Comparison of nucleocapsid antigen with strand-specific reverse-transcription PCR for monitoring SARS-CoV-2 infection

Abstract

Background: Tests that sensitively detect the presence of actively replicating SARS-CoV-2 may improve patient care by allowing the safe and timely discontinuation of isolation. Correlates of active replication include nucleocapsid antigen and virus minus-strand RNA. Methods: Qualitative agreement of the DiaSorin LIAISON SARS-CoV-2 nucleocapsid antigen chemiluminescent immunoassay (CLIA) with minus-strand RNA was determined using 402 upper respiratory specimens from 323 patients previously tested using a laboratory-developed SARS-CoV-2 strand-specific RT-qPCR. Nucleocapsid antigen levels, minus-strand and plus-strand cycle threshold values, as well as virus culture, were used to evaluate discordant specimens. Receiver operating characteristic curves were also used to identify virus RNA thresholds for active replication, including values harmonized to the World Health Organization International Standard. Results: Overall agreement was 92.0% [95% confidence interval (CI): 89.0 - 94.5], positive percent agreement was 90.6% (95% CI: 84.4 - 95.0), and negative percent agreement was 92.8% (95% CI: 89.0 - 95.6). The kappa coefficient was 0.83 (95% CI: 0.77 - 0.88). Discordant specimens contained low levels of nucleocapsid antigen and minus-strand RNA. 84.8% (28/33) were negative by culture. Sensitivity-optimized plus-strand RNA thresholds for active replication were 31.6 cycles or 3.64 log10 IU/mL; resulting in 100.0% sensitivity (95% CI: 97.6 to 100.0) and 55.9 specificity (95% CI: 49.7 to 62.0). Conclusions: Detection of nucleocapsid antigen by CLIA performs equivalently to minus-strand detection via strand-specific RT-qPCR, though these methods may overestimate replication-competent virus compared to culture. Careful implementation of biomarkers for actively replicating SARS-CoV-2 has the potential to inform infection control decision-making and patient management.