Antibody responses to SARS-CoV-2 vaccines in 45,965 adults from the general population of the United Kingdom

Abstract

We report that in a cohort of 45,965 adults, who were receiving either the ChAdOx1 or the BNT162b2 SARS-CoV-2 vaccines, in those who had no prior infection with SARS-CoV-2, seroconversion rates and quantitative antibody levels after a single dose were lower in older individuals, especially in those aged >60 years. Two vaccine doses achieved high responses across all ages. Antibody levels increased more slowly and to lower levels with a single dose of ChAdOx1 compared with a single dose of BNT162b2, but waned following a single dose of BNT162b2 in older individuals. In descriptive latent class models, we identified four responder subgroups, including a 'low responder' group that more commonly consisted of people aged >75 years, males and individuals with long-term health conditions. Given our findings, we propose that available vaccines should be prioritized for those not previously infected and that second doses should be prioritized for individuals aged >60 years. Further data are needed to better understand the extent to which quantitative antibody responses are associated with vaccine-mediated protection.

Fig. 1. Predicted probability of anti-spike IgG positivity by time from first vaccination for comparisons of age by vaccine type and prior infection status.

The data are from 40,131 participants without prior infection and 5,834 participants with prior infection. a, No prior infection and received one dose of ChAdOx1. b, No prior infection and received one dose of BNT162b2. c, No prior infection and received two doses of BNT162b2. d, With prior infection and received one dose of ChAdOx1. e, With prior infection and received one dose of BNT162b2. f, For those who received two doses of BNT162b2 without prior infection, the chart shows the percentage of participants having had two vaccine doses by each time point (grey, had two doses; blue, had only one dose). Different x axis scales reflect different durations of follow-up post-vaccination in the different cohorts. Line colour indicates antibody response predicted for ages 20, 40, 60 and 80 years (see Extended Data Fig. 2 for the full model across all ages and comparisons of vaccine type by age). The 95% CIs are calculated by prediction ± 1.96 × standard error of prediction. Source data

Fig. 2. Predicted probability of anti-spike IgG positivity after first vaccination in participants without evidence of prior infection from a multivariable logistic regression model.

Predicted probability with 95% CIs of anti-spike IgG positivity 14–60 days after first vaccination in 24,977 participants without evidence of prior infection. a, Predicted probability of anti-spike IgG positivity by age and sex. b, Predicted probability of anti-spike IgG positivity by age and vaccine type. Age was fitted using natural cubic spline with four internal knots placed at the 20th, 40th, 60th and 80th percentile (30, 44, 57 and 71 years) and two boundary knots at the 5th and 95th percentile (19 and 82 years). The 95% CIs were calculated by prediction ± 1.96 × standard error of the prediction and are shown as the shaded area. Testing for an interaction between sex and age was performed using a likelihood ratio test (P = 0.02). Values are plotted at the reference category for other variables (BNT162b2 one dose (a)/female (b), white ethnicity, index of multiple deprivation = 55, household size = 1, did not work in patient-facing healthcare or social care, did not work in a care home, no long-term health condition). Source data

Fig. 3. Predicted anti-spike IgG levels by time from first vaccination for comparisons of age by vaccine type and prior infection status.

Predicted anti-spike IgG levels (mAb45 ng ml^–1 equivalent units) by time from first vaccination based on data from 40,131 participants without prior infection and 5,834 participants with prior infection. a, No prior infection and received one dose of ChAdOx1. b, No prior infection and received one dose of BNT162b2. c, No prior infection and received two doses of BNT162b2. d, With prior infection and received one dose of ChAdOx1. e, With prior infection and received one dose of BNT162b2. f, For those who received two doses of BNT162b2 without prior infection, the chart shows the percentage of participants having had two vaccine doses by each time point (grey, had two doses; blue, had only one dose). Different x axis scales reflect different durations of follow-up post-vaccination in the different cohorts. Predicted levels are plotted on a log scale. Black dotted line indicates the threshold of IgG positivity (42 ng ml^–1). Line colour indicates response predicted for ages 20, 40, 60 and 80 years (see Extended Data Fig. 8 for all ages and Fig. 4 for comparisons of vaccine type by age). The 95% CIs are calculated by prediction ± 1.96 × standard error of the prediction. Source data

Fig. 4. Predicted anti-spike IgG levels by time from first vaccination for comparisons of vaccine type and prior infection status by age.

a–d, Predicted anti-spike IgG levels (mAb45 ng ml^–1 equivalent units) by time from first vaccination based on data from 40,131 participants without prior infection and 5,834 participants with prior infection. Data shown for 20 year olds (a), 40 year olds (b), 60 year olds (c) and 80 year olds (d). Black dotted line indicates the threshold of IgG positivity (42 ng ml^–1). Line colour indicates predicted response for the different vaccine type and prior infection status (full models shown in Extended Data Fig. 8, plotted by vaccine in Fig. 3). The 95% CIs are calculated by prediction ± 1.96 × standard error of the prediction. Data identical to Fig. 3, but Fig. 3 represent age rather than vaccine type. Source data

Fig. 5. Predicted anti-spike IgG trajectory in participants without prior infection by class identified from latent class mixed models.

Predicted anti-spike IgG trajectory in 36,518 participants without prior infection by class identified from latent class mixed models, using data from 14 days before vaccination to the 90th percentile of the observed time points after vaccination. a, One dose ChAdOx1 vaccine and no evidence of prior infection (N = 22,424 participants). b, One dose BNT162b2 vaccine and no evidence of prior infection (N = 14,094 participants). Black dotted line indicates the threshold of IgG positivity (42 ng ml^–1). The 95% CIs are calculated by Monte Carlo approximation of the posterior distribution of the predicted values. The distribution of factors by class membership is shown in Supplementary Table 4. Class 1 = plausibly previously infected group (3.9% ChAdOx1, 3.9% BNT162b2), 2 = high-response group (31.6% ChAdOx1, 63.5% BNT162b2), 3 = medium-response group (58.7% ChAdOx1, 27.5% BNT162b2), 4 = low-response group (5.8% ChAdOx1, 5.1% BNT162b2). Source data

Extended Data Fig. 1. Flowchart of the study cohort.

N represents the number of participants, and n represents the number of antibody measurements.

Extended Data Fig. 2. Predicted probability of anti-spike IgG positivity by time from first vaccination based on data from 40,131 participants without prior infection and 5,834 participants with prior infection.

a, 20-year-old. b, 40-year-old. c, 60-year-old. d, 80-year-old. Line colour indicates different vaccine type and prior infection status. The 95% confidence intervals are calculated by prediction ± 1.96*standard error of the prediction. Data identical to Fig. 1, but Fig. 1 panels represent age rather than vaccine type as here. e, Predicted probability of anti-spike IgG positivity by time from first vaccination and age, according to vaccine type and prior infection status (full model). Predictions shown for specific ages in Fig. 1. Observed data in Extended Data Fig. 3–7. Source data

Extended Data Fig. 3. Percentage (95% CI) anti-spike IgG positive by days after first vaccination in 23,368 participants receiving a single dose of ChAdOx1 vaccine and without evidence of prior infection.

Results were divided into four age groups: 16–34, 35–54, 55–74 and >75 years. Exact binomial test was performed to obtain the 95% confidence intervals (shown as ‘error bars’). Numbers above the x-axis represent positive number/total number, and the percentage of positivity (%). Source data

Extended Data Fig. 4. Percentage (95% CI) anti-spike IgG positives by days after first vaccination in 3,767 participants receiving a single dose of ChAdOx1 vaccine and with evidence of prior infection.

Results were divided into four age groups: 16–34, 35–54, 55–74 and >75 years. Exact binomial test was performed to obtain the 95% confidence intervals (shown as ‘error bars’). Numbers above the x-axis represent positive number/total number, and the percentage of positivity (%). Source data

Extended Data Fig. 5. Percentage (95% CI) anti-spike IgG positives by days after first vaccination in 14,894 participants receiving a single dose of BNT162b2 vaccine and without evidence of prior infection.

Results were divided into four age groups: 16–34, 35–54, 55–74 and >75 years. Exact binomial test was performed to obtain the 95% confidence intervals (shown as ‘error bars’). Numbers above the x-axis represent positive number/total number, and the percentage of positivity (%). Source data

Extended Data Fig. 6. Percentage (95% CI) anti-spike IgG positives by days after first vaccination in 2,067 participants receiving a single dose of BNT162b2 vaccine and with evidence of prior infection.

Results were divided into four age groups: 16–34, 35–54, 55–74 and >75 years. Exact binomial test was performed to obtain the 95% confidence intervals (shown as ‘error bars’). Numbers above the x-axis represent positive number/total number, and the percentage of positivity (%). Source data

Extended Data Fig. 7. Percentage (95% CI) anti-spike IgG positives by days after first vaccination in 1,869 participants receiving two doses of BNT162b2 vaccine and without evidence of prior infection.

Results were divided into four age groups: 16–34, 35–54, 55–74 and >75 years. Exact binomial test was performed to obtain the 95% confidence intervals (shown as ‘error bars’). Numbers above the x-axis represent positive number/total number, and the percentage of positivity (%). Source data

Extended Data Fig. 8. Predicted anti-spike IgG levels (ng/ml) by time from first vaccination and age based on data from 40,131 participants without prior infection and 5,834 participants with prior infection, according to vaccine type and prior infection status (full model).

Predictions shown for specific ages in Fig. 3. Source data

Extended Data Fig. 9. Individual trajectories by classes identified from latent class mixed models for single dose ChAdOx1 and BNT162b2 vaccine.

In Class 1 and Class 4, all participants were plotted (ChAdOx1-Class 1: N = 867, Class 4: N = 1297. BNT162b2- Class 1: 547, Class 4: 720). In Class 2 and 3, 1,000 randomly selected participants were plotted for visualisation due to the large number. Colours: navy blue, Class 1; orange, Class 2; mid blue, Class 3; red, Class 4. Source data

Extended Data Fig. 10. Age distribution by classes identified from latent class mixed models for single dose ChAdOx1 (N = 22,424) and BNT162b2 (N = 14,094) vaccines.

1 = ’plausibly previously infected’ group, 2 = ’high response’ group, 3 = ’medium response’ group, 4 = ’low response group’. Same area for each violin, see Supplementary Table 4 for numbers in each group. For the box and whisker inserts: centre line, median; box limits, upper and lower quartiles; whiskers, 1.5x interquartile range; points, outliers. Colours: navy blue, Class 1; orange, Class 2; mid blue, Class 3; red, Class 4. Source data