Antibodies to SARS-CoV-2 are associated with protection against reinfection

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

Abstract

Background: It is critical to understand whether infection with Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) protects from subsequent reinfection.

Methods: We investigated the incidence of SARS-CoV-2 PCR-positive results in seropositive and seronegative healthcare workers (HCWs) attending asymptomatic and symptomatic staff testing at Oxford University Hospitals, UK. Baseline antibody status was determined using anti-spike and/or anti-nucleocapsid IgG assays and staff followed for up to 30 weeks. We used Poisson regression to estimate the relative incidence of PCR-positive results and new symptomatic infection by antibody status, accounting for age, gender and changes in incidence over time.

Results: A total of 12,219 HCWs participated and had anti-spike IgG measured; 11,052 were followed up after negative and 1,246 after positive antibody results, including 79 who seroconverted during follow-up. 89 PCR-confirmed symptomatic infections occurred in seronegative individuals (0.46 cases per 10,000 days at risk) and no symptomatic infections in those with anti-spike antibodies. Additionally, 76 (0.40/10,000 days at risk) anti-spike IgG seronegative individuals had PCR-positive tests in asymptomatic screening, compared to 3 (0.21/10,000 days at risk) seropositive individuals. Overall, positive baseline anti-spike antibodies were associated with lower rates of PCR-positivity (with or without symptoms) (adjusted rate ratio 0.24 [95%CI 0.08-0.76, p=0.015]). Rate ratios were similar using anti-nucleocapsid IgG alone or combined with anti-spike IgG to determine baseline status.

Conclusions: Prior SARS-CoV-2 infection that generated antibody responses offered protection from reinfection for most people in the six months following infection. Further work is required to determine the long-term duration and correlates of post-infection immunity.

Introduction

Early in the COVID-19 pandemic, a critical unknown was the nature and durability of "natural immunity." While it was observed that most infected individuals developed detectable immune responses, it remained unclear whether these responses provided robust protection against a second infection. Specifically, researchers needed to determine if prior infection would prevent reinfection entirely (sterile immunity) or simply reduce disease severity.

Understanding these correlates of protection is vital for public health modeling, hospital staffing strategies, and defining the long-term trajectory of the pandemic. This study sought to quantify the relative risk of reinfection by comparing the incidence of new PCR-positive results between seropositive (antibody-present) and seronegative (antibody-absent) individuals in a high-exposure healthcare environment.

Methods

The researchers conducted a prospective longitudinal cohort study of 12,219 healthcare workers (HCWs) at Oxford University Hospitals. Participants were categorized based on their baseline serology using two primary markers: IgG antibodies against the SARS-CoV-2 trimeric spike (S) protein (via an Oxford-developed ELISA) and the nucleocapsid (N) protein (via a commercial Abbott assay).

Participants were followed for up to 30 weeks, undergoing regular asymptomatic PCR screening every two weeks alongside symptomatic testing as needed. To ensure data accuracy and avoid counting lingering viral RNA from a primary infection, the researchers implemented a 60-day window; a "new" infection was only recorded if it occurred at least 60 days after a previous positive PCR or serology result. The team utilized Poisson regression to estimate the Incidence Rate Ratio (IRR), adjusting for age, gender, and calendar time to account for the fluctuating background incidence of the pandemic.

Key Results

The study found a significant divergence in infection rates based on antibody status:

  • Incidence in Seronegative Individuals: 165 PCR-positive results occurred (0.86 per 10,000 days at risk), 89 of which were symptomatic.
  • Incidence in Seropositive Individuals: Only 3 PCR-positive results occurred (0.21 per 10,000 days at risk), and notably, zero symptomatic infections were recorded in this group.
  • Relative Risk: The adjusted IRR was 0.24 (95% CI 0.08–0.76, p=0.015), representing a 76% reduction in the risk of testing PCR-positive for those with baseline antibodies.
  • Durability: This protective effect was maintained for the duration of the 30-week study, suggesting that natural immunity lasts at least six months for the vast majority of individuals.

Discussion and Insights

The most significant finding was the total absence of symptomatic reinfections among seropositive individuals. This suggests that while "sterile immunity" (the prevention of any viral presence) may not be absolute, natural immunity is highly effective at preventing clinical disease.

Furthermore, the researchers observed a titer-dependent relationship. The risk of infection began to decline as antibody titers rose, even when those titers were below the official diagnostic threshold for a "positive" result. This implies that low levels of antibodies, which might be classified as negative in a clinical setting, still confer some degree of biological protection. The study also validated the use of various commercial assays, as both anti-spike and anti-nucleocapsid antibodies served as reliable proxies for protection.

To address potential "outcome ascertainment bias"—the possibility that seropositive individuals might seek testing less frequently—the authors performed a sensitivity analysis matching testing rates between groups. The protective effect remained robust, confirming the biological reality of the immune response.

Limitations and Future Work

While the results are encouraging, the study has several limitations. The cohort consisted primarily of working-age adults, meaning the results may not generalize to children, the elderly, or immunocompromised populations. Additionally, the study focused on humoral immunity (antibodies) and did not measure T-cell responses, which likely play a major role in the protective mechanism.

As the data was collected in 2020, the authors noted that the emergence of new viral variants could potentially escape the immunity generated by earlier strains. Finally, despite the high level of protection observed, the authors caution that seropositive individuals should continue to follow standard infection control measures, such as masking and distancing, until the risks of onward transmission are better understood.

Summary updated

Abstract

Abstract Background It is critical to understand whether infection with Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) protects from subsequent reinfection. Methods We investigated the incidence of SARS-CoV-2 PCR-positive results in seropositive and seronegative healthcare workers (HCWs) attending asymptomatic and symptomatic staff testing at Oxford University Hospitals, UK. Baseline antibody status was determined using anti-spike and/or anti-nucleocapsid IgG assays and staff followed for up to 30 weeks. We used Poisson regression to estimate the relative incidence of PCR-positive results and new symptomatic infection by antibody status, accounting for age, gender and changes in incidence over time. Results A total of 12219 HCWs participated and had anti-spike IgG measured, 11052 were followed up after negative and 1246 after positive antibody results including 79 who seroconverted during follow up. 89 PCR-confirmed symptomatic infections occurred in seronegative individuals (0.46 cases per 10,000 days at risk) and no symptomatic infections in those with anti-spike antibodies. Additionally, 76 (0.40/10,000 days at risk) anti-spike IgG seronegative individuals had PCR-positive tests in asymptomatic screening, compared to 3 (0.21/10,000 days at risk) seropositive individuals. Overall, positive baseline anti-spike antibodies were associated with lower rates of PCR-positivity (with or without symptoms) (adjusted rate ratio 0.24 [95%CI 0.08-0.76, p=0.015]). Rate ratios were similar using anti-nucleocapsid IgG alone or combined with anti-spike IgG to determine baseline status. Conclusions Prior SARS-CoV-2 infection that generated antibody responses offered protection from reinfection for most people in the six months following infection. Further work is required to determine the long-term duration and correlates of post-infection immunity.
BibTeX
@misc{lumley2020antibodies,
  author = {Sheila Lumley 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 Fiona C Warren and Liam J Peck and Thomas Ritter and Zoe de Toledo and Laura Warren and David Axten and Richard J. Cornall and E. Yvonne Jones and David I. Stuart and Gavin Screaton and Daniel Ebner and Sarah Hoosdally and Meera Chand and Derrick W. Crook and Colm P. O’Donnell and Christopher P. Conlon and Koen B. Pouwels and A. Sarah Walker and Tim Peto and Susan Hopkins and A Sarah Walker and Katie Jeffery and David W. Eyre},
  title = {{Antibodies to SARS-CoV-2 are associated with protection against reinfection}},
  year = {2020},
  doi = {10.1101/2020.11.18.20234369},
  howpublished = {medRxiv},
}
Figure