Influence of individuals’ determinants including vaccine type on cellular and humoral responses to SARS-CoV-2 … – Nature.com

Identification of individual correlates of immunogenicity

Characteristics of the 115 participants with humoral and cellular data included in the analyses are presented in Table 1. The median age was 66.4 years (IQR 61.068.9), 47 (41%) males and 68 (59%) females, 57 (49.6%) had a BMI less than 25, 45 (39.1%) had a BMI of 2530 and a further 13 (11.3%) had a BMI of greater than 30. Of the participants 77 (66.9%) received ChAdOx1 and 38 (33.0%) received BNT162b2. When individuals were separated according to vaccine type received, the individual characteristics were similar (Supplementary Table 1). To identify demographic and technical factors associated with humoral and cellular responses to COVID-19 vaccination we first performed univariate analysis of ten factors: age, sex, ethnicity, general health category, vaccine type, number of days between 1st and 2nd vaccination [inter-vaccine days], days post second vaccine, pre-vaccine SARS-CoV-2 serostatus, BMI value, and BMI category. This analysis identified seven factors, which had a significant association with cellular and/or humoral immune responses: age, sex, pre-vaccine SARS-CoV-2 serostatus, BMI category, vaccine type, inter-vaccine days, and days post second vaccine (Supplementary Tables 2ai). These seven factors were adjusted for in subsequent analyses to identify independent associations.

First, we investigated whether there was any relationship between humoral and cellular immune responses we measured and post-vaccine anti-Spike and neutralising titres, adjusting for the seven baseline and post-vaccination covariates identified in univariate analyses described above. Post-vaccination titres of anti-Spike combined IgG/A/M antibody ratio responses were identified to be associated with neutralising antibody concentrations, three spike-specific CD4+T cell phenotypes and spike-stimulation induced IFN secretion (Fig. 1, Supplementary Table 3; quadratic regression for general linear models with adjustment for covariates). In addition, neutralising antibody concentrations significantly correlated with SARS-CoV-2-specific TNF+CD8+ T cell frequency (p=0.0003, Supplementary Table 4; quadratic regression for general linear models with adjustment for covariates). As expected, post-vaccination titres of neutralising and anti-Spike IgG/A/M antibody ratio correlated positively with each other (R=0.47, p<0.0001, Fig. 1a; Pearson correlation and quadratic regression for general linear models with adjustment for covariates respectively). In addition, anti-S IgG/A/M antibody ratio correlated positively with IFNy secretion from SARS-CoV-2 peptide-stimulated whole blood (R=0.31, p=0.001; Fig. 1b; Pearson correlation and quadratic regression for general linear models with adjustment for covariates respectively). The three spike-specific CD4+T cell populations from PBMC SARS-CoV-2 peptide-stimulated cultures that positively correlated with anti-Spike IgG/A/M antibody ratio post-vaccination were all double-positive cytokine producers: IFN-+IL2+ (R=0.37, p=0.0003; Pearson correlation and quadratic regression for general linear models with adjustment for covariates respectively); TNF+IL-2+ (R=0.35, p=0.0004; Pearson correlation and quadratic regression for general linear models with adjustment for covariates respectively); IFN-+TNF+ (R=0.32, p=0.001; Pearson correlation and quadratic regression for general linear models with adjustment for covariates respectively) (Fig. 1c). There was also a trend (q=0.1) for a positive correlation between anti-Spike IgG/A/M and SARS-CoV-2 peptide-specific CD4+IL-2 single positive T cells (Supplementary Table 3; quadratic regression for general linear models with adjustment for covariates). Significant correlations were maintained between anti-Spike combined IgG/A/M antibody and neutralising antibody concentrations and spike-stimulation induced IFN secretion when not including vaccine type as a covariate adjustment (R=0.56; Pearson correlation and Supplementary Fig. 4).

Correlation between post-COVID-19 vaccine anti-S IgG/A/M antibody ratio and a, neutralising antibody titre IC50, b, whole blood IFN production after S peptide stimulation and c, percent of cytokine positive CD4+ T after PBMC stimulation with S peptide as determined by intracellular cytokine staining. Coloured according to IFN production from S peptide-stimulated whole blood (SARS-S WB). Data presented on x and y axes are normalised including log2 transformation and adjusted for the baseline and post-vaccination covariates (age, sex, BMI category, pre-vaccine SARS-CoV-2 serostatus, vaccine type, vitamin D randomisation, inter-vaccine days, and days post second vaccine), p values derived using the quadratic regression for general linear models with adjustments for covariates, all q<0.01. Trend line indicates Pearson correlation (R-statistic).

We next determined whether there was a relationship between pre-vaccine anti-Spike IgG/A/M antibody ratio on post-vaccine humoral and cellular responses (Supplementary Table 5), including adjustment for baseline and post-vaccination covariates (age, sex, BMI category, vitamin D randomisation, vaccine type, inter-vaccine days, and days post second vaccine). Pre-vaccine anti-Spike IgG/A/M antibody ratios were significantly correlated with post-vaccine anti-Spike IgG/A/M antibody ratios (p<0.001; Pearson correlation and quadratic regression for general linear models with adjustment for covariates respectively) (Fig. 2a), frequency of IFN+CD4+ and IFN+CD8+SARS-CoV-2-specific T cells (p<0.001; quadratic regression for general linear models with adjustment for covariates) (Fig. 2b, c) and frequency of effector memory CD4+T cells in unstimulated PBMC (p=0.006; Supplementary Table 5; quadratic regression for general linear models with adjustment for covariates). Comparing those who were considered seropositive (anti-Spike IgG/A/M antibody ratio 1) vs seronegative pre-vaccination, post-vaccine anti-Spike IgG/A/M antibody ratios and frequency of IFN+CD4+SARS-CoV-2-specific T cells remained significantly different (p<0.001) (Supplementary Table 6; t-test for general linear models with adjustment for covariates). Together, these data demonstrate that individuals with baseline seropositivity due to prior SARS-CoV-2 infection had a stronger absolute cellular immune response following SARS-CoV-2 vaccination than those who were anti-S seronegative. These responses will be a combination of the original infection induced T cell memory and that expanded by the vaccination, as baseline PBMC were not taken, T cell expansion could not be assessed.

Correlation between pre-COVID-19 vaccine anti-S IgG/A/M antibody ratio and post-COVID-19 vaccine. a anti-S IgG/A/M antibody ratio, and frequency of IFNy+ b, CD4+ and c, CD8+ T cells after S peptide stimulation, b, c coloured according to anti-S IgG/A/M antibody ratio normalised to mean 0 and variance 1. Data presented on x and y axes are normalised including log2 transformation and adjusted for the baseline and post-vaccination covariates (age, sex, BMI category, vaccine type, vitamin D randomisation, inter-vaccine days, and days post second vaccine), p value derived using thequadratic regression for general linear models with adjustments for covariates, all q<0.01. Trend line indicates Pearson correlation (R-statistic).

Having identified cellular and humoral correlates of post-vaccination anti-S and neutralising antibody titre, irrespective of vaccine type, we next investigated whether there were any differences in cellular and humoral correlates between those who received BNT162b2 and ChAdOx1-nCoV-19. Adjusting for the other baseline and post-vaccination covariates, as above, we found BNT162b2 induced significantly higher anti-S IgG/A/M antibody ratio and neutralising antibody titres compared to ChAdOx1-nCoV-19 (Fig. 3a, b, Supplementary Table 7; t-test for general linear models with adjustment for covariates). There were, however, no significant differences in unstimulated or antigen-stimulated cellular responses that we measured in PBMC or whole blood between those who received BNT162b2 compared with those who received ChAdOx1-nCoV-19 (Supplementary Table 7). Separately analysing post-vaccination anti-S IgG/A/M antibody ratio correlations for each vaccine type, BNT162b2 had a stronger correlation compared to ChAdOx1-nCoV-19 with neutralising antibody titres (R=0.691 vs R=0.467, respectively; Pearson correlation) (Fig. 3c) and spike-stimulation induced IFN secretion (R=0.418 vs R=0.345, respectively) (Fig. 3d), although BNT162b2 Pearsons correlations were less significant due to smaller sample size compared to ChAdOx1-nCoV-19. Post-vaccination anti-S IgG/A/M antibody ratio correlations with the three polyfunctional spike-specific CD4+T cell populations previously identified, irrespective of vaccine type (Fig. 1c), showed stronger correlation for BNT162b2 with TNF+IL-2+CD4+T cells (R=0.51, vs R=0.31 ChAdOx1-nCoV-19; Pearson correlation) and IFN-+IL2+CD4+T cells (R=0.39, vs R=0.35 ChAdOx1-nCoV-19; Pearson correlation); whereas ChAdOx1-nCoV-19 had stronger for IFN-+TNF+CD4+T cells (R=0.33, vs R=0.14 BNT162b2; Pearson correlation) (Supplementary Fig. 5).

Cumulative data showing a anti-Spike IgG/A/M antibody ratio and b neutralising antibody titre IC50 in participants post-vaccination with either BNT162b2 or ChAdOx1-nCoV-19. Line indicated Median. Data points are plotted without covariate adjustment. Correlations plotted separately for participants who received either BNT162b2 (blue) or ChAdOx1-nCoV-19 (yellow) comparing post-vaccination anti-S IgG/A/M antibody ratio, and post-vaccination. c neutralising antibody (NAB) titre IC50. d whole blood IFN production after S peptide stimulation. Data presented on x and y axes are normalised including log2 transformation and adjusted for the baseline and post-vaccination covariates (age, sex, BMI category, pre-vaccine SARS-CoV-2 serostatus, vitamin D randomisation, inter-vaccine days, and days post second vaccine), p value derived using t-test (a and b) or quadratic regression (c and d) for general linear models with adjustment for covariates. Trend line indicates Pearson correlation (R-statistic).

When analysing whether there was an impact of the number of days between vaccinations and measured immune responses, we only identified a trend for the frequency of SARS-CoV-2 peptide-specific TNF+IFN+CD4+T cells and TNF+CD8+ T cells (p0.007; q=0.16) when adjusting for the other baseline and post-vaccination covariates (Supplementary Table 8; quadratic regression for general linear models with adjustment for covariates). However, we did find that the delay from the date of the second vaccine dose to the date of blood draw was positively correlated with the level of SARS-CoV-2 peptide-induced whole blood secretion of IL-6, IL-8 and TNF (p0.0004), and a trend for negative correlation with Neutralising antibody titres (p=0.01; q=0.14) (Supplementary Table 9; linear regression for general linear models with adjustment for covariates).

Having determined the vaccine type and timing variables independently associated with vaccine-induced humoral and cellular immune responses, we next analyzed demographic correlates adjusting for all other baseline and post-vaccination covariates as previously. We found that increasing age was independently associated with lower anti-S antibody titres post-vaccination (R=0.277, all participants adjusting for vaccine type; Pearson correlation) (Fig. 4a). There was no difference when analysing each vaccine type independently (R=0.26, BNT162b2; R=0.27, ChAdOx1-nCoV-19, Supplementary Fig. 6A; Pearson correlation). However, due to lower number of individuals who received BNT162b2, this did not reach statistical significance. For all participants, the frequency of polyfunctional spike-specific CD4+T cells from PBMC SARS-CoV-2 peptide-stimulated cultures post-vaccination also negatively correlated with increasing age for IFN-+IL-2+ (R=0.24, p=0.012; Pearson correlation and linear regression for general linear models with adjustment for covariates respectively) and TNF+IL-2+CD4+T cells (R=0.24, p=0.013; Pearson correlationand linear regression for general linear models with adjustment for covariates respectively) (Supplementary Table 10; and Fig. 4b). Supplementary Fig. 6B, C shows the same analysis separating by vaccine type, ChAdOx1-nCoV-19 had a stronger negative correlation for both CD4+T cell populations with age, compared to BNT162b2 (IFN-+IL-2+, R=0.34 vs R=0.15 and TNF+IL-2+, R=0.32 vs 0.14; Pearson correlation), although the larger and younger age range of those who received ChAdOx1-nCoV-19 may have improved the strength of correlation for ChAdOx1-nCoV-19. As would be expected, we also found increasing age was independently associated with lower frequency of naive CD8+ T cells (p<0.0001) and higher frequency of CD8+EM (p=0.004) and EMRA T cells (p=0.004). These data collectively show that increasing age was associated with reduced humoral immunity associated with reduced double-positive cytokine-producing spike-specific CD4+T cells, and vaccine type having no or only minor effect on these age-related differences.

a Correlation between age and post-vaccine anti-Spike IgG/A/M antibody ratio, coloured according to frequency of naive (CD45RA+CD27+) CD8+ T cells present in the peripheral blood of the individuals. b percent of cytokine positive CD4+T after PBMC stimulation with S peptide as determined by intracellular cytokine staining, coloured according to anti-S IgG/A/M antibody ratio normalised to mean 0 and variance 1. Data presented on y axis is normalised including log2 transformation and adjustment for baseline and post-vaccination covariates (sex, BMI category, pre-vaccine SARS-CoV-2 serostatus, vaccine type, vitamin D randomisation, inter-vaccine days, and days post second vaccine), p values derived using linear regression for general linear models with adjustments for the same covariates, all q<0.1. Trend line indicates Pearson correlation (R-statistic).

Finally, we tested for associations between sex and BMI with SARS-CoV-2 cellular and humoral immunity, with analyses adjusted for baseline and post-vaccination co-variates as previously. We found a highly significant lower frequency of SARS-CoV-2-specific TNF+CD4+ and CD8+ T cells (p<0.0001) correlated with increasing BMI category (Supplementary Table 11; ANOVA for general linear models with adjustment for covariates), as well as trends of lower SARS-CoV-2-specific IL-2+CD4+T cells and higher CRP (p0.014, q<0.156). There was no significant association between higher BMI category and anti-Spike or neutralising antibodies. (Supplementary Table 11; ANOVA for general linear models with adjustment for covariates), despite our finding that the level of neutralising antibodies significantly correlated with SARS-CoV-2-specific TNF+CD8+ T cells (Supplementary Table 5), which were decreased with increasing BMI. This remained true if we analysed each vaccine type separately.

When analysing associations between sex immune correlates, the most significant difference was in our control assay with higher LPS induced IL-6 secretion in whole blood stimulated plasma (p<0.0001). Males also had a higher frequency of CD8+ (p=0.0001) and CD4+ (p=0.011) EM T cells, whilst females had a higher frequency of naive CD4+ (p=0.012) and CD8+ (p=0.002) T cells (Supplementary Table 12; t test for general linear models with adjustment for covariates). However, these differences did not impact upon SARS-CoV-2-specific cellular or humoral responses post-vaccination between males and females with no significant differences observed (Supplementary Table 11). Collectively these data show sex had no impact on SARS-CoV-2 vaccine immunogenicity, whilst BMI had significant effects on single cytokine producing SARS-COV-2-specific T cell functions we independently identified to be associated with SARS-CoV-2 neutralising antibody concentrations post-vaccination.

See more here:

Influence of individuals' determinants including vaccine type on cellular and humoral responses to SARS-CoV-2 ... - Nature.com