SARS-CoV-2 Antigen Test Results to Infer Active or Non-Active Virus Replication Status in COVID-19 Patients

We used nasopharyngeal swab samples of patients with a symptomatic (n = 82) or asymptomatic (n = 20) coronavirus disease 2019 (COVID-19) diagnosis to assess the ability of antigen detection tests to infer active (potentially transmissible) or inactive (potentially non-transmissible) infection by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Using the subgenomic RNA (sgRNA) as an active replication marker of SARS-CoV-2, 48 (76.2%), 56 (88.9%), and 63 (100%) of 63 samples with sgRNA positive results tested positive with the SD BIOSENSOR STANDARD Q COVID-19 Ag (Standard Q), the SD BIOSENSOR STANDARD F COVID-19 Ag FIA (Standard F), or the Fujirebio LUMIPULSE G SARS-CoV-2 Ag (Lumipulse) assay, respectively. Conversely, 37 (94.9%), 29 (74.4%), and 7 (17.9%) of 39 samples with sgRNA negative results tested negative with Standard Q, Standard F, or Lumipulse, respectively. Stratifying results by the number of days of symptoms before testing revealed that most antigen positive/sgRNA positive results were among samples tested at 2–7 days regardless of the assay used. Conversely, most antigen negative/sgRNA negative results were among samples tested at 16–30 days only when Standard Q or Standard F were used. In conclusion, based on our findings, a negative antigen test, especially with the Lumipulse assay, or a positive antigen test, especially with the Standard F assay, may suggest, respectively, the absence or presence of replication-competent SARS-CoV-2.


Introduction
Reverse transcription-PCR (RT-PCR)-based molecular assays offer a reliable means of detecting the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) genomic RNA (gRNA) in respiratory tract samples from patients who develop coronavirus disease 2019 (COVID-19) within 1 to 2 weeks of SARS-CoV-2 exposure [1]. Whereas there is no established RT-PCR cycle threshold (Ct) value at which individuals are likely to no longer be infectious [2], growing evidence shows that RT-PCR detection of SARS-CoV-2 subgenomic RNA (sgRNA), as with viral culture (i.e., the currently accepted reference standard for virus replication) [3], may be reliably used to measure infectivity [4][5][6]. Unlike diagnostic (gRNAdetecting) RT-PCR, detection of the SARS-CoV-2 antigen, which is per se an indicator of active infection, offers an accurate but less sensitive means of identifying acutely infected individuals [7] and, therefore, the possibility to identify individuals who are shedding the infectious virus and are likely to transmit SARS-CoV-2 [2]. Compared to RT-PCR assays, which have widely replaced viral culture to diagnose respiratory infections [2], antigen assays have 100% specificity, whereas their sensitivity may be moderately high (84%) or low (53%) for symptomatic (i.e., tested within the first 7 days of illness) or asymptomatic patients, respectively [8].
Only one study so far [9], to the best of our knowledge, has compared SARS-CoV-2 sgRNA with SARS-CoV-2 antigen detection in symptomatic COVID-19 patients, showing that antigen was concordant with sgRNA for samples from a midturbinate nasal swab but not from a nasopharyngeal swab (NPS). To clarify this matter, we used NPS samples of asymptomatic or symptomatic COVID-19 patients to assess the antigen detection tests' ability for differentiating between active (potentially transmissible) and inactive (potentially non-transmissible) SARS-CoV-2 infection. We assessed the SD BIOSENSOR STANDARD Q COVID-19 Ag, the SD BIOSENSOR STANDARD F COVID-19 Ag FIA, and the Fujirebio LUMIPULSE G SARS-CoV-2 Ag in comparison with SARS-CoV-2 sgRNA, here used as a marker of active virus replication.

Materials and Methods
Study design and clinical samples. This retrospective study was carried out at the Fondazione Policlinico Universitario A. Gemelli IRCCS hospital of Rome, Italy, and approved by the Institutional Ethics Committee (reference no. 0040205/21), during a 2021 one-month (24 November through 24 December) period. We randomly selected NPS samples from patients who had a laboratory-confirmed COVID-19 diagnosis, which relied on a positive RT-PCR result through a Panther Fusion SARS-CoV-2 assay (Hologic S.r.l., Roma, Italy). Residual 2-mL aliquots from originally collected samples in 3 mL of universal transport medium (UTM, Copan, Brescia, Italy) were put into anonymized blank vials and immediately processed for or kept at −80 • C until testing (see below). One hundred and two NPS samples from patients diagnosed with asymptomatic (n = 20), mild (n = 14), moderate (n = 17), severe (n = 32), or critical (n = 19) COVID-19, according to a median (interquartile range) RT-PCR Ct value of 24.6 (18.6-30.2) for SARS-CoV-2 gRNA detection, were included (Table 1). Only for symptomatic patients (n = 82), disease severity was established according to the criteria from US National Institutes of Health (https://www.covid19treatmentguidelines.nih.gov accessed on 10 May 2022).
Statistical analysis. For SARS-CoV-2 antigen or sgRNA assays' results, differences between a priori established groups were assessed using the chi-square test or the Wilcoxon test, as appropriate. Percent agreement values, with their respective confidence intervals (CIs), were calculated comparing results from antigen assays with those from the sgRNA assay, which was used as the reference method. Statistical analysis was conducted using Stata 15 (StataCorp, College Station, TX, USA) or GraphPad Prism 7 (GraphPad Software, San Diego, CA, USA) software. p < 0.05 was considered statistically significant.
Of the 102 samples studied (Table 1)  SARS-CoV-2, severe acute respiratory syndrome coronavirus 2; sgRNA, subgenomic RNA. A p-value (positive vs negative) of ≤0.05 was considered statistically significant. b Available only for samples with sgRNA levels detected (n = 63, in total), which were distributed among antigen positive or negative testing groups (as detailed in Table 2). All samples in the Lumipulse antigen negative group (n = 7) had sgRNA levels not detected.   Stratifying antigen results by number of days of symptoms before sample collection (Table 2 and Figure 2) showed that Standard F for samples tested at 2-7 days had positive and negative percent agreement values of 96.7% and 100%, which in combination were found to be better than those of Standard Q (93.3% and 100%) or Lumipulse (100% and 0.0%). As illustrated in Figure 2, antigen positive/sgRNA positive results were most frequent among the samples tested at 2-7 days regardless of the (Standard Q, Standard F, or Lumipulse) test used, and antigen negative/sgRNA negative results were most frequent among the samples tested at 16-30 days only when Standard Q or Standard F were used.  Stratifying antigen results by number of days of symptoms before sample collection (Table 2 and Figure 2) showed that Standard F for samples tested at 2-7 days had positive and negative percent agreement values of 96.7% and 100%, which in combination were found to be better than those of Standard Q (93.3% and 100%) or Lumipulse (100% and 0.0%). As illustrated in Figure 2, antigen positive/sgRNA positive results were most frequent among the samples tested at 2-7 days regardless of the (Standard Q, Standard F, or Lumipulse) test used, and antigen negative/sgRNA negative results were most frequent among the samples tested at 16-30 days only when Standard Q or Standard F were used. Stratifying antigen results by number of days of symptoms before sample collection (Table 2 and Figure 2) showed that Standard F for samples tested at 2-7 days had positive and negative percent agreement values of 96.7% and 100%, which in combination were found to be better than those of Standard Q (93.3% and 100%) or Lumipulse (100% and 0.0%). As illustrated in Figure 2, antigen positive/sgRNA positive results were most frequent among the samples tested at 2-7 days regardless of the (Standard Q, Standard F, or Lumipulse) test used, and antigen negative/sgRNA negative results were most frequent among the samples tested at 16-30 days only when Standard Q or Standard F were used.

Discussion
We comparatively assessed SARS-CoV-2 antigen detection tests, i.e., three among a myriad of commercially available tests [13], whose analytical sensitivity (Lumipulse > Standard F > Standard Q) depends on how advanced the immunoassay's technology is (lateral flow assay < immunofluorescence < chemiluminescence). Two are rapid antigen tests [13] and the third (Lumipulse) is a laboratory-based antigen test [12]. Our findings show that a negative antigen test, especially with the Lumipulse assay, or a positive antigen test, especially with the Standard F assay, may be highly suggestive of the absence or presence of replication-competent SARS-CoV-2, respectively. Differentiation between the persistent shedding of viral RNA (i.e., non-infectious status), which is usually associated with a positive SARS-CoV-2 molecular assay [7], and active virus replication (i.e., infectious status) is essential for the clinical management, discontinuation of isolation, and/or work reintegration of SARS-CoV-2-infected individuals. A negative antigen result may be particularly important in the context of a previously positive SARS-CoV-2 molecular assay, i.e., within 10-14 days following symptom onset when both the detection of the infectious virus and contagiousness (or transmissibility) are almost unlikely [14].
Previous studies' findings supported us to adopt SARS-CoV-2 sgRNA RT-PCR as the reference method for our antigen tests' assessment [6,[15][16][17]. One study [15] using respiratory tract samples obtained from COVID-19 patients within 7 days of symptom onset showed the correlation between the SARS-CoV-2 antigen (using the Becton Dickinson BD Veritor System for Rapid Detection of SARS-CoV-2 test) and SARS-CoV-2 culture (using a VeroE6TMPRSS2 cell line-based high-sensitive culture test) positivity. Another study [16] showed that three lateral flow assays and a (microfluidic) immunofluorescence assay correlated with cell-culture infectivity more significantly than SARS-CoV-2 gRNA RT-PCR. Finally, in Santos Bravo et al.'s study [6], sgRNA RT-PCR proved to detect replicationcompetent SARS-CoV-2 in COVID-19 patients' NPS or other respiratory samples with a sensitivity of 97% and a positive predictive value of 94% compared to SARS-CoV-2 culture. However, to comply with the assumption that culture may be suboptimal to detect the presence of the infectious virus due to poor sensitivity [2], we used a highly sensitive RT-PCR method (i.e., the sgRNA detection assay) to assess antigen assays in our study. Similar to us, Santos Bravo et al. [6] chose to measure E-gene sgRNA, i.e., the sgRNA species known as the best option to detect infectivity so far [2].
Our study has some intrinsic limitations. First, the sample size is relatively small, particularly for patients without COVID-19 compatible symptoms who represent only 19.6% of the patients included in the study. This precludes our results from being informative for study populations that primarily consist of asymptomatic patients. Second, antigen assays studied by us are very different platforms, implying their analytical sensitivities are bound to be different. This was especially noted for asymptomatic patients, but it could also explain the high rates of Standard Q or Standard F disagreement with the sgRNA RT-PCR assay that occurred for samples tested after 7 days or more of symptoms. Third, our results suggest that antigen assays can be very helpful at the onset of symptoms, which is pretty much expected.
In conclusion, our findings reinforce the potential of SARS-CoV-2 antigen tests to serve as a marker for active SARS-CoV-2 replication. However, further studies are required before using such tests to predict SARS-CoV-2 infectivity and, thus, inform the discontinuation of COVID-19 transmission-based precautions.