We have shown that the candidate clade 1 H5N1 inactivated split-virion vaccine adjuvanted with an oil-in-water based emulsion adjuvant system can induce neutralising antibodies against recombinant strains derived from three recently emergent clade 2 viruses belonging to different subclades. Furthermore, this cross-clade immunity was induced at a low HA dose of 3.8 µg. These results are encouraging, as they demonstrate that a vaccine based on an existing H5N1 strain could potentially protect against a range of different emerging H5N1 strains. This is the concept on which the pre-pandemic immunisation strategy is based.
To be optimal for use in a pre-pandemic immunisation strategy, we need vaccines that i) are safe, ii) are highly immunogenic, iii) exhibit broad cross-immunity and iv) have long- lived immunity. As with other split-virion or whole-virion H5N1 vaccines based on strains derived by reverse genetics 
, the candidate adjuvanted vaccine derived from the 2004 H5N1 isolate A/Vietnam/1194/2004 seems to be well tolerated with an acceptable safety profile 
. The adjuvanted vaccine was also shown to be highly immunogenic, a dose of 3.8 µg HA was sufficient to achieve immunity against the vaccine strain at a level that was acceptable for licensing in US and Europe 
. An inactivated split-virion H5N1 vaccine has been licensed by the FDA, however this vaccine, which is administered without adjuvant is poorly immunogenic 
. Two 90 µg doses are needed to achieve the level of immunity required for licensing compared to one dose of 15 µg for conventional seasonal split-virion vaccines. Adjuvantation with aluminium was shown to only modestly improve the immunogenicity of inactivated split-virion H5N1 vaccine 
although more promising results were achieved with whole-virion H5N1 vaccines administered with aluminium 
As previously reported 
, after 2 administrations of 3.8 µg HA of the AS adjuvanted rH5N1 vaccine, 84% of the 50 volunteers presented seroprotective HAI titres against A/Vietnam/1194/2004 vaccine strain and 86% presented a four-fold seroconversion rate for neutralising antibodies while in the group of volunteers administered with the non-adjuvanted vaccine these percentages were 4% and 22%, respectively 
In this report we now provide evidence that the adjuvanted clade 1 candidate vaccine exhibits a broad cross-immunity against circulating strains shown to be responsible for human cases 
. The effect of the oil-in-water emulsion based adjuvant system in promoting this cross-immunity was contrasting with the absence of a response in the non-adjuvanted group. We demonstrated that, in addition to the recombinant subclade 2.1 A/Indonesia/5/2005 strain, the vaccine also induced neutralising antibodies against two other recombinant strains derived from the recent drift H5N1 strains A/turkey/Turkey/1/2005 and A/Anhui/1/2005 which are, as defined by WHO, representatives of subclades 2.2 and 2.3 respectively. The ability of the vaccine to induce immunity against these three phylogenetic subclades is of relevance as, together with clade 1, they account for the majority of recent circulating H5N1 isolates and also human H5N1 cases 
Following the first dose of the vaccine a neutralising response against the subclades 2.2 and 2.3 was evidenced in 35%-45% of subjects. It has been estimated that a pandemic vaccine that provides even partial cross-protection (about 30%) could have substantial impact on attack rates 
. Thus in a critical situation where there is not sufficient time or supply of vaccine to administer a second dose, even one dose of the vaccine may help to reduce transmission of the pandemic virus. A high level of cross-immunity (75%–85%) against all three subclades was evident following the second dose. Furthermore we provide evidence that this cross-clade immunity is long-lived as it could still be detected in the majority of subjects at six months following vaccination. The neutralising antibody titres against A/Vietnam/1194/04 homologous virus follow the same trend as the cross-reactive antibody titres against clade 2 viruses (data not shown, manuscript under preparation). Humoral immunity for influenza vaccines has conventionally been assessed by HAI. Our previous experience with A/Indonesia/5/2005 H5N1 strain has shown that cross-reactivity is stronger when assessed by the more sensitive neutralisation assay 
which provides an evaluation of the vaccine activity against both the HA and the NA antigens and consequently, gives a more comprehensive evaluation of the biological activity of the vaccine. This was confirmed in this present study for the recombinant A/turkey/Turkey/1/2005 and A/Anhui/1/2005 strains.
A recent pre-clinical study provides further evidence that vaccination with H5 and N1 antigens from one clade can induce a broadly protective immune response against wild type viruses from another clade 
. Suguitan and colleagues showed that vaccines developed from attenuated strains containing H5 and N1 components from 1997 clade 3, 2003 clade 1 or 2004 clade 1 isolates protected mice from lethal challenge with both homologous and heterologous wild type viruses including more recent 2005 clade 1 and clade 2 viruses 
. Similar data were generated for protection against pulmonary replication following challenge with these different strains in vaccinated mice and ferrets. The authors suggested that the high level of protection afforded by vaccination with the 1997 clade 3 vaccine against challenge with the clade 1 and clade 2 H5N1 viruses isolated over a span of 8 years, indicates that the H5N1 viruses are evolving to infect different birds, and not predominantly to evade antibodies as they do in humans 
. Earlier evidence that an avian influenza vaccine could exhibit cross-immunity came from a study where a surface-antigen vaccine based on the antigenically related H5N3 influenza virus (influenza A/duck/Singapore/97) and adjuvanted with MF59 induced cross-reactive antibodies against H5N1 
. Whereas we did not observe any cross-reactive response following administration of two doses of non-adjuvanted vaccine, these authors did measure some degree of cross-reactivity after three doses of non-adjuvanted vaccine.
We have previously demonstrated the significant antigen dose-sparing effect of including an oil-in-water emulsion based adjuvant system in the candidate vaccine formulation 
. This is now re-enforced by the results of this present study which confirm that the adjuvant enhances the effectiveness of a low antigen dose in broadening the immune response. Baras et al 
recently documented in a stringent preclinical model that the AS adjuvanted candidate vaccine described in the present paper provides protection against cross-clade heterologous challenge in ferrets. The availability of H5N1 vaccines that can elicit a broad cross-protective immunity against different currently circulating H5N1 viruses, including newly emerged strains, is a pre-requisite for the development of a successful pre-pandemic vaccination strategy. Deployment of such vaccines for pre-emptive vaccination could play a key role in pandemic mitigation during the several months that it would take to produce an H5N1 vaccine exactly matched to a pandemic strain.