Abstract
Cross-reactive antibody responses of 3 trivalent, live attenuated intranasal influenza vaccine (FluMist) formulations containing 3 different H1N1 strains (A/Texas/36/91, A/Shenzhen/227/95, and A/Beijing/262/95) were evaluated in initially seronegative children. FluMist containing A/Shenzhen/227/95 was more likely to induce cross-reactive hemagglutination inhibition (HAI) antibody against A/Texas/36/91 than against A/Beijing/262/95, and FluMist containing A/Beijing/262/95 induced low levels of cross-reactive HAI antibody against A/Shenzhen/227/95 and A/New Caledonia/20/99. The observed differences in HAI cross-reactivity seem to be partly related to the number of amino acid (aa) differences on the hemagglutinin 1 domain (328 aa residues) rather than the hemagglutinin protein (550 aa residues)
Influenza is a major cause of hospitalization and lower respiratory tract illnesses in humans [1]. Vaccination is the primary method for preventing influenza and its severe complications [1, 2]. Current influenza vaccines contain 3 virus strains (i.e., 2 type A strains [H1N1 and H3N2] and 1 type B strain) representing the influenza viruses likely to circulate in the upcoming winter and need to be updated frequently on the basis of virus surveillance data [1, 2]. Because influenza viruses frequently undergo antigenic drift and because heterologous strains can cocirculate in an epidemic season [1, 2], an influenza vaccine that can induce cross-reactive antibody against heterologous strains would be optimal
The hemagglutinin (HA) protein of influenza viruses is the major surface protein that induces a protective antibody response. The HA protein is a homotrimer and is synthesized as a single polypeptide (HA0) that is subsequently cleaved into 2 chains (HA1 and HA2) [3]. The 3-dimensional (3-D) structure of the HA protein of strain A/Aichi/2/68 (H3N2) has been determined, and 5 antigenic sites on the HA1 polypeptide have been proposed [3]. The 3-D structure of the HA protein of H1N1 viruses has not been determined but has been inferred to be similar to the 3-D structure of the HA protein of H3N2 viruses [3, 4]. Thirty-two amino acid positions in the 5 antigenic sites have been mapped on the basis of laboratory variants selected in the presence of mouse monoclonal antibodies [4]. However, the significance of the 32 aa positions in polyclonal antibody response is unclear
A trivalent, live attenuated intranasal influenza vaccine (FluMist) has been shown to be generally safe and highly efficacious in children [5–11]. The present study evaluated cross-reactive antibody responses of 3 FluMist formulations, containing 3 different H1N1 strains, in initially seronegative children. Because amino acid–sequence data of the HA protein could be useful for predicting antigenicity of influenza viruses in humans, amino acid differences in the HA protein, between the vaccine and heterologous strains, also were assessed
Subjects, materials, and methodsCold-adapted (ca), trivalent, live attenuated influenza vaccines were supplied by MedImmune Vaccines (formerly Aviron) as intranasal sprays. The mean TCID50 of each of the 3 virus strains included in the vaccines was ∼107.0 [5–11]. The chosen strains matched the antigens recommended by the US Public Health Service, for different seasons, as shown in table 1. Wild-type (wt) virus seeds were acquired from the US Centers for Disease Control and Prevention (CDC), and vaccine reassortants were produced as described elsewhere [12]. Serum samples from healthy children aged 1–3 years who received 2 doses of vaccine were selected from 3 different studies (table 1). All studies used a 2-dose regimen, ∼28–60 days apart, and all participants had not received influenza vaccine previously
Human serum hemagglutination inhibition (HAI) antibody cross-reactivity and amino acid differences in the hemagglutinin (HA) protein of influenza H1N1 viruses
Human serum hemagglutination inhibition (HAI) antibody cross-reactivity and amino acid differences in the hemagglutinin (HA) protein of influenza H1N1 viruses
Informed consent was obtained from participants’ guardians. Human-experimentation guidelines of the US Department of Health and Human Services were followed in the conduct of this clinical research
Serum samples were obtained just before the first dose was given and ∼4 weeks after the second dose was given and were assayed for hemagglutination inhibition (HAI) antibody levels against the vaccine strains and selected heterologous strains, as shown in table 1 [13]. The heterologous strains were selected because they have been recommended as vaccine strains for different seasons
Nucleotide sequences of the HA genome were determined and further translated into amino acid sequences. Influenza viral RNA was isolated using the RNA STAT-50 LS reagent (Tel-Test) or the QiaAmp Viral RNA mini kit (Qiagen). Viral RNA was amplified by use of the GeneAmp RNA polymerase chain reaction (PCR) kit (PE Applied Biosystems) and specific primers (available from the authors upon request). The amplified gene segments were either gel-purified or column-purified, by use of a Qiagen PCR purification kit before sequencing. Nucleotide sequencing was conducted by cycle sequencing using reverse-transcriptase PCR–amplified cDNA products and type-specific sequencing primers (available from the authors upon request). The cycle sequencing was performed by use of the ABI PRISM dRhodamine Terminator Cycle Sequencing Ready Kit (PE Applied Biosystems), the MicroSpin G-50 column (Amersham Pharmacia Biotech), and the ABI 377 DNA Sequencer (PE Applied Biosystems). Nucleotide sequence data were processed and translated into amino acid sequences by use of DNA Sequencing Analysis (PE Applied Biosystems), Sequencher (Gene Code), and MacDNASIS (Hitachi). Amino acid sequence data were aligned with CLUSTAL_W software (available at: http://www.ebi.ac.uk), using A/PR/8/34 (H1N1) as the template for amino acid position (GenBank accession no. AAA43194), because this virus has been used to map 32 aa positions in the 5 antigenic sites [4]. The 5 antigenic sites and the corresponding amino acid positions are as follows: Cb, aa 70, 71, 73–75, and 115; Ca1, aa 165, 169, 178, 203, 236, and 270; Ca2, aa 136, 139, 141, 220, and 221; Sa, aa 124, 125, 154, 156, 158, 159, and 161–163; and Sb, aa 152, 155, 188, 189, 192, and 194
The starting serum dilution in the HAI assay was 1:4, and HAI titers <4 were assigned to be 2 for the calculation of geometric mean titer (GMT). HAI titers ⩽4 were defined as seronegative. Seroconversion was defined as a ⩾4-fold increase in HAI titers measured before and after vaccination. To exclude the effect of preexisting influenza antibody, only children who were seronegative to the vaccine and heterologous strains before vaccination were included for analysis. The proportions of children who experienced seroconversion to homologous and heterologous strains were tested for statistical significance by use of McNemar’s test. All statistical analyses were performed using Epi Info (version 6.02; available at: http://www.cdc.gov) or StatXact-3 (Cytel Software)
ResultsOf 78 seronegative children vaccinated with ca A/Texas/36/91, 41% and 17% seroconverted to ca A/Texas/36/91 and ca A/Shenzhen/227/95, respectively (P<.01, McNemar’s test) (table 1). Of 80 seronegative children vaccinated with ca A/Shenzhen/227/95, 89% and 55% seroconverted to ca A/Shenzhen/227/95 and ca A/Texas/36/91, respectively (P<.01, McNemar’s test) (table 1). Of 218 seronegative children vaccinated with ca A/Shenzhen/227/95, 76% and 18% seroconverted to ca A/Shenzhen/227/95 and ca A/Beijing/262/95, respectively (P<.01, McNemar’s test) (table 1). Of 21 seronegative children vaccinated with ca A/Beijing/262/95, 71%, 10%, and 5% seroconverted to ca A/Beijing/262/95, ca A/Shenzhen/227/95, and wt A/New Caledonia/20/99, respectively (71% vs. 10% [P<.01]; 71% vs. 5% [P<.01]; and 10% vs. 5% [P=1.00, McNemar’s test]) (table 1)
Overall, FluMist containing A/Shenzhen/227/95 is more likely to induce cross-reactive HAI antibody against A/Texas/36/91 than against A/Beijing/262/95. In addition, FluMist containing A/Beijing/262/95 induced low cross-reactive HAI antibody against A/Shenzhen/227/95 and A/New Caledonia/20/99
The HA proteins of ca A/Texas/36/91 and ca A/Shenzhen/227/95 have 14 aa differences, located at positions 47, 66, 125, 133, 145, 152, 160, 221, 268, 272, 372, 435, 450, and 497 (table 2); 10 of the 14 aa differences are located on the HA1 domain, and 3 of them are located on the suggested antigenic sites (aa 125, 152, and 221) (tables 1 and 2). Differences between homologous and heterologous seroconversion rates for these 2 viruses were 24% (41% vs. 17%) for ca A/Texas/36/91 vaccinees and 34% (89% vs. 55%) for ca A/Shenzhen/227/95 vaccinees (table 1). The HA proteins of ca A/Shenzhen/227/95 and ca A/Beijing/262/95 have 15 aa differences, located at positions 43, 47, 71, 80, 129a, 145, 162, 268, 270, 272, 344, 372, 431, 435, and 450; 10 of the 15 aa differences are located on the HA1 domain, and 3 of them are located on the suggested antigenic sites (aa 71, 162, and 270) (tables 1 and 2). Differences between homologous and heterologous seroconversion rates for these 2 viruses were 58% (76% vs. 18%) for ca A/Shenzhen/227/95 vaccinees and 61% (71% vs. 10%) for ca A/Beijing/262/95 vaccinees (table 1). Although the HA proteins of ca A/Beijing/262/95 and ca A/Shenzhen/227/95 have 15 aa differences, as mentioned above, the HA proteins of ca A/Beijing/262/95 and wt A/New Caledonia/227/95 have 12 aa differences, located at positions 69, 125, 132, 152, 162, 182, 186, 190, 221, 272, 309, and 344; 11 of the 12 aa differences are located on the HA1 domain, and 4 of them are located on the suggested antigenic sites (aa 125, 152, 162, and 221) (tables 1 and 2). Difference between homologous and heterologous seroconversion rates for these 2 viruses was 66% (71% vs. 5%) for ca A/Beijing/262/95 vaccinees (table 1). These 2 viruses had the largest differences in cross-reactive antibody responses measured by seroconversion or GMT ratio and also had the largest amino acid differences on the HA1 domain and on the 5 suggested antigenic sites, but not on the HA protein
Deduced amino acid differences in the hemagglutinin (HA) protein of influenza H1N1 viruses compared in the present study
Deduced amino acid differences in the hemagglutinin (HA) protein of influenza H1N1 viruses compared in the present study
Overall, 22 (79%) of the 28 aa positions detected among the 4 viruses are located on the HA1 domain (table 2). In addition, the differences in human HAI antibody cross-reactivity seem to be related, in part, to the number of amino acid differences on the HA1 domain and the suggested antigenic sites, but not to the number of amino acid differences on the HA protein (table 1)
DiscussionA/Texas/36/91 virus was the vaccine strain for the 1996–1997 season and it was replaced with A/Bayern/7/95-like viruses (potential epidemic strains) for the 1997–1998 season because, in adults receiving inactivated vaccines, it induced 50% lower HAI GMT against A/Bayern/7/95 virus than against the vaccine virus [14]. In the present study, in immunologically unprimed children, live attenuated influenza vaccine containing A/Texas/36/91 virus induced 50% lower HAI GMT against A/Shenzhen/227/95 virus (A/Bayern/7/95-like) than against the vaccine virus (HAI GMT, 3 vs. 6). A/Bayern/7/95-like viruses were replaced with A/Beijing/262/95-like viruses (potential epidemic strains) for the 1998–1999 season because, in adults receiving inactivated vaccines, it induced 80% lower HAI antibody titers against A/Beijing/262/95 virus than against the vaccine virus [14]. In the present study, in immunologically unprimed children, live attenuated influenza vaccine containing A/Shenzhen/227/95 virus induced 77% lower HAI GMT against A/Beijing/262/95 virus than against the vaccine virus (HAI GMT, 3 vs. 13). A/Beijing/262/95-like viruses were replaced with A/New Caledonia/20/99-like viruses (potential epidemic strains) for the 2000–2001 season because, in adults receiving inactivated vaccines, it induced 65% lower HAI GMT against A/New Caledonia/20/99-like viruses than against the vaccine virus [14]. In the present study, in immunologically unprimed children, live attenuated influenza vaccine containing A/Beijing/262/95 virus induced 91% lower HAI GMT against A/New Caledonia/20/99 virus than against the vaccine virus (HAI GMT, 2 vs. 22). Overall, the cross-reactivity profiles in adults receiving inactivated vaccines were comparable to those in immunologically unprimed children receiving 2 doses of FluMist, as noted above
For children receiving inactivated influenza vaccines, few cross-reactive antibody data were available for comparison. It has been documented that inactivated vaccines are less likely to induce cross-reactive antibody against antigenic variants in immunologically unprimed children than in primed children [15]. Belshe et al. [8] have documented that, in immunologically unprimed children, FluMist induced higher cross-reactive antibody responses against H3N2 antigenic variants than did inactivated vaccines. It remains unknown whether FluMist and inactivated vaccines would induce comparable cross-reactive antibody to H1N1 antigenic variants in immunologically unprimed children. Head-to-head studies are needed to clarify this point
Correlates of immunity induced by live attenuated influenza vaccines have not been fully defined. In a challenge study, FluMist was 83% (95% confidence interval, 60%–93%) efficacious at preventing shedding of the homologous H1N1 vaccine virus [7]. The presence of homologous serum HAI antibodies correlated with protection against virus shedding. However, some vaccinated children without detectable homologous serum HAI antibodies were still protected against virus shedding. These findings indicate that serum HAI is not necessarily predictive of FluMist-induced immunity and suggest that FluMist vaccinees without detectable cross-reactive serum HAI antibodies may be protected against antigenic variants
Ferrets have been widely used to characterize antigenicity of influenza viruses and predict antigenic variants among influenza viruses [14]. On the basis of ferret serum HAI data, A/Texas/36/91 does not differ antigenically from A/Shenzhen/227/95, A/Shenzhen/227/95 differs antigenically from A/Beijing/262/95, and A/Beijing/262/95 does not differ antigenically from A/New Caledonia/20/99 [11, 14]. Human serum HAI data from the present study has shown that ferret cross-reactive HAI data could not precisely predict cross-reactive antibody responses to FluMist in immunologically unprimed children
It has been well documented that the HA1 domain mutates more frequently than does the HA2 domain and that the HA1 domain plays a major role in the process of natural selection [3]. The present study has shown that the differences in human HAI antibody cross-reactivity seem to be related, in part, to the number of amino acid differences on the HA1 domain rather than the HA protein. On the basis of abundant epidemiological and molecular data, Wilson and Cox [3] proposed that a drift variant of epidemiological importance usually contains ⩾4 aa changes located on ⩾2 of the suggested antigenic sites on the HA1 domain. The present study has further shown that amino acid changes at the suggested antigenic sites could be related, in part, to the magnitude of cross-reactive antibody responses to influenza vaccination in immunologically unprimed children. Only 5 pair-wise comparisons among 4 H1N1 influenza viruses were available in the present study. More human cross-reactive antibody data are needed for better understanding the correlation between antibody cross-reactivity and amino acid differences, in the HA protein
In conclusion, in immunologically unprimed children, live attenuated influenza vaccine induced variable cross-reactive antibody response against heterologous H1N1 viruses. The differences in human HAI antibody cross-reactivity seems to be related, in part, to the number of amino acid differences on the HA1 domain and on the suggested antigenic sites, but not to the number of amino acid differences on the HA protein
Acknowledgments
We thank Terry Nolan (University of Melbourne, Australia); Keith Reisinger (Primary Physicians Research, Pittsburgh, PA); Kenneth Zangwill (University of California, Los Angeles Center for Vaccine Research); the study coordinators and research personnel at the 3 clinical study sites (University of Melbourne, Australia; University of California, Los Angeles; and Primary Physicians Research); Marilyn August, Iksung Cho, Julie Cordova, Denise Dawson, Sharon Mathie, Paul Mendelman, and Jackie Zhao (study personnel at MedImmune); and Kathy Coelingh, Hong Jin, George Kemble, Paul Mendelman, Richard Spaete, and Robert Walker (MedImmune), for their review and discussion
