Performance of the LIAISON® SARS-CoV-2 Antigen Assay vs. SARS-CoV-2-RT-PCR

Abstract

We aimed to evaluate the LIAISON^® SARS-CoV-2 antigen assay (DiaSorin), comparing its performance to real-time polymerase chain reaction (RT-PCR) for the detection of SARS-CoV-2 RNA. 182 (110 PCR-positive and 72 PCR-negative) nasopharyngeal swab samples were taken for the detection of SARS-CoV-2. RT-PCR and antigen assay were performed using the same material. The sensitivity and specificity of the antigen assay were calculated for different cut-offs, with RT-PCR serving as the reference method. Stored clinical samples that were positive for other respiratory viruses were tested to evaluate cross-reactivity. One third (33/110, 30%) were falsely classified as negative, while no false positives were found using the 200 TCID₅₀/mL cut-off for the SARS-CoV-2 antigen as proposed by the manufacturer. This corresponded to a sensitivity of 70% (60-78%) and a specificity of 100% (94-100%). Lowering the cut-off for positivity of the antigen assay to 22.79 or 57.68 TCID₅₀/mL increased the sensitivity of the method, reaching a sensitivity of 92% (85-96%) vs. 79% (70-86%) and a specificity of 81% (69-89%) vs. 99% (91-100%), respectively. The antigen assay reliably detected samples with high SARS-CoV-2 viral loads (≥10⁶ copies SARS-CoV-2/mL), while it cannot differentiate between negative and low positive samples. Cross-reactivity toward other respiratory viruses was not detected.

Keywords: COVID-19; LIAISON® SARS-CoV-2 antigen assay; RT-PCR; SARS-CoV-2; antigen.

Figure 1

(a) SARS-CoV-2 antigen concentration normalized to baseline values over a four-day period. Values are presented as median (interquartile range). The comparison between the groups was performed using the Friedmann test (p = 0.02). A significant difference can be observed when comparing the first to the fourth day (p = 0.02). All other comparisons did not reach statistical significance. (b) Serial dilution and linearity of the assay. The line represents the actual dilution values on the x- and y-axis, and the triangles and dots represent the measured values for each dilution. The triangles and dots present the average value from two technical duplicates. The assay demonstrates linearity, especially for values over the 200 Tissue Culture Infectious Dose (TCID₅₀/mL) cut-off.

Figure 1

(a) SARS-CoV-2 antigen concentration normalized to baseline values over a four-day period. Values are presented as median (interquartile range). The comparison between the groups was performed using the Friedmann test (p = 0.02). A significant difference can be observed when comparing the first to the fourth day (p = 0.02). All other comparisons did not reach statistical significance. (b) Serial dilution and linearity of the assay. The line represents the actual dilution values on the x- and y-axis, and the triangles and dots represent the measured values for each dilution. The triangles and dots present the average value from two technical duplicates. The assay demonstrates linearity, especially for values over the 200 Tissue Culture Infectious Dose (TCID₅₀/mL) cut-off.

Figure 2

Receiver operating characteristic (ROC) curves for SARS-CoV-2 antigen values of SARS-CoV-2 RT-PCR-positive samples.

Figure 3

A linear relationship was observed between the log values of antigen concentration and viral load (n = 93, values outside the quantified range of antigen concentration (22–100,000) were excluded) (a). Stratifying the samples according to their PCR status as negative (n = 72), positive (n = 110), as well as the more differentiated view of the positive samples, including low positive samples from >10⁴ copies/mL (n = 92), moderately high positive samples from >10⁵ copies/mL (n = 79), high positive samples from >10⁶ copies/mL (n = 66), and very high positive samples from >10⁷ copies/mL (n = 43), revealed that the antigen assay can differentiate between negative samples and samples with high viral load, while a significant overlap can be observed between negative and low positive samples (b).