Background
Coronavirus disease 2019 (COVID-19) is a respiratory illness caused by the β-coronavirus severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and is the cause of the pandemic [
1]. Humoral immunity is vital to combat and protect from SARS-CoV-2 infection [
2]. Therefore, understanding clinical factors affecting humoral protection over time is essential to understanding the disease caused by SARS-CoV-2.
Antibodies (Ab) against the receptor-binding domain (RBD) of the SARS-CoV-2 spike protein are crucial for developing immunological protection [
3]. Several factors may influence humoral responses to SARS-CoV-2 infection and vaccination, including increasing age, male sex and immunosuppression [
4‐
8]. However, how disease severity influences humoral responses, such as neutralizing antibodies (NAb) production, is not fully understood [
2]. A positive association between NAb titers, using different laboratory assays, and disease severity has been well described for up to 90 days after symptom onset [
9‐
14]. However, the effect of disease severity on NAb titers beyond 90 days after symptom onset is currently lacking.
An extensive review concluded that the association between viral load and disease severity is inconsistent [
15]. Therefore, assessing viral shedding and clinical characteristics affecting it is essential to identify and isolate infectious patients correctly and to further assess the inconsistent relationship between viral load, disease severity and humoral responses over time.
We conducted a prospective cohort study to evaluate humoral responses and live viral shedding in patients hospitalized with COVID-19. In addition, we explored whether clinical characteristics, such as disease severity, could affect NAb titers for up to 180 days and viral shedding for up to 30 days after study inclusion.
Materials and methods
Study design and population
Patients 18 years or older hospitalized at Copenhagen University Hospital—North Zealand, Denmark, between May 24, 2020, and May 5, 2021, were screened for COVID-19 at admission by routine collection and analysis of oropharyngeal swabs or tracheal aspirate samples. The swabs and aspirates were locally analyzed in a diagnostic reverse transcriptase-polymerase chain reaction (RT-PCR) assay as part of the hospital routine at admission. Inclusion criteria for the study were: (1) positive SARS-CoV-2 respiratory tract specimen (virological criteria) within 48 h of study inclusion, (2) consolidations on chest X-ray described by a radiologist or physician (radiological criteria) and (3) the presence of one or more of the following: temperature ≥ 38.0 °C, new-onset cough, pleuritic chest pain, dyspnea or altered breath sounds on auscultation (clinical criteria). Exclusion criteria were: (1) cognitive impairment prohibiting giving informed consent to participation and (2) by December 14, 2020, and onwards, if the time since symptom onset was more than seven days at the time of inclusion.
Variables and outcomes
Clinical variables extracted from the patient’s electronic medical records and the definition of immunocompromised status are described in the Additional file
1: appendix. Disease severity was defined based on the maximum required oxygen treatment during the hospitalization. Patients defined as having severe disease received high-flow nasal cannula (HFNC), invasive or non-invasive mechanical ventilation (NIV) treatment during the admission. The remaining patients were defined as having a mild disease.
Primary outcomes were defined as (1) NAb titers on days 0, 30, 90 and 180 and (2) viral load during the initial 30 days of inclusion. The secondary outcome was defined as the number of successful viral culturing attempts during the initial 30 days of inclusion.
Sample collection
Oropharyngeal swabs were collected using flocked swabs in a universal transportation medium (COPAN Italia S.p.A, Brescia, Italy). Oropharyngeal swabs and serum samples were collected on inclusion (day 0), days 3, 7, 10, 14, 17, 24 and 30 and serum was furthermore collected 90 and 180 days after study inclusion. In addition, a control oropharyngeal SARS-CoV-2 RT-PCR sample for immediate analysis was taken on day 14; if negative, no further oropharyngeal sampling was performed.
Laboratory analyses
See Additional file
1: appendix for a detailed description of reverse transcriptase-quantitative polymerase chain reaction (RT-qPCR), viral culturing and NAb assay methods.
RT-qPCR
All collected oropharyngeal samples were stored at − 80 °C. RT-qPCR and an attempt to culture virus from RT-PCR positive samples were performed on all swab samples using in-house analyses. Briefly, the RT-qPCR analysis targeted the SARS-CoV-2 RNA-dependent-RNA-polymerase (RdRp)-helicase gene region and two samples with known viral load were included in each PCR-run for quantification of patient samples [
16].
Viral cultures
SARS-CoV-2 was cultured in African green monkey kidney cells (VERO-E6) with incubation for 3–4 days and daily microscopic inspection for cytopathogenic effect (CPE) in accordance with the in-house procedures. A total of three passages were made before the virus was interpreted as non-replicant. In addition, cells with CPE were tested for SARS-CoV-2 RNA by RT-qPCR.
SARS-CoV-2 Ab
The presence of specific Ab against SARS-CoV-2 in serum was assessed by determining total-Ab by ELISA according to the manufacturer’s instructions (Wantai, Beijing, China). The Wantai ELISA used was reported to have 96.7% and ≥ 99% sensitivity and specificity, respectively. Detailed methods regarding the Wantai ELISA have been published elsewhere [
17].
Microneutralization assay
The microneutralization assay methods and validation used in this study has been published as a separate paper [
18]. Briefly, levels of neutralizing antibodies were determined using a median tissue culture infectious dose (TCID
50) microneutralization assay with an ELISA readout, further described in the Additional file
1: Appendix. Briefly, the 50% neutralization titers were calculated as the interception between a 4-parameter logistic regression curve fitted optical density values from each serum serial dilution and a 50% cut-off value, calculated from quadruplicate virus and cell control wells included on each plate. The titers were normalized according to a positive control on each assay plate to minimize inter-assay variation [
19,
20].
Statistical analysis
Mann–Whitney U test and Fisher’s exact test were used to compare groups. To present results in relation to symptom onset, the median time from symptom onset to sampling time point was added, as appropriate. A linear mixed-effect model (LME) with an unstructured covariance pattern was used to explore associations between repeated NAb titer measurements (dependent variable) and sample day, disease severity, age, sex, and disease severity/sample day interaction (fixed effects). Patient was used as random effect. The LME NAb model was further used to predict mean NAb titers at median days 7, 37, 97 and 187 from symptom onset. Samples exclusively from non-vaccinated patients at the time of sample collection were used in the NAb LME model. A generalized linear model (GLM) was used to assess the association between peak viral load, age, sex, and disease severity (dependent variable). Missing data analysis was conducted and missing completely at random (MCAR) was concluded for the dependent variable (NAb titer). All statistical analyses were performed in R Statistical Software (version 3.6.1) [
21].
Discussion
The major finding of this study was that high disease severity during admission was associated with higher NAb titers for up to 6 months after symptom onset in patients hospitalized due to COVID-19. Also, viral culturing from oropharyngeal swabs taken at hospital admission was difficult due to a long time between symptom onset and hospital admission. Finally, no association between peak viral load during admission and disease severity was observed.
Few studies have addressed whether NAb titers remain higher over time in patients with severe disease [
22]. Our data suggest that patients admitted with critical COVID-19 develop higher NAb titers and retain higher titers for at least six months after symptom onset compared to non-critically ill patients. These findings may indicate that patients with critical COVID-19 are better protected against reinfection after discharge as NAbs are strongly correlated with protection from reinfection [
23,
24]. Previous studies have found a strong correlation between the levels of anti-spike Ab and disease severity [
14,
25,
26]. Current studies also report similar findings regarding the association between the levels of anti-spike NAb and disease severity [
9‐
14], with one exception [
27]. These reports are however almost entirely based on pseudovirus assays. This is primarily due to live virus assays requiring BSL-3 facilities and are more time and resource-consuming. Live virus NAb assay is the most accurate method to assess antibody/virus interactions by assessing neutralization of the SARS-CoV-2 spike protein and all other parts of the SARS-CoV-2 virus [
28]. This study presents NAb results based exclusively on a live virus assay, which is the method closest to describing the reality of antibody/virus interactions during SARS-CoV-2 infection [
29].
The association between NAb titers and disease severity are not entirely understood, but two main explanations have been suggested. First, high disease severity could result from hyperinflammation, independent of viral load [
30‐
32] or second, high viral load leads to increased disease severity, which then, in turn, promotes antibody production [
15,
33]. However, our findings did not find associations between peak viral load during admission and disease severity, which then, in turn, would affect NAb titers. Therefore, our findings suggest that hyperinflammation is likely involved in the positive association between increasing NAb titers and disease severity.
In our study, the median time from symptom onset to first viral sample collection was seven days. Previous studies had suggested that a successful viral culturing attempt is highly dependent on samples with high viral load, where the probability was described as < 5% when the sample cycle threshold (ct) value was > 24 [
34‐
36]. Only six of all samples collected had a ct value < 24, of which one ended up being the only successful viral culturing attempt in the study. A plausible explanation for the abundance of samples with low admission viral loads is the time between symptom onset and sample collection. In addition, viral samples were stored for a median time of 8 months at − 80 °C, which could further affect the sample viral load at the time of analysis. Our data is in line with other studies suggesting that severe and critical SARS-CoV-2 infection can be characterized as a biphasic illness with a viral replication phase and a hyperinflammatory phase [
37]. Our results suggest that almost all patients were in the hyperinflammatory phase at admission. Future studies investigating SARS-CoV-2 infectiousness should focus on collecting viral samples 1–4 days after symptom onset to maximize the success rate of viral culturing attempts. Large-scale studies are needed to fully assess the risk of SARS-CoV-2 transmission at admission and further explore the clinical characteristics associated with the difference in NAb titers between disease severity groups.
In our study, none of the participants were known to be immunosuppressed. Immunosuppression is well known to affect both humoral responses after natural infection and vaccination and also the persistency of viral shedding and the neutralizing activity of antibodies, all factors that could have influenced our results in case of immunocompromised patients were included [
4,
36,
38‐
41]. Studies focusing on antibody responses and viral shedding in immunocompromised individuals are warranted in the future.
None of the included patients in this study were vaccinated or have had a previous SARS-CoV-2 infection. Furthermore, none of the included patients were vaccinated prior to admission. We therefore assume that none of the patients included in this study were primed by a previous infection, which otherwise could affect the results.
The emergence of new dominant SARS-CoV-2 variants has been described to be associated with changes in disease severity and the effectiveness of vaccines [
42,
43]. We did not have specific variant information at the individual patient level. However in Denmark, as was also the case worldwide, the wildtype-like variant containing the S:D614G mutation (formerly referred to as the Wuhan variant) was the dominant circulating variant until December 2020, where the alpha variant (B.1.1.7) quickly took over [
44]. The latter dominated until June 2021. In our study, the vast majority of patients were included when the wildtype-like variant by far still was the dominating circulating variant and were therefore most likely infected with this variant. None of the patients were vaccinated at the time of inclusion, which allowed for insight into the natural humoral response and association to the severity of disease of the SARS-CoV-2 wildtype-like variant.
Evidence regarding antibody and T cell cross-reactivity between SARS-CoV-2 and the four endemic coronaviruses (NL63, 229E, OC43, and HKU1) was established during the first year of the pandemic [
45‐
49]. Since then, studies exploring the clinical significance of antibody cross-reactivity have led to mixed results with no conclusive evidence regarding clinical outcomes [
50]. In our study, we could not perform the serological analyses necessary to assess cross-reactivity with other coronaviruses besides SARS-CoV-2. Therefore, we cannot rule out the possibility that cross-reactivity was present and potentially affecting our results. Our study had no specific inclusion criteria based on previous infections. Therefore, we assume that if antibody cross-reactivity was present, it would have been randomly distributed in the study population resulting in no overall changes in our comparisons between patient groups and sample time points.
The primary strength of our study is the prospective design with sample collection at predetermined time points for up to six months after inclusion. The included patients represent the general COVID-19 population hospitalized with x-ray confirmed pneumonia during the inclusion period. Furthermore, fully validated gold standard methods were used throughout the study. However, our findings are limited by relatively low sample size, preventing the possibility of generalization. The study was also affected by missing samples. In addition, our findings were mainly from a non-vaccinated population infected primarily with the wild-type (Wuhan-Hu-1) or alpha (B.1.1.7) SARS-CoV-2 variants. Therefore, the findings will not necessarily be translatable to a vaccinated population or populations infected with a different SARS-CoV-2 variant.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.