Performance of the SD-Biosensor RAT
In order to evaluate the sensitivity and specificity of the SD-Biosensor RAT, paired samples were tested in 4845 individuals. The sensitivity and specificity of the SD-Biosensor RAT was evaluated against three PCR kits; SeeGene (n = 2427), Taqpath (n = 1667) and SD Biosensor (n = 692), as different kits were available in our laboratory during different time periods. All three of these kits have the RdRP gene of the SARS-CoV2 as a target, while SeeGene has N and E genes as the other target genes, Taqpath has N and S as the other genes and SD-Biosensor E gene as the other target gene.
Of the 4845 individuals, 893 gave a positive RT-qPCR result, 3925 were negative and 27 gave an inconclusive result. Of the 893 RT-qPCR positive individuals, the RAT was positive in 324 (36.3%) individuals, negative in 567 (63.5%) and invalid result in 2 individuals. Therefore, the overall sensitivity of the SD-Biosensor RAT was 36.4% (95% CI -33.3–39.8%). Of those who had a negative RT-qPCR result, 96/3925 (2.4%) gave a positive RAT result and therefore, the specificity was 97.7% (95% CI − 97–98%). The positive predictive value (PPV) of the assay compared to RT-qPCR was 76.5% and the negative predictive value (NPV), was 87.6%. The sensitivity of this RAT at different Ct values for different PCR kits is shown in Table 1.
Table 1
Positivity rate of the SD-Biosensor rapid Ag kit for different RT-qPCR kits at different Ct values
PCR Kit | Avg Ct Value for PCR | Sensitivity of the SD-Biosensor Ag kit | 95% CI Lower Limit | 95% CI Upper Limit |
SeeGene Kit | < 25 | 76.6% | 69.5% | 82.5% |
26–30 | 36.2% | 28.1% | 45.1% |
31–38 | 6.1% | 3.3% | 10.6% |
TAQPATH Kit | < 25 | 71.3% | 60.4% | 80.2% |
26–30 | 13% | 6.7% | 23% |
31–38 | 9.5% | 1.7% | 31.8% |
SD-Biosensor RT-qPCR Kit | < 25 | 72.1% | 59% | 82.5% |
26–30 | 7.8% | 2.5% | 19.7% |
31–38 | 0% | 0% | 6.4% |
The viral load in a respiratory sample inversely correlates with the cycle threshold (Ct). As individuals with lower Ct values, have higher viral loads and likely to be more infectious [12], we evaluated the sensitivity of this RAT at different Ct values of the RdRp gene, E gene, N gene and S that was used in the RT-qPCR assay (Fig. 1). For the RdRp gene, the median Ct value of SD-Biosensor RAT positive samples was 26 (IQR 23 to 40) by the SeeGene kit, 25 (IQR 19 to 40) by the Taqpath kit and 10 (IQR 10 to 23) by the SD-Biosensor RT-qPCR kit (Fig. 1A). For the RAT negative samples, the median Ct values for all three kits was 40 (IQR 40 to 40). The S gene target was only present in the Taqpath kit and the median Ct values for RAT positive samples was 26 (IQR 19.75 to 40) (Fig. 1B). Two PCR kits had primers targeting the N gene and the median Ct value of SD-Biosensor RAT positive samples was 17 (IQR 16 to 20) for the SeeGene kit and 26 (IQR 22 to 39) by the Taqpath kit and 26 (IQR 20 to 40) (Fig. 1C). Two kits had primers targeting the E gene and the median Ct value of SD-Biosensor Ag positive samples in the SeeGene kit was 24 (IQR 19 to 35) and for the SD-Biosensor RT-qPCR kit it was 19 (IQR11 to 20) (Fig. 1D).
Performance of the Abbot RAT
For determining the sensitivity and specificity of the Abbott SARS-CoV2 RAT, the RAT and RT-qPCR were carried out in 3625 individuals. The sensitivity and specificity of the Abbott RAT were evaluated against three PCR kits; SeeGene (n = 555), Taqpath (n = 2490) and Hyrbribio (n = 280), as different kits were available in our laboratory during different time periods. Of the 3625 individuals, 327 gave a positive RT-qPCR result, 3288 were negative and 10 gave an inconclusive result. Of the 327 RT-qPCR positive individuals, the RAT was positive in 166 (50.76%) individuals and negative in 161(49.24%). Of those who had a negative RT-qPCR result, 20/3288 gave a positive RAT result. Therefore, the overall sensitivity of the RAT was 50.76% (95% CI − 45.2–56.3%) and the specificity was 99.4% (95% CI − 99.0–99.6%). The PPV of the assay compared to RT-qPCR was 89.2% and the NPV was 95.3%. The sensitivity of this RAT at different Ct values for different PCR kits is shown in Table 2.
Table 2
Positivity rate of the Abbott rapid Ag kit for different RT-qPCR kits at different Ct value
PCR Kit | Avg Ct Value for PCR | Sensitivity of the Abbott Ag kit | 95% CI Lower Limit | 95% CI Upper Limit |
SeeGene Kit | < 25 | 77.3% | 54.2% | 91.3% |
26–30 | 7.1% | 0.4% | 35.8% |
31–38 | 0% | 0% | 22.9% |
TAQPATH Kit | < 25 | 81.4% | 72.9% | 87.7% |
26–30 | 18.7% | 10.9% | 29.7% |
31–38 | 25% | 4.5% | 64.4% |
Hybri-Bio Kit | < 25 | 88.9% | 63.9% | 98.1% |
26–30 | 60% | 17% | 92.7% |
31–38 | 11.1% | 0.6% | 49% |
The sensitivity and specificity of this RAT was also evaluated for different Ct values of the RdRp gene, E gene, N gene and S that was used in the RT-qPCR assay (Fig. 2). For the RdRp gene, the median Ct value of Abbott RAT positive samples was 17 (IQR 14 to 18.25) for the SeeGene kit, 20.5 (IQR 18 to 24) by the Taqpath kit and 22 (IQR 19.25 to 24.75) by the Hybribio kit (Fig. 2A). For the RAT negative samples, the median Ct values for all three kits was 40 (IQR 40 to 40). For the S gene, which was only targeted by the Taqpath kit, the median Ct value of RAT positive samples was 22 (IQR 18 to 26) (Fig. 2B). All three PCR kits had primers targeting the N gene and the median Ct value of Abbott RAT positive samples was 17 (IQR 16 to 20) for the SeeGene kit, 21 (IQR 18 to 26) by the Taqpath kit and 23 (IQR 21 to 26) by the Hybribio kit (Fig. 2C). The SeeGene kit only had primers targeting the E gene and the median Ct value of RAT positive samples was 16.5 (IQR 14 to 18) (Fig. 2D).
Comparison of the Ct values in the samples which gave a positive result by the SD-Biosensor vs the Abbot RAT
In order to determine the differences in detection by the two RATs at different Ct values, we compared the Ct values of the samples which gave a positive result for the SD-Biosensor RAT and the Abbott RAT for the RdRP, S, N and the E genes. This comparison was only carried out for the SeeGene and the Taqpath kits as the SD-Biosensor RT-qPCR kits were not used to validate the Abbott RAT and the Hybribio PCR kit was not used to validate the SD-Biosensor RAT. For RdRP, S, N and the E gene, the Ct values of the Abbott RAT positive samples was significantly less for the SD-Biosensor RAT positive samples for both PCR kits (Fig. 3). For the RdRP gene the Ct values of the Abbot RAT positive samples was significantly less for the SeeGene kit (p < 0.0001) than the Taqpath kit (p = 0.001) (Fig. 3A). The Ct values of the Abbott RAT positives were again significantly lower for the S gene with the Taqpath kit (p = 0.0007) (Fig. 3B) and also for the N gene in the SeeGene kit (p < 0.0001) and the Taqpath kit (p < 0.0001) (Fig. 3C). The Ct values of the Abbott RAT positives were again significantly lower for the S gene with the Taqpath kit (p = 0.0007) (Fig. 3B) and also for the N gene in the SeeGene kit (p < 0.0001) and the Taqpath kit (p < 0.0001) (Fig. 3C). Only the Seegene kit targeted the E gene and again the Ct values of the Abbott RAT positives were again significantly lower in the Abbott RAT positive samples compared to the SD-Biosensor Ag positive samples (p < 0.0001) (Fig. 3D).
Presence of SARS-CoV2 antibodies and the performance of the RATs
Individuals infected with the SARS-CoV2 virus may shed virus for a prolonged period [13, 14], but are often thought to shed dead, or non-culturable virus [15, 16]. The appearance of neutralizing antibodies are thought to associate with the presence of non-culturable and therefore, virus particles that are non-infectious [15]. Therefore, it has been proposed that seroconversion and Ct values of > 24 could be used to release individuals from isolation/quarantine [15]. As one of the most important uses of the SARS-CoV2 RATs it to detect people who are infectious, we proceeded to investigate their performance in individuals from the community and in those who are primary and secondary contacts of the index patient. We obtained blood samples from 2721 individuals at the same time as paired nasopharyngeal samples were obtained for RT-qPCR and antigen testing. The presence of SARS-CoV2 specific antibodies were detected using the Wantai ELISA, which detects the presence of virus specific IgA, IgM and IgG antibodies and was shown to have a sensitivity of 98% [11]. The specificity of this assay was evaluated in the Sri Lankan population in 81 serum samples obtained in 2018 and was found to be 100% specific.
Of the study participants, 435 (15.98 %), were found to have SARS-CoV-2 specific antibodies, although they were not aware of any recent or past COVID-19 illness. 209 (48.04%) individuals who had antibodies gave a positive RT-qPCR result, while 226 (52.4%) SARS-CoV2 antibody positive individuals were negative for RT-qPCR. The presence of SARS-CoV2 antibodies for the three different PCR kits, at different Ct values is shown in Table 3.
Table 3
The presence of SARS-CoV2 specific antibodies for different RT-qPCR kits and different Ct values
Avg Ct Values | Seegene | TAQPATH | SD-Biosensor |
< 25 | 21/80 = 26.25% | 15/63 = 23.81% | 10/24 = 41.67% |
26–30 | 41/52 = 78.85% | 22/62 = 35.48% | 12/17 = 70.59% |
31–38 | 58/70 = 82.86% | 6/12 = 50% | 18/39 = 46.15% |
Of those who had a positive result with the SD-Biosensor RAT 60/187 (32.1%) had detectable SARS-CoV-2 antibodies, while antibodies were detected in 432/1751 (24.6%) of those who tested negative result (Table 4). Of those who tested positive with the Abbott RAT 10/38 (26.3%) were also positive for antibodies, and 109/602 (18.1%) who tested negative for the Abbott RAT, had SARS-CoV2 specific antibodies.
Table 4
The presence of SARS-CoV2 specific antibodies, in those who were tested positive vs negative for the SARS-CoV2 rapid antigen kits
| Ab Positive | Ab Negative | Total |
SD-Biosensor Ag Positive | 60 (32.1%) | 127 | 187 |
SD-Biosensor Ag Negative | 432 (24.6%) | 1319 | 1751 |
Abbott Ag Positive | 10 (26.3%) | 28 | 38 |
Abbott Ag Negative | 109 (18.1%) | 493 | 602 |
Disease symptoms and sensitivity of RAT compared to RT-qPCRs
Four symptoms were recorded at the time of testing in all participants and if they reported at least one of the four symptoms (fever, sore throat, cough or diarrhea) they were considered to be symptomatic. Of the 8470 individuals, 249 (2.94%) were symptomatic and 8221 (97.1%) were asymptomatic at the time of testing. Of those who were asymptomatic 1115 (13.2%) were PCR positive. Of those who were PCR positive, 847 (75.9%) were positive by the SD-Biosensor and 268 (24.03%) by the Abbott RAT. Of those who were asymptomatic and PCR negative, 88/7106 (1.23%) tested positive for the SD RAT and 20/7106 (0.28%) tested positive for the Abbott RAT.
209 (48.4%) of the PCR positive individuals had SARS-CoV-2 antibodies at the time of testing. 8 of them were symptomatic while 201 (96.2%) were asymptomatic. While numbers of RT-qPCR positive, antibody-positive symptomatic individuals is small as expected, there was no significant difference in RT-qPCR positivity and presence or absence of symptoms or antibody positivity.