key: cord-0255058-b5nnamx1
authors: Favresse, J.; Bayart, J.-L.; David, C.; Gillot, C.; Wieërs, G.; Roussel, G.; Sondag, G.; Elsen, M.; Eucher, C.; Dogne, J.-M.; Douxfils, J.
title: Serum SARS-CoV-2 Antigens for the Determination of COVID-19 Severity
date: 2021-11-21
journal: nan
DOI: 10.1101/2021.11.18.21266478
sha: c32e6958ba1aea788dbf9995b0a30e2f3d7d0437
doc_id: 255058
cord_uid: b5nnamx1

The diagnostic of SARS-CoV-2 infection relies on reverse transcriptase polymerase chain reactions (RT-PCR) performed on nasopharyngeal (NP) swabs. Nevertheless, false negative results can be obtained with inadequate sampling procedures making the use of other matrices of interest. This study aims at evaluating the kinetic of serum N antigen in severe and non-severe patients and compare the clinical performance of serum antigenic assays with NP RT-PCR. Ninety patients were included and monitored for several days. Disease severity was determined according to the WHO clinical progression scale. The serum N antigen was measured with a chemiluminescent assay (CLIA) and the Single Molecular Array (Simoa). Thresholds for severity were determined. In severe patients, the peak antigen response was observed 7 days after the onset of symptoms followed by a decline. No peak response was observed in non-severe patients. Severity threshold for the Simoa and the CLIA provided positive likelihood ratio of 30.0 and 10.9 for the timeframe between day 2 and day 14, respectively. Compared to NP RT-PCR, antigenic assays were able to discriminate the severity of the disease (p = 0.0174, 0.0310 and p = 0.1551 with the Simoa, the CLIA and the NP RT-PCR, respectively). Sensitive N antigen detection in serum thus provides a valuable new marker for COVID-19 diagnosis and evaluation of disease severity. When assessed during the first 2 weeks since the onset of symptoms, it may help in identifying patients at risk of developing severe COVID-19 to optimize better intensive care utilization.

The diagnostic of SARS-CoV-2 infection still relies on molecular assays with reverse 53 transcriptase polymerase chain reaction (RT-PCR) performed on nasopharyngeal (NP) swabs 54 being considered as the gold standard detection method.(1) Other techniques like 55 chemiluminescent immunoassays (CLIA) or Single Molecular Array (Simoa) have 56 demonstrated good correlation with RT-PCR on NP swabs at least for cycle thresholds (Ct) 57 below 33.(2, 3) Other matrices like saliva, plasma or dried blood spots have been used in order 58

to detect SARS-CoV-2 infection and also revealed interesting performance versus RT-PCR.(4-59 6) Nevertheless, the clinical performance depends on the detection methods. Namely, lateral 60 flow assay (LFA) permits to provide results within a couple of minutes, but their performance 61 is relatively weak with some rapid antigen detection (RAD) assays showing clinical sensitivity 62 below 30%.(7) However, while almost all RT-PCR techniques show excellent sensitivity with 63 various methods demonstrating limit of detection (LOD) from 10 2 to 10 5 copies/mL according 64 to manufacturer package inserts and reference panels,(8) the correlation between RT-PCR 65 results from NP swabs and the disease severity has been questioned.(9, 10) The Ct values for a 66 specimen vary between different kits and techniques (including target genes, primers and 67 threshold fluorescence values) and Ct values may vary between different runs of the same 68 kit.(10) The Ct value also depends on the method of collection of the sample and hence there 69 may be a variation in Ct values between two different samples obtained from the same person 70 on the same day and run on the same kit.(11) In addition, concerns have been raised about the 71 risk of false-negative results associated with the use of nasal and NP swabs, especially before 72 symptom onset.(12) Thus, there is a place for improvement in the diagnostics of SARS-CoV-2 73 infection, the "ideal" test being able to detect the presence of the virus with a similar or better 74 sensitivity than the RT-PCR performed on NP swabs, but with a better prognostic value 75 . CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 in serum-gel tubes (BD SST II Advance®, Becton Dickinson, NJ, USA) and centrifuged for 10 126 min at 1740×g on a Sigma 3-16KL centrifuge. Sera were stored in the laboratory serum biobank 127 at −20°C from the collection date. Frozen samples were thawed for 1 hour at room temperature 128 on the day of the antigen analysis. Re-thawed samples were vortexed before the analysis. 129

Single Molecular Array 131

The SARS-CoV-2 nucleocapsid (N) antigen was quantified automatically in patient sera by a 132

Single Molecular Array (Simoa) immunoassay using the Simoa HD-X analyzer (Quanterix, 133 Massachusetts, USA). Samples were analyzed using the commercial SARS-CoV-2 N-Protein 134

Advantage kit (item 103806), a paramagnetic microbead-based sandwich ELISA. Briefly, 135 diluted sample and anti-N protein antibody coated capture beads and detector antibodies are 136 combined for 35 minutes in the cuvette in a first step. Beads are then washed, and a conjugate 137 of streptavidin-ß-galactosidase is added to label the captured N protein. After washing, beads 138 are resuspended with resorufin-ß-galactopyranozide for signal generation. Finally, beads are 139 loaded in microarrays of femtoliter reaction wells. The fraction of bead-containing microwells 140 counted with an enzyme is converted into "average enzymes per bead" (AEB). AEB values are 141 converted into N protein concentration by interpolation from the calibration curve obtained by 142 4-parameter logistical regression fitting. The results are quantitative and expressed as pg/mL. 143

This assay uses 8 calibrators ranging from 0 pg/mL to 200 pg/mL. Serum samples that provided 144 results upper the calibration range were retested with a 1,000-fold dilution. The LOD of the 145 assay is 0.099 pg/mL. The within-run coefficient of variation (CV) is less than 10%. Positivity 146 cut-offs in serum samples were not disclosed by the manufacturer. 147

The SARS-CoV-2 N antigen was detected automatically in patient sera by CLIA using the 149 iFlash 1800 automated magnetic CLIA (MCLIA) analyzer (Shenzhen YHLO Biotech Co. Ltd., 150 . CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 Shenzhen, China). Samples were analyzed using the commercial iFlash-2019-nCOV Antigen 151 assay kit. Antigens in the sample will react with anti-2019-nCoV antibodies coated on 152 paramagnetic particles and with anti-2019-nCoV acridinium-ester-labeled conjugate to form a 153 sandwich complex. Under magnetic field, particles are absorbed to the wall of the reaction 154 chamber, and unbound materials are washed away. Afterwards, the Pre-Trigger and Trigger are 155 added to the reaction mixture. The chemiluminescent signal is then measured as relative light 156 units (RLUs). Results are determined via a 2-point calibration curve. The results are only 157 qualitative and are expressed as a cut-off index (COI). The within-run CV ranges from 2.7 to 158 3.6%. (19) The manufacturer's positivity cut-off is > 1.0 COI. The indented use only includes 159 the detection of antigens in NP swabs. In our study, we explored the possibility of detecting the 160 antigen in serum (RUO setting). 161

SARS-CoV-2 Spike IgG antibodies were quantified automatically in patient sera by a Simoa 163 immunoassay using the Simoa HD-X analyzer (Quanterix, Massachusetts, USA) . Samples were 164 analyzed using the commercial SARS-CoV-2 Spike IgG Advantage kit (item 103769), using a 165 similar procedure as for the SARS-CoV-2 N antigen. The positivity cut-off of 924 ng/mL 166 corresponding to the maximal value obtained in pre-pandemic serum samples was used.(20) 167 SARS-CoV-2 antibodies were available for 77 patients out of the 81 (95.1%) due to insufficient 168 residual serum sample. 169

Reverse transcriptase polymerase chain reaction for SARS-CoV-2 determination in NP swab 171 samples was performed on a LightCycler 480 Instrument II (Roche Diagnostics, Rotkreuz, 172

Switzerland) using the LightMix Modular SARS-CoV E-gene set (for samples originating from 173

Clinique Saint-Luc Bouge) and on the GeneXpert instrument (Cepheid, California, USA) using 174 the Xpress SARS-CoV-2 assay targeting N2 and E genes (for samples originating from Clinique 175 . CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 Saint-Luc Bouge and Clinique Saint-Pierre Ottignies). Cycle threshold (Ct) values obtained by 176

RT-PCR were used as a proxy for the viral load. 177

Descriptive statistics were used to analyze the data. Smoothing splines with four knots were 179 used to estimate the time kinetics curves using all longitudinal samples from the study 180 population. Difference between disease severity per time intervals (i.e. < 3 days, 4 to 10 days, 181 11 to 20 days and > 20 days) and difference between time intervals per severity was assessed 182

using an ordinary two-way ANOVA with Šidák's multiple comparison tests with individual 183 variances computed for each comparison. In the subpopulation A, antigen results have been 184 used to define cut-offs for severity. Receiver operating characteristics (ROC) curve for antigen 185 assays were performed and the corresponding area under the curves (AUC) was calculated. The 186

Youden index was used to determine the optimal severity cut-off. Additionally, a positive 187 likelihood ratio (sensitivity/1-specificity) was calculated based on the severity cut-off for 188 prediction of severe forms of In the subpopulation B, the comparison of RT-PCR and antigen results at the time of diagnosis 190 (i.e. at the time of the RT-PCR), according to disease severity was tested using an ordinary one-191 way ANOVA. A Tukey multiple testing comparison was applied with multiplicity adjusted p 192 value. Antigen and SARS-CoV-2 Spike IgG values were log-transformed. ROC curves were 193 also performed and the Youden index was used to determine the optimal positivity cut-offs is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10. 1101 /2021 Antigen level comparison between patients with or without SARS-CoV-2 Spike IgG response 201 has been performed using a Welch's t-test. Pearson's correlation coefficients were used to 202 investigate the correlation between RT-PCR, SARS-CoV-2 Spike IgG and N antigen results 203 and between the two antigen assays. 204

Data analysis was performed using GraphPad Prism software (version 9.2.0, California, USA) 205

and MedCalc software (version 14.8.1, Ostend, Belgium). P < 0.05 was used as a significance 206 level. Our study fulfilled the Ethical principles of the Declaration of Helsinki. 207

Results 208

The kinetics of the N antigen in the studied population is presented in Figure 2 for both the 210

Simoa and the iFlash assay. The peak antigen response was observed at day 7 in severe patients 211 using both assays. Afterwards, a decline was observed up to day 20. In non-severe patients, the 212 antigen response corresponded to a plateau phase that slowly decreased over time. The 213 difference of kinetics between severe and non-severe patients was more distinguishable using 214 the Simoa assay (Figure 2) . Using ROC curves analyses on samples collected in patients from 215 day 2 to day 14 (subpopulation B), the optimal cut-off to identify severe patients have been 216 found to be 5,043 pg/mL for the Simoa ( Figure 2) . The application of these severity cut-offs on kinetic models 221 permitted to identify the best timing since symptom onset to identify severe patients (i.e. from 222 4 to 10 days) (Figure 2 -red dotted lines) . 223 . CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10. 1101 /2021 By splitting timing since symptom onset into categories (i.e. < 3 days, 4 to 10 days, 11 to 20 224 days and > 20 days), the antigen level in severe patients observed between days 4 and 10 was 225 significantly higher compared to non-severe patients, whatever the time category. These 226 observations were similar for both antigen assays (Figure 3 ). Using the Simoa assay, the 227 sensitivity of antigen in serum up to 10 days since symptom onset was 100% in severe patients 228 and ranged from 91.3% (< 3 days) to 100% (from 4 to 10 days) in non-severe patients (global 229 sensitivity = 96.9%). Between days 11 and 20, the sensitivities decreased to 88.5% and 86.5% 230 and after 20 days waned at 75.0% and 35.3% in severe and non-severe patients (Figure 3) . 231

Using the iFlash assay with the optimized positivity cut-off of 0.31 COI (see section below), 232 the sensitivity of antigen in serum up to 10 days since symptom onset was 100% in severe 233 patients and ranged from 93.5% (< 3 days) to 96.0% (from 4 to 10 days) in non-severe patients 234 (global sensitivity = 96.2%). Between days 11 and 20, the sensitivities decreased at 80.8% and 235

86.5% and after 20 days dampened at 75.0% and 23.5% in severe and non-severe patients. 236

These two antigen assays showed a highly significant correlation with a Pearson r of 0.96 (p < 237 0.0001) (Supplemental Figure 3) . Among the 243 samples, only 12 (4.9%) were discordant. 238

Four results were positive on the iFlash assay (range: 0.33-0.54 COI) but negative on the Simoa 239 and 8 were negative on the iFlash assay but positive on the Simoa (range: 0.32-30.9 pg/mL). 240

The mean Ct values obtained by RT-PCR were significantly lower in case of serum antigen 243 positivity using the two antigen assays (p < 0.0001) (Figure 4 -panel A) . The clinical 244 sensitivity was 100% and the clinical specificity was 92.3% for both antigen assays ( Table 1) . 245

Highly significant correlations were found between RT-PCR results and antigen levels (Pearson is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10. 1101 /2021 Based on the Youden index recommended values, the following positivity cut-offs were found 248 for the iFlash and the Simoa assays: > 0.310 COI and > 0.099 pg/mL. The application of these 249 cut-offs on the pre-pandemic cohort resulted in specificities of 95.8% (95%CI: 88.1-99.1; 3 250 false positive results) and 98.6% (95%CI: 92.4-99.9; 1 false positive result), respectively (Table  251 1). False positive pre-pandemic samples were different between the iFlash and the Simoa 252

Antigen assays were also confronted to RT-PCR for determination of disease severity at the 254 time of diagnosis (i.e. at the time of the RT-PCR). Considering NP RT-PCR, the mean Ct value 255 of asymptomatic patients was significantly higher compared to severe patients (p = 0.0037). No 256 significant difference was observed between mild, moderate, and severe patients based on NP 257

RT-PCR ( Figure 5 ). Serum antigen assays showed higher antigen levels in severe patients 258 ( Figure 5) . The difference was statistically significant between severe patients and 259 asymptomatic or moderate patients. Moderate patients also had significantly higher antigen 260 levels compared to asymptomatic subjects. 261

For the 77 samples for which the serological status has been obtained at the time of diagnosis 264 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10. 1101 /2021 correlations between SARS-CoV-2 Spike IgG and antigen results were also identified (Pearson 273 r of -0.60 [iFlash] to -0.66 [Simoa]; p < 0.0001) (Supplemental Figure 5 -panel B) . 274

Interestingly, patients with a Ct value > 33 along with negative antigen in serum all had positive 275 SARS-CoV-2 Spike IgG (n = 12) (Figure 4) . Contrariwise, most patients (96.6%) with negative 276 SARS-CoV-2 Spike IgG were positive for antigen in serum using both assays (Supplemental 277 (Figure 2 and 3) . Difference between 295 severe and non-severe patients was especially noticeable between days 4 and 10. We also 296 estimated that cut-offs for identifying patients more at risk to be severe were 5,043 pg/mL and 297 . CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10.1101/2021.11.18.21266478 doi: medRxiv preprint 313.8 COI on the Simoa and the iFlash assays. The likelihood ratio for these cut-offs were 30.0 298 and 10.9 which reflects their good capacity to distinguish severe from non-severe patients 299 during the day 2 to day 14 timeframe. The likelihood ratio was lower using the iFlash assay and 300 the larger range of antigen concentration of the Simoa technology (from 0.099 to 100,000 301 pg/mL) might explain why the Simoa may be a better predictor of severe outcomes compared 302

to the iFlash, even if a high correlation was found between the two techniques (Supplemental 303 were the first to report on the Simoa technology and found a specificity of 82.4%. (6) The 308 method they used was, however, home-made and do not represent the current performance of 309 the Simoa assay. As a matter of fact, Shan et al. found a specificity of 100% using the 310 commercial kit from Quanterix on the Simoa.(5) In our evaluation, specificities of the iFlash 311 and Simoa were 95.8% and 98.6% and were therefore consistent with the data from the 312

The clinical performance of these antigen assays was also directly compared to RT-PCR 314 performed on NP swabs. (5, 15, 25, 26 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10.1101/2021.11.18.21266478 doi: medRxiv preprint clinical sensitivities ranging from 41.0% to 74.0%. (6, 22, 24, 25, 27, 29) This lower 323 performance is explained by the design of these studies. Indeed, the timing since symptoms was 324 either not disclosed (22, 24, 25, 29) or was long (i.e. up to 30 days since symptoms), decreasing 325 the probability of having positive samples knowing the kinetic of the SARS-CoV-2 N antigen 326 in blood.(6, 27) Given that the peak of N antigen is reached after 7 days, as for the viral load in 327 NP samples,(30) and that a continuous decline is observed afterwards, the timing since 328 symptoms is a paramount information for the evaluation of antigenemia (Figure 2) . In our 329 evaluation, we confirmed that the clinical sensitivity of N antigen as compared to RT-PCR was 330 excellent (96.2% for iFlash and 96.9% for Simoa) in samples collected earlier since symptom 331 onset (i.e. < 10 days), especially in more severe patients (i.e. 100% for both assays) (Figure 3) is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10.1101/2021.11.18.21266478 doi: medRxiv preprint (Supplemental Figures 4 and 5) . As expected from previous studies in COVID-19 subjects, 348 the level of antibodies started to increase approximately 10 days after the onset of symptoms. 349

It also correlates with the waning of N antigen levels as observed in Supplemental Figure 6  350 and in another study.(23) These data support the concept that antigenemia may be a prognostic 351 marker of severity to identify patients more likely to require intensive care in the first few days 352 after the onset of symptoms. 353

The clinical specificity of N antigen assays ranged from 68.0% to 99.8% (mean clinical 354 specificity = 90.9%), (6, 22, 23, 26, 28, 29) and were in line with the results we reported in this 355 study (92.3%) ( Table 1) patients. (22) The detection of blood antigen can therefore be a valuable alternative to RT-PCR, 364 especially in case of negative results. In our study, we identified a patient with positive N 365 antigen in serum (both on iFlash and Simoa, low positive sample: 2.14 COI and 10.5 pg/mL) 366 but with a Ct value > 33 (i.e. Ct of 34) and positive for SARS-CoV-2 Spike IgG antibodies 367 (Figure 4) . 368

In a previous study, the application of ROC curve adapted cut-offs (1.85 to 10 pg/mL) allowed 369 to significantly increase the clinical sensitivity (from 76 to 92%) with a limited impact on the 370 specificity (from 100% to 96.84%).(29) In our study, we increased the clinical performance of 371 . CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10.1101/2021.11.18.21266478 doi: medRxiv preprint the iFlash (1.0 to 0.31 COI). This approach has also been successfully used for SARS-CoV-2 372 antibody assays.(33) 373

Sensitive N antigen detection in serum provides a valuable new marker for COVID-19 375 diagnosis, only requiring a blood draw, that is scalable in all clinical laboratories. It allows 376 potential new developments to design rapid antigen blood tests or combined ELISA assays, 377 detecting both antigens and antibodies. Measuring antigen in blood present several advantages 378 including a more standardized process of obtaining blood samples compared to RT-PCR with 379 possible development to dry blood spots sampling strategies. Importantly, antigenemia, when 380 assessed during the first 2 weeks since the onset of symptoms, may help in identifying patients 381 at risk of developing severe COVID-19. The severity cut-offs proposed in our investigation 382 need to be confirmed at subsequent studies, but it already supports the rational that compared 383 to the NP RT-PCR, the detection of N antigen in blood allows a better discrimination of severe 384 cases from non-severe ones. In addition, these techniques are at least as accurate than NP RT-385 PCR for diagnosing SARS-CoV-2 infection, giving us the possibility to develop more 386 convenient testing strategies for patients. It may finally facilitate patient triage to optimize better 387 intensive care utilization. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 Conflict of interest disclosures: YHLO provided the kits for the iFlash antigen assays and 390 JDO received honorarium from YHLO outside the submitted work. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. 

The grey dotted lines correspond to the positivity cut-off of each antigen assay, as found by ROC curves 548 analyses. The black dotted line corresponds to the positivity cut-off of the iFlash assay, as declared the 549 manufacturer. The red dotted lines correspond to the severity cut-off of each antigen assay, as found by 550 ROC curve analyses for the day 2 -day 14 window. Only patients with symptoms and negative for 551 SARS-CoV-2 Spike IgG directed against the spike protein were included in this kinetics representation. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10.1101/2021.11.18.21266478 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 Supplemental Figure 1 : Timing between the collection of NP and blood samplings, in minutes. For 586 41 samples, the NP was collected before the blood sample. For the other 40 samples, the blood sample 587 was collected before the NP swab. The median delta collection time was 1 minute (10 th -90 th percentile: is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10.1101/2021.11.18.21266478 doi: medRxiv preprint Supplemental Figure 6 : Kinetics of antigenemia and SARS-CoV-2 Spike IgG antibodies since the onset of symptoms in non-severe and severe patients determined according to the WHO clinical progression scale.(18) The continuous orange lines correspond to the severity cut-off of each antigen assay, as found by ROC curves analyses. The purple dotted lines correspond to the positivity cut-off of the SARS-CoV-2 Spike IgG assay. The continuous red and blue lines correspond to antigen kinetics in severe and non-severe patients. The dotted red and blue lines correspond to the kinetics of SARS-CoV-2 Spike IgG in severe and non-severe patients. Only patients with symptoms and negative for IgG directed against the Spike protein were included in this representation.

. CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint On what basis potentially eligible participants were identified (such as symptoms, results from previous tests, inclusion in registry) 5 8

Where and when potentially eligible participants were identified (setting, location and dates) 5 9

Whether participants formed a consecutive, random or convenience series 5 Test methods 10a Index test, in sufficient detail to allow replication 6-8 10b Reference standard, in sufficient detail to allow replication 6-8 11 Rationale for choosing the reference standard (if alternatives exist) 3 12a Definition of and rationale for test positivity cut-offs or result categories of the index test, distinguishing pre-specified from exploratory 8-9

12b Definition of and rationale for test positivity cut-offs or result categories of the reference standard, distinguishing pre-specified from exploratory 8-9

13a Whether clinical information and reference standard results were available to the performers/readers of the index test 5 13b Whether clinical information and index test results were available to the assessors of the reference standard 5 Analysis 14 Methods for estimating or comparing measures of diagnostic accuracy 8-9 15 How indeterminate index test or reference standard results were handled 8-9 16 How missing data on the index test and reference standard were handled 8-9 17 Any analyses of variability in diagnostic accuracy, distinguishing pre-specified from exploratory 8-9

18 Intended sample size and how it was determined / RESULTS Participants 19 Flow of participants, using a diagram 5 and Figure 1 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 21, 2021. ; https://doi.org/10. 1101 /2021 

Molecular, serological, and biochemical diagnosis and 409 monitoring of COVID-19: IFCC taskforce evaluation of the latest evidence

Simoa® SARS-CoV-2 N Protein Antigen Test -Instruction for use 412

Head Comparison of Rapid and Automated Antigen Detection Tests for the Diagnosis 416 of SARS-CoV-2 Infection

Femtomolar SARS-CoV-2 Antigen Detection 421 Using the Microbubbling Digital Assay with Smartphone Readout Enables Antigen 422 Burden Quantitation and Tracking

N-protein presents early in blood, dried blood 427 and saliva during asymptomatic and symptomatic SARS-CoV-2 infection

Ultra-Sensitive Serial Profiling of SARS-CoV-2 Antigens

Antibodies in Plasma to Understand Disease Progression in COVID-19 Patients with 433 Severe Disease

2020. 435 Analytical sensitivity and clinical sensitivity of the three rapid antigen detection kits for 436 detection of SARS-CoV-2 virus

Limits of 438 Detection of 6 Approved RT-PCR Kits for the Novel SARS-Coronavirus-2 (SARS-439 CoV-2)

SARS-CoV-2 Viral Load in Upper 442 Respiratory Specimens of Infected Patients

RT-PCR for SARS-CoV-2: quantitative 444 versus qualitative

Virological assessment of 448 hospitalized patients with COVID-2019

False Negative Tests for SARS-CoV-2 450 Infection -Challenges and Implications

Distinct Early Serological Signatures Track with SARS-CoV-2 455 Survival

Clinical Predictors of COVID-457 19 Severity and Mortality: A Perspective. Frontiers in Cellular and Infection 458 Microbiology 11

Houhou-Fidouh N, French Covid cohort management 462 committee C-Csg, members of the French Ccsg, member of the Co VCsgPi

Detection of SARS-CoV-2 N-antigen in blood during acute COVID-19 465 provides a sensitive new marker and new testing alternatives

A patient with asymptomatic severe acute respiratory 469 syndrome (SARS) and antigenemia from the 2003-2004 community outbreak of SARS 470 in

Clinical features of patients 474 infected with 2019 novel coronavirus in Wuhan

A minimal common outcome 482 measure set for COVID-19 clinical research

Instruction for Use -Simoa® SARS-CoV-2 Spike IgG Advantage 486 Kit

Jin 491 R, Feng Y. 2021. Quantification of SARS-CoV-2 antigen levels in the blood of patients 492 with COVID-19

Nucleocapsid Plasma Antigen for Diagnosis and Monitoring of COVID-19

Daily monitoring of viral load measured as SARS-CoV-2 antigen and RNA in blood, 500 IL-6, CRP and complement C3d predicts outcome in patients hospitalized with COVID-501 19

Serum levels

Supplemental Table 1: Demographic data of the included population

 No  Item  Reported on page  #  TITLE OR  ABSTRACT  1 Identification as a study of diagnostic accuracy using at least one measure of accuracy (such as sensitivity, specificity, predictive values, or AUC) 3 ABSTRACT 2 Structured summary of study design, methods, results, and conclusions (for specific guidance, see STARD for Abstracts)