Probabilistic classification of anti-SARS-CoV-2 antibody responses improves seroprevalence estimates

Abstract

Objectives: Population-level measures of seropositivity are critical for understanding the epidemiology of an emerging pathogen, yet most antibody tests apply a strict cutoff for seropositivity that is not learnt in a data-driven manner, leading to uncertainty when classifying low-titer responses. To improve upon this, we evaluated cutoff-independent methods for their ability to assign likelihood of SARS-CoV-2 seropositivity to individual samples.

Methods: Using robust ELISAs based on SARS-CoV-2 spike (S) and the receptor-binding domain (RBD), we profiled antibody responses in a group of SARS-CoV-2 PCR+ individuals (n = 138). Using these data, we trained probabilistic learners to assign likelihood of seropositivity to test samples of unknown serostatus (n = 5100), identifying a support vector machines-linear discriminant analysis learner (SVM-LDA) suited for this purpose.

Results: In the training data from confirmed ancestral SARS-CoV-2 infections, 99% of participants had detectable anti-S and -RBD IgG in the circulation, with titers differing > 1000-fold between persons. In data of otherwise healthy individuals, 7.2% (n = 367) of samples were of uncertain serostatus, with values in the range of 3-6SD from the mean of pre-pandemic negative controls (n = 595). In contrast, SVM-LDA classified 6.4% (n = 328) of test samples as having a high likelihood (> 99% chance) of past infection, 4.5% (n = 230) to have a 50-99% likelihood, and 4.0% (n = 203) to have a 10-49% likelihood. As different probabilistic approaches were more consistent with each other than conventional SD-based methods, such tools allow for more statistically-sound seropositivity estimates in large cohorts.

Conclusion: Probabilistic antibody testing frameworks can improve seropositivity estimates in populations with large titer variability.

Keywords: COVID‐19; SARS‐CoV‐2; antibody responses; antibody testing; probability; serology.

Figure 1

Anti‐SARS‐CoV‐2 Ab responses in RT‐PCR+ and vaccinated individuals are spread over a wide titer range. (a) IgM, IgG and IgA anti‐S and –RBD responses in individuals RT‐PCR+ for SARS‐CoV‐2 RNA (COVID‐19 patients and hospital staff, n = 138). A small number of healthy controls (HC, pre‐pandemic samples) are shown for each assay and isotype. (b) Anti‐S isotype‐level responses according to COVID‐19 clinical status. Cat 1: mild/asymptomatic. Cat 2: hospitalized. Cat 3: Intensive care. (c) Anti‐RBD isotype‐level responses according to COVID‐19 clinical status. (d) Anti‐S responses in RT‐PCR+ cases (n = 105), RT‐PCR+ hospital staff (HS, n = 33), blood donor (n = 1000) and pregnant women (n = 1000) serum samples collected during the first three months of the pandemic. 3 and 6 SD cutoffs calculated from n = 595 historical control samples are shown by dashed and solid red lines, respectively. (e) Neutralizing ID₅₀ titers in RT‐PCR+ individuals and a subset of healthy donors (n = 56) collected during the first three months of the pandemic. (f) Anti‐S and ‐RBD IgG responses 3 months post‐boost in individuals vaccinated with either BNT162b2 (n = 10), mRNA‐1273 (n = 10) or ChAdOx1 (n = 10) COVID‐19 vaccines. 3 and 6 SD cutoffs are shown by dashed and solid red lines, respectively. Error bars represent the geometric mean with 95% CIs.

Figure 2

Low‐titer responses are difficult to classify using conventional assay thresholds. (a) Anti‐S and RBD‐ IgG responses in pre‐pandemic negative controls (n = 595) and blood donor and pregnant women test data (n = 5100). 3 and 6 SD cutoffs based on all negative control values are shown by dashed and solid red lines, respectively. (b) Anti‐S vs. ‐RBD IgG responses. 10% of samples were seropositive against both antigens at 6 SD, while 7.2% of values were of uncertain serostatus, depending on the assay and cutoff used.

Figure 3

Likelihood of past SARS‐CoV‐2 infection at the individual level in blood donors and pregnant women test data. (a) Schematic representation of the probabilistic learner strategy for estimating probability of seropositivity. Training data consisted of n = 138 RT‐PCR cases and n = 595 pre‐pandemic negative controls. (b) Individual probability of past infection in blood donor and pregnant women test data (n = 5100) according to the SVM‐LDA learner. (c) Number of samples per % chance interval in the test data according to SVM‐LDA. (d) For test samples with > 50% chance of past infection, the proportion in different intervals according to SVM‐LDA.