Acute Respiratory Infections (Update September 2009)
- Meetings reports
- Other WHO Links
- Tables on the clinical trials of pandemic/potentially pandemic and seasonal influenza vaccines
The repetitive occurrence of yearly, seasonal influenza epidemics is due to the fact that influenza viruses are continuously changing antigenically. This was well illustrated by the emergence of the influenza A H3N2 'Fujian' strain, which appeared in July 2003 in the southern hemisphere then spread to the northern hemisphere a few months later to become a dominant strain  . To face this continuous change, virus strains to be included in the vaccine are updated annually so as to match the circulating virus strains. An alternative would be to develop "universal" influenza vaccines which would cover all possible influenzavirus strains (see below). Meanwhile, the recent emergence of highly pathogenic avian influenza H5N1 in poultry farms and markets in several countries in Asia, from where it spread to Africa and Europe, and its transmission to humans with a 60% case fatality rate, has revived the fear of a possible new pandemic of influenza with high mortality and planetary consequences    .
The burden of influenza is currently estimated to be 25-50 million cases per year (~ 20% of the population) in the USA alone, leading to 150 000-200 000 hospitalizations and 30,000-40,000 deaths  . If one extrapolates these figures to the rest of the world, the average global burden of seasonal influenza comes to be on the order of 600 million cases, 3 million cases of severe illness and 250 000-500 000 deaths per year. Hospitalization rates from severe illness can be as high as 3 per 1000 for 6 to 23 months old children and as high as 9 per 1000 for children younger than 6 months .
Influenza is highly contagious  and is readily transmitted via aerosols and droplets from the respiratory tract of infected persons by direct contact, through coughing or sneezing, or by hands contaminated with respiratory secretions. Adults are most infectious from 1 day before symptom onset up to 7 days afterwards. When influenza is introduced into a household, 20-60% of exposed persons will eventually show virologic or serologic evidence of infection. The disease can affect all age groups, but rates of infections are highest among young children who shed virus and are a potential source of infection in older age cohorts  , whereas rates of serious illness, complications and death are highest in persons aged 65 years and older, as well as in persons with chronic cardiac or respiratory conditions. Data collected in Michigan (USA) in Japan and in Russia indicate that mass vaccination of school-aged children correlates with a reduced rate of respiratory illness in all age groups, suggesting that larger-scale immunization in childhood could favourably affect influenza epidemics .
Epidemics and outbreaks of influenza occur in different seasonal patterns depending on the region in the world: in temperate climate zones, seasonal epidemics typically begin in the late Fall and peak in mid-winter, infecting about 5-15% of the population each season. In tropical zones, seasonal patterns are less pronounced and the virus can be isolated year-round  . The disease, characteristically a febrile illness with respiratory symptoms, ranges in severity from mild to debilitating and can lead to lethal primary fulminant pneumonia, particularly in persons with underlying pulmonary or cardiopulmonary pathologies.
The repetitive occurrence of yearly influenza epidemics is maintained through the ongoing process of "antigenic drift", which results from the accumulation of point mutations in the genes that encode the two viral surface proteins haemagglutinin (HA) and neuraminidase (NA), and leads to the constant emergence of new virus variants against which there is little or no pre-existing immunity in the population. For this reason, major seasonal epidemics of influenza continue to occur each year and the virus strains to be included in the vaccine of the year must be chosen to match the emerging new variants   Recent molecular studies suggest that new virus strains emerge in East and South-East Asian countries, from which they spread around the world, first to Europe and North America which are reached within 6-9 months then to South America .
At unpredictable intervals, due to the segmented nature of the influenza virus genome, circulating human influenzavirus A strains also can acquire new genes from an avian or other animal influenza virus. This process is believed to occur most readily in pigs, as pigs have the complete set of syalilated receptors for avian, swine and human influenza virus strains. Co-infection in pigs can thus result in the emergence of a virus with a completely new glycoprotein subtype, which is referred to as an "antigenic shift". If the reassortant virus can efficiently spread into the human population, a worldwide pandemic can occur, as was the case in 1918, 1957, and 1968  .
The impact of a new influenza pandemic has grossly been estimated at 1-2 billion cases of flu, 5-12 million cases of severe illness, and 1.5-3.5 million deaths worldwide  . It could result in 1 million to 2.3 million hospitalizations and 250 000 to 650 000 deaths in industrialized nations alone. In the USA, the impact of a new pandemic, assuming it would be of a similar magnitude as the 1957 or the 1968 pandemics, and not like the 1918 pandemic, is projected to be 18-42 million outpatient visits, 314 000-734 000 hospitalizations and 89 000-207 000 deaths. Its impact would be even more devastating in developing countries  . Historians estimate that more than 50 million people died in the 1918-1920 influenza pandemic  .
Of note is the observation that in three studies of blood cultures taken from living soldiers with pandemic influenza-associated pneumonia in 1918, pneumococci were isolated from 46% of patients. The data from influenza hospitalization in children and from pandemic influenza mortality in 1918 suggest a significant role for pneumococci in influenza-associated pneumonia  . This is most likely due to inhibition of macrophage-mediated antibacterial defense during the recovery stage from influenza infection, which is characterized by pulmonary infiltration of T cells that secrete high amounts of IFN-?  . This suggests that the prevention of pneumococcal super-infection should be an essential part of pandemic influenza preparedness  .
The emergence of the avian H5N1 influenzavirus A strain with pandemic potential occurred in 1997 in Hong Kong, SAR, resulting in the death of 6 of the 18 affected patients, mainly young adults  . The virus was fortunately not able to spread from person-to-person and the outbreak was controlled through massive culling of poultry. The H5N1 strain however reemerged in 2003-2004 in China, Japan, South Korea then Thailand, Vietnam, Indonesia, Cambodia and Malaysia, leading to massive culling and attempts at vaccination of poultry. More than 60% of human patients diagnosed with the virus died. In May, 2005, a highly pathogenic (HP) H5N1 variant emerged in wild birds in Quinghai Lake, China    , that not only killed domestic poultry but also wild aquatic birds. The HP strain also is pathogenic for ferrets, cats and tigers. Cats can be infected both by the respiratory route and by feeding on virus-infected birds. The HP H5N1 strain spread to a great many countries in Asia, Africa and Europe where it was repeatedly recovered from migratory birds and was the cause of multiple outbreaks in poultry  . A significant part of transmission has been linked to commercial trade of poultry and derived products  .
Cases of human H5N1 infections have been reported in many countries but no human-to-human transmission has been evidenced so far, except in a couple of instances among close contacts. As of October, 2008, a total of 387 confirmed human cases from 15 countries resulting in 245 deaths had been reported to WHO, of which more than 100 cases and 50 deaths had occurred in Vietnam. The virus seems to now have become endemic in several countries, including Nigeria, Egypt, Bangladesh, Vietnam and Indonesia, while showing continuous sequence evolution leading to the emergence of different molecular clades and sub-clades  . Significant concerns therefore remain that the virus could adapt to infect humans or exchange genetic material with co-circulating human influenza viruses, resulting in the emergence of a highly pathogenic pandemic strain  .
Other avian influenza viruses have occasionally caused a human outbreak, such as a H9N2 strain in 1999 in Hong Kong, a H7N7 strain in 2003 in the Netherlands, which caused 89 confirmed human cases with conjunctivitis and one death, and H7N2 and H7N3 in 2003-2004 in North America. No one can know how severe the next pandemic will be, nor which influenza virus will cause it -it could be an H2, H5, H7, H9 or another subtype. A report from the Centers for Disease Control and Prevention (CDC) indicates that North American H7N2 and H7N3 strains have now partially adapted to the human sialic acid receptor (see below), a characteristics that would enhance the virus potential to infect and spread among humans  . At the same time, the H5N1 virus continues to be a zoonotic virus, not a human-adapted one, and human infections remain rare. Still, given the alarming 60% average case fatality rate of H5N1 influenza cases in humans so far, it probably is prudent to prepare for the 'worst case' scenario  .
Influenza viruses are enveloped viruses with a segmented genome made of eight single-stranded negative RNA segments, most of which encode only one viral protein (for a review, see  ). They form the family Orthomyxoviridae, which includes three genera, Influenzavirus A, Influenzavirus B, and Influenzavirus C, that differ antigenically by two of the structural proteins, the matrix (M) protein and the nucleoprotein (NP). Influenza A viruses are further divided into subtypes according to the antigenicity of their major envelope glycoproteins, the haemagglutinin (HA) and neuraminidase (NA). Sixteen HA subtypes (H1 to H16) and nine NA subtypes (N1 to N9) have been identified. The nomenclature of human influenza virus strains includes the type of the isolate, the place where it was isolated, the year of isolation, an identification number and for influenza A viruses, the subtype of both HA and NA, e.g. "A/Panama/2007/99 (H3N2)"  .
Influenza A viruses of the H1N1, H1N2, H3N2 and H2N2 subtypes naturally infect humans, but only the first three types are currently circulating in the human population. In contrast, influenza A viruses of all 16 HA subtypes and all 9 NA subtypes have been recovered from aquatic birds, which serve as a natural virus reservoir and a potential source of new genes for pandemic influenza viruses. Study of the epidemiological dynamics of human influenza A virus shows that genomic segments encoding the non-structural proteins (NS1 and NS2), the matrix proteins (M1 and M2), and two of the polymerase proteins, PB2 and PA, have remained unchanged since the global pandemic of 1918   . In contrast, new RNA segments encoding the haemagglutinin (HA) and neuraminidase (NA) glycoproteins, as well as the PB1 polymerase, have been acquired through reassortment with avian influenza virus, coinciding with the global pandemics in 1957 (H2N2) and 1968 (H3). Seasonal epidemics of influenza A virus since 1968 have been dominated by H3N2 viruses and characterized by punctual mutations, natural selection and frequent reassortment that underlie antigenic drift    .
Influenza A viruses also can infect poultry, pigs, horses, dogs, and sea mammals. Interspecies transmission has been well documented  . Aquatic birds, in which the virus multiplies in the gut, usually have an asymptomatic infection and excrete the virus in their faeces. They can transport the virus over large geographical distances. Influenza B viruses appear to naturally infect only humans, although infection of seals was documented in the Netherlands  . Influenza C viruses only appear to infect humans and pigs and usually cause sporadic cases of upper respiratory disease.
The HA glycoprotein molecules are present at the surface of the influenza virion in the form of a trimeric HA0 precursor which must undergo proteolytic cleavage to generate functional subunits HA1 and HA2. HA1 bears the receptor-binding site and neutralization epitopes, whereas HA2 is responsible for the fusion of the viral envelope with the host-cell membrane at acidic pH     . Classical avian virus strains have a HA0 cleavage site which is trypsin-like, hence their tropism for the gastrointestinal tract. In contrast, highly virulent avian strains such as the 2004 H5N1 strains from Thailand and Viet Nam or the HP H5N1 isolates have acquired through spontaneous mutations an ubiquitous furin-like cleavage site, which allows them to multiply in many tissues including the respiratory tract  . A recent post-mortem study of two fatal H5N1 human cases showed that in addition to the lungs, the virus had disseminated to other organs including the intestine, T lymphocytes in lymph nodes and circulating monocytes and macrophages, as well as cerebral neurons  . The virus could also be transmitted from mother to foetus across the placenta. Viremia and extra-respiratory complications appear to be more common with the avian H5N1 virus than with usual human influenzaviruses  .
Influenza virus entry into cells of the respiratory tract is mediated by binding of the trimeric haemagglutinin spikes on the virion to specific sialic acid receptors at the cell surface. The difference between the avian (H5) and human (H1, H3) hemagglutinins lies in their specificity for ?-2,3-Gal or ?-2,6-Gal sialic acid linkages, respectively  . In mammals, cells with ?-2,3 receptors only occur deep in the lungs, whereas ?-2,6 receptors are found in the upper respiratory tract, explaining why the H5N1 influenza subtype may cause infection of the lower respiratory tract and severe pneumonia in humans but does not spread readily from human-to-human  . Two amino acid mutations that caused a switch in receptor binding preference from the human ?-2,6 to the avian ?-2,3 sialic acid resulted in a 1918 H1N1 virus incapable of respiratory droplet transmission between ferrets, suggesting that a predominant ?-2,6 sialic acid-binding preference is essential for optimal transmission of influenza viruses among mammals  . Three mutations in the H5 haemagglutinin were both necessary and sufficient to change its receptor binding specificity from the avian ?-2,3 to an human ?-2,6 pattern  .
The pathogenicity of influenza A virus is clearly due to a polygenic effect   . Studies of virulence of avian influenza virus strains shows that in addition to the HA cleavage site (see above), changes in NS1 are important determinants   , as well as mutations in the PA and PB1 genes  . Similarly, the HA, NS, NP, PA, PB1 and PB2 proteins contribute to the virulence of influenza A virus strains in mice, pigs and monkeys (reviewed in  ). The NS1 protein is critical for the pathogenicity of H5N1 influenza viruses in mammalian hosts as it antagonizes host cell interferon induction and the double-stranded RNA-mediated activation of the NF-kappaB and IFN pathways   .
The currently available influenza vaccines are made from either inactivated, detergent-split or whole virion influenza virus or live attenuated influenza virus (LAIV) propagated in the allantoic cavity of 9-12 days embryonated chicken eggs from certified farms under strict veterinary control  . They include two currently circulating influenza A virus strains and one influenza B virus strain. These trivalent vaccines, which have been used for decades in industrialized countries to prevent seasonal influenza infection, provide a high benefit/cost ratio in terms of preventing hospitalizations and deaths, as shown in numerous studies on vaccination of the elderly and of individuals at high risk for severe outcomes of influenza  . Recommended recipients of influenza vaccines are people at high risk of influenza-related complications, namely adults and children with chronic health conditions such as cardiac or pulmonary disorders, asthma, cancer, immunodeficiency or immunosuppression, people who are residents of nursing homes and other chronic care facilities, people over 65 years of age, healthy children aged 6-23 months   and pregnant women. Household contacts of individuals at high-risk, as well as providers of health care in facility or community settings who are potentially capable of transmitting the virus to vulnerable populations should also be vaccinated, as well as those providing regular child care to children under 24 months of age.
WHO estimates that there globally are about 1.2 billion people at high risk for severe influenza outcomes: 385 million elderly over 65 years of age, 140 million infants, and 700 million children and adults with an underlying chronic health problem. In addition, 24 million health-care workers ought to be immunized to prevent spreading of the disease to the high-risk population.
Seasonal inactivated vaccines.
Continual antigenic drift of the virus means that a new vaccine updated to contain the most current circulating strains is needed every year to protect against new seasonal infections. This is done by reassortment or through the use of the techniques of reverse genetics, which allow one to transfer the HA and NA genomic segments from a circulating wild-type virus into a well characterized, high-yield egg-adapted or cell-culture-adapted influenza virus strain such as the PR8 virus strain, then to select the reassortant virus with the desired HA and NA gene combination. The reassortant virus is usually grown in the allantoic cavity of embryonated hen's eggs, although some vaccines are now produced using mammalian cell lines such as MDCK, PERC-6 or Vero cells.
Monovalent vaccine lots are inactivated using either formalin or ?-propiolactone, then purified and combined to make the final trivalent vaccine   . Most seasonal vaccines produced today are 'split-vaccines' (subvirion vaccines), which result from treatment of the virions with a detergent or a solvent to dissolve the viral lipid envelope. Less frequently used are whole-virion vaccines made of intact inactivated virus, which may be more reactogenic but have consistantly proved to be better immunogens and confer more efficient protective responses than split vaccines .
Other inactivation procedures have been reported, such as treatment with hydrophobic 1,5-iodonaphtyl-azide (INA), which selectively targets the transmembrane segments of proteins in the viral envelope while preserving the structure of the antigens on the surface of the virus. INA is activated by UV irradiation, resulting in a non-infectious intact influenza virus particle with full neuraminidase activity able to induce potent anti-influenza virus serum neutralizing antibody and T cell responses and to provide protection against heterosubtypic challenge after a single Intranasal immunization, similar to live attenuated virus immunization .
The trivalent split inactivated influenza vaccines have a remarkable safety profile, including in 6 to 23 months old children  , as recently confirmed in a retrospective study bearing on 45,356 vaccinations in children that age  . Multiple studies show the vaccine to be approximately 60% to 90% efficacious against influenza illness in healthy children and adults, depending on the antigenic match between the circulating and vaccine viral strains    . Vaccine effectiveness was found to actually vary substantially across successive seasons from a low 10% to a high 52% depending on the antigenic match between the circulating virus strains and the strains used in the vaccine  . Vaccination was also found to decrease the incidence of pneumonia, physician visits, hospitalizations and deaths in the elderly    . In vaccine-naive children less than 9 years old, a two dose schedule is required   . In contrast, a second dose of vaccine in elderly individuals does not boost immunity .
The inactivated split influenza vaccine also is quite immunogenic when administered as a full dose (15µg HA) to elderly volunteers by the ID route using a Beckton-Dickinson microneedle delivery system   . The use of the ID route was found to enhance the immunogenicity of the vaccine in the elderly population as compared to the classical IM route of administration, as judged by higher geometric mean antibody titers and higher percentage of vaccinees with a > 4-fold rise in HA antibody titer  . Similar data were generated in recipients of renal transplants. A randomized, open-label study in 112 healthy children aged 3 to 18 years showed that the immunogenicity of ID vaccination at one fifth of a dose was comparable to that of standard-dose IM vaccination in children as young as 3 years of age  . Attempts at vaccinating healthy adults by the intranasal route using the inactivated split vaccine in combination with detoxified E coli enterotoxin (LTK63) as an adjuvant also are in progress  .
At this time, the vaccination of people at high risk of complications from influenza is a key public health strategy in industrialized countries. Almost 350 million trivalent vaccines are distributed worldwide each year. Universal influenza vaccination has been proposed as one strategy to improve vaccination coverage and disease prevention. Most economic analyses show that universal vaccination of healthy adults and children would be cost-effective. Also, herd immunity resulting from vaccination of school-age children would substantially reduce the incidence of disease and mortality among adults   . The world's total vaccine production capacity at this time, however, is only about 350 million doses of trivalent vaccine, of which close to 50% are produced by just one manufacturer, Sanofi Pasteur.
Live attenuated influenza vaccines (LAIVs)
The second major approach to influenza vaccines has been the development of cold-adapted (ca) virus strains which grow well in primary chicken kidney cells and embryonated eggs at 25-33°C, have a reduced replication titre at 37°C, and show attenuated virulence in ferrets. Cold adaptation was found to be a reliable and efficient procedure for the derivation of LAIVs, which are obtained by reassorting the HA and NA genomic segments from a circulating influenza A virus strain with the other six genomic segments from a well characterized laboratory strain that carries the ca mutation. Thus, a bivalent ca LAIV was developed by Microgen and used for decades in Russia to immunize more than 100 million people every year.
A trivalent ca LAIV (FlumistTM) has been developed for intra-nasal spray delivery by MedImmune and Wyeth in the USA where it officially is licensed for 2-59 years-old persons. The FlumistTM vaccine was proven highly efficacious in Phase III trials, showing a 92% overall protection rate over a 2-year study in children   . It was well tolerated and effective in preventing culture-confirmed influenza illness in children as young as 6 months of age who attended day care centers  and was shown to provide herd immunity to adults when used in children  . It also demonstrated superior efficacy compared to the trivalent inactivated vaccine in preventing influenza illness in children 12 to 59 months of age  , in young children with recurrent respiratory tract infections  and in children with asthma   . One dose of FlumistTM induced significant protection against influenza illness and pneumonia in 5 to 18 years old children when administered during an influenza outbreak due to a circulating influenza virus strain distinct from the vaccine strain, whereas the trivalent inactivated vaccine was not able to provide protection  . Nasal shedding of LAIV in individuals 5-49 years of age was found to generally be of short duration (days 1-10 post immunization) and at low titers. LAIV recipients should therefore avoid contact with severely immunosuppressed persons .
The FlumistTM vaccine shows remarkable genetic stability, but it has to be kept at -18°C. A new, heat-stable version of the vaccine has been developed, which showed good efficacy in clinical trials in Asia and Europe, but was associated with an increase in asthma episodes in young children  . The current vaccine was also tested for safety in infants aged 6 to 24 weeks, who received two doses of vaccine intranasally 1 month apart: the vaccine was generally well tolerated  . In children, LAIV provided sustained protection against influenza illness caused by antigenically related strains for 9-12 months, and showed meaningful efficacy although at a lesser level through a second season without revaccination .
Another ca LAIV also is in development at Applied Microbiology in Austria by growth of the virus in Vero cells at 25 °C  . Still another cold-adapted LAIV grown in Madin-Darby canine kidney (MDCK) cells on microcarrier beads in serum-free medium is at an advanced preclinical development stage at the Vector Scientific Center in the Russian Federation.
Preliminary studies indicate that the IFN?-ELISPOT assay, a measure of cell-mediated immunity, may be a more sensitive measure of influenza immune memory responses to LAIVs than serum antibody. The role of cell-mediated immunity in actual protection against culture-confirmed clinical influenza by LAIV was investigated in a large efficacy trial involving 2172 6 to 36 months-old children in the Philippines and Thailand who were vaccinated with an intranasal dose of 107 FFU of LAIV: the majority of infants with >100 IFN-? spot-forming cells/106 peripheral blood mononuclear cells were protected against influenza .
Other types of influenza vaccines
Berna Biotech, now Crucell, is commercializing an influenza vaccine formulated in virosomes, with the surface spikes of the three currently circulating influenza virus strains inserted into the vesicle membrane of liposomes. A nasal formulation of the vaccine had to be withdrawn from the market, due to undesirable neurological side effects (Bell's palsy) linked to the presence of the E. coli labile toxin (LT) used as an adjuvant, most likely because the B subunit of LT could bind to GM1 ganglioside receptors in neuronal tissues associated with the olfactory tract. Other formulations of inactivated influenza vaccine for mucosal delivery are in progress including immunostimulating complexes (ISCOMs).
Yeda, an Israeli R&D company, is developing a synthetic peptide influenza vaccine for nasal administration. The vaccine has shown protective efficacy in humanized mice and is planned to enter clinical trials soon.
A "universal" human influenza virus A vaccine was initially developed at the University of Ghent, Belgium, based on transmembrane viral protein M2, which is scarcely present on the virion but is abundantly expressed on virus-infected cells  . The extracellular M2 domain, M2e, a 23 amino acid-long peptide, is remarkably conserved between H1N1, H2N2 and H3N2 influenza A virus strains  . Passive administration of anti-M2e antibodies affords significant protection against influenza A virus challenge in animal models. The mechanism is believed to be NK cell-mediated antibody-dependent cellular cytotoxicity (ADCC), not virus neutralization.
A recombinant particulate vaccine has been engineered by genetically fusing copies of the influenza virus M2e domain to the hepatitis B core antigen (HBc). The (M2)-HBc fusion protein was found to spontaneously assemble into highly immunogenic virus-like particles (VLPs) that provided complete protection against a lethal influenza A virus challenge in mice and ferrets  . The VLP vaccine, ACAM-FLU-ATM, which is developed by Acambis, Cambridge, MA, USA, was recently tested in a two-dose immunization schedule clinical study in 18-40 years-old volunteers.
Another formulation involving a fusion protein based on the CTA1-DD adjuvant and containing tandem repeats of the M2e sequence, still is at a preclinical stage of development   . These vaccines could theoretically serve as universal influenza A vaccine in view of the high degree of conservation of the M2e sequence among human influenza A viruses. The question of whether they would also protect against a pandemic virus is however open, as the M2e sequence in avian virus strains including H5N1 appears to be divergent from that in humans viruses.
The M2e peptide has also been conjugated to other protein carriers, such as the Neisseria meningitidis outer membrane protein complex (OMPC) and derivatives of the cholera toxin. Another approach has been developed at VaxInnate, USA, linking four tandem copies of M2e to Salmonella typhimurium flagellin type 2 protein, a potent TLR5 ligand. The resulting STF2.4xM2e fusion protein, which can be produced in high yields in E coli fermentors, was shown to induce a robust and long-lasting protective antibody response in mice .
A subunit vaccine made of the conserved NP internal protein, a known target for cytotoxic T cells, which is covalently coupled to the M2e peptide and to the ISS adjuvant, is in development at Dynavax Technologies, Berkeley, CA, USA. ISS, a TLR9 activator, has been shown to activate NK cell secretion of IFN-?  and was successfully tested as an adjuvant in the HeplisavTM hepatitis B vaccine.
Influenza VLPs are produced in insect cell using recombinant baculovirus vectors that express viral proteins HA, NA and M1   . Influenza VLPs induced mucosal IgG and cellular immune responses in mice, as well as long-lived antibody-secreting cells that were detected in the bone marrow of the immunized animals .
A two-dose immunization regimen with influenza VLPs developed by Novavax, Rockville, MD, USA, was highly immunogenic in mice and ferrets, even at the low dose of 6µg HA  . In contrast, an experimental trivalent influenza vaccine made of HA0 haemagglutinins H1, B and H3 produced in serum-free insect cell cultures using three recombinant baculoviruses (Protein Science), required 75 µg of each of the three HAs to elicit HAI antibody responses in only 51%, 65% and 81% of vaccine recipients, respectively .
Live recombinant vaccines
Other influenza vaccines, such as DNA vaccines, live recombinant vaccines based on non-replicative Ad5  , modified vaccinia Ankara (MVA) virus  or Newcastle disease virus vectors  are still at a preclinical or early clinical development stage.
Pandemic Influenza vaccines
No current influenza vaccine would offer protection against a pandemic triggered by an emerging avian virus strain, such as H5N1, H7N7 or H7N2. Specific 'pandemic' vaccines need therefore to be prepared for that purpose. In addition, two doses of pandemic strain vaccine spaced by one month will likely be necessary as no humans have any underlying immunity to avian influenza virus strains  . Given the world's current vaccine production capacity, if a monovalent inactivated pandemic influenza vaccine were produced according to the formulation of seasonal vaccines, i.e. with 15 ?g haemagglutinin per dose, only about 475 million people could be vaccinated. The WHO has been active at encouraging pandemic influenza vaccine R&D and reviewing progresses through a number of meetings. Updated results on clinical trials of pandemic influenza vaccines can be found at .
Inactivated split vaccines
Immunogenicity trials with candidate split inactivated vaccines prepared with a H5N1 virus strain (A/Vietnam/2004 H5N1) showed disappointingly low immune potency, as two 45 ?g haemagglutinin doses of split vaccine at four weeks interval elicited only 38% seroconversion among 2-9 years old children and two doses of 90 ?g H5 haemagglutinin only 35% responses in elderly volunteers    . The need to enhance the immunogenicity of inactivated H5N1 vaccines and to search for 'antigen-sparing' formulations was therefore obvious   .
Attempts at using aluminium salts as adjuvants in inactivated H5N1 influenza vaccines had a limited effect on the potency of the vaccines, as two doses of 30µg or even 45 µg HA were required for induction of six-month antibody persistence  . In contrast, remarkable results were obtained with adjuvants such as polyoxidonium (Microgen), MF59 (Novartis), AS03 (GSK) or AF03 (Sanofi Pasteur), which are based on oil-in-water emulsions. These adjuvants considerably increased the immunogenicity of the inactivated split H5N1 vaccines, allowing a significant reduction in the dose of haemagglutinin needed to elicit protection.
Thus, two doses of MF59-adjuvanted Novartis vaccine (FluadTM) containing 7.5 ?g of H5 haemagglutinin elicited a protective influenza neutralizing antibody response in 77% of recipients  ; the antibody response to two doses of vaccine containing 15µg HA (H5N1) with MF59 were higher than the response to 45 µg of vaccine alone  . Quite similarly, two doses of AS03-adjuvanted GSK H5N1 vaccine containing 3.8 ?g and 7.5 ?g H5 haemagglutinin elicited haemagglutination-inhibiting (HAI) antibodies in 82% and 90% of the volunteers, as compared to 4% and 16% with the non adjuvanted vaccine controls, respectively   ; and two doses of AF03-adjuvanted H5N1 Sanofi Pasteur vaccine containing 1.9 ?g or 3.75 ?g of antigen generated a high level seroprotective immune response in over 70% and over 80% of the participants, respectively  . In addition, the antibodies elicited by the adjuvanted vaccines were broadly cross-neutralizing antibodies that neutralized virus strains belonging to the various H5N1 virus clades in cell culture and could protect ferrets from challenge with virus strains from different H5N1 clades .
There is therefore no more theoretical impossibility to manufacture 1.5-2 billion doses of split pandemic influenza vaccine using the currently available manufacture facilities if need be. Further improvements aimed at up-scaling vaccine production in case of a pandemic would be to develop cell-culture vaccines, which would bypass the bottleneck of limited egg supply. Influenza virus can be adapted to grow in a variety of cell lines including Vero cells and PER.C-6 cells, which would allow large-scale production of the virus.
The possibility is also entertained that the addition of adjuvants to seasonal influenza vaccines might be of benefit to populations at risk such as persons with underlying chronic diseases  . Inactivated whole-virion vaccines adjuvanted with aluminium hydroxide were found to have better immunogenicity in naive individuals than split vaccines, but may be associated with febrile reactions, particularly in childen  . New adjuvants based on TLR ligands are been developed that could eventually be used for influenza vaccines, such as the deoxynucleotide IC31TM (Intercell, Austria)  or the Salmonella typhimurium flagellin type 2 (STF2, VaxInnate Corp, USA).
Inactivated whole-virion vaccines
Candidate pandemic inactivated whole-virion vaccines also have been produced by several companies, including Baxter, Austria, Nobilon International, The Netherlands, Sinovac, China, and Omnivest, Hungary. Adjuvantation with aluminium hydroxide was found to be devoid of effect on the immunogenicity of the Vero cell-derived Baxter H5N1 vaccine in mice and humans  , whereas it provided high immunogenicity to the Nobilon vaccine in ferrets. The Omnivest H5N1 vaccine (FluvalTM), an egg-based whole-virion vaccine adjuvanted with aluminium phosphate, was found to elicit significant protective cross-clade immune responses in 70% of volunteers, including 60-90 years-old persons and children, after only a single IM dose of 6µg HA  . The Baxter vaccine  and the Sinovac vaccine  both necessitated two doses of vaccine with somewhat higher antigenic content 3-4 weeks apart to elicit protective neutralizing antibody levels in volunteers (For review, see ).
Candidate pandemic LAIVs have also been prepared. A single-dose immunization of ferrets with a candidate Flumist LAIV prepared from the A/Hong Kong/97 (H5N1) strain provided 100% protection against challenge of the animals with a variety of H5N1 strains belonging to different clades. A series of LAIVs were produced by MedImmune (USA) by reassortment between avian influenza A viruses H5N1, H7N3 or H9N2 and the ca Ann Arbor virus strain (H1N1). Phase I clinical trials of resulting ca reassortants were performed on volunteers kept in an isolation unit at Johns Hopkins Medical Center, Baltimore, MD, USA. Virus shedding into nasal secretions was detected for one day in 60% of volunteers and for up to four days in another 25%. After a second immunization a few weeks later, no viral shedding could be detected. A systematic comparison between avian-human and human-human ca reassortants in human volunteers however evidenced reduced infectivity and immunogenicity of the avian-human LAIVs, as compared with human-human H1N1 or H3N2 LAIVs. The H5 LAIV was the least immunogenic of the five ca reassortants tested (for a review, see ). A two-dose immunization with the H5N1/influenza A/Ann Arbor 60 ca (H3N2) reassortant LAIV fully protected mice and ferrets against pulmonary replication of homologous and heterologous wild-type H5N1 viruses .
Another avian-human ca reassortant influenza virus strain was developed as LAIV by Microgen, Russia  , that elicited strong cross-clade protection in mice. The resulting LAIV strain, which carries a H5N2 antigenicity, was tested in Phase I and Phase II clinical trials using two immunizations 21 days apart and showed reasonably good immunogenicity.
Still another type of H5 LAIV was developed by Green Hills Biotech, Vienna, Austria, by deleting the NS1 viral gene, a virulence factor known to antagonize interferon, from a H5N1 avian strain. The resulting ?NS1 H5N1 strain can only be grown in Vero cells, shows attenuation in mice and ferrets and provides protection of the animals against virulent challenge with wild-type H5N1 virus strains .
A H5N1 VLP containing the HA, NA and M1 viral proteins was developed at Novavax, USA, and found to elicit strong HAI and cell-mediated immune responses in mice and ferrets and to protect 100% of the animals against cross-clade H5N1 virus challenge. A two-dose Phase I trial is planned to take place shortly. A new recombinant VLP candidate vaccine was recently developed by LigoCyte Pharmaceuticals in collaboration with the Batelle Biomedical Resaearch Center based on the expression of a fusion protein between the murine leukemia virus Gag protein and the influenzavirus HA glycoprotein in a baculovirus-insect cell expression system. The resulting HA-pseudotyped Gag VLPs were purified as 100 nm spherical particles that elicited robust anti-influenza immunity in ferrets and protected the animals against cross-clade H5N1 virus challenge .
Live vectored vaccines
Live recombinant H5N1 vaccines also have been developed but still are at preclinical development stage. These include a mixture of three Ad5 recombinants that expressed the HA, M1 and NP proteins from a H5N1 strain, respectively; the vaccine elicited strong protection against lethal H5N1 challenge in mice and chickens .
An MVA recombinant that expressed the H5 HA protein elicited protection against challenge with three antigenically distinct strains of H5N1 influenza viruses in mice  . Another MVA recombinant which will express all five M1, M2, HA, NA and NP proteins is being developed and should be tested shortly.
Last, but not least, a NDV recombinant that expresses the H5 HA protein was found to be highly immunogenic and protective in chickens and could be used as bivalent vaccine against both Newcastle disease and highly pathogenic H5N1 avian influenza virus infection in poultry  .
A subunit recombinant H5 vaccine based on HA protein expressed in a baculovirus expression system was well tolerated in human volunteers but showed mediocre immunogenicity  , probably due to lack of oligomerization of the antigen .
Many plans have been made on how to mitigate an influenza pandemic once it breaks out, including stockpiling of antiviral drugs, and restricting population movements until a vaccine is available for the specific strain that has broken out. As strain-specific vaccines would be available only several months into the pandemic and would be in short supply, the possibility has been entertained of using 'pre-pandemic' vaccines which would not match the exact pandemic strain but an earlier variant. As seen above, H5N1 vaccines do show cross-clade protection in animal models and thus might confer sufficient protection against death or severe disease due to a new emerging pandemic H5N1 variant. The World Health Organization is planning to stockpile more than 100 million doses of pre-pandemic H5N1 vaccines and several nations are considering the same.