Correlates of protection, thresholds of protection, and immunobridging in SARS-CoV-2 infection

David S. Khoury, Timothy E. Schlub, Deborah Cromer, Megan Steain, Youyi Fong, Peter B. Gilbert, Kanta Subbarao, James A. Triccas, Stephen J. Kent, Miles P. DavenportmedRxiv 2022 · DOI

Abstract

Several studies show neutralizing antibody levels are an important correlate of immune protection from COVID-19 and have estimated the relationship between neutralizing antibodies and protection. However, a number of these studies appear to yield quite different estimates of the level of neutralizing antibodies required for protection. Here we show that after normalization of antibody titers current studies converge on a consistent relationship between antibody levels and protection from COVID-19.

Introduction

As the COVID-19 pandemic progressed, the traditional pathway for vaccine approval—large-scale, placebo-controlled Phase 3 clinical trials—became increasingly difficult to execute. The emergence of new variants and the ethical concerns of withholding vaccines from control groups necessitated a more streamlined approach: immunobridging. This process allows researchers to predict a vaccine's efficacy by measuring a specific immune marker, known as a correlate of protection (CoP), rather than waiting for clinical infection outcomes.

While neutralizing antibody (nAb) levels were early candidates for a CoP, initial research presented a fragmented picture. Different studies suggested widely varying "thresholds" of antibodies required for protection, leading to confusion in both regulatory and clinical settings. This paper reconciles these differences, demonstrating that the underlying biology of protection is remarkably consistent once measurement variability is addressed.

The Challenge of Standardization

The primary obstacle in establishing a universal threshold for protection was the lack of assay standardization. Even with the introduction of the WHO’s International Units (IU/ml), laboratories using different testing platforms reported vastly different absolute values. For instance, reported titers for the Moderna (mRNA-1273) vaccine ranged from 247 IU/mL to 1404 IU/mL across different studies.

To resolve this, the authors employed a pragmatic normalization approach. Instead of relying on absolute units, they scaled antibody levels as a "fold-of-convalescent" titer. By comparing vaccine-induced antibodies to the levels found in individuals who had recovered from natural infection within the same study, the researchers were able to translate results into a common language.

Reconciling Models of Protection

The study synthesized data from two primary modeling perspectives:

  1. Vaccine-Comparison Models: These analyze the average antibody levels produced by various vaccine platforms (e.g., mRNA, viral vector, protein subunit) and correlate them with the overall efficacy observed in their respective Phase 3 trials.
  2. Breakthrough-Infection Models: These look at individual-level data within a single trial (such as the Moderna, AstraZeneca, or Pfizer trials), comparing the antibody levels of participants who experienced "breakthrough" infections against those who did not.

By applying the fold-of-convalescent normalization, the authors found that these two seemingly different approaches converged. The "protection curves" generated from individual breakthrough data showed remarkable agreement with the broader vaccine-comparison models originally proposed in 2021.

Key Findings

  • A Gradient of Risk: The research clarifies that there is no "hard" threshold for immunity. Protection is not binary; rather, it exists on a gradient where higher antibody titers correlate with a lower probability of symptomatic infection.
  • The 50% Benchmark: The unified model estimates that the antibody level associated with 50% protection from symptomatic COVID-19 is approximately 20% (0.2-fold) of the mean convalescent titer.
  • Predictive Power: This consistency allows for the calculation of non-inferiority margins. For example, to achieve 80% efficacy against the ancestral strain, a vaccine candidate needs to elicit approximately 0.44 times the antibody levels seen in the Moderna vaccine trials.

Implications and Limitations

This unified framework provides a vital tool for the rapid approval of boosters and variant-specific vaccines. However, the authors note several caveats:

  • The Variant Factor: While the shape of the protection curve remains consistent, variants like Omicron require much higher absolute antibody levels because they are less sensitive to neutralization (the "fold-drop" effect).
  • Severe Disease vs. Infection: The study focuses on symptomatic infection. Protection against severe disease and death likely involves lower antibody thresholds and a more significant contribution from cellular immunity (T-cells), which were not the primary focus of this analysis.
  • Clinical Utility: Because individual lab assays still vary, it remains difficult for a single patient to use a commercial antibody test to determine their exact level of protection. Standardization at the laboratory level remains a work in progress.

Conclusion

By normalizing disparate datasets, this study proves that neutralizing antibodies are a highly reliable and consistent correlate of protection across different vaccine platforms. This alignment provides a scientific foundation for immunobridging, ensuring that the global response to COVID-19 can keep pace with an evolving virus without the constant need for massive clinical trials.

Summary updated

Abstract

Abstract Several studies show neutralizing antibody levels are an important correlate of immune protection from COVID-19 and have estimated the relationship between neutralizing antibodies and protection. However, a number of these studies appear to yield quite different estimates of the level of neutralizing antibodies required for protection. Here we show that after normalization of antibody titers current studies converge on a consistent relationship between antibody levels and protection from COVID-19.
BibTeX
@misc{khoury2022correlates,
  author = {David S. Khoury and Timothy E. Schlub and Deborah Cromer and Megan Steain and Youyi Fong and Peter B. Gilbert and Kanta Subbarao and James A. Triccas and Stephen J. Kent and Miles P. Davenport},
  title = {{Correlates of protection, thresholds of protection, and immunobridging in SARS-CoV-2 infection}},
  year = {2022},
  doi = {10.1101/2022.06.05.22275943},
  howpublished = {medRxiv},
}
Figure