SARS-CoV-2 antibodies and breakthrough infections in the Virus Watch cohort

Abstract

A range of studies globally demonstrate that the effectiveness of SARS-CoV-2 vaccines wane over time, but the total effect of anti-S antibody levels on risk of SARS-CoV-2 infection and whether this varies by vaccine type is not well understood. Here we show that anti-S levels peak three to four weeks following the second dose of vaccine and the geometric mean of the samples is nine fold higher for BNT162b2 than ChAdOx1. Increasing anti-S levels are associated with a reduced risk of SARS-CoV-2 infection (Hazard Ratio 0.85; 95%CIs: 0.79-0.92). We do not find evidence that this antibody relationship with risk of infection varies by second dose vaccine type (BNT162b2 vs. ChAdOx1). In keeping with our anti-S antibody data, we find that people vaccinated with ChAdOx1 had 1.64 times the odds (95% confidence interval 1.45-1.85) of a breakthrough infection compared to BNT162b2. We anticipate our findings to be useful in the estimation of the protective effect of anti-S levels on risk of infection due to Delta. Our findings provide evidence about the relationship between antibody levels and protection for different vaccines and will support decisions on optimising the timing of booster vaccinations and identifying individuals who should be prioritised for booster vaccination, including those who are older, clinically extremely vulnerable, or received ChAdOx1 as their primary course. Our finding that risk of infection by anti-S level does not interact with vaccine type, but that individuals vaccinated with ChAdOx1 were at higher risk of infection, provides additional support for the use of using anti-S levels for estimating vaccine efficacy.

Fig. 1. Geometric mean (with 95% confidence intervals) for anti-S samples (U/ml) over time since second dose of vaccination, and levels predicted by a linear mixed effect model amongst N-seronegative individuals by vaccine type.

Wanting anti-S levels followed a mean log-linear decline from 4 weeks after the second dose of vaccination for both vaccine types. The geometric mean of the anti-S samples peaked at 3 weeks after the second dose of vaccine BNT162b2 at 10555 (95% CI: 9291–11992) U/ml and at 4 weeks for ChAdOx1 at 1069 (95% CI: 925–1236) U/ml. At 20 weeks after the second dose of vaccine, the mean anti-S levels were 1611 (95% CI: 1515–1712) U/ml for BNT162b2 and 387 (95% CI: 363–414) U/ml for ChAdOx1.

Fig. 2. Geometric mean (with 95% confidence intervals) for anti-S samples (U/ml) over time since second dose of vaccination, and levels predicted by a linear mixed effect model amongst N-seronegative individuals by vaccine type and sex, age and clinical risk group.

(Note: Different y-axis scales for BNT162b2 and ChAdOx1). Twenty weeks after the second dose of vaccine, mean BNT162b2 anti-S levels were 1875 (95% CI: 1686–2085) U/ml in 18–64 year olds and 1479 (95% CI: 1373–1593) U/ml in participants over 65 years of age (Fig. 2). Anti-S levels at 20 weeks post second dose of ChAdOx1 were 420 (95% CI: 380–464) U/ml in 18–64 year olds and 356 (95% CI: 328–390) U/ml in participants over 65 years of age. There was no evidence of a difference in the rates of waning by age, sex or clinical risk group for BNT162b2 or ChAdOx1. ChAdOx1 Oxford, AstraZeneca COVID-19 vaccine. BNT162b2 BioNTech, Pfizer COVID-19 vaccine. Legend description: Panels report on geometric mean anti-S samples (U/ml) over time since second dose of vaccination for Oxford, AstraZeneca by a sex, c age, e clinical risk group and BioNTech, Pfizer by b sex, d age, f clinical risk group. CEV Clinically Extremely Vulnerable, CV Clinically Vulnerable. 18–64 individuals aged 18 to 64. 65+ individuals aged 65 and over. Predictions are the results from linear mixed effect models with a random intercept for age, sex and clinical vulnerability.

Fig. 3. Kaplan–Meier survival curve of risk of breakthrough infection by Spike antibody level in quartiles.

The first anti-S level for participants after their second dose of vaccine and 1st July 2021 was used. The risk of breakthrough infections for the lowest quartile (upper limit 413 U/ml) began to diverge from the risk for higher quartiles after around 20 days of follow-up. The risk of breakthrough infections for the second lowest quartile (414–934 U/ml) began to diverge from the risk for higher quartiles after around 70 days of follow-up.