Influenza viruses cause annual epidemics and occasional pandemics of acute respiratory disease in human populations leading to significant health threats worldwide [1] . Vaccination is the primary intervention method for preventing clinical disease and associated complications [2] [3] . The virus is capable of evading immune recognition through continual antigenic drift of its surface glycoproteins, hemagglutinin (HA) and neuraminidase (NA), resulting in a loss of immunity and vaccine mismatch. As a consequence, monitoring antigenic and genetic variations of circulating influenza viruses are crucial for the selection of annual strains [4] .

While two types of influenza A viruses (H3N2 and H1N1) have been in circulation since 1968 and 1977 respectively, two distinct evolutionary lines of the influenza B virus represented in human populations by the reference strains B/Victoria/2/87 and B/Yamagata/16/88 have gained attention since 2001/2002 [5] . Influenza virological surveillance evidenced the co-circulation of the two lineages of the influenza B virus during the same season in many parts of the world, including Europe, Italy and Taiwan [6] -[8] . Since Asia has been considered the gene pool for influenza viruses, the Taiwan Centers for Disease Control (Taiwan-CDC) have been coordinating a laboratory-based virological surveillance network since 2000 to provide updated information about viral characteristics and activities [9] [10] . After the co-circulation of B/Victoria/2/87- and B/Yamagata/16/88-like viruses in the 2004-2005 influenza season in Taiwan, both lineages circulated with alternating presence, and new genetic clades with amino acid substitutions appeared [11] [12] .

Current trivalent inactivated influenza virus vaccines (TIV) contain only a single influenza B virus component, either a Yamagata-like (influenza season 2004/2005 and 2005/2006) or a Victoria-like (influenza season 2003/ 2004, 2006/2007 and 2007/2008) strain [13] . The selection was based on epidemiological data on the geographic advance of infections, and antigenic and genetic results of recently and most frequently isolated B viruses. These two lineages are distinct antigenically, raising the concern that re-immunization might be required when antigenically altered strains appear [14] -[16] . The current recommendation for TIV vaccination in previously unvaccinated children under the age of 9 years is to receive 2 doses of TIV when vaccinated for the first time and 1 dose annually thereafter. This recommendation assumes effective prime-boosting across related antigens of the 2 influenza A/subtypes, but the declining number of children receiving the second dose 4 weeks later after the first dose makes it logistically challenging.

Since 2008, the Advisory Committee on Immunization Practices (ACIP) recommends annual influenza vaccination for all children of 6 months to 18 years, including all school-aged children [17] . High vaccination coverage of school-aged children is an important public health objective because of the direct benefits to immunized children and the potential indirect protection by reducing student and teacher absenteeism, allowing schools to better accomplish their primary mission and reducing morbidity/mortality for the broader community [18] -[20] . In Taiwan, the national free pediatric immunization programme of TIV has been implemented since 2007, and all children under the age of 9 years from the elementary schools will receive one dose of TIV annually. So far, limited study has been conducted to understand the immunogenicity after receiving different lineages of B antigens in school-aged children less than the age of 9 years―the group considered the primary community spreader of influenza viruses [18] [21] . Most among them are not naïve to influenza viruses and have antibodies against B of both lineages [21] [22] . The current non-randomized sero-epidemiological study conducting during 2008-2010 influenza seasons provided a good opportunity to evaluate the immunological response in school- aged children who received two antigenic-distinct influenza B virus lineages included in trivalent inactivated influenza vaccine (TIV) in two consecutive seasons.

In order to determine whether immunization with a B viral antigen of one lineage is sufficient to produce antibodies with cross-reactivity against both lineages, we analyzed the haemagglutination inhibition (HI) antibody response of school-aged children who received different B components of TIV in the 2008/09 and 2009/10 influenza seasons. Immunogenicity was measured for local strains of influenza B viruses of the B/Victoria lineage (B/Taiwan/5908/2009) and of the B/Yamagata lineage (B/Taiwan/29/2008) circulating in the community before and after vaccination. In the current study, we aimed to discover the immunological responses of children immunized with B viral antigens of one lineage, and subsequently with the other in the following influenza season, as well as the antibody protection during winter epidemics in Taiwan.

In the current study, we followed school-aged children with a well-defined vaccination history to assess subsequent immunological responses to TIV containing the alternate B/Lineage. Our results revealed that children under the age of 9 years in years receiving TIV showed higher HI titres and sero-protection rates of homologous strains from the same lineage than those of heterologous strains from the other lineage while no differences were observed for children in the older age group (≥9 years old). Since 2008, the ACIP recommends annual influenza vaccination for all children of 6 months to 18 years old, including all school-aged children [17] . The results shown in our study provided important information for countries which are planning, or currently implementing the TIV immunization programme in school-aged children.

Influenza B viruses currently circulating worldwide are of two distinct evolutionary hemagglutinin lineages, B/Yamagata and B/Victoria [13] [27] [28] . The distinct antigenicity between B/Yamagata and B/Victoria lineages has also been established based on antisera produced in ferrets, which were found to be highly specific against the homologous lineage with little or no cross-reactivity, suggesting a remarkable antigenic difference between these two lineages [7] [29] [30] . The results were further supported by the lack of cross-reactivity and neutralization activity observed in immunologically naïve volunteers [16] . Since the current trivalent influenza vaccines contain only one B vaccine antigen, the mismatch between vaccine and wild-type antigens and the concurrent appearance of influenza B virus of both lineages during the same winter season are of major concern [14] . The question has been raised as to whether both lineages of B antigen should be represented in the annual influenza vaccine formulation. In this study, we examined whether immunization with a B viral antigen of one lineage is sufficient for producing antibodies with cross-reactivity against both lineages of B viruses. Our results confirmed that children receiving TIV showed higher HI titres and sero-protection rates of homologous strains from the same lineage than those of heterologous strains from the other lineage―either post-vaccination or post-influenza season. The need for the quadrivalent TIV requires further consideration.

Influenza B virus infections and immunizations often produce relatively low HI titres. This effect is not only the result of strain variability but also the method of measuring HI antibodies. Previous studies suggested a more sensitive but less specific assay when applying antigens from influenza B viruses after disruption by treatment with an organic solvent such as ether [31] [32] . The HI titres obtained with antigens after ether or detergent disruption correlate extremely well with neutralization antibody titres, and the HI titres at dilutions exceeding 1:32 after immunization with detergent-disrupted antigens can be associated with protection against illness and infection [31] [33] . In our study, sero-protection rates in subjects after immunization in 2008-09 and 2009-10 sea- sons were shown to be only 69% and 64% by measuring HI with non-disrupting B viruses as depicted in Table 2. Since the presence of a HI antibody is the best surrogate measure of protection and the desired outcome of immunization [34] , the antibody responses detected by HI with non-detergent-disrupted antigens suggest that the sero-protection rates against influenza B virus strains can be under-estimated in our study subjects.

The lower antibody response to the heterologous B/Yamagata lineage antigen among the vaccinated subjects during 2009-10 season as shown above could be related to different priming histories as suggested by previous

studies [16] [35] [36] . Our study also attempted to understand whether priming with one dose of a specific influenza B/Lineage provided children with an adequate boost by a single dose of the other lineage in the consecutive year; the subjects were further divided into four groups: 1) receiving both lineages of B component of TIV in 2008/09 and 2009/10 influenza seasons; 2) receiving TIV containing B/Yamagata lineage only in 2008/09 but not receiving TIV containing B/Victoria lineage in 2009/10; 3) not receiving TIV containing B/Yamagata lineage only in 2008/09 but receiving TIV containing B/Victoria lineage in 2009/10; 4) not receiving TIV in both 2009/09 and 2009/10 influenza seasons. Although the sample size was small, the results revealed that receiving vaccination of both lineages in two consecutive years showed a significantly higher GMT of HI (107.67; 95% CI: 72.5 - 144.9) against the vaccine strain post vaccination compared with children having no vaccination (35.54; 95% CI: 24.9 - 51.5) for children under the age of 9 years, with statistical significance (p < 0.05) (data not shown). The significant differences between previous and the current studies included age of the study subjects and a 2-dose schedule was administered and repeated during the season that influenza B/Lineage changed. Evaluation of cohorts of children with a defined history of prior influenza B/Lineage prime-boost experiences across additional seasons will therefore be important for confirming these results and understanding their clinical relevance.

The main limitation of our study relates to the small sample sizes among subsets of the participants receiving alternating TIV, which reduced both the statistical power and precision. Furthermore, the sample bias could not be excluded since the current study design was non-randomized and different exposure experiences between vaccinated and unvaccinated groups might confound the results. Although influenza B was not a major contributor to virus circulation in Taiwan during the span of these sequential studies as shown in Figure 1, low level transmission in the community was still possible even though the HI titres in post-season sera were similar between vaccinated and non-vaccinated groups. Particularly, children are usually asymptomatic once infected by influenza viruses, and are therefore considered the primary community spreaders of the viruses [19] [37] [38] . At baseline antibody titres before vaccination during 2009-10 season (Table 2), a lower level of GMT and sero- protection rate for children under the age of 9 years in vaccinated group than that in un-vaccinated group was observed, which suggested asymptomatic infections among un-vaccinated subjects. Finally, we also cannot rule out the observations in this study could be particular to the specific influenza B strains and further evaluation in additional seasons is therefore warranted.

The burden of influenza B infection falls mostly on children although the overall mortality rates associated with influenza B infection were less than those of A/H3N2, but greater than those of A/H1N1 [39] -[41] . In this study, the significantly higher HI titres and sero-protection rates of homologous strains from the same lineage than those of heterologous strains from the other lineage were showed among children under the age of 9 years in years after receiving TIV, while no differences were observed for children in the older age group (>9 years old). Our finding provides both practical and scientific implications for the evaluation of influenza immunization programmes in school-aged children with more diverse exposures and antibody population of immunologic experience to both influenza B lineages.

This study was supported by National Science Council of Taiwan (NSC 97-2118-M-039-004) and China Medical University, Taiwan (CMU 97 323). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. We declare that we have no conflict of interest.

None.

This study was supported by National Science Council of Taiwan (NSC 97-2118-M-039-004) and China Medical University, Taiwan (CMU 97 323). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.