Longitudinal detection of SARS‐CoV‐2‐specific antibody responses with different serological methods

Abstract Serological testing for anti‐severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) antibodies is used to detect ongoing or past SARS‐CoV‐2 infections. To study the kinetics of anti‐SARS‐CoV‐2 antibodies and to assess the diagnostic performances of eight serological assays, we used 129 serum samples collected on known days post symptom onset (dpso) from 42 patients with polymerase chain reaction‐confirmed coronavirus disease 2019 (COVID‐19) and 54 serum samples from healthy blood donors, and children infected with seasonal coronaviruses. The sera were analyzed for the presence of immunoglobulin G (IgG), immunoglobulin M (IgM), and immunoglobulin A (IgA) antibodies using indirect immunofluorescence testing (IIFT) based on SARS‐CoV‐2‐infected cells. They were further tested for antibodies against the S1 domain of the SARS‐CoV‐2 spike protein (IgG, IgA) and against the viral nucleocapsid protein (IgG, IgM) using enzyme‐linked immunosorbent assays. The assay specificities were 94.4%–100%. The sensitivities varied largely between assays, reflecting their respective purposes. The sensitivities of IgA and IgM assays were the highest between 11 and 20 dpso, whereas the sensitivities of IgG assays peaked between 20 and 60 dpso. IIFT showed the highest sensitivities due to the use of the whole SARS‐CoV‐2 as substrate and provided information on whether or not the individual has been infected with SARS‐CoV‐2. Enzyme‐linked immunosorbent assays provided further information about both the prevalence and concentration of specific antibodies against selected antigens of SARS‐CoV‐2.


| INTRODUCTION
In the current pandemic, direct pathogen detection via reverse transcription and polymerase chain reaction amplification as well as Seroconversion of anti-SARS-CoV-2 antibodies can occur at different points in time after virus contact. 1,2 The features of immune responses to SARS-CoV-2 infections vary significantly between individuals, 3 especially regarding the kinetics, immunoglobulin classes, and antigen specificity. In the majority of COVID-19 patients, anti-SARS-CoV-2 antibodies are detectable within two weeks after infection. [4][5][6] Usually, specific immunoglobulin M (IgM) and immunoglobulin A (IgA) antibodies are detectable earlier than specific immunoglobulin G (IgG) antibodies. 5,7,8 In individual cases, anti-SARS-CoV-2 antibodies are either only detectable more than four weeks after onset of symptoms or not at all due to generally absent antibody secretion. [8][9][10] Anti-SARS-CoV-2 antibodies target different structural proteins of SARS-CoV-2. The main immunogens are the spike and nucleocapsid proteins. The highly immunogenic S1 domain of the spike protein of SARS-CoV-2 is a major target for neutralizing antibodies and is being used as the antigen in many serological assays. 11 The immunologically relevant receptor-binding domain (RBD) represents another important target antigen for virus-neutralizing antibodies. 12 The nucleocapsid protein (NCP) of SARS-CoV-2 is the antigen with the strongest immune dominance among Coronaviridae 13 and contains diagnostically relevant epitopes of SARS-CoV-2. Previous studies suggested heterogeneous binding antibody responses to S1/ RBD and NCP viral antigens, 14  Currently, knowledge about SARS-CoV-2 antibody persistence is scarce, although it would help to understand the possible role of humoral immunity in the protection against reinfection. The aim of this study was to study the kinetics of antibodies against SARS-CoV-2 and to explore the characteristic features of eight serological assays.

| Human serum samples
Panel A comprised 82 sequential and single serum samples from 25 German patients (Table 1). Infection with SARS-CoV-2 was confirmed by PCR 15 by regional health authorities. These patients had mild to moderate COVID-19 symptoms.
These patients required hospitalization.
All patient samples were also serologically precharacterized by indirect immunofluorescence testing (IIFT).
Panel C comprised serum samples taken before August 2019 from 42 healthy German blood donors (Table 1).

Panel D comprised serum samples taken between January and
March 2020 from twelve German children (Table 1) positive for IgG against seasonal coronaviruses (e.g., HCoV 229-E) by indirect immunofluorescence testing (IIFT, for research use only).

| Detection of anti-SARS-CoV-2 antibodies
The detection of antibodies against SARS-CoV-2 (genus: Betacoronavirus, family: Coronaviridae) using IIFT was performed with anti-IgG-, anti-IgA-, and anti-IgM-fluorescein isothiocyanate-labeled secondary antibodies on infected Vero E6 cells fixed in acetonemethanol. 16  The detection of SARS-CoV-2-specific antibody responses was also investigated with respect to the infection phase. As the diagnostic window for serological testing opens several days after pathogen contact, only samples taken later than ten dpso were  Table 1).
The overall agreement between the qualitative results obtained with the anti-SARS-CoV-2 ELISA (IgG) and Anti-SARS-CoV-2 Quan-tiVac ELISA (IgG) was calculated, their degree of agreement was quantified using Cohen's κ including borderline results, 18 and the statistical association between results was described using Pearson correlation and 95% confidence intervals as determined by Clopper-Pearson interval.
Positive results for anti-SARS-CoV-2 IgG antibodies against S1     IgM ELISA by 46.1%, again reflecting the wider antigenic spectrum in IIFT.
The ELISAs, in contrast, provide information about the prevalence of specific antibodies against selected antigens of SARS-CoV-2. Hence, lower sensitivities of the S1-specific ELISAs compared to the NCP-specific ELISA probably reflect the known fact that not all infected individuals produce antibodies against the S1 domain of SARS-CoV-2. 10 Importantly, previous research showed that responses of specific IgG against S1 and NCP may be heterogeneous between individuals, time-delayed and do not always coincide with each other. 8,12,14 In the present panels, the prevalence of specific IgG antibodies against NCP in the early phase of infection was higher than that against S1 (Table 3) Exclusively in the early phase of infection, the prevalence of specific IgA antibodies against S1 was higher than that of specific IgG antibodies against NCP as well as S1. This observation reflects that of Okba et al. 1 However, it is in contrast to a previous study that showed a higher sensitivity of the Anti-SARS-CoV-2 IgG compared to the Anti-SARS-CoV-2 IgA ELISA in patient samples taken later than fourteen dpso, 19 whereby the discrepancy might be due to heterogeneous definitions regarding the early phase of infection.
The IgA IIFT showed a pronounced decrease in the antibody detection rate after 60 dpso, which was not observed for the IgA ELISA (Table 3). A possible explanation for this might be that the IgA antibody response against the S1 protein largely remains constant, while the production of IgA antibodies against other antigens of SARS-CoV-2 decreases.
More patients were seropositive for IgM by IIFT than by ELISA ( Table 2, Table 3), which could be accounted for by the low sensitivity of the NCP IgM ELISA, warranting further investigations.
However, the continuously low sensitivity of the NCP-specific IgM ELISA (Table 3)   | 5821 sensitivity of 55% at week 3-4 after disease onset. 20 Liu et al also observed a higher sensitivity of an ELISA based on the spike protein compared to an NCP-based ELISA for detection of IgM antibodies. 8 Two months after symptom onset, we observed a decline in the sensitivity of both IgM-specific assays ( Table 3). Independent of the serological method, the two IgM-specific assays reached maximal sensitivities between 11 and 20 dpso (Table 3) and could therefore especially be applied to detect antibodies in samples taken during the early phase of infection. If patients develop specific IgM against NCP, these antibodies seem to be present for only a short time during the early phase of infection. A sharp decline in the IgM prevalence is to be expected because isotype switching of virus-specific B-cells from IgM to IgG antibody production causes a decline in circulating IgM. 21 The fact that SARS-CoV-2-specific IgM is detected mostly in the early infection phase but only in rare cases 22 invites the question of whether all isotypes should be measured during serodiagnostics.
The agreement analysis revealed a very high correlation between results obtained with the anti-SARS-CoV-2 ELISA (IgG) and the QuantiVac ELISA (IgG). The two samples that showed inconsistent qualitative results between these assays (Table 4, Figure 2) were taken relatively early and late (7 and 116 dpso) in the course of the disease. An explanation for these inconsistencies might be that the assays were incubated using the same aliquot but on different days, hence the experimental conditions might have differed slightly.
Another reason might be that the artificial division between positive and negative results does not match the natural range of activity of some samples.
In general, the use of cells infected with the whole SARS-CoV-2 as a substrate has the great advantage of obtaining a high sensitivity due to the presence of the complete antigenic spectrum, as evident in the present IIFT results ( Table 2). This is, however, linked to the disadvantage that a positive IIFT result does not allow for a con- The presence of anti-SARS-CoV-2 S1/RBD IgG antibodies seems to correlate with the development of both virus neutralization and immunity. 1,3,23 Previous research found that titers of neutralizing antibodies were significantly correlated with the levels of anti-RBD IgG, 12 and RBD-specific IgG titers were suggested as a surrogate of neutralization potency against SARS-CoV-2 infection. 24 Nevertheless, it is possible that a patient does not develop antibodies against S1 of SARS-CoV-2, but only against NCP. However, this would suggest that neutralizing antibodies might not be present since binding antibodies against NCP seem to correlate to a lesser degree with immunity than binding antibodies against S1/RBD. 25 The A detailed analysis of potential associations between antibody kinetics and disease severity was not performed because symptoms were not systematically recorded and the disease severity could therefore not be rated other than that patients in Panel A had no or mild symptoms and patients in Panel B required hospitalization.
Nevertheless, the assay sensitivities were also reported for each panel separately ( Germany). All samples were processed anonymously.