The duration, dynamics and determinants of SARS-CoV-2 antibody responses in individual healthcare workers

Sheila Lumley, Jia Wei, Denise O’Donnell, Nicole Stoesser, Philippa C. Matthews, Alison Howarth, Stephanie B. Hatch, Brian D. Marsden, Stuart Cox, Tim James, Liam J Peck, Thomas Ritter, Zoe de Toledo, Richard J. Cornall, E Yvonne Jones, David I. Stuart, Gavin Screaton, Daniel Ebner, Sarah Hoosdally, Derrick W. Crook, Christopher P. Conlon, Koen B. Pouwels, A. Sarah Walker, Tim Peto, Katie Jeffery, David W. EyremedRxiv 2020 · DOI

Abstract

Understanding the longevity of the immune response to SARS-CoV-2 is critical for predicting the risk of reinfection and interpreting population-level seroprevalence data. This study analyzed longitudinal data from 3,217 healthcare workers (HCWs) in the UK to quantify the decay of IgG antibodies against the SARS-CoV-2 nucleocapsid protein. Using Bayesian mixed linear models, we found that antibody levels peak approximately 24 days post-infection and subsequently wane with a mean half-life of 85 days. The estimated time to seroreversion—the loss of a detectable positive result—was 137 days. Factors such as older age, Asian ethnicity, and symptomatic infection were associated with higher peak antibody levels and slower waning. These findings suggest that antibody-based surveys may significantly underestimate cumulative incidence as time passes since the initial infection.

Introduction

In the first year of the COVID-19 pandemic, a fundamental question for public health policy and individual risk assessment was the duration of "natural immunity." While the development of antibodies following infection was well-documented, the rate at which these markers declined remained a subject of significant debate. If antibody levels drop rapidly, "snapshot" seroprevalence studies—used to estimate what percentage of a population has been infected—become increasingly inaccurate over time. Furthermore, the rate of waning has direct implications for the potential window of protection against reinfection.

This study sought to characterize the quantitative decay of SARS-CoV-2 anti-nucleocapsid IgG antibodies in a large, longitudinal cohort of working-age adults. By moving beyond binary "positive/negative" classifications, the researchers aimed to model the biological half-life of the antibody response and identify the demographic and clinical determinants of its duration.

Methods and Cohort Selection

The study followed a cohort of 3,217 healthcare workers (HCWs) at Oxford University Hospitals between April and October 2020. Participants provided serial blood samples to track antibody trajectories over time.

Assay and Measurement

Antibody levels were measured using the Abbott Architect i2000 chemiluminescent microparticle immunoassay (CMIA), specifically targeting IgG antibodies against the SARS-CoV-2 nucleocapsid (N) protein. The nucleocapsid protein is a major structural component of the virus and a common target for diagnostic serology.

Statistical Modeling

To account for individual variability in immune responses, the researchers employed Bayesian mixed linear models. This statistical framework allowed for the estimation of "random effects" (individual-specific deviations) and "fixed effects" (population-wide trends associated with specific variables). Antibody levels were $log_2$-transformed to model exponential decay, enabling the calculation of a biological half-life for each participant.

The analysis focused on a sub-cohort of 452 HCWs who were seropositive and had at least two measurements available to track the decline from their maximum observed titre.

Results: The Dynamics of Antibody Waning

The study identified a clear, predictable trajectory for the anti-nucleocapsid IgG response in this cohort.

Peak and Half-Life

Antibody levels typically reached their peak at approximately 24 days (95% CrI 19-31) following the first positive PCR test. Following this peak, levels began a steady decline. The mean estimated antibody half-life was 85 days (95% CrI 81-90).

The Seroreversion Clock

A critical finding for public health surveillance was the "time to loss" of a positive result. The mean time from peak antibody levels to seroreversion (dropping below the assay's threshold for a positive result) was 137 days (95% CrI 127-148). The model estimated that 33% of previously infected individuals would test negative by three months, and 53% would test negative by six months. This rapid seroreversion indicates that the "visibility" of past infections in a population is relatively short-lived when using anti-nucleocapsid assays.

Determinants of the Immune Response

The study identified several key factors that influenced both the magnitude and the duration of the antibody response.

The "Rich Get Richer" Phenomenon

A significant correlation was observed between the maximum antibody titre achieved and the rate of waning. Individuals who generated a more robust initial response (higher peak) tended to have longer antibody half-lives. This suggests that a strong initial immune challenge leads to a more durable marker of immunity.

Demographics and Symptoms

  • Age: Increasing age was independently associated with higher peak antibody levels. For every 10-year increase in age, the antibody half-life increased by approximately 3.4 days. Older workers in this cohort (up to age 69) retained detectable antibodies longer than their younger colleagues.
  • Ethnicity: Participants of Asian ethnicity were found to have higher maximum antibody levels compared to other groups. The authors suggest this might be related to differences in initial viral load or potential cross-reactivity from prior exposure to seasonal coronaviruses.
  • Symptomatic Status: Asymptomatic individuals exhibited both lower peak antibody levels and faster waning. This implies that mild or silent infections may leave a less permanent "imprint" on the circulating antibody pool.

Discussion and Limitations

These results provide a quantitative basis for understanding the limitations of seroprevalence studies. Because antibodies wane within months, cross-sectional surveys will naturally underestimate the true extent of a population's exposure to the virus, especially as the time since the peak of an outbreak increases.

However, several caveats must be considered:

  1. Nucleocapsid vs. Spike: This study focused on the nucleocapsid (N) protein. Other research suggests that antibodies against the Spike (S) protein—the primary target of most vaccines—may be more stable and wane at a slower rate.
  2. Cellular Immunity: Circulating antibodies are only one component of the immune system. The study did not measure T-cell responses or memory B-cells, which can persist even after antibody levels become undetectable and may provide significant protection against severe disease upon reinfection.
  3. Cohort Age: The study population consisted of working-age adults (17-69). The dynamics of waning in children or the elderly (over 70) may differ.

Conclusion

The anti-nucleocapsid IgG response to SARS-CoV-2 wanes significantly within the first six months following infection, with a half-life of 85 days. The rate of decay is influenced by the initial peak titre, age, and the presence of symptoms. While the loss of detectable antibodies does not necessarily equate to a total loss of immunity, it does pose a significant challenge for long-term serological surveillance. Ongoing longitudinal studies are essential to determine how these waning trajectories correlate with the risk of reinfection over time.

Summary updated

Abstract

Abstract Background SARS-CoV-2 IgG antibody measurements can be used to estimate the proportion of a population exposed or infected and may be informative about the risk of future infection. Previous estimates of the duration of antibody responses vary. Methods We present 6 months of data from a longitudinal seroprevalence study of 3217 UK healthcare workers (HCWs). Serial measurements of IgG antibodies to SARS-CoV-2 nucleocapsid were obtained. Bayesian mixed linear models were used to investigate antibody waning and associations with age, gender, ethnicity, previous symptoms and PCR results. Results In this cohort of working age HCWs, antibody levels rose to a peak at 24 (95% credibility interval, CrI 19-31) days post-first positive PCR test, before beginning to fall. Considering 452 IgG seropositive HCWs over a median of 121 days (maximum 171 days) from their maximum positive IgG titre, the mean estimated antibody half-life was 85 (95%CrI, 81-90) days. The estimated mean time to loss of a positive antibody result was 137 (95%CrI 127-148) days. We observed variation between individuals; higher maximum observed IgG titres were associated with longer estimated antibody half-lives. Increasing age, Asian ethnicity and prior self-reported symptoms were independently associated with higher maximum antibody levels, and increasing age and a positive PCR test undertaken for symptoms with longer antibody half-lives. Conclusion IgG antibody levels to SARS-CoV-2 nucleocapsid wane within months, and faster in younger adults and those without symptoms. Ongoing longitudinal studies are required to track the long-term duration of antibody levels and their association with immunity to SARS-CoV-2 reinfection. Summary Serially measured SARS-CoV-2 anti-nucleocapsid IgG titres from 452 seropositive healthcare workers demonstrate levels fall by half in 85 days. From a peak result, detectable antibodies last a mean 137 days. Levels fall faster in younger adults and following asymptomatic infection.
BibTeX
@misc{lumley2020duration,
  author = {Sheila Lumley and Jia Wei and Denise O’Donnell and Nicole Stoesser and Philippa C. Matthews and Alison Howarth and Stephanie B. Hatch and Brian D. Marsden and Stuart Cox and Tim James and Liam J Peck and Thomas Ritter and Zoe de Toledo and Richard J. Cornall and E Yvonne Jones and David I. Stuart and Gavin Screaton and Daniel Ebner and Sarah Hoosdally and Derrick W. Crook and Christopher P. Conlon and Koen B. Pouwels and A. Sarah Walker and Tim Peto and A Sarah Walker and Katie Jeffery and David W. Eyre},
  title = {{The duration, dynamics and determinants of SARS-CoV-2 antibody responses in individual healthcare workers}},
  year = {2020},
  doi = {10.1101/2020.11.02.20224824},
  howpublished = {medRxiv},
}
Figure