Vaccination against influenza viruses annually: Renewing or narrowing the protective shield?

The influenza virus has loomed large in the minds of scientists and public health policymakers since the 1918–1920 Great Influenza Pandemic. The latter half of the 20th century brought significant developments in influenza virus vaccines. Since 2010, the Advisory Committee on Immunization Practices has recommended annual influenza vaccination for all persons aged ≥6 mo (Fiore et al., 2010). Vaccination has undoubtedly reduced the influenza virus disease burden, but variable vaccine efficacy due to antigenic drift (Hensley et al., 2009) necessitates yearly reformulations of the vaccine (Gerdil, 2003). Seasonal influenza outbreaks continue to cause significant morbidity and mortality, with 3–5 million cases of severe illness and 290,000–650,000 deaths globally each year (Iuliano et al., 2018). Additionally, the zoonotic spillover of highly pathogenic avian influenza viruses to mammals (Caserta et al., 2024) continues to threaten to cause future pandemics (Guan et al., 2010). The antibody response to influenza virus is primarily directed against the hemagglutinin (HA) glycoprotein that mediates viral entry into host cells (Krammer, 2019), rendering this antigen the main target for influenza virus vaccines. Additionally, immune responses to the neuraminidase (NA) enzyme have proven to play a role in protection (Rajendran et al., 2021). HA and NA are antigenically variable and mutate in response to shifting immunity in the population, although NA evolves slower than HA (Kilbourne et al., 1990). While vaccine research has focused on improving immune protection and cross-reactivity to these antigens, a full understanding of why it is difficult to maintain high year-to-year vaccine efficacy or durable, broadly protective immunity persistently remains out of grasp.

It has been recognized since 1960 that childhood exposure to influenza viruses shapes subsequent antibody responses to other strains later in life. This “Doctrine of Original Antigenic Sin,” described by Thomas Francis Jr (Francis, 1960), has evolved conceptually to “immunological imprinting” (reviewed in Koutsakos and Ellebedy [2023]) and applies to pathogens beyond influenza viruses as a potential fundamental adaptive immunity mechanism (Yewdell and Santos, 2021). The original observation made by Francis Jr. that the antibodies or immunoglobulins (Igs) produced against the first influenza A virus subtype encountered in life dominate over antibody titers produced against subsequent influenza virus subtypes can mechanistically be explained by the recall of influenza-specific memory B cells (Abreu et al., 2020). Antigen-experienced memory B cells exist at higher precursor frequencies than their naive counterparts, and they possess enhanced survival programming and lower B cell receptor signaling thresholds (Good et al., 2009) that facilitate the rapid differentiation into antibody-secreting plasmablasts (Wrammert et al., 2008; Phan and Tangye, 2017). These factors contribute to the rapid anamnestic responses observed after secondary antigen exposure. Furthermore, it has been demonstrated that the imprint of childhood influenza A virus can reduce the risk of severe disease in subsequent exposures to the same strain in adulthood or exposure to novel strains within the same phylogenetic group (Gostic et al., 2016, 2019). Thus, epidemiological studies and basic immunology data reinforce that prior influenza virus exposure can alter subsequent immune responses for both positive and negative disease outcomes.

The implication of exposure history affecting secondary influenza virus immunity has repeatedly brought into question the recommended policy of influenza virus vaccination annually. The earliest study to suggest annual revaccination may not provide long-term benefits was published in 1979 (Hoskins et al., 1979), but data on the subject have conflicted in the following decades. Nevertheless, 118 of 194 member countries of the World Health Organization report a seasonal influenza virus vaccination policy (Morales et al., 2021), although vaccination rates vary (Chen et al., 2022). This review will examine the data available from epidemiological studies in support of or opposition to the benefits of influenza virus vaccination annually. Further, immunological data will be considered to propose potential mechanisms explaining why suboptimal immune responses may be generated after multiyear vaccination. This review does not argue that influenza virus vaccines do not work. Additionally, this review does not call into question the value of annual influenza virus vaccination. When major antigenic shifts are observed in circulating influenza virus strains (Bedford et al., 2015; Chen et al., 2022), particularly in response to emerging variants reflected in northern and southern hemisphere infection rates (Chan et al., 2010), the recommended policy of annual vaccine updates is certainly warranted. Influenza virus vaccines are efficacious (DiazGranados et al., 2012; Gross et al., 1995); this review considers whether the heterogeneity in vaccine efficacy could be explained or improved upon by modifying or maintaining the policy of annual influenza virus vaccination.

Unsurprisingly, it has been reported that compared with placebo, influenza virus vaccination reduces incidences of upper respiratory illness (Nichol et al., 1995). However, concerns raised by prior studies that there may be drawbacks to repeat vaccination (Hoskins et al., 1979) prompted research on the value of annual revaccination. A 5-year study conducted from 1983 to 1988 found that repeat vaccination was associated with reductions in influenza virus shedding and infection in adults, supporting the recommendation for annual immunization (Keitel et al., 1997). Even if annual immunization does not reduce the incidence of disease, reductions in viral shedding argue in favor of repeat vaccination to reduce transmission and protect at-risk populations. There have been limited studies since the publication of Keitel et al. that have demonstrated the benefits of repeat immunization with influenza vaccines. A series of studies covering the 2011–2012, 2012–2013, and 2013–2014 influenza seasons in Canada did find evidence for an increase in vaccine effectiveness for A/H1N1 among individuals who had received multiple vaccinations; however, this was over several seasons in which the influenza virus HA antigen was relatively conserved, and lower vaccine effectiveness was observed among repeat vaccinees for A/H3N2 (Skowronski et al., 2014a, 2014b, 2015).

The general support for annual vaccination stems from the following: (1) antigenic drift necessitates updating the vaccine to counter the annual epidemics, so it is safer to recommend yearly vaccination, and (2) repeat vaccinees still appear to be better protected than the unvaccinated. The main advantage of repeat vaccination appears to be when the HA antigen is well conserved across multiple seasons, such as for A/H1N1 or influenza B virus. Otherwise, those with repeat vaccination histories do not appear to reap substantial benefits concerning vaccine effectiveness or antibody titers compared with those vaccinated only in the current season.

The first major studies to raise concerns about repeat vaccination concerned three outbreaks of influenza A virus in 1972, 1974, and 1976 among children at a British boarding school (Hoskins et al., 1973, 1979). The summation of the findings indicated that there was no significant difference in influenza virus attack rates between boys unvaccinated during the outbreaks and boys vaccinated before each outbreak. It was also found that boys with no prior vaccination history fared better in attack rates when vaccinated before outbreaks than boys with repeat vaccination histories (Hoskins et al., 1979). Thus, this study called into question the long-term benefits of annual vaccination. Further studies have also identified susceptibility in populations with repeat influenza immunization. A study of influenza virus vaccine effectiveness carried out over eight seasons (2004–2013, omitting 2009–2010) of influenza virus infection in a Wisconsin community found that current season vaccine effectiveness against A/H3N2 was highest among individuals with no prior vaccination history compared to those with frequent vaccination history (McLean et al., 2014). Assessment of the 2014–2015 influenza season in Canada, an epidemic of antigenically drifted A/H3N2, found that vaccine effectiveness was 52% in patients with no prior vaccination history but significantly lower if patients were vaccinated one year prior (−32%) or vaccinated each year since 2012–2013 (−54%) (Skowronski et al., 2016). Vaccine effectiveness was similarly lower in children in schools in China who received vaccinations in the prior years versus those who only received vaccinations in 2014–2015 (Zhang et al., 2017). Finally, a recent systematic review of 83 studies and meta-analysis of 43 studies reported reduced vaccine effectiveness for influenza A and influenza B viruses after consecutive vaccination in previous seasons; however, protection was still substantial over those unvaccinated, and the effects of attenuation were not deemed severe (Jones-Gray et al., 2023).

A study of 20 participants based on seasonal influenza virus immunization history reported potentially blunted immune responses due to repeat vaccination. Participants who had received minimal vaccinations in the past 5 years displayed higher frequencies of antibody-secreting cells and greater fold changes in hemagglutination inhibition (HAI) titers, although repeatedly vaccinated individuals displayed higher HAI titers at baseline (Sherman et al., 2020). Similarly, assessment of patients who received inactivated or live attenuated influenza virus vaccines found that those patients who had received two consecutive years of prior vaccination exhibited lower HAI titers after vaccination, but rates of decline were similar between those vaccinated 2 years prior (2004–2005, 2005–2006) and those that were not (2005–2006) (Petrie et al., 2015). Another household study in Michigan, USA, over the 2012–2013 influenza season also found reduced vaccine effectiveness and lower HAI titers against A/H3N2 in repeat vaccinees compared with those who only received the 2012–2013 vaccine (Ohmit et al., 2014, 2015). This effect was consistent with a 2010–2011 study of healthcare workers receiving seasonal influenza virus vaccines, which also observed blunted responses in repeat vaccinees to A/H3N2 (Thompson et al., 2016). Finally, a 1996–1999 study comparing vaccine responses in healthy young adults (20–40 years) and healthy old adults (≥65 years) suggested that differences in responses between the age groups may be better explained by vaccination history than immunosenescence (Mosterín Höpping et al., 2016). While young adults exhibited more robust titers after vaccination compared with old adults, this difference declined with repeat immunizations until almost no effect due to age was observed, suggesting that the initial difference may be due to vaccination history rather than declining immune competence (Mosterín Höpping et al., 2016). The negative impact of repeat vaccination may be an influenza virus-specific phenomenon, as a study of healthcare workers in Japan from 2003 to 2004 observed negative effects of repeat vaccination on influenza virus HAI titers but no effect on antibody titers to hepatitis virus (Nabeshima et al., 2007). It should be noted that reduced HAI titers do not always correlate with a reduction in vaccine effectiveness (Gilbert et al., 2019). An HAI titer ≥40 is considered the primary correlate of protection for influenza virus vaccines, but this may not always be the most appropriate measure to predict protection rates or vaccine efficacy (Tsang et al., 2014b; Memoli et al., 2016). Non-neutralizing antibodies may also protect against influenza virus infection (Carragher et al., 2008; Ko et al., 2021), and their relative fold change after repeat vaccination would not be indicated by HAI titer.

The disadvantages of repeat vaccination appear to be linked to antigenic changes in the HA glycoprotein; thus, the effects are mostly seen for A/H3N2, which tends to be more antigenically variable than A/H1N1 or influenza B virus (Bedford et al., 2014, 2015). There appears to be a clear negative effect on vaccine effectiveness when individuals are previously vaccinated in multiple seasons. The evidence suggests the reduction in vaccine effectiveness is derived from decreased HAI titers in repeat vaccinees compared with those who receive only the current seasonal influenza virus vaccine. The antigenic distance hypothesis has been proposed to explain the reduction in vaccine effectiveness observed in some cases after multiple immunizations.

The antigenic distance hypothesis was proposed by Smith et al. (1999) to reconcile the different results regarding repeat vaccination reported by Hoskins et al. (1979) (not protective) and Keitel et al. (1997) (protective). Smith et al. employed computer modeling of B cell clones with different antigen specificities to test how their hypothesis could explain the conflicting reports. The hypothesis considers the antigenic distance between two vaccines, vaccine 1 (v1) and vaccine 2 (v2), as well as a possible epidemic influenza virus strain (e). When antigenic distance v1–v2 is low (v1 and v2 are antigenically similar), antibodies produced by v1 cross-react with v2, termed negative interference. When antigenic distance v1−e is low, antibodies produced by v1 cross-react with e and help to clear the virus, termed positive interference. The effect of repeat vaccination is thus dependent on the combination of negative and positive interference (Smith et al., 1999). What this hypothesis predicts is that homologous boosting can be detrimental to the immune response compared with a single dose because v1–v2 is 0 and negative interference is maximized. If the circulating strain then exhibits high antigenic distance from v1 (v1−e > 0), positive interference is minimal and possibly masked by negative interference. The hypothesis also helps to resolve some of the different observations in Hoskins et al. versus Keitel et al. (and subsequent studies) by suggesting that heterogeneity in immune responses to repeat vaccination versus single immunization is caused by varying degrees of antigenic distance between vaccine strains and circulating strains (Fig. 1).

This conceptual framework underscores that seasonal influenza virus vaccine efficacy is likely highly variable due to varying degrees of antigenic similarity between the vaccination strain and circulating, potentially epidemic, strains year to year, as well as any compounding effects due to antigenic distance between vaccine/circulating strains and immunologically imprinted strains (Gostic et al., 2016, 2019). This is without considering other factors contributing to immune heterogeneity (e.g., age) and vaccination history. As the case studies above demonstrate, there have been multiple influenza virus seasons globally in which repeat vaccination appears to have disadvantaged individuals compared to those with a recent lack of immunization.

Repeat vaccination does not necessarily need to be proven beneficial over single-season immunization to justify its use in public health policy. However, the evidence that repeat vaccination may reduce vaccine effectiveness suggests there are immunological mechanisms worth considering for why this result has been observed multiple times in case studies. Rather than using these potential mechanisms as an argument for rejecting annual vaccination, it is more beneficial to consider how these mechanisms could be subverted or manipulated to improve vaccine efficacy. Indeed, even without achieving broad cross-reactivity, vaccines that reduce transmission would advance the public health benefits of seasonal influenza virus vaccination. The following sections discuss potential immunological mechanisms that may contribute to reduced vaccine effectiveness after repeat influenza virus immunization.

Studies have documented that annual seasonal influenza virus vaccination may reduce or blunt subsequent immune responses. However, the immunological mechanism behind this phenomenon is not known. The following sections detail possible explanations for why repeated immunization may not be as beneficial as single immunization (potential mechanisms illustrated in Fig. 2).

Preexisting influenza virus–specific antibody titers are the most widely reported potential immunological mechanism for reduced vaccine effectiveness after repeat influenza virus immunization. Even without vaccination, lifetime exposure to seasonal influenza viruses results in baseline circulating influenza virus–specific antibody titers. However, vaccination has been shown to boost antibody titers less effectively when participants have been previously immunized, particularly with homologous virus strains.

Several studies have demonstrated the impact of preexisting antibody titers on subsequent influenza virus vaccination. A study conducted from 1990 to 1992 in Finland demonstrated preexisting antibodies from natural infection could blunt postvaccination HAI titers; subsequent vaccinations were also less likely to boost HAI titers (Pyhälä et al., 1994). Cohort studies performed on 884 young and elderly adults from 1986 to 1989 across the Netherlands and Switzerland demonstrated that based on HAI titers, prevaccination status and antibody titer were significantly correlated with postvaccination HAI titers (Beyer et al., 1996). Repeat vaccination had a positive effect on A/H3N2 postvaccination titers, no impact on A/H1N1 postvaccination titers, and a negative effect on influenza B virus postvaccination titers compared with single vaccination controls (Beyer et al., 1996). A subsequent study of healthcare workers in Portland, OR, USA, from 2010 to 2011 found that prior receipt of a 2009 influenza A/H1N1 pandemic virus (A/H1N1/pdm09) vaccine resulted in higher baseline HAI titers before 2010–2011 vaccination and ultimately decreased probability of maintaining an HAI titer ≥40 at the end of the influenza virus season compared with participants that were only immunized once (Gaglani et al., 2014). Finally, an in-depth analysis of 63 volunteers contributing samples before and after the 2009 seasonal and pandemic A/H1N1 vaccination found that even with modeling multiple parameters incorporating gene expression data, the magnitude of postvaccination antibody responses remained tightly linked with prevaccination status and baseline serum titers (Tsang et al., 2014a).

Evidence from human studies suggests that influenza virus–specific preexisting antibodies are modulating an inhibition mechanism. Sasaki et al. reported that lower baseline HAI titers were associated with a greater fold increase in postvaccination HAI titers and frequencies of antibody-secreting cells as measured by ELISpot (Sasaki et al., 2008). This decrease in antibody-secreting cells as baseline antibody titers increase would suggest that circulating antibodies directly or indirectly inhibit the proliferation of influenza virus–specific B cells rather than simply saturating the circulating blood concentration of antibodies. Further functional inhibition of new antibody responses by preexisting antibodies was demonstrated by Khurana et al. in 16 individuals who received repeat influenza virus vaccination (Khurana et al., 2019). Participants who were vaccinated twice exhibited a decline in HAI seroconversion rates after a second immunization despite limited antigenic distance among the vaccine strains, as well as a decrease in antibody affinities in year 2 compared with year 1 (Khurana et al., 2019). Antibody affinities not only declined from postvaccination year 1 to prevaccination year 2, suggesting a loss of high-affinity antibodies in the circulating serum population before a second immunization, but the maturation of antibodies postvaccination year 2 was less than the maturation observed after vaccination in year 1 (Khurana et al., 2019). This suggests a limited capacity to increase the binding affinities of influenza virus–specific polyclonal antibodies and even a potential blunting of affinity maturation after repeat vaccination.

Thus, the literature indicates a reduction in vaccine effectiveness after repeat immunization; the degree of this reduction is potentially related to the antigenic distance between the vaccine and circulating virus strains. Why repeat immunization reduces vaccine effectiveness may be linked to preexisting antibody titers. Vaccinated once, an individual produces influenza-specific antibodies that persist at baseline until the next vaccination. Upon secondary immunization, these baseline antibodies inhibit the fold increase in serum titers after vaccination, resulting in reduced protection (particularly, in theory, if a circulating strain is antigenically distinct from the vaccine-specific antibodies). Why would preexisting antibodies be expected to exercise such an effect over subsequent vaccination when, potentially, stimulation of prior memory B cells would be anticipated to be just as robust as after the first immunization, if not more so? It is possible preexisting antibodies bind vaccine antigens resulting in sequestration in immune complexes, limiting antigen access and stimulation of influenza virus–specific memory B cells.

Antigen-specific antibodies in circulation can mediate their effector function by binding the immunogen via the fragment antigen binding region. Multiple antibodies bound to a single (or multiple) antigen(s) will generate immune complexes, which, via the exposed fragment crystallizable (Fc) region of the antibodies, can activate the complement pathway or bind Fc receptors (FcRs) to activate or inhibit immune cells (reviewed in Bournazos et al. [2017]). Preexisting antibodies generated by influenza virus vaccination can then modulate subsequent vaccine responses by forming vaccine-induced immune complexes (reviewed in Wang et al. [2018]).

One possible mechanism by which immune complexes hinder postrepeat immunization antibody responses may be the capture and clearance of influenza virus vaccine antigens. Influenza virus vaccines are primarily delivered by intramuscular injection. This delivery method can potentially introduce vaccine antigens to the tissues at the injection site and the circulation via draining lymph or blood. It is possible for circulating antibodies to encounter antigens in any of these sites. Upon forming immune complexes, the antigen can be sequestered away from lower affinity memory B cells, limiting recall antibody titers, or rapidly cleared from circulation (Davies et al., 1990). However, it is also possible for immune complexes to enhance antigen delivery to lymph nodes and capture by antigen-presenting cells in animal models (Phan et al., 2007), so it would be predicted that preexisting antibodies might have the opposite effect in humans vaccinated against influenza virus. Experiments in mice suggest that whether preexisting antibodies have an enhancing or suppressive effect on subsequent B cell responses depends upon the clonality, affinity, and titers of the antibodies produced from the primary challenge (Tas et al., 2022). Since the prevailing data in humans suggest higher titers of preexisting influenza virus–specific antibodies are inversely correlated with a fold increase in HAI titers after repeat immunization, it would support the hypothesis that the mechanism is primarily antigen capture and clearance rather than improved antigen delivery to lymph nodes.

It is also possible that preexisting antibodies inhibit subsequent responses through an FcR-mediated mechanism. A study found that distinct glycoforms of HA-specific antibodies can modulate FcR expression on the surface of B cells (Wang et al., 2015). Sialylated anti-HA IgGs produced in human vaccinees complexed with A/H1 HA protein stimulated the increased expression of FcγRIIB, an inhibitory type I FcR, on human CD19⁺ peripheral blood mononuclear cells (Wang et al., 2015). The upregulation of FcγRIIB was determined to be driven by the binding of sialylated Fcs to the type II FcR CD23 (Wang et al., 2015). Based on experiments in mice, the authors concluded that this mechanism could enhance influenza virus–specific antibody responses by favoring the secretion of high-affinity antibodies and suppressing low-affinity B cells, which are inhibited by FcγRIIB engagement; however, mice may not accurately represent the dynamics of preexisting immunity in humans with lifetime exposure to seasonal influenza virus. At a minimum, the results suggest immune complexes can (1) indirectly inhibit postvaccination titers by upregulating inhibitory receptors on responding B cells and (2) directly inhibit postvaccination titers by binding these inhibitory receptors. A direct inhibition mechanism has been observed for other viral antigens in animal models (Kim et al., 2011). This would provide a second alternative explanation for why high baseline antibody titers are associated with lower fold increases in HAI titers after repeat vaccination.

However, fold change in antibody titers after vaccination may not be the primary reason repeat vaccination is associated with reduced vaccine effectiveness. A decrease in the fold change of antibody titers postrepeat vaccination compared with primary vaccination does not mean the HAI titers induced are insufficient for protection. If circulating baseline antibodies have a negative impact on vaccine responses by capturing antigens, they would be equally capable of binding infectious viruses (although possibly not as effective at neutralizing viruses). It is also not inevitable that preexisting antibodies would have a negative effect on vaccine responses. Evidence suggests when vaccine strains are divergent from previous years, preexisting antibodies may block previously encountered epitopes, increasing the magnitude of new B cell responses to variant strains (Andrews et al., 2015). Further, this emphasizes that the negative effects of repeat vaccination may be relative to the antigenic distance between current and previously encountered influenza viruses. Ultimately, these negative effects may have more to do with the quality of the antibody repertoire generated rather than the quantity of antibodies secreted.

One of the predictions of the antigenic distance hypothesis is that repeat vaccination shifts the antibody repertoire toward the boosted vaccine virus strain, which, when there is a substantial antigenic distance between the vaccine strain and the circulating strain, disadvantages the host in comparison with a single immunization because the response is directed further away from beneficial cross-reactive clonotypes (Smith et al., 1999). The immunological level at which this effect operates would likely be the influenza virus–specific memory B cell repertoire.

Memory B cell clones recalled in vaccine responses are certainly retained in circulation across multiple years of immunizations (Vollmers et al., 2013). There is also likely an “antigenic seniority” to the recallability of the memory B cell repertoire: neutralization titers are highest against strains encountered early in life, and these titers tend to increase over time (Lessler et al., 2012; Miller et al., 2013). Additionally, variable exposure history, whether through vaccination or infection, results in unique “antibody landscapes” for each individual; again, these landscapes are characterized by biases in the highest pre-exposure titers to strains encountered earlier in life (Fonville et al., 2014). Both results demonstrate that prior influenza virus exposure, whether by infection or by vaccination, establishes a baseline memory B cell repertoire, which dictates the possible dynamics of subsequent antibody responses. As these cells primarily contribute to recall responses that peak after vaccination (Wrammert et al., 2008; Ellebedy et al., 2016), they may play a greater role in the effects of repeat immunization over preexisting antibody titers.

The correlative effect of preexisting titers may be due to the observation that after boosting with influenza virus vaccines, the majority of serological antibodies consist of preexisting clonotypes from prior exposure, rendering newly elicited clonotypes a smaller percentage of the circulating antibody titers (Lee et al., 2016). Thus, the preexisting titers do not cause a reduction in vaccine effectiveness per se, but the overwhelming dominance of preexisting (but not necessarily cross-reactive) memory B cells in recall responses “drowns out” potentially protective newly elicited clonotypes. These results also suggest that current influenza virus vaccines engage a limited repertoire of memory B cells, with a minority of the most frequent clonotypes dominating the serological response (Lee et al., 2016). This would be useful when the repeated immunizing strain is protective against an epidemic strain but detrimental if the circulating strain evades these clonotypes. Cross-reactive HA stem antibodies can persist in the serological repertoire over multiple years (Lee et al., 2019), but again, this could be a negative factor if the dominant antibody profile induced by vaccination renders the repertoire resistant to an increase in the proportion of clonotypes recognizing drifted viruses.

The possible immunological mechanism behind reduced vaccine effectiveness after repeat immunization is thus not the preexisting antibody titers alone but the composition of the memory B cell repertoire engaged by vaccination. The repertoire is likely both imprinted by strain exposure early in life, as documented by the “original antigenic sin” phenomenon (Francis, 1960), and resistant to major shifts in clonal dynamics if current vaccines only engage a few dominating clones (Lee et al., 2016). Therefore, a single immunization might be less counterproductive than multiple immunizations (particularly with a homologous strain) because it would be less likely to entrench the memory B cell repertoire toward nonprotective clonotypes. Then, in subsequent years, when the circulating strain escapes vaccine-induced immunity, it would be predicted that those with less vaccine exposure would have more mutable B cell repertories capable of mounting protective antibody titers in the recall response. Cross-reactive antibody clonotypes may also represent a greater proportion of the circulating antibody repertoire in those receiving a single immunization. Vaccine design and efficacy likely need to be evaluated not only by the fold change in antibody titers induced after vaccination but also by the clonal composition of the serological antibody repertoire and its capacity to compensate for viral escape.

This review has primarily considered B cell–mediated mechanisms driving a reduction in protection after repeat immunization. However, some attention must be given to the potential for T cell–mediated effects on seasonal influenza virus vaccination. B cells are more likely to be responsible for the effects of repeat immunization because antibody-mediated immunity is primarily homosubtypic within influenza virus subgroups (and usually directed against the antigenically variable HA head); cross-reactive heterosubtypic clones exist in the human memory B cell compartment, but their antibodies are rare in serum (McCarthy et al., 2018). Subsequently, B cell–mediated immunity has greater potential to be biased toward one virus strain over another. T cells, on the other hand, are more likely to facilitate protective heterosubtypic immunity in animal models (Weinfurter et al., 2011; Seo et al., 2002; Laidlaw et al., 2013). This is due to T cell receptor recognition of conserved internal proteins of the influenza virus (Assarsson et al., 2008) compared with antibody recognition of the variable HA head. Evidence exists for T cell–mediated heterosubtypic protection against seasonal influenza viruses in humans (Hayward et al., 2015; Sridhar et al., 2013; McMichael et al., 1983).

Heterosubtypic protection has been studied in cytotoxic CD8 T cells (Sridhar et al., 2013; McMichael et al., 1983) capable of killing virally infected cells or secreting interferon-γ (Assarsson et al., 2008; Gotch et al., 1987; Zweerink et al., 1977). The lack of HA as a major target for CD8 T cell responses (Assarsson et al., 2008) makes it unlikely that these effector cells would significantly contribute to biases in immune responses after repeat immunization. Although CD8 T cell epitopes to HA and NA exist (Bui et al., 2007), it is difficult to propose a mechanism whereby repeated immunization would be detrimental compared with single immunization due to CD8 T cell recognition of conserved epitopes. Data also suggest that current influenza virus vaccines do not efficiently induce CD8 T cell responses (He et al., 2006; Forrest et al., 2008; Bodewes et al., 2009), making it unlikely that repeat vaccination will have significant detrimental effects. While CD4 T cells appear to primarily recognize epitopes corresponding to conserved internal proteins of the influenza virus (Assarsson et al., 2008), they also recognize HA and NA epitopes (Richards et al., 2010), possibly to a greater extent than CD8 T cells (Bui et al., 2007). This may be relevant to repeat seasonal influenza virus vaccination in the context of CD4 Tfh cell repertoires.

Tfh cells provide costimulation and survival signals to antigen-specific B cells during germinal center (GC) reactions in draining lymph nodes, facilitating clonal selection and affinity maturation of antibody responses (reviewed in Crotty [2019]). Helper T cells are thought to be a limiting factor in regulating B cell selection based on the amount of antigen presented by GC B cells to Tfh cells (reviewed in Victora and Nussenzweig [2022]), rendering the Tfh cell repertoire a likely important factor in controlling both naive and memory B cell responses to influenza virus vaccines. Levels of CD4 T cells are correlated with protection and antibody titers to the influenza virus (Mettelman et al., 2023), indicating they are functionally relevant. Additionally, recombinant HA protein-alone vaccines can induce robust CD4 T cell responses and antibody titers, even superior to inactivated viral vaccines (Richards et al., 2020). This suggests that not only can CD4 Tfh cells recognize HA and probably NA epitopes presented by B cells, but these antigens may also exert immunodominance effects on the Tfh repertoire. A CD4 T cell immunodominance hierarchy study in children versus young adults provided evidence that the CD4 T cell repertoire can be imprinted early in life (Shannon et al., 2019). Annual influenza virus vaccination definitively generates circulating memory Tfh cells (some of which are HA-specific) (Bentebibel et al., 2016; Herati et al., 2014) with the maintenance of oligoclonal lineages over subsequent years (Herati et al., 2017). These circulating Tfh cells are thought to be derived from lymph node GC Tfh cells (Heit et al., 2017). Further evidence demonstrates that influenza virus vaccine–induced GC Tfh cell clonal lineages can persist in lymph nodes over multiple years of vaccination (Schattgen et al., 2024), highlighting the potential for multiple immunizations to shift or bias the Tfh cell repertoire.

While a greater understanding of the impact of repeat immunization on influenza virus–specific Tfh cells is needed, the available literature suggests that (1) Tfh cells can respond to HA and NA epitopes and are directly correlated with antibody responses; and (2) the potential exists for the CD4 Tfh repertoire to be influenced by immunodominance effects. Studies in mice indicate that Tfh cells can migrate between multiple GCs (Shulman et al., 2013), suggesting a limited Tfh cell repertoire may have a disproportionately large bottleneck effect on the B cell clones that receive costimulation. As a potential limiting factor providing T cell help and survival factors to lymph node influenza virus–specific memory B cells and GC B cells, the Tfh cell repertoire may be responsible for the observed effects of repeat immunization. Multiple immunizations may skew the repertoire toward limited T cell epitopes, thereby biasing the memory B cells that preferentially receive costimulation even when a greater proportion of the memory B cell pool can theoretically respond to a circulating strain. Multiple annual immunizations “cement” the Tfh cell repertoire in draining lymph nodes, and these Tfh cells then limit the future recallability of the B cell repertoire at both the memory B cell and GC level. A single immunization, in contrast, would be expected to have less of a reinforcing effect on the immunodominance of the Tfh cell repertoire, thus maintaining the broad capacity of the recall response. Therefore, it may be that neither preexisting antibodies nor memory B cells are directly responsible for a reduction in vaccine effectiveness after repeat immunization, but a relatively limited repertoire of Tfh cells. Future studies should focus on the effects of repeat immunization and Tfh cells to determine the validity of this hypothesis.

This review examined both the possibility that repeat seasonal influenza virus vaccination results in reduced vaccine effectiveness and the potential immune mechanisms that could be responsible for observed decreases in vaccine efficacy. Numerous case studies suggest there is a reduction in vaccine effectiveness after repeat immunization compared with single immunization. The antigenic distance hypothesis predicts this risk is likely increased when vaccine strains significantly vary antigenically from circulating viruses. Possibly, preexisting antibodies limit future responses after repeat immunization by capturing and clearing vaccine antigens or directly inhibiting B cells via FcRs. If repeat immunization bypasses these antibodies, the next culprit may be the memory B cell repertoire. Multiple doses may push the memory B cell repertoire toward a limited number of clones that do not confer broad cross-reactivity. Finally, even if the memory B cell repertoire has the potential to recall many lineages against epidemic strains, repeat immunizations may bias the Tfh cell repertoire in such a manner that only a limited pool of B cells can access costimulation, proliferate, differentiate to plasma cells, and secrete antibodies.

While these mechanisms are not mutually exclusive, and more work is needed to support their role in repeat immunization, the overarching theme is that current seasonal influenza virus vaccines likely stimulate a limited diversity of B and T cells in draining lymph nodes. It is not the “annual” aspect of vaccines that reduces their effectiveness but their inability to broadly stimulate the available clones that could protect against circulating strains in vaccine mismatch years. One potential change to current vaccination policies could be to provide influenza virus vaccines covering all types (A/H1, A/H3, and B) to those without a recent (∼2–3 years) history of vaccination, but separate monomeric vaccines encompassing major antigenic variant strains for those vaccinated in the previous year. This may mitigate any effects of immunodominance or epitope-focusing exacerbated by repeat vaccination. While this may not be practical for current public health policy implementation and vaccine manufacturing, it could be tested experimentally in smaller cohorts to determine any benefits or mitigation of reduced vaccine effectiveness. There is also a need to evaluate whether new vaccine platforms, such as mRNA-based vaccines (Lee et al., 2023), multivalent vaccines incorporating HA proteins from multiple strains (Yang et al., 2024), or chimeric HA antigen vaccines (Guthmiller et al., 2025), increase antibody titers against both the vaccine and variant strains while engaging a more diverse proportion of the memory B cell and Tfh cell repertoires. Such increased diversity could potentially expand the cross-reactivity of the circulating antibody population. This would likely mitigate reduced vaccine efficacy in years when circulating strains are significantly drifted from vaccine strains, hopefully steering a course toward a truly “universal” influenza virus vaccine.

Author contributions: H.C. Matz: conceptualization and writing—original draft, review, and editing. A.H. Ellebedy: conceptualization and writing—original draft, review, and editing.