Influenza is a major public health burden causing millions of illnesses and outpatient visits and resulting in tens of thousands of hospitalizations and deaths in the U.S. annually [1]. Vaccination is the primary influenza prevention method. Advances in vaccine technology have resulted in multiple vaccine types licensed for use in the U.S., comprising intramuscularly administered inactivated influenza vaccines (IIV), including a quadrivalent cell culture-based inactivated vaccine (ccIIV4), and nasally administered, egg-based live attenuated influenza vaccine (LAIV4). The Advisory Committee on Immunization Practice (ACIP), which sets civilian immunization policy for the United States, recommends vaccination with any licensed vaccine, without a stated preference [2].

Several factors can affect the immune response to either vaccine type, including pre-existing antibodies, prior vaccination status, previous infection, and age [3–5]. Nasally administered LAIV4, which is absorbed through the nasal mucosa, generates an immune response that differs from direct systemic exposure from intramuscularly administered IIV [6]. Moreover, the manufacturing process may have an impact on immune response. For example, the viruses used for vaccine antigens can be grown in mammalian cell culture or embryonated chicken eggs. Growth in cell culture has the advantage of limiting the virus mutations that often occur during the egg-based manufacturing process. While, egg-based LAIV4 has been shown to be more effective against antigenically different or drifted strains [7], egg-induced mutations in influenza A(H3N2) vaccine viruses have resulted in reduced efficacy of egg-based vaccines [8].

Previous research comparing the hemagglutination inhibition (HAI) antibody responses of IIV and LAIV4 has historically demonstrated lower HAI titers in response to LAIV compared to egg-based IIV [3,6,9,10]. These studies were conducted before the reformulation of LAIV4 necessitated by poor vaccine effectiveness [11]. In a previous study among children, we found no differences in seroconversion in children receiving egg- or cell-grown IIV in 2018–19 [12]. The cell-grown IIV4 used in that study contained A/H1N1 that was derived from an egg-based seed leading to a 3:1 cell to egg formulation. The 2019–20 season is the first in which the ccIIV4 formulation is exclusively cell-based. Furthermore, the current LAIV4 reformulation and interest in avoiding egg-induced mutations to the A/H3N2 antigen should be considered. Thus, an assessment of the HAI response to A/H3N2 egg- and cell-based antigens for each vaccine in a clinical setting of children and young adults is warranted.

The purpose of this randomized controlled trial was to quantify and compare pre- and post-vaccination HAI responses to 2019–20 LAIV4 and ccIIV4, among a racially diverse group of children and young adults during the 2019–20 influenza season.

The Institutional Review Boards at the University of Pittsburgh and the Centers for Disease Control and Prevention (CDC) approved this study. Written informed consent and assent, where appropriate, were obtained from all participants and/or their parents/legal guardians prior to beginning study procedures.

Of 234 persons aged 4 to 21 years assessed for eligibility, 16 were excluded and 218 were randomized: 110 received ccIIV4 and 108 LAIV4. After randomization, 20 participants were excluded from the per-protocol analysis: 19 failed to return for post-vaccination specimen collection or were missing paired sera, and one participant randomized to the ccIIV4 group received a non-study IIV4. Per-protocol analysis included 198 individuals: 100 received ccIIV4 and 98 received LAIV4 (Fig. 1).

Patient characteristics were similar for the two vaccine arms (Table 1). Overall, 55% of participants were age 18–21 years, 63% were female, 48% were non-white and 16% were obese. Nearly half of participants were publicly insured, 14% were exposed to household smoking and the majority of parents had less than college education. Differences in prior season influenza vaccination (52% vs 47%) were not statistically significant (p = 0.48).

Pre- and post vaccination GMTs and GMFR are shown by vaccine group in Table 2. Pre-vaccination titers were similar by vaccine group across all antigens; 72%−85% had titers ≥ 1:40 and 14%−61% had titers ≥ 1:110. After vaccination, GMTs increased significantly against all antigens except among LAIV4 recipients against B/Phuket (Yamagata lineage). All measures of post-vaccination titers, i.e., GMTs, GMFR and proportions with HAI titers ≥ 1:40 and ≥ 1:110, were significantly higher among ccIIV4 recipients than LAIV4 recipients across all antigens (all p ≤ 0.02; Table 2). There was no significant difference between GMFR in antibody titer for A/H3N2 tested against cell-grown antigen and A/H3N2 tested against egg-grown antigen. Responses to all antigens in ccIIV4 and LAIV4 recipients for Day 0, Day 28 and GMFR are shown in Fig. 2.

Day 28 HAI titers of ≥ 1:40 and ≥ 1:110 to all antigens were higher in response to ccIIV4 compared to LAIV4 (Table 2, p < 0.05). Seroconversion with titers ≥ 1:40 occurred in<12% of all LAIV4 participants compared to 28–47% of ccIIV4 participants. Seroconversion ≥ 1:110 occurred in<9% of all LAIV4 participants compared to 22–39% of ccIIV4 participants. Among participants seronegative to each antigen at baseline (Table 3), HAI titers were higher at Day 28 than Day 0 for both vaccines. Day 28 HAI titers, GMFR, and seroconversion to ≥ 1:40 were higher in ccIIV4 recipients. Among this subset, seroconversion to ≥ 1:40 was observed in 64–100% ccIIV4 recipients and 14–33% for LAIV4 recipients. Regression analyses for seroconversion ≥ to 1:40 were controlled for baseline titer, age, and prior year vaccination status because those vaccinated in the prior season were older (19.4 years, 95% CI 18.8–20.0) than those unvaccinated in the prior season (14.5 years, 95% CI 13.7–15.2, p = 0.02). Controlling for these factors among those with HAI titers < 1:40 at baseline for each antigen by each vaccine type, baseline HAI titers and age did not predict seroconversion (Supplemental Table 1). Prior year vaccination predicted seroconversion to ≥ 1:40 for ccIIV4 for B/Colorado-B/Victoria (ß = −0.82, 0.001) p < and A/Kansas cell grown antigens (ß = −0.48, p = 0.03).

Figs. 3a and 3b show patterns of antibody response to vaccination for all antigens. For both ccIIV4 and LAIV4, antibody titers for most participants increased or remained similar to prevaccination; greater than two-fold decreased titers were observed in few participants in both vaccine groups. However, patterns of antibody response to vaccination differed significantly by vaccine type across all antigens, including against cell-grown and egg-grown A/H3N2 antigens. For the ccIIV4 recipients, a strong “ceiling” effect was observed, with post vaccination titers reaching similar levels across a range of prevaccination titers; whereas antibody titers among LAIV recipients increased more variably from pre- to post vaccination. In regression analyses, lower pre-vaccination titers were associated with greater GMFR in titers in both vaccine groups, although the association was much stronger in the ccIIV4 recipients (Supplemental Table 2). Controlling for pre-vaccination HAI titer, and age, prior season (2018–19) vaccination was associated with decreased GMFR against B/Colorado-B/Victoria in ccIIV4 recipients (ß =−1.21, p < 0.001) and B/Yamagata-B/Phuket (ß =−0.53, p < 0.05). With similar adjustments, prior season vaccination was associated with increased GMFR against A/H3N2-A/Kansas egg-grown antigens (ß =0.18, p < 0.01) in LAIV4 recipients, but was not independently associated with GMFR against other antigens.

This randomized controlled trial compared antibody responses to cell-grown (ccIIV4) and egg-grown (LAIV4) influenza vaccine in a racially diverse group of children and young adults. Compared to previous work, which used a ccIIV4 formulation with three cell-grown antigens and one egg-grown antigen (A/H1N1), this study used a ccIIV4 composed entirely of cell-based antigens. LAIV4 was composed entirely of egg-grown antigens. A/H3N2 strains have become increasingly glycosylated as a result of adaptations required for growth in eggs. These adaptations have led to antigenic differences between vaccine strains and the wild virus [16] and reduced vaccine effectiveness. Thus, cell-grown vaccine strains offer the potential to avoid HA protein mutations that occur in egg-based systems [17].

For all measures of HAI response, ccIIV4 elicited significantly larger increases than LAIV4 at 28 days post vaccination. For LAIV4, HAI response was attenuated and seroconversion was infrequent. While titers of ≥ 1:40 have long been used as a correlate of protection of 50% protection from influenza infection in response to influenza vaccines, titers of ≥ 1:110 have been shown to be required to reach this level or protection in children ages 6–72 months [14], and attainment of these levels was more common in response to ccIIV4. However, to date no HAI marker as a correlate of protection for LAIV4 [18].

Lower HAI responses to LAIV compared to IIV have been reported in prior seasons [3,6,9,10]. Our findings with the 2019–20 LAIV4 formulation and fully cell-based ccIIV4 follow a similar pattern. Prior studies have similarly shown a lower HAI response to LAIV in countries where influenza vaccines are routinely administered, although a few earlier studies showed a significant HAI response to LAIV prior to the initiation of routine seasonal vaccination in children [18]. Because of different routes of administration for the two vaccine types, secretory IgA and cellular immune processes stimulated by administration via the nasal route may be a better measure of immune response than HAI for LAIV4 influenza vaccine [6,10,19]. Measures of immunoglobulins (IgG and IgA) and cellular immune responses would help delineate the potential for differences between the vaccine groups not observed in HAI titers alone.

Despite the lower rise in HAI titers typically found in response to LAIV, vaccination with LAIV is effective at limiting influenza infection [7,11]. This disconnection between serological measures of immunity in response to LAIV and vaccine effectiveness suggests other aspects of the immune response may play a greater role for vaccine efficacy for LAIV. For example, mucosal immunity due to the route of administration and cellular immunity may play a role in vaccine efficacy in response to LAIV [10,18]. Differences in inflammatory mediator expression [9] and gene expression [20] in comparisons of the immune response to LAIV and IIV have also been reported.

When comparing the relationship of pre- to post vaccination titers, with ccIIV4 in particular, those with lower pre-vaccination titers had the greatest increase. For both ccIIV4 and LAIV4, participants with high baseline titers remained high. Pre-vaccination titer was predictor of post-vaccination titer (positively correlated) and GMFR (negatively correlated). We found similar relative GMFR increases in A/H3N2 cell- and egg-grown antigens within vaccine type in the 2019–20 season. Analyses to determine predictors of seroconversion among those seronegative at baseline did not show an effect of age, after controlling for prior year vaccination status, on rates of seroconversion. In this subgroup, prior year vaccination predicted a lower rate of seroconversion to B/Colorado (B/Victoria), a component of the 2018–19 vaccine in the ccIIV4 recipients.

A prior study in 2018–19, comparing ccIIV4 with 3 cell-grown and 1 egg-grown strains and LAIV4, which incorporated microneutralization measures, found that the GMFR measured using egg-grown A/H3N2 antigen was higher for egg-based vaccine than cell-based vaccine recipients [12]. In the current study, GMFR responses to egg- and cell-grown A/H3N2 antigens were similar within vaccine type. Thus, the question of response based on the antigen growth medium is still unresolved.

Due to the known effect of baseline titers and prior vaccination history on vaccine response [19], we specifically recruited participants who were not vaccinated in the prior season anticipating that they may have lower titers pre-vaccination and be more likely to mount an HAI response. In post hoc analyses, pre-vaccination titer was a negative predictor of vaccine response for both vaccines, with a greater rise in titer among those with lower prevaccination titers. Despite the large proportion of participants unvaccinated in the prior year, only the HAI response to B/Colorado-B/Victoria and B/Yamagata-B/Phuket, components of both the 2018–19 and 2019–20 vaccines, had an independent associations with a smaller GMFR. For LAIV4, a slightly higher response to A/H3N2-A/Kansas egg-grown antigens was for those vaccinated in the prior season was found. Pre-vaccination titers as well as prior year vaccination status should be considered in determining degree of vaccine response.