Comparative Immunogenicity of COVID-19 Vaccines in a Population-Based Cohort Study with SARS-CoV-2-Infected and Uninfected Participants

Abstract

To assess vaccine immunogenicity in non-infected and previously infected individuals in a real-world scenario, SARS-CoV-2 antibody responses were determined during follow-up 2 (April 2021) of the population-based Tirschenreuth COVID-19 cohort study comprising 3378 inhabitants of the Tirschenreuth county aged 14 years or older. Seronegative participants vaccinated once with Vaxzevria, Comirnaty, or Spikevax had median neutralizing antibody titers ranging from ID50 = 25 to 75. Individuals with two immunizations with Comirnaty or Spikevax had higher median ID50s (of 253 and 554, respectively). Regression analysis indicated that both increased age and increased time since vaccination independently decreased RBD binding and neutralizing antibody levels. Unvaccinated participants with detectable N-antibodies at baseline (June 2020) revealed a median ID50 of 72 at the April 2021 follow-up. Previously infected participants that received one dose of Vaxzevria or Comirnaty had median ID50 to 929 and 2502, respectively. Individuals with a second dose of Comirnaty given in a three-week interval after the first dose did not have higher median antibody levels than individuals with one dose. Prior infection also primed for high systemic IgA levels in response to one dose of Comirnaty that exceeded IgA levels observed after two doses of Comirnaty in previously uninfected participants. Neutralizing antibody levels targeting the spike protein of Beta and Delta variants were diminished compared to the wild type in vaccinated and infected participants.

Keywords: Comirnaty; SARS-CoV-2; Spikevax; Vaxzevria; immunogenicity; neutralizing antibodies; population-based cohort; vaccination; variants of concern VoC.

Figure 1

Antibody responses in infected participants. Shown are antibody levels to the receptor-binding domain (RBD) of the S protein (Roche Elecsys Anti-SARS-CoV-2 S test; arbitrary units AU) (A) and neutralizing antibody levels (ID50) (B) in seropositive participants receiving the indicated number of doses of Vaxzevria (VX), Comirnaty (COR), or SpikeVax (SV). Median and interquartile range are indicated by red bars. The Kruskal–Wallis test including all five groups with more than six participants indicated significant differences for the RBD and neutralizing antibody levels (p < 0.0001 for both, judged at 5% significance level). Pairwise comparisons were performed to (i) analyze a potential waning of immune responses after infection (0 dose T3 vs. 0 dose T1), (ii) determine and compare the effect of one dose of Vaxzevria and Corminaty (0 dose T1 vs. 1 dose VX vs. 1 dose COR) and (iii) explore the effect of a second dose of Cormirnaty (1 dose COR vs. 2 dose COR). p-values of these pairwise group comparisons with the Mann–Whitney test are shown (judged for significance at a Bonferroni-corrected level of 0.05/5 = 0.01 to account for 5 tests). N antibody (ab) status: T1: participants that were N-antibody positive already at the June 2020 visit; T3: participants negative for N antibodies at the June and November 2020 visits, but positive at the April 2021 visit. Dashed lines in (A) mark the upper and lower limits of the linear range of the assay, the dashed line in (B) mark the upper and lower limits of detection. Supplementary Tables S1 and S2 provide number of participants, basic demographic characteristics and the median time interval between the last immunization and blood sampling. # doses indicates the number of applied doses.

Figure 2

RBD binding and neutralizing antibodies in uninfected participants after vaccination. (A) Binding antibody levels to RBD (arbitrary units, AU) and (B) neutralizing antibodies (ID50) in N-antibody seronegative participants vaccinated once (1×) or twice (2×) with Vaxzevria (VX), Comirnaty (COR), or Spikevax (SV). The Kruskal–Wallis test including all five groups indicated significant differences for the RBD and neutralizing antibody levels ((A) p < 0.0001, (B) p < 0.0001, judged at 5% significance level). Pairwise comparisons were performed to (i) compare antibody response to one vaccination between vaccine types (COR vs. VX, SV vs. VX, SV vs. COR), (ii) compare two versus one vaccination (COR-2x vs. COR-1x, SV-2x vs. SV-1x) and (iii) compare two vaccinations of different vaccine types (SV-2x vs. COR-2x). p-values of these pairwise group comparisons with the Mann–Whitney test are shown (judged for significance at a Bonferroni-corrected level of 0.05/6 = 0.008 to account for 6 tests). Dashed lines mark the upper and lower limits of the linear range of the assay.

Figure 3

Model-based RBD binding and neutralization antibody levels related to age and time since vaccination. We applied a generalized additive model with assumed log-normal distribution of the antibody levels and non-linear associations with age and time since vaccination on the cross-sectional data from T3. On the example of individuals vaccinated twice with COV, we show predicted levels of (A) RBD binding antibodies (arbitrary units AU/mL) and (B) neutralizing antibodies (ID50) for the age of 30, 55, and 80 years and over the observed range of days since second vaccination (n = 272 and 269, respectively). Shaded regions indicate approximate 95%-prediction intervals (Prediction ± 2×SE).

Figure 4

Breadth of neutralization after infection or vaccination. We randomly selected four groups of participants totaling n = 120 (30 infected and unvaccinated (A), 90 vaccinated twice with Comirnaty from three age groups (B–D)). The neutralization activity was determined in a lentiviral pseudotype assay using the spike proteins of wild type (D614G) SARS-CoV-2 and the indicated VOCs. Shown are the reciprocal of the 50% inhibitory dilution (ID₅₀) of each serum sample by WT or VOCs. Person-specific measurements for WT and VOCs are connected by gray lines. Dashed black lines indicate the lower limit of detection (ID₅₀ = 20) and median ID₅₀ values per group are stated below this line. We conducted paired tests (Wilcoxon rank-sum test) comparing VOCs with WT and indicate significant differences in the medians by *, *** for p-values of <0.01 and <0.0001, respectively.

Figure 5

IgA antibody levels. Shown are IgA signal to cut-off ratios (S/Co) for serum samples from the indicated groups of participants. Numbers on top of the graph give the percentage of IgA-positive participants of each group. Fisher’s exact tests were used for pairwise comparison of all groups without correcting for multiple testing (see Supplementary Table S5) and judged at Bonferroni-corrected significance of 0.05/55 = 9.09 × 10⁻⁴. For clarity reasons, highly significant differences (p < 10⁻¹²) between groups are only shown for comparisons of participants that were N-antibody (ab) seropositive at the June 2020 time point (N ab status: T1) and vaccinated once with Comirnaty (COR). T3: participants negative for N antibodies at the June and November 2020 visits, but positive at the April 2021 visit. N ab status: T0: participants seronegative for N antibodies at all study time points analyzed. VX: Vaxzevria; SV: Spikevax; The dashed line separates IgA seropositive from seronegative samples. Supplementary Tables S5 and S6 provide p-values of all pairwise comparisons as well as number of participants, basic demographic characteristics and the median time interval between the last immunization and blood sampling. # doses indicates the number of applied doses.