key: cord-0965891-wsetqnpp
authors: Carreno, J. M.; Mendu, D. R.; Simon, V.; Shariff, M. A.; Singh, G.; Menon, V.; Krammer, F.
title: Longitudinal analysis of SARS-CoV-2 seroprevalence using multiple serology platforms
date: 2021-02-26
journal: nan
DOI: 10.1101/2021.02.24.21252340
sha: fda29ff8c23fd1ded3d681fca64064bbbfa02d8b
doc_id: 965891
cord_uid: wsetqnpp

Serological tests are important tools helping to determine previous infection with severe acute respiratory disease coronavirus 2 (SARS-CoV-2) and to monitor immune responses. The current tests are based on spike (S), the receptor binding domain (RBD), or the nucleoprotein (NP) as substrate. Here, we used samples from a high seroprevalence cohort of health care workers (HCWs) to perform a longitudinal analysis of the antibody responses using three distinct serological assays. 501 serum samples were tested using: a) a research-grade RBD and spike based tandem enzyme-linked immunosorbent assay (MS-RBD ELISA, MS-spike ELISA), b) a commercial RBD and spike based tandem ELISA (Kantaro-RBD, -spike), and c) a commercial NP-based chemiluminescent microparticle immunoassay (CMIA, Abbott Architect). Seroprevalence ranged around 28% during the early stage of the pandemic (a: 28.4% positives; b: 28.1%; c: 27.3%). Good correlation was observed between the MS and Kantaro RBD ELISAs and between the MS and Kantaro spike ELISAs. By contrast, modest correlations were observed between the Abbott Architect and both RBD and spike-based assays. A proportion of HCWs (n=178) were sampled again 3-5 months after the first time point. Although antibody levels declined in most of the positive individuals, the overall seroprevalence measured by RBD-spike based assays remained unchanged. However the seroprevalence of NP-reactive antibodies significantly declined. Lastly, we tested six samples of individuals who received two doses of SARS-CoV-2 mRNA vaccine and found that seroconversion was detected by the RBD-spike assays but, of course as expected, not the NP based assay. In summary, our results consolidate the strength of different serological assays to assess the magnitude and duration of antibodies to SARS-CoV-2.

In the advent of the current pandemic caused by the severe acute respiratory syndrome coronavirus 44 2 (SARS-CoV-2), pandemic, methods to detect the prevalence of recent and past infections are key 45 to determine public health and social countermeasures. Nucleic acid amplification tests (NAAT) 46 provide an accurate estimation of acute infections (1, 2), but they fail to inform about past infections.

Serological tests that detect antibodies directed against structural targets of the virus, not only are 48 useful to estimate the overall viral seroprevalence and rates of infection in the population (3-5), but 49 also help to assess responses to vaccination (6), to determine correlates of protection (7, 8) , and to 50 test and standardize therapeutic approaches such as monoclonal antibody and plasma transfer 51 therapies (9). Moreover, estimation of viral seroprevalence and quantification of antibody levels 52 adds to our understanding of the immune response and protection at the individual and population 53 levels (10). 54 55 Currently, serological assays to detect antibodies against SARS-CoV-2 are based on recombinant 56 versions of the spike (S), the receptor binding domain (RBD) of S, or the nucleoprotein (NP) as 57 substrate (11). A variety of research grade and commercial S-based and NP-based assays are now 58 available, but antibodies to these two targets have different characteristics. Antibodies directed 59 against the viral spike are retained for several months after infection (12) (13) (14) (15) (16) (17) and correlate with 60 virus neutralization and protection against re-infection (6, 13, (18) (19) (20) (21) . Moreover, vaccination relies 61 uniquely on the viral spike, evidencing the importance of detecting antibodies against this target 62 with high levels of sensitivity and specificity (6). 63 64 Several studies evaluate the sensitivity and specificity of individual assays, either S-or NP-based, 65 however longitudinal side-by-side comparisons of different serological platforms are scarce. Here, 66 we employed samples from a high-risk cohort of health care workers (HCWs) using three different 67 serological assays. In addition, SARS-CoV-2 post-vaccination samples were included in the analysis. 68

We compared a research grade RBD and spike based tandem enzyme-linked immunosorbent assay 69 (ELISA) developed at Mount Sinai (MS ELISA, research grad version), the Seroklir commercial RBD-70 spike based ELISA from Kantaro Biosciences, and the commercial NP-based chemiluminescent 71 microparticle immunoassay (CMIA) Abbott Architect. 72 73 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 26, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 Methods 74

Detection of receptor binding domain (RBD) and full-length spike (S) antibodies in plasma was 76 performed with a research-grade two-step ELISA developed at Mount Sinai closely resembling an 77 assay used in Mount Sinai's CLIA-certified Clinical Pathology Laboratory, which received FDA 78 Emergency Use Authorization in April 2020 (22, 23 incubation, plates were washed three times with PBS-T and 50μl/well of anti-human IgG (Fab-88 specific) horseradish peroxidase antibody (Sigma, A0293) diluted 1:3,000 in PBS-T 1% milk powder 89 were added. Plates were incubated for 1 h at room temperature, followed by three times washing 90 with PBS-T and addition of developing solution (100μl/well) sigmafast o-phenylenediamine 91 dihydrochloride (Sigma). The reaction was led to proceed for 10 min, and stopped using 50μl/well 92 of 3-molar hydrochloric acid (Thermo Fisher). Optical density was measured at 490 nm using an 93 automated plate reader (BioTek). Samples with an OD490nm above 0.15 (cut-off value) were 94 considered as presumptive positives and were further tested in the confirmatory ELISA using the 95 full-length recombinant spike protein. 96 97 Briefly, to perform the confirmatory ELISAs, plates were coated and blocked as described above, 98 but using full-length spike protein for coating. Presumptive positive plasma samples were serially 99 diluted (1:3) in 1%-milk prepared in PBS-T, starting at an initial dilution of 1:80. Serial dilutions 100 (100μl/well) were added to the plates, followed by 2-hour incubation at room temperature. The 101 remaining steps were performed as described above. Data was analyzed using GraphPad Prism 7. 102

Samples with an OD490nm above 0.15 (cut-off value) at a 1:80 plasma dilution were considered 103 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted February 26, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 positive. Samples with an OD490nm above 0.15 at the last dilution were further diluted (1:2160  104 initially) and re-tested. Only samples positive in both steps of the assay were considered positive. 105 106

ELISAs to detect antibodies in plasma against the receptor binding domain (RBD) and the full-length 108 spike (S) based on the commercial Kantaro Quantitative SARS-CoV-2 IgG Antibody Kit SeroKlir, Kantaro Biosciences) were used. The assay was performed according to manufacturer's 110 instructions except for additional serum dilution steps in highly reactive individuals. All reagents 111 and microplates were included with the commercial kit. Briefly, for qualitative RBD ELISAs, samples 112 were diluted in sample buffer (1:100) using 96-well microtitre plates, and 100μl/well of pre-diluted 113 samples were transferred to the RBD pre-coated microplates. Positive and negative controls were 114 added to every plate. Samples were incubated for 2 hours at room temperature, followed by 115 removal of plasma dilutions and washing three times with wash buffer. RBD conjugate was diluted 116 in conjugate buffer and 100μl/well were added to the plates. After 1h incubation, conjugate was 117 removed and plates were washed three times with wash buffer. Substrate solution was added 118 buffer. Dilutions were added in duplicate to the pre-coated microplates. Low, medium and high 126 controls, as well as spike calibrators used to generate a standard curve, were added to every 127 microplates. After 2h incubation at room temperature, the remaining steps were performed as 128 described above. Data was analyzed using GraphPad Prism 7. Concentration of spike-reactive 129 antibodies was calculated using a four parameter logistic (4-PL) curve fit. Samples exceeding the 130 range of the standard curve were further diluted (1:5400) and re-tested. Only samples positive in 131 both steps of the assay were considered positive. 132 133 Abbott Architect CMIA 134 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 26, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 The Architect test (Abbott Laboratories) consists of an automated, two-step, qualitative CMIA for 135 qualitatively detecting IgG against the nucleoprotein (N) antigen from SARS-CoV-2. This test has a 136 reported sensitivity of 100% (CI 95.8-100%) and specificity of 99.6 (CI 99-99.9%) 14 days after 137 symptom onset. The assay was performed according to manufacturer's instructions. All reagents 138 were included with the kit. Briefly, sample, SARS-CoV-2 antigen coated paramagnetic 139 microparticles, and assay diluent were combined and incubated. The IgG antibodies to SARS-CoV-2 140 present in the sample bind to the SARS-CoV-2 antigen coated microparticles. The mixture is washed. 141

Anti-human IgG acridinium-labeled conjugate is added to create a reaction mixture and incubated. 142 Correlations of antibody levels in the different assays were calculated using a standard Pearson's 163 correlation. Coefficients of correlation (r) are presented. Paired t-test for comparison of phase 1 and 164 phase 2 antibody levels was used. All adjusted P values of <0.05 were considered statistically 165 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 26, 2021. ; https://doi.org/10.1101/2021.02.24.21252340 doi: medRxiv preprint significant with a confidence interval of 95%. Statistical analyses were performed using Prism 7 166 (GraphPad, USA). 167 168

Longitudinal comparison of SARS-CoV-2 seroprevalence using RBD/spike and NP based 170 assays 171

Seroprevalence of SARS-CoV-2 across different regions of the world has been described using 172 multiple serological assays based either on the spike protein (S), its receptor-binding domain (RBD), 173 or the nucleoprotein (NP). Here we compared side-by-side a research grade MS ELISA based on RBD 174 and spike, an RBD/spike based SeroKlir assay from Kantaro Biosciences and the NP based Abbott 175 Architect test. We used a set of 501 samples from frontline healthcare workers (HCW) collected 176 after the first pandemic wave in the New York City area (May 2020). Seroprevalence in this set of 177 samples using a research grade ELISA from Mount Sinai was 28.4%, (142/501), 28.1%, using the 178 SeroKlir test from Kantaro Biosciences (141/501) and 27.3% using the Abbott Architect test 179 (137/501) (Fig. 1A) . A subset of the initial participants (n=178) attended provided a second serum 180 sample at a follow up visit in August-October 2020 allowing assessment of seroprevalence at two 181 different time points. Of note, the seroprevalence in the smaller subset of participants was higher 182 compared to the initial cohort (N=501). This is likely due to higher compliance of participants that 183 knew their sero-status in the first phase, since they were informed about their antibody levels. 184

Overall, the seroprevalence measured by the Mount Sinai and the Kantaro ELISAs did not vary 185 significantly between the two time points (Fig. 1B-1C) , but the seroprevalence of NP reactive 186 antibodies measured by the Abbott Architect test declined (Fig. 1D) . 187

We further compared the antibody levels in samples obtained during the first phase (May 2020) 189 and the second phase (Aug-Oct, 2020) in the subset of 178 subjects (the distribution of antibody 190 levels is shown in Sup. Fig. 1 and concordance analyses among the different assays are shown in 191 Sup. Fig. 2) . As expected, antibody levels in the second phase declined in the majority of participants 192 in a manner that was consistent in the three different assays ( Figs. 2A-2C) . A sharper decline of NP 193 reactive antibody levels as measured by the Abbott Architect test (Fig. 2C) . Moreover, the 194 percentage of subjects that were seropositive initially and whose antibodies became undetectable 195 in the second phase did not vary significantly in the Mount Sinai and Kantaro ELISAs ( Figs. 2A-2B ) 196 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 26, 2021. ; https://doi.org/10.1101/2021.02.24.21252340 doi: medRxiv preprint but 30% of the samples that were positive initially in the Abbott Architect test became negative in 197 the second phase (Fig. 2B) . is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 26, 2021. ; https://doi.org/10. 1101 /2021 To date only limited data is available about how commercial antibody assays respond to antibodies 227 developed in response to vaccination. In order to determine how the three assays perform against 228 vaccine-induced antibodies, we measured reactivity in serum of individuals who had received two 229 doses of SARS-CoV-2 mRNA vaccines. The expectation was, that the RBD/spike based assays would 230 detect a signal while the NP based assay would not. Indeed, we measured high titers using the Spike 231 based assay platforms (the Mount Sinai and Kantaro assays) but the samples produced no signal in 232 the NP-based assay (Fig. 5A-5C ). Of note, the spike titers measured in the Mount Sinai and Kantaro 233 assay correlated very well (Fig. 5D) . While antibody responses to acute SARS-CoV-2 infection are relatively well understood, less data is 237 available regarding antibody kinetics over longer time frames against different viral antigens. We 238 determined seroprevalence and antibody titers in SARS-CoV-2 infected individuals at two time 239 points (1-2 months and 3-4 months post infection) using three different assays. One assay, the 240 Mount Sinai ELISA, is a laboratory-developed assay that uses an initial ELISA at a single serum 241 dilution against the RBD followed by a confirmation and titration against the full-length spike 242

protein. The second assay tested, the Kantaro SeroKlir assay, is based on the same principle, but 243 commercially available. The third assay, the Abbott Architect, targets the NP and is a CMIA. 244

There was high concordance among the three assays with respect to seroprevalence during phase 245 1. However, the titers only correlated well for the two spike-based assays. During phase 2, the two 246 spike-based assays identified all (Mount Sinai Research grade) or the vast majority (Kantaro) of 247 previously seropositive individuals as seropositive, while the NP based assay (Abbott) failed to 248 detect a signal above the cut-off in approximately 30% of previously positive individuals. These 249 findings mirror similar results recently published by Grandjean and colleagues, suggesting that the 250 NP antibody response is short-lived (24). However, it could also be a reflection of a high cut-off 251 required to ensure high specificity for SARS-CoV-2 in the NP-based assay. Importantly, and as 252 expected since no NP is included in the FDA EUA approved vaccines used in the US, only the spike-253 based assays were able to detect antibodies induced by SARS-CoV-2 mRNA vaccines (6). Our data 254 highlight the need to understand assay performance before a specific assay is used to study specific 255 aspects of SARS-CoV-2 immunity. All three assays are very valuable to assess seroconversion 256 shortly after infection, but only the two spike-based assays were reliable months after recovery. 257 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 26, 2021. 

The Icahn School of Medicine at Mount Sinai has filed patent applications relating to SARS-CoV-2 284 serological assays (the "Serology Assays") and NDV-based SARS-CoV-2 vaccines which list 285 Florian Krammer (the "Serology Assays", vaccines) and Viviana Simon ("Serology Assays") as co-286 inventors. The foundational "Serology Assay" intellectual property (IP) was licensed by the Icahn 287

School of Medicine at Mount Sinai to commercial entities including Kantaro Biosciences, a 288 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 26, 2021.

serologic tests based on the Mount Sinai IP. It is anticipated that the medical school will receive 290 payments related to commercialization of the "Serology Assay" IP and, as faculty inventors, Drs. 291

Krammer and Simon will be entitled to a portion of these payments. 292 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 26, 2021. October 2020. Antibodies against RBD-Spike were measured using a research grade ELISA from 413 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 26, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 against NP using the Abbott-Architect CMIA (C). Antibody levels in specimens obtained early during 415 the pandemic on May 2020 or in a follow up visit on August-October 2020 are shown. Samples with 416 a value above or below the cutoff of the corresponding assay (doted line) are shown. **** P<0.0001. 417

The percentage of seropositive samples that turned negative (red) CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 26, 2021. ; https://doi.org/10.1101/2021.02.24.21252340 doi: medRxiv preprint 

Immunological memory to SARS-CoV-2 assessed for up 332 to eight months after infection. bioRxiv

Robust neutralizing antibodies to SARS-CoV-2 infection 336 persist for months

Persistence of serum and saliva antibody 342 responses to SARS-CoV-2 spike antigens in COVID-19 patients

Dynamics and significance of the antibody response to SARS-348

CoV-2 infection

Neutralizing antibody responses to SARS-CoV-2 in a COVID-19 recovered 351 patient cohort and their implications

Longitudinal observation and decline of 358 neutralizing antibody responses in the three months following SARS-CoV-2 infection in 359 humans

Primary exposure to SARS-CoV-2 protects against reinfection in rhesus macaques

SARS-CoV-2 infection protects against rechallenge in rhesus macaques

Do antibody positive

Distribution of 455 antibody levels (spike-reactive) early during the pandemic on

Supplementary Figure 2. Concordance analysis of positive samples among the different 459 serology assays. Venn diagrams depicting the number of seropositive samples measured by a RBD-460

Spike research grade ELISA from Mount Sinai (blue); a RBD-Spike commercial ELISA from Kantaro 461

Concordance of positivity among the 462 three different assays using samples obtained early during the pandemic on

Seroprevalence in a subset of subjects (n=178) who attended a follow up visit on