SARS-CoV-2 booster vaccine dose significantly extends humoral immune response half-life beyond the primary series

Abstract

SARS-CoV-2 lipid nanoparticle mRNA vaccines continue to be administered as the predominant prophylactic measure to reduce COVID-19 disease pathogenesis. Quantifying the kinetics of the secondary immune response from subsequent doses beyond the primary series and understanding how dose-dependent immune waning kinetics vary as a function of age, sex, and various comorbidities remains an important question. We study anti-spike IgG waning kinetics in 152 individuals who received an mRNA-based primary series (first two doses) and a subset of 137 individuals who then received an mRNA-based booster dose. We find the booster dose elicits a 71-84% increase in the median Anti-S half life over that of the primary series. We find the Anti-S half life for both primary series and booster doses decreases with age. However, we stress that although chronological age continues to be a good proxy for vaccine-induced humoral waning, immunosenescence is likely not the mechanism, rather, more likely the mechanism is related to the presence of noncommunicable diseases, which also accumulate with age, that affect immune regulation. We are able to independently reproduce recent observations that those with pre-existing asthma exhibit a stronger primary series humoral response to vaccination than compared to those that do not, and further, we find this result is sustained for the booster dose. Finally, via a single-variate Kruskal-Wallis test we find no difference between male and female humoral decay kinetics, however, a multivariate approach utilizing Least Absolute Shrinkage and Selection Operator (LASSO) regression for feature selection reveals a statistically significant (p $<1\times {10}^{-3}$ ), albeit small, bias in favour of longer-lasting humoral immunity amongst males.

Keywords: Asthma; Booster dose; Humoral immunity; Immunogenicity; Immunosenescence; SARS-CoV-2.

Figure 1

Primary series and booster dose model fit results. (A) Example individual model fits (Eq. 1) for the primary series (blue) and booster dose (orange) Anti-S trajectory data. Dashed lines for D1, D2, and D3 indicate timing of dose 1, dose 2, and dose 3, respectively. (B) Distribution of all individual half lives (Eq. 2) for primary series (blue) and booster dose (orange). (C) Individual Anti-S half life as a function of chronological age for primary series and booster dose, where a decreasing trend as a function of increased age is found. (D) Violin plots of Anti-S half lives and number of chronic morbidities. Points are plotted with a jitter function to display the density of individuals with each chronic morbidity count. (E) Parametric plot of chronic morbidity count and chronological age, where the size of each point corresponds to the number of data points at a given age with the same chronic morbidity count. The number of individuals within each chronic morbidity category are shown to the right of the plot.

Figure 2

Single-variate Kruskal-Wallis test results for primary series and booster dose: sex, age, and status (HCW, resident, senior). (A) Cross-sectional statistical comparisons. In order from left to right: population comparisons between primary series and booster dose decay, primary series sex-based comparison, booster sex-based comparison, primary series age-based comparison, booster age-based comparison. (B) Left: statistical comparisons between HCW, residents and seniors for primary series decay rates. Right: statistical comparisons between HCW, residents and seniors for booster-elicited decay rates. We note the y-axis is decay rate, which is inversely proportional to half life (see Eq. 2).

Figure 3

Lasso regression coefficients for Primary and Booster Dose responses across various health and demographic variables. Each color-coded curve corresponds to a specific variable, illustrating its coefficient value as the regularization parameter ($\lambda$) changes. The red dashed line represents the optimal $\lambda$ (min $\lambda$) where the model demonstrates the best fit, while the blue dashed line denotes a $\lambda$ that is 1 standard deviation away from the minimum.

Figure 4

The outcomes of the multivariate regression analysis, conducted with factors selected via lasso regression at the optimal lambda value. 95% confidence intervals are displayed for each coefficient. The primary and booster series are depicted in blue and gold, respectively. Statistical significance is indicated by stars: ‘***’ for $p<0.001$, ‘**’ for $p<0.01$, and ‘*’ for $p<0.05$. On the x-axis, the effect size is centered around zero; positive values to the right suggest an increased antibody decay rate compared to the reference category, while negative values to the left suggest a decrease. The magnitude of the associated effect is proportional to the distance from the center line. The reference category is specified in brackets. For dummy variables, each coefficient represents the change in the outcome relative to this reference within the categorical factor. For a table of numerical results see Table S2.