The starting assumption for the use of animal models is that, if immunogenicity follows similar patterns in animal models and humans, it may be reasonable to extrapolate efficacy as well. The animal models used to study influenza include ferrets and mice, much less frequently guinea-pigs and non-human primates. The assessment of vaccine efficacy in ferrets includes challenge with high doses (105 ID50 or a lethal dose) of homologous and heterologous pandemic viruses followed by measurements of clinical signs and viral replication (serial nasal washes).
To provide data on viral replication (peak, kinetics of viral replication and clearance) in the upper and lower respiratory tract, animals need to be sacrificed at serial time points (at least two time points are recommended).
The existing consistencies and inconsistencies between the animal models and humans were summarised as following.
Consistent findings from animal and human studies include: i) inactivated vaccines that elicit robust HI and VN responses also protect animals from experimental challenge; ii) adjuvanted vaccines are more immunogenic than unadjuvanted vaccines allowing antigen sparing; iii) unadjuvanted split vaccines are less immunogenic than adjuvanted split vaccines, whilst the latter and whole virion vaccines show similar immunogenicity. For example unadjuvanted whole virion vaccines were immunogenic at low doses and protected against live virus challenge in mice14; both immunogenicity and protection in this model were comparable to that of a whole virus H5N1 vaccine, which had previously been demonstrated to induce high titres of antibodies in clinical studies.
A 2 dose-regimen of the split virion pH1N1-AS03 vaccine (3.75 or 1.9 μg) was required to protect naïve ferrets against homologous challenge whereas higher doses of unadjuvanted split vaccine (15 μg) or one dose of adjuvanted vaccine conferred only 17-33% survival in ferrets after challenge15.
Inconsistencies between animal and human studies include: a) a single dose of 15μg unadjuvanted split virion pH1N1 vaccine was immunogenic in older children and adults; b) a single dose of 3.75 μg of AS03- adjuvanted split virion pH1N1 vaccine was immunogenic in older children and adults; c) unadjuvanted and adjuvanted split virion pH1N1 vaccines were effective in humans in protecting against MAARI (Medically Attended Acute Respiratory Illness); d) immunogenicity of LAIV in animal models exceeds that seen in clinical studies; and e) alum enhances vaccine immunogenicity significantly in ferrets, but the benefit in humans is equivocal.
Examples of the above include an immunogenicity and efficacy study of pH1N1 LAIV (p-LAIV) in mice and ferrets16 that showed that the p-LAIV was immunogenic in both species and conferred complete protection against challenge with pH1N1, in contrast to seasonal (s-)TIV that did not confer protection in either animal model and to s-LAIV that did not confer protection in ferrets (in mice, 2 doses of s-LAIV led to complete protection in upper respiratory tract and partial protection in the lungs). In another ferret study17 (n=118) it was found that alum improved the efficacy of an H5N1 (A/Vietnam/1203/ 2004) split virion vaccine as measured by immunogenicity (HI, VN), mortality, morbidity and brain invasion, whereas alum is not an effective adjuvant for H5N1 vaccines in humans. In contrast Cox et al. reported that alum enhanced immunogenicity and/or efficacy of a H7N1 split virion cell-grown vaccine in mice and ferrets (IRV 2009) as well as immunogenicity in humans (Vaccine, 2009), whilst as expected the non-adjuvanted formulation was poorly immunogenic if at all.
Concerning the importance of priming in animal models, Dr Subbarao described five preclinical studies to show that in animal models prior experience with influenza infection or vaccination played an important role in the response to pH1N1 vaccines. For example some authors18 found that prior infection with seasonal H1N1 or s-LAIV (but not s-TIV) primed for a robust response to a single dose of p-LAIV and protection from challenge in mice. Similarly two different ferret studies19 showed that prior infection with seasonal H1N1 (but not vaccination with s-TIV) primed for a more robust response to split virion pH1N1 vaccine and altered morbidity to a pH1N1 challenge, despite the detection of only minimal levels of cross-reactive antibodies. Suguitan et al., PLoS One 2011, evaluated different prime-boost regimens in ferrets and found that priming with DNA vaccine boosts immunogenicity of H5N1 LAIV in those animals.
In a study in mice on the impact of prior vaccination against s-TIV on antibody responses against pH1N1 vaccine, vaccination with one/two doses of pH1N1 unadjuvanted split virion vaccine induced consistently higher antibody response in s-TIV primed animal vs. naïve animals, whereas one low dose of an AF03-adjuvanted pH1N1 was able to induce higher titres vs. the unadjuvanted vaccine regardless of the priming status of the animals.
In summary, extrapolations from preclinical studies need to be carefully balanced over a number of aspects, most importantly that animals are immunologically naïve, whereas most humans have prior experiences with influenza, and this is difficult to model in animals. In addition viral challenge is quite stringent (i.e. high virus dose) and the intra-tracheal route of inoculation is not physiological; sometimes the vaccine preparations used in animals differ from the vaccine in clinical use and specifics of dose sparing cannot be extrapolated to humans. Overall the assessment in animal models is most useful for predicting efficacy in humans if correlates of protection are known and can be measured. A challenge for the future is to identify immune correlates of protection for LAIV and newer vaccine platforms that can be measured in animal models. Other challenges include developing models that reflect the immunological priming that occurs in humans, including responses to T and B cell epitopes, conserved internal proteins and HA stem. However it should be noted that there is no standardization of assays in animals to detect different immune responses and that there are a number of open questions, including how to properly prime and whether it’s possible to cross-prime.
After Kanta Subbarao, Laboratory of Infectious Diseases NIAID, NIH, USA in EMA highlights, May 2013
14 Kistner et al., PLoS One 2010
15 Baras et al., Vaccine 2011
16 Chen et al., J Inf Dis 2011
17 Layton et al., PLoS One 2011
18 Chen et al., PNAS 2011
19 Ellebedy et al., CIV 2010 and Vaccine 2011