Antibody Response and Protection After Receipt of Inactivated Influenza Vaccine: A Systematic Review

We searched Medline, Embase, and Cochrane Library databases to include literature published up to April 2, 2019, without limits on the year or language of publication. Search key terms included “influenza vaccine(s)”; “infant,” “child,” or “pediatrics”; and “dose” or “two-doses.” The search was limited to articles with available abstracts and, we excluded animal studies, commentaries, editorials, letters, and conference abstracts or articles. We also searched references from relevant articles to identify studies not captured in our database search.

The intervention(s) or exposure(s) of interest included trivalent or quadrivalent IIV containing 2 influenza A viruses (H1N1 and H3N2 subtypes) and either 1 or 2 influenza B viruses (Victoria and/or Yamagata lineage). Eligible studies included randomized controlled trials (RCTs) and observational studies (case-control and cohort studies) in which researchers compared the effectiveness or immunogenicity of 2 doses versus 1 dose of influenza vaccine among healthy children aged 6 months through 8 years. Immunogenicity studies were restricted to those in which researchers measured the HAI antibody response after each vaccine dose in the same group of children. Studies with broader age groups were included if stratified data were presented on children aged 6 months–8 years. We excluded studies that were non-English, focused exclusively on participants ≥9 years, restricted to special populations (eg, immunosuppressed or asthma), interim, focused on pandemic H1N1 vaccines, or had <30 participants. Studies were also excluded if they were focused exclusively on vaccines that were not licensed for use in all children aged 6 months through 8 years in the United States, including live attenuated influenza vaccines (LAIVs), which are currently licensed for persons aged 2 to 49 years, and adjuvanted vaccines, currently licensed for persons aged ≥65 years.

Two authors (D.J.W. and F.S.D.) assessed the risk of bias for the studies included in this systematic review using 3 quality assessment tools created by the National Heart, Lung, and Blood Institute. Observational studies were assessed by using the Quality Assessment of Case-Control Studies tool (Supplemental Table 6), cohort studies were assessed by using the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies (Supplemental Table 7), and RCTs were assessed by using the Quality Assessment of Controlled Intervention tool (Supplemental Table 8).⁹  Studies were rated as good, fair, or poor, depending on the level of risk of bias. Both authors compared their risk of bias assessments and resolved all discrepancies through discussion.

We included observational studies fulfilling search criteria in which researchers estimated vaccine effectiveness (VE) among fully vaccinated (FV) and partially vaccinated (PV) children, compared with unvaccinated children. VE was assessed against either laboratory-confirmed influenza virus infection or medically attended influenza-like illness (ILI) or acute respiratory infection (ARI) that did not require laboratory confirmation. Laboratory-confirmed cases were determined by viral culture or reverse transcription polymerase chain reaction. In the studies, researchers used either cohort or test-negative case-control designs to assess VE. In the test-negative design, the exposure of interest is influenza vaccination, and case-patients and controls are defined as persons with respiratory illness with and without laboratory-confirmed influenza, respectively.¹⁰  This design is typically used to assess VE against medically attended influenza illness and is less susceptible to confounding from health care seeking behaviors.¹⁰ 

Vaccination status was typically defined by ACIP criteria for each season. FV was defined as 2 doses received at least 4 weeks apart in the study season; depending on the season, an alternative definition for FV was at least 1 dose in a previous season and 1 dose at least 14 days before the outcome. PV was defined as those children who were not vaccinated in any previous season and received 1 dose in the study season at least 14 days before the outcome or 2 doses in the study season but with the second dose administered <14 days before the outcome. Children were considered unvaccinated if they received no influenza doses in the study season.

To evaluate the immunogenicity of 2 vs 1 dose of IIV, we examined serum HAI antibody titers against influenza A(H1N1), influenza A(H3N2), and influenza B-lineage vaccine antigens by using geometric mean titers (GMTs) after 1 and 2 doses of IIV in the same group of children. In all studies, researchers administered the second dose of IIV ∼4 weeks after the first. Serum was collected 3 to 4 weeks after each dose. When studies did not publish HAI GMTs, we attempted to contact the authors to obtain these data for analysis. When contact was not successful, we used GetData Graph Digitizer 2.26 software to estimate original (x and y) data from figures and graphs. To compare antibody titers after 1 and 2 doses, we calculated the geometric mean titer ratio (GMR). GMR was estimated as the log₂ of the postvaccination GMT after 2 doses divided by the log₂ of the postvaccination GMT after 1 dose. A GMR >1.0 indicates a higher GMT after the second dose. We calculated the mean-fold rise (MFR) from the baseline to postvaccination titers after the first dose and from the baseline to postvaccination titers after the second dose when baseline GMTs were available. MFR was estimated as the log₂ of the postvaccination GMT after 1 or 2 doses divided by the log₂ of the prevaccination baseline GMT. Seroconversion rates (SCRs) were identified after 1 and 2 doses, which was defined by either a prevaccination HAI titer <1:10 and postvaccination HAI titer ≥1:40 or a minimum fourfold rise in postvaccination HAI titer. We compared antibody response by age and baseline antibody titers. Undetectable baseline antibody titers were defined as either a titer <1:10^11,12  or <1:8.¹³ 

We identified 724 articles from the database searches and 10 more after searching relevant bibliographies (Fig 1). After duplicate articles were excluded (n = 7), 727 were screened. After screening titles and abstracts, 82 were chosen for full-text review. Reasons for exclusion included a failure to meet the inclusion criteria (n = 41), a sample size <30 (n = 7), a lack of age group specification (n = 4), no available English text (n = 3), and vaccine not being approved for clinical use (n = 1). In total, 26 studies met inclusion criteria. 10 were VE studies,^14–23  and 16 were immunogenicity trials (Table 1).^{11–13,24–36}  We included 1 study that measured HAI antibody titers in children from 6 to 9 years of age.³² 

A total of 8 of 10 studies used a test-negative design to estimate VE against medically attended, laboratory-confirmed influenza virus infection.^{15–18,20–23}  These studies measured VE against influenza-associated outpatient medical visits,^21,23  both outpatient and inpatient medically attended encounters^15,17,18,22  and influenza-associated hospitalizations,^16,20  In the 2 remaining studies, researchers estimated VE for preventing medically attended ILI and pneumonia and/or influenza medical visits using International Classification of Diseases, Ninth Revision, Clinical Modification, codes without laboratory confirmation of influenza.^14,19  Vaccine-status definitions were consistent across the majority of trials, with the exception of 1 study, in which researchers considered the spacing of 2 doses of influenza vaccine ≥14 days,²⁰  instead of 4 weeks. In the majority of the VE studies, researchers used either medical records or immunization registries to determine vaccination status.^{14–20,22,23}  Only 1 of the studies did not specify how authors determined vaccination status.²¹ 

VE against any influenza was generally higher among FV children than PV children (Table 2). Wide confidence intervals (CIs) were reported across studies. In many studies, researchers did not detect statistically significant VE among PV children.^{15,17,20–22}  Studies were typically underpowered to identify significant differences between PV and FV children, although point estimates were generally higher in FV children.^16,18,19,23  In 1 study, researchers did not find significant VE estimates for PV or FV children across the 2 study seasons when there were suboptimal matches between vaccine strains and circulating viruses.¹⁵  VE stratified by influenza subtype (ie, A[H1N1], A[H3N2], and B-lineage viruses) was also typically higher for FV children compared with PV children (Supplemental Table 9).^16,20,23  However, Thompson et al²³  observed a higher VE against A(H3N2) viruses among PV children (VE = 83; 95% CI: 60 to 93) than FV children (VE = 36; 95% CI: 6 to 56; P = .021), with a 2.7-fold (95% CI: 1.2 to 6.3) increased odds of being influenza A(H3N2)–positive if FV, compared with those PV. Buchan et al¹⁶  also found higher VE against A(H3N2) for PV children (VE = 70; 95% CI: 25 to 88), compared with FV children (VE = 53; 95% CI: 4 to 77); however, all other subanalyses showed higher VE in FV children. In this study, the effect of FV status on influenza subtype appeared the strongest against influenza B-lineage viruses (P = .03), with higher VE in FV children (VE = 58%; 95% CI: 28 to 75), compared with PV children (VE = 12%; 95% CI: −45 to 46).¹⁶ 

In the youngest age group, 6 to 23 months of age, VE estimates were consistently higher for FV children, compared with PV children (Table 2). None of the 7 studies that estimated VE in children 6 to 23 months of age showed significant VE for PV children in this age group.^{14–17,19,21,22}  Differences in VE by age were variable in children aged 6 to 23 months of age, compared with older children. Buchan et al¹⁶  identified higher VE for children 24 to 59 months of age compared with children 6 to 23 months of age (P = .012). In contrast, Shuler et al²¹  found higher VE among FV children 6 to 23 months of age (VE = 52%; 95% CI: 20 to 70), compared with PV children (VE = −7%; 95% CI: −280 to 10). However no significant differences were observed between PV (VE = 65%; 95% CI: 30 to 80) and FV (VE = 45%; 95% CI: 10 to 70) children aged 24 to 59 months.²¹  Similarly, Ritzwoller et al¹⁹  found infants aged 6 to 11 months were 4 to 5 times more likely to present with ILI, compared with older children aged 7 to 8 years. In contrast, in other studies, researchers found no VE differences across age groups in children <8 years.^17,20 

In 2 studies, researchers considered the impact of previous season influenza vaccination on current-season VE.^16,23  Buchan et al¹⁶  found similar VE estimates overall on the basis of vaccination in the current season only, compared with the 2 sequential seasons (66% vs 62%). However, there was variation across seasons. For example, in the 2010–2011 season, VE was higher when vaccinated in the current season only, compared with children vaccinated in current and previous season (83% vs 72%). In contrast, VE was 67% for children vaccinated in current and previous season and only 27% for children vaccinated in the current season during 2012–2013.¹⁶  Thompson et al²³  measured VE against influenza infection on the basis of priming or doses received in previous seasons. In the study, the researchers considered 4 potential definitions for priming and estimated VE for each combination of current season and priming definition.²³  Significant effect modification on current-season VE was seen only when priming was defined as ≥1 doses received in the previous season only (P < .1) and 2 doses of influenza vaccine received in the same previous season (P < .05).²³  Furthermore, VE point estimates were consistently higher for children who received 2 doses in a previous season (VE = 58% to 80%), compared with children who did not (VE = 33% to 42%).²³  Similarly, the odds of testing positive for A(H3N2) were 2.4 times (95% CI: 1.4 to 4.3) higher among unprimed children who received current-season vaccination, compared with children who received the current-season vaccination but were primed with 2 doses in a previous season.²³  With these findings, it is suggested that priming with 2 doses of influenza vaccination in young children may be more effective than 1 dose.

Immunogenicity results separated by vaccine strain are summarized in Tables 3–5. In 5 RCTs, researchers compared antibody responses between inactivated and adjuvanted influenza vaccines^{25,31,34,36,37} ; data are only included from the groups that received 2 doses of IIV. In the majority of RCTs, researchers administered a one-half dose (0.25 mL) for children aged <3 years and 0.5 mL dose for children aged ≥3 years. In 1 study, researchers administered a full 0.5 mL dose regardless of age,²⁵  and, in another study, researchers administered a 0.1 mL dose for children aged <1 year and 0.2 mL dose for children aged ≥1 year.³⁰  In 2 studies, researchers stratified data by age groups,^25,36  and, in 3 studies, researchers stratified results by baseline titers.^11–13  Of note, the 2 RCTs in which researchers stratified by baseline antibody titers had small numbers in some strata.^12,13 

Overall, 2 doses of IIV provided better antibody responses, compared with 1 dose for A(H1N1) (Table 3), A(H3N2) (Table 4), and B-lineage vaccine antigens (Table 5). GMRs ranged from 0.9 (95% CI: 0.8 to 1.1) to 8.7 (95% CI: 7.8 to 9.7) against A(H1N1) vaccine antigens, 0.7 (95% CI not reported) to 5.3 (95% CI: 4.7 to 6.0) against A(H3N2) vaccine antigens, and 0.8 (95% CI: 0.5 to 1.5) to 6.8 (95% CI: 6.0 to 7.8) against B-lineage vaccine antigens.^{11–13,24–36}  MFR or value increase of GMTs from baseline were typically higher after the second dose, compared with the first. The only exceptions were 2 studies in which researchers stratified GMTs by baseline antibody titers.^11,12  MFR calculations in these studies increased after each dose in the strata with undetectable baseline titers. In contrast, MFR changed minimally after doses 1 and 2 in the strata with higher baseline antibodies.

In general, higher GMRs were seen in younger children and groups with low prevaccination baseline titers. Lower GMRs were observed among older children and groups with higher baseline titers. For example, GMR calculations for one study in which researchers calculated GMTs stratified by age group revealed a GMR of 6.1 (95% CI: 4.4 to 8.4) for infants aged 6 to 11 months, 6.0 (95% CI: 3.9 to 9.3) for children aged 12 to 35 months, and 1.7 (95% CI: 1.0 to 2.9) for children aged 36 to 71 months for HAI titers against the A(H1N1) vaccine antigen, with a similar pattern for A(H3N2) and B-lineage vaccine antigens.²⁵  In addition, all 3 studies that stratified results by baseline antibody titers revealed an inverse relationship between GMR and baseline titers.

In 7 studies, researchers calculated SCR (Supplemental Table 10).^{12,25,28,31,32,34,36}  SCRs were heterogeneous across studies, ranging from 4% (95% CI: 0.10 to 19.6) to 70% (95% CI: 46 to 88) after 1 dose and 73% (95% CI: 52.2 to 88.4) to 98% (95% CI: 92.6 to 99.7) after 2 doses for A(H1N1); from 43% (95% CI: 33 to 53) to 89% (95% CI: 86 to 91) after 1 dose and 74% (95% CI: 62 to 85) to 100% (95% CI: 89 to 100) after 2 doses for A(H3N2); and from 2% (95% CI: 0.0 to 5.2) to 68% (95% CI: 58 to 77) after 1 dose and 19% (95% CI: 12 to 28) to 97% (95% CI: 91 to 99) after 2 doses for B-lineage antigens (Supplemental Table 10). In general, a greater proportion of children seroconverted after 2 doses of IIV, compared with 1 dose for all influenza strains. SCRs were higher for older children and those with detectable prevaccination titers (ie, titers ≥1:10 or ≥1:8).^12,25,32  Although both of these groups had higher SCRs after 1 dose, compared with younger children and those with undetectable baseline titers, all SCRs increased after the second dose, irrespective of age or baseline antibody titer status.

In 3 trials, researchers compared antibody responses after alternative dosing schedules, comparing a spring-fall 2-dose regimen to the standard dosing regimen within the same season.^26,35  The studies differed in that vaccine antigen composition for at least 1 of the 3 viruses were either similar or different during the 2 seasons.^26,35  When antigenically identical vaccines were administered, no significant differences in GMTs were observed. In contrast, antibody responses were lower in the spring-fall group, compared with the standard group when the antigens changed between the 2 seasons for A(H1N1) (57 vs 48), A(H3N2) (129 vs 115), and B-lineage vaccine antigens (28 vs 24).^26,35  When antigen composition was different across 2 seasons, differences in antibody response were observed for A(H3N2) (spring-fall group GMTs = 57 ± 4.1 versus standard group GMTs = 78 ± 3.7) and, to a greater extent, for B-lineage viruses (spring-fall group GMTs = 18 ± 2.4 versus standard group GMTs= 62 ± 2.5).³⁵ 

In another study, results were comparable for antibody response between groups for the unchanged A(H1N1) antigen (GMTs 75 vs 69).²⁷  The A(H3N2) GMTs were significantly higher in the group that received 2 doses across seasons (GMT = 156; 95% CI: 105 to 231), compared with the antigenically identical vaccine group (GMT = 54; 95% CI: 41 to 70).²⁷  However, the former group was older, and a higher percentage were alive during the previous influenza season when A(H3N2) circulated. In contrast, the antibody response to B-lineage vaccine antigens (antigen components differed greatly across seasons) was significantly lower after the second IIV dose for the group that received doses across seasons (GMT = 14; 95% CI: 11 to 17), compared with the group given identical vaccines in the same season (GMT = 49; 95% CI: 41 to 59).²⁷ 

In the vast body of evidence, it is indicated immunologically naïve children need 2 doses of influenza vaccine to induce a protective immune response.^{11,17,21,22,25,26,33–35}  Similar findings of 2 dose requirements for optimal priming in age groups that are immunologically naïve are observed against pandemic and novel influenza viruses.³⁸  Because influenza infections increase with age,^39,40  vaccinating children early with 2 doses improves immune response. In line with this evidence, the ACIP and World Health Organization recommend 2 doses of influenza vaccine before the start of influenza season for children 6 months through 8 years, during their first season of vaccination.^7,41 

In our systematic review, we identified 4 key findings. First, a full 2-dose influenza vaccine series in children receiving influenza vaccine for the first time provides optimal protection against influenza-related medical visits in the first season of vaccination. Higher effectiveness is observed among FV children, compared with PV children, and 2 doses of vaccine are more immunogenic than 1 dose. Second, young children aged <2 years, who have a less likelihood of having acquired their first influenza infection, may benefit the most from a second dose of influenza vaccine. Third, children with positive prevaccination titers and who are older (ie, more likely to be exposed to influenza naturally) may have a protective immune response after a single dose of influenza vaccine.⁴²  Finally, the timing between 2 doses of influenza vaccine or schedule (ie, administering doses across seasons) does not appear to negatively affect the immune response when antigen components are similar.

Vaccine effectiveness was consistent across a variety of study designs, with most studies showing higher point estimates among FV versus PV children. Whether partial vaccination or a single dose of influenza vaccine provides protection in a child’s first season of influenza vaccinations remains unclear, largely because of inadequate power to address that question. However, some studies did observe significant effectiveness against influenza for PV children.^16,18,19,23  Two studies found relatively higher effectiveness against A(H3N2) for PV children, compared with FV children, although sample sizes were small.^16,23  When considering the youngest age group of 6 to 23 months, effectiveness was consistently low for PV children, highlighting the importance of full vaccination in young children, who are less likely to have experienced natural infection.

We also documented increased antibody responses after 2 doses of IIV, compared with 1 dose, for children aged 6 months through 8 years. The GMR was typically >1 across all 3 vaccine antigens, indicating higher GMTs after the second dose. The benefits of a second dose are the greatest in children without a previous history of a primary infection and who are, thus, immunologically naïve. Consistent with this, in studies in this review, researchers did not identify consistent increases in antibody responses after the second dose among children with higher baseline titers^11–13  or those with a history of influenza vaccination.³⁰  Antibody responses were also higher among the youngest age group (children aged 6 to 23 months) and children with nondetectable baseline titers, indicating true immunologic naivety. With these findings, the importance of vaccinating children with 2 doses early in life before their first exposure to influenza to gain the optimal benefits of vaccination is highlighted. Vaccination during early childhood may also result in immune imprinting or the creation of immune memory that results in more robust responses to future exposures to antigenically similar influenza viruses.^40,43 

In the 2 studies in which researchers compared a spring-fall vaccination schedule with a standard vaccine schedule, researchers found similar antibody responses irrespective of the timing between doses when vaccine antigens were identical across doses.^26,35  However, decreased antibody responses were observed when antigen differences increased across doses.³⁵  This suggests that primary vaccine doses may provide antibody responses primarily to homotypic viruses and subsequent doses with drifted antigens may not adequately boost primed doses in young children. Practically, the spring-dose 2 dose series is also difficult to implement. ACIP policy allows 2 doses to be administered across seasons for children who received an influenza vaccine in previous seasons. However, with the findings in our review, we highlight the importance of receiving 2 doses in a single season, especially for immunologically naïve children, which is particularly relevant when vaccine composition changes sufficiently across seasons.

This review has several limitations. In few studies did researchers include the oldest group within the 6 months through 8 years age range, and most of the studies were focused on children aged ≤5 years. Data were limited to make conclusions about children aged 5 to 8 years. In most immunogenicity trials, researchers administered a 0.25 mL dose for children aged <3 years and 0.5 mL dose for children aged ≥3 years. Recently, several vaccines have been licensed in the United States to be administered at a full 0.5 mL dose (rather than 0.25 mL) in children aged 6 through 35 months, and therefore it is unclear whether our findings apply to higher doses. Studies in our review were only focused on HAI antibodies, which is not a perfect correlate of protection. Other mediators of immunity that likely play a role in protection, such as cell-mediated immunity and anti-neuraminidase antibodies, were not evaluated. Finally, in our review, we did not include LAIV trials because LAIV is only licensed for children aged ≥2 years. Although there is some evidence to suggest 1 dose of LAIV provides optimal protection against influenza in vaccine-naïve children,⁴⁴  other data support the importance of 2 doses of LAIV in this population.⁴⁵ 

Current evidence suggests that 2 doses of IIV improves antibody responses and clinical protection against influenza-related medical illness, compared with 1 dose in children aged 6 months through 8 years. Vaccinating children with 2 doses in the same season before their second birthday provided optimal protection. These findings support the current ACIP recommendations. Further efforts are needed to determine if the imprinting benefits of early vaccination in immunologically naïve children could provide durable and broad protection against homosubtypic and heterosubtypic strains. Imprinting children through optimal vaccination strategies could have far reaching benefits for the control of seasonal influenza and improving public health.

ACIP

Advisory Committee on Immunization Practices

ARI

acute respiratory infection

CI

confidence interval

FV

fully vaccinated

GMR

geometric mean titer ratio

GMT

geometric mean titer

HAI

hemagglutination inhibition

IIV

inactivated influenza vaccine

ILI

influenzalike illness

LAIV

live attenuated influenza vaccine

MFR

mean-fold rise

PV

partially vaccinated

RCT

randomized controlled trial

SCR

seroconversion rate

VE

vaccine effectiveness