in migratory birds would pose a constant threat to poultry holdings. This would only be met by strict biosecurity measures including a prohibition of free-roaming poultry holdings. Alternatively, mass vaccination of poultry must be considered. As a second line, endemicity in wild birds may also lead to the presence of HPAI H5N1 virus in the environment (lakes, sea shores etc.) and might pose an additional potential risk of exposure for humans. So far, there are no reports of transmission from wild birds or environmental sources to humans. All reported human infections, including the most recent ones from Turkey, seemed to be acquired following virus amplification in, and close contact to, household poultry.
The complexity and the potential impact of the current, zooanthroponotic HPAI H5N1 virus semi-pandemic in birds, demands concerted and prudent actions from scientists, politicians, and the public.