The assessment of vaccine efficacy is generally based on the European regulatory requirements for annual licensing of influenza vaccine, i.e., >40% seroconversion (4-fold increase of HI titer), >70% seroprotection (HI titer, ≥40), and >2.5 GMT increase of HI titer between pre- and postvaccination sera (11
). These criteria mainly cover the immunogenicity of vaccines against the homologous vaccine viruses, but not the cross-reactivity or protective efficacy of serum antibodies against circulating epidemic viruses. The most important criterion in the assessment of vaccine efficacy is the evaluation of whether the vaccine can produce high titers of broad-spectrum HI and NT antibodies against circulating epidemic viruses, including antigenically variant viruses. In Japan, the licensing process for the clinical use of influenza vaccine is based on the antigen protein contents; the requirements include a minimum content of HA protein (greater than 15 μg/dose), a maximum total viral protein content (less than 240 μg/ml), and some safety assessments of the vaccine. However, neither immunogenicity testing nor preclinical studies of vaccines in volunteers are required for vaccine licensing in Japan. Consequently, low-immunogenic B vaccine, which cannot satisfy any of the CHMP criteria required for use in European countries (), can be distributed in Japan and has been used in clinics. Moreover, we have recently observed by HI tests that the ferret antisera immunized by X-187 (A/H3N2) and egg-adapted B/Brisbane/60/2008 vaccine strains exhibited greatly decreased cross-reactivity to recent circulating epidemic viruses ( and ), and we anticipated that the A/H3N2 and B vaccines for the 2010-2011 season could be less effective against circulating viruses. Although studies have assessed immunogenicity against vaccine viruses after the administration of vaccines (3
), studies assessing cross-reactivity of postvaccinated human serum antibodies against epidemic viruses are quite limited (2
). Consequently, in the present study we assessed the cross-reactivity of serum HI and NT antibodies of adult, middle-aged, and elderly individuals who received the 2010-2011 season vaccine.
During egg adaptation of the X-187 vaccine virus, an amino acid substitution, Ser228Thr, in the globular head of the HA molecule occurred; this substitution is evident in a comparison to the wild-type HA of A/Victoria/210/2009 virus. Amino acid position 228 of H3 subtypes of influenza HA has been implicated as a critical residue for receptor specificity and host range restriction (25
). Three-dimensional modeling analysis showed that the substitution is located at a site where it could cause critical structural changes in the receptor-binding surface due to changes in the amino acid side chain (). Such structurally changed X-187 HA protein would induce antibodies with reduced reactivity to the original wild-type A/Victoria/210/2009 virus and antigenically similar prototype viruses A/Perth/16/2009 and A/Niigata/403/2009 as well as many epidemic viruses (). Human serum antibodies induced by the X-187 vaccine also showed decreased HI titers to the majority of test viruses in all age groups (). Many studies have shown that susceptibility to influenza virus infection is inversely related to the initial titer of serum IgG HI antibody (38
). Most results indicate that, following immunization with inactivated virus vaccines, HI antibody titers of approximately 1:30 to 1:40 represent the 50% protective level of antibody (3
). Similarly, the NT antibodies induced in human sera exhibited narrow cross-reactivity against epidemic A/H3N2 viruses in all age groups, since half of the test viruses showed greater than 50% reduction of GMT from the GMT of the vaccine virus (). These results suggest that the X-187 vaccine possesses a small inhibitory effect against circulating epidemic A/H3N2 viruses, resulting in the decreased protective efficacy of the A/H3N2 vaccine. However, three MDCK-grown viruses (A/Kobe/357/2010, A/Akita/10/2010, and A/Fukuoka-C/35/2010) showed higher NT antibody titers than did the homologous X-187 vaccine virus. These three viruses were also relatively better inhibited in HI tests with all reference ferret antisera raised to vaccine-like viruses (). Although these three viruses did not possess the consensus amino acid substitutions in the HA protein that are suggestive of broad reactors, other epitopes could be recognized by NT antibody and result in elevated reactivity in the NT test.
Generally, the antibody response to influenza B vaccine is significantly lower than the antibody response to influenza A vaccine (7
). Our routine serology studies of human sera for vaccine strain selection by WHO have also indicated the low immunogenicity of influenza B vaccine (data not shown). A recent report by Xie et al. (71
) indicated that the B/Brisbane/60/2008 vaccine used for the 2009-2010 influenza season was less immunogenic than the two influenza A vaccine components. By HI tests in the present study, human serum antibodies induced by B/Brisbane/60/2008 vaccine included in the vaccine composition of the 2009-2010 season showed overall very low GMTs against all test B viruses, including homologous vaccine virus, particularly for middle-aged and elderly individuals, and the titers were as low as the marginal level (). Moreover, no test viruses, including vaccine virus, exceeded the GMT of 40, indicating a poor protective efficacy of the vaccine. Similarly, for the adult group (mean age, 38.1 years) and elderly group (mean age, 82.3 years), serum antibodies induced by B/Brisbane/60/2008 vaccine included in the vaccine composition of the 2010-2011 season revealed remarkably low immunogenicity and low GMTs to test viruses isolated in the 2010-2011 season (data not shown). We therefore concluded that the efficacy of the influenza B vaccine should not be assessed by HI test.
HI antibody substantially recognizes the receptor-binding region in the HA1 molecule and inhibits virus binding to the receptor of host cells, while NT antibody includes not only HI antibody but also the antibodies recognizing other regions outside antigenic sites in the HA1 molecule (23
) and plays a substantial role in the protective efficacy of vaccines. Consequently, the proper assessment of vaccine efficacy for human use should be done by NT test. NT tests of human serum antibodies showed a sufficient level of GMTs against homologous vaccine virus like the GMTs of H3 vaccine (), so that the cross-reactivity of the serum antibody induced by B/Brisbane/60/2008 vaccine was able to be assessed by NT test. The overall reactivity of the serum antibodies in all age groups was high with egg-grown B viruses, whereas the reactivity against MDCK-grown B viruses, except for the original wild-type virus B/Brisbane/60/2008, was significantly lower (). In particular, a marked difference was seen between the pair of egg- and MDCK-grown viruses of B/Hiroshima/9/2010 and B/Mie/6/2010. Our results were consistent with the previous report that egg-grown influenza B vaccine virus induced narrow-spectrum serum antibodies that were more specific to homologous egg-grown viruses (1
). From the assessment by NT test, it was strongly suggested that the efficacy of the B/Brisbane/60/2008 vaccine for the 2009-2010 and 2010-2011 seasons was significantly decreased.
Previous studies have demonstrated that egg passage of B/Victoria/2/87 lineage viruses is associated with the loss of an N-glycosylation site (Asn-X-Thr/Ser) at residues 197 to 199 of the HA protein in mammalian-cell isolates, resulting in the antigenic change from the original wild-type virus (55
). Our amino acid alignment of the egg-adapted B/Brisbane/60/2008 vaccine strain and all other egg-grown isolates used in the present study confirmed the change from Asn or an Asn/Ser mixture to Ser at residue 197 or from Thr or a Thr/Ile mixture to Ile at residue 199, resulting in the loss of the N-glycosylation site. Between the HAs of egg- and MDCK-grown B/Brisbane/60/2008 viruses, only one amino acid at residue 197 was different, and the difference was reflected in the absence or presence of N-glycan in antigenic site B (). Attachment of N-glycan at the antigenic site of HA protein could hide the epitope and affect immune recognition by the antibody induced by non-N-glycosylated virus. In fact, the reactivity of egg-grown B/Brisbane/60/2008 ferret antiserum against the glycosylated MDCK-grown B/Brisbane/60/2008 virus was dramatically affected (). However, such a striking change of B/Brisbane/60/2008 virus was not crucial for NT antibody, since MDCK-grown B/Brisbane/60/2008 virus showed high cross-reactivity in the NT test (). It is likely that the NT antibodies react with another homologous part of the HA protein in MDCK-grown B/Brisbane/60/2008 virus to restore cross-reactivity. On the other hand, other MDCK-grown test viruses possessed 3 to 4 amino acid differences in the HA1 region from B/Brisbane/60/2008 vaccine virus together with the retained N-glycan at residues 197 to 199. This heterogeneity of HA protein might not be covered by the NT serum antibodies induced by B/Brisbane/60/2008 vaccine. Several studies point to the importance of the second most abundant surface influenza glycoprotein neuraminidase (NA) in conferring cross-reactive immunity (10
). Anti-N2 serum antibodies provide protection against antigenically distinct viruses belonging to the same subtype (10
). However, the NA gene sequences of A/Victoria/210/2009 and X-187 vaccine virus and MDCK-grown and egg-adapted B/Brisbane/60/2008 viruses were identical, respectively. Therefore, we conclude that anti-NA antibodies had no effect on the different reactivities of these viruses.
Egg-grown virus vaccine has been suggested to produce less cross-reactive antibody and inferior protective efficacy compared to MDCK-grown virus vaccine (1
). Unfortunately, this was the case for the A/H3N2 X-187 vaccine virus as assessed by our serology study. In addition, recent A/H3N2 viruses have become extremely difficult to isolate in eggs because of the change in receptor-binding properties related to amino acid substitutions at residue 156, 186, 194, or 196 (16
). Such a viral change causes difficulties in the development of a high-growth reassortant vaccine virus and in the timely supply of vaccine seed virus to manufacturers so that the vaccine can be produced before the annual outbreak of influenza. Moreover, most B viruses of both B/Victoria/2/87 lineage and B/Yamagata/16/88 lineage lose the N-glycosylation site in their important antigenic site by egg adaptation (55
). Although an amino acid substitution from Gly to Arg at residue 141 in the HA protein of B virus can stabilize the glycosylation site without affecting antigenicity (9
), the production of influenza vaccine in eggs has encountered difficult problems, and these problems will continue to arise in every influenza season. Many studies of influenza vaccine efficacy have been conducted on the basis of virological and statistical studies (13
). However, the possibility that altered antigenicity caused by egg adaptation during the vaccine production process has harmed the efficacy of influenza vaccine (as shown in our present study) has not been well explored. Ideally, any new influenza vaccine for humans should be evaluated for potential alteration of immunogenicity prior to use, as described in this study. However, manufacturers may not have sufficient time to conduct exhaustive clinical trials. Therefore, it is important that HI and NT tests using ferret antisera and HA gene sequencing of vaccine viruses be performed to (i) confirm that the vaccines have not suffered any modifications that decrease their efficacy and (ii) confirm that the vaccine viruses are comparable to the circulating viruses or viruses grown in MDCK cells. Consequently, to solve the problems of egg adaptation, the development and supply of influenza vaccines grown in MDCK cells or other surrogate cells should be globally propelled to replace current egg-grown virus vaccines.