Participant Recruitment and Sample Collection
This study was approved by the University of British Columbia Research Ethics board (REB number: H20-03951) and was performed in accordance with the Declaration of Helsinki. Responses to COVID-19 vaccination were quantified in participants enrolled to the ‘PRospEctiVe EvaluatioN of immuniTy after COVID-19 vaccines’ (PREVENT-COVID) study over five study visits. Participants were recruited via sending study information through local networks; advertising through public media, existing volunteer or participant databases, institute-related websites and social media outlets; posters in physician offices, local hospitals, university areas and other COVID-19 immunization sites. Inclusion criteria were: ≥19 years old at first visit; and had an immunization provider administer a COVID-19 vaccine. All participants provided written informed consent. Exclusion criteria consisted of bleeding disorders that contraindicated blood collection, immunocompromising conditions or currently taking medication that may affect immune responses to COVID-19 vaccination. This analysis only included participants aged ≥ 50 years
Using early published binding antibody data from the mRNA-1273 vaccine 31, 50 individuals per group was determined to be sufficient to establish superiority between groups with 0.95 power (0.05 level of significance) and allowing for 10% loss to follow-up at each study visit. Data collected included age, biological sex, gender, ethnicity, health status (excellent, very good, good, fair, mildly poor, poor, very poor) 32–34, height, weight, COVID-19 vaccines received, dates of vaccinations, health conditions, and SARS-CoV-2 PCR testing data. Participants attended a maximum of five visit timepoints from April to December 2021. Participants could enter the study at any time from prior to their first vaccine dose to one-month after their second dose. Pre-vaccination timepoints were eligible for collection up to 24 hrs pre-vaccination. One-month post-dose vaccination were within window for collection between three- and six-weeks post vaccination. The last sample (four-months post-dose two) was eligible for collection between three- and five-months post-dose two. Participants were separated based on vaccine series as well as previous history of SARS-CoV-2 infection (Fig. 1). There was a total of four participant samples that were collected out of the one-month post-dose two window by approximately one week. A sensitivity analysis was performed, it was determined that there was no significant difference in results when these participants were removed, therefore they remained in the final study analyses.
Determination of COVID-19 Infection Status
To ensure that responses to vaccination were not due to potential SARS-CoV-2 infection,
participant provincial health records were screened for any positive COVID-19 PCR tests35. The following commercial assays were also performed on serum collected at each visit; Siemens ADVIA Centaur SARS-CoV-2 spike protein S1 RBD (total antibody detection), Ortho Vitros SARS-CoV-2 spike S1 RBD (total antibody detection) and Abbott Architect SARS-CoV-2 nucleocapsid (total IgG detection) to identify SARS-CoV-2 previously infected participants. For pre-vaccination samples, participants were identified as SARS-CoV-2 previously infected if any of the commercial assays had a reactive result. At post-vaccination visits, samples had to be reactive for both the nucleocapsid (Abbott) assay and either spike protein S1 RBD (Seimens or Ortho) assays to be considered previously infected36,37. Once a participant was identified as SARS-CoV-2 previously infected, they were considered previously infected for subsequent study visits.
Enzyme-linked immunosorbent assay (ELISA)
An indirect ELISA was performed for all available serum samples. Ninety-six well flat bottom microtiter plates (Immulon 2 HB, Thermo Fisher Scientific) were coated with SARS-CoV-2 index virus spike protein (SMT1-1 SARS-CoV-2 spike glycoprotein reference material, National Research Council Canada). Serum and standard (20/136, National Institute for Biological Standards and Control) were diluted in antibody buffer (1% skim milk in 1X PBS- 0.1% Tween20, Thermo Fisher Scientific). Plates were washed with 1X-PBS-0.1% Tween20 (wash buffer) before the addition of the secondary antibody (horseradish peroxidase conjugated mouse anti-human IgG, Life Technologies). The substrate solution was added (O-Phenylenediamine dihydrochloride (OPD), Sigma Aldrich) prior to the addition of of 3M hydrochloric acid (Thermo Fisher Scientific). Optical densities of individual wells were analyzed at 490 nm (infiniteM200, Tecan). Concentrations of anti-spike protein specific IgG (S-IgG), expressed in binding antibody units per millilitre (BAU/mL). Samples below 2 BAU/mL were assigned a value of 1 BAU/mL for statistical purposes.
Antibody avidity assay
Using the ELISA described previously, 1X PBS, 0.25M, 0.5M, 0.75M, 1.0M, 1.5M and 2.0M ammonium thiocyanate (NH4SCN, Fisher Scientific) was added to seven separate wells prior to the addition of the secondary antibody, similarly to Abu-Raya et al. 201938. The results were reported as the total relative avidity index (TRAI) expressed in arbitrary units (AU) and the total absolute avidity (TAA) expressed in absolute AU per mL (AAU/ mL). The minimum concentration needed to perform the assay was not reached for any of the baseline samples, therefore no comparisons could be made between baseline and subsequent visits. Fractional absolute avidity (FAA) levels of S-IgG present at each concentration of NH4SCN were reported in BAU/mL. Supplementary Fig. 1 displays a sample calculation.
Antibody dependent cellular phagocytosis (ADCP) assay
This assay was optimized based on previously published material39–42. Briefly, fluorescent beads (FluoSpheres™ NeutrAdavin ™, 1µm, yellow-green, 505/ 515, Invitrogen) were conjugated with biotinylated SARS-CoV-2 index virus spike protein (Biotinylated SARS-CoV-2 S protein, His,Avitag™, Super stable trimer, MALS verified, ACROBiosystems) before being added to participant serum. THP-1 cells (American Type Culture Collection) were added to each well with the bead-antibody complexes overnight to allow the cells to phagocytose the beads. Samples were analyzed in triplicate with the BD LSR Fortessa flow cytometer and FlowJo Software (BD Biosciences). The gating strategy is displayed in Supplementary Fig. 2. Results were reported as the mean phagocytic score of duplicates resulting the lowest percent coefficient of variation (CV) (percentage of bead positive events multiplied by the mean fluorescent intensity of the bead positive events). Percent CV of duplicates had to be below 20% to be accepted. Results below a mean phagocytic score of 300 were assigned a score of 150 for statistical purposes.
ACE2 inhibition assay
Serum samples were diluted in Meso Scale Discovery (MSD) Diluent 100 prior to serological testing. Following the manufacturer recommended protocol, we measured the capacity of spike specific SARS-CoV-2 antibodies directed against the index virus and nine additional variants (WHO label 43, pango lineage 44) index virus (GISAID Accession ID: EPI_ISL_402124), Alpha (B.1.1.7), Beta (B.1.351), Gamma (P.1), Delta (B.1.617.2), Zeta (P.2), Iota (B.1.526.1), Kappa (B.1.617.1), B.1.617, and B.1.617.3 to inhibit the interaction of the ACE2 receptor with the spike protein using the MSD V-PLEX SARS-CoV-2 Panel 13 ACE2 assay. Inhibition capacity was measured using the MSD QuickPlex SQ120 instrument and was reported as concentration of ACE2 inhibiting antibodies, in µg/mL for the index virus and in units/mL for SARS-CoV-2 variants. Results below the following lower limits of quantification were assigned half of the following values: index virus (0.02262 µg/mL), Alpha (0.01620 units/mL), Beta (0.00762 units/mL), Gamma (0.01439 units/mL), Delta (0.03334 units/mL), Zeta (0.01655 units/mL), Iota (0.01761 units/mL), Kappa (0.007091 units/mL), B.1.617 (0.02540 units/mL), B.1.617.3 (0.01044 units/mL).
Activation induced marker (AIM) assay
Whole blood was collected in sodium heparin tubes, kept at room temperature and processed within 4 hours of collection. Blood was mixed at 1:1 ratio with IMDM media (Gibco), and 200 µL of whole blood-media mixture was added to a single well per condition in a 48-well tissue culture plate. Wells were stimulated with 1 µg/mL of SARS-CoV-2 (index virus) overlapping peptide pool of the immunodominant regions of the spike glycoprotein (Prot_S PepTivator®, Miltenyi Biotec), or the complete spike glycoprotein complete (Complete Prot_S PepTivator®, Miltenyi Biotec). Negative control wells were left unstimulated. Assays were incubated at 37°C for 44–48 hours. Anti-CD137-APC mAb (BioLegend, Supplementary table 11) was added to each condition at stimulation, as upregulation of this surface receptor peaks at 24 hours. Following stimulation, whole blood was surface stained with mAb panel for 20 min at room temperature. 1 mL of OptiLyse C (Beckman Coulter) was added to each tube and incubated for 10 min at room temperature in the dark. Cells were washed twice with 2 mL of PBS and fixed with 0.5% paraformaldehyde in PBS and stored at 4°C until acquisition. Samples were acquired within 48 hours of staining on a 5-laser A5 Symphony cytometer (BD Biosciences), fresh single colour controls were used for compensation. Phenotypic analysis was performed using FlowJo version 10.8.1 (BD Biosciences). Assay cut-off for a positive response (CD4+CD25+OX40+ or CD8+CD69+4-1BB+) was 0.02%, being the mean + 3SD of the negative control wells.
Statistical analyses- humoral immune responses
To evaluate humoral immune responses to COVID-19 vaccines in adults over 50 years old, S-antibody concentration and function was quantified. Participants were separated into three groups based on the types of first and second COVID-19 vaccine doses received. The first group received either two doses of BNT162B2 or two doses of mRNA-1273 or a combination of BNT162B2 and mRNA-1273 (mRNA/mRNA), the second group received ChAdOx1-S for their first dose and either BNT162B2 or mRNA-1273 as their second dose (ChAdOx1-S/mRNA), and the third group received two doses of ChAdOx1-S (ChAdOx1-S/CAadOx1-S).
To assess the suggested protection elicited by a two-dose series of COVID-19 vaccines, S-IgG concentrations were compared to suggested literature correlates of protection. Goldblatt et al. (2022) suggested correlates of protection corresponding to 154 BAU/mL based on homologous mRNA (mRNA-1273 or BNT162b2) or viral vector (ChAdOx1-S or Ad.26COV2.S). They also removed mRNA vaccine responses from their analyses which resulted in a correlate of 60 BAU/mL (95% CI 35–102) because it may be useful when comparing vaccines with lower vaccine efficacy11. Wei et al. (2022) suggested potential correlates of protection based on S-IgG concentrations by estimating two-thirds (67%) of individuals were protected against infection. Homologous ChAdOx1-S or homologous BNT162b2 required estimated concentrations of 107 BAU/mL or 94 BAU/mL. Using models that pooled data for both vaccines, 100 BAU/mL was suggested12.
Prior to statistical analyses, all data were log10 transformed. Statistical analyses were performed with R (V.4.1.1; The R Project for Statistical Computing, Vienna, Austria). R version 4.1.3 (2022-03-10). Statistical analyses were performed on groups with a minimum of two participants. Results were considered statistically significant at P-values < 0.05. A Welch’s t-test was used to compare responses to vaccination based on S-IgG concentrations, antibody avidity, ADCP scores and ACE2 inhibiting antibody concentrations between timepoints for each group, responses pre-dose two, as well as compare previously infected and infection-naive participants within the same group at each visit. A One-way ANOVA was used to compare responses to vaccination with baseline (Dunnett post-hoc) as well as compare the three groups at each visit (Tukey-Kramer post-hoc). Multivariable linear regression analyses were performed to determine the effect of demographic factors on S-IgG concentrations, TRAI, ADCP scores and S-inhibiting antibody concentrations for the index virus. Based on the distribution of the data, demographic factors were separated into the following groups: age (≥ 70 years old vs. ≤ 69 years old), biological sex (male vs. female), ethnicity (white vs. non-white), BMI (normal (18.5–24.9) and underweight (< 18.5) vs. overweight and obese (≥25), vaccine series (ChAdOx1-S/ ChAdOx1-S vs. mRNA/mRNA and ChAdOx1-S/mRNA combined), vaccine interval (≥ 13 weeks vs. < 13 weeks), health status (excellent vs. very good/ good/ fair/ mildly poor) and infection status (previously infected vs. infection-naive ) at both one-month and four-months post-dose two. This was achieved by fitting the multivariable model via priori-define backwards selection protocol with the Bayesian Information Criterion (BIC) statistic used to select the model with the best fit. From univariable models, the initial multivariable model included all demographic factors with a P < 0.10. The reduced models were selected by removing the demographic factor with the highest p-value at each stage. The final model consisted of the demographic factors with the lowest BIC value45.
To determine the contributions of both antibody concentration and avidity on antibody function, multivariable linear regression analyses were completed for one-month and four-months post-dose two between FAA levels and ADCP scores as well as ACE2 inhibiting antibody concentrations. Multivariable models were fit using a conservative confounding model selection approach. From univariable models, the initial multivariable model included all demographic factors with a P < 0.10. A stepwise approach was applied to fit a series of reduced models. The value of the coefficient was compared between reduced model for each FAA level, the secondary demographic factor associated with the lowest relative change were removed. This process was repeated until the minimum change exceeded 5%46.
Pearson correlations were performed to determine the relationship between S-IgG concentrations and antibody function (TRAI, ADCP scores and concentrations of index virus ACE2 inhibiting antibodies) at one-month and four-months post-dose two. The same approach was taken to evaluate the correlation between TAA and antibody function (ADCP scores and concentrations of index virus ACE2 inhibiting antibodies).
Statistical Analyses (T-cell responses)
All statistical analyses were performed in R studio (version 4.0.2 and below). T cell data were tested for normal distribution and equal variance using Shapiro-Wilk (shapiro.test()) and Levene Test (leveneTest()) respectively. Kruskal Wallis (kruskal.test()) and ANOVA(anov()) were performed based on distribution and variance for multiple comparisons. If significance was found ( P < 0.005), Wilcoxon Rank Sum Test (pairwise.wilcox.test()) or Tukey’s Multiple Comparisons test (glth() using mcp = “TUKEY”) were carried out. Mann-Whitney (wilcox.test()) tests were used when comparing only two groups. Multilinear regression models were generated to determine the impact of demographic data on T cell responses and T cell responses on antibody data. The associations between T cell data from one-month post-dose two and four-months post-dose two and antibody data at the same time point were evaluated. Additionally, the associations between T cell data from one-month post-dose two and antibody data from four-months post-dose two were evaluated. Heatmaps were generated and scaled using lm() function.