H3N2 Mismatch of 2014-15 Northern Hemisphere Influenza Vaccines and Head-to-head Comparison between Human and Ferret Antisera derived Antigenic Maps

Abstract

The poor performance of 2014-15 Northern Hemisphere (NH) influenza vaccines was attributed to mismatched H3N2 component with circulating epidemic strains. Using human serum samples collected from 2009-10, 2010-11 and 2014-15 NH influenza vaccine trials, we assessed their cross-reactive hemagglutination inhibition (HAI) antibody responses against recent H3 epidemic isolates. All three populations (children, adults, and older adults) vaccinated with the 2014-15 NH egg- or cell-based vaccine, showed >50% reduction in HAI post-vaccination geometric mean titers against epidemic H3 isolates from those against egg-grown H3 vaccine strain A/Texas/50/2012 (TX/12e). The 2014-15 NH vaccines, regardless of production type, failed to further extend HAI cross-reactivity against H3 epidemic strains from previous seasonal vaccines. Head-to-head comparison between ferret and human antisera derived antigenic maps revealed different antigenic patterns among representative egg- and cell-grown H3 viruses characterized. Molecular modeling indicated that the mutations of epidemic H3 strains were mainly located in antibody-binding sites A and B as compared with TX/12e. To improve vaccine strain selection, human serologic testing on vaccination-induced cross-reactivity need be emphasized along with virus antigenic characterization by ferret model.

Figure 1. Cross-reactive hemagglutination inhibition (HAI) responses against the latest epidemic influenza A (H3) isolates in humans vaccinated with the 2014–15 Northern Hemisphere (NH) seasonal influenza vaccines.

Sera were collected from healthy children (6 months to <9 years), adults (18 to <65 years), and older adults (≥65 years), vaccinated with 2014–15 NH egg- or cell-based vaccine. Post-vaccination (post-vac) HAI titers against the H3 vaccine prototype virus and representative isolates of the latest H3N2 epidemic strains were determined by using 1% guinea pig erythrocytes. There were 30 post-vac sera each from adult and older adult populations and 52 post-vac sera from children administered 2014–15 NH egg-based vaccine. There were also 24 post-vac sera each from adult and older adult populations administered 2014–15 NH cell-based vaccine. The H3 strains in the testing panel were as follows: egg-grown A/Texas/50/2012 (TX/12e), A/Switzerland/9715293/2013 (SWZ/13e), A/Palau/6759/2014 (PL/14e), A/North Carolina/13/2014 (NC/14e), and cell-grown A/Texas/50/2012 (TX/12c), A/Switzerland/9715293/2013 (SWZ/13c), A/North Carolina/13/2014 (NC/14c), and A/Michigan/15/2014 (MI/14c). *Indicates the H3 prototype virus of 2014–15 NH egg-based vaccine. Cross-reactive post-vac geometric mean titers (GMTs) against testing H3 viruses are expressed as % of TX/12e post-vac GMT in the bar graphs. Numbers shown at the bottom of each bar graph are individual post-vac GMTs and GMT ratios relative to TX/12e. Red dashed horizontal line indicates 50% of TX/12e-specific GMT.

Figure 2. Cross-reactive hemagglutination inhibition (HAI) responses of 2009–10, 2010–11, and 2014–15 Northern Hemisphere (NH) seasonal influenza vaccines against epidemic influenza A (H3) viruses.

Serum samples were collected from healthy adults immunized with 2009–10, 2010–11, and 2014–15 NH egg-based or cell-based vaccine. Post-vaccination (post-vac) HAI titers against the H3 vaccine prototype virus and representative isolates of the latest H3N2 epidemic strains were determined by using 1% guinea pig erythrocytes. There were 24 post-vac sera each from adults vaccinated with 2009–10 and 2010–11 NH egg-based vaccines and 2014–15 NH cell-based vaccine. There were also 30 post-vac sera from adults vaccinated with 2014–15 NH egg-based vaccine. The H3 strains in the testing panel included egg-grown A/Texas/50/2012 (TX/12e), A/Switzerland/9715293/2013 (SWZ/13e), A/Palau/6759/2014 (PL/14e), A/North Carolina/13/2014 (NC/14e), and cell-grown A/Texas/50/2012 (TX/12c), A/Switzerland/9715293/2013 (SWZ/13c), A/North Carolina/13/(NC/14c), and A/Michigan/15/2014 (MI/14c). The proportions of subjects with post-vac HAI titer of ≥40, ≥80 and ≥160 were plotted. *Indicates the H3 prototype virus of 2014–15 NH egg-based vaccine. Dotted horizontal line indicates 50% achievement. ND: not determined due to limited volumes of sera.

Figure 3. Antigenic characterization of influenza A (H3N2) viruses by cartography.

Antigenic maps were constructed on the basis of human hemagglutination inhibition (HAI) or ferret HAI data using AntigenMap (http://sysbio.cvm.msstate.edu/AntigenMap). An HAI titer of 10 was set as the cutoff for negative reaction in the HAI assay. Each entry in the HAI table was normalized by the maximum value from individual serum samples. Noise in the HAI data was minimized by implementing low-rank matrix completion. A two-dimensional map with multidimensional scaling was used to reflect antigenic distances among influenza A (H3) viruses. Each gridline (horizontal and vertical) is one antigenic unit distance corresponding to a 2-fold difference in HAI titers. Dots indicate egg-grown strains A/Uruguay/716/2007 (NYMCX175C) (URY/07e), A/Perth/16/2009 (PE/09e), A/Victoria/361/2011 (VIC/11e), A/Texas/50/2012 (TX/12e), A/Costa Rica/4700/2013 (CRI/13e), A/Utah/07/2013 (UT/13e), A/Switzerland/9715293/2013 (SWZ/13e), A/Palau/6759/2014 (PL/14e), and A/North Carolina/13/2014 (NC/14e). Triangles indicate cell-grown strains A/Victoria/361/2011 (VIC/11c), A/Texas/50/2012 (TX/12c), A/Costa Rica/4700/2013 (CRI/13c), A/Utah/07/2013 (UT/13c), A/Switzerland/9715293/2013 (SWZ/13c), A/North Carolina/13/2014 (NC/14c), and A/Michigan/15/2014 (MI/14c). Red indicates H3 vaccine prototype viruses, and underlining indicates prototype strains TX/12e for 2014–15 NH vaccines and SWZ/13e for 2015–16 NH vaccines. (Panel A), ferret post-infection sera–derived map. The post-vaccination sera from adults who had pre-vaccination HAI titers of <40 against all H1, H3, and B viruses tested in the study, defined as H1/H3/B-unprimed, were selected to construct human serology–based antigenic maps. There were 23 H1/H3/B-unprimed adult sera from 2009–10, 2010–11, and 2014–15 NH egg-based vaccine trials (Panel B), 8 H1/H3/B-unprimed adult sera from 2014–15 NH cell-based vaccine trial (Panel C), and a total of 31 samples from the combined trials with egg- and cell-based vaccinations (Panel D). the numbers of mutations in antibody-binding sites (ABS) (A–E) and antigenic distances relative to NC/14c shown in panel (A–D).

Figure 4. Molecular modeling of influenza A (H3N2) viruses.

The three-dimensional structures of hemagglutinin 1 (HA1) of egg-grown A/Texas/50/2012 (TX/12e) and A/Switzerland/9715293/2013 (SWZ/13e) were constructed based on the template PDB ID 4WE8 by homology modeling using MODELLER. Antibody-binding sites (ABS) and receptor-binding site (RBS) were adapted from previous annotations. The changes between TX/12e and SWZ/13e were highlighted at the structural superposition. (Panel A) shows the locations of 5 annotated ABS (A–E) on H3 HA1; (Panel B) highlighted mutations at the ABS and RBS among all H3N2 viruses in the testing panel; (Panel C) shows structural locations of amino acid differences on ABS between TX/12e and SWZ/12e. The conserved residues are gray-outlined boxes; changes between TX/12e and SWZ/13e are in boxes outlined with magenta-colored dashes. Egg-grown viruses are A/Uruguay/716/2007 (NYMCX175C) (URY/07e), A/Perth/16/2009 (PE/09e), A/Victoria/361/2011 (VIC/11e), A/Costa Rica/4700/2013 (CRI/13e), A/Utah/07/2013 (UT/13e), A/Palau/6759/2014 (PL/14e), and A/North Carolina/13/2014 (NC/14e) in addition to TX/12e and SWZ/13e. Cell-grown strains are A/Victoria/361/2011 (VIC/11c), A/Texas/50/2012 (TX/12c), A/Costa Rica/4700/2013 (CRI/13c), A/Utah/07/2013 (UT/13c), A/Switzerland/9715293/2013 (SWZ/13c), A/North Carolina/13/2014 (NC/14c), and A/Michigan/15/2014 (MI/14c).