key: cord-266348-tbr2ynx0
authors: Stroemer, A.; Grobe, O.; Rose, R.; Fickenscher, H.; Lorentz, T.; Krumbholz, A.
title: Diagnostic accuracy of six commercial SARS-CoV-2 IgG/total antibody assays and identification of SARS-CoV-2 neutralizing antibodies in convalescent sera
date: 2020-06-17
journal: nan
DOI: 10.1101/2020.06.15.20131672
sha: 
doc_id: 266348
cord_uid: tbr2ynx0

The reliable detection of immunoglobulin G (IgG) or total antibodies directed against the novel severe acute respiratory syndrome coronavirus type 2 (SARS-CoV-2) is important for clinical diagnostics and epidemiological studies. Here, we compare the diagnostic accuracy of six commercially available SARS-CoV-2 IgG (Abbott SARS-CoV-2 IgG; Diasorin Liaison SARS-CoV-2 S1/2 IgG; Epitope EDI Novel Coronavirus COVID-19 IgG ELISA Kit; Euroimmun Anti-SARS-CoV-2 ELISA (IgG); Mikrogen recomWell SARS-CoV-2 IgG) or total SARS-CoV-2 antibody assays (Roche Elecsys Anti-SARS-CoV-2). The test sensitivities were analyzed with a set of 34 sera obtained from 26 patients after PCR-confirmed SARS-CoV-2 infection and varied from 76.9% (Euroimmun) to 96.2% (Abbott). The majority of assay results were confirmed in a laboratory-developed plaque reduction neutralization test and by a SARS-CoV-2 IgG-specific line assay including measurement of generally low IgG avidities (Mikrogen recomLine Coronavirus IgG [Aviditaet], prototype). Moreover, 100 stored sera collected during summer 2018 (N = 50) and winter season 2018/2019 (N = 50) were included to demonstrate test specificities. These varied from 96.0% (DiaSorin) to 100% (Epitope EDI). A subset of sera were retested with a lateral flow test (STANDARD Q COVID-19 IgM/IgG Duo) and a considerably lower sensitivity was noted. Overall, the diagnostic accuracy of the six SARS-CoV-2 IgG/total antibody assays was good and varied from 92.9% (Euroimmun) to 98.4% (Abbott). Due to the different specificities, results of commercially available SARS-CoV-2 antibody tests should be interpreted with caution. A high proportion of antibody-positive patient sera demonstrated neutralizing capacity against SARS-CoV-2.

At the end of the year 2019, Chinese local health authorities reported the occurrence of a cluster of 41 pneumonia cases in Wuhan, a megapolis in the Hubei province [1] . Shortly after, a novel beta coronavirus 42 -which is now designated as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) [2] -was 43 discovered in the bronchoalveolar lavage fluid of a patient with pneumonia [3, 1] . This virus emerged More recently, the measurement of immune response against SARS-CoV-2 came into the focus of clinical 51 diagnostics, particularly by the detection of virus-specific antibodies. The use of SARS-CoV-2 antibody tests 52 could clarify the etiology of the disease in patients who present late, after two weeks from onset of 53 symptoms. Moreover, these tests can demonstrate the viral spread in the community and may even 54 identify individuals who are potentially protected from re-infection by neutralizing antibodies [4] . 55

It is believed that the majority of developed antibodies is directed against the abundant viral nucleocapsid 56 protein while antibodies directed against the spike protein are considered more specific and may possess 57 neutralizing capacity [4] . The diagnostic value of serological tests, however, is limited due to their potential 58 cross-reactivity with other human coronaviruses. Furthermore, it is still unclear, how long SARS-CoV-2 59 antibodies will persist and whether they are protective at all [4] . 60

In the following, we present data on the diagnostic performance of six commercially available SARS-CoV-61 2 IgG or total SARS-CoV-2 antibody tests. Variations in their diagnostic sensitivity and specificity were 62 observed. The latter could lead to a considerable number of false positive results in countries with a 63 currently low SARS-CoV-2 prevalence such as Germany. Moreover, we demonstrate SARS-CoV-2 64 neutralizing antibodies in a marked proportion of convalescent sera. 65 66 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. between four and 60 days (median 19 days) after a positive real-time RT-PCR result in which parts of the 69 SARS-CoV-2 E or N genes [5] were detected in respiratory samples. Only sera from individuals that 70 exhibited a cycle threshold (CT) ≤ 35 (median CT 26.8) in the initial real-time RT-PCR were included. These 71 samples should contain SARS-CoV-2 IgG/total antibodies and, therefore, were considered to determine 72 serological assay sensitivities. Seven subjects with repeated sample entries were included in the study 73 group. Among them were two patients, each with a serum taken 0 and 9 days before the diagnosis of Germany) were included to ensure that cross-reactivity is not occurring. None of these 102 sera were 79 expected contain SARS-CoV-2 antibodies. 80

Test results were also used to calculate the accuracy of the assays, i.e. the proportion of correctly identified 81 samples from all samples. 82

Twelve sera that were sent to us for routine SARS-CoV-2 antibody tests without a previous SARS-CoV-2 83 PCR result were also included to show test performance. 84

The Ethics committee of the Medical Faculty of the Kiel University approved the setting of this study (AZ 85 D467/20). All sera were stored at -20°C until testing. 86

The testing was done with five SARS-CoV-2 IgG assays (Abbott SARS-CoV-2 IgG; DiaSorin Liaison® SARS-87

CoV-2 S1/2 IgG, Diasorin, Dietzenbach, Germany; Epitope EDI™ Novel Coronavirus COVID-19 IgG ELISA Kit, (which was not certified by peer review) 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 June 17, 2020. . https://doi.org/10.1101/2020.06.15.20131672 doi: medRxiv preprint

In general, borderline results were counted as positive. In order to allow a better comparison of the test 97 results, the raw data were converted to relative indices according to the decision limits specified by the 98 manufacturer. A signal/cut-off value of <1 was valued as negative and ≥1 as positive, which corresponds 99 to a previous study [6] . 100

A subset of sera was tested with and without avidity reagent in an IgG line assay ( to an untreated cell-control and a virus control. A serum dilution ≥ 1:20 that prevented the formation of 126 plaques by 50% compared to the virus control (PRNT50) was considered likely to be protective. 127 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 June 17, 2020. . https://doi.org/10.1101/2020.06.15.20131672 doi: medRxiv preprint

Thirty-seven samples taken from 26 patients with a PCR-confirmed SARS-CoV-2 infection were analyzed 129 with six SARS-CoV-2 IgG or total antibody assays. Samples #1, #7, and #9 were taken nine days before to 130 four days after PCR and were all found to be free of SARS-CoV-2 antibodies by the assays. These samples 131

were not considered for calculation of sensitivity but clearly demonstrated seroconversion (Figure 1 ). Out 132 of the remaining 34 samples, only one serum (#20; Figure 1 ) which was obtained ten days after a positive 133 RT-PCR was tested negative for SARS-CoV-2 IgG/total antibodies in the six assays. However, this sample 134 exhibited a PRNT50 1:10 to 1:20. All other samples were reactive in at least one assay (Figure 1 Overall, ten sera were found to be reactive in the six assays, and four of them were found to be reactive 146 in one assay. None contained SARS-CoV-2 neutralizing antibodies (Figure 2, Supplementary material) . 147

Thus, assay specificities ranged from 96.0% to 100.0% (Table 1) . Two samples with a typical constellation 148 of an acute EBV infection did not show cross reactivity with any of the SARS-CoV-2 IgG/total antibody 149 assays (data not shown). 150

Taken together, assay accuracies varied from 92.5% to 98.5% (Table 1) . 151

Twelve routine samples obtained from individuals who were interested in their SARS-CoV-2 antibody 152 status were re-evaluated by all six assays (Figure 3 ). Six of them -including three family members of a 153 confirmed COVID-19 case (#22; Figure 1 ) -were classified SARS-CoV-2 IgG/total antibody positive by the 154 majority of the tests. Isolated reactivity was observed in two sera (#2, #10; Figure 3 ). 155

Thirty of the 37 sera obtained from COVID-19 cases were retested in the IgG recomLine assay. Among 156 them, 27 (90%) were confirmed as SARS-CoV-2 IgG positive (Figure 1, Supplementary material) . One out 157 of the ten archived sera that were reactive in at least one of the six SARS-CoV-2 IgG/total antibody tests 158 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 June 17, 2020. . https://doi.org/10.1101/2020.06.15.20131672 doi: medRxiv preprint was also classified as SARS-CoV-2 IgG positive by the blot results. The underlying sample was collected in 159 winter 2018/2019 and was found to be reactive for SARS-CoV-2 IgG/total antibodies by the Mikrogen 160 (S/SCO 1.33) and Roche (S/SCO 3.46) assays in parallel ( Figure 2 ). As well, nine of the twelve pretested 161 routine samples were analyzed by the blot and pre-known results were largely confirmed (Figure 3,  162 Supplementary material). All measured SARS-CoV-2 IgGs were of low avidity (Supplementary material). 163

Eighteen samples, including eight sera from confirmed COVID-19 cases were retested in a lateral flow 164 assay. Of the eight COVID-19 sera, two were clearly positive for IgG, while three were tested weakly 165 positive and three negative (Figure 1, Supplementary) . Five archived samples that were valued reactive in 166 at least one of the six SARS-CoV-2 IgG/total antibody tests were also re-evaluated by this lateral flow assay. (which was not certified by peer review) 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 June 17, 2020. . https://doi.org/10.1101/2020.06.15.20131672 doi: medRxiv preprint Sample #7 was taken nine days before PCR and the corresponding sample #8 19 days after the PCR. Serum 180 #9 was taken on the day of the PCR and serum #10 29 days thereafter. Results of the rapid lateral flow 181 test, the plaque reduction neutralization assay (PRNT50) and the line assay (blot) are shown in the table. P, 182 positive; wp, weakly positive; n, negative; nd, not determined. 183 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 June 17, 2020. . https://doi.org/10.1101/2020.06.15.20131672 doi: medRxiv preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) 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 June 17, 2020. . https://doi.org/10.1101/2020.06.15.20131672 doi: medRxiv preprint (which was not certified by peer review) 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 June 17, 2020. . https://doi.org/10.1101/2020.06.15.20131672 doi: medRxiv preprint

In a rather short period, various assays for detection of SARS-CoV-2 antibodies were introduced to the 203 market. A meta-analysis of 38 studies on the performance of different format SARS-CoV-2 antibody tests 204 mainly manufactured from Chinese companies was reported recently [9] . Furthermore, studies on the 205 sensitivity and specificity of SARS-CoV-2 IgG tests, including the ELISAs from Epitope, Euroimmun, and 206

Mikrogen, were published previously [6,10,8,7]. Here, we compare six different commercially available 207 tests, all of which have been released within the past few months. Three of them (Abbott, DiaSorin, Roche) 208 were applied in a random access manner which reduces the needed hands-on-time markedly. A subset of 209 samples were retested in a laboratory-developed PRNT50, in an immunoblot including the determination 210 of SARS-CoV-2 IgG avidities and in a rapid lateral flow assay. To our knowledge, such a comprehensive 211 study has not yet been carried out. 212

The sensitivities between the six tests differ from 76.9% to 97.1% (Table 1) , which is in accordance to other 213 studies [6,8,10] and to the meta-analysis [9] . Our data indicate that assays based on the more abundant 214 viral nucleocapsid protein as an epitope are slightly more sensitive compared to tests using domains of 215 the spike protein as an antigen. Furthermore, a lack of reactivity in the latter IgG tests does not necessarily 216 mean that SARS-CoV-2 neutralizing antibodies are missing (compare Figure 1) . However, three sera from 217 two COVID-19 patients, which were only recognized as IgG positive in the Abbott test, could not be 218 The specificities of SARS-CoV-2 IgG/total antibody tests are at a comparatively high level between 96.0 226 and 100.0%, as has also been reported by others [6,8,10]. Interestingly, one sample collected in winter 227 2018/2019 was found to be reactive in the Mikrogen and Roche assays as well as in the blot but could not 228 be confirmed by the PRNT50 (Figure 2, Supplementary material) . For all other 99 archived samples, a 229 random pattern of rare, isolated reactivity was demonstrated and none was confirmed as possessing SARS-230

CoV-2 neutralizing antibodies by PRNT50 (Figure 2, Supplementary material) . In addition, two sera obtained 231 from patients with acute EBV infection were tested negative for SARS-CoV-2 IgG/total antibodies. Thus, 232 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 June 17, 2020. . https://doi.org/10.1101/2020.06.15.20131672 doi: medRxiv preprint

China Novel Coronavirus I, Research T (2020) A Novel Coronavirus from 279 Patients with Pneumonia in China

The species Severe 281 acute respiratory syndrome-related coronavirus: classifying 2019-nCoV and naming it SARS-CoV-2

A novel 284 coronavirus genome identified in a cluster of pneumonia cases -Wuhan

Interpreting Diagnostic Tests for SARS-CoV-2

Detection of 2019 291 novel coronavirus (2019-nCoV) by real-time RT-PCR

Comparison of four new 294 commercial serologic assays for determination of SARS-CoV-2 IgG

Haagmans BL (2020) Severe Acute Respiratory Syndrome Coronavirus 2-299

Specific Antibody Responses in Coronavirus Disease

Clinical performance of SARS-CoV-2 IgG 302 antibody tests and potential protective immunity

Neutralizing antibody responses to SARS-CoV-2 in a COVID-19 recovered patient cohort and their 309 implications

cross-reactivity to epitopes of endemic human coronaviruses or triggered by active EBV infection may not 233 represent a major problem. Nevertheless, it should be noted that due to the currently estimated low 234 prevalence of SARS-CoV-2 antibodies in the overall German population, even a rather high specificity of 235 99% would produce a relevant number of false positive results. Thus, these assays -that all show a 236 comparable high accuracy of 92.5%-98.5% (Table 1 ) -should be preferentially used for testing of patients 237 with a history of a probable infection. In case of doubt, the implementation of the labor-intensive and 238 time-consuming PRNT50 should be considered. 239The majority of SARS-CoV-2 IgGs in sera from COVID-19 patients were confirmed by an immunoblot but for SARS-CoV-2 IgG/total antibodies by at least one of the six assays were all negative by this lateral flow 256 assay. Thus, this rapid antibody test is believed to have a good specificity but sensitivity is reduced. 257Particularly the occurrence of faint IgG bands is a diagnostic challenge. The interpretation of results 258 obtained from this assay should be done with caution. 259Taken together, all six tested SARS-CoV-2 IgG/total antibody assays demonstrate a good performance. 260Their slightly reduced specificities, however, may produce a relevant number of false positive results in 261 low prevalence countries. Future studies should address the long-term course of SARS-CoV-2 antibodies 262 as well as the virus-specific cellular immune response. For the latter, the development of routine test 263 All rights reserved. No reuse allowed without permission.(which was not certified by peer review) 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 June 17, 2020. . https://doi.org/10.1101/2020.06.15.20131672 doi: medRxiv preprint 14 procedures is urgently required. The PRNT results indicate that a high proportion of antibody-positive sera 264 are actually able to neutralize the virus in culture and, thus, may be relevant for immunity. 265 All rights reserved. No reuse allowed without permission.(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

All authors declare no conflict of interest. 276 All rights reserved. No reuse allowed without permission.(which was not certified by peer review) 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 June 17, 2020. . https://doi.org/10.1101/2020.06.15.20131672 doi: medRxiv preprint