key: cord-0867128-w5o82mk7
authors: Drevinek, P.; Hurych, J.; Kepka, Z.; Briksi, A.; Kulich, M.; Zajac, M.; Hubacek, P.
title: The sensitivity of SARS-CoV-2 antigen tests in the view of large-scale testing
date: 2020-11-24
journal: nan
DOI: 10.1101/2020.11.23.20237198
sha: 7989cead485d1d022cc51023135765b67c5fce55
doc_id: 867128
cord_uid: w5o82mk7

Objectives: Antigen tests have recently emerged as an interesting alternative to SARS-CoV-2 diagnostic PCR, thought to be valuable especially for the screening of bigger communities. To check appropriateness of the antigen based testing, we determined sensitivity of two point-of-care antigen tests when applied to a cohort of COVID-19 symptomatic, COVID-19 asymptomatic and healthy persons. Methods: We examined nasopharyngeal swabs with antigen test 1 (Panbio Covid-19 Ag Rapid Test, Abbott) and antigen test 2 (Standard F Covid-19 Ag FIA, SD Biosensor). An additional nasopharyngeal and oropharyngeal swab of the same individual was checked with PCR (Allplex SARS-nCoV-2, Seegene). Within a 4-day period in October 2020, we collected specimens from 591 subjects. Of them, 290 had COVID-19 associated symptoms. Results: While PCR positivity was detected in 223 cases, antigen test 1 and antigen test 2 were found positive in 148 (sensitivity 0.664, 95% CI 0.599 - 0.722) and 141 (sensitivity 0.623, 95% CI 0.558 - 0.684) patients, respectively. When only symptomatic patients were analysed, sensitivity increased to 0.738 (95% CI 0.667 - 0.799) for the antigen test 1 and to 0.685 (95% CI 0.611 - 0.750) for the antigen test 2. The substantial drop in sensitivity to 12.9% (95% CI 0.067 - 0.234) was observed for samples with the PCR threshold cycle above > 30. Conclusions: Low sensitivity of antigen tests leads to the considerable risk of false negativity. It is advisable to implement repeated testing with high enough frequency if the antigen test is used as a frontline screening tool.

In our pilot study, we aimed to evaluate the performance of two antigen tests in a scenario 1 close to the population-wide testing. In this regard, we tested a group of nearly 600 people 2 who had no common epidemiological link between each other. 3 4

Subjects. Within a 4-day period in October 2020, we tested 591 individuals of 10 years of 6 age or older, who attended a single collection site, dedicated to the SARS-CoV-2 specimen 7 collection at the Motol University Hospital, Prague, Czech Republic, and consented to the 8 study. The main reasons for their SARS-CoV-2 collection site visit were either the suspicion 9 of COVID-19 infection (273 patients) or contact tracing (290 cases). While 511 persons were 1 0 referred by general practitioner or public health officer, 54 individuals were self-payers. The 1 1 mean age of the cohort was 40 years (age range 12 to 78 years), 44.7% were males. Nearly 1 2 one half of the population (290 subjects) self-reported presence of one or more of the 1 3 following symptoms: cough, pain of muscles and/or joints, chills, diarrhoea and/or vomiting, 1 4 elevated body temperature, loss of smell and/or taste. The study was approved by the 1 5 hospital Ethics Committee (ref no EK-1286/20).

Antigen tests and PCR. Upon the subject's consent, we sampled three separate 1 7 nasopharyngeal swabs and one additional swab from the oropharynx. Two nasopharyngeal 1 8 samples were used onsite for two antigen detection assays according to the manufacturers' 1 9

instructions: Panbio Covid-19 Ag Rapid Test (Abbott, Germany; hereafter referred to as "Ag 2 0 test 1") and Standard F Covid-19 Ag FIA (SD Biosensor, Republic of Korea; hereafter 2 1 referred to as "Ag test 2"). Briefly, the swab was first inserted into an extraction buffer 2 2 provided with the kit, then the amount of 5 (Ag test 1) or 4 drops (Ag test 2) was loaded on 2 3 the test device. The results were read after 15 minutes incubation at room temperature by a 2 4 naked eye (Ag test 1) or after 30 minutes incubation at room temperature on the bench (Ag 2 5

test 2) in the Standard F200 Analyser ('read-only' mode). In order not to unnecessarily lose 2 6 the sensitivity of the assays, we ran antigen tests immediately upon collecting the sample 1 and without the optional step of inserting the swab into the viral transport medium that may 2 lead in undesirable antigen dilution. 3

The remaining nasopharyngeal swab along with the oropharyngeal swab were sampled in 4 accordance with the international specimen collection guidelines (CDC, (9)). They were both 5 inserted into the viral transport medium (10) and transported to the hospital microbiology 6 laboratory for the PCR analysis. RNA extraction was performed with Viral Nucleic Acid 7

Extraction kit (Zybio, China) on the EXM3000 instrument (Zybio, China). The extracts were 8 subjected to the reverse transcription PCR, targeting N, E and RdRP/S genes (Allplex SARS-9 nCoV-2; Seegene, Republic of Korea), run on the CFX96 PCR cycler (Bio-Rad, USA). The 1 0 sample was deemed positive if at least one of the genes was detected with a threshold cycle 1 1 (Ct) value < 40; to define a single Ct for a respective sample, we used the lowest Ct out of 1 2 the three detected targets. 1 3 All rights reserved. No reuse allowed without permission. 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 this version posted November 24, 2020. ; https://doi.org/10.1101/2020.11.23.20237198 doi: medRxiv preprint

The PCR positivity was detected in 223 cases (37.7%). Out of them, 168 people had one or 2 more COVID-19 related symptoms (57.9% of all individuals with symptoms and 75.3% of all 3 PCR positive cases), while 55 PCR positive subjects reported no symptoms at the time of 4 sampling. Ag test 1 and Ag test 2 were found positive in 148 and 141 cases, respectively 5 (Table 1) 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 this version posted November 24, 2020. ; https://doi.org/10.1101/2020.11.23.20237198 doi: medRxiv preprint Test sensitivity increased to 0.738 (95% CI 0.667 -0.799) for the Ag test 1 and to 0.685 4 (95% CI 0.611 -0.750) for the Ag test 2 if only a subgroup of symptomatic patients (290 5 subjects) was analysed (data not shown). On the contrary, a low sensitivity value of 0.436 6 (95% CI 0.314 -0.567) for either of the Ag tests was found in asymptomatic persons (301 7 subjects). 8

The likelihood of detecting the SARS-CoV-2 antigen in a PCR positive person increased with 9 decreasing PCR threshold cycle (Ct) ( 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 this version posted November 24, 2020. ; https://doi.org/10.1101/2020.11.23.20237198 doi: medRxiv preprint < 40 223

168 55 see Table 1a see are added to the samples with lower Ct (Table 2a) , and if samples with lower Ct are taken 5 away from the samples with higher Ct (Table 2b) . 6 7 8 All rights reserved. No reuse allowed without permission. 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 this version posted November 24, 2020. ; https://doi.org/10.1101 https://doi.org/10. /2020 Discussion 1

With the surge of the SARS-CoV-2 epidemic wave in autumn 2020, novel testing strategies 2 to tackle the community transmission are being sought, including the option of population-3 wide screening with the aid of antigen tests (4). In our study, we mimicked a situation of 4 mass screening in that we tested each individual, attending the hospital COVID-19 collection 5 site, regardless of the presence or absence of clinical symptoms. Within 4 days, we enrolled 6 nearly 600 individuals out of 800 eligible people who were 10 years or older. Our PCR 7 positivity rate was almost 38% which was well in accordance with over 30% observed on a 8 national level at the time of the study performance (daily reports on (11)). 9

In our study, we worked with two different antigen tests out of which one enabled europium 1 0 fluorescence-based detection (Ag test 2), the detection that was believed to improve 1 1 sensitivity. However, this mode of result visualisation did not have any impact on the change 1 2 of sensitivity. Overall mean sensitivity values were 66.7% for Ag test 1 and 62.6% for Ag test 1 3

2. If the parameter of the presence of clinical symptoms is a criterion for performing the test, 1 4 sensitivity increased only modestly to 73.8% for Ag test 1 and 68.5% for Ag test 2, while it 1 5 dropped below 50% if asymptomatic, but PCR positive persons are tested.

Our findings fit well with the known sensitivity characteristics of antigen tests for other 1 7 respiratory viruses like influenza or respiratory syncytial virus where rapid 1 8 immunochromatography tests reach the sensitivity of 54.4% and 80%, respectively (12, 13). 1 9

Such discrepancies between sensitivity values reported in research articles and 2 0 manufacturers' leaflets come from differences in selection of tested samples. For instance, a 2 1 clinical evaluation of the Ag test 2 with the claimed sensitivity of 100% was performed on a 2 2 positive spiked material, not on real subjects. Interim data on Ag test 1 showed the sensitivity 2 3 of 85.5% (95% CI 78.2, 90.6), based on testing of 535 patients with suspicion of COVID-19 2 4 (14). Of them, 77% were checked by the antigen test within three days from the onset of the 2 5 symptoms, and 75% had their Ct < 25.

All rights reserved. No reuse allowed without permission. 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 this version posted November 24, 2020. ; https://doi.org/10.1101 https://doi.org/10. /2020 Similarly to Scohy et al. (8) , we found out that the main attribute affecting the antigen test 1 sensitivity is a viral load as estimated by the Ct. Lower viral load, represented in our study by 2 high Ct values above 30, became hardly detectable by any of the two antigen tests used. 3

Thus, patients with samples of late Ct values would be largely left undiagnosed, although 4 many of them also presented with clinical symptoms in our study. Regardless of their clinical 5 state, it is important to point out that they all might be also infectious as documented in 6 studies on asymptomatic and presymptomatic persons (15, 16) . 7

Because of the recent findings on infectivity (17), we can speculate that patients with low Ct 8 were actually at the end of their infection stage and no longer posed the risk to others. 9

However, our patient records show that only 5 of 62 patients with Ct > 30 visited the 1 0 collection site to monitor the course of their infection (3 of them had still clinical symptoms), 1 1 so we can assume that most of them were near the onset of the disease and cannot rule out 1 2 their infectivity. In reference to the aim of our study, i.e. to check the effectiveness of the 1 3 mass testing with antigen tests, another limitation of the study is the bias in selection of the 1 4 subjects who were in most cases indicated for the examination due to symptoms or contact 1 5 tracing. One can expect that the population-wide screening would include higher rate of 1 6 asymptomatic people with low viral loads as well as healthy persons which in turn would lead 1 7 in even higher false negativity and false positivity rates (18). 1 8

To conclude, in our opinion, the risk and rate of false negativity of antigen tests may have a 1 9 significant negative impact on the effectiveness of outbreaks containment as it is crucial to 2 0 early identify any positive person, including the ones with initially low viral load. Not to miss 2 1 them, a single round of testing, which is likely the case when a population-wide screening is 2 2 ordered, seems insufficient and inadequate. Instead, the strategy based on repeated testing 2 3 with high enough frequency (2) needs to be implemented if the antigen test is used as a 2 4 frontline screening tool. 

Presymptomatic SARS-CoV-2 Infections and Transmission in a Skilled Nursing Facility

Duration of Severe Acute Respiratory 6

Syndrome Coronavirus 2 (SARS-CoV-2) Infectivity: When Is It Safe to Discontinue Isolation?

All rights reserved. No reuse allowed without permission. 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 this version posted November 24, 2020. ; https://doi.org/10.1101 https://doi.org/10. /2020 

WHO. Antigen detection in the diagnosis of SARS-CoV-2 infection using rapid 1 4 immunoassays 2020. Available from: https://www.who.int/publications/i/item/antigen-1 5 detection-in-the-diagnosis-of-sars-cov-2infection-using-rapid-immunoassays. 1 6

ECDC. Population-wide testing of SARS-CoV-2: country experiences and potential 1 7 approaches in the EU/EEA and the United Kingdom 2020. Available from: 1 8 https://www.ecdc.europa.eu/sites/default/files/documents/covid-19-population-wide-testing-1 9 country-experiences.pdf. All rights reserved. No reuse allowed without permission. 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 this version posted November 24, 2020. ; https://doi.org/10.1101/2020.11.23.20237198 doi: medRxiv preprint 6.Lambert-Niclot S, Cuffel A, Le Pape S, Vauloup-Fellous C, Morand-Joubert L, Roque-1 Clin Infect Dis. 2020.In press. 8

Sudlow All rights reserved. No reuse allowed without permission. 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 this version posted November 24, 2020. ; https://doi.org/10. 1101