Exploratory assessment of serological tests to determine antibody titer against SARS‐CoV‐2: Appropriateness and limits

Abstract Background Serological tests can be used to detect antibodies in the serum of subject's after SARS‐CoV‐2 infection and vaccination. Currently, variability in antibody titers and the availability of a multiplicity of serological tests have made it necessary to highlight their appropriateness and limitations in various diagnostic settings. Methods This study is part of Covidiagnostix, a multicenter project aimed at the assessment of the health technology used in SARS‐CoV‐2 serological tests. Based on data gained from the analysis of over 5000 subjects, a selected number of serum samples, representative of different diagnostic settings, were analyzed first by qualitative immunoassays (IgA, M, and G MILLIPLEX® SARS‐CoV‐2 tests based on Luminex®) to define the immunoglobulins serum composition and subsequently by four serological diagnostic tests (Elecsys Anti‐SARS‐CoV‐2 and Elecsys Anti‐SARS‐CoV‐2 S by Roche, SARS‐CoV‐2 IgG by Siemens Healthcare, and CHORUS SARS‐CoV‐2 “NEUTRALIZING” Ab by DIESSE). The first WHO International Standard for SARS‐CoV‐2 was also analyzed using the same methods. Results This study evaluated the antibody content and titer of the WHO Standard and serum of subjects with/without previous infection and before/after vaccination for SARS‐CoV‐2. Conclusion The definition of antibodies in the WHO standard and the analysis of serum samples allowed for the identification of the appropriateness of serological tests in each diagnostic setting, increasing the effectiveness of the resulting laboratory data. Furthermore, we found that it would be optimal to produce new international standards against the S1 domain and RBD of the SARS‐CoV‐2 spike protein for a more effective serological monitoring of vaccination.


| INTRODUC TI ON
The current gold standard for diagnosing severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection is "real-time reverse transcription polymerase chain reaction (RT-PCR)," which identifies the viral genome in samples taken from the respiratory tract and is particularly effective in the acute phase. On the other hand, serological tests allow for the detection of the presence of antibodies in the subject's serum from one to several weeks after infection or vaccination, which is the time necessary to produce antibodies. Serology is essential both for diagnosis, especially for patients with mild/moderate coronavirus disease 2019 (COVID- 19), who may present beyond the first 2 weeks of illness onset, 1 as well as for the monitoring of the host immune response to viral antigen exposure.
It is essential to underline that the antibody titer varies between immunity due to natural infection and vaccination. 2  Therefore, humoral immunoglobulin G (IgG) forms later than IgM but is characterized by a higher specificity and guarantees a longer term protection than IgM. Long-lasting protection is ensured by B cells that differentiate into memory B cells. In the event of a new encounter with the same antigen, they differentiate into plasma cells to rapidly produce high specificity IgG. The efficacy of vaccines that guarantee long-term protection and the production of specific IgG through the involvement of memory B cells is based on this mechanism. 3 Circulating IgA antibodies appear 4-24 days after infection, appearing after 11 days in most cases. The levels of IgM antibodies are detectable from 4 to 14 days after infection and increase until about the 20th day (peaking between 2 and 5 weeks), after which they begin to disappear, declining over 3-5 weeks post-symptom onset.
The IgG antibodies become detectable 12-15 days from infection, that is, at a later time compared to the IgM, with a peak between 3 and 7 weeks and the ability to persist for at least 8 weeks. 4 Considering the antigen specificity of the immunoglobulins, upon viral infection, the humoral immune system responds by producing antibodies against multiple SARS-CoV-2 proteins, including the spike (S) and the nucleocapsid (N) protein. The spike (S) proteins form the characteristic "corona," or crown, of the virus and are composed of subunit S1, which contains the receptor-binding domain (RBD), and subunit S2, containing the fusion peptide. The spikes surround the membrane glycoprotein and the envelope protein, containing the viral RNA encased by the N protein. 5 Upon vaccination, the humoral immune system is able to potentially develop antibodies against spike proteins but not against N proteins. 6 All IgA, IgM, and IgG can be measured in blood serum and plasma samples. 4 In vitro serological tests detecting the presence of specific antibodies are used to reveal past infections and vaccine reactivity.
By testing the response of each type of immunoglobulin against specific antigenic regions of SARS-CoV-2, it is also possible to track the immune response to the virus during COVID-19 infection and recovery. Therefore, these tests are essential for epidemiological assessments of population seroprevalence and forward-looking estimates of global therapeutic needs. and total Ig in serum or plasma samples with a generally high sensitivity and specificity. Although their sensitivity is very low in the first week after the onset of symptoms, this improves in the second week and reaches its maximum at 21 days after infection 7 or vaccination.
However, many of these commercially available serologic tests have been recalled due to their poor performance. In fact, the majority have not been fully evaluated with large panels of samples, stressing the importance of their systematic validation or a health technology assessment (HTA) approach. 8 To maximize the informativeness of the serological tests, there is a need to specify the antibodies that they determine and at which limits. As the antibody responses reflect exposure to virus and vaccination, to ensure the appropriateness of the request, the effectiveness, and the correct interpretation/communication of the laboratory data, it is essential to assess serological tests in different diagnostic settings.
Aiming to improve the effectiveness of laboratory data, this study focuses on the definition of the most appropriate serological test to be used in the different diagnostic settings (patient seropositive to SARS-CoV-2 or vaccinated subjects). To reach this objective, the main goal of the present study is to perform a qualitative analysis of the antibodies (class and antigen recognition) present in the sera of specific groups of patients (with and without previous SARS-CoV-2 infection and before/after vaccination) and in the WHO standard was performed.

| Study design
This study is part of Covidiagnostix, a multicenter national project granted by the Italian Ministry of Health aimed to evaluate several diagnostic serological tests available on the market for SARS-CoV-2 using a health technology assessment (HTA) approach. 8

| Serum samples
The samples were selected in the context of different diagnostic settings: • Nine serum samples obtained from subjects infected by the virus and not vaccinated, denoted as "natural seropositive," were sampled 20 days after the onset of symptoms.
• Five serum samples obtained from subjects who recovered from a previous SARS-CoV-2 infection and then received one dose of the BNT162b2 mRNA COVID-19 vaccine, denoted as "vaccinated natural seropositive"; they were sampled 21 days after the vaccination.

| Multiplex assays
The three different MILLIPLEX ® SARS-CoV-2 tests (Millipore Sigma), a bead-based multiplex assay based on Luminex ® xMAP ® technology, were used to detect the presence of IgA, IgG, and IgM against SARS-CoV-2 spike protein subunits S1 and S2, RBD and N in the 38 samples and the first WHO international standard.
Intra-assay precision results for all three panels were declared to be <15% CV, calculated from the mean of the %CVs from eight reportable results in a single assay. Inter-assay precision for all three panels was <20% CV, calculated from the mean of the %CVs across four different assays.

| Diagnostic serological tests
The serum samples and the first WHO International standard were also tested to determine the antibody titer against SARS-CoV-2 antigens, using: • The Elecsys Anti-SARS-CoV-2 (Roche Diagnostics, Inc.), a highthroughput ECLIA qualitative method to detect Pan-Ig against the Nucleocapsid protein.
• The SARS-CoV-2 IgG (sCOVG) (Siemens Healthcare Diagnostics Inc.), a high-throughput CLIA qualitative and semi-quantitative approach to detect IgG against the spike RBD.
The characteristics of the diagnostic serological tests are listed in Table 2.

| Statistical analysis
Receiver-operating characteristic was used to set putative thresholds in a multiplex assay to identify positive and negative samples.
The pre-pandemic and natural seropositive samples were used to establish experimental cutoffs. Thresholds for positivity were determined maximizing Youden's index ( Table 1).
Kolmogorov-Smirnov test was used to assess the normality of the data distribution, and unpaired ANOVA or Kruskal-Wallis tests were used to compare multiplex assay data among natural seropositive, seronegative, or seropositive vaccinated subjects.
One-vs-one Spearman's correlation coefficients were calculated between the values obtained from each method and the experimen- Antibody titers demonstrated a beta distribution determined by comparing theoretical and empirical distribution functions. Then, beta regression models (using R package "betareg") were used to assess the contribution of the different antibody classes against the different antigens to determine the test results.

| RE SULTS
The serum samples and the first WHO international standard were simultaneously tested by the MILLIPLEX ® SARS-CoV-2 antigen panels for IgA, IgM, and IgG antibodies against SARS-CoV-2 S1 S2, RBD, and N antigenic proteins, as reported in Figure 1 (panels A-E).
The first WHO International standard sample ( Figure 1B) shows MFI levels above the determined threshold for all the antigenic proteins of IgA, IgM, and IgG, proving that all these antibodies are present in the standard. The MFI levels defined in the natural seropositive were similar to the WHO standard ( Figure 1C).
The natural seropositive vaccinated subjects ( Figure 1D) showed MFI levels above the IgA, IgM, and IgG threshold against S1, S2, and RBD antigens.
The seronegative vaccinated subjects ( Figure 1E) show MFI levels above the IgA, IgM, and IgG threshold against S1, S2, and RBD

antigens.
Both seronegative and seropositive subjects were negative for IgM against N antigen. Vaccinated seronegative subjects were also negative for IgA and IgG against N antigen. Table 3 summarizes the positivity and negativity for IgA, IgM, and IgG against different epitopes in natural seropositive patients and vaccinated subjects. Thresholds obtained from the ROC curves (Table 1) were used as internal references to define the results shown in Table 3.
In addition to the differences concerning the presence/absence of antibodies against N antigen, the comparison of seronegative vaccinated subjects with natural seropositive showed significantly lower levels (p < 0.05) of all the three immunoglobulins against S2 antigen in seronegative vaccinated. On the contrary, higher IgG levels against S1 and RBD antigens were observed in seropositive vaccinated subjects compared to natural seropositive subjects. The comparison of data obtained from seronegative and seropositive vaccinated subjects showed that those that were vaccinated seropositive had higher levels of IgG against S1 (p < 0.05), S2 (p < 0.001), and RBD (p < 0.05) compared with vaccinated seronegative subjects.
Using multiple beta regression models to assess the antibodies contributing the most to the results of each specific test, it emerged that the Elecsys Anti-SARS-CoV-2 S test was significantly influenced by IgG against RBD only (p < 0.0001), the CHORUS SARS-CoV-2 "NEUTRALIZING" Ab test results depended substantially on all three antibody class anti-S1 (IgG: p < 0.0001, IgA: p = 0.004, IgM: p < 0.0001), and the results of the SARS-CoV-2 IgG (sCOVG) depended mainly on anti-S1 IgG (p < 0.0001) rather than IgG anti-RBD (p = 0.0001).

Finally, the only test against N antigen was Elecsys Anti-SARS-
CoV-2; consequently, the serological results obtained did not correlate with any other diagnostic serological test.

| DISCUSS ION
This study provides for the first time representative expected antibody profile (class and antigen recognition) of the serum of subjects with and without previous infection by SARS-CoV-2 and before/ after vaccination with BNT162b2. Further information concerning the composition of the WHO standard in terms of immunoglobulins against specific antigens was also reported and is fundamental to harmonize the results obtained from the wide range of inherently different assays available.
The main findings of the study confirmed that IgG against N antibodies are the best choice to identify subjects naturally exposed to SARS-CoV-2, while also indicating and suggested that IgG against S1, S2, and RBD should be used to monitor the antibody response to the vaccine.

Roche Diagnostic
Inc.

High-throughput ECLIA (qualitative)
Nucleocapsid Pan-Ig To assess the appropriateness of the request, serological tests were divided into those that recognize the nucleocapsid protein and those that recognize the spike protein or its components. The natural seropositive samples strictly relate to WHO standards for all antibody classes and antigens, suggesting that these samples are representative of the expected antibody content for subjects who were infected by SARS-CoV-2. Therefore, the N antibody test identifies these natural seropositive individuals compared to uninfected subjects, whether vaccinated or not. 7,14,15 In subjects who received the BNT162b2 mRNA COVID-19 vaccine without having previously contracted the virus, these tests give In our samples, IgG against S1, S2, and RBD were present in natural seropositive, vaccinated seropositive, and vaccinated seronegative subjects, with IgG against S1 and RBD highly present in the upper respiratory tract and adds to growing evidence that a vaccination that elicits a mucosal response is required. [18][19][20][21] TA B L E 3 Positivity and negativity for IgA, IgM, and IgG against S1, S2, RBD, and N in natural seropositive, natural seropositive vaccinated, and seronegative vaccinated subjects obtained by multiplex assays IgA IgM IgG S1 S2 RBD N S1 S2 RBD N S1 S2 RBD N Natural seropositive + + ± ± ± + ± + + + + +

Natural seropositive vaccinated
In the presence of some samples below the positivity threshold, this was indicated as ±.

F I G U R E 2 Correlation plot
SARS-CoV-2 IgG (sCOVG) by Siemens, and CHORUS SARS-CoV-2 "NEUTRALIZING" Ab by Diesse), a particularly strong correlation was observed between the serological tests by Siemens and Diesse; these two tests were set up to recognize IgG and all classes of immunoglobulins, respectively, against S1 antigen (and RBD in the case of Siemens).
On the other hand, the Elecsys Anti-SARS-CoV-2 S by Roche specifically recognizes RBD antigen (all Ig classes) and showed only a strict relationship with IgG. The lack of S1 recognition may explain the observed reduced correlation with the other anti-S1 tests.
Nevertheless, the Roche test would be less sensitive to variations in IgA and IgM titer occurring during time compared to the other Pan-Ig test by Diesse.
The SARS-CoV-2 IgG (sCOVG) results by Siemens against S1 and RBD antigens appeared to depend strongly on anti-S1 antibodies rather than anti-RBD.
According to its design, the Diesse test effectively identifies all anti-S1 antibody classes. Based on these observations, assays based on IgG against S1 antigens (or eventually designed exploiting IgG against RBD) should be selected to monitor the immune response after the vaccine.
The small sample size represents the main limitation of the study; thus, these preliminary data will need to be verified on a larger cohort of subjects. Other limitations are related to the fact that the current data represent a reference only for subjects undergoing vaccines based on mRNA or DNA to produce S proteins, and should be considered as strictly related to the time-point explored after vaccination and to the seronegative vaccinated subjects in terms of serological observations.

| CON CLUS ION
Serological analysis is capable of defining the antibody profile in classes of immunoglobulins and antigens against SARS-CoV-2 in natural seropositive, vaccinated natural seropositive, and vaccinated seronegative subjects and the WHO standard. As expected, the natural seropositive samples strictly relate to the WHO standard for all antibody classes; moreover, the most appropriate tests to identify individuals naturally exposed to SARS-CoV-2 are those based on recognizing IgG against N antibodies.
Due to the relevant presence of IgG against S1, S2, and RBD in natural seropositive, vaccinated seropositive, and vaccinated seronegative subjects, with differences among groups and to the observed correlation among serological results, tests exploiting IgG against S1 or RBD antigens should be selected to monitor the immune response after vaccination.
This study highlights the need to produce new international standards against the S1 and RBD domains of the SARS-CoV-2 spike protein, preferably on an industrial scale, allowing for the use of serology to monitor vaccination reactivity more effectively.

CO N FLI C T O F I NTE R E S T S
The authors declare that they have no conflict of interest. SARS-CoV-2 IgG (sCOVG) IgG, S1 and RBD CHORUS SARS-CoV-2 Pan-Ig, S1

PATI E NT CO N S E NT S TATE M E NT
Informed consent was obtained from all subjects involved in the study.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data presented in this study are available on request from the corresponding author.