key: cord-0310178-96b0ezkf
authors: Lopera, T. J.; Diaz, F. J.; Alzate-Angel, J. C.; Rugeles, M. T.; Aguilar-Jimenez, W.
title: The usefulness of antigen testing in predicting contagiousness in COVID-19
date: 2021-10-18
journal: nan
DOI: 10.1101/2021.10.16.21265067
sha: 3b904059a565d13766b0825e93f08784c4f46c7b
doc_id: 310178
cord_uid: 96b0ezkf

Increasing the diagnostic capacity of COVID-19 (SARS-CoV-2 infection) is required to improve case detection, reduce COVID-19 expansion, and boost the world economy. Rapid antigen detection tests are cheaper and easier to implement, but their diagnostic performance has been questioned compared to RT-PCR. Here, we evaluate the performance of the Standard Q COVID-19 antigen test for diagnosing SARS-CoV-2 infection and predicting contagiousness compared to RT-PCR and viral culture, respectively. The antigen test was 100.0% specific but only 40.9% sensitive for diagnosing infection compared to RT-PCR. Interestingly, SARS-CoV-2 contagiousness is highly unlikely with a negative antigen test since it exhibited a negative predictive value of 99.9% than viral culture. Furthermore, a cycle threshold (Ct) value of 18.1 in RT-PCR was shown to be the one that best predicts contagiousness (AUC 97.6%). Thus, screening people with antigen testing is a good approach to prevent SARS-CoV-2 contagion and allow returning to daily activities.

Coronavirus disease 2019 caused by SARS-CoV-2 has resulted in a global 22 health crisis that requires substantial efforts worldwide to increase the diagnosis capacity and 23 . 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 October 18, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. improve detection of cases by using inexpensive, easy, and rapid testing (1). On the other hand, 24 the economic re-opening requires a diagnostic test before returning to daily activities. Therefore, 25 a negative result in a diagnostic test has become the entrance door to many countries or to other 26 activities that involve some risk of transmission (2). 27 28 Reverse transcription-polymerase chain reaction (RT-PCR) has become the standard gold 29 test for SARS-CoV-2 detection due to its high sensitivity and specificity (3). However, it has been 30 observed that RT-PCR remains positive well beyond the period of contagiousness. Therefore, the 31 return to daily activities is delayed, generating unnecessary restrictions for patients in the 32 convalescent period. In these patients, viral RNA can be detected in low loads (~Ct value > 24 in 33 RT-PCR), and contagiousness is unlikely. For this reason, classical techniques such as viral 34 isolation are used as a better predictor of viral infectivity. However, these methodologies are 35 expensive, risky, and time-consuming. 36

In 2020 point-of-care (POC) tests, such as antigen detection, were developed and approved 38

(4). The first antigen test approved for diagnosis use in Colombia was the Standard Q COVID-19 39

Ag test (SD Biosensor, Republic ok Korea), which is based on immunochromatography and 40 detects the nucleocapsid (N) antigen of SARS-CoV-2 using monoclonal antibodies (5). Antigen 41 detection tests have shown desirable diagnostic characteristics such as reasonable specificity, fast 42 execution, and easy processing (5,6). 43

The sensitivity and specificity of diagnostic tests are essential parameters used to guide 45 decision-making regarding diagnostic tests. However, antigen tests are considered inferior to . 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 October 18, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 performance of antigen tests can be affected by the days of symptoms onset (DSO) and viral load 48 during infection (9,10). Likewise, the usefulness of the antigen test for predicting contagiousness 49 remains to be determined (11). Thus, we aimed to evaluate the performance of a rapid antigen 50 detection test approved for use in Colombia using RT-PCR and viral culture as the standard gold 51 tests for diagnosing infection and contagiousness, respectively. 52 53

Population 55 A sample size of at least 250 specimens with a ratio of positive: negative of 1:1 was 56 calculated, anticipating a 97% specificity according to reported previously (12,13), a probability 57 of type I error of 5%, and a precision of 3%. The inclusion criteria to participate in the study were: 58 adults (≥18 years) with suspicion of SARS-CoV-2 infection either by clinical or epidemiological 59 criteria (clinical symptoms or recent exposure to a confirmed case, respectively). We analyzed 306 60 nasopharyngeal samples from 282 ambulatory subjects (some were sampled more than once as 61 part of their follow-up) recruited at the Grupo Inmunovirología, University of Antioquia, Medellin, 62

Colombia, between September 2020 and January 2021. We excluded pediatric patients and 63 individuals who were unwilling to provide their written consent. The study was designed and 64 conducted following the Declaration of Helsinki and Colombia legislation (Ministry of Health 65 resolution 008430 of 1993). It was approved by the Ethics Committee of the Universidad de 66 Antioquia (Act. 004, 02-04-2020). After explaining the project and clarifying doubts about the 67 research, all included subjects signed the informed consent. The collected biological material was 68 encoded to ensure privacy. 69 . 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) supplemented with 2% fetal bovine serum and 1% Penicillin-Streptomycin. This solution was 105 inoculated into 1 x 10 5 Vero E6 cells monolayers in 12-well plates and incubated for 90 min at 37° 106 C with 5% CO2. Subsequently, the inoculum was removed, and 1,5 mL of DMEM culture medium 107 supplemented with 2% fetal bovine serum and 1% Penicillin-Streptomycin was added. The culture 108 was kept at 37°C with 5% CO2 for 5 days. The monolayer was observed daily for a cytopathic 109 effect (CPE; observed as rounding and detachment of infected cells) indicative of SARS-CoV-2 110 infection. After CPE observation, the supernatants were harvested and stored at -80°C. For 111 confirmation purposes, some samples with or without CPE were evaluated for the presence of 112 SARS-CoV-2 by indirect immunofluorescence (18) or by qRT-PCR on cell supernatants; all of 113 these samples showed a perfect correlation between the presence of CPE and detection of the virus 114 in the cell culture (data not shown). 115 . 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 October 18, 2021. ; https://doi.org/10.1101/2021.10.16.21265067 doi: medRxiv preprint

The Standard Q COVID-19 Ag test was evaluated using the sensitivity, specificity, 118 predictive value, and likelihood ratio calculated using GrapdPad Prism (version 9, California, 119 USA), R v4.1.0, and the Integrated Development Environment Rstudio. To explore variables 120 associated with the result of each test, bivariate analyses were carried-out, and according to 121 statistical (p-value <0.05) and plausibility criteria, they were included in a binary logistic 122 regression model. Receiver Operating Characteristic (ROC) curves were constructed to 123 discriminate the best cycle threshold (Ct) value predicting contagiousness. 124

Results 126

In this study, 306 nasopharyngeal samples were analyzed. All samples were blinded and codified 128 before the analyzes. The median age of subjects was 38 years [Range 18 -96], and 58.9% were 129 female. Considering that the performance of diagnostic tests may fluctuate depending on DSO, we 130 allocate the samples in four diagnostic sceneries: i) 108 samples came from subjects with 1-5 DSO, 131

ii) 50 from individuals with 6-11 DSO, iii) 9 samples were from individuals with more than 11 132 DSO, and iv) 139 samples were from SARS-CoV-2-exposed subjects without symptoms (some 133 symptomatic subjects also disclose previous close contact with a person diagnosed with COVID-134 19) (Figure 1 ). 135

The percentage of positivity in RT-PCR was higher than 70% in individuals with symptoms 137 but below 40% in asymptomatic ones. In contrast, less than 40% positivity by antigen test and viral 138 . 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 October 18, 2021. ; https://doi.org/10.1101/2021.10.16.21265067 doi: medRxiv preprint culture were seen in subjects with either 1-5 or 6-11 DSO, and no positivity was observed in 139 subjects with more than 11 DSO. As expected, the positivity in viral culture decreased as days 140 with symptoms increased (Figure 1 ). 141

The definitive diagnosis of infection was made by RT-PCR. In total, 176 samples were 143 SARS-CoV-2 RT-PCR-positive (57.5% positivity), most of them from outpatients who resolved 144 COVID-19 at home, and seven were from hospitalized patients. The clinical and demographic 145 characteristics of subjects in the study were recorded in search of associations with the result of 146 antigen test (Table S1 ) and viral culture (Table S2) . ageusia, compared to those with a negative antigen test. No differences in gender between subjects 153 with positive or negative antigen tests were observed (Table S1) . 154

The performance of the SARS-CoV-2 antigen test for infection diagnosis was evaluated 156 with RT-PCR as the reference standard. From 176 RT-PCR positive specimens, 72 were positive 157 by antigen test, revealing a low sensitivity (40.9%, confidence interval (CI) 95%= 33.6-48.6%). 158

No negative RT-PCR sample was positive in the antigen test, demonstrating high specificity for 159 SARS-CoV-2 infection (100% CI 95% = 96.4%-100%). Indeed, the probability of COVID-19 was 160 high in the participants with a positive result in the antigen test (positive predictive value, PPV 161 . 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) 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 October 18, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 We carried out a multiple logistic regression model (Table S3 ) with variables such as the 207 Ct value in RT-PCR, and those showing a crude association with antigen test or viral culture results 208 (Tables S1 and S2 ). Although positivity in antigen test was associated with positivity in viral 209 culture [crude OR=134.79 (30.45, 596.64) ], the association was not statistically significant after 210 covariates adjustment [adjusted OR=5.79 (0.5, 66.85) ] (Table S3 ). In contrast, the categorized Ct 211 value was associated with the positivity in viral culture after adjustment. Indeed, the samples with 212 a Ct<20 and <15 had adjusted ORs of 25. 07 (2.27, 277.23) and 290.45 (17.19, 4907.16) ] 213 respectively (Table S3) . We evaluated the Standard Q antigen test performance to diagnose SARS-CoV-2 infection 225 and predict contagiousness. This rapid and simple antigen test is widely used in Colombia and 226 other countries, especially in high prevalence populations; its sensitivity has varied widely across 227 different evaluations (13, 16, (19) (20) (21) (22) . In this study, the sensitivity for infection diagnosis in all 228 samples was low, 40.9% (CI 95%= 33.6,48.6%), much lower than the 84% sensitivity reported by 229 . 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 October 18, 2021. Remarkably, the antigen test and viral culture performance were similar; in both cases, the 250 distribution of Ct values had a median <20 in positive results. Whereas there were no specific 251 symptoms typical in COVID-19 or DSO related to the probability of viral isolation, the antigen 252 . 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 October 18, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 test showed a specificity of 91% to predicts viral infectivity; from the 306 samples, only 2 were 253 discordant between antigen test and viral culture results. In addition, our findings are consistent 254 with studies such as Pekosz et al (28) , who reported that the antigen tests correlate with viral 255 culture of SARS-CoV-2 in 27/28 samples assessed. Yamayoshi et al (26) , in samples with a Ct 256 value <22.5 found that antigen test result was similar to viral isolation in culture (11/11 were 257 positive in antigen test and 8/11 in viral culture). Therefore, these results suggest that an antigen 258 test is a promising tool to discharge patients from quarantine because of the low probability of 259 transmitting the virus. 260

Although viral culture is the most obvious method to assess contagiousness, it is complex, 262 expensive, and requires particular biosafety conditions. The Ct value of RT-PCR, despite the 263 usefulness in the prediction of contagiousness in samples with high viral load, is intrinsically 264 variable (25, 29, 30) . The variation in Ct can be due to the conditions of the RT-PCR reaction, the 265 fluorescence threshold, the fluorochrome used, the gene target, and the virus lineage (10). Antigen 266 test, which targets the more conserved nucleocapsid protein, is potentially a more consistent and 267 cost/benefit method of assessing contagiousness. 268 269 Although some authors have reported viral isolation in culture up to Ct value close to 32 270 (25, 31, 32) , most studies have associated viral isolation with Ct 18-24 (5.4 -7.0 log RNA copies) 271 (33, 34) . Our study indicates that the Ct value that best predicts the result in viral culture is 18.1, 272 with a progressive decrease in the possibility of contagiousness above this Ct value. Hence, we 273

propose that a Ct value higher than 23.5 could be a safe threshold of contagiousness since it was 274 the highest Ct in which we could isolate the virus in culture. 275 . 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 October 18, 2021. ; https://doi.org/10. 1101 /2021 In this study, the time of exposure was not controlled since it was reported by each 277 individual, considered from the last contact with someone with a COVID-19 confirmed diagnosis. 278

Due to this, many individuals did not know precisely the time of exposure to the virus. 279

Additionally, the data correspond to a study of diagnostic tests in which the prevalence of infection 280 is given by the proportion of infected and uninfected subjects who participated in the study, which 281 could not be an exact reflection of the actual prevalence in the population. Further studies should 282 include a serial sampling of the participants in other prevalence scenarios to improve the scope of 283 the results. 284

This study demonstrated that the Standard Q COVID-19 antigen test has excellent 287 agreement with viral culture, indicating that it can be used as a marker of contagiousness. Due to 288 its high positive predictive value in situations of a high prevalence of infection, positive results do 289 not require confirmation with another test. Likewise, its high negative predictive value for 290 contagiousness makes it possible to use this test as a criterion to discharge patients in isolation and 291 screen people moving into environments that could facilitate the transmission of the virus. 292

Additionally, we found that 18.1 is the Ct value in RT-PCR that best predicts the possibility of in 293 vitro viral isolation, related to a higher probability of contagion and viral transmission. 294 295 Acknowledgments 296 . 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 October 18, 2021. ; https://doi.org/10.1101/2021.10.16.21265067 doi: medRxiv preprint . 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 October 18, 2021. ; https://doi.org/10.1101/2021.10.16.21265067 doi: medRxiv preprint The Receiver operating characteristic (ROC) curve shows the Ct value where sensitivity and 432 specificity are higher than 95% to predict the contagiousness using viral culture as reference. Area 433

Under the Curve (AUC) 97.6% (95.6%-99.5%). 434 . 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 October 18, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 Manuscript: The usefulness of antigen testing in predicting contagiousness in 1 COVID-19 2 

Virological assessment of hospitalized cases of coronavirus disease 2019. medRxiv

Nasopharynx of Symptomatic Neonates, Children, and Adolescents. Emerg Infect

Viral Load Analysis of SARS-CoV-2 in Infected Patients

Address for correspondence: Wbeimar Aguilar-Jiménez, Grupo Inmunovirología

) the author/funder, who has granted medRxiv a license to display the preprint in perpetuity

The authors thank the patients and volunteers who kindly participated in this study. We also thank 297 the clinicians who support the recruitment of patients and laboratory assistants who contributed to 298 the sampling. 299 . 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) . 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 October 18, 2021. Evaluation of the accuracy, ease of use and limit of detection of novel, rapid, antigen-detecting 383 point-of-care diagnostics for &lt; em&gt; SARS-CoV-2&lt;/em&gt; medRxiv [Internet] . 2020 384 . 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 October 18, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 . 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 October 18, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 The copyright holder for this preprint this version posted October 18, 2021. ; https://doi.org/10.1101/2021.10.16.21265067 doi: medRxiv preprint . 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 October 18, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021