Despite widespread vaccination programs, older adults remain susceptible to the serious complications of influenza and experience dramatic increases in hospitalization rates when A/H3N2 is the predominant circulating strain (2017; 2018). Historically, influenza vaccines have been designed to stimulate antibody responses to the critical protective epitopes surrounding the receptor-binding domain of the globular head of hemagglutinin (HA) that permits infection of the host cell. However, although the major protective mechanism of vaccines in young adults is antibody-mediated, older adults also need cytotoxic T lymphocyte (CTL)-mediated clearance of the virus once infection occurs; this provides “clinical protection” against disease in this vulnerable population (Rimmelzwaan and McElhaney, 2008). Nonetheless, when adjusted for the effects of frailty, influenza vaccine effectiveness for the prevention of hospitalization in older adults is similar to that of young adults, suggesting that it is primarily frailty that impacts cellular immunity (Andrew et al., 2017).

Consistent with this notion, prior exposure to influenza through infection or vaccination affects antibody titers and antibody responses to vaccination more so than does aging (Skowronski et al., 2011). In contrast, the decline in cell-mediated immune responses to influenza is related to aging rather than previous exposures to the virus; after age 60, there is a progressive age-related decline in the levels of interferon-γ (IFNγ), IL-10, and the IFNγ:IL-10 ratio secreted by PBMC challenged ex vivo with A/H3N2, A/H1N1, or pH1N1 (Skowronski et al., 2011). We have shown that in older adults, vaccination with current inactivated influenza vaccines provides only a weak stimulus to the production of IFNγ by both CD4⁺ and CD8⁺ T cells Kumar et al. (2017) and to the production of the cytolytic mediator, granzyme B (GrB), as well as the generation of cytolytic CD8⁺ T cells in response to ex vivo influenza A/H3N2 challenge (McElhaney et al., 2006; McElhaney et al., 2009; Shahid et al., 2010; Zhou and McElhaney, 2011; Haq et al., 2017; Kumar et al., 2017). In contrast, the anti-inflammatory IL-10 response is better preserved and contributes to an overall decline with aging in the IFNγ:IL-10 ratio in response to ex vivo influenza challenge (Skowronski et al., 2011). These poor T cell responses have been attributed to the effects of aging [reviewed in (McElhaney et al., 2016; McElhaney et al., 2020a)], persistent cytomegalovirus infection [reviewed in (Merani et al., 2017)], and frailty [reviewed in (McElhaney et al., 2020b)].

Given their role in the cytolysis of infected lung epithelial cells and overall clearance of the virus from the lungs, CD8⁺ CTLs are key to protection against the serious complications of influenza [reviewed in (Altenburg et al., 2015)]. These cells secrete granzyme B (GrB) together with the pore-forming effector, perforin, to enter infected cells and induce apoptosis. Even though only 2–3% of CD8⁺ T cells become positive for both GrB and perforin upon infection, active GrB activity is specifically directed to virus-infected target cells into which it is delivered by perforin; this is a functional readout of physiologically relevant, virus-specific T cell activation (Kula et al., 2019). We have previously shown that GrB activity in ex vivo A/H3N2-challenged peripheral blood mononuclear cells (PBMC) correlates with protection against influenza in older adults (McElhaney et al., 2006; McElhaney et al., 2009; Shahid et al., 2010). However, under conditions associated with aging and/or augmented inflammation Hiebert and Granville. (2012), GrB accumulates extracellularly and retains its activity. Thus, in the absence of perforin, GrB is released into the extracellular space and degrades proteins key to maintaining cell-cell junctions Buzza et al.(2005), disrupts epithelial barrier function Matsubara et al. (2020), and increases susceptibility to infection, inflammation, and injury.

We have shown that GrB activity in circulating T cells (bGrB) correlates with the frequency of late-differentiated T cells in older adults and is significantly increased in those seropositive for persistent cytomegalovirus infection relative to those that are seronegative. Upon infection, ∼40% of CD8⁺ T cells become GrB-positive and perforin-negative Kumar et al. (2017) and release GrB (McElhaney et al., 2012). In the extracellular milieu, GrB plays an important role in the loss of tissue function and integrity in the response to an inflammatory stimulus in the lungs (Granville, 2010; Hendel et al., 2010; Kim et al., 2013). This is particularly relevant to acute viral respiratory infection as it has been previously shown that the number of GrB-positive/perforin-negative cells in the lungs and levels of GrB in bronchoalveolar lavage fluid are elevated in patients with acute respiratory distress syndrome Rasmuson et al. (2016) Recently, plasma levels of GrB were identified as one of six key inflammatory analytes in critically ill coronavirus disease 2019 patients (Fraser et al., 2020).

Previous randomized studies comparing HAI antibody and IFNγ-mediated T cell responses to influenza vaccination in older adults have demonstrated the benefit of high-dose (HD) over standard-dose (SD) formulations of seasonal influenza split-virus vaccine (SVV) (Chen et al., 2011; Kim et al., 2016; Cowling et al., 2019). The purpose of our study conducted over four consecutive influenza seasons was to compare antibody, IFNγ, IL-10, and GrB responses to vaccination among older adults randomized to receive the seasonal formulations of HD-SVV vs. SD-SVV, and a young adult control group who received SD-SVV. We have previously reported the antibody responses from this study and showed that frailty was associated with higher titers and increased antibody responses at 4 weeks following influenza vaccination, which was partially dependent on vaccine dosage (Loeb et al., 2020). We now report differences related to age, vaccine dose, CMV serostatus, and frailty in the cell-mediated immune response to influenza vaccination using ex vivo PBMC challenged with live influenza A/H3N2.

Based on the humoral response, several influenza vaccine formulations have demonstrated enhanced immunogenicity Nicolay et al.(2019) and superior efficacy over standard dose comparator vaccines in randomized trials in older adults (DiazGranados et al., 2014; Dunkle et al., 2017). However, there have been few randomized trials that have compared cytokine-mediated immune responses to these approved enhanced influenza vaccines against a standard dose comparator vaccine in older adults Kumar et al.(2017); Cowling et al.(2019), and none have studied the cytolytic T cell response. To this end, we conducted a randomized trial of HD-SVV vs. SD-SVV over 4 years. The primary comparison planned was between those older adults randomized to the HD vs. SD formulation of seasonal influenza vaccine using our established methods for measuring cytokine (IFNγ, IL-10) and cytolytic T cell (iGrB) responses to vaccination. We found that HD-SVV significantly increased the IFNγ response (produced by both CD4⁺ and CD8⁺ T cells) to vaccination but this result may have been affected by differences in the baseline levels of IFNγ (HD-SVV < SD-SVV) and may explain why there was no difference in the iGrB response to vaccination in HD-SVV recipients when compared to younger and older SD-SVV recipients. Thus, HD-SVV would not be expected to provide additional benefit over SD-SVV in the stimulation of cytolytic CD8 T cells needed to clear the virus from the lungs. In contrast to the GrB activity detected in unstimulated T cells (bGrB), our unpublished data has shown that there are no detectable levels of IFNγ or IL-10 in the supernatants of unstimulated T cells. Further, GrB activity in unstimulated PBMC is largely derived from T cells and thus we have used bGrB activity (in isolated unstimulated T cells) to adjust for this effect in ex vivo challenged PBMC.

Previous randomized trials have measured the CMI response to influenza vaccination in older adults according to changes in IFNγ⁺, CD4⁺, or CD8⁺ T cell frequencies (Cowling et al., 2019; Pillet et al., 2019). Although these IFNγ⁺ T cell frequencies have been shown to be a correlate of protection against influenza A in young adults with low antibody titers, the ex vivo response is specific for peptides derived from the internal proteins, matrix (M1), and nucleoprotein (NP), which are shared between influenza A/H1N1 and A/H3N2 strains but distinct from those of influenza B (Wilkinson et al., 2012; Sridhar et al., 2013). This becomes important when considering the content of the vaccine: SVV formulations studied in this trial contain significant quantities of M1 and NP Co et al.(2009) while subunit Kumar et al.(2017) and recombinant HA formulations contain no internal proteins. In addition, we have shown that IL-10 is an important mediator of the response to vaccination; a low IFNγ:IL-10 ratio in A/H3N2-challenged PBMC following vaccination with SD-SVV is an important predictor of risk for febrile LCII in older adults (Shahid et al., 2010). The production of IL-10 may curtail undesirable inflammatory responses to A/H3N2 infection simulated in A/H3N2-challenged PBMC. In a randomized, double-blind, placebo-controlled trial, we showed that influenza vaccination attenuated the inflammatory response in patients undergoing cardiopulmonary bypass surgery; there was a reduction in serum levels of troponin (as a measure of myocardial injury) and pro-inflammatory mediators [IL-6, IL-8, tumor necrosis factor (TNF), C-reactive protein], and an increase in IL-10 (Atoui et al., 2020). However, given the kinetics of the IL-10 response, which peaks at 4-weeks vaccination, this response as a correlate of protection appears to vary depending on timing of vaccination relative to the onset of the influenza season (McElhaney et al., 2006; Shahid et al., 2010).

The level of GrB activity measured in ex vivo A/H3N2 PBMC (exGrB) reflects the activation of the 2–3% of CD8⁺ T cells that are GrB⁺Perforin⁺ and the 40% of CD8⁺ T cells that are GrB⁺Perforin− (Kumar et al., 2017). We have previously shown that the ex vivo activation of influenza-specific memory T cells corresponds to the cytolytic activity of GrB⁺Perforin⁺ CD8⁺ T cells, which is reduced in both magnitude and duration of the response to SD-SVV in older adults relative to the young (Zhou and McElhaney, 2011). In contrast, GrB⁺Perforin− CD8⁺ T cells degranulate in response to ex vivo influenza challenge McElhaney et al.(2012) and thus have no influenza-specific cytolytic activity in the absence of perforin, while also contributing to local tissue damage. We have therefore developed an assay of bGrB activity in circulating T cells that correlates with the proportion of late/terminally differentiated CD8⁺ T cells (Haq et al., 2017). These GrB⁺Perforin ^- CD8⁺ T cells are present in the circulating T cell pool McElhaney et al.(2012) and contribute to basal levels of GrB (bGrB) activity which are: 1) significantly higher in CMV-seropositive vs. CMV-seronegative older adults; 2) positively correlated with the frequency of late/terminally differentiated CD8⁺ T cells; and 3) negatively correlated with the frequency of memory CD8⁺ T cells (Haq et al., 2017). The potential limitations of this analysis of the cell-mediated immune response to influenza vaccination include the absence of measures of different T cell subset frequencies that we have correlated with the cytolytic activity of CD8⁺ T cells Zhou and McElhaney,(2011) and other studies demonstrating that IFNγ⁺, CD4⁺, and CD8⁺ T cells responding to M1 and NP peptides are correlates of protection against influenza (Wilkinson et al., 2012; Sridhar et al., 2013).

To adjust for the contribution of GrB activity presumably derived from the non-specific inflammatory response to ex vivo A/H3N2 challenge in late/terminally differentiated CD8⁺ T cells, we calculated iGrB levels as exGrB activity divided by bGrB activity. We observed the expected age-related decline in the iGrB response to ex vivo challenge both pre- and post-vaccination. In the absence of this adjustment, exGrB activity was highest in frail older adults and lowest in younger adults. Taken together, these results suggest that bGrB activity accounts for the effect of age and increasing frailty in the iGrB response to ex vivo influenza challenge. Consistent with this finding, all three vaccination groups showed a significant increase in iGrB levels 4 weeks post-vaccination. There was a trend toward a higher fold-increase in iGrB levels 4-weeks post-vaccination in HD vs. SD recipients; the overall weak stimulus to memory CD8 T cells provided by inactivated influenza vaccines may explain why this did not reach statistical significance.

In the regression analysis, we found that increasing age and CMV seropositivity are significant correlates of the decreased iGrB response to ex vivo influenza challenge. CMV has been shown to be the main driver of oligoclonal expansions of CD8⁺ T cells with altered phenotypes and functions (Pawelec et al., 2005). As a component of the Immune Risk Phenotype, CMV seropositivity is an important predictor of mortality (Pawelec et al., 2005). We have since shown that bGrB activity is increased in CMV seropositive vs. seronegative older adults and correlates with the frequency of late/terminally differentiated CD8⁺ T cells that result from oligoclonal expansion. Using the Frailty Index, which is also a strong predictor of mortality Clegg et al.(2013), we now show that although bGrB activity increases with age and increased levels of frailty, CMV seropositive status is a significant contributor to this relationship. In the extracellular space, GrB contributes to enhanced inflammatory responses Hiebert and Granville,(2012) that may suppress the response influenza vaccination. In addition, increased frailty is associated with a lower IFNγ response to ex vivo A/H3N2 challenge. Our previous studies suggest that the addition of a toll-like receptor (TLR4) agonist to SVV can overcome the suppressive effects of IL-10 and improve the IFNγ response and markedly improve the IFNγ:IL-10 ratio and GrB response to influenza challenge (Behzad et al., 2012).

In light of the current controversy over the effects of repeated annual influenza vaccination potentially leading to a loss of vaccine-mediated protection against influenza, we analyzed the effect of vaccination in each study year according to the absolute levels of antibody, cytokines (IFNγ, IL-10, and IFNγ:IL-10) and iGrB across the four study time points and the pre-to 4 weeks post-vaccination response in older adults. We found a significant decline only in the A/H3N2 iGrB and antibody response to influenza vaccination. Our previous study showed that this decline in antibody responses to influenza vaccination in older adults is an effect of repeated vaccination and not age (Mosterin Hopping et al., 2016). In contrast, absolute antibody titers and cytokine levels across the four study time points, increased with each successive year of vaccination in the study. This is important because previous studies have shown that there is no significant impact of prior-season vaccination on vaccine effectiveness for the prevention of influenza hospitalizations in older adults (Cheng et al., 2017; Song et al., 2020). Consistent with our iGrB results showing a decline in the iGrB levels with prior season vaccination, Nichols et al reported that prior season influenza vaccination resulted in a non-significant decline in vaccine effectiveness for the prevention of hospitalization. However, prior season vaccination was still more effective than not receiving the current season’s vaccine (Nichols et al., 2019). These results suggest that the decline in iGrB levels with prior season influenza vaccination correlates with protection against hospitalization but this decline in iGrB levels does not appear to be clinically significant. Further, these findings highlight the importance of re-stimulating the iGrB response with annual influenza vaccination. Relevant to our clinical trial, it has been previously shown in a large clinical trial of HD-SVV vs. SD-SVV in older adults over two influenza seasons where A/H3N2 was the predominant circulating strain, that vaccination with HD-SVV compared to SD-SVV was associated with lower risk of LCII irrespective of prior-season vaccination (DiazGranados et al., 2016). Thus, we are planning to analyze the effect of prior-season vaccination on all of our immunologic measures included in our trial.

The strengths of the study are the kinetic analysis of the response to vaccination and also the study of immune responses in the context of prior season vaccination. However, there are important limitations to the study. The first limitation of this study relates to the recruitment of a relatively younger segment of the over 65 population. Thus, these results may not extend to adults age 85 years and older as is often the case in randomized trials of influenza vaccines. Even so, half of the sample were pre-frail or frail, suggesting that the trial participants were not limited to fit younger seniors. In addition, the trial was not designed to study the effect of repeat vaccination and thus we were limited to studying the effect of prior year vaccination without distinguishing between HD vs. SD vaccines.

The second limitation is that we have not accounted for the contribution of NK cells to the total amount of GrB activity measured in ex vivo A/H3N2 challenged PBMC. Our unpublished data has shown that with ex vivo A/H3N2 challenge, there is a 40-fold increase of GrB activity in isolated T cells, most of which is derived from CD8⁺ T cells. In contrast, a recent study showed that the NK cell response to influenza infection resulted in a 2-fold increase in the Mean Fluorescence Intensity (MFI) of intracellular GrB (with no change in the MFI for perforin) in NK cells (Scharenberg et al., 2019). Further, NK cells contribute to GrB activity in influenza-stimulated PBMC but the co-expression of Perf (needed for cytolytic activity) in these NK cells has been shown to be T cell dependent (He et al., 2004). We have shown that older compared to young adults have an enhanced GrB⁺ CD8⁺ T cell response to ex vivo A/H3N2 challenge when compared to unstimulated PBMC, whereas the GrB⁺ NK cell response is similar in young and older adults (McElhaney et al., 2009). The age-related increase in the GrB⁺ CD8⁺ T cell response appears to be related GrB⁺ CD8⁺ T cells that do not co-express Perf. Taken together, these results suggest that the cytolytic potential based on the amount of GrB produced on a per cell basis in response to ex vivo challenge is much greater in CD8⁺ T cells compared to NK cells but further studies are needed to understand how aging may affect the cytolytic potential (GrB⁺Perf⁺) vs. potential damaging effects (GrB⁺Perf⁻) of CD8⁺ T cells and NK cells.

A third limitation inherent to the study design is that the source of the cytokines cannot be identified. B1 cells, regulatory T cells and NK cells which are part of PBMC cultures, may also produce the cytokines measured. A more detailed analysis using intracellular cytokine staining and flow cytometric analysis of ex vivo stimulated PBMC is needed to resolve this issue. However, we have already found that IL-10-producing T cells are not detectable using standard methods for intracellular cytokine staining.

In summary, this study highlights the age-related decline in the IFNγ and iGrB response to influenza challenge which corresponds to the increased risk for serious complications of influenza in older adults. These responses are only weakly stimulated by SVV, independent of vaccine dose. In contrast, HD-SVV amplified the IL-10 response, with little effect on the iGrB response relative to SD-SVV in either younger or older adults. Thus, the benefit of HD over SD vaccines in older adults may be largely antibody-mediated but not necessarily limited to the neutralizing effects measured in hemagglutination inhibition assays. In light of these results, our future direction is to characterize the IFNγ and iGrB responses to vaccination according to the frequency of different CD4⁺ and CD8⁺ T cell subsets and understand how inflammatory (IL-6, IL-8, and TNF) and anti-inflammatory cytokines (IL-10) produced in ex vivo challenged PBMC, affect the frequency of these T cell subsets. In addition, we want to advance our studies of the effect of prior season vaccination to study the effect of repeated annual influenza vaccination. Ultimately, these biomarkers of protection could be incorporated into early phase clinical trials in the development of more effective influenza vaccines for older adults.