Immunogenicity and reactogenicity of SARS-CoV-2 vaccines BNT162b2 and CoronaVac in healthy adolescents

Abstract

We present an interim analysis of a registered clinical study (NCT04800133) to establish immunobridging with various antibody and cellular immunity markers and to compare the immunogenicity and reactogenicity of 2-dose BNT162b2 and CoronaVac in healthy adolescents as primary objectives. One-dose BNT162b2, recommended in some localities for risk reduction of myocarditis, is also assessed. Antibodies and T cell immune responses are non-inferior or similar in adolescents receiving 2 doses of BNT162b2 (BB, N = 116) and CoronaVac (CC, N = 123) versus adults after 2 doses of the same vaccine (BB, N = 147; CC, N = 141) but not in adolescents after 1-dose BNT162b2 (B, N = 116). CC induces SARS-CoV-2 N and N C-terminal domain seropositivity in a higher proportion of adolescents than adults. Adverse reactions are mostly mild for both vaccines and more frequent for BNT162b2. We find higher S, neutralising, avidity and Fc receptor-binding antibody responses in adolescents receiving BB than CC, and a similar induction of strong S-specific T cells by the 2 vaccines, in addition to N- and M-specific T cells induced by CoronaVac but not BNT162b2, possibly implying differential durability and cross-variant protection by BNT162b2 and CoronaVac, the 2 most used SARS-CoV-2 vaccines worldwide. Our results support the use of both vaccines in adolescents.

Fig. 1. Humoral immunogenicity outcomes for adolescents were mostly non-inferior in adolescents in comparison to adults.

a One dose of BNT162b2 (B) in adolescents was non-inferior by S-RBD IgG (adolescent B N = 107, adult BB N = 115), sVNT (adolescent B N = 107, adult BB N = 115) and S IgG avidity (adolescent B N = 88, adult BB N = 114) but not by S IgG (adolescent B N = 101, adult BB N = 115), PRNT90 (adolescent B N = 63, adult BB N = 13), PRNT50 (adolescent B N = 63, adult BB N = 13) and S IgG FcγRIIIa-binding (adolescent B N = 101, adult BB N = 115) (all P < 0.0001, except S IgG avidity with P = 0.13), which failed the non-inferiority comparison to adults. Additionally, although non-inferiority was satisfied for S-RBD IgG and sVNT for B in adolescents, their CIs were also within the inferior ranges (both P < 0.0001). b In contrast, humoral responses in adolescents were non-inferior to adults after 2 doses of BNT162b2 (BB), as measured by S IgG (also superior, P = 0.0036, adolescent BB N = 103), S-RBD IgG (P = 0.23, adolescent BB N = 103), sVNT (also superior, P < 0.0001, adolescent BB N = 104), PRNT90 (also superior, P = 0.018, adolescent BB N = 60), PRNT50 (P = 0.25, adolescent BB N = 60), S IgG avidity (also superior, P < 0.0001, adolescent BB N = 103) and S IgG FcγRIIIa-binding (also superior, P = 0.0005, adolescent BB N = 103). c After 2 doses of CoronaVac (CC), adolescents also had non-inferior humoral responses to adults as assessed by S IgG (also superior, P = 0.0049, adolescent CC N = 116, adult CC N = 50), S-RBD IgG (P = 0.96, adolescent CC N = 119, adult CC N = 51), sVNT (also superior, P < 0.0001, adolescent CC N = 119, adult CC N = 51), PRNT90 (P = 0.08, adolescent CC N = 64, adult CC N = 19), PRNT50 (P = 0.12, adolescent CC N = 64, adult CC N = 19), S IgG avidity (also superior, P < 0.0001, adolescent CC N = 109, adult CC N = 41) and S IgG FcγRIIIa-binding (P = 0.086, adolescent CC N = 116, adult CC N = 50). Additionally for adolescent CC, N and N-CTD IgGs were non-inferior and superior for adolescents compared to adults (both P < 0.0001, adolescent CC N = 60, adult CC N = 36). GMR geometric mean ratio, CI confidence interval, S spike protein, RBD receptor-binding domain, N nucleocapsid protein, CTD C-terminal domain, sVNT surrogate virus neutralisation test, PRNT plaque reduction neutralisation test, FcγRIIIa Fcγ receptor IIIa.

Fig. 2. Cellular immunogenicity outcomes for adolescents were mostly non-inferior or inconclusive in adolescents in comparison to adults.

a, b Adolescents receiving one dose of BNT162b2 (B; N = 58) and 2 doses of BNT162b2 (BB; N = 56) and c adolescents receiving 2 doses of CoronaVac (CC; N = 60) were tested for IFN-γ⁺ and IL-2⁺ CD4⁺ and CD8⁺ T cells on flow-cytometry-based intracellular cytokine staining assays specific to S (and N and M for CC) for 21 days after dose 1 and 28 days after dose 2. The results of SNM-specific T cell responses were calculated from the sum of responses of the individual S, N and M peptide pools. S-specific IFN-γ⁺CD4⁺, IL-2⁺CD4⁺ and IFN-γ⁺CD8⁺ T cell responses were non-inferior for adolescent BB in comparison to adults (N = 47). For adolescent CC compared to adults (N = 36), SNM-specific IL-2⁺CD4⁺, IFN-γ⁺CD8⁺ and IL-2⁺CD8⁺, N-specific IFN-γ⁺CD4⁺, IL-2⁺CD4⁺, IFN-γ⁺CD8⁺ and IL-2⁺CD8⁺, M-specific IFN-γ⁺CD8⁺ and IL-2⁺CD8⁺ were non-inferior. The remaining cellular immunogenicity outcomes were inconclusive. Dots and error bars show GMR estimates and two-sided 95% CI respectively. GMR geometric mean ratio, CI confidence interval, S spike protein, N nucleocapsid protein, M membrane protein, IFN-γ interferon-γ, IL-2 interleukin-2.

Fig. 3. Antibody levels against S were higher for BNT162b2 than CoronaVac in adolescents.

Humoral and cellular immunogenicity was compared between vaccines in adolescents at 21-28 days after 1 dose and 28 days after 2 doses. a There were lower humoral responses after CC than BB as measured by S IgG (adolescent CC N = 116, adolescent BB N = 103) (GM OD450 0.54 vs 1.21; GMR 0.44, 95% CI 0.40–0.49), S-RBD IgG (adolescent CC N = 119, adolescent BB N = 104) (GM OD450 1.20 vs 2.64; GMR 0.46, 95% CI 0.41-0.50), sVNT (adolescent CC N = 119, adolescent BB N = 104) (GM % inhibition 71.2% vs 97.1%; GMR 0.73, 95% CI 0.68-0.79), PRNT90 (adolescent CC N = 64, adolescent BB N = 60) (GM PRNT90 9.58 vs 115; GMR 0.08, 95% CI 0.07–0.11), PRNT50 (adolescent CC N = 64, adolescent BB N = 60) (GM PRNT50 28.0 vs 331; GMR 0.08, 95% CI 0.07–0.11), S IgG avidity index (adolescent CC N = 109, adolescent BB N = 103) (GM % avidity 20.5% vs 29.7%; GMR 0.69, 95% CI 0.63–0.76) and S IgG FcγRIIIa-binding (adolescent CC N = 116, adolescent BB N = 103) (GM OD450 0.75 vs 2.07; GMR 0.36, 95% CI 0.31–0.42) (all P < 0.0001). Most outcomes except S IgG avidity were also lower in C compared to B. b Cellular immunogenicity outcomes were similar between vaccine types except for the S-specific IL-2⁺CD4⁺ T cell response (adolescent CC N = 60, adolescent BB N = 56), which was lower after CC (GM % T cells 0.015% vs 0.032%; GMR 0.45, 95% CI 0.28-0.72) (P = 0.001). Nucleocapsid (N) and membrane (M)-specific T cell responses were not compared since only CC but not BB had induced non-spike responses, as expected. Data labels and centre lines show GM estimates, and error bars show 95% CI. P-values were derived from two-tailed unpaired t test after natural logarithmic transformation. GM geometric mean, GMR geometric mean ratio, CI confidence interval, B 1 dose of BNT162b2, BB 2 doses of BNT162b2, C 1 dose of CoronaVac, CC 2 doses of CoronaVac, S spike protein, RBD receptor-binding domain, sVNT surrogate virus neutralisation test, PRNT plaque reduction neutralisation test, FcγRIIIa Fcγ receptor IIIa, IFN-γ interferon-γ, IL-2 interleukin-2. ***P < 0.001; ****P < 0.0001.

Fig. 4. Significant increases in N- and M-specific T cell responses after CoronaVac in adolescents.

a For adolescents who received each vaccine, when compared to their own baseline values, BB (N = 56) and CC (N = 60) had significant increases in T cell responses for S-specific IFN-γ⁺CD4⁺, IL-2⁺CD4⁺ and IFN-γ⁺CD8⁺ (all P < 0.0001). Additionally, a significant increase in S-specific IL-2⁺CD8⁺ T cells was observed for CC (P = 0.023). b When added together, SNM-specific IFN-γ⁺CD4⁺ (P < 0.0001), IL-2⁺CD4⁺ (P < 0.0001) and IFN-γ⁺CD8⁺ T cells (P < 0.0001) increased significantly for CC (N = 60). c These marked increases were likely due to post-CC’s combined increases in S-specific T cell responses as well as N-specific increases in IFN-γ⁺CD4⁺ (P < 0.0001), IL-2⁺CD4⁺ (P < 0.0001), IFN-γ⁺CD8⁺ (P = 0.042) and d M-specific IL-2⁺CD4⁺ (P = 0.021). On the other hand, no significant N- and M-specific T cell responses were elicited by BB, an expected result. Centre lines show GM estimates, and error bars show 95% CI. P-values were derived from two-tailed paired t test after natural logarithmic transformation. GM geometric mean, CI confidence interval, B 1 dose of BNT162b2, BB 2 doses of BNT162b2, C 1 dose of CoronaVac, CC 2 doses of CoronaVac, S spike protein, N nucleocapsid protein, M membrane protein, IFN-γ interferon-γ, IL-2 interleukin-2. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Fig. 5. Vaccine efficacy estimates based on neutralising antibody titres for BNT162b2 (after 1 dose or 2 doses) and CoronaVac (after 2 doses) were ≥50% in adolescents.

Neutralising antibodies have been established as a reliable correlate of protection that can predict VEs against symptomatic COVID-19. The mean neutralising levels (fold of convalescent) were derived by dividing the geometric mean titres of PRNT90 in healthy evaluable adolescents who received the vaccines with that of 102 convalescent sera collected on days 28–59 post-onset of illness in patients aged ≥18 years. A point estimate of VE was extrapolated from the best fit of the logistic model in Khoury et al.^,,. Adolescent B has been considered completion of primary series, but not adolescent CC or adult B, for a time period in HK and the UK due to elevated myocarditis risks after youths received 2 doses of BNT162b2. Therefore, the VE of adolescent B, but not adolescent C or adult CC, was also extrapolated, along with adolescent BB and CC. The mean neutralisation levels (fold of convalescent) for adolescents after receiving 2 doses of BNT162b2, 2 doses of CoronaVac and 1 dose of BNT162b2 were 2.39, 0.20 and 0.30, respectively. Extrapolation of these mean neutralisation levels using the logistic model resulted in VEs of 93% after 2 doses of BNT162b2, 50% after 2 doses of CoronaVac and 59% after 1 dose of BNT162b2. VE, vaccine efficacy.

Fig. 6. Adverse reactions 7 days after each dose of BNT162b2 and CoronaVac were solicited from adolescents in the healthy safety population.

In the adolescent healthy safety population, pain at the injection site was the most common adverse reaction (ARs) reported for both vaccines, which was significantly more for those who received BNT162b2 (N = 116) than CoronaVac (N = 123) (B: 89.7% vs C: 54.5%, P < 0.0001; BB: 87.9% vs CC: 52.9%, P < 0.0001). BNT162b2 was also associated with more reporting of several other ARs, including swelling, erythema, induration and pruritis at the injection site, headache, fatigue, myalgia, nausea, diarrhoea, vomiting, arthralgia, chills, fever, reduced appetite, and abdominal pain. More participants had antipyretics use after either dose of BNT162b2 than CoronaVac (B: 9.5% vs C: 1.6%, P = 0.009; BB: 22.4% vs CC: 0.8%, P < 0.0001). Data are shown as percentages and error bars show two-sided 95% CI of the total frequency of the respective AR of any severity. CI confidence interval, B 1 dose of BNT162b2, BB 2 doses of BNT162b2, C 1 dose of CoronaVac, CC 2 doses of CoronaVac. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.