Vaccination of Macaques with Adjuvanted Formalin-Inactivated Influenza A Virus (H5N1) Vaccines: Protection against H5N1 Challenge without Disease Enhancement

No vaccine-related adverse clinical signs or effects on body weight gain were observed with any of the vaccines. All animals lost weight after the challenge, with no differences between groups. One accidental death occurred among controls.

Responses against the vaccine strain in samples obtained before and after vaccination and before and after challenge were assessed by evaluating hemagglutination inhibition (HI) and by using an enzyme-linked immunosorbent assay (ELISA) as previously described (5, 11) (Fig. 1). Functional (HI) responses were detected only after the second dose, with strong titer increases consistent with results in humans, providing additional evidence that a two-dose strategy is needed to elicit functional antibodies against pandemic strains (12) (Fig. 1). The lack of difference in responses to the two adjuvanted vaccines was probably due to the high antigenic contents, which may have led antibody responses to plateau. In contrast, without an adjuvant, HI titers remained undetectable until after the challenge, although antibodies were detected by ELISA and partial protection against infection was observed upon histopathological analysis (see below). These results are in agreement with recent mouse and ferret data demonstrating protection in the absence of strain-specific functional antibodies (4, 8, 13), suggesting that low or undetectable levels of antibodies after immunization with H5 vaccines do not necessarily indicate a lack of vaccine effectiveness. The HI results presented in Fig. 1 were obtained using chicken erythrocytes in the assay. Assays of pooled subsets of sera collected just before and after challenge (data not shown) were also performed using horse erythrocytes (or horse red blood cells [HRBC]). Samples that were negative in assays using chicken erythrocytes remained negative when HRBC were used. Detectable HI titers were approximately 10-fold higher when assessed using HRBC; 15 days after the challenge, geometric mean HI titers determined using HRBC (expressed as reciprocal dilutions) were 320 for controls, 1,280 for the nonadjuvanted vaccine group, 2,560 for the aluminum-adjuvanted vaccine group, and 5,120 for the AdjA-adjuvanted vaccine group.

The cross-reactivity of antibodies against a clade 2 strain (influenza virus A/Indonesia/5/05/RG-2) was assessed by evaluating HI (using HRBC) in sera pooled by group before and 15 days after challenge. Before challenge, cross-reactive antibodies were detected in samples from both adjuvanted vaccine groups (with a titer of 80 for each adjuvanted vaccine group) but not in samples from the other two groups, whereas after challenge, cross-reactivity was detected in samples from all groups (titers were 40 for the PBS group, 160 for the nonadjuvanted vaccine group, and 640 for each adjuvanted vaccine group).

Analyses of H5N1-specific cell-mediated responses were performed before challenge by gamma interferon (IFN-γ) enzyme-linked immunospot assay and ELISA determination of IFN-γ, interleukin-5 (IL-5), and IL-13 titers in supernatants of stimulated peripheral blood mononuclear cells. No differences in cytokine secretion between groups were detected (data not shown). However, it is possible that early responses were missed, as the earliest sample tested was obtained 25 days after the first vaccination. Day 7 or 14 samples would have been preferable, but the primary objective of the blood sampling was the determination of serum antibody titers, and the number of blood draws and associated anesthetizations were limited for ethical reasons. In quantitative (TaqMan) PCR analysis of IFN-γ, IL-4, and IL-13 mRNA in lung samples, no differences between treatment groups on days 5 and 15 postchallenge were observed: IFN-γ and IL-4 gene expression levels were low, and IL-13 was undetectable in all animals (data not shown). These observations contrast with the enhanced Th2 cytokine production seen during the RSV disease exacerbation phenomenon (3, 9).

After sacrifice, lungs were weighed and examined and samples were fixed and preserved in 10% neutral buffered formalin and then analyzed using hematoxylin and eosin slide staining (Fig. 2). Animals vaccinated with either adjuvanted vaccine had lower relative lung weights than controls, which correlated with decreased lung pathology (Table 1). The results of macroscopic examination correlated with microscopic observations of bronchointerstitial pneumonia, which was acute on day 5 postchallenge and subacute on day 15.

Evidence of bronchointerstitial pneumonia was seen in all groups, particularly among controls. Animals vaccinated with the emulsion-adjuvanted vaccine had the fewest and mildest findings (Table 1). There were no appreciable differences in the incidence or severity of eosinophilic infiltrates in the lungs between vaccinated and unvaccinated animals. While increased eosinophilic infiltrates were observed in the lungs of some animals from all groups, these findings did not correlate with the severity of pneumonia, as there was no detectable difference in the incidence or severity of infiltrates between groups (Table 1). Similar to the findings regarding Th2 cytokine production, this observation contrasts with the features seen during the RSV disease exacerbation phenomenon, which include increased pulmonary eosinophila (3, 9).

Quantitative PCR and Madin-Darby canine kidney (MDCK) cell virus culture methods were used to quantify virus in pharyngeal and nasal swabs collected before and after challenge and in cranioventral, craniodorsal, caudoventral, and caudodorsal lung samples. RNA was extracted from samples using a Magnapure LC isolation station (Roche Applied Science, Penzberg, Germany); H5N1 RNA content was quantified using a Taqman PCR assay. Virus culture was performed as previously described (10).

Results were expressed as influenza virus TCID₅₀ equivalents per milliliter, determined using a standard curve produced from a titrated stock of virus which was serially diluted, with each dilution undergoing nucleic acid extraction and PCR amplification in the same manner as test samples. Both adjuvanted vaccines protected against viral replication in the lungs: no virus in the lungs of animals receiving vaccines with adjuvants was detected by either method, whereas virus in the lungs of control animals was detected by both methods (Fig. 3). In animals vaccinated without an adjuvant, H5N1 was detected in the lungs but at lower levels than those detected among controls and by PCR only (Fig. 3). Both adjuvanted vaccines also partially protected the upper respiratory tract, as for the groups receiving vaccines with adjuvants, low levels of viral material in pharyngeal swabs from a few animals were detected by PCR and only on day 2 postchallenge (Fig. 4). Virus culture results for pharyngeal swabs were negative, except for swabs collected on day 2 from two animals (Fig. 4) vaccinated without an adjuvant (180 TCID₅₀ equivalents/ml was found in the sample from each animal). Results for nasal swabs from all animals were negative by both methods, except for a sample from one animal in the nonadjuvanted group in which viral material was detected by PCR on day 2 postchallenge only (data not shown).

These observations suggest that the nonadjuvanted vaccine provided a low level of protection that was insufficient to prevent initial viral replication and shedding in the pharynx and lungs but was able to prevent further replication. This hypothesis is supported by the detection of virus in the pharynges of the animals receiving the nonadjuvanted vaccine on day 2 postchallenge only, while it was present for up to 10 days in the control group.

In summary, the vaccination of monkeys with a formaldehyde-inactivated H5N1 pandemic influenza vaccine at the dose level of 30 μg of hemagglutinin with aluminum hydroxide or an oil-in-water emulsion adjuvant did not exacerbate pneumonia when the animals were subjected to intratracheal challenge with the parental wild-type influenza virus A/Vietnam/1194/2004 (H5N1) strain. Furthermore, vaccination—in particular with a vaccine including a new oil-in-water emulsion adjuvant—was associated with a marked reduction in the incidence and severity of interstitial pneumonia and protected against infection in the lungs and upper respiratory tract. Protection against disease and the viral load correlated with the ability to induce the production of neutralizing antibodies. The adjuvanted vaccines, especially the emulsion-adjuvanted vaccine, offered a greater degree of protection than the nonadjuvanted vaccine, but even the nonadjuvanted vaccine provided partial protection. Vaccination, therefore, did not exacerbate the disease and, moreover, protected against infection. We have demonstrated that the influenza A/H5N1 viral load can be controlled by vaccination and that this control results in protection against pneumonia. Recent results clearly highlight the central role of high viral loads and the associated high levels of cytokine production (cytokine bursts) and the resulting intense inflammatory reactions in the pathogenesis of human H5N1 disease (1, 6). Other studies indicate that the successful control of influenza viral replication is associated with a better clinical outcome (2). We can therefore expect that the immunization of humans with this clade 1 H5N1 aluminum-adjuvanted vaccine capable of reducing the lung viral load and inducing cross-reactive antibodies may attenuate disease caused by homologous and heterologous strains and may also reduce the aerosol transmission of virus to susceptible humans. Vaccines including the emulsion-based adjuvant may result in greater protection than nonadjuvanted vaccines and those incorporating other adjuvants.

We thank the following contributors: at sanofi pasteur, Christophe Charnay for technical support during the in vivo phase of the studies, Isabelle Legastelois for the HI assays, Marie-Françoise Klücker for the formulation of vaccines, and Grenville Marsh for editorial assistance in preparing the manuscript, and at ViroClinics, Geert van Amerongen and Frank Pistoor for the animal and laboratory work. Finally, we thank W. Lim for providing the wild-type H5N1 strain.