We have reported the first human trial of a live attenuated H7 influenza A virus vaccine. The live attenuated, cold-adapted, temperature sensitive vaccine, H7N3 BC 2004/AA ca elicited adverse events that were generally mild, was restricted in replication and immunogenic in seronegative healthy adult volunteers.
Several other H7 vaccines have been developed and evaluated in pre-clinical studies [38
] and several H7 vaccines have also been developed for use in chickens [42
]. All of these are inactivated whole or split virus vaccines that contain the HA and NA from avian H7 viruses and the internal protein genes from PR8 vaccine donor virus. To date, 2 H7N7 vaccines, an H7N1 vaccine, and an H7N2 vaccine have been studied, including both non-adjuvanted vaccine preparations and preparations containing alum or immune stimulatory complexes (ISCOMs). In animal studies, these vaccines are of variable immunogenicity, yet generally protect against challenge with either homologous or heterologous H7 viruses [38
]. Several inactivated H7 virus vaccines are currently being evaluated in Phase I clinical trials.
The H7N3 BC 2004/AA ca
vaccine was generally well tolerated by our subjects. The majority of symptoms were unlikely to be related to the vaccine: 6 of the 8 subjects who had upper respiratory tract symptoms after the first dose had rhinovirus detected by rRT-PCR, but only 1 of the 8 was shedding vaccine virus at the time. The laboratory abnormalities are also unlikely to be related to the vaccine. One subject had an episode of transient neutropenia during the study that resolved by the next day. He had a history of lower neutrophil counts, both before and long after vaccination, and likely has benign ethnic neutropenia, which is common and well documented in those of African or Mediterranean descent [46
]. Four subjects had 5 episodes of elevated ALT during the study. Only 2 of these had vaccine virus detected by rRT-PCR after dose 1. This elevation of serum transaminases is not likely to be due to vaccine virus, as this was not a phenomenon seen with wild-type H7 influenza infection. Rather, this is most likely due to alterations in diet and activity commonly seen in Phase I inpatient studies [48
Similar to the other avian LAIVs tested, including H9N2 G9/AA ca[27
], H5N1 VN 2004/AA ca
, and H5N1 HK 2003/AA ca
(Karron et al., manuscript submitted for publication), the H7N3 BC 2004/AA ca
vaccine influenza A virus was more restricted in replication than the H1N1/AA ca
and H3N2/AA ca
influenza A viruses contained in the intranasal seasonal LAIV in seronegative subjects. The reason for this higher restriction is not completely understood, but may be related to decreased affinity of the avian HAs for receptors in the upper airways of humans. Human influenza virus HAs preferentially bind to receptors terminating in N-acetylneuraminic acid α2,6 galactose (α2,6Gal), which are abundant on non-ciliated epithelial cells in the upper and lower human respiratory tract. Many avian influenza viruses, on the other hand, preferentially bind to receptors terminating in α2,3Gal [34
], which are found in type II pneumocytes in humans but in limited numbers of epithelial cells in the upper respiratory tract [50
]. However, this interpretation is offered with the caveat that human parainfluenza viruses that bind α2,3Gal replicate to high titer in the upper respiratory tract of humans [52
]. Alternatively, the constellation of the avian influenza HA and NA genes combined with internal protein gene segments from a human influenza virus may have resulted in a vaccine with a host-range restricted phenotype for humans. It is possible that an avian virus that emerges as a pandemic virus (i.e., acquires genetic mutations that allow it to transmit readily among the human population) will have HA or NA alterations that control infectivity and spread in humans. Such HA or NA genes, when incorporated into LAIV may be more infectious and immunogenic in humans than the LAIV vaccines containing the avian HA and NA genes of the present report.
Despite this restriction, the H7N3 vaccine was detected more frequently by viral culture than either the H9N2 or H5N1 LAIVs. This may be because contemporary North American H7 HAs, such as the one included in this vaccine, possess an increased affinity for α2,6Gal receptors when compared to Eurasian H7 HAs [55
]. Since H9N2 G9 HK 1997 influenza A viruses can also bind to α2,6Gal receptors, this cannot be the only explanation for the observed differences in replication between vaccine strains; rather, it may be a contributing factor. In addition, the H7N3 vaccine has an NA antigen that is not contained in circulating human strains of influenza so individuals are unlikely to have preexisting immunity to the NA that could limit viral replication.
A surprising finding in our study was the difference in the rate of vaccine virus detection by rRT-PCR following the first and second dose of vaccine. Seventeen of 21 subjects had virus detected by rRT-PCR on day 1 after the first vaccination, 13 of whom did not have any virus detected on subsequent days. In contrast, none of the subjects had vaccine virus detected by rRT-PCR following the second dose of vaccine. In previous studies, the detection of vaccine virus on day 1 only was attributed to detection of input virus[27
]. However, the findings from this study suggest that low-level replication of vaccine virus occurs after the first dose of vaccine, but that replication following the second dose is inhibited, perhaps by the adaptive immune response. The pattern of ASC responses (discussed below) supports this hypothesis. While we chose not to include individuals with nasal wash specimens that were positive by rRT-PCR on day 1 only in our calculations of percentages of infected vaccinees, these data suggest that perhaps the H7N3 BC 2004/AA ca
influenza A virus was more infectious than we have estimated.
Although the co-circulating rhinovirus had the potential to confound our data, we do not believe that it affected either vaccine virus replication or the immune response to the vaccine. Of those who shed rhinovirus, 4/6 (67%) were positive for vaccine virus by rRT-PCR on day 1. One of these subjects (17%) also had vaccine virus detectable by culture. Of those who were rhinovirus negative, 4/15 (27%) had vaccine virus detectable by culture, and 13/15 (87%) had positive rRT-PCR for vaccine virus (). Those who shed rhinovirus were neither more nor less likely to have a serological response to the vaccine ().
One of the difficulties in evaluating potential vaccines for avian influenza lies in determining the correlates of vaccine-induced immunity [3
]. For inactivated seasonal influenza vaccines, the correlate of protection accepted by regulatory agencies has been defined as an HI titer of
1:40, measured in a standard assay using 4U of HA [56
]. However, seasonal LAIV has been shown to protect against natural infection or wt
challenge even when HI titers
1:40 were not achieved, presumably through the induction of local mucosal antibody and cellular immunity [21
]. Moreover, this may not be the best correlate of protection against infection with H5 and H7 avian influenza viruses, since individuals infected with these viruses often fail to develop HI responses of this magnitude [14
]. For this reason, alternative assays, such as the microneutralization assay, ELISA [35
] and, in the case of H7, a modified version of the HI assay using horse RBC and 2U of HA have been developed [14
To assess the antibody response to the H7N3 BC 2004/AA ca
vaccine, we used several assays in addition to the standard HI and microneutralization assays, in order to further understanding of the immune response to this novel antigen. We tested sera in an HI assay using 2 HAU of antigen and horse erythrocytes, since a reanalysis of serum from the H7N7 virus outbreak in The Netherlands showed that this was the most sensitive and specific method for detecting infection. In that study, the modified HI assay using 2 HAU was positive in 85% of patients with culture-proven H7N7 infection. However, most individuals had only modest increases in HI titer, such that only 6% had titers of 1:40 [14
]. Using 2 HAU of antigen, we found that 62% of volunteers had a 4-fold or greater rise in HI titers after either 1 or both doses of vaccine. The mean post-vaccination titer at 2 months was 1:8 in all subjects and 1:14 in the vaccinees who had a 4-fold or greater rise in titer after any dose (). Titers of 1:32 were achieved in 14% of vaccinees, comparable to what was observed following natural infection.
We also measured neutralizing antibody, H7-specific IgG and IgA ELISA antibody[59
], and production of ASCs against the vaccine virus, AA ca
, and rH7 HA antigen. Interestingly, the most sensitive measure of response was the serum IgA assay, despite the fact that a heterologous H7 was used (derived from the Eurasian H7N7 A/Netherlands/219/03 virus). The Archetti-Horsfall value of antigenic relatedness for these two HA antigens, tested with post-infection mouse sera, is 33%, which represents a 4 to 8-fold difference in cross-neutralizing titer [60
]. Eleven of 21 volunteers had a 4-fold or greater rise in serum IgA titer after the first dose, and 5 of 17 responded after the second dose. Serum IgA has been used as a marker for response to parainfluenza and respiratory syncytial virus vaccines in infants who may have circulating maternal IgG [61
]. The biological significance of this response remains to be determined. Serum IgA may be a surrogate marker for a mucosal immune response that is triggered by the intranasal vaccine. Clements and Murphy found that there was a significant association between serum and nasal wash IgA and IgG after either live or inactivated H1N1 or H3N2 influenza vaccine [63
]. Although we also measured nasal wash H7-specific IgA, our assay may not be sensitive enough to detect a mucosal response in most vaccinees.
Measurement of vaccine-specific ASC showed a greater response to the first dose than to the second dose of vaccine (). As discussed above, this might reflect the decreased infectivity of the second dose of vaccine. The magnitude of the vaccine-specific IgG ASC response observed in H7-naïve individuals after the first dose of vaccine was comparable to that previously observed in older children administered the intranasal seasonal LAIV [36
]. Although serum IgA was the most sensitive measure of antibody response, this was not reflected in the IgA ASC data. This may be due to timing: the ASC were measured on day 7 based on work by Sasaki et al [36
], and the ELISA for IgA was performed on day 28 sera. The kinetics of the ASC response to AI viruses may be different than to the human influenza viruses.
In our study, the immunological response to the vaccine in a few individuals increased gradually, such that 2-fold rises in titer were observed following each dose of vaccine. Moreover, delayed rises in HI and/or neutralizing antibody titers were observed in 3 of 4 individuals who received only 1 dose of vaccine, with higher titers achieved at the 2 month than at the 1 month time-point. These data, combined with the rRT-PCR data discussed above, suggest that although the best responses occurred in those who received two doses, perhaps a single dose of vaccine would be sufficient to induce an immune response, which would be ideal in the event of a pandemic. This is supported by preclinical studies of this vaccine, in which mice that received only 1 dose of the H7N3 BC 2004/AA ca
vaccine had increasing neutralizing antibody titers between days 28 and 56, rising to nearly the same level as those that had received 2 doses [26
]. The immune response of humans to a single dose of the H7N3 BC 2004/AA ca
vaccine should be explored further.
In summary, we have shown that 2 doses of a live attenuated cold adapted H7N3 BC 2004/AA ca vaccine are generally well tolerated and immunogenic when given to healthy adults. Replication of the vaccine virus was restricted, especially after the second dose. The detection of vaccine virus by rRT-PCR on day 1 after the first vaccination may reflect a low level of replication not detected by culture. We found that serum IgA antibody to rH7 HA antigen was the most sensitive measure of the immune response to this vaccine. Our data suggest that H7N3 BC 2004/AA ca should be investigated further and that it might be useful to assess the immunogenicity of a single dose of this vaccine.