Comparison of fourteen Rapid Point-of-Care Antigen Tests for SARS-CoV-2: Use & Sensitivity

Fast sensitive techniques are advisable for SARS-CoV-2 detection. Various rapid SARS-CoV-2 antigen detection tests have been developed, but type and quality of the sample, stage of the disease and viral load can all have an impact on their sensitivity. For this study, a total of 486 swabs were processed and checked with various commercially available tests, and then compared with q(RT)-PCR (the gold-standard method). Total sensitivity varied considerably, for example, 42.10% (nal von minden and Tody Laboratories), 68.42% (Cahnos) and 84.78% (PCL). Sensitivity reached 100% when cycle threshold (Ct) was lower than 22 in almost all tests, although this dropped considerably when Ct was higher above 30, where only three tests identified 40% or more positive samples and in 5 cases it was 0%. What is more, only two cases were 100% accurate when viral load was higher than 5 log/103 cells and accuracy was 0% in 12 cases where viral load was lower than 4 log/103 cells. These results, particularly taking into consideration the fact they used normalized viral load, suggest that antigen detection tests have their role in the fast triage of positive patients, but that considerable care should be taken with negative results, which is even more important if they are used for massive screening.


Introduction
Since 11 March 2020, when the WHO declared COVID-19 a pandemic, and now with over 150 million infections and more than 3 million deaths worldwide (1)(2), the long term management of this disease is imperative.
Although q(RT)-PCR is the gold-standard method, it can take at least 3 hours from the point at which the sample suspected of containing SARS-CoV2 arrives at the laboratory and is properly processed for the result to be obtained (3). In contrast, with the various rapid antigen detection tests that have been developed a preliminary result can be obtained in 10-30 minutes (4)(5), providing quick information for the triaging of patients.
However, the type and quality of the sample, stage of the disease and viral load can all have an impact on this latter type of test (5)(6)(7)(8). In order to provide more information on this, here, various different antigen detection tests were compared with q(RT)-PCR.

Material & Methods
Samples between September 2020 and January 2021, 457 positive and 29 negative nasopharyngeal samples were collected from different patients at the Hospital Universitario Central de Asturias.
Genome detection all samples were extracted and puri ed using MagNa Pure 96 system (Roche, Geneva), following the manufacturer instructions.
An in-house q(RT)-PCR test able to detect two targets of the SARS-CoV2 genome (ORF1ab and N gene) as well as the human β-globin gene was performed for each sample. Brie y, 5µL of the sample was added to 10 µL of TaqMan Fast 1-Step Master Mix (Life Technologies, Carlsbad, CA) and supplemented with a mixture of primers (Thermo Fisher Scienti c, Walthman, MA) and taqman MGB probes (Applied Biosystems, Foster City, CA) ( Table 1)
With respect to Ct: total mean sensitivity was 58.36% ± 11.90 with a con dence interval (CI) of 95% (52. 13  The full results are shown in Table 3 and Figure 2.
Speci city was always 100% except for with PCL test, where there was one false positive of 4 samples tested.

Discussion
Under the current emergency measures in force in different parts of the world, and with the fourth wave rising or feared in many countries, the time it takes to process samples is crucial to triage patients, making rapid antigen detection tests can be a very useful assay. But all efforts should be made to ensure the tests used are as sensitive as possible.
When comparing the different tests examined in this study and using the Ct of the q(RT)-PCR as validation, a large range of sensitivity was found, from 42.10% using the tests by the nal von minden and Tody Laboratories to 84.78% with the COVID-19 Ag Gold Saliva from PCL.
As would be expected, we found the correlation between sensitivities on rapid antigen detection test and q(RT)-PCR for Cts under 23 to be practically complete (only the Operon test was not 100% accurate, although it did correctly identify over 90% of positives). However, as soon as Ct is higher, sensitivity decreases notably. With Cts over 23, four of the 14 tests correctly identi ed only 50% of positives, and just ve were accurate in 80% or more of cases. The situation is even more marked when Ct is over 30, where just 2 tests correctly identi ed 50% or more of positives (PCL with 63.63% and Beright with 57.14%).
When the accuracy of the various tests was compared on the basis of the quanti cation of human βglobin, which allows the true measurement of viral load and validates the quality of the sample extraction (13)(14), the results are somewhat different.
In this case, not all tests have a 100% correlation for samples with over 5 log/103 cells, as might be expected. This is very important, both clinically and epidemiologically, because it implies a not insigni cant percentage of false negatives which correspond to contagious patients (15)(16). This indicates, therefore, that these rapid antigen detection tests are not recommended for massive screenings.
For samples with VL of below 4 log, only two methods can detect SARS-CoV-2 antigens, and both at low rates: 31% (Operon) and 20% (PCL). This is to be expected, and is not of great epidemiological signi cance given that patients with low viral load are not considered to be transmitters (12), even though they may in fact be at the beginning of the infection, so could become transmitters at a relatively short later date.
To our knowledge this is the rst study which compares so many rapid antigen detection tests for SARS-CoV2. We counteracted any potential bias of the low number of samples used by checking each sample with each commercial test and, despite the variance derived from the manual procedure of this kind of probes, our results for speci c tests are similar to those obtained in other studies (4-5, 10-11, 17). They are, though, far from the promising results published by the manufacturers themselves, or claimed by certain authors (17) and this emphasises that special care should be taken with the general lack of sensitivity with higher Cts (or low viral load), as results shows that the reliability of obtaining positive PCR and contagion capacity in this range are as yet not well known (15)(16).
This concern is con rmed when a normalized measure of viral load is taken into account. These results also con rm that sampling procedures are very important and rapid tests using easily recovered samples can have compromised sensitivity.
In conclusion, this kind of immunoassay for antigen detection can be useful to ensure the quick isolation of positive patients, but the lack of sensitivity of some tests, even in patients with high viral load, means they miss identifying patients who are positive for SARS-CoV2 who might be infectious, so they must be used with great care.

Declarations
Funding: Not applicable Con ict of interest statement: Authors declare no potential con icts of interest.    Figure 1 Representation of sensitivity percentage of each test by cycle threshold.