We screened 3283 older adults, of whom 2695 were eligible and 2039 agreed to participate. We obtained signed informed consent from 2008 persons and were able to randomize and vaccinate 1861 participants from 7 October 2017 through 12 January 2018, of whom 1826 provided Day 30 postvaccination serum samples (Figure 1). About 10% of participants were asked to give additional blood; 168 contributed PBMCs for analyses of cell-mediated immunity.

Participants were balanced by age, sex, underlying medical conditions, and prior vaccination history between the 4 study groups, with no statistically significant differences in these key variables overall or among those contributing to the primary antibody findings reported here (Table 1). The most common underlying conditions were hypertension, osteoarthritis, and diabetes. Most participants (66%) had received an influenza vaccination in the preceding year (2016–2017), and only 22% had not received any influenza vaccination in the preceding 5 years. All prior influenza vaccinations were with an SD vaccine. We selected 200 participants from each group for a serologic analysis, and the characteristics of this subset were representative of the entire cohort (Table 2). Pre- and post-vaccination GMTs and corresponding MFRs are described in Table 3 for each vaccine group, and Figure 2 presents comparisons of the MFRs between the enhanced vaccine groups and the SD group. Table 4 shows the proportions of participants achieving specific titer thresholds after vaccination.

There were no statistically significant baseline differences between the groups in prevaccination GMTs for each of the antigens tested, except that the baseline GMTs to the B/Victoria antigen were statistically significantly lower in the SD group (Table 3). As shown in Table 3, among SD QIV recipients, the HAI GMT to A(H1N1) increased from 17 to 69. MF59-adjuvanted and HD recipients achieved a significantly higher postvaccination GMT to A(H1N1) than SD recipients; recombinant-HA recipients did not. Nonetheless, MFRs for A(H1N1) were statistically significantly higher for all 3 enhanced vaccines (range, 5.3–6.1) compared to the SD (4.1; Table 3; Figure 2). The proportions with ≥4-fold rises to titers ≥40 were statistically significantly higher for all 3 enhanced vaccines (range, 59–60%) compared to the SD QIV (42%). As shown in Table 4, the proportions achieving elevated titers of ≥40 were higher for the MF59-adjuvanted (82%) and HD (83%) groups, compared to the SD recipients (72%). Of note, the proportions achieving very high titers of ≥160 were significantly higher for all 3 enhanced vaccines (45–55%), compared to the SD vaccine (35%).

Recipients of all 3 enhanced vaccines achieved higher postvaccination GMTs and greater MFRs to both the egg-propagated A(H3N2) by HAI and the cell-propagated A(H3N2) by MN than recipients of SD vaccines (Table 3; Figure 2). The directions of the effects were similar using other indicators. Of note, only 28% of SD recipients achieved ≥4-fold rises in MN responses to cell-propagated A(H3N2), compared to 39% of MF59-adjuvant, 47% of HD QIV, and 57% of recombinant-HA QIV recipients (P < .01; Table 3).

On several indicators, antibody responses to cell-propagated A(H3N2) by MN were significantly higher among recombinant-HA QIV recipients, compared to SD vaccine, MF59-adjuvant, and HD recipients (Table 3). Specifically, recipients of recombinant-HA QIV achieved a postvaccination GMT to cell-propagated A(H3N2) that was 2.57-fold higher than that of SD recipients; the magnitude of this difference was significantly higher than the 1.43- and 1.33-fold differences of MF59-adjuvanted and HD recipients, respectively, compared to SD recipients (Figure 2). Recombinant-HA recipients were the only enhanced vaccine group with a significantly larger proportion (74%) of very high titers (≥160) against cell-propagated A(H3N2), compared to SD recipients (48%; Table 4).

Responses to the B/Victoria component of both the trivalent influenza vaccine (TIV) and QIV in the study year (B/Brisbane/60/2008) were similar across vaccines, with the exception of a significantly higher GMT of HD TIV, compared with SD QIV (Figure 2). Although the MFRs against B/Phuket/3073/2013 were statistically significantly >1 for both SD and enhanced vaccines (Table 3), the GMTs were significantly lower, compared with the SD QIV, for the MF-59-adjuvanted and HD TIVs, which did not include this B/Yamagata component (Figure 2). The response to recombinant-HA QIV that did include B/Yamagata was not significantly different to SD QIV on any indicator.

For T-cell responses at 7 and 30 days postvaccination, we noted a trend whereby enhanced influenza vaccines elicited increased peak acute IFN-γ⁺ T-cell responses and maintained higher average memory responses for selected time points and viruses (Figure 3; Supplementary Appendix Table 5). For influenza-specific IFN-γ⁺ CD8⁺ T-cell responses, statistically significant rises in T-cell responses from Day 0 to Day 30 were noted for the recombinant-HA vaccine against A(H1N1) and B/Victoria and for the MF59-adjuvanted vaccine against B/Victoria; no significant rises in the SD group were observed. IFN-γ⁺ CD8⁺ T cells to B/Victoria increased from Day 7 to Day 30 for all 3 enhanced vaccines; this trend was statistically significant when contrasted with a decline in T cells among SD vaccinees.

For influenza-specific IFN-γ⁺ CD4⁺ T-cell responses, statistically significant rises in T-cell responses were noted at 7 and/or 30 days postvaccination for the enhanced vaccines, though none of these were statistically significantly higher than the trend for the SD vaccine. Overall, the recombinant-HA vaccine caused sustained memory IFN-γ⁺ CD4⁺ T-cell responses by significantly increased Day 30 mean fold rises for 3 of 3 viruses tested, whilst the HD vaccine increased CD4⁺ T-cell Day 30 memory fold rises for 2 of 3 viruses and the MF59-adjuvanted vaccine resulted in early Day 7 rises for 2 of 3 viruses and sustained memory rises for only 1 of 3 viruses. SD group had a significant rise in CD4⁺ T-cell response to 1 virus at day 7 and no significant rises for all 3 viruses tested at Day 30.

The rate of contraction of T-cell responses from acute Day 7 to memory Day 30 was determined by the fold change from Day 30, divided by Day 7 (Supplementary Appendix Table 5). We did not observe statistically significant contractions (or expansions) for virus-specific CD4⁺ T cells for any vaccines. For virus-specific CD8⁺ T cells, a significant difference for the fold change from acute Day 7 to memory Day 30 responses was noted for B/Victoria only. Among those receiving the SD vaccine, B/Victoria-specific CD8⁺ T cells significantly contracted by 0.65-fold from 7 to 30 days postvaccination. In contrast, among those receiving the enhanced vaccines, B/Victoria-specific CD8⁺ T cells continued to increase by 1.34- to 1.40-fold from Day 7 to Day 30 (P values all <.05, compared to the standard vaccine group); therefore, enhanced vaccines maintained and expanded a larger pool of circulating B/Victoria-specific CD8⁺ T cells than the SD vaccine.

The most frequently reported adverse events following vaccination included local reactions, such as tenderness (range across vaccine groups, 12–20%), pain (10–22%) and swelling (4–10%; Supplementary Appendix Figure 4). Compared to the SD group, recipients of the MF59-adjuvanted vaccine and HD vaccine experienced more tenderness and recipients of the HD vaccine experienced more swelling, mostly of a mild nature. Hospitalizations were not uncommon, but the rates of postvaccination hospitalizations were not statistically different between the 4 vaccine groups (Supplementary Appendix Table 3). We did not identify any potentially vaccine-related serious adverse events.

We found that all 3 of the 2017–2018 Northern hemisphere–enhanced vaccines had improved immunogenicity for influenza A(H1N1) and A(H3N2), compared to the SD vaccine, among older adults, while the recombinant-HA vaccine elicited a particularly high antibody response to the cell-like H3N2 strain, relative to the other enhanced vaccines and the SD vaccine (Figure 2). Improved CD4⁺ T-cell responses against H3N2 were also seen at Day 7 or Day 30 for each enhanced vaccine (Figure 3). Influenza A(H3N2) virus infection causes the greatest morbidity and mortality effects in older adults [22] and there have been particular concerns in recent years over poorer vaccine effectiveness against influenza A(H3N2) [16, 23], which may be at least partially due to egg adaptations in A(H3N2) vaccine strains and/or poorer immunogenicity in repeated vaccinees [17, 24–27].

Enhanced vaccines resulted in higher antibody responses on almost all indicators, compared to a SD vaccine, to the A/Michigan/45/2015(H1N1)-like virus, which was introduced in the 2017–2018 season. The ratios of postvaccination titers against the A(H1N1) component for the MF59-adjuvanted vaccine and the HD vaccine, compared to the SD vaccine (1.37-fold higher and 1.80-fold higher, respectively), were very similar to pooled estimates in a recent meta-analysis for this ratio across previous immunogenicity trials (1.28 and 1.72, respectively) [5]. The antibody response to A(H1N1) among recombinant-HA recipients in our study was better than the A(H1N1) response among older adults in a previous recombinant-HA trial [15]. Our findings for A(H3N2) were also within the range of postvaccination ratios for the HD vaccine and MF59-adjuvanted vaccine, compared to the SD vaccine, when observed across prior trials [5]. Our findings reinforce other studies suggesting that enhanced vaccines offer improved antibody responses over standard vaccines in older adults. However, our study was not designed to demonstrate efficacy against polymerase chain reaction–confirmed influenza, and comparisons of clinical efficacy and effectiveness are still needed.

However, the findings for influenza B viruses were less clear. Antibody responses to the influenza B/Victoria-lineage component, which was in the TIV and QIV formulations in 2017–2018 and the previous season, were high for all vaccines. There were also significant improvements over the SD vaccine in CD8⁺ T-cell responses to B/Victoria (Figure 3). Not surprisingly, the quadrivalent SD vaccine resulted in higher antibody responses to the influenza B/Yamagata-lineage strain, compared to the trivalent MF59-adjuvanted and HD vaccines, which did not include this component; nonetheless, both vaccines were associated with significant rises in B/Yamagata titers (Table 3). The QIV recombinant-HA vaccine did not result in a higher antibody response to the influenza B/Yamagata lineage, compared to the QIV SD vaccine, consistent with a previous trial [13].

Our study had a number of limitations. First, vaccine immunogenicity is not equivalent to vaccine efficacy or effectiveness, nor is there yet a way to directly extrapolate our findings into an epidemiologic measure of differences in actual vaccine protection. Second, we only tested sera from a subset of 800 participants, determining that this would be sufficient to compare vaccines in the first year and allowing us to save more of the sera for future cross-season analyses. Two-thirds of participants were vaccinated during the previous season; examining the association between vaccination histories and immune response will also require additional future testing. Finally, here we only examined the immediate response to vaccination; we did not examine the duration of protection or effects on subsequent vaccinations with enhanced or SD vaccines.

In conclusion, this randomized, controlled trial provides a unique comparison of the immune response to multiple enhanced influenza vaccine options for older adults. Although distinguishing the relative preventive value of alternative vaccine types among older adults will ultimately depend on future efficacy trials and/or well-controlled effectiveness studies, the findings from this ongoing immunogenicity trial can help to establish the conditions for those more definitive evaluations, including what differences in efficacy/effectiveness might plausibly be expected.