# Comprehensive analysis of human coronavirus antibody responses in ICU and non-ICU COVID-19 patients reveals IgG3 against SARS-CoV-2 spike protein as a key biomarker of disease severity

## TL;DR

This study finds that higher levels of IgG3 antibodies against the SARS-CoV-2 spike protein are linked to severe COVID-19, offering a potential biomarker for disease severity.

## Contribution

The study identifies IgG3 antibodies against the SARS-CoV-2 spike protein as a novel biomarker for predicting ICU admission in COVID-19 patients.

## Key findings

- ICU patients showed higher anti-S1 IgG and IgA responses compared to non-ICU patients.
- IgG3 against the SARS-CoV-2 spike protein was the most influential variable in predicting ICU status.
- Machine learning models using antibody subclass data effectively distinguished ICU from non-ICU patients.

## Abstract

Introduction. Pre-existing immunity to human coronaviruses (HCoVs) may shape the immune response in COVID-19 patients. Increasing evidence suggests that immune cross-reactivity between SARS-CoV-2 and other coronaviruses may determine clinical prognosis.

Hypothesis. SARS-CoV-2 disease severity is influenced by pre-existing immunity to HCoVs, with distinct antibody profiles and cross-reactivity patterns.

Aim. To investigate the antibody response of ICU and non-ICU SARS-CoV-2 patients against different HCoV proteins and assess the potential impact of pre-existing immunity on SARS-CoV-2 disease outcomes.

Methodology. This study used a comprehensive HCoVs antigen bead array to measure antibody response to pathogenic Middle East respiratory syndrome coronavirus (MERS-CoV), SARS-CoV, SARS-CoV-2 and the four seasonal HCoVs in 70 ICU and 63 non-ICU COVID-19 patients.

Results. Our analysis demonstrates an overall higher antibody response in ICU than in non-ICU COVID-19 patients. Interestingly, the anti-S1 IgG and IgA were significantly higher among ICU than in non-ICU patients. Similarly, the anti-S1 IgG against NL63 showed a lower response among ICU compared to non-ICU. Cross-reactivity was evident between SARS-CoV-2 and SARS-CoV antibodies but not with MERS-CoV and seasonal HCoVs. The subclass analysis of antibodies recognizing SARS-CoV-2 revealed that anti-S1 IgG1, IgG3, IgA1 and IgA2 were significantly higher in ICU compared to non-ICU. The predominant IgA subtype among SARS-CoV-2 patients was IgA1. We applied machine learning algorithms to subclass serological responses to build classifiers that could distinguish between ICU patients and patients with milder COVID-19. Out of 90 variables used in two different types of models, the variable of highest influence in determining the ICU status was IgG3 against SARS-CoV-2 S, and the top 8 variables of influence included the presence of IgG3 against S-trimer as well as IgA against SARS-CoV-2 S.

Conclusion. Understanding the complexities of humoral immunity in various patients is critical for early medical intervention, disease management, selective vaccination and passive immunotherapy.