Background
Since the surge of the SARS Corona Virus 2 (SARS-CoV-2) pandemic, considerable progress has been made regarding diagnosis, treatment and prevention of COVID-19. Although by mid-2022, more than 545 million people have been infected and more than 6 million died, serological responses following infection or vaccination are still not fully understood and a correlate of protection has not been identified yet [
1,
2].
Describing the natural course of the disease in detail may be key to understanding the immune mechanisms and subsequent protection, either through previous infection or vaccination, or both. Natural infection with SARS-CoV-2 reduces the risk of subsequent infections with the wild-type virus by 82–89% for at least 6 months [
3,
4]. In addition, SARS-CoV-2 vaccines protection against symptomatic COVID-19 disease was reported to be 95% for BNT162b2, 94% for mRNA 1273, 70% for ChAdOx1 and 50% for Sinovac [
5‐
9]. The difference in the estimated protective effect of the vaccines correlates with the elicited immune responses, which have been considerably higher in the mRNA vaccines compared to vector-based products [
10]. The longevity of this protective effect however, is a matter of debate.
In addition, viral variants of SARS-CoV-2 have emerged since and acquired immune protection was found to be reduced [
11]. Large studies have now demonstrated breakthrough infections in vaccinated individuals even during the peak of the antibody response, i.e., weeks or months after completion of the vaccination course [
12‐
15]. A waning of the immune response against SARS CoV-2 was suggested by Mizrahi et al. [
14], demonstrating a 1.5 times increased risk for breakthrough infections with the Delta-variant 6 month after vaccination with BioNTech/Pfizer, compared to a 3 month time lag. In an in-house study in early 2022 we observed many breakthrough infections with the Omicron variant regardless of vaccination status or previous infections, including recent infection with the Delta-variant.
Several serological studies have tried to estimate the duration and dynamics of antibody responses following SARS-CoV-2 infection, yielding ambiguous results. Long et al. [
16] reported rapid waning of nucleocapsid antibodies in the first 3 months after infection [
16,
17]. Similarly, Ibarrondo et al. [
18] described a half-life of antibodies against the receptor binding domain (RBD) of 36 days. In contrast, Dan et al. [
19], Flehmig et al. [
20] and Ripperger et al. [
21] reported that immunity against RBD and the anti-Spike domain persisted for at least 7 months. The reasons for these different reports may be e.g. heterogeneity of population, assays used etc.
Several studies highlighted considerable differences in the readout of serological assays, indicating a hampered cross-comparison. In a report by Harris et al. [
22], anti-nucleocapsid antibodies measured with the SARS-CoV-2 IgG Assay from Abbott (Abbott Diagnostic, IL, USA), or anti S1 antibodies measured with the Euroimmun Anti-SARS- CoV-2 ELISA IgG (Euroimmun, Lubeck, Germany) were declining within few months. Similarly, plasma from the same subjects measured for anti-nucleocapsid or anti RBD antibodies respectively using the Elecsys Anti-SARS-CoV-2 Roche assays (Roche, Mannheim, Germany), demonstrated stable values over the same time [
22,
23].
To improve standardization of serological anti-spike measurements, the WHO made available a set of well-characterized samples deriving from SARS-CoV-2-recovered individuals and shipped by late 2020/early 2021 to laboratories requesting it [
24]. These samples were subsequently used to normalize results of different quantitative anti-Spike test systems to standardized units termed “BAU” (Binding Antibody Units) [
25]. Many different manufacturers have since published correction factors or formulas to calculate BAU values from their quantitative anti-Spike assays [
26]. In addition, laboratories have provided SARS-CoV-2 serology results in BAU to patients and physicians in routine care [
27,
28]. Of note, the use of this WHO standard was encouraged to cross validate internal standards, effectively generating a chain of standards [
29].
Following the suggested approach, we compared anti-spike antibody titres quantitatively at defined and standardized time points spanning over 18 months after infection using different commercially available test kits. Therefore, we used samples derived from ninety-nine SARS-CoV-2-infected individuals and from forty-five participants with no history of previous infection but with two vaccinations. The anti-spike quantitative responses were calculated to BAU units as suggested by the manufacturers and compared. Two assays reacting only to infection were added to the analysis.
Discussion
In this study, we compared serological dynamics using samples from ninety-nine non-vaccinated PCR-positive participants and from forty-five participants with no history of previous infection but with two vaccinations against SARS-CoV-2. Serum samples were analyzed with a total of six different assays. To follow infection, assay readouts of Ro-N-Ig and GS-cPass were performed from the same sample in a head-to-head comparison. Participants showed positive antibodies against these assays for at least 400 days. The remaining four assays detect responses to both infection and vaccination. In previous studies the EI-S1-IgA showed to be less reliable [
23] and was therefore performed only for samples of the PCR-positive participants. The EI-S1-IgG assay is per manufacturer’s definition non-quantitative and its quantitative version is defined by the EI-S1-IgG-quant assay [
31]. It was possible to measure samples of the PCR-positive participants only with the non-quantitative version. Measurements were hence transformed to quantitative values using paired samples presented in [
31]. A comparison with the vaccinees was possible thereafter, together with direct readouts of the Ro-RBD-Ig-quant for both cohorts. As a results, the longitudinal dynamics of EI-S1-IgG-quant and Ro-RBD-Ig-quant in the PCR-positive cohort present completely different trends, while for vaccinees the trends a very similar.
Duration and magnitude of serological responses in relation to different testing systems and antigen-target has been subject to dissent. Harris et al. [
22] demonstrated a rapid decline of anti-N antibodies using the ELISA from Abbot, with only 51% of SARS-CoV-2 infected individuals having a sero-response after 6 months. In contrast, Favresse et al. [
32] showed a positivity rate of 94% after 10 months but using the Ro-N-Ig test. Muecksch et al. [
33] compared four different assays: the Ro-N-Ig and the Abbott SARS-CoV-2 immunoglobulin (Ig) G assay for anti-N detection, and the DiaSorin SARS-CoV-2 IgG together with the Siemens SARS-CoV-2 RBD assay for anti-S comparisons. Similarly to our analysis, the shapes of the curves strongly differ between assays. Dan et al. [
19] described a half-life of binding anti-S antibodies of 103 days. In contrast Ibarrondo et al. [
18] described a rapid decay of anti-RBD antibodies with a half-life of only 36 days.
In our cohort of non-vaccinated SARS-CoV-2 RT-PCR-positive individuals direct readout values are coherent to previous published literature, comparing the same testing platform [
33]. If compared to other assays, same discrepancies as in the rest of the literature appear. In addition, clinical characteristics of the underlying cohorts differed greatly. Several authors described a correlation between magnitude of antibody responses and degree of clinical symptoms in SARS-CoV-2 infected individuals [
16,
17,
32,
34‐
38]. This was replicated in our cohort, where we could observe a trend towards higher antibodies titers in individuals with more severe symptoms. As we solely enrolled oligo-symptomatic participants, these findings did not reach statistical significance. Of importance, exactly that group of oligo-symptomatic patients is the overwhelming majority of the population which might be subject to serological testing for different reasons.
A correlate of protection of SARS-CoV-2 has not been established yet and it is still debated whether the protection after natural infection is different or perhaps even superior to vaccination [
39]. Natural infection is likely to elicit a broader response against more epitopes of the virus [
40]. However, several studies describe the immune response after vaccination to be characterized by higher antibody levels compared to natural infections, especially following vaccination with mRNA- based vaccines [
5,
7,
9,
33,
41]. Recent reports describe waning protection already shortly after the second dose and the decay seems more pronounced than after a natural infection [
42‐
45]. Similarly, when comparing naturally infected to vaccinated participants, we observed pronouncedly higher antibody levels in the latter compared to the former. Antibody levels remained positive for at least seven months after vaccination.
Initial reports on SARS-CoV-2 infected cohorts declared a high level of protection of 82–89% for approximately 6 months against the wild type virus [
3]. Similarly, data from Israel suggested a high level of protection after vaccinating with Pfizer-BioNTech [
46]. Since the surge of new virus variants protection against Delta and Omicron variants was reduced [
14]. Mizrahi et al. [
14] described a 1.5 times increased risk for breakthrough infections with Delta variant for subjects 6 months after vaccination with Pfizer-BioNTech compared to subjects 3 months after vaccination. Shrotri et al. [
42] compared protection of vaccinated individuals with anti-RBD antibodies above and below 500 BAU (ELYSYS Ro-Ig) and found significantly more infected participants with antibodies below 500 BAU. However, our data suggests that an antibody lever of 500 BAU is usually not reached after natural infection. Our SARS-CoV-2 RT-PCR-positive cohort only included participants infected with the original wild type strain. A comparison between variants is therefore not possible, but would also only generate data unclear to compare, as the vaccines and the antigens used in the tests are all also wild type.
Modelling the temporal evolution of the antibody levels, the serological dynamics of the vaccinated cohort is completely different than that of the PCR-positive vaccine naive infected participants. After vaccination, we observed a short initial peak phase, followed by a very slow decline of antibody levels in both quantitative tests EI-S1-Ig-quant and Ro-RBD-Ig. All measurements were above the positivity threshold even seven months after the second dose. The curves representing the antibody dynamics of both quantitative tests were very similar. This is in contrast to our observations in the SARS-CoV-2 RT-PCR-positive cohort described here. The ELISA-based Euroimmun test suggested a rapid decline of antibodies with more than 50% of the samples dropping below the threshold for positivity within less than one year while the Ro-RBD-Ig assay yielded positive readouts after 200 days almost without declining. An explanation for the slow EI-S1-Ig-quant antibody decrease in vaccinated versus the steady state suggested by Ro-RBD-Ig readout in the RT-PCR-positive cohort could be the rise in avidity as also described by Scheiblauer et al. [
47]. The authors hypothesized that two vaccine doses lead to an antibody response dominated by highly specific and highly avid IgG directed against the S-protein. Thus, the antibody-signal dynamics over time reflect the overall amount of antibodies in both tests. In contrast, natural infection will likely elicit a much broader response which maturates over time [
47], including detection of the RBD-domain which in turn might lead to an increase in avidity, while the overall antibody amount is dropping [
47]. Those two opposing effects may compensate each other at different rates depending on the assay format. The Ro-RBD-Ig assay reportedly detects the binding of few antibodies but favors high avidity [
47], potentially resulting in a persistently high assay readout, while the ELISA-based Euroimmun assay values are biased towards whole antibody binding and thus decline. Persisting non-declining Roche RBD-antibodies detectable for more than 300 days after natural infection have been repetitively described with a level of ~ 100 BAU [
32,
47].
Summarising, we present the results of a well-characterized cohort to investigate dynamics in serological responses to non-vaccinated SARS-CoV-2 infected individuals compared to vaccinated healthy controls. For quantitative anti-Spike assays, we used BAU standardization which is provided by the manufacturer.
However, we observe distinct differences both in the magnitude and dynamics of the measured antibody response, although BAU standardization for anti-S1/RBD tests was used. Interestingly, these differences were negligible for samples taken two months after symptom onset. The standardization however is less accurate before and after this time period, resulting in differences of up to one order of magnitude in supposedly standardized and comparable values. These differences disappeared in the vaccinated cohort. One potential explanation could be the fact that the assays measure different targets. While the EI-S1-IgG detects the overall amount of binding antibodies in an ELISA-format, Ro-RBD-Ig-quant is an ELECSYS based double-antigen sandwich-test, detecting highly avid antibodies [
47]. Importantly, the BAU-standard provided by the WHO is derived from a group of donors relatively shortly after the infection. Subsequently, standardization performed for an assay will likely be accurate for tests with a similar profile of antibodies, regarding both subclass as well as avidity, hence only in individuals few months after natural infection [
25]. Therefore, it is not be the best standard for clinical cohorts, including samples from individuals very early or late after the infection, or after vaccination.
Acknowledgements
This work was conducted within the general umbrella of the KoCo19 representative SARS-CoV-2 cohort in Munich. The whole project was supported by the Bavarian State Ministry of Science and the Arts, the European Commission through the ORCHESTRA grant (The ORCHESTRA project has received funding from the European Union’s Horizon 2020 research and innovation programme under Grant Agreement No 101016167. The views expressed in this publication are the sole responsibility of the author and the Commission is not responsible for any use that may be made of the information it contains), the University Hospital of the LMU (KUM), the Helmholtz Centre Munich and the University of Bonn as well as University of Bielefeld. The German Ministry for Education and Research (Project No. 01KI20271) supported the project as did the Medical Biodefense Research Program of the Bundeswehr Medical Service. For this project Roche Diagnostics provided kits and machines for analyses at discounted rates. The funders did not influence the scientific conduct of the experiments or presentation/interpretation of the results or the decision to publish the data.
KoCo19/ORCHESTRA Study group: Mohamed IbraheemMohamed Ahmed, Emad Alamoudi, Jared Anderson, Valeria Baldassarre, Abhishek Bakuli, Maximilian Baumann, Marc Becker, Franziska Bednarski, Marieke Behlen, Olimbek Bemirayev, Jessica Beyerl, Patrick Bitzer, Rebecca Böhnlein, Isabel Brand, Anna Brauer, Vera Britz, Jan Bruger, Franziska Bünz, Friedrich Caroli, Josephine Coleman, Lorenzo Contento, Alina Czwienzek, Flora Deák, Maximilian N. Diefenbach, Jana Diekmannshemke, Paulina Diepers, Anna Do, Gerhard Dobler, Jürgen Durner, Tabea Eser, Ute Eberle, Judith Eckstein, Philine Falk, Manuela Feyereisen, Volker Fingerle, Stefanie Fischer, Jonathan Frese, Felix Forster, Günter Fröschl, Christiane Fuchs, Otto Geisenberger, Mercè Garí, Marius Gasser, Sonja Gauder, Raffaela Geier, Kristina Gillig, Christof Geldmacher, Keisha Gezgin, Leonard Gilberg, Kristina Gillig, Philipp Girl, Elias Golschan, Vitus Grauvogl, Jessica Michelle Guggenbuehl Noller, Elena Maria Guglielmini, Pablo Gutierrez, Anslem Haderer, Celina Halfmann, Marlene Hannes, Lena Hartinger, Timm Haselwarter, Jan Hasenauer, Alejandra Hernandez, Luca Heller, Arlett Heiber, Matthias Herrmann, Leah Hillari, Stefan Hillmann, Christian Hinske, Janna Hoefflin, Tim Hofberger, Michael Höfinger, Larissa Hofmann, Sacha Horn, Kristina Huber, Christian Janke, Lilian Karger, Ursula Kappl, Antonia Keßler, Zohaib Khan, Charlotte Kiani, Isabel Klugherz, Norah Kreider, Johanna Kresin, Arne Kroidl, Pratik Kunder, Magdalena Lang, Clemens Lang, Silvan Lange, Ekaterina Lapteva, Michael Laxy, Ronan Le Gleut, Reiner Leidl, Leopold Liedl, Felix Lindner, Xhovana Lucaj, Elisabeth Lucke, Fabian Luppa, Alexandra Sophie Nafziger, Alexander Maczka, Petra Mang, Alisa Markgraf, Paula Matcau, Rebecca Mayrhofer, Anna-Maria Mekota, Dafni Metaxa, Emily Mohr, Hannah Müller, Katharina Müller, Nathalia Nascimento, Kasimir Niermeyer, Sophia Nikolaides, Ivan Noreña, Leonie Pattard, Michael Plank, Claire Pleimelding, Michel Pletschette, Viona Poll, Stephan Prückner, Kerstin Puchinger, Konstantin Pusl, Peter Pütz, Katja Radon, Elba Raimúndez, Julius Raschka, Jakob Reich, Christina Reinkemeyer, Friedrich Riess, Camilla Rothe, Viktoria Ruci, Elmar Saathoff, Nicole Schäfer, Yannik Schälte, Paul Schandelmaier, Benedikt Schluse, Annika Schneider, Lara Schneider, Sophie Schultz, Mirjam Schunk, Lars Schwettmann, Josefin Sedlmeier, Linda Sintu-Sempta, Alba Soler, Peter Sothmann, Katharina Strobl, Aida Strüber, Laura Strüber, Jeni Tang, Fabian Theis, Verena Thiel, Eva Thumser, Niklas Thur, Sophie Thiesbrummel, Julian Ullrich, Vincent Vollmayr, Emilia Von Lovenberg, Jonathan Von Lovenberg, Carsten Vos, Julia Waibel, Claudia Wallrauch, Nikolas Weigl, Roman Wölfl, Julia Wolff, Pia Wullinger, Tobias Würfel, Patrick Wustrow, Sabine Zange, Eleftheria Zeggini, Anna Zielke, Thorbjörn Zimmer, Thomas Zimmermann, Anna Zielke, Lea Zuche.